Strategies for partner notification for sexually transmitted infections, including HIV

Summary of findings for the main comparison. Enhanced patient referral compared with simple patient referral for partner notification for STIs, including HIV

Enhanced patient referral compared with simple patient referral for partner notification for STIs, including HIV
Health problem: partner notification for STIs, including HIV Settings: people in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services Intervention: enhanced patient referral Comparison: simple patient referral
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Simple patient referral	Enhanced patient referral
Re‐infection in index patient ‐ home sampling vs. simple patient referral Follow‐up: 12 months	Study population		RR 2.14 (0.91 to 5.05)	220 (1 study)	⊕⊕⊝⊝ low^1,2
	64 per 1000	136 per 1000 (58 to 321)
	Moderate
	64 per 1000	137 per 1000 (58 to 323)
Re‐infection in index patient ‐ information booklet vs. simple patient referral Follow‐up: 8 weeks	Study population		RR 0.55 (0.22 to 1.33)	942 (2 studies)	⊕⊕⊝⊝ low^3,4
	180 per 1000	99 per 1000 (40 to 239)
	Moderate
	156 per 1000	86 per 1000 (34 to 207)
Re‐infection in index patient ‐ patient referral (DIS/health advisor) vs. patient referral (nurse) Follow‐up: 6 weeks	Study population		RR 0.35 (0.01 to 8.51)	140 (1 study)	⊕⊕⊝⊝ low⁵
	14 per 1000	5 per 1000 (0 to 118)
	Moderate
	14 per 1000	5 per 1000 (0 to 119)
Re‐infection in index patient ‐ disease‐specific website vs. simple referral Follow‐up: 1 weeks	Study population		RR 3.12 (0.17 to 58.73)	105 (1 study)	⊕⊕⊝⊝ low⁶
	0 per 1000	0 per 1000 (0 to 0)
	Moderate
	0 per 1000	0 per 1000 (0 to 0)
Re‐infection in index patient ‐ additional counselling vs. simple patient referral Follow‐up: 6 months	Study population		RR 0.49 (0.27 to 0.89)	600 (1 study)	⊕⊕⊕⊝ moderate⁷
	101 per 1000	50 per 1000 (27 to 90)
	Moderate
	101 per 1000	49 per 1000 (27 to 90)
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; DIS: disease intervention specialist; RR: risk ratio; STI: sexually transmitted infection.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ Method of allocation concealment was not reported. 70% completed follow‐up, some were lost to follow‐up and some withdrew from the study, reasons for withdrawal were not reported. Study was not blinded. ² Assuming alpha of 0.05 and beta of 0.2. For relative risk reduction of 20% with best estimate of control event rate of 0.2 approximately 3000 participants were required. The total sample size was 220 and did not meet the optimal information size. ³ High attrition rate and no information given on method of allocation concealment in one of the studies. Different methods were used for outcome assessment ⁴ I² = 76% (P value = 0.06) and minimal overlap of CIs. ⁵ Sample size less than 400, there were very few events and CIs around both relative and absolute estimates include both appreciable benefit and appreciable harm. ⁶ Sample size was very small and optimal information size was not met. There were very few events and CIs overlapped, therefore, no effect both for absolute and relative estimates. ⁷ Risk for selective reporting and unclear method of allocation concealment.

Summary of findings 2. Expedited partner therapy compared with simple patient referral for partner notification for STIs, including HIV

Expedited partner therapy compared with simple patient referral for partner notification for STIs, including HIV
Health problem: partner notification for STIs, including HIV Settings: people in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services Intervention: expedited partner therapy Comparison: simple patient referral
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Simple patient referral	EPT
Re‐infection in index patients Follow‐up: 2‐12 months	Study population		RR 0.71 (0.56 to 0.89)	6018 (6 studies)	⊕⊕⊕⊝ moderate¹
	110 per 1000	78 per 1000 (62 to 98)
	Moderate
	84 per 1000	60 per 1000 (47 to 75)
Re‐infection in index patients ‐ chlamydia Follow‐up: 3‐12 months	Study population		RR 0.9 (0.6 to 1.35)	2007 (2 studies)	⊕⊕⊕⊝ moderate²
	114 per 1000	102 per 1000 (68 to 154)
	Moderate
	92 per 1000	83 per 1000 (55 to 124)
Re‐infection in index patients ‐ trichomonas	Study population		RR 0.67 (0.34 to 1.28)	631 (2 studies)	⊕⊕⊝⊝ low^3,4
	67 per 1000	45 per 1000 (23 to 85)
	Moderate
	67 per 1000	45 per 1000 (23 to 86)
Re‐infection in index patients ‐ chlamydia or gonorrhoea Follow‐up: 4‐18 weeks	Study population		RR 0.61 (0.39 to 0.94)	3380 (2 studies)	⊕⊕⊝⊝ low^5,6
	116 per 1000	71 per 1000 (45 to 109)
	Moderate
	164 per 1000	100 per 1000 (64 to 154)
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ There was high attrition rate in three of the studies. Methods of sequence generation and allocation concealment not reported in two of the studies. ² CI includes possibility of no effect (i.e. RR of 1.0). ³ Method of sequence generation and allocation concealment not reported in one of the studies. There was high attrition rate in one of the studies. ⁴ Sample size was greater than 400 but CI overlaps, therefore, no effect (i.e. RR of 1.0). ⁵ There were no details on method of sequence generation and allocation concealment. One of the studies had a high attrition rate. ⁶ I² = 74%

Summary of findings 3. Expedited partner therapy compared with enhanced patient referral for partner notification for STIs, including HIV

Expedited partner therapy compared with enhanced patient referral for partner notification for STIs, including HIV
Health problem: partner notification for sexually transmitted infections, including HIV Settings: people in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services Intervention: expedited partner therapy Comparison: enhanced patient referral
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Enhanced patient referral	EPT
EPT vs. enhanced patient referral ‐ re‐infection in index patients Follow‐up: 1‐12 months	Study population		RR 0.96 (0.6 to 1.53)	1220 (3 studies)	⊕⊕⊝⊝ low^1,2
	92 per 1000	88 per 1000 (55 to 140)
	Moderate
	86 per 1000	83 per 1000 (52 to 132)
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; EPT: expedited partner therapy; RR: risk ratio.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ No details on method of sequence generation in one of the studies. One study had high attrition rate and one study used different methods for outcome assessment. ² Sample size is high but CI includes appreciable benefit and harms with both relative risk reduction and increase being greater than 25%.

Summary of findings 4. Contract referral compared with expedited partner therapy for partner notification for STIs, including HIV

Contract referral compared with expedited partner therapy for partner notification for STIs, including HIV
Health problem: partner notification for sexually transmitted infections, including HIV Settings: people in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services Intervention: contract referral Comparison: expedited partner therapy
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	EPT	Contract referral
Re‐infection in index patient Follow‐up: 3 months	Study population		RR 2 (0.7 to 5.72)	322 (1 study)	⊕⊕⊝⊝ low^1,2
	99 per 1000	198 per 1000 (69 to 565)
	Moderate
	99 per 1000	198 per 1000 (69 to 566)
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ Method of sequence generation and allocation concealment not reported. The study had high attrition rate. No blinding. ² Imprecision owing to small sample size.

Background

Description of the condition

Sexually transmitted infections (STI) have a negative impact on the social, health and economic well‐being of a country. Every year an estimated 499 million new cases of the four most common curable STI, trichomoniasis, chlamydia, syphilis and gonorrhoea, are acquired (WHO 2012). Furthermore, two to three million new cases of human immunodeficiency virus (HIV) occur per year (UNAIDS 2010). Up to 4000 infants become blind annually due to eye infections attributable to underlying gonococcal and chlamydial infections in the mother (WHO 2007).

The term STI includes both infections that remain latent or asymptomatic and those that progress to a clinical manifestation (disease). In this update, we used the term STI instead of sexually transmitted diseases (STD), which was used in the original review. STI are more prevalent in countries and communities where socio‐economic conditions are poor (Glasier 2006; Low 2006a). Curable STIs are often overshadowed by the burden of HIV, but are important causes of morbidity in their own right (Table 1).

Table 1. Burden of disease

Disease	DALYs
HIV	58.5 million
Chlamydia trachomatis	3.7 million
Gonorrhoea	3.5 million
Other	280,000

Source: WHO 2004.

DALY: disability adjusted life years.

Clinical symptoms of STIs can be non‐specific and, where possible, the diagnosis needs to be confirmed by laboratory testing. In lower‐income countries, laboratory testing is not always available and women and men reporting symptoms suggestive of an STI are often treated according to algorithms without confirmatory tests. For male urethritis and genital ulcers, this approach is effective but with vaginal discharge the risk of misdiagnosis is high. Syndromic management of STI can therefore lead to over‐treatment and adverse social consequences such as stigma and intimate partner violence (Trollope‐Kumar 2006). Women are more likely than men to suffer from reproductive tract complications of STIs such as chlamydia and gonorrhoea if the infection ascends to the upper genital tract; pelvic inflammatory disease (PID), ectopic pregnancies and infertility are the most commonly documented complications (Gerbase 1998). STIs are, however, often asymptomatic in both women and men (WHO 2007). As a result, disclosing a diagnosis of an STI to sexual partners and partner treatment play a critical part in the comprehensive management of STI. Willingness to disclose varies according to the STI and gender (Alam 2010). In one study among people with a diagnosis of HIV, 85% of people living with HIV were sexually active, but only 58% revealed their HIV status to recent sexual partners (Simbayi 2007). In a study in Connecticut, US, 25% of females with chlamydia intended not to notify their partners (Niccolai 2007) as most (46%) thought it unimportant and 43% were not willing to discuss the condition. In a study in India, the patient characteristics most likely to increase the odds of referring a partner were having a diagnosis of genital ulcer disease (odds ratio (OR) 2.78, 95% confidence interval (CI) 1.08 to 7.13, P value = 0.033) and having the intention to inform the regular partners (OR 16.9, 95% CI 3.29 to 86.70, P value = 0.001) (Sahasrabuddhe 2002).

Description of the intervention

"Partner notification is a process that includes informing sexual partners of infected people of their exposure, administering presumptive treatment, and providing advice about the prevention of future infection" (UNAIDS 1999). Partner notification (PN) is also known as contact tracing, partner management or partner information. A person with a newly diagnosed STI is often referred to as an 'index case' or 'index patient'. The index patient has one or more sexual partners. The sexual partners of the index patient might have been the source of the infection in the index patient or they might have acquired the infection from the index patient.

A variety of approaches has been used to notify sexual partners and to ensure that they receive treatment. In principle, managing infection in people with more than one current sexual partner should have the greatest impact on the spread of STI (Fenton 1997). The use of different approaches depends partly on the STI for which they were originally intended. There are other influences at the country level, including cultural factors, the structure and financing of health systems, and clinical consensus. At the individual level, factors such as patient choice influence choice of PN strategies. Traditionally, three main approaches have been defined: patient referral, provider referral and contract (or conditional) referral. Definitions and explanations of these PN methods are given below.

Patient referral (patient‐led referral) refers to an approach in which health service personnel encourage index patients to notify their own partners. In this review, we used the term simple patient referral to refer to spoken advice from health service personnel about the need for sexual partners to receive treatment. This can be seen as a minimum standard for a PN intervention. There is, however, no agreement about the content of a consultation for simple patient referral. Patient referral was developed in the 1970s when rates of gonorrhoea in the US were very high and the capacity of specialist PN personnel was exceeded. Patient referral has since become the preferred method of PN for gonorrhoea and subsequently chlamydia in many countries. There has been great interest in developing methods to support index patients so that the outcomes of patient referral can be improved or enhanced (Trelle 2007). Patient referral can, therefore, be split into two categories (simple and enhanced), according to the level of support given to the patient. Expedited partner therapy (EPT) has developed in the US since the late 1990s as a new patient‐led strategy to help index patients to get their partners treated more quickly.

Enhanced patient referral refers to a group of strategies that supplement the spoken advice with the aim of improving patient referral success, including educational material such as videos viewed in waiting rooms, written disease‐specific information for index patients to give to their partners, home sampling kits for partners, disease‐specific websites, theory‐based counselling and reminders by telephone or other means (Trelle 2007).

EPT is a group of strategies to enhance the success of patient referral by increasing the numbers of partners treated and speeding up the time to treatment (CDC 2006). The EPT strategies include: patient‐delivered partner medication (PDPM) or patient‐delivered partner therapy (PDPT), where the index patient receives antibiotics (often in a package with condoms and written information) to give to their partner without the need for a medical examination of the partner (Golden 2005); or additional prescriptions given to index patients for their partner(s). EPT can reduce loss to follow‐up of index cases (Young 2007), and reduce the risk of repeated infection in the index case (Golden 2005). There are, however, disadvantages, including the risk of adverse drug reactions, other underlying disease remaining undetected and a missed opportunity for counselling and testing for other STIs including HIV (Golden 2005). In some countries, such as the UK, EPT is not legal unless the partner is assessed before receiving antibiotic treatment (ECDC 2013).

Provider referral (provider‐led referral) uses third parties (usually specialist health service personnel) to notify partners. The name of these health professionals differs between countries, for example; 'disease intervention specialists' (DIS) in the US; 'health advisers' in the UK and 'Kurators' in Sweden. Provider referral originated in Scandinavia and the UK as a method to trace and refer the sexual partners of people with syphilis when treatment first became available. More recently, it has been used for other clinically severe STIs such as HIV infection and hepatitis B. It can also be used for other STIs such as gonorrhoea and chlamydia when the index patient is unable to notify partners by themselves. Provider referral should only be done with the explicit consent of the index patient. In some countries, for example France, provider referral does not occur because it is seen as an invasion of privacy (ECDC 2013).

Contract referral (conditional referral) refers to an approach in which there is an agreement (contract) between the patient and the health professional. Health service personnel encourage index patients to notify their partners, with the understanding that health service personnel will notify those partners who do not visit the health service by an agreed date. Contract referral is, in practice, difficult to define as a separate PN approach. It can be difficult to distinguish from provider referral if the time window for patient referral is very short (two or three days) (Peterman 1997). In contrast, contract referral is often used as an extension to simple patient referral, rather than a separate strategy, if the index patient has not been able to inform their partner(s) when they are followed up.

How the intervention might work

There are different aims of PN, depending on the level at which it is targeted and the infection (Low 2006a). At the level of the index patient with a curable STI the aim is to provide concurrent antibiotic treatment to the sexual partner(s) so that infection can be eradicated in both people and re‐infection prevented in the index patient, which is a clinical goal. For the sexual partner(s) the aim is to identify and treat infection that might have been the source of infection in the index patient, or might have been acquired from the index patient. At the level of sexual networks and populations, the aim is to interrupt chains of transmission and reduce the spread of STIs, which is a public health goal. For viral STIs, the aim is to identify previously undiagnosed infections, which can provide early access for sexual partners to treatment and prevent onward transmission through behavioural change by the infected person.

To succeed, PN strategies need to first elicit from the index patient details of all sexual partners from whom he/she may have acquired the infection, or whom he/she might have subsequently infected. Identifying partners in the latent period of infection (usually three months for primary syphilis and one month for acute urethritis) (Toomey 1996), should identify those from whom infection was acquired, while identifying partners after the onset of symptoms will identify those who were likely to have been infected by the index case. The time period for identifying partners differs between countries for different STIs.

For most PN strategies, eliciting partner information from infected people is a prerequisite to notifying sexual partners. For example, when health service personnel notify partners, they rely on the index patient to count, name and provide details to enable all his/her partners to be traced. Once partners have been elicited, PN strategies need to provide either the index patient or the health service personnel with the necessary knowledge, skills or resources to enable them to locate, notify, medically evaluate and test or treat these partners.

Communication between partners, during which the index patient encourages them to consider screening or treatment, has been identified as a critical point in effective PN strategies (Young 2007). The communication usually requires the index patient to disclose their STI diagnosis. Disclosure can lead to benefits other than successful partner treatment, such as emotional support and protecting the health of others. Disclosure can also lead to stigma, rejection, physical abuse and discrimination (Arnold 2008).

