Hum. Reprod. Advance Access originally published online on July 4, 2006
Human Reproduction 2006 21(10):2686-2693; doi:10.1093/humrep/del231
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Measuring quality of life in women with endometriosis: tests of data quality, score reliability, response rate and scaling assumptions of the Endometriosis Health Profile Questionnaire
1 School of Health and Related Research (ScHARR), University of Sheffield, Sheffield 2 Health Services Research Unit, Department of Public Health and Primary Care, University of Oxford, UK 3 School of Psychology and Sociology, Central Queensland University, Rockhampton, Australia 4 School of Medicine and Biomedical Sciences, University of Sheffield, Sheffield and 5 Nuffield Department of Obstetrics and Gynaecology, Women’s Centre, John Radcliffe Hospital, University of Oxford, Oxford, UK
6 To whom correspondence should be addressed at: School of Health and Related Research (ScHARR), University of Sheffield, Regent Court, 30 Regent Street, Sheffield, S1 4DA, UK. E-mail: g.l.jones{at}sheffield.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
BACKGROUND: To test the data quality, scaling assumptions and scoring algorithms underlying the Endometriosis Health Profile-30 (EHP-30) questionnaire: a questionnaire developed to measure the health-related quality of life (HRQoL) of women with endometriosis. METHODS: A cross-sectional postal survey to 727 women with surgically confirmed endometriosis recruited from an existing genetic linkage study (OXEGENE), The National Endometriosis Society (NES), UK and the outpatient gynaecology clinics of the Women’s Centre, John Radcliffe Hospital, Oxford. Tests of data quality included secondary factor analysis, internal reliability consistency, descriptive statistics of the data, missing data levels, floor and ceiling effects and corrected item to total correlation scores. RESULTS: Six hundred and ten women (83.9%) returned the questionnaire. Secondary factor analysis verified the domain structure of the EHP-30. All 11 dimensions were internally reliable with Cronbach’s
scores ranging from 0.80 to 0.96. Missing response rates ranged from 0.2 to 1.3%, and all items were found to be most highly correlated with their own (corrected) scale. CONCLUSIONS: Results confirmed the factor structure, scoring and scaling assumptions of the questionnaire. The high rate of data completeness indicated that the EHP-30 was acceptable and understandable to the respondents, thereby verifying its suitability for measuring the HRQoL of women with endometriosis.
Key words: data quality/endometriosis/quality of life
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Endometriosis is one of the most common chronic gynaecological conditions. Although the prevalence is difficult to determine, it may be as high as 10% (Olive and Schwartz, 1993
). It typically affects women of reproductive age, that is, from the onset of menstruation to the menopause, and it usually regresses after the menopause.
The symptoms typically associated with endometriosis—chronic pelvic pain, painful periods (dysmenorrhoea), pain on defaecation, pain on intercourse (dyspareunia) and subfertility—can have a negative impact upon psycho-social parameters (Low and Edelmann, 1990; Low et al., 1993) and lead to a significant reduction in health-related quality of life (HRQoL) (Burry, 1992; Bodner et al., 1997; Garry et al., 2000; Jones et al., 2002). HRQoL is a multi-dimensional, dynamic concept that encompasses physical, psychological and social aspects associated with a disease or its treatment (Colwell et al., 1998).
Recently, we developed a new, disease-specific questionnaire to measure the HRQoL of women with endometriosis: the Endometriosis Health Profile-30 (EHP-30) questionnaire. We reported on the empirical research undertaken to develop and evaluate the psychometric tests of the EHP-30 (Jones et al., 2001) and the responsiveness of the instrument (Jones et al., 2004). Although the questionnaire demonstrated good reliability, validity and responsiveness, it has become increasingly important to evaluate the quality of the data and the assumptions underlying the scoring and structure of the instrument’s dimensions.
For example, Kline (2000) argued that factor analysis should be replicated in a new and different data set to verify the factor structure and compositions of the dimensions produced from the first analysis. As a test of data quality, such an analysis has been carried out to verify the factor content of the 36-item short-form health survey (SF-36) scales when applied in clinical trials of patients with rheumatoid arthritis and osteoarthritis (Kosinski et al., 1999) and across patient groups with different severities of disease and sociodemographic characteristics (McHorney et al., 1994).
To evaluate the data quality of health-status instruments, most studies have focused on the following psychometric tests: internal consistency reliability, secondary factor analysis and item total correlations. However, other variables have been identified which influence the scoring and structure of a questionnaire’s dimensions including data completeness, that is, levels of missing data and floor and ceiling effects. A high percentage of missing data on an item or scale may indicate problems with a questionnaire (Gandek et al., 1998) or that such an item is confusing, offensive or inappropriate (Oppenheim, 1992). Floor and ceiling effects refer to the extent to which patients score at the extreme ends of the questionnaire, that is, the lowest score (floor) or the highest score (ceiling). If respondents all score at these extreme ends, then the extent of ill health in the sample can be over- or under-represented, respectively (Rowan, 1994). Consequently, it is not possible to report an improvement or deterioration in health state in subsequent assessments (Bindman et al., 1990).
