Participants

Participants from National Cheng Kung University Hospital had all attended the Department of Obstetrics & Gynecology and met the inclusion criteria: (1) female, (2) aged 18–45 years, (3) competent in the Chinese language, (4) had a diagnosis of PCOS, and (4) were regularly (at least two outpatient visits before enrollment) followed at outpatient clinics. Exclusion criteria were: (1) diagnosed with a similar clinical presentation, including congenital adrenal hyperplasia, androgen secreting tumors, Cushing syndrome, thyroid dysfunction, and hyperprolactinaemia, (2) previously diagnosed diabetes or Fasting Plasma Glucose Test > 126 mg/dL at the time of inclusion, (3) took any medication having an effect on insulin levels or hormonal medications, including contraceptive pills, at least two months before participating this study, (4) known to have suffered a major traumatic event at least 6 months prior to data collection, such as divorce, separation, or death of someone close.

All participants provided written informed consent regarding their willingness to participate in the research. A well-trained interviewer assessed each participant, and confirmed that all participators completed the Chi-PCOSQ, WHOQOL-BREF, EQ-5D, and demographic questions regarding age, gender, residence, highest education level, disease duration, and disease subtype.

Measurements

PCOSQ [23] is a disease-specific HRQoL questionnaire that contains 26 questions with a seven-point rating scale (1: maximum impairment and 7: no impairment of HRQoL) in the following five domains: emotion (7 items), hair growth (5 items), body weight (5 items), infertility (5 items), and menstruation (4 items). The psychometric properties of the PCOSQ show good test-retest reliability (all intraclass coefficients > 0.8), acceptable internal reliability (all Conbroach’s α values > 0.7), and satisfactory concurrent validity with the SF-36 [21]. The responsiveness of the PCOSQ was also established in a randomized controlled trial of troglitazone [6]. The PCOSQ was responsive to treatment effects. HRQoL was found to be improved for the domains that corresponded to the treated symptoms (e.g., infertility, emotion, and menstruation), and no improved HRQoL was found for the domains without treatment effects (e.g., weight and hirsutism).

WHOQOL-BREF [31] is a short-form version of the WHOQOL-100 and has 26 items (the WHOQOL-BREF Taiwan version additionally includes two domestic items) distributed into two generic items and four domains: physical health (7 items), psychological health (6 items), social relations (4 items), and environment (11 items). The internal consistency (Cronbach’s α = 0.70–0.91), the test-retest reliability (r = 0.76–0.80), and the construct validity (comparative fit index [CFI] = 0.89) have been established for the Taiwan version scores [32].

5-dimension EuroQoL questionnaire (EQ-5D) is a multi-attribute utility instrument with five dimensions (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression), each of which has three levels of severity (health states). The Taiwan version was validated in a previous study [33]. Using the scoring function from Taiwan, these health state parameters were transformed into a utility value ranging from 0 to 1, where 0 represents death and 1 indicates full health.

Translating PCOSQ into Chinese and statistical analyses

Permission for the study from the Institutional Review Board of the National Cheng Kung University Hospital was obtained before study commencement (A-ER-103-287). The User Agreement and Translation Agreement of the English version of the PCOSQ was obtained from the original authors [23]. The translation process was guided by an expert panel, which included two coordinators (one researcher and one physician), one psychometric expert, four translators, and two interviewers.

Comparison of HRQoL of women with PCOS between Chinese-speaking and other countries

As compared to other ethnic origins, the clinical manifestations of PCOS on HRQoL in Iranian women in general are lesser extent (Fig 1). East Asian patients, as presented by our Chinese participants, struggled more with infertility and acne. However, the impact of hirsutism on the HRQoL of Chinese-speaking patients was trivial. This finding is consistent with previous studies on South Chinese, who were less hirsute and had lower Ferriman-Gallwey (F-G) scores [35]; fewer than 10% of Chinese patients had a modified F-G score of more than 5 [36]. Although previous studies had examined the impact of acne problems on PCOS patients’ HRQoL in the US [5] and Brazil [24], our results indicate that acne problems for Chinese patients may be more severe. Hence, the present study highlights the importance of managing emotional disturbances due to clinical manifestations of PCOSQ, especially infertility and acne problems, on the HRQoL of East Asians.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Comparisons across ethnicities in HRQoL of PCOS women measured using PCOSQ.

https://doi.org/10.1371/journal.pone.0137772.g001

Psychometric properties

Our rigorous translation procedure and psychometric evaluation ensured that the Chi-PCOSQ is scientific and convincing. All aspects of assessment indicate that the Chi-PCOSQ has good measurement properties, and the results of psychometric properties agree with previous validation studies of the PCOSQ [6, 21]. Although our proposed 6-factor solution structure for the Chi-PCOSQ is slightly different from the 5-factor solution structure in most previous studies [6, 21], we are confident that our structure is suitable for East Asians based on the following reasons. First, the factor loadings of individual items on the five dimensions of emotional disturbance, hirsutism, weight concerns, infertility, and menstrual problems were identical to those in previous findings [6, 21, 23], expect for one item, “tired easily”, which was loaded on the acne and hair loss domain. As for the five added items, “afraid of getting diabetes” was loaded on the weight domain, while the other four were loaded on the acne and hair loss domain. Second, our EFA results suggest that the factor loadings of five additional items (those related to acne, hair loss, and afraid of getting diabetes) on the corresponding domains are satisfactory. Although “tired easily” was loaded on the acne and hair loss domain, rather than on the emotion domain, as in the original scale, this loading is acceptable. Third, the internal-consistency of the Chi-PCOSQ was excellent in all six domains.