Why it is important to do this review

PN has been practised as a measure to control STIs since the early 1900s (ECDC 2013), but there is limited evidence of its public health impact. Many evaluations have not been conducted as randomised controlled trials (RCTs) and many were conducted in developed countries before the HIV/acquired immunodeficiency syndrome (AIDS) pandemic. It is not known whether interventions developed for high‐income countries are applicable to resource‐limited settings.

There are several published systematic reviews of PN. The first included only studies conducted in developed countries (Oxman 1994). Another included only published studies conducted in the US after 1980 (Macke 1999). The original Cochrane Review by Mathews et al. was assessed as up to date in July 2001 (Mathews 2001). Trelle et al. systematically reviewed studies of enhanced methods of patient referral, including EPT, to improve the effectiveness of simple patient referral (Trelle 2007). The latest systematic review only studied curable STIs in developing countries (Alam 2010). Considering the ongoing developments in this field, the Cochrane Review was updated in line with recommendations of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Objectives

To assess the effects of alternative PN strategies.

Methods

Criteria for considering studies for this review

Types of studies

We included RCTs that compared at least two PN strategies.

Types of participants

People in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services with any of the following STI: gonorrhoea (Neisseria gonorrhoeae), chlamydia (Chlamydia trachomatis), trichomoniasis (Trichomonas vaginalis), syphilis (Treponema pallidum), chancroid (Haemophilus ducreyi), genital herpes, hepatitis B and HIV. We also included diagnoses of the following STI syndromes: genital ulcer syndrome ‐ non‐vesicular or vesicular, urethral discharge syndrome, vaginal discharge syndrome and lower abdominal pain in women. Studies conducted in any type of health service were included.

Types of interventions

Strategies directed at patients (patient‐led) or health workers (provider‐led) were included. The following types of strategies were included:

strategies to enhance the effectiveness of patient referral through, for example, health education and counselling, health education materials (such as pamphlets, posters, video and audio productions), patient assistance strategies directed at facilitating patient referral (such as referral cards, incentives, reminders, video and audio productions). EPT was included as a specific type of enhanced patient referral;
contract referral strategies;
provider referral strategies;
combinations of the above.

Types of outcome measures

Primary outcomes

Number of index patients with curable STIs given a clinical or laboratory diagnosis of re‐infection. Re‐infection implies re‐infection of the index patient with the same STI from an untreated sexual partner. In practice, the outcome measured is repeated detection of the STI at some time interval after the index case has been treated. Repeated detection of an STI could also result from a new infection in the index case acquired from a new sexual partner, or treatment failure due to antibiotic resistance or subtherapeutic dosing. These causes cannot be reliably distinguished and the term re‐infection is used to include repeated detection from any cause.

Secondary outcomes

Numbers of partners elicited (sexual partners that the health professional obtains from the index patient for the recall period in question), located (sexual partners that the index patient was able to find; this number is likely to be a subset of partners elicited), notified (sexual partners that the index patient informed of their possible exposure to an STI; this number is likely to be a subset of partners located), presenting for care, testing positive or treated per index case; delay in partners presenting for care; incidence of STIs; changes in the index patient's or partner's behaviour with regard to condom use, abstinence in the presence of symptomatic infections, the number of partners, the number of concurrent partners; emotional impact on the index patient or partner in their relationship; harm to the patient or partners, such as domestic violence, abuse or suicide; ethical outcomes (patient autonomy vs. beneficence).

Search methods for identification of studies

Electronic searches

Search method for original review (Mathews 2001)

The original review authors searched MEDLINE (1966 to 24 July 2001), EMBASE (1974 to 24 July 2001), Psychological Abstracts (1967 to 24 July 2001) and Sociological Abstracts (1963 to 24 July 2001). The Cochrane Controlled Trials register was searched with the text words 'sexual partners', 'partner notification', 'contact‐tracing' and 'contact tracing'. The Effective Practice and Organisation of Care (EPOC) register of studies was searched, as was the register of the HIV and AIDS Cochrane Review Group.

Search method for the review update

We searched three electronic databases, MEDLINE, EMBASE and CENTRAL, from 5 January 2001 to 31 August 2012. Search strategies are shown in Appendix 1, Appendix 2 and Appendix 3.

Searching other resources

Original Cochrane review (Mathews 2001)

The original review authors handsearched the Proceedings of the International AIDS Conferences (1996 to 24 July 2001) and the International Society for STD Research meetings (ISSTDR) (1991 to 24 July 2001). Bibliographies of studies and previous reviews were examined for references to other trials. Experts in the field were contacted.

Review update

We searched all reference lists of potential studies and previous reviews for relevant RCTs and contacted experts in the field. We searched the International Clinical Trials Registry Platform (ICTRP) from 18 March 2011 to 31 August 2012 to identify ongoing studies (www.who.int/ictrp/en/). We searched the ICTRP for the protocols of the 16 new studies. Trial registries were not searched for the protocols of the original included studies because these were all published before 1998.

Data collection and analysis

Selection of studies

Two review authors (Cathy Mathews, CM and Riabatu Abdullah, RA (original review); and Adel Ferreira, AF and Taryn Young, TY or CM or Moleen Zunza, MLZ (update)) independently screened titles and abstracts of the electronic search results. We obtained all the eligible abstracts of comparative studies in full‐text format, and two review authors (CM and RA original review and AF and TY or CM update) independently reviewed them for inclusion using prespecified eligibility criteria. We included all studies that reported random allocation. We assessed the risk of bias in the methods of sequence generation and allocation, as described in the section 'Assessment of risk of bias in included studies' and considered risk of bias interpreting the strength of evidence for each intervention.

Data extraction and management

Two review authors (CM and Nicol Coetzee, NC or Merrick Zwarenstein, MZ (original review) and AF and TY or CM or MLZ (update)) independently abstracted study characteristics and outcomes including information on: social context (developing (World Bank classification: countries with low or middle levels of gross national product (GNP) per capita as well as five high‐income developing economies ‐ Hong Kong (China), Israel, Kuwait, Singapore and the United Arab Emirates. These five economies are classified as developing despite their high per‐capita income because of their economic structure or the official opinion of their governments. Several countries with transition economies are sometimes grouped with developing countries based on their low or middle levels of per‐capita income, and sometimes with developed countries based on their high industrialisation (World Bank 2012)) or developed country); access to health services; legislative context (permissive or proscriptive public health legislation); methodological quality of study; type of health facility; type of provider (for example, nurse, physician, DIS); participants; type of interventions; outcome measure; results and correspondence required using a data extraction form.

We resolved disagreements by discussion. We summarised data from included studies in the Characteristics of included studies table and data from excluded studies in the Characteristics of excluded studies table. We summarised studies with insufficient information in the Characteristics of studies awaiting classification table. Where there were missing data, we attempted to contact study authors by email.

Assessment of risk of bias in included studies

Two review authors (AF and CM or MLZ) independently evaluated the risk of bias using The Cochrane Collaboration's tool (Higgins 2011a). We made judgements about the presence of bias by selecting one of three categories of risk of bias: low risk, high risk and unclear risk of bias. We resolved disagreements by discussion. If we could not reach consensus, we involved a third independent review author (TY). We contacted trial authors if there were any unclear issues and, if we received no response, we made a judgement of 'unclear risk of bias'.

We assessed and summarised the following main items in the 'Risk of bias' table: sequence generation, allocation concealment, blinding of participant and personnel, blinding of outcome assessment, whether incomplete outcome data were adequately addressed, selective reporting and any other bias. We searched the ICTRP for protocols of the 16 additional studies to assess selective reporting bias. Figure 1 and Figure 2 show the 'Risk of bias' graphs, which illustrate the proportions of studies with low, high and unclear risk of bias. In the 10 studies of the original review, the ICTRP was not searched; instead, the methods and result sections were compared to evaluate if the same outcomes were reported in these two sections. If the protocol was not available, the methods and results sections were compared to assess selective reporting bias.

Figure 1

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Measures of treatment effect

The review authors prepared tables summarising the results of each study for each comparison.

We defined re‐infection rate in index patients as the percentage of index patients with a repeated diagnosis of the same STI divided by the number of index patients retested.

Partners elicited, notified, presenting for care, tested, treated or harmed: we assumed that the number of units of each outcome per index patient was a random variable following a Poisson distribution. We assumed that the index patients from the groups within a study had similar distributions for exposure time to partners, for time to notify their partners, and that the same assumption held for partners with respect to the time taken to present to the health service. The value of the mean and the variance of a Poisson distribution are the same.

To calculate a CI for the difference in relevant outcomes, we used the normal approximation to the Poisson distribution since only summarised data from the included RCTs were available.

The approximate 95% CI for the rate difference is given by:

(Lamda1 ‐ Lamda2) ± 1.96√ (lamda1/n1 + lamda2/n2),

where lamda1 and lamda2 are the rates of partners per index patient in two groups, and n1 and n2 the number of index patients.

To calculate the standard error (SE) the formula used was:

(upper limit of 95% CI ‐ lower limit of 95% CI)/3.92.

To calculate the standard deviation (SD) the formula used was:

SE/√ (1/Nexp+ 1/Ncont),

where Nexp is the number of index patients randomised to the experimental group and Ncont is the number of index patients randomised to the control group

For continuous outcomes (number of partners elicited, notified, presenting for care, tested, treated or harmed), we recorded the mean (in number of partners per index patient randomised), SE and sample size. Where the exact numbers of partners were not available, we contacted study authors. If authors did not respond or could not provide the exact numbers, the mean difference (MD) could not be calculated and we reported the study findings descriptively. In studies where the rate of partners elicited per index patient was not reported, we used the number of contact cards given to the index patient as a proxy indicator.

We described the delay in partners presenting for care as the mean or median number of days after index patient enrolment.

Unit of analysis issues

We dealt with studies with multiple intervention groups as recommended in the Cochrane Handbook for Systematic Intervention Reviews (Higgins 2011b). We compared each intervention arm with another.

Where this resulted in shared intervention groups, we did not perform a meta‐analysis to prevent 'double‐counts' of participants. In these studies, we described the results in narrative form (Ellison undated; Montesinos 1990). We did not include any cluster randomised trials and, therefore, no adjustments were necessary.

Dealing with missing data

Where there were missing data, we attempted to obtain the data by contacting study authors by email. We contacted the authors of eight trials and authors provided requested data for five of the eight trials.

Assessment of heterogeneity

We assessed sources of clinical and methodological heterogeneity by looking at characteristics of studies, evaluating similarity between type of participants, intervention used and outcomes. We calculated the Chi² test for heterogeneity (Deeks 2011), and the I² statistic to evaluate statistical heterogeneity. Values of the I² statistic were interpreted as follows (Deeks 2011): 0% to 40%: might not be important; 30% to 60%: might represent moderate heterogeneity; 50% to 90%: might represent substantial heterogeneity; 75% to 100%: might represent considerable heterogeneity.

Assessment of reporting biases

We did not find a sufficient number of studies to produce funnel plots to investigate publication bias for specific comparisons.

Data synthesis

We analysed data according to paired partner referral strategies (Table 2). We organised the comparisons first according to the four main PN strategies (1. enhanced patient referral, 2. EPT, 3. contract referral, 4. provider referral). Each main strategy was compared with simple patient referral and then with each other, if trials were available. We compared each enhanced patient referral with another enhanced patient referral. This resulted in 10 comparisons (Table 2).

Table 2. Summary of comparisons with data available and STI studied

Partner notification strategy, intervention	Partner notification comparator, comparison number (number of trials)					STI included in trials
Partner notification strategy, intervention	Simple patient referral	Enhanced patient referral	Expedited partner therapy	Contract referral	Other enhanced patient referral	STI included in trials
Enhanced patient referral	1 (16)	‐	‐	‐	2 (2)	Gonorrhoea, chlamydia, non‐gonococcal urethritis, trichomonas, pelvic inflammatory disease, STI syndromes
Expedited partner therapy	3 (8)	4 (5)*	‐	‐	Not applicable	Gonorrhoea, chlamydia, trichomonas, STI syndromes
Contract referral	5 (5)	6 (1)	7 (1)	‐	Not applicable	Gonorrhoea, trichomonas, HIV
Provider referral	8 (3)†	9 (1)	No trials	10 (2)	Not applicable	Non‐gonococcal urethritis, syphilis, HIV

* Comparison includes one trial comparing combinations of expedited partner therapy and patient referral.

† Comparison Includes one trial comparing a choice between provider or simple patient referral and simple patient referral.

‐ Indicates combinations of an intervention and comparison that are covered elsewhere in the table; HIV: human immunodeficiency virus; STI: sexually transmitted infection.

The largest group of trials (Table 2; comparison 1, enhanced patient referral versus simple patient referral) included several different interventions to enhance the outcomes of patient referral. We grouped these into six categories: (1) patient referral with DIS or health adviser, (2) postal testing kit, (3) information booklet, (4) disease‐specific website, (5) additional counselling or (6) showing a videotape.

We performed meta‐analyses where appropriate using random‐effects models to report the pooled MD (for continuous outcomes) or risk ratio (RR for dichotomous outcomes) with 95% CI. When there was a moderate or low level of heterogeneity (I² ≤ 50%), we pooled results. If there was more substantial evidence of heterogeneity (I² > 50%), we pooled the results of individual studies if appropriate or described in the narrative. We reported results of tests for heterogeneity (Tau², Chi² test with number of degrees of freedom (df), P value and I² statistic).

Subgroup analysis and investigation of heterogeneity

We used subgroup analyses to explore possible sources of heterogeneity. These included: age of participant, gender, specific STIs investigated, setting (developed vs. developing country) and category of healthcare worker.

Sensitivity analysis

We performed a sensitivity analysis on the primary outcome, re‐infection rate of index patient with curable STIs. Given the limited numbers of trials and meta‐analyses, the sensitivity analysis examined only the effect of attrition bias. We repeated meta‐analyses excluding trials with more than 20% attrition and compared results with the primary analysis.

'Summary of findings' table

We interpreted results using a 'Summary of findings' table, which provided key information about the quality of evidence for the studies included in a comparison, the magnitude of effect of the interventions examined and the sum of available data on the primary outcome. We imported data from Review Manager 5 (RevMan 2011), using the GRADE profiler (GRADE 2004). We selected the primary outcome of re‐infection in the index case for the 'Summary of findings' table.

Results

Description of studies

Results of the search

The initial search (1966 to 24 July 2001; Mathews 2001) identified 11 RCTs, including 8041 participants. The updated search (5 January 2001 to 31 August 2012) identified an additional 16 RCTs (9597 participants; 6841 women and 2756 men). One study was listed as awaiting classification (Characteristics of studies awaiting classification). In the original review, Levy 1998 (with 60 participants) was listed as under 'Included studies' but, in this update, it was placed under 'Characteristics of studies awaiting classification' because no results were available. We found four ongoing studies in trial registers (Characteristics of ongoing studies).

Included studies

Twenty‐six RCTs (Figure 3) were included in the review including 17,578 participants (Characteristics of included studies). Most of the trials (14) were conducted in the US, four in the UK, two in Denmark, and one each in Australia, Malawi, South Africa, Uganda, Zambia and Zimbabwe. Most trials (21) were based in public health clinics. One was conducted in a large academic medical centre (Trent 2010), three in general practice (Andersen 1998; Low 2006b; Ostergaard 2003), and one on a university campus (Montesinos 1990).

Figure 3

Flow diagram detailing the updated search and selection of studies.

