Microeconomic institutions and personnel economics for health care delivery: a formal exploration of what matters to health workers in Rwanda

See for example Jaffre and Olivier de Sardan [45] for urban areas in Francophone Africa, McPake et al. [59] on coping strategies, Ferrinho et al. [36] on dual practice, Belli et al. [12] on informal charging and corruption, Franco et al. [39] on motivation and Lindelow and Serneels [54] on performance problems of health workers in general in Ethiopia. Early quantitative evidence on absenteeism (see [22]) has been confirmed and extended, documenting poor performance in other dimensions, by recent work for Kenya, Mozambique, Senegal, Tanzania, Togo and Uganda and under way for other African countries (see http://​www.​sdindicators.​org/​/). These studies also show weaknesses in terms of case load, diagnostic accuracy and adherence to clinical guidelines, as well as challenges with the availability of drugs and quality of infrastructure.

North’s [73] classic definition of institutions refers to rules, incentive structures and organisations and is well suited for the issues that come up in this study. Emerging work on personnel economics of the state underline the promise of this approach (see for instance [38]).

Note that institutions can be seen as building blocks of systems, which focus on the linkages between these building blocks ([43]).

It had less than one fourth physicians per 1000 inhabitants compared to Sub-Sahara Africa as a whole, less than one half nurses and midwives, less that a sixteenth dentists and technicians and about half the pharmacists and technicians, in 2005 [106].

A key strength of group interviews lies in the interaction between group members. Group discussions—in contrast to individual interviews—allow researchers to elicit a multitude of views and contrast these. Group discussions may also function as an ‘information quality filter’: highly individual or extreme points of view will provoke disagreement from other participants. Pragmatically, they permit the collection of wide data in a relatively short time span.

Selection criteria were well respected overall. Some difficulties arose in finding rural female doctors and health workers active in rural private facilities. Equality of participants in each group is important to enable comparison between groups. Since the group with PLWHA will be considered on its own, nine participants were allowed.

Each script focused on a number of issues to which a prompt or trigger question is associated. Five different scripts were developed and can be obtained upon simple request. At the start of the group discussions, participants were informed about the objective of the study, its independent academic character warranted full confidentiality, and were invited to sign a consent form. At the end of the discussion, participants filled out an information sheet and were reimbursed for transport costs.

The service index is a weighted sum of the volume of the main health services produced in the primary health care facilities. This downward trend has to be interpreted with caution as it cannot say anything about quality. A high indicator may coincide with low quality of care.

The study focuses on adult and adolescents primary care delivery in eight rural health centres in southeast Rwanda and observed close to 500 consultations.

Carrying out surprise visits, the studies find that 45% of health providers were absent in Uganda, with the highest absence rate (55% absence) among nurses in rural areas. In Tanzania, 21% was absent. In Uganda, the majority of these were approved absence, and the study concludes that these ‘are thus within management’s power to influence suggesting that the problem of low provider effort is largely a reflection of suboptimal management of human resources.’ An earlier pilot study found time spent counselling patients per clinician per day to be 36 (26) minutes in urban (rural) Tanzania.

Case load is often seen as a potential indicator of the consequences of health worker shortages, which may cause a rise in caseload and compromise service quality.

Focusing on specific conditions, health workers were found to provide a correct diagnosis for 58% of the tracer conditions in Uganda and 57% in Tanzania (for instance, diarrhoea with dehydration and malaria with anaemia). In just over a third of the cases (35%) did they provide the correct clinical guidelines for these conditions in either country. For maternal and neonatal complications, only 19% followed the correct treatment actions in Uganda.

The project’s budget rose from USD 0.8 million in 2004 to USD 8.9 million in 2007 [60].

Rusa et al. [80] evaluating an early pilot version of the intervention—which was strongly adapted before scale-up—argue that improvements in utilization during the pilot were primarily in areas that had not previously been well organised.

While the scheme goes under the name of ‘Community based health insurance’, it does not have the typical characteristics ascribed to these types of schemes. It is strongly managed from above, owned by the state and seems to function more like a parastatal, with compulsory membership through payment of a flat fee. Some have argued that the scheme acts more like a tax.

Two overview studies indicate that the responsiveness of labour supply of nurses to a rise in wages may be limited in high-income countries, around 0.30 [91] or often close to zero [2].