All the coefficients of the Chi-PCOSQ domains from test-retest reliability analysis were greater than 0.7, implying that the scale had satisfactory reproducibility. However, as to individual items, the coefficients of test-retest reliability values for “Q20: late menstrual period” and “Q8: irregular menstrual periods” were relatively low. This may be because most of the respondents had begun taking metformin after the baseline interview and most reported an improved menstrual cycle when they were back for their second interview (when the re-test was administrated). As a result, test-retest analysis for menstrual-related questions showed some inconsistency between the two tests.

In construct validity analysis, the Chi-PCOSQ was correlated with generic HRQoL measures (WHOQOL-BREF, EQ-5D) and clinical parameters (e.g., weight and blood pressure). Also, the Chi-PSOCQ had the ability to differentiate between patients according to their HRQoL at a point in time. In the known-group validity analysis, the Chi-PCOSQ provided support to all hypotheses depending upon known performances of the construct that the scale is intended to measure. This is, the respondents who had sexual experiences were more worried about infertility problems (as shown by lower scores for the infertility domain). Those experiencing the problem of acne or hair loss reported significantly lower scores for the acne and hair loss domain of the Chi-PCOSQ. Family history is important in the baseline evaluation of Asian women with PCOS [37], which should be linked to their long-term metabolic management. Our results show that a patient with a family history of diabetes rated her menstruation more problematic.

Modifications of original PCOSQ

Another important contribution of this study is the enhancement of the face validity of the PCOSQ in its application in Chinese-speaking countries by incorporating five additional items into the original scale, presented as “Q14_2: frightened of getting diabetes,” “Q29: growth of visible acne,” “Q27: excess hair loss,” “Q30: feel acne is a problem,” and “Q28: excess hair loss is a problem”. The factor loadings of these new items were expected; Q27, Q28, Q29, and Q30 were loaded on the acne and hair loss domain and Q14_2 was loaded on the weight domain.

PCOS increases the risks of metabolic syndrome, insulin resistance, and diabetes, especially in Chinese patients [38], with elevated chronic inflammation markers such as interleckin-6 [39]. Asian women with PCOS at reproduction age are at risk of developing impaired glucose tolerance and type 2 diabetes [30]. Also, PCOS in Asian women is a significant risk factor of gestational diabetes [40]. In this regards, PCOS patients’ psychological disturbances due to the fear of developing diabetes may need to be concerned. It may be important to incorporate this issue into the study of HRQoL of Asian women affected by PCOS. However, the original PCOSQ only addresses PCOS patients’ concern about getting cancer, while their worries about the risk of getting diabetes is not assessed. In fact, our pilot participants rated the issue of developing diabetes more important than that of getting cancer and, overall, our study patients were a little more worried about having diabetes than cancer. Hence, it is important to understand PCOS patients’ emotional impairment due to their fear of getting diabetes. In this study, “afraid of getting diabetes” was loaded on the weight domain. PCOS is considered to be associated with weight, as many women with PCOS are overweight or obese, which makes them prone to getting diabetes [41]. As a result, PCOS patients, especially those with obesity, might be afraid of getting diabetes and present their underlying concerns or worry about their weight. In this regard, “afraid of getting diabetes” is considered acceptable to belong to the weight domain, although this item also represents an emotional disturbance.

Unlike the suggestions from previous studies that “tired easily” be included in the emotion domain, our results show that “tired easily” was loaded on the acne and hair loss domain. Because “tired easily” intuitively represents a type of emotional tiredness, it seems appropriate to be included in the emotion domain. However, previous studies showed that the factor loadings of this item on the emotion domain were borderline acceptable. For example, the loadings were 0.47 in its original development study [23], 0.39 in the validation study by Guyatt et al. [6], and 0.345 in another validation study by Jones et al. [21]. We further proposed possible reasons why “tired easily” was loaded on the acne and hair loss domain in this study. First, acne and hair loss (i.e., male-pattern hair loss) are important clinical presentations of PCOS, which are likely to provoke anxiety or make individuals feel self-conscious about their physical appearance [5]. In this regards, “tired easily” would represent patients’ being exhausted due to underlying acne or hair loss problems and is considered acceptable to be included in the acne and hair loss domain in the present study. Second, because most items in the emotion domain mention the term “PCOS”, while “tired easily” does not, there might be a wording effect in the emotion domain [42]. That is, the respondents may directly link their emotional disturbances to PCOS for the emotion domain, while they might not consider that “tired easily” is related to PCOS. As a result, “tired easily” performed better in the acne and hair loss domain than in the emotion domain.