Participants

Trials were conducted among patients with gonorrhoea (three trials, Cleveland undated; Potterat 1977; Solomon 1988); gonorrhoea or non‐gonococcal urethritis (one trial, Montesinos 1990); non‐gonococcal urethritis only (one trial, Katz 1988); chlamydia (six trials, Andersen 1998; Apoola 2009; Cameron 2009; Low 2006b; Ostergaard 2003; Schillinger 2003); syphilis (one trial, Peterman 1997); HIV (two trials, Brown 2011; Landis 1992); chlamydia or gonorrhoea, or both (four trials, Golden 2005; Kerani 2011; Kissinger 2005; Wilson 2009); trichomonas (two trials, Kissinger 2006; Schwebke 2010); PID (one trial, Trent 2010); and chlamydia or non‐gonococcal urethritis (one trial, Tomnay 2006). Four trials in developing countries where syndromic diagnoses are made included patients with any STI syndrome (Ellison undated; Faxelid 1996; Moyo 2002; Nuwaha 2001). In six studies, STI diagnoses were made clinically, based on symptoms or clinic tests (Ellison undated; Faxelid 1996; Katz 1988; Moyo 2002; Nuwaha 2001; Trent 2010). In the other 20 trials, STI diagnoses (other than non‐gonococcal urethritis) were confirmed with laboratory testing. There were no RCTs among patients with laboratory‐diagnosed hepatitis B, genital herpes or chancroid.

Six trials included male patients only, or reported over 90% male index patients (Cleveland undated; Katz 1988; Kerani 2011; Kissinger 2005; Potterat 1977; Solomon 1988). Seven trials included female index patients only (Andersen 1998; Apoola 2009; Cameron 2009; Kissinger 2006; Schillinger 2003; Schwebke 2010; Trent 2010).The remaining trials included male and female index patients. Two trials included men who had sex with men (Kerani 2011; Landis 1992) and one included male and female injecting‐drug users (Landis 1992).

Types of interventions

Included studies investigated the effects of various PN strategies (Table 2; Table 3):

See: Summary of findings for the main comparison Enhanced patient referral compared with simple patient referral for partner notification for STIs, including HIV; Summary of findings 2 Expedited partner therapy compared with simple patient referral for partner notification for STIs, including HIV; Summary of findings 3 Expedited partner therapy compared with enhanced patient referral for partner notification for STIs, including HIV; Summary of findings 4 Contract referral compared with expedited partner therapy for partner notification for STIs, including HIV

Table 3. Summary of included studies and outcomes reported by authors, according to partner notification strategies and comparisons

Partner notification strategy Comparison number, comparison	N (studies)	n (participants)	Outcomes, as reported in any included RCT	Study ID
ENHANCED PATIENT REFERRAL
1. Enhanced patient referral vs. simple patient referral	16	7642	Index patient returning for a test of cure Knowledge of the index patient Number of partners notified and referral of partners for treatment Proportion of index patients with at least 1 partner tested Proportion of index cases with at least 1 sexual partner treated Proportion of index patients with at least 1 partner positive for C. trachomatis Number of partners treated per index patient 6 weeks after randomisation Number of partners elicited Proportion of index cases with a positive chlamydia test result 6 weeks after randomisation Proportion of index cases with all sexual partners treated Acceptability of Internet for use in standard partner notification Partners located Index re‐infection Harms ‐ adverse effects of medication Index patient 72‐hour follow‐up Medication adherence Temporary abstinence from sexual intercourse as evidence of self care Behavioural change Partners contacted Partners tested Partners testing positive Time until testing of partners Number of partners treated per index case Number of partners identified per index Number of traceable partners Number of partners treated within 28 days Proportion of index patients with at least 1 partner treated within 28 days per index case	Andersen 1998 Apoola 2009 Cleveland undated Cameron 2009 Ellison undated Kerani 2011 Katz 1988 Kissinger 2005 Kissinger 2006 Low 2005 Moyo 2002 Ostergaard 2003 Solomon 1988 Tomnay 2006 Trent 2010 Wilson 2009
2. Enhanced patient referral vs. other enhanced patient referral method	2	1336	Partners presenting for care Partners elicited Partners treated	Montesinos 1990 Ellison undated
EXPEDITED PARTNER THERAPY
3. EPT vs. simple patient referral	8	6537	Re‐infection rate of index patient Number of partners notified Partner treatment Sexual outcomes such as having unprotected sex before partner took medication, re‐initiated sex with partner, unprotected sex with any partner Partners elicited Index patient 2‐week post‐treatment return Harms ‐ fighting and refusal of intercourse Side effects of drugs Partner testing	Cameron 2009 Golden 2005 Kerani 2011 Kissinger 2005 Kissinger 2006 Nuwaha 2001 Schillinger 2002 Schwebke 2010
4.1 EPT vs. enhanced patient referral	4	1253	Re‐infection rate of index patient Number of partners notified Partner testing Partner treatment Sexual outcome (unprotected sex, re‐initiated sex with untreated partner)	Cameron 2009 Kerani 2011 Kissinger 2005 Kissinger 2006
4.2 EPT and enhanced patient referral vs. simple patient referral	1	41	Number of partners notified Number of partners treated Method (telephone or in person) of partner notification used Partner tested for HIV/syphilis Adverse events	Kerani 2011
CONTRACT REFERRAL
5 Contract referral vs. simple patient referral	5	2006	Number of partners notified Partners presenting to health service Partners testing positive	Brown 2011 Cleveland undated Landis 1992 Potterat 1977 Schwebke 2010
6. Contract referral vs. enhanced patient referral	1	1266	Partners presenting for care Partners testing positive	Cleveland undated
7. Contract referral vs. EPT	1	324	Re‐infection index patient	Schwebke 2010
8. PROVIDER REFERRAL
8.1 Provider referral vs. simple patient referral	2	596	Partners located Partners treated Partner visit to the clinic during the 30 days after index enrolment Harms Partners testing positive	Brown 2011 Katz 1988
8.2 Choice between provider or simple patient referral vs. simple patient referral	1	396	Partners elicited Number of partners notified Partners treated Harms	Faxelid 1996
9. Provider referral vs. enhanced patient referral	1	461	Partners elicited Partners testing positive Partners treated	Katz 1988
10. Provider referral vs. contract referral	2	2206	Partners tested Partners treated Partner presenting for care Harms Partners testing positive	Brown 2011 Peterman 1997

The outcomes listed are those reported by the authors of the RCTs. Not all were named primary or secondary outcomes in the review.

EPT: expedited partner therapy; HIV: human immunodeficiency virus; RCT: randomised controlled trial.

Enhanced patient referral versus simple patient referral;
Enhanced patient referral versus other enhanced patient referral method;
EPT versus simple patient referral;
EPT versus enhanced patient referral;
EPT and enhanced patient referral versus simple patient referral;
contract referral versus simple patient referral;
contract referral versus enhanced patient referral;
contract referral versus EPT;
provider referral versus simple patient referral;
choice between provider or simple patient referral versus simple patient referral;
provider referral versus enhanced patient referral;
provider referral versus contract referral.

Outcomes

Outcomes assessed are reported in Table 3. The comprehensive details of included studies can be seen in the Characteristics of included studies table.

One study from the original review was classified as a study awaiting assessment because there were no results available (Levy 1998) (Characteristics of studies awaiting classification).

Four ongoing studies were identified from the trial register (Characteristics of ongoing studies).

Excluded studies

We excluded 11 studies (see Characteristics of excluded studies for details).

Risk of bias in included studies

The risk of bias for each study is presented in the 'Risk of bias' table in the section Characteristics of included studies. Figure 1 and Figure 2 illustrate the summary of risk of bias in all the studies.

Allocation

Random sequence generation

Eleven trials reported adequate generation of the random allocation sequence (Apoola 2009; Brown 2011; Cameron 2009; Faxelid 1996; Kissinger 2005; Kissinger 2006; Low 2006b; Nuwaha 2001; Tomnay 2006; Trent 2010; Wilson 2009). Of these trials, eight used blocked randomisation (Apoola 2009; Brown 2011; Cameron 2009; Kissinger 2005; Kissinger 2006; Low 2006b; Tomnay 2006; Wilson 2009), two trials used computer‐generated random numbers tables (Nuwaha 2001; Trent 2010), and, in one study, lots were drawn by index patient (Faxelid 1996). Sequence generation was adequate in six of nine trials reporting the primary outcome of re‐infection with a bacterial STI (Cameron 2009; Kissinger 2005; Kissinger 2006; Low 2006b; Tomnay 2006;Wilson 2009).

In 13 trials, random sequence generation was unclear (Cleveland undated; Ellison undated; Golden 2005; Katz 1988; Kerani 2011; Landis 1992; Montesinos 1990; Moyo 2002; Ostergaard 2003; Peterman 1997; Schillinger 2003; Schwebke 2010; Solomon 1988) and two trials reported methods used that can introduce a high risk of bias (Andersen 1998; Potterat 1977). In Andersen 1998, the date of birth of index patient was used and, in Potterat 1977, assignment of index patient was performed alternately to specific intervention arms. Both of these trials reported secondary outcomes only.

Allocation concealment

Five trials reported adequate allocation concealment (Apoola 2009; Brown 2011; Low 2006b; Schillinger 2003; Tomnay 2006). Of these, four trials reported the use of sealed, opaque, sequentially numbered envelopes (Apoola 2009; Brown 2011; Schillinger 2003; Tomnay 2006), and one trial reported the use of a centralised telephone service (Low 2006b). In 18 trials, the methods used for allocation concealment were not adequately described (Andersen 1998; Cameron 2009; Cleveland undated; Faxelid 1996; Golden 2005; Katz 1988; Kerani 2011; Kissinger 2005; Kissinger 2006; Landis 1992; Moyo 2002; Nuwaha 2001; Ostergaard 2003; Potterat 1977; Schwebke 2010; Solomon 1988; Trent 2010; Wilson 2009). Allocation concealment was adequate in three of nine trials reporting the primary outcome of re‐infection with a bacterial STI.

Three studies reported methods that could introduce a high risk of bias (Ellison undated; Montesinos 1990; Peterman 1997). In Ellison et al., the interventions were allocated in turn to each consecutive patient according to a printed schedule, which could have influenced enrolment or exclusion and hence the intervention received by the index patients (Ellison undated). In Montesinos et al., the protocol used in the intervention was colour coded and the counsellor removed the protocol for the next index patient from a randomly ordered set (Montesinos 1990). Peterman et al. reported that the assignment was known to the interviewer before contact with index patients and sequentially adapted (Peterman 1997).

Blinding

Blinding of participants and personnel (performance bias)

Twenty‐five trials did not have blinding of the participants or the personnel (Andersen 1998; Apoola 2009; Brown 2011; Cameron 2009; Cleveland undated; Ellison undated; Faxelid 1996; Golden 2005; Katz 1988; Kerani 2011; Kissinger 2005; Kissinger 2006; Landis 1992; Low 2006b; Montesinos 1990; Moyo 2002; Nuwaha 2001; Peterman 1997; Potterat 1977; Schillinger 2003; Schwebke 2010; Solomon 1988; Tomnay 2006; Trent 2010; Wilson 2009). In one trial, the index patient received identical specimen collection kits to be given to their partners, and was, therefore, blinded to the intervention in which they were taking part (Ostergaard 2003).

Blinding of outcome assessment (detection bias)

Eleven trials did not report blinding of the outcome assessors (Apoola 2009; Faxelid 1996; Katz 1988; Kerani 2011; Kissinger 2005; Kissinger 2006; Montesinos 1990; Moyo 2002; Nuwaha 2001; Peterman 1997; Potterat 1977). In five trials, the outcome assessors were blinded (Cleveland undated; Ellison undated; Low 2006b; Solomon 1988; Wilson 2009). Cameron et al. reported that the laboratory personnel (primary outcome) were blinded but not the interviewers (Cameron 2009). We judged the risk of bias as low. In six studies, the blinding of outcome assessors was unclear (Andersen 1998; Landis 1992; Ostergaard 2003; Schwebke 2010; Tomnay 2006; Trent 2010). In the remaining three studies, the outcome assessor was not blinded but we judged the risk of bias as low because the primary outcome was objectively assessed (Brown 2011; Golden 2005; Schillinger 2003).

Incomplete outcome data

Seven trials had a high (> 20%) attrition rate (Cameron 2009; Golden 2005; Kerani 2011; Kissinger 2005; Moyo 2002; Schwebke 2010; Trent 2010), including four of nine trials reporting re‐infection with a bacterial STI as an outcome. In Cameron 2009, 65% of index patients submitted at least one urine sample in 12 months, while in Golden 2005, 68% of index patients completed the study. In Kerani 2011, 71% of index patients completed baseline and follow‐up interviews. In Kissinger 2005, 79% of index patients had a follow‐up interview but only 37.5% were retested, and in Moyo 2002, only 50% of index patients had a follow‐up interview. In Schwebke 2010, 40% of index patients completed the study. In Trent 2010, 62% of index patients had a follow‐up interview.

Selective reporting

We compared the trial protocols with published trial results sections to assess reporting bias. If the trial protocol was not available, we compared the methods and results sections of the trial. We searched three trial registries for the protocols of the 16 additional studies included in this update. Protocols were available for five of these studies (Apoola 2009; Kissinger 2006; Low 2006b; Schwebke 2010; Wilson 2009).

We judged 21 trials to have a low risk of reporting bias either because the primary outcome stated in the protocol was reported in the trial result sections (Apoola 2009; Schwebke 2010), or the outcomes stated in the method sections were reported in the result sections (Andersen 1998; Brown 2011; Cameron 2009; Cleveland undated; Ellison undated; Faxelid 1996; Katz 1988; Kerani 2011; Landis 1992; Montesinos 1990; Moyo 2002; Nuwaha 2001; Ostergaard 2003; Peterman 1997; Potterat 1977; Schillinger 2003; Schwebke 2010; Solomon 1988; Tomnay 2006; Trent 2010).

We considered four trials to have an unclear risk of reporting bias because the outcomes reported in the results sections differed from those stated in the method sections (Golden 2005; Kissinger 2005), or protocols (Kissinger 2006; Low 2006b). In Kissinger 2006, the protocol had primary and secondary outcomes whereas in the trial report outcomes were not divided into primary and secondary. Furthermore, additional sexual and behavioural outcomes were reported. Low et al. reported some outcomes in the published paper that differed from the protocol (Low 2006b).

We assessed one trial as being at high risk of reporting bias. In Wilson 2009, the primary outcomes stated in the protocol differed from those stated in trial report; in the protocol there were also three intervention arms described but only two were reported in the trial publication.

Other potential sources of bias

One study had a high potential for other bias (Peterman 1997). The authors of the study reported contamination between the three groups caused by overlap of partners common to index patients. In eight studies, it was unclear if there was any other potential source of bias (Andersen 1998; Cleveland undated; Golden 2005; Kerani 2011; Landis 1992; Nuwaha 2001; Potterat 1977; Solomon 1988). Of these seven studies, in five no comparisons of baseline characteristics between study arms were given (Andersen 1998; Cleveland undated; Landis 1992; Potterat 1977; Solomon 1988). In Golden 2005, selective reporting of subgroups might have introduced bias and in Nuwaha 2001, partners of the patient referral group could have been treated elsewhere leading to misclassification bias. In the remainder of the studies, the risk for potential sources of bias was low.

Effects of interventions

Enhanced patient referral

1. Enhanced patient referral versus simple patient referral

Sixteen studies looked at different types of enhanced patient referral compared with simple patient referral among patients with gonorrhoea (Cleveland undated; Solomon 1988), chlamydia (Andersen 1998; Apoola 2009; Cameron 2009; Low 2006b; Ostergaard 2003), non‐gonococcal urethritis (Katz 1988), gonorrhoea or chlamydia (Kerani 2011; Kissinger 2005; Wilson 2009), trichomoniasis (Kissinger 2006), chlamydia or non‐gonococcal urethritis (Tomnay 2006), PID (Trent 2010), or any STI syndrome (Ellison undated; Moyo 2002).

There were seven different types of enhanced patient referral interventions for patients or partners: 1) an additional counselling session (Cleveland undated; Ellison undated; Moyo 2002; Wilson 2009); 2) a home testing kit for the partners to use and send back to a laboratory (Andersen 1998; Cameron 2009; Ostergaard 2003), or for the partners to bring back to the clinic (Apoola 2009); 3) an additional information booklet to be given to the partner (Kissinger 2005; Kissinger 2006); 4) a videotape shown to the index patient (Solomon 1988; Trent 2010); 5) a disease‐specific website was available to the partner (Kerani 2011; Tomnay 2006); 6) health education messages for the index case (Ellison undated); and 7) health education plus counselling for the index patient (Ellison undated). In addition, two studies compared patient referral performed by a contact tracer (DIS or health adviser) with patient referral performed by a nurse (Katz 1988; Low 2006b).