Chaudhury et al. [22] were the first to provide systematic evidence of health worker absenteeism across countries using unannounced visits to primary health clinics in Bangladesh, Ecuador, India, Indonesia, Peru, and Uganda. They found that more than a third of health workers were absent. A study on India reports close to 40% absenteeism rates, ranging from 30% in Madhya Pradesh to over 67% in Bihar. An earlier study in rural Udaipur District in Rajasthan, measures absence more intensively through weekly unannounced visits on random days for one year, and found average absence rate of 36%, and 45% in rural subcenters [5]. Similar figures are found in a more recent survey of health centres in Rajasthan, India (27% in general and 36% among doctors, see [23]). In Sub-Saharan Africa. Bjorkman and Svensson [15] observe an absence rate of 47% in Uganda. Another study finds a similar figure (48%) in Bushenyi District in Uganda [98]. A 2008 study in Zambia, measuring absenteeism on the day of the survey (and not through strict surprise visits), finds 31% of doctors, 20% of clinical officers, and 14% of midwives and other clinical staff to be absent.

The earlier mentioned study in Ethiopia, using contingent valuation, indicated that to get 80% of health workers to take up a rural position, public sector salaries for nurses and doctors need to increase 57 and 83%, respectively, suggesting that small changes in salaries will have a limited effect on the distribution of health workers. One reason seems to be that in most low-income countries, the discrepancy in amenities between rural and urban areas may be too large to be compensated by salaries alone [84]. A discrete choice experiment study focusing on doctors in Ethiopia finds that doubling wages would increase the share of doctors willing to work in rural areas from 7% to over 50%, while for nurses doubling salaries would increase those willing to work in rural areas from 4 to 27% [42]. Analysis for Tanzania confirms similar strong effects of increasing salary and hardship allowance, with women less responsive to these financial incentives [47]. Focusing on Vietnam, another survey [104] finds that increased pay would be the single most powerful incentive to get health workers willing to locate in rural areas. An older study for Indonesia, Chomitz et al. [24] finds that modest cash incentives can make health workers more likely to work in moderately remote areas but that staffing of very remote areas would be prohibitively expensive.

Among benefits, housing appears to be important. The study of doctors in Ethiopia finds that providing high-quality housing would increase the share of doctors willing to work in rural areas to 27% or the equivalent of a wage bonus of 46% [42]. In Tanzania and Malawi, housing is also found to be important. Being able to have a second job is another factor that seems to play an important role [47, 57].

Continued access to professional training are found to be important motivating factors for job choice among future health workers in Ethiopia, Kenya, Malawi, South Africa, Tanzania, Thailand and Vietnam [16, 47, 57, 86, 104].

See for example Dixit [31] and Le Grand [52] for a discussion on these themes for civil servants in general.

For example, on teacher absenteeism [33], on the management and allocation of funds at the school level [79] and on student-learning outcomes [10]. Local monitoring by head teachers as well as communities of beneficiaries have had substantial effects on teacher presence and student-learning outcomes (see [10, 25]).

Recent work confirms possible collusion between workers and monitors. A recent study on environmental auditors in Gujarat shows that random allocation of monitors to plants resulted in more reduced pollution than when monitors were chosen by the plants themselves [32].

In response to these concerns, Rwanda established a national zero-tolerance for corruption in 2006, which announced a number of measures including regular unannounced evaluations, investigation of suspicious reports, on-site verification of data and validation of receipts of services by a random sample of patients who are recorded. It is unclear to what extent these measures have been followed through, and we know of no evaluation. Bucagu et al. [21] hold a plea for further research to develop systems that improve monitoring but keep the consequences of gaming to a minimum

See Bartel et al. [11], Barr and Serneels [9] and Bowles et al. [18] for work on workplace norms outside the health sector.

See for example Deci [30], Krepps [49] and Benabou and Tirole [14]. Intrinsic motivation is typically seen as an individual characteristic that is a long term product of both nature and nurture, and that cannot be changed in the short term.

Which is more likely to arise in some environments—such as NGOs—than in others because these environments are better suited to overcome agency problems see Glaeser, [41, 75].

Burundi (2), DRC (2), Egypt, Philippines (2), Rwanda

See de Bruin et al. [29]; Eijkenaar et al. [34]; Kondo et al. [48]; Mendelson et al. [61].

The poverty of causal evidence on participatory development initiatives is not restricted to the health sector, as documented by Mansuri and Rao [56]

The operational plan sets out a number of focused activities, including a health worker labour market study to take stock of numbers, needs and distribution of health worker and a second study to identify malpractices and unethical behaviour. It plans the development of innovative incentives to attract and retain health personnel, develop an award system for the best performing health professionals and draft a law setting out a career development system for health professionals. Other activities include the plans to develop a certified ethics course, integrating modules of leadership, management and ethics in the training of health professionals [68].