Also, the lack of questions about acne may weaken the validity of the PCOSQ because acne is a common symptom of PCOS and leads to psychological disturbances in patients [21]. Hence, four acne items were incorporated into the mPCOSQ by Barnard et al. [5], which accounted for 10.38% of variance in their study. A difference between the mPCOSQ and the Chi-PCOSQ is the number of items related to acne problems. The mPCOSQ [5] has four items representing patients’ difficulties with acne problems: (i) To what extent was acne a problem for you in the last 2 weeks? (ii) In relation to your last menstruation, how much was acne a problem for you? (iii) How much time during the last 2 weeks did you feel unattractive because of acne? (iv) How much time during the last 2 weeks did you feel depressed as a result of acne? The first two items ask the extent of acne problems with different recall periods (e.g., in the last 2 weeks vs. the last menstruation) and the last two items focus on patients’ emotional disturbances (e.g., unattractive, depression) due to acne problem. Although the Chi-PCOSQ only contains two items associated with acne, it is believed that these two items are sufficient to understand acne-associated disturbances. This low number of items lowers the burden of participants. In addition, the acne-related items use “in the past 2 weeks” as the recall period, the same as that used for all remaining items in the scale.

Moreover, hair loss (androgenic alopecia, in a classic male pattern) as characterized by hyperandrogenism is another issue to be considered. Female-pattern baldness or hair loss is often associated with the development of PCOS and can have a significant psychological impact on a person, such as low self-esteem, depression, introversion, and feelings of unattractiveness. However, little research has assessed how such clinical presentation influences the HRQoL of women with PCOS. In fact, some of our pilot participants reported their concerns regarding hair loss and on average our study respondents felt moderate impairment due to this problem. This indicates the importance of incorporating the assessment of hair-loss-related items into the scale.

Furthermore, it has been recognized that racial and ethnic backgrounds may influence the impact of PCOS on HRQoL [25–28]. It is clear that symptoms commonly associated with PCOS, including acne, hirsutism, irregular menses, amenorrhea, obesity, and subfertility, are major sources of psychological and behavioral disorder, and that these clinical presentations vary in ethnic groups, which may result in different impact on HRQoL. One study showed that South Asian women with PCOS seemed not to have identifiably poorer HRQoL than did Caucasians with PCOS; the burden of hirsutism on HRQoL in South Asians was especially to a lesser extent than that in Caucasians, with weight and infertility as the worst domains on the condition-specific questionnaire for both ethnic groups [26]. Although women affected by PCOS commonly report excess hair growth (hirsutism), PCOS patients in China [35, 36] had low rates of hirsutism. Consistent with previous findings [35, 36], our Chinese respondents rated their concerns about the excess growth of body hair to a lesser extent, especially the aspect of “growth of visible hair on chin”.

Although the scoring approach in the Chi-PCOSQ is consistent with that in previous studies [6, 43] using the original PCOSQ, the difference in total number of items in the Emotional and Weight domains between the Chi-PCOSQ and the original PCOSQ may result in different domain and total scores between the two scales. Despite this difference, we found that the domain scores of Emotional and Weight domains in the present study were within the range of those in previous studies using original PCOSQ (Fig 1). An added Acne domain score seems to be higher in an Iranian study [24] which used the mPCOSQ [5], than that domain score of the Chi-PCOSQ in the present study (Fig 1). However, this difference may be attributable to ethnicity instead of the scoring method.

Potential limitations

Several limitations of this study need to be addressed. First, the sample size was limited and all participants were from one medical center. Most of the participants had a high educational level and might have had good health literacy in terms of their self-care and/or health-seeking behaviors. It is crucial for future researchers to validate the Chi-PCOSQ using a large sample from different regions with various education and socioeconomic statuses. Although our sample size was based on a > 2:1 ratio of subjects to items [21, 34], others suggested that the ratio should be ≥ 5:1 [44]. Future studies containing such a ratio of respondents to items are recommended. Second, the translators for back translation were not native English speakers, which may affect the language fluency of the back-translated English versions of the PCOSQ. However, the back translation may be somewhat acceptable for the following reasons: (1) the concept of the back-translated English versions was consistent with the original scale; (2) the back translators had studied or worked in the United States for over 5 years, and (3) they had received high scores on the Test of English as a Foreign Language. Third, because our work was only a preliminary study, we did not have any priori anticipation for the magnitude of the correlations among measures. However, priori predictions are important for examining construct validity [6]. Therefore, we suggest that future studies need to have such priori anticipations before testing construct validity of the Chi-PCOSQ. Fourth, some of the results relating to factor analysis have produced item-domain linkages that some might find counter-intuitive (i.e., Q4: tiring easily loaded on the Acne&Hair loss domain in the present study). Future research is warranted to examine and verify our study results. Lastly, the responsiveness of the Chi-PCOSQ was not assessed in this study due to the limited study time period. Future research, along with sufficient follow-up time, should evaluate the ability of the Chi-PCOSQ to describe the health status of patients over time and its sensitivity to detect changes in HRQoL because of treatment or intervention, and determine the minimal clinical importance (MID) score to justify whether these changes are clinically relevant.