Primary outcome

Six studies (2007 participants) assessed the index patient re‐infection rate (Cameron 2009; Kissinger 2005; Kissinger 2006; Low 2006b; Tomnay 2006; Wilson 2009) (Figure 4). Owing to substantial heterogeneity (Tau² = 0.38; Chi² = 16.86, df = 5 (P value = 0.005); I² = 70%), the results of individual studies were not pooled. In one comparison, the risk of re‐infection in the index patients was 51% lower in the enhanced patient referral (additional counselling) compared with the simple patient referral group (RR 0.49, 95% CI 0.27 to 0.89) (Wilson 2009). In two smaller studies, the risk of re‐infection was higher in index patients receiving the enhanced patient referral strategy but CIs included the possibility of no difference (Cameron 2009; Tomnay 2006). In the other three studies, there was no statistical evidence of a difference between enhanced and simple patient referral (Table 4).

Figure 4

Forest plot: 1 Enhanced patient referral versus simple patient referral, outcome: 1.1 Re‐infection in index patient, by STI.

Table 4. Enhanced patient referral versus simple patient referral, re‐infection in the index patient, effect size

Comparison	N (studies)	n (participants)	Study ID	RR (95% CI)	Test for heterogeneity I²; Chi², P value
Home sampling kit vs. simple patient referral	1	220	Cameron 2009	2.14 (0.91 to 5.05)	n/a
Information booklet vs. simple patient referral	2	942	Kissinger 2005; Kissinger 2006	0.55 (0.22 to 1.33)	76%; 4.19, P value = 0.04
Patient referral (DIS/health adviser) vs. patient referral (nurse)	1	140	Low 2005	0.35 (0.01 to 8.51)	n/a
Disease‐specific website vs. simple patient referral	1	105	Tomnay 2006	3.12 (0.17 to 58.73)	n/a
Additional counselling vs. simple patient referral	1	600	Wilson 2009	0.49 (0.27 to 0.89)	n/a

Enhanced patient referral is taken as the experimental group. Risk ratio (RR) < 1 indicates a lower re‐infection risk after enhanced patient referral than simple patient referral. If RR = 1, the risk of re‐infection is the same in both groups. If RR > 1, there is a higher risk of re‐infection in the enhanced patient referral group. In the trial by Low et al., the outcome was assessed in a minority of index patients.

CI: confidence interval; DIS: disease intervention specialist; n/a: not applicable; RR: risk ratio.

We judged the quality of evidence for the primary outcome, using the GRADE approach, as low for four of the five enhanced patient referral interventions. We judged additional counselling to provide moderate evidence of a beneficial effect when compared with simple patient referral but there was only one trial in this group (Wilson 2009) (summary of findings Table for the main comparison).

Secondary outcomes

Twelve studies (6045 participants) used five different comparisons and assessed the number of partners elicited (Andersen 1998; Apoola 2009; Cameron 2009; Cleveland undated; Ellison undated; Katz 1988; Kerani 2011; Kissinger 2005; Low 2006b; Moyo 2002; Solomon 1988; Tomnay 2006). There was no evidence of clinically relevant differences between enhanced and simple patient referral strategies (Table 5). When simple patient referral delivered by a nurse was compared with specialist contact tracer (DIS or health adviser) (Katz 1988; Low 2006b), the number of partners elicited was slightly higher in the simple patient referral (nurse) group. We conducted a sensitivity analysis, removing the trial by Andersen 1998 (high risk of bias in random sequence generation), but there was no appreciable difference in the results.

Table 5. Enhanced patient referral versus simple patient referral, number of partners elicited per index patient randomised, effect size

Comparison	N (studies)	n (participants)	Study ID	MD (95% CI)	Test for heterogeneity I²; Chi², P value
Home sampling kit vs. simple patient referral	3	516	Cameron 2009; Andersen 1998; Apoola 2009	0.00 (‐0.19 to 0.19)	0%; 0.32, P value = 0.85
Additional counselling vs. simple patient referral	3	2401	Cleveland undated; Ellison undated; Moyo 2002	0.1 (0.00 to 0.19)	0%; 1.17, P value = 0.56
Patient referral (DIS) vs. patient referral (nurse)	2	597	Katz 1988; Low 2005	‐0.40 (‐0.57 to ‐0.24)	0%; 0.03, P value = 0.87
Information booklet vs. simple patient referral	1	633	Kissinger 2005	0.0 (‐0.22 to 0.22)	n/a
Disease‐specific website vs. simple patient referral	2	140	Kerani 2011; Tomnay 2006	‐0.15 (‐0.72 to 0.42)	13%; 1.15, P value = 0.28

Enhanced patient referral is taken as the experimental group. Mean difference (MD) < 0 indicates that simple patient referral resulted in more partners elicited; MD = 0 indicates no difference between groups; MD > 0 indicates more partners elicited in the enhanced patient referral group.

CI: confidence interval; MD: mean difference; n/a indicates not applicable.

In Ellison et al. there were four intervention arms comparing three different enhanced patient referral methods with simple patient referral: (1) patient referral with a health education message, (2) patient referral with counselling and (3) patient referral with health education message and counselling (Ellison undated). Small increases in the number or partners elicited per index patient were observed in the enhanced patient referral strategy with a health education message (MD 0.25, 95% CI 0.10 to 0.39) and health education message plus counselling (MD 0.6, 95% CI 0.45 to 0.76). In Solomon et al. the authors reported that there was no evidence of differences between enhanced patient referral (videotape) and simple patient referral group for number of partners elicited (Solomon 1988).

Six studies (1885 participants) assessed number of partners notified (Cameron 2009; Moyo 2002; Ostergaard 2003; Tomnay 2006; Trent 2010; Wilson 2009). In Trent 2010 and Wilson 2009, the exact number of partners notified was not reported so we could not calculate the MD. In three studies (Table 6), there was no evidence of a difference in the number of partners notified per index patient between the groups (Cameron 2009; Ostergaard 2003; Tomnay 2006). In Moyo et al. additional counselling resulted in slightly more partners being notified (Moyo 2002).

Table 6. Enhanced patient referral versus simple patient referral, number of partners notified per index patient randomised, effect size

Comparison	N (studies)	n (participants)	Study ID	MD (95% CI)	Test for heterogeneity I²; Chi², P value
Home sampling kit vs. simple patient referral	2	782	Cameron 2009; Ostergaard 2003	0.01 (‐0.12 to 0.14)	0%; 0.01, P value = 0.93
Additional counselling vs. simple patient referral	2	272	Moyo 2002; Wilson 2009	0.21 (0.05 to 0.36) data not available	n/a
Disease‐specific website vs. simple patient referral	1	105	Tomnay 2006	‐0.17 (‐0.68 to 0.35)	n/a
Videotape vs. simple patient referral	1	77	Trent 2010	data not available	n/a

Enhanced patient referral group is taken as the experimental group. Mean difference (MD) < 0 indicates that simple patient referral resulted in more partners notified; MD = 0 indicates no difference between groups; MD > 0 indicates more partners notified in the enhanced patient referral group.

CI: confidence interval; MD: mean difference; n/a indicates not applicable.

Five studies (2684 participants) assessed the number of partners who presented for care (Andersen 1998; Apoola 2009; Cameron 2009; Cleveland undated; Solomon 1988). Data were only available for four studies (Andersen 1998; Apoola 2009; Cameron 2009; Cleveland undated). There was no evidence that one group resulted in more partners who presented for care compared with another (MD 0.1, 95% CI ‐0.08 to 0.28; heterogeneity: Tau² = 0.02; Chi² = 12.59, df = 3 (P value = 0.006); I² = 76%). In Solomon 1988), the authors reported no difference in number of partners presenting for care when a videotape was used.

Five studies (2601 participant) assessed the number of partners who tested positive (Andersen 1998; Cameron 2009; Cleveland undated; Katz 1988; Ostergaard 2003). There was no evidence that there were more partners testing positive in one group than the other (MD 0.04, 95% CI ‐0.01 to 0.09; heterogeneity: Tau² = 0.00; Chi² = 8.49, df = 4 (P value = 0.08); I² = 53%).

Six studies (3275 participants) assessed the number of partners treated (Table 7) (Apoola 2009; Ellison undated; Katz 1988; Kissinger 2005; Low 2006b; Trent 2010). In Trent 2010, the exact number of partners treated was not reported so we could not calculate the MD. The enhanced group receiving the information booklet had slightly more partners treated compared with simple patient referral (Kissinger 2005). The combination of a health education message and counselling also resulted in slightly more partners treated (Ellison undated) (MD 0.08, 95% CI 0.01 to 0.14). There was no evidence of a difference in partners treated with the other enhanced patient referral strategies.

Table 7. Enhanced patient referral versus simple patient referral, number of partners treated per index patient randomised, effect size

Comparison	N (studies)	n (participants)	Study ID	MD (95% CI)	Test for heterogeneity I²; Chi², P value
Home sampling kit vs. simple patient referral	1	200	Apoola 2009	‐0.03 (‐0.25 to 0.19)	n/a
Additional counselling vs. simple patient referral	1	863	Ellison undated	0.04 (‐0.02 to 0.1)	n/a
Patient referral (DIS) vs. patient referral (nurse)	2	597	Katz 1988; Low 2005	‐0.05 (‐0.13 to 0.03)	0%; 0.71, P value = 0.40
Information booklet vs. simple patient referral	1	633	Kissinger 2005	0.22 (0.08 to 0.36)	n/a
Videotape vs. simple patient referral	1	12,677	Trent 2010	not reported	n/a

Enhanced patient referral group is taken as the experimental group. Mean difference (MD) < 0 indicates that simple patient referral resulted in more partners treated; MD = 0 indicates no difference between groups; MD > 0 indicates more partners treated in the enhanced patient referral group.

CI: confidence interval; MD: mean difference; n/a indicates not applicable.

In one study (902 participants), 14.5% of partners in the simple patient referral group and 3.3% in the enhanced group (videotape) attended the clinic eight or more days after the index patient (Solomon 1988).

Five studies (1138 participants) assessed the number of harmful events reported (Kerani 2011; Moyo 2002; Tomnay 2006; Trent 2010; Wilson 2009). In two of these, no harms were reported (Kerani 2011; Tomnay 2006). In Wilson et al., no evidence of a difference of the amount of harm (argument, fight or physical violence) was found in the group receiving the enhancement (additional counselling) compared with simple patient referral group (Wilson 2009). In the fourth trial, complications due to medicine or symptoms worsening were equally distributed between two groups (Trent 2010). The fifth study did not specify the number of harms (physical and verbal abuse) reported but stated that it was not associated with the study arm assignment (Moyo 2002).

No information was available for incidence of STI, changes in behaviour emotional impact and ethical outcomes.

2. Enhanced patient referral versus other enhanced patient referral method

Two studies (1351 participants) compared one enhanced patient referral method with another enhanced patient referral method among patients with any STI syndrome (Ellison undated) and gonorrhoea or non‐gonococcal urethritis (Montesinos 1990).

Secondary outcomes

Both studies assessed the number of partners elicited. In Ellison et al., a health education message plus counselling elicited a slightly higher number of partners compared with counselling alone (MD 0.48, 95% CI 0.32 to 0.64) or to a health education message alone (MD 0.35, 95% CI 0.19 to 0.52) (Ellison undated). There was no difference between the groups receiving health education messages alone compared with the group receiving counselling alone (MD ‐0.12, 95% CI ‐0.27 to 0.03). In Montesinos 1990, there was no evidence of differences in the number of partners elicited when counselling was compared with a combination of counselling plus incentive plus contact cards, and with counselling plus no incentive plus follow‐up call.

One study (65 participants) assessed the number of partners who presented for care (Montesinos 1990), and found no difference between groups when index patients received counselling plus follow‐up call plus no incentive plus contact cards compared with counselling alone or counselling plus incentive plus contact cards.

One study (1286 participants) assessed number of partners treated (Ellison undated). There was no difference between the groups receiving counselling plus health education message compared with health message alone (MD 0.05, 95% CI ‐0.01 to 0.12) or with counselling alone (MD 0.03, 95% CI ‐0.03 to 0.1). No evidence of a difference between groups receiving health message alone compared with counselling alone was found (MD 0.02, 95% CI ‐0.04 to 0.08).

No information was available for index patient re‐infection rate, partners notified, delay in partners presented for care, partners testing positive, incidence of STI, changes in behaviour, emotional impact, harms or ethical outcomes.

Expedited partner therapy

3. Expedited partner therapy versus simple patient referral

Eight studies compared EPT versus simple patient referral among patients with chlamydia (Cameron 2009; Schillinger 2003), trichomoniasis (Kissinger 2006; Schwebke 2010), gonorrhoea or chlamydia (Golden 2005; Kerani 2011; Kissinger 2005) and any STI syndrome (Nuwaha 2001).

Primary outcome

Six studies (6018 participants) assessed the index patient re‐infection rate (Cameron 2009; Golden 2005; Kissinger 2005; Kissinger 2006; Schillinger 2003; Schwebke 2010). Index patients in the EPT group had a 29% lower risk of being re‐infected compared with index patients in simple patient referral group (RR 0.71, 95% CI 0.56 to 0.89; heterogeneity: Tau² = 0.03; Chi² = 8.15, df = 5 (P value = 0.15), I² = 39%) (Figure 5). When a sensitivity analysis was performed and only studies with attrition less than 20% were included (Kissinger 2006; Schillinger 2003), the effect of EPT was attenuated and CIs were wider (RR 0.8, 95% CI 0.62 to 1.04; heterogeneity: Tau² = 0.00; Chi² = 18, df = 1 (P value = 0.67), I² = 0%).

Figure 5

Forest plot: 3 Expedited partner therapy versus simple patient referral, outcome 3.1 Re‐infection in index patients, by STI.

The GRADE quality of the overall evidence for six studies reporting the primary outcome of re‐infection was moderate. We downgraded the quality of the evidence because of the serious risk of bias resulting from attrition and from inadequately described methods in several of the studies. When stratified according to type of STI (two studies each), there was low‐quality evidence suggesting no difference between EPT and simple patient referral for chlamydia, and low‐quality evidence favouring EPT for trichomonas and a combined outcome of either chlamydia or gonorrhoea (summary of findings Table 2).

Secondary outcomes

Six studies (4339 participants) assessed the number of partners elicited (Cameron 2009; Golden 2005; Kerani 2011; Kissinger 2005; Nuwaha 2001; Schwebke 2010). There was no evidence of a difference between the two groups (MD ‐0.02, 95% ‐0.09 to 0.04; heterogeneity: Tau² = 0.00; Chi² = 1.05, df = 5 (P value = 0.96); I² = 0% (Figure 6).

Figure 6

Forest plot: 3 Expedited partner therapy versus simple patient referral, outcome 3.2 Number of partners elicited.

In one small study of men who have sex with men (75 men, Kerani 2011), a slightly higher number of partners was elicited when the index patient received EPT compared with simple patient referral (MD 0.42, 95% CI 0.05 to 0.79).

Three studies (3600 participants) assessed number of partners notified (Cameron 2009; Golden 2005; Kissinger 2005). These three studies showed inconsistent results (heterogeneity: Tau² = 0.07; Chi² = 29.71, df = 2 (P value < 0.001); I² = 93%) (Figure 7). Heterogeneity was explored by setting, STI and gender, and it could not be explained by subgroup analysis. In one study, slightly more partners of index patients in the EPT group were notified (MD 0.45, 95% CI 0.28 to 0.62) (Kissinger 2005). In two studies, there was no significant difference (Cameron 2009: MD 0.13, 95% CI ‐0.06 to 0.32; Golden 2005: MD ‐0.05, 95% CI ‐0.12 to 0.01).

Figure 7

Forest plot: 3 Expedited partner therapy versus simple patient referral, outcome 3.3 Number of partners notified.

One study (220 participants) found no evidence of a difference in the number of partners who presented for care between the groups (MD 0.05, 95% CI ‐0.12 to 0.23) (Cameron 2009).

Four studies (4085 participants) assessed the number of partners treated (Golden 2005; Kissinger 2005; Nuwaha 2001; Schwebke 2010). The studies showed results in the same direction but were very heterogeneous (heterogeneity: Tau² = 0.07; Chi² = 59.57, df = 3 (P value < 0.001); I² = 95%) (Figure 8). Subgroup analysis (setting, STI, gender) did not explain the heterogeneity. In three of the four trials, there was a moderate difference favouring EPT (Kissinger 2005: MD 0.43, 95% CI 0.28 to 0.58; Nuwaha 2001: MD 0.50, 95% CI 0.34 to 0.67; Schwebke 2010: MD 0.51, 95% CI 0.35 to 0.67). The difference between groups was very small in the fourth trial (Golden 2005: MD 0.06, 95% CI 0.01 to 0.12).

Figure 8

Forest plot: 3. Expedited partner therapy versus simple patient referral, outcome 3.5 Number of partners treated.

One of the studies included a measure of harm (Nuwaha 2001). This study (383 participants) found no statistical evidence of a difference in harm between simple patient referral and EPT (MD 0.06, 95% CI 0.0 to 0.12). The index patients in the EPT group reported 23 incidents of quarrelling compared with 11 incidents of quarrelling reported in simple patient referral group. Side effects were reported by index patients in 20 partners in the EPT group and in 10 partners in the simple patient referral group.

No information was available for: partners testing positive, changes in behaviour, emotional impact, ethical outcomes, delay in partners presenting for care or incidence of STI.

4. Expedited partner therapy versus enhanced patient referral

Four studies compared EPT versus enhanced patient referral among patients with gonorrhoea or chlamydia (Kerani 2011; Kissinger 2005), trichomoniasis (Kissinger 2006) or chlamydia (Cameron 2009).

Primary outcome

Three studies (1220) assessed the index patient re‐infection rate (Cameron 2009; Kissinger 2005; Kissinger 2006). There was no evidence of a difference between the two groups (RR 0.96, 95% CI 0.6 to 1.53; heterogeneity: Tau² = 0.06; Chi² = 2.99, df = 2 (P value = 0.22); I² = 33%) (Figure 9). Sensitivity analysis including only studies with attrition less than 20% (Kissinger 2006) also found no evidence of a difference between the two groups (RR 0.73, 95% CI 0.30 to 1.76).

Figure 9

Forest plot: 4 Expedited partner therapy versus enhanced patient referral, outcome: 4.1 Re‐infection in index patients.

The GRADE assessment suggests low‐quality evidence that there was no difference between EPT and enhanced patient referral for preventing re‐infection in patients with curable STI (three studies). The evidence was downgraded because of the risk of bias in the methods and imprecision in the effect estimate (summary of findings Table 3).

Secondary outcomes

Three studies (945 participants) assessed the number of partners elicited (Cameron 2009; Kerani 2011; Kissinger 2005). There was no evidence of a difference between the two groups (MD 0.07, 95% CI ‐0.180 to 0.32; heterogeneity: Tau² = 0.02; Chi² = 3.33, df = 2 (P = 0.19); I² = 40%) (Figure 10).

Figure 10

Forest plot: 4 Expedited partner therapy versus enhanced patient referral: 4.2 Secondary outcomes.

One study (220 participants) measured the number of partners notified and found no evidence of a difference between the groups (MD 0.11, 95% CI ‐0.08 to 0.3) (Cameron 2009).

One study (220 participants) measured the effect on number of partners presenting for care. There was no evidence of a difference between groups (MD 0.01, 95% CI ‐0.02 to 0.03) (Cameron 2009).

One study (692 participants) found a small increase in the number of partners treated per index patient randomised to the EPT group compared with the enhanced patient referral group (MD 0.22, 95% CI 0.21 to 0.23) (Kissinger 2005).

No information was available for delay in partners presenting for care, partners testing positive, changes in behaviour, emotional impact, harms, ethical outcomes and incidence of STI.

One study compared EPT plus enhanced patient referral or simple patient referral among men who have sex with men with chlamydia or gonorrhoea (Kerani 2011). A website, 'inSPOT' was used to enhance the patient referral intervention. The primary outcome assessed was the number of partners treated or notified. In the comparison of EPT and inSPOT (41 participants), a moderately higher number of partners was elicited in the combination group compared with inSPOT alone (MD 1.15, 95% CI 0.22 to 2.08). There was no evidence of differences in the number of partners treated or notified for the comparisons of EPT and inSPOT versus EPT alone (40 participants; MD 0.17, 95% CI ‐0.89 to 1.23); or EPT and inSPOT versus simple patient referral (42 participants, MD 0.58, 95% CI ‐0.4 to 1.57).

No information was available for index patient re‐infection rate, incidence of STI, partners notified, partners presenting for care, number of partners tested, number of partners testing positive, partners treated, delay in partners presented for care, changes in behaviour, emotional impact, harms or ethical outcomes.

Contract referral

5. Contract referral versus simple patient referral

Five trials compared contract referral versus simple patient referral among patients with HIV (Brown 2011; Landis 1992), gonorrhoea (Cleveland undated; Potterat 1977) or trichomoniasis (Schwebke 2010).

Primary outcome

The index patient re‐infection rate was assessed in one trial (322 participants) among women with trichomoniasis (Schwebke 2010). There was no statistical evidence of a difference in the risk of re‐infection in the women receiving contract referral or simple patient referral at either one month (RR 1.65, 95% CI 0.74 to 3.65) or three months (RR 1.65, 95% CI 0.4 to 6.77).

The GRADE level of evidence was very low because the findings were from one small trial with a serious risk of bias in the methods (summary of findings Table 4).

Secondary outcomes

All five studies (2006 participants) assessed the number of partners elicited per index patient. Slightly fewer partners were elicited in the contract referral than the simple patient referral group (MD ‐0.22, 95% CI ‐0.37 to ‐0.06; heterogeneity: Tau² = 0.01; Chi² = 5.27, df = 4 (P value = 0.26); I² = 24%) (Figure 11). We conducted a sensitivity analysis, removing the trial by Potterat et al. (high risk of bias in random sequence generation), but there was no appreciable difference in the results.

Figure 11

Forest plot: 5 Contract referral versus simple patient referral, outcome: 5.1 Number of partners elicited.

One study (74 participants) assessed the number of partners notified per index patient among patients with HIV (Landis 1992). There were more partners notified per index patient in the contract referral group than those that were asked to refer partners themselves (MD 1.71, 95% CI 1.24 to 2.19).

Three studies (1610 participants) assessed the number of partners who presented for care (Brown 2011; Cleveland undated; Potterat 1977). Contract referral resulted in slightly more partners presenting for care (MD 0.25, 95% CI 0.18 to 0.32; heterogeneity: Tau² = 0.00; Chi² = 0.85, df = 2 (P value = 0.65); I² = 0%). We conducted a sensitivity analysis, removing the trial by Potterat 1977 (high risk of bias in random sequence generation), but there was no appreciable difference in the results.

Two studies (481 participants) assessed the time delay between enrolment of the index patient and presentation of the partner for care (Brown 2011; Schwebke 2010). In both studies, authors reported that the partner presented sooner in the simple patient referral than the contract referral group. In one study, the median time between enrolment of the index patient and partner presentation was 3 days (interquartile range (IQR) 2 to 7 days) in the simple patient referral group compared with 7 days (IQR 3 to 11 days) in the contract referral group (Brown 2011), and, in the other trial, the mean time was 5 days in the simple patient referral group and 7.25 days in the contract referral group (P value = 0.19) (Schwebke 2010).

Four studies (1684 participants) assessed the number of partners who tested positive (Brown 2011; Cleveland undated; Landis 1992; Potterat 1977). Contract referral resulted in slightly more partners who tested positive (MD 0.12, 95% CI 0.07 to 0.17; Heterogeneity: Tau² = 0.00; Chi² = 2.73, df = 3 (P value = 0.44); I² = 0%). We conducted a sensitivity analysis, removing the trial by Potterat 1977 (high risk of bias in random sequence generation), but there was no appreciable difference in the results.

Two studies (509 participants) assessed the number of partners treated (Potterat 1977; Schwebke 2010). In one study, slightly more partners of women with trichomoniasis were treated in contract referral than in the simple patient referral group (MD 0.28, 95% CI 0.14 to 0.42) (Schwebke 2010). In the trial of men with gonorrhoea (Potterat 1977), there was no evidence of a difference between groups (MD 0, 95% CI ‐0.04 to 0.03). These two studies were not summarised in a meta‐analysis (heterogeneity: Tau² = 0.04; Chi² = 14.61, df = 1 (P value = 0.0001); I² = 93%). These two studies were both performed in the US but in different gender groups reporting different STI.

One study (159 participants) reported on harms with one event in each group (Brown 2011). One episode of abandonment was reported in the simple patient referral group and the police were contacted to placate the partner of one index patient in the contract referral group.

No information was available for incidence of STI, changes in behaviour and emotional impact.

6. Contract referral versus enhanced patient referral

One study (1266 participants) compared contract referral versus enhanced patient referral (counselling) among patients with gonorrhoea (Cleveland undated).

Secondary outcomes

The number of partners elicited per index patient randomised was moderately lower in contract referral group than the enhanced patient referral (counselling) group (MD ‐0.40, 95% CI ‐0.59 to ‐0.21).

The number of partners who presented for care per index patient (MD 0.25, 95% CI 0.17 to 0.33) and the number of partners who tested positive per index patient (MD 0.11, 95% CI 0.05 to 0.18) were slightly higher in the contract referral group than the enhanced patient referral (counselling) group.

7. Contract referral versus expedited partner therapy

One study (324 participants) compared contract referral with EPT among patients with trichomoniasis (Schwebke 2010).

Primary outcome

There was no statistical evidence of a difference in index patient re‐infection rate at one or three months after treatment comparing EPT with contract referral (one month: RR 0.40, 95% CI 0.16 to 1.01 and three months: RR 2.0, 95% CI 0.7 to 5.72).

Secondary outcomes

There was no statistical evidence of a difference between the two groups in the number of partners elicited per index patient (MD 0.11, 95% CI ‐0.11 to 0.33). The number of partners treated per index patient was slightly higher in the EPT group compared with the contract referral group (MD 0.23, 95% CI 0.05 to 0.41).

No information was available for incidence of STI, changes in behaviour, emotional impact, harms, ethical outcomes, partners notified, partners presented for care, delay in partners presented for care or partners testing positive.

Provider referral

8. Provider referral versus simple patient referral

Two studies compared provider referral versus simple patient referral among patients with HIV (Brown 2011), and non‐gonococcal urethritis (Katz 1988). One study compared a choice between simple patient or provider referral with counselling versus simple patient referral among patients with STI syndromes (Faxelid 1996). None of these studies rinvestigated the primary outcome.

Secondary outcomes

Both studies comparing provider referral with simple patient referral (Brown 2011; Katz 1988) (596 participants) assessed the number of partners elicited per index patient. The results of these two studies showed effects in the opposite direction (heterogeneity: Tau² = 0.14; Chi² = 7.76, df = 1 (P value = 0.005); I² = 87%). Subgroup analysis showed that these two studies included index patients with different STI, different settings and participants, and these studies were reported individually. Among women and men with HIV infection (Brown 2011), there was no evidence of a difference in the number of partners elicited per index patient (MD 0.21, 95% CI ‐0.15 to 0.57). Among men with non‐gonococcal urethritis (Katz 1988), those receiving provider referral reported fewer partners (MD ‐0.36, 95% CI ‐0.55 to ‐0.17).

In one study (158 participants), the time delay in partners presenting for care was measured (Brown 2011). Partners presented sooner for care in the simple patient referral group (median time from index patient enrolment to partners presenting for care 3 days (IQR 2 to 7 days)) compared with the provider referral group (median time 4 days (IQR 2 to 8 days)).

In both trials, there was a small increase in the number of partners testing positive per index patient in the provider group compared with the simple patient referral group (MD 0.06, 95% CI 0.02 to 0.11; heterogeneity: Tau² = 0.00; Chi² = 0.35, df = 1 (P value = 0.55); I² = 0%).

Among men with non‐gonococcal urethritis (438 participants) there was a moderate increase in the number of partners treated per index patient in the provider referral group compared with the simple patient referral group (MD 0.5, 95% CI 0.37 to 0.63) (Katz 1988).

One trial (158 participants) reported harms among patients with HIV infection (Brown 2011). In the provider referral group, no harms were reported and in the simple patient referral group one episode of abandonment was reported.

No information was available for index patient re‐infection rate, partners notified, partners presenting for care, incidence of STI, changes in behaviour, emotional impact or ethical outcomes.

In the study that compareda choice between simple patient or provider referral with counselling versus simple patient referral (Faxelid 1996) (396 participants), there was evidence of a difference between the two groups in the number of partners elicited (MD ‐0.03, 95% CI ‐0.3 to 0.23).

The number of partners notified per index patient (MD 0.41, 95% CI 0.18 to 0.64) and the number of partners who presented for care per index patient (MD 0.46, 95% CI 0.24 to 0.69) were moderately higher in those given a choice than the simple patient referral group. The number of harms reported per male index patient randomised was slightly higher in choice option compared with patient referral option (MD 0.15, 95% CI 0.06 to 0.25). The trial authors did not report individual data but stated that there was no difference between the two groups in the number of harms reported.

No information was available for delay in partners presenting for care, partners testing positive, partners treated, changes in behaviour, emotional impact or ethical outcomes.

9. Provider referral versus enhanced patient referral

One study (461 participants) compared provider referral versus enhanced patient referral (contact tracer (DIS)) among men with non‐gonococcal urethritis (Katz 1988). This study did not investigate the primary outcome.

Secondary outcomes

No evidence of a difference was found in the two groups when comparing the number of partners elicited per index patient (MD ‐0.05, 95% CI ‐0.21 to 0.11). The number of partners who tested positive per index patient (MD ‐0.06, 95% CI ‐0.11 to ‐0.02) and number of partners treated per index patient (MD ‐0.54, 95% CI ‐0.66 to ‐0.42) were slightly lower in the provider referral group than the enhanced patient referral (contact tracer) group.

No information was available for index patient re‐infection rate, partners notified, partners presented for care, delay in partners presented for care, incidence of STI, changes in behaviour, emotional impact, harms or ethical outcomes.

10. Provider referral versus contract referral

Two studies (1491 participants) compared provider referral versus contract referral among patients with HIV (Brown 2011), and syphilis (Peterman 1997). Peterman et al. also compared a strategy of enhanced provider referral (field testing) with contract referral (1224 participants) and with provider referral alone (1380 participants). Neither of these studies investigated the primary outcome.

Secondary outcomes

Both studies that compared provider referral with contract referral (Brown 2011; Peterman 1997) assessed the number of partners elicited. The results were inconsistent (heterogeneity: Tau² = 3.03; Chi² = 127.21, df = 1 (P value < 0.001); I² = 100%). Subgroup analysis (setting, STI, gender) could not explain heterogeneity. Brown et al. (163 participants) found no evidence of a difference between the two groups in patients with HIV infection (MD ‐0.27, 95% CI ‐0.62 to 0.07). The other study (1328 participants) found that the number of partners elicited per index patient with syphilis was higher in the contract referral group than the provider referral group (MD 2.2 95% CI 1.95 to 2.45) (Peterman 1997).

Both studies (1491 participants) assessed the number of partners located. No evidence of a difference between the two groups was found (MD 0.10, 95% CI ‐0.01 to 0.2; heterogeneity: Tau² = 0.00; Chi² = 0.05, df = 1 (P value = 0.82); I² = 0%).

One of these studies (163 participants) compared the number of partners who presented for care and found no evidence of a difference between the two groups (MD 0.03, 95% CI ‐0.19 to 0.25) (Brown 2011).

Peterman et al. (1328 participants) assessed the number of partners tested per index patient and found no evidence of a difference between two groups (MD 0.05, 95% CI ‐0.05 to 0.15) (Peterman 1997).

Both studies (1491 participants) compared the number of partners testing positive (Brown 2011; Peterman 1997). No evidence was found of a difference between two groups (MD 0.02, 95% CI ‐0.03 to 0.06; heterogeneity: Tau² = 0.00; Chi² = 0.07, df = 1 (P value = 0.79); I² = 0%).

Peterman et al. (1328 participants) compared the number of partners treated per index patient and found no evidence of a difference between the two groups (MD 0.06, 95% CI ‐0.03 to 0.15) (Peterman 1997).

One of these studies (163 participants) reported on harms (Brown 2011). The study reported no harms reported in the provider referral arm and one episode of abandonment reported in the contract referral arm (MD 0.01, 95% CI ‐0.01 to 0.04).

One study (163 participants) assessed the time delay in partners presented for care after enrolment of index patient (Brown 2011). The study found that the partners of the index patient in the provider referral arm presented sooner for care (median time between enrolment of index patient and partner presenting 4 days (IQR 2 to 8 days) compared with the contract referral arm (median time between enrolment of index patient and partner presenting for care 7 days (IQR 3 to 11 days).

No information was available for ethical outcomes, index patient re‐infection rate, partners presenting for care, incidence of STI, changes in behaviour or emotional impact.

In Peterman et al., the number of partners elicited per index patient was moderately higher in the enhanced provider (field testing) referral group compared with the contract referral group (MD 0.5, 95% CI 0.21 to 0.79) and much higher than in the group receiving provider referral alone (MD 2.7, 95% CI 2.45 to 2.95) (Peterman 1997).

There was no evidence of a difference between enhanced provider (field testing) referral group compared with contract referral or provider referral alone in the number of partners located per index patient (contract referral: MD ‐0.10, 95% CI ‐0.22 to 0.02; provider referral: MD 0.0, 95% CI ‐0.11 to 0.11), in the number of partners who were tested (contract referral: MD ‐0.06, 95% CI ‐0.17 to 0.05; provider referral: MD ‐0.01, 95% CI ‐0.11 to 0.09), in the number of partners testing positive (contract referral: MD ‐0.02 95% CI ‐0.07 to 0.03; provider referral: MD 0.0; 95% ‐0.05 to 0.04) or in the number of partners receiving treatment (contract referral: MD ‐0.05, 95% CI ‐0.14 to 0.04; provider referral: MD 0.01, 95% CI ‐0.07 to 0.09).

No information was available for incidence of STI, partners notified, index patient re‐infection rate, partners presenting for care, delay in partners presenting for care, changes in behaviour, emotional impact, harms, and ethical outcomes.

Discussion

Summary of main results

Twenty‐six RCTs including 17,578 participants (9015 women and 8563 men) conducted in 10 countries were included in this systematic review.

Summary of evidence according to type of partner notification strategy

EPT for index patients, in trials including those with gonorrhoea, chlamydia, gonorrhoea or chlamydia, or trichomonas (six trials) was better than simple patient referral for the prevention of re‐infection of the index patient. EPT also increased the number of partners treated per index patient (four trials). The re‐infection rate after EPT was similar to that with enhanced patient referral (three trials) but EPT resulted in more partners treated (one trial). When contract referral was compared with EPT (one trial), there was no difference in re‐infection rates among index patients but EPT resulted in more partners being treated. There was insufficient evidence to determine the most effective components of an enhanced patient referral intervention.

We found some evidence that more partners were treated with provider referral (one trial of non‐gonococcal urethritis) compared with simple patient referral. In patients with syphilis (one trial), contract referral elicited more partners than provider referral but the number of partners presenting for care and receiving treatment was the same in the two groups. There was no consistent evidence for the relative effects of provider, contract or patient referral for other STI.

The results of four trials comparing home sampling kits for partners with simple patient referral found no evidence of a reduction in re‐infection rates in index cases or higher numbers of partners elicited, notified or treated. We found no studies evaluating provider training. Only seven trials assessed potential harms; we could not combine the results but there was no evidence of differences in the incidence of adverse effects in any of the individual trials.

Summary of evidence, by infection

There were 11 different categories of STI included in the review. Fifteen studies assessed strategies for PN in individual STI and 11 studies assessed combinations of STI or syndromic diagnoses. There were no RCTs among patients with laboratory‐diagnosed hepatitis B, genital herpes or chancroid.

HIV

Only two studies evaluated PN strategies among patients with HIV (314 participants) (Brown 2011; Landis 1992). Both contract referral and provider referral resulted in more partners presenting for care and testing positive than simple patient referral.

Chlamydia

There was no evidence of a difference in index patient re‐infection rates in two trials that compared EPT with simple patient referral (2007 participants) (Cameron 2009; Schillinger 2003). Four studies compared home sampling kits for partners with simple patient referral (1058 participants) (Andersen 1998; Apoola 2009; Cameron 2009; Ostergaard 2003). One study found no reduction in re‐infection in index patients (Cameron 2009). There was no difference between groups in numbers of partners elicited, notified or treated.

Gonorrhoea

There was no evidence about index patient re‐infection rates. Three studies were performed among patients with gonorrhoea (Cleveland undated; Potterat 1977; Solomon 1988). One study compared simple patient referral versus contract referral (Potterat 1977), another study compared simple patient referral versus enhanced patient referral (videotape) (Solomon 1988), and the third study compared simple patient referral versus contract referral versus enhanced patient referral (additional counselling) (Cleveland undated). Simple patient referral elicited a slightly higher number of partners if compared with contract referral (Cleveland undated; Potterat 1977). The authors of one study using the enhanced patient referral (videotape) did not report results (Solomon 1988).

Chlamydia or gonorrhoea

In trials that included index patients with either chlamydia or gonorrhoea, there was evidence that EPT reduced the index patient re‐infection rate compared with simple patient referral (3380 participants) (Golden 2005; Kissinger 2005). There was also evidence from one trial (600 participants) that index patient re‐infection rates were reduced by patient referral enhanced by additional counselling compared with simple patient referral (Wilson 2009). In one trial among men who had sex with men, more partners were elicited when a combination of EPT and enhanced patient referral (inSPOT website) was used compared with enhanced patient referral (inSPOT website) alone (Kerani 2011).

Trichomonas

There was no statistical evidence that EPT resulted in a lower re‐infection rate in female index patients in comparisons of: EPT versus patient‐booklet enhanced patient referral versus simple patient referral (463 participants) (Kissinger 2006); or EPT versus contract referral versus simple patient referral (484 participants) (Schwebke 2010). Slightly more partners were treated when EPT was used compared with contract referral (Schwebke 2010).

Non‐gonococcal urethritis

One study (678 participants) compared simple patient referral delivered by a nurse versus enhanced patient referral delivered by a DIS versus provider referral (Katz 1988). Provider referral resulted in slightly more partners who tested positive and who received treatment when compared with simple patient referral delivered by a nurse. Simple patient referral by a nurse was superior to enhanced patient referral in the number of partners elicited, but there was no evidence of a difference between groups in number of partners who tested positive.

Non‐gonococcal urethritis or gonorrhoea

There was no evidence about index patient re‐infection rates. One study (65 participants) compared patient referral enhanced by additional counselling alone, counselling with incentives and counselling with a follow‐up telephone call (Montesinos 1990). There was no evidence of superiority of any of the different method assessed in eliciting partners or increasing the number of partners who presented for care.

Non‐gonococcal urethritis or chlamydia

In one study (105 participants), there was no evidence that the use of a website reduced index patient re‐infection rates, or increased the number of partners elicited or notified (Tomnay 2006).

Syphilis

One study (1966 participants) was performed among patients with syphilis (Peterman 1997). This study compared contract referral versus provider referral versus enhanced provider referral (with field testing). Contract referral elicited more partners than provider referral. There was no difference between the numbers of partners who were tested, who tested positive or who received treatment between the contract and provider referral group.

Pelvic inflammatory disease

There was no evidence about index patient re‐infection rates. One study (126 participants) was included on PID, but exact numbers of partners notified were not available from trial authors (Trent 2010).

Any sexually transmitted infections syndrome

Four studies (2770 participants) in developing countries in Africa were performed among patients with a syndromic diagnosis of a STI (Ellison undated; Faxelid 1996; Moyo 2002; Nuwaha 2001). One study (396 participants) found that index patients given a choice between patient and provider referral, compared with simple patient referral resulted in slightly more partners notified and presenting for treatment (Faxelid 1996). One study (383 participants) found that EPT resulted in slightly more partners treated (Nuwaha 2001). In two studies (1991 participants), simple patient referral was compared with enhanced patient referral with additional counselling (Ellison undated; Moyo 2002). In one study (858 participants), a combination of giving additional counselling and health education messages compared with simple patient referral resulted in slightly more partners elicited and treated (Ellison undated).

Overall completeness and applicability of evidence

We identified 16 new trials in the update in addition to the 11 in the first Cochrane review (Mathews 2001). We found studies on the four most common curable STIs: chlamydia (six trials), gonorrhoea (three trials), chlamydia or gonorrhoea (four trials), trichomoniasis (two trials) and syphilis (one trial). We included only two trials among people with HIV and we identified no studies on chancroid, genital herpes or hepatitis B. Only five of the 26 trials were conducted in developing countries. Only one trial included in this review enrolled men who had sex with men who were infected with chlamydia or gonorrhoea (Kerani 2011). One of the trials among people with HIV infection included men who had sex with men (Landis 1992). We added EPT as a new strategy to enhance the effectiveness of patient referral in this update. In addition, we separated patient referral interventions into those that added components such as counselling, written information, websites and specimen testing kits (enhanced patient referral), and those restricted to spoken advice about the need for partners to receive treatment (simple patient referral). We found no studies on provider training. Nine studies reported index patient re‐infection rate, the primary outcome for curable STIs. Few of the studies assessed the proportion of partners who were infected, but both studies of patients with HIV infection reported this outcome. Instead, most studies relied on surrogate outcomes such as partners presenting for medical evaluation, or reports by index patients of partners presenting. Secondary outcomes reported on infrequently or not at all included delays in partners presenting for care, incidence of STIs, changes in behaviour, emotional impact and ethical outcomes.

Quality of the evidence

In every study, there were risks to the validity of the findings and assessment of risk of bias was hampered by incomplete reporting in more than half of the included studies. Sequence generation was adequate in 11 studies while allocation concealment was only adequate in five studies. Inadequate methods of allocation concealment are an important source of potential bias for RCTs of PN interventions, where those enrolling participants might preferentially allocate selected patients to one particular intervention. Blinding of investigators and patients was not feasible for the types of interventions studied and only six studies reported blinding of outcome assessors. Where outcomes can be subjective, for example judging patient‐reported outcomes, unblinded outcome assessment could introduce bias. Re‐infection is an objective biological outcome, so lack of blinding of outcome assessors would be less important. Seven studies reported loss to follow‐up of more than 20%. Most studies had a low risk of selective outcome reporting. In addition, methods and sensitivity of tests used to diagnose STIs varied across studies.

When the body of evidence about PN strategies was considered, there were only four comparisons reporting the primary outcome of re‐infection of index patients with curable STI. EPT compared with simple patient referral was the comparison with the largest number of trials, showing moderate‐quality evidence that EPT reduces re‐infection more than simple patient referral when we pooled results from trials of all curable STIs. We downgraded the quality of evidence because of the risk of bias resulting from attrition and inadequately described methods. There was also low‐quality evidence (limited by the small number of studies and attrition bias), that effect size might differ for different STI. There was also low‐quality evidence from three trials that the effect of EPT was similar to that of enhanced patient referral strategies. Comparisons of enhanced versus simple patient referral were limited to one or two trials for each strategy. There was moderate‐quality evidence that additional counselling reduced re‐infection more than simple patient referral. There was low‐quality evidence from one trial that the effect of contract referral was similar to patient referral.

Potential biases in the review process

We conducted an extensive and comprehensive search strategy with no language restrictions of electronic databases to identify all published and unpublished trials. We contacted experts in the field and searched trial registries to identify ongoing studies. We contacted trial authors, where necessary, to obtain missing data. To minimise bias in the review process, two review authors independently performed all study selection, eligibility assessment, data extraction and assessment of risk of bias. If consensus could not be reached, we consulted a third review author. We used standardised eligibility and data extraction forms.

Agreements and disagreements with other studies or reviews

Our findings were consistent with the findings of the two most recently published systematic reviews (Alam 2010; Trelle 2007). The first found that counselling increased partner referral and was reasonable for developing countries where it was well received by index patient, easily integrated and cost effective (Alam 2010). It also found that EPT resulted in more partners treated compared with simple patient referral alone. Barriers to partner referral were mainly cultural and psychosocial (fear of rejection and abuse). The second review also found that EPT resulted in fewer re‐infections of the index patient and more partners treated than simple patient referral and that the outcomes of EPT were similar to those with enhanced patient referral (Trelle 2007). Consistent with this update, both Trelle 2007 and Alam 2010 reported the inappropriateness of summarising the evidence in a meta‐analysis due to the differences in PN methods used and the way outcomes were reported. Two observational studies reported on adverse effects, 9% of index patients reported physical violence (Kissinger 2003), and 44% reported negative emotional reactions by partners (Rosenthal 1995). In Trelle et al., the authors suggested that labour‐intensive methods, such as provider and contract referral, could be considered for more serious conditions, such as HIV and syphilis, even though evidence for their superiority was inconsistent (Trelle 2007).

Furthermore, Trelle et al. argued for more studies on the use of EPT in chlamydia and gonorrhoea, as well as large RCTs on PN and HIV and syphilis, and that adverse effects need to be reported specifically (Trelle 2007).

Figure 1

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3

Flow diagram detailing the updated search and selection of studies.

Figure 4

Forest plot: 1 Enhanced patient referral versus simple patient referral, outcome: 1.1 Re‐infection in index patient, by STI.

Figure 5

Forest plot: 3 Expedited partner therapy versus simple patient referral, outcome 3.1 Re‐infection in index patients, by STI.

Figure 6

Forest plot: 3 Expedited partner therapy versus simple patient referral, outcome 3.2 Number of partners elicited.

Figure 7

Forest plot: 3 Expedited partner therapy versus simple patient referral, outcome 3.3 Number of partners notified.

Figure 8

Forest plot: 3. Expedited partner therapy versus simple patient referral, outcome 3.5 Number of partners treated.

Figure 9

Forest plot: 4 Expedited partner therapy versus enhanced patient referral, outcome: 4.1 Re‐infection in index patients.

Figure 10

Forest plot: 4 Expedited partner therapy versus enhanced patient referral: 4.2 Secondary outcomes.

Figure 11

Forest plot: 5 Contract referral versus simple patient referral, outcome: 5.1 Number of partners elicited.

Analysis 1.1

Comparison 1 Enhanced patient referral versus simple patient referral, Outcome 1 Re‐infection in index patient.

Analysis 1.2

Comparison 1 Enhanced patient referral versus simple patient referral, Outcome 2 Number of partners elicited.

Analysis 1.3

Comparison 1 Enhanced patient referral versus simple patient referral, Outcome 3 Number of partners notified.

Analysis 1.4

Comparison 1 Enhanced patient referral versus simple patient referral, Outcome 4 Number of partners presenting for care.

Analysis 1.5

Comparison 1 Enhanced patient referral versus simple patient referral, Outcome 5 Number of partners testing positive.

Analysis 1.6

Comparison 1 Enhanced patient referral versus simple patient referral, Outcome 6 Number of partners treated.

Analysis 2.1

Comparison 2 Enhanced patient referral versus other enhanced patient referral, Outcome 1 Partners elicited.

Analysis 2.2

Comparison 2 Enhanced patient referral versus other enhanced patient referral, Outcome 2 Number of partners presenting for care.

Analysis 2.3

Comparison 2 Enhanced patient referral versus other enhanced patient referral, Outcome 3 Number of partners treated.

Analysis 3.1

Comparison 3 Expedited partner therapy (EPT) versus simple patient referral, Outcome 1 Re‐infection in index patients.

Analysis 3.2

Comparison 3 Expedited partner therapy (EPT) versus simple patient referral, Outcome 2 Number of partners elicited.

Analysis 3.3

Comparison 3 Expedited partner therapy (EPT) versus simple patient referral, Outcome 3 Number of partners notified.

Analysis 3.4

Comparison 3 Expedited partner therapy (EPT) versus simple patient referral, Outcome 4 Number of partners presenting for care.

Analysis 3.5

Comparison 3 Expedited partner therapy (EPT) versus simple patient referral, Outcome 5 Number of partners treated.

Analysis 3.6

Comparison 3 Expedited partner therapy (EPT) versus simple patient referral, Outcome 6 Number of harmful events reported.

Analysis 4.1

Comparison 4 Expedited partner therapy (EPT) versus enhanced patient referral, Outcome 1 EPT vs. enhanced patient referral.

Analysis 4.2

Comparison 4 Expedited partner therapy (EPT) versus enhanced patient referral, Outcome 2 EPT vs. enhanced patient referral.

Analysis 4.3

Comparison 4 Expedited partner therapy (EPT) versus enhanced patient referral, Outcome 3 Enhanced patient referral plus EPT vs. simple patient referral.

Analysis 5.1

Comparison 5 Contract referral versus simple patient referral, Outcome 1 Number of partners elicited.

Analysis 5.2

Comparison 5 Contract referral versus simple patient referral, Outcome 2 Number of partners notified.

Analysis 5.3

Comparison 5 Contract referral versus simple patient referral, Outcome 3 Number of partners presenting for care.

Analysis 5.4

Comparison 5 Contract referral versus simple patient referral, Outcome 4 Number of partners testing positive.

Analysis 5.5

Comparison 5 Contract referral versus simple patient referral, Outcome 5 Number of partners treated.

Analysis 5.6

Comparison 5 Contract referral versus simple patient referral, Outcome 6 Number of harmful events reported.

Analysis 6.1

Comparison 6 Contract referral versus enhanced patient referral, Outcome 1 Number of partners elicited.

Analysis 6.2

Comparison 6 Contract referral versus enhanced patient referral, Outcome 2 Partners presenting for care.

Analysis 6.3

Comparison 6 Contract referral versus enhanced patient referral, Outcome 3 Partners testing positive.

Analysis 7.1

Comparison 7 Contract referral versus expedited partner therapy (EPT), Outcome 1 Re‐infection in index patient.

Analysis 8.1

Comparison 8 Provider referral versus simple patient referral, Outcome 1 Provider referral vs. simple patient referral.

Analysis 8.2

Comparison 8 Provider referral versus simple patient referral, Outcome 2 Choice between provider or simple patient referral vs. simple patient referral.

Analysis 9.1

Comparison 9 Provider referral versus enhanced patient referral (disease intervention specialist), Outcome 1 Number of partners elicited.

Analysis 9.2

Comparison 9 Provider referral versus enhanced patient referral (disease intervention specialist), Outcome 2 Number of partners testing positive.

Analysis 9.3

Comparison 9 Provider referral versus enhanced patient referral (disease intervention specialist), Outcome 3 Number of partners treated.

Analysis 10.1

Comparison 10 Provider referral versus contract referral, Outcome 1 Number of partners elicited.

Analysis 10.2

Comparison 10 Provider referral versus contract referral, Outcome 2 Number of partners presenting for care.

Analysis 10.3

Comparison 10 Provider referral versus contract referral, Outcome 3 Number of partners located.

Analysis 10.4

Comparison 10 Provider referral versus contract referral, Outcome 4 Number of partners tested.

Analysis 10.5

Comparison 10 Provider referral versus contract referral, Outcome 5 Partners testing positive.

Analysis 10.6

Comparison 10 Provider referral versus contract referral, Outcome 6 Number of partners treated.

Analysis 10.7

Comparison 10 Provider referral versus contract referral, Outcome 7 Number of harmful events reported.

Summary of findings for the main comparison. Enhanced patient referral compared with simple patient referral for partner notification for STIs, including HIV

Enhanced patient referral compared with simple patient referral for partner notification for STIs, including HIV
Health problem: partner notification for STIs, including HIV Settings: people in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services Intervention: enhanced patient referral Comparison: simple patient referral
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Simple patient referral	Enhanced patient referral
Re‐infection in index patient ‐ home sampling vs. simple patient referral Follow‐up: 12 months	Study population		RR 2.14 (0.91 to 5.05)	220 (1 study)	⊕⊕⊝⊝ low^1,2
	64 per 1000	136 per 1000 (58 to 321)
	Moderate
	64 per 1000	137 per 1000 (58 to 323)
Re‐infection in index patient ‐ information booklet vs. simple patient referral Follow‐up: 8 weeks	Study population		RR 0.55 (0.22 to 1.33)	942 (2 studies)	⊕⊕⊝⊝ low^3,4
	180 per 1000	99 per 1000 (40 to 239)
	Moderate
	156 per 1000	86 per 1000 (34 to 207)
Re‐infection in index patient ‐ patient referral (DIS/health advisor) vs. patient referral (nurse) Follow‐up: 6 weeks	Study population		RR 0.35 (0.01 to 8.51)	140 (1 study)	⊕⊕⊝⊝ low⁵
	14 per 1000	5 per 1000 (0 to 118)
	Moderate
	14 per 1000	5 per 1000 (0 to 119)
Re‐infection in index patient ‐ disease‐specific website vs. simple referral Follow‐up: 1 weeks	Study population		RR 3.12 (0.17 to 58.73)	105 (1 study)	⊕⊕⊝⊝ low⁶
	0 per 1000	0 per 1000 (0 to 0)
	Moderate
	0 per 1000	0 per 1000 (0 to 0)
Re‐infection in index patient ‐ additional counselling vs. simple patient referral Follow‐up: 6 months	Study population		RR 0.49 (0.27 to 0.89)	600 (1 study)	⊕⊕⊕⊝ moderate⁷
	101 per 1000	50 per 1000 (27 to 90)
	Moderate
	101 per 1000	49 per 1000 (27 to 90)
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; DIS: disease intervention specialist; RR: risk ratio; STI: sexually transmitted infection.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ Method of allocation concealment was not reported. 70% completed follow‐up, some were lost to follow‐up and some withdrew from the study, reasons for withdrawal were not reported. Study was not blinded. ² Assuming alpha of 0.05 and beta of 0.2. For relative risk reduction of 20% with best estimate of control event rate of 0.2 approximately 3000 participants were required. The total sample size was 220 and did not meet the optimal information size. ³ High attrition rate and no information given on method of allocation concealment in one of the studies. Different methods were used for outcome assessment ⁴ I² = 76% (P value = 0.06) and minimal overlap of CIs. ⁵ Sample size less than 400, there were very few events and CIs around both relative and absolute estimates include both appreciable benefit and appreciable harm. ⁶ Sample size was very small and optimal information size was not met. There were very few events and CIs overlapped, therefore, no effect both for absolute and relative estimates. ⁷ Risk for selective reporting and unclear method of allocation concealment.

Summary of findings for the main comparison. Enhanced patient referral compared with simple patient referral for partner notification for STIs, including HIV

Summary of findings 2. Expedited partner therapy compared with simple patient referral for partner notification for STIs, including HIV

Expedited partner therapy compared with simple patient referral for partner notification for STIs, including HIV
Health problem: partner notification for STIs, including HIV Settings: people in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services Intervention: expedited partner therapy Comparison: simple patient referral
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Simple patient referral	EPT
Re‐infection in index patients Follow‐up: 2‐12 months	Study population		RR 0.71 (0.56 to 0.89)	6018 (6 studies)	⊕⊕⊕⊝ moderate¹
	110 per 1000	78 per 1000 (62 to 98)
	Moderate
	84 per 1000	60 per 1000 (47 to 75)
Re‐infection in index patients ‐ chlamydia Follow‐up: 3‐12 months	Study population		RR 0.9 (0.6 to 1.35)	2007 (2 studies)	⊕⊕⊕⊝ moderate²
	114 per 1000	102 per 1000 (68 to 154)
	Moderate
	92 per 1000	83 per 1000 (55 to 124)
Re‐infection in index patients ‐ trichomonas	Study population		RR 0.67 (0.34 to 1.28)	631 (2 studies)	⊕⊕⊝⊝ low^3,4
	67 per 1000	45 per 1000 (23 to 85)
	Moderate
	67 per 1000	45 per 1000 (23 to 86)
Re‐infection in index patients ‐ chlamydia or gonorrhoea Follow‐up: 4‐18 weeks	Study population		RR 0.61 (0.39 to 0.94)	3380 (2 studies)	⊕⊕⊝⊝ low^5,6
	116 per 1000	71 per 1000 (45 to 109)
	Moderate
	164 per 1000	100 per 1000 (64 to 154)
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ There was high attrition rate in three of the studies. Methods of sequence generation and allocation concealment not reported in two of the studies. ² CI includes possibility of no effect (i.e. RR of 1.0). ³ Method of sequence generation and allocation concealment not reported in one of the studies. There was high attrition rate in one of the studies. ⁴ Sample size was greater than 400 but CI overlaps, therefore, no effect (i.e. RR of 1.0). ⁵ There were no details on method of sequence generation and allocation concealment. One of the studies had a high attrition rate. ⁶ I² = 74%

Summary of findings 2. Expedited partner therapy compared with simple patient referral for partner notification for STIs, including HIV

Summary of findings 3. Expedited partner therapy compared with enhanced patient referral for partner notification for STIs, including HIV

Expedited partner therapy compared with enhanced patient referral for partner notification for STIs, including HIV
Health problem: partner notification for sexually transmitted infections, including HIV Settings: people in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services Intervention: expedited partner therapy Comparison: enhanced patient referral
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	Enhanced patient referral	EPT
EPT vs. enhanced patient referral ‐ re‐infection in index patients Follow‐up: 1‐12 months	Study population		RR 0.96 (0.6 to 1.53)	1220 (3 studies)	⊕⊕⊝⊝ low^1,2
	92 per 1000	88 per 1000 (55 to 140)
	Moderate
	86 per 1000	83 per 1000 (52 to 132)
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; EPT: expedited partner therapy; RR: risk ratio.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ No details on method of sequence generation in one of the studies. One study had high attrition rate and one study used different methods for outcome assessment. ² Sample size is high but CI includes appreciable benefit and harms with both relative risk reduction and increase being greater than 25%.

Summary of findings 3. Expedited partner therapy compared with enhanced patient referral for partner notification for STIs, including HIV

Summary of findings 4. Contract referral compared with expedited partner therapy for partner notification for STIs, including HIV

Contract referral compared with expedited partner therapy for partner notification for STIs, including HIV
Health problem: partner notification for sexually transmitted infections, including HIV Settings: people in rural and urban areas, given a diagnosis of STI (clinically or by a laboratory) in health services Intervention: contract referral Comparison: expedited partner therapy
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect (95% CI)	No of Participants (studies)	Quality of the evidence (GRADE)	Comments
	Assumed risk	Corresponding risk
	EPT	Contract referral
Re‐infection in index patient Follow‐up: 3 months	Study population		RR 2 (0.7 to 5.72)	322 (1 study)	⊕⊕⊝⊝ low^1,2
	99 per 1000	198 per 1000 (69 to 565)
	Moderate
	99 per 1000	198 per 1000 (69 to 566)
The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: confidence interval; RR: risk ratio.
GRADE Working Group grades of evidence High quality: Further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: We are very uncertain about the estimate.
¹ Method of sequence generation and allocation concealment not reported. The study had high attrition rate. No blinding. ² Imprecision owing to small sample size.

Summary of findings 4. Contract referral compared with expedited partner therapy for partner notification for STIs, including HIV

Table 1. Burden of disease

Disease	DALYs
HIV	58.5 million
Chlamydia trachomatis	3.7 million
Gonorrhoea	3.5 million
Other	280,000
Source: WHO 2004. DALY: disability adjusted life years.

Table 1. Burden of disease

Table 2. Summary of comparisons with data available and STI studied

Partner notification strategy, intervention	Partner notification comparator, comparison number (number of trials)					STI included in trials
Partner notification strategy, intervention	Simple patient referral	Enhanced patient referral	Expedited partner therapy	Contract referral	Other enhanced patient referral	STI included in trials
Enhanced patient referral	1 (16)	‐	‐	‐	2 (2)	Gonorrhoea, chlamydia, non‐gonococcal urethritis, trichomonas, pelvic inflammatory disease, STI syndromes
Expedited partner therapy	3 (8)	4 (5)*	‐	‐	Not applicable	Gonorrhoea, chlamydia, trichomonas, STI syndromes
Contract referral	5 (5)	6 (1)	7 (1)	‐	Not applicable	Gonorrhoea, trichomonas, HIV
Provider referral	8 (3)†	9 (1)	No trials	10 (2)	Not applicable	Non‐gonococcal urethritis, syphilis, HIV
* Comparison includes one trial comparing combinations of expedited partner therapy and patient referral. † Comparison Includes one trial comparing a choice between provider or simple patient referral and simple patient referral. ‐ Indicates combinations of an intervention and comparison that are covered elsewhere in the table; HIV: human immunodeficiency virus; STI: sexually transmitted infection.

Table 2. Summary of comparisons with data available and STI studied

Table 3. Summary of included studies and outcomes reported by authors, according to partner notification strategies and comparisons

Partner notification strategy Comparison number, comparison	N (studies)	n (participants)	Outcomes, as reported in any included RCT	Study ID
ENHANCED PATIENT REFERRAL
1. Enhanced patient referral vs. simple patient referral	16	7642	Index patient returning for a test of cure Knowledge of the index patient Number of partners notified and referral of partners for treatment Proportion of index patients with at least 1 partner tested Proportion of index cases with at least 1 sexual partner treated Proportion of index patients with at least 1 partner positive for C. trachomatis Number of partners treated per index patient 6 weeks after randomisation Number of partners elicited Proportion of index cases with a positive chlamydia test result 6 weeks after randomisation Proportion of index cases with all sexual partners treated Acceptability of Internet for use in standard partner notification Partners located Index re‐infection Harms ‐ adverse effects of medication Index patient 72‐hour follow‐up Medication adherence Temporary abstinence from sexual intercourse as evidence of self care Behavioural change Partners contacted Partners tested Partners testing positive Time until testing of partners Number of partners treated per index case Number of partners identified per index Number of traceable partners Number of partners treated within 28 days Proportion of index patients with at least 1 partner treated within 28 days per index case	Andersen 1998 Apoola 2009 Cleveland undated Cameron 2009 Ellison undated Kerani 2011 Katz 1988 Kissinger 2005 Kissinger 2006 Low 2005 Moyo 2002 Ostergaard 2003 Solomon 1988 Tomnay 2006 Trent 2010 Wilson 2009
2. Enhanced patient referral vs. other enhanced patient referral method	2	1336	Partners presenting for care Partners elicited Partners treated	Montesinos 1990 Ellison undated
EXPEDITED PARTNER THERAPY
3. EPT vs. simple patient referral	8	6537	Re‐infection rate of index patient Number of partners notified Partner treatment Sexual outcomes such as having unprotected sex before partner took medication, re‐initiated sex with partner, unprotected sex with any partner Partners elicited Index patient 2‐week post‐treatment return Harms ‐ fighting and refusal of intercourse Side effects of drugs Partner testing	Cameron 2009 Golden 2005 Kerani 2011 Kissinger 2005 Kissinger 2006 Nuwaha 2001 Schillinger 2002 Schwebke 2010
4.1 EPT vs. enhanced patient referral	4	1253	Re‐infection rate of index patient Number of partners notified Partner testing Partner treatment Sexual outcome (unprotected sex, re‐initiated sex with untreated partner)	Cameron 2009 Kerani 2011 Kissinger 2005 Kissinger 2006
4.2 EPT and enhanced patient referral vs. simple patient referral	1	41	Number of partners notified Number of partners treated Method (telephone or in person) of partner notification used Partner tested for HIV/syphilis Adverse events	Kerani 2011
CONTRACT REFERRAL
5 Contract referral vs. simple patient referral	5	2006	Number of partners notified Partners presenting to health service Partners testing positive	Brown 2011 Cleveland undated Landis 1992 Potterat 1977 Schwebke 2010
6. Contract referral vs. enhanced patient referral	1	1266	Partners presenting for care Partners testing positive	Cleveland undated
7. Contract referral vs. EPT	1	324	Re‐infection index patient	Schwebke 2010
8. PROVIDER REFERRAL
8.1 Provider referral vs. simple patient referral	2	596	Partners located Partners treated Partner visit to the clinic during the 30 days after index enrolment Harms Partners testing positive	Brown 2011 Katz 1988
8.2 Choice between provider or simple patient referral vs. simple patient referral	1	396	Partners elicited Number of partners notified Partners treated Harms	Faxelid 1996
9. Provider referral vs. enhanced patient referral	1	461	Partners elicited Partners testing positive Partners treated	Katz 1988
10. Provider referral vs. contract referral	2	2206	Partners tested Partners treated Partner presenting for care Harms Partners testing positive	Brown 2011 Peterman 1997
The outcomes listed are those reported by the authors of the RCTs. Not all were named primary or secondary outcomes in the review. EPT: expedited partner therapy; HIV: human immunodeficiency virus; RCT: randomised controlled trial.

Table 3. Summary of included studies and outcomes reported by authors, according to partner notification strategies and comparisons

Table 4. Enhanced patient referral versus simple patient referral, re‐infection in the index patient, effect size

Comparison	N (studies)	n (participants)	Study ID	RR (95% CI)	Test for heterogeneity I²; Chi², P value
Home sampling kit vs. simple patient referral	1	220	Cameron 2009	2.14 (0.91 to 5.05)	n/a
Information booklet vs. simple patient referral	2	942	Kissinger 2005; Kissinger 2006	0.55 (0.22 to 1.33)	76%; 4.19, P value = 0.04
Patient referral (DIS/health adviser) vs. patient referral (nurse)	1	140	Low 2005	0.35 (0.01 to 8.51)	n/a
Disease‐specific website vs. simple patient referral	1	105	Tomnay 2006	3.12 (0.17 to 58.73)	n/a
Additional counselling vs. simple patient referral	1	600	Wilson 2009	0.49 (0.27 to 0.89)	n/a
Enhanced patient referral is taken as the experimental group. Risk ratio (RR) < 1 indicates a lower re‐infection risk after enhanced patient referral than simple patient referral. If RR = 1, the risk of re‐infection is the same in both groups. If RR > 1, there is a higher risk of re‐infection in the enhanced patient referral group. In the trial by Low et al., the outcome was assessed in a minority of index patients. CI: confidence interval; DIS: disease intervention specialist; n/a: not applicable; RR: risk ratio.

Table 4. Enhanced patient referral versus simple patient referral, re‐infection in the index patient, effect size

Table 5. Enhanced patient referral versus simple patient referral, number of partners elicited per index patient randomised, effect size

Comparison	N (studies)	n (participants)	Study ID	MD (95% CI)	Test for heterogeneity I²; Chi², P value
Home sampling kit vs. simple patient referral	3	516	Cameron 2009; Andersen 1998; Apoola 2009	0.00 (‐0.19 to 0.19)	0%; 0.32, P value = 0.85
Additional counselling vs. simple patient referral	3	2401	Cleveland undated; Ellison undated; Moyo 2002	0.1 (0.00 to 0.19)	0%; 1.17, P value = 0.56
Patient referral (DIS) vs. patient referral (nurse)	2	597	Katz 1988; Low 2005	‐0.40 (‐0.57 to ‐0.24)	0%; 0.03, P value = 0.87
Information booklet vs. simple patient referral	1	633	Kissinger 2005	0.0 (‐0.22 to 0.22)	n/a
Disease‐specific website vs. simple patient referral	2	140	Kerani 2011; Tomnay 2006	‐0.15 (‐0.72 to 0.42)	13%; 1.15, P value = 0.28
Enhanced patient referral is taken as the experimental group. Mean difference (MD) < 0 indicates that simple patient referral resulted in more partners elicited; MD = 0 indicates no difference between groups; MD > 0 indicates more partners elicited in the enhanced patient referral group. CI: confidence interval; MD: mean difference; n/a indicates not applicable.

Table 5. Enhanced patient referral versus simple patient referral, number of partners elicited per index patient randomised, effect size

Table 6. Enhanced patient referral versus simple patient referral, number of partners notified per index patient randomised, effect size

Comparison	N (studies)	n (participants)	Study ID	MD (95% CI)	Test for heterogeneity I²; Chi², P value
Home sampling kit vs. simple patient referral	2	782	Cameron 2009; Ostergaard 2003	0.01 (‐0.12 to 0.14)	0%; 0.01, P value = 0.93
Additional counselling vs. simple patient referral	2	272	Moyo 2002; Wilson 2009	0.21 (0.05 to 0.36) data not available	n/a
Disease‐specific website vs. simple patient referral	1	105	Tomnay 2006	‐0.17 (‐0.68 to 0.35)	n/a
Videotape vs. simple patient referral	1	77	Trent 2010	data not available	n/a
Enhanced patient referral group is taken as the experimental group. Mean difference (MD) < 0 indicates that simple patient referral resulted in more partners notified; MD = 0 indicates no difference between groups; MD > 0 indicates more partners notified in the enhanced patient referral group. CI: confidence interval; MD: mean difference; n/a indicates not applicable.

Table 6. Enhanced patient referral versus simple patient referral, number of partners notified per index patient randomised, effect size

Table 7. Enhanced patient referral versus simple patient referral, number of partners treated per index patient randomised, effect size

Comparison	N (studies)	n (participants)	Study ID	MD (95% CI)	Test for heterogeneity I²; Chi², P value
Home sampling kit vs. simple patient referral	1	200	Apoola 2009	‐0.03 (‐0.25 to 0.19)	n/a
Additional counselling vs. simple patient referral	1	863	Ellison undated	0.04 (‐0.02 to 0.1)	n/a
Patient referral (DIS) vs. patient referral (nurse)	2	597	Katz 1988; Low 2005	‐0.05 (‐0.13 to 0.03)	0%; 0.71, P value = 0.40
Information booklet vs. simple patient referral	1	633	Kissinger 2005	0.22 (0.08 to 0.36)	n/a
Videotape vs. simple patient referral	1	12,677	Trent 2010	not reported	n/a
Enhanced patient referral group is taken as the experimental group. Mean difference (MD) < 0 indicates that simple patient referral resulted in more partners treated; MD = 0 indicates no difference between groups; MD > 0 indicates more partners treated in the enhanced patient referral group. CI: confidence interval; MD: mean difference; n/a indicates not applicable.

Table 7. Enhanced patient referral versus simple patient referral, number of partners treated per index patient randomised, effect size

Comparison 1. Enhanced patient referral versus simple patient referral

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Re‐infection in index patient Show forest plot	6		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.1 Home sampling vs. simple patient referral	1	220	Risk Ratio (M‐H, Random, 95% CI)	2.14 [0.91, 5.05]
1.2 Information booklet vs. simple patient referral	2	942	Risk Ratio (M‐H, Random, 95% CI)	0.55 [0.22, 1.33]
1.3 Patient referral (DIS/health advisor) vs. patient referral (nurse)	1	140	Risk Ratio (M‐H, Random, 95% CI)	0.35 [0.01, 8.51]
1.4 Disease‐specific website vs. simple referral	1	105	Risk Ratio (M‐H, Random, 95% CI)	3.12 [0.17, 58.73]
1.5 Additional counselling vs. simple patient referral	1	600	Risk Ratio (M‐H, Random, 95% CI)	0.49 [0.27, 0.89]
2 Number of partners elicited Show forest plot	11		Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 Home sampling vs. patient referral	3	516	Mean Difference (IV, Random, 95% CI)	‐0.01 [‐0.19, 0.18]
2.2 Additional counselling vs. patient referral	3	4108	Mean Difference (IV, Random, 95% CI)	0.23 [0.03, 0.43]
2.3 Patient referral (DIS) vs. patient referral (nurse)	2	597	Mean Difference (IV, Random, 95% CI)	‐0.40 [‐0.57, ‐0.24]
2.4 Information booklet vs. patient referral	1	633	Mean Difference (IV, Random, 95% CI)	0.0 [‐0.22, 0.22]
2.5 Disease‐specific website vs. patient referral	2	140	Mean Difference (IV, Random, 95% CI)	‐0.15 [‐0.72, 0.42]
3 Number of partners notified Show forest plot	5	1236	Mean Difference (IV, Random, 95% CI)	0.07 [‐0.06, 0.20]

3.1 Home sampling vs. patient referral	2	782	Mean Difference (IV, Random, 95% CI)	0.01 [‐0.12, 0.14]
3.2 Additional counselling vs. patient referral	1	272	Mean Difference (IV, Random, 95% CI)	0.21 [0.06, 0.36]
3.3 Disease‐specific website vs. patient referral	1	105	Mean Difference (IV, Random, 95% CI)	‐0.17 [‐0.68, 0.34]
3.4 Videotape vs. patient referral	1	77	Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
4 Number of partners presenting for care Show forest plot	4		Mean Difference (IV, Random, 95% CI)	Totals not selected

4.1 Home sampling vs. patient referral	3		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
4.2 Additional counselling vs. patient referral	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
5 Number of partners testing positive Show forest plot	5		Mean Difference (IV, Random, 95% CI)	Subtotals only

5.1 Home sampling vs. patient referral	3	878	Mean Difference (IV, Random, 95% CI)	0.11 [0.05, 0.17]
5.2 Additional counselling vs. patient referral	1	1266	Mean Difference (IV, Random, 95% CI)	0.01 [‐0.05, 0.07]
5.3 Patient referral (DIS) vs. patient referral (nurse)	1	457	Mean Difference (IV, Random, 95% CI)	0.0 [‐0.03, 0.03]
6 Number of partners treated Show forest plot	6		Mean Difference (IV, Random, 95% CI)	Totals not selected

Comparison 1. Enhanced patient referral versus simple patient referral

Comparison 2. Enhanced patient referral versus other enhanced patient referral

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Partners elicited Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Totals not selected

2 Number of partners presenting for care Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

3 Number of partners treated Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

Comparison 2. Enhanced patient referral versus other enhanced patient referral

Comparison 3. Expedited partner therapy (EPT) versus simple patient referral

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Re‐infection in index patients Show forest plot	6	6018	Risk Ratio (M‐H, Random, 95% CI)	0.71 [0.56, 0.89]

1.1 Chlamydia	2	2007	Risk Ratio (M‐H, Random, 95% CI)	0.90 [0.60, 1.35]
1.2 Trichomonas	2	631	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.34, 1.28]
1.3 Chlamydia or gonorrhoea	2	3380	Risk Ratio (M‐H, Random, 95% CI)	0.61 [0.39, 0.94]
2 Number of partners elicited Show forest plot	6	4339	Mean Difference (IV, Random, 95% CI)	‐0.02 [‐0.09, 0.04]

3 Number of partners notified Show forest plot	3		Mean Difference (IV, Random, 95% CI)	Totals not selected

4 Number of partners presenting for care Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

5 Number of partners treated Show forest plot	4		Mean Difference (IV, Random, 95% CI)	Totals not selected

5.1 Chlamydia or gonorrhoea	2		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
5.2 Trichomonas	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
5.3 Any STI syndrome	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
6 Number of harmful events reported Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

Comparison 3. Expedited partner therapy (EPT) versus simple patient referral

Comparison 4. Expedited partner therapy (EPT) versus enhanced patient referral

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 EPT vs. enhanced patient referral Show forest plot	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

1.1 Re‐infection in index patients	3	1220	Risk Ratio (M‐H, Random, 95% CI)	0.96 [0.60, 1.53]
2 EPT vs. enhanced patient referral Show forest plot	3		Mean Difference (IV, Random, 95% CI)	Subtotals only

2.1 Number of partners elicited	3	945	Mean Difference (IV, Random, 95% CI)	0.07 [‐0.18, 0.32]
2.2 Number of partners notified	1	220	Mean Difference (IV, Random, 95% CI)	0.11 [‐0.08, 0.30]
2.3 Number of partners presenting for care	1	220	Mean Difference (IV, Random, 95% CI)	0.01 [‐0.01, 0.03]
2.4 Number of partners treated	1	692	Mean Difference (IV, Random, 95% CI)	0.22 [0.21, 0.23]
3 Enhanced patient referral plus EPT vs. simple patient referral Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

3.1 Number of partners elicited	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]

Comparison 4. Expedited partner therapy (EPT) versus enhanced patient referral

Comparison 5. Contract referral versus simple patient referral

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number of partners elicited Show forest plot	5	2006	Mean Difference (IV, Random, 95% CI)	‐0.22 [‐0.37, ‐0.06]

2 Number of partners notified Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

3 Number of partners presenting for care Show forest plot	3	1610	Mean Difference (IV, Random, 95% CI)	0.25 [0.18, 0.32]

4 Number of partners testing positive Show forest plot	4	1684	Mean Difference (IV, Random, 95% CI)	0.13 [0.07, 0.18]

5 Number of partners treated Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Totals not selected

6 Number of harmful events reported Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

Comparison 5. Contract referral versus simple patient referral

Comparison 6. Contract referral versus enhanced patient referral

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number of partners elicited Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

2 Partners presenting for care Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

3 Partners testing positive Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

Comparison 6. Contract referral versus enhanced patient referral

Comparison 7. Contract referral versus expedited partner therapy (EPT)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Re‐infection in index patient Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

Comparison 7. Contract referral versus expedited partner therapy (EPT)

Comparison 8. Provider referral versus simple patient referral

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Provider referral vs. simple patient referral Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Subtotals only

1.1 Number of partners elicited	2	596	Mean Difference (IV, Random, 95% CI)	‐0.10 [‐0.65, 0.46]
1.2 Number of partners testing positive	2	596	Mean Difference (IV, Random, 95% CI)	0.06 [0.02, 0.11]
1.3 Number of partners treated	1	438	Mean Difference (IV, Random, 95% CI)	0.5 [0.37, 0.63]
1.4 Number of harmful events reported	1	158	Mean Difference (IV, Random, 95% CI)	‐0.01 [‐0.04, 0.02]
2 Choice between provider or simple patient referral vs. simple patient referral Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

2.1 Number of partners elicited	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.2 Number of partners notified	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]
2.3 Number of partners presenting for care	1		Mean Difference (IV, Random, 95% CI)	0.0 [0.0, 0.0]

Comparison 8. Provider referral versus simple patient referral

Comparison 9. Provider referral versus enhanced patient referral (disease intervention specialist)

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number of partners elicited Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

2 Number of partners testing positive Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

3 Number of partners treated Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

Comparison 9. Provider referral versus enhanced patient referral (disease intervention specialist)

Comparison 10. Provider referral versus contract referral

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number of partners elicited Show forest plot	2		Mean Difference (IV, Random, 95% CI)	Totals not selected

2 Number of partners presenting for care Show forest plot	1	163	Mean Difference (IV, Random, 95% CI)	0.03 [‐0.19, 0.25]

3 Number of partners located Show forest plot	2	2129	Mean Difference (IV, Random, 95% CI)	0.10 [‐0.00, 0.20]

4 Number of partners tested Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

5 Partners testing positive Show forest plot	2	2129	Mean Difference (IV, Random, 95% CI)	0.02 [‐0.02, 0.06]

6 Number of partners treated Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

7 Number of harmful events reported Show forest plot	1		Mean Difference (IV, Random, 95% CI)	Totals not selected

Comparison 10. Provider referral versus contract referral