Recently, such tests of data quality on health status instruments have been carried out on the Amyotrophic Lateral Sclerosis Assessment Questionnaire (ALSAQ-40) (Jenkinson et al., 2001) and the SF-36 (Bjorner et al., 1998; Gandek et al., 1998; Kosinski et al., 1999; Jenkinson et al., 2002). In relation to the SF-36, numerous studies have been undertaken to test the data quality of translated versions of the questionnaire, to assess whether the data gained are suitable for use in other countries other than the one in which it was originally developed.
Consequently, this study was conducted to evaluate the data produced from the EHP-30. The aims were to carry out secondary factor analysis, tests of reliability and validity and to determine response rates and completeness of the data in a new postal survey. These results were then compared with the data from the initial postal survey involving women recruited via the National Endometriosis Society (NES) (Jones et al., 2001) to verify the quality of the data and the assumptions underlying the scoring and structure of the questionnaire’s dimensions.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
A postal survey was administered to 727 women who had ever been given a surgical confirmation of endometriosis. All the women who participated were recruited from an existing genetic linkage study (OXEGENE) (Kennedy, 1997
), NES, UK and gynaecology clinics at the Women’s Centre, John Radcliffe Hospital, Oxford. A minimum sample size of 530 was required as there are 53 items in the EHP-30 questionnaire, and it has been postulated that factor analysis ideally requires a 10 : 1 ratio of subjects to items (Kline, 2000).
Statistical methods
The following criteria were used to evaluate the data quality of the EHP-30 questionnaire: (i) secondary factor analysis; (ii) internal reliability consistency; (iii) descriptive statistics of the data including skewness; (iv) data completeness including the levels of scales and individual items missing data; (v) floor and ceiling effects and (vi) corrected item to total correlations.
The EHP-30 is comprised of two parts: a core questionnaire containing 5 scales that is applicable to all women with endometriosis (30 items) and a modular part containing 6 scales which do not necessarily apply to all women with endometriosis (23 items) (Supplementary data available at www.endometriosisguide.com/EHP-30.pdf). As occurred when the EHP-30 was developed, each scale was transformed on a range from 0 (indicating the best health status) through to 100 (indicating the worst health status), so that the extent of ill health could be measured. Each scale was calculated as follows: scale score is equal to the total of the raw scores of each item in the scale divided by the maximum possible raw score of all the items in the scale, multiplied by 100.
Factor analysis is a statistical procedure which enables the underlying dimensions of a questionnaire to be determined (Kline, 2000). Secondary factor analysis was performed (principal component analysis, varimax rotation) to verify the scales produced from the first analysis in the development of the questionnaire. Internal reliability consistency, which is the extent to which a scale taps an underlying dimension, was evaluated using Cronbach’s statistic (Cronbach, 1951). It has been argued that for analysis at the group level a minimum coefficient of 0.70 is acceptable; however, this increases to a reliability coefficient of 0.90 for analysis at the individual level (Nunnally, 1978).
In addition to internal consistency reliability, it is important to evaluate the item total correlation (i.e. the extent to which there is a linear relationship between an item and its scale total, which has been corrected for overlap (Gandek et al., 1998). To correct for overlap, the item to be correlated is omitted from the scale total. It has been suggested that a correlation coefficient of 0.40 is indicative of item total consistency (Ware et al., 1980). In relation to ceiling and floor effects, if a high percentage of the sample score at either extreme of the questionnaire, this would indicate the limited ability of the item/domain to assess change over time.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Of the 727 questionnaires that were administered, 610 were returned giving a response rate of 83.9%. The sample of women differed slightly to that recruited via the NES (Jones et al., 2001
). Firstly, these patients were slightly older. The mean age of the patients was 34.7 years (SD = 8.15 years: range = 17–64 years, n = 600). In comparison, the mean age of the NES sample was 32.5 years (SD = 7.2: range = 17–58 years, n = 325). However, Mann–Whitney U-tests revealed that these differences were significant (z = –4.030, P < 0.001). The mean period since diagnosis was 48.9 months (SD = 64.8: range 0–384 months, n = 597) and 32 months (SD = 44: range = 1–362, n = 323), respectively, for the two groups (z = –4.649, P < 0.001). The percentage of women who were widowed and married were similar in both groups; 0.3% were widowed and 53.3% were married in the new data set, compared with 0.3 and 44.3%, respectively, in the original NES survey.
Secondary factor analysis
Data for the core questionnaire were factor analysed using principal component analysis (varimax rotation) (Table I). Initially, only factors which gained an eigenvalue of 1 or more were retained. This procedure identified four factors, which accounted for 72.5% of the variance. Only those questions which obtained a value 
0.50 on any of the factors were initially retained, and any factors which scored less than this were omitted.
|
From this analysis, three dimensions (emotions, social support and self-image) were identical to the initial factor analysis carried out during development of the EHP-30 with the same items loading on each factor. Although all the items for the pain and control and powerlessness dimensions obtained a value
0.50, both domains loaded on the same factor (Factor 1).
The underlying dimensions of the 23-item modular questionnaire were determined in the same way (Table II). The factor analysis identified five factors, accounting for 84.4% of the variance. Only those factors which obtained a value 0.50 on any of the factors were initially retained.
|
The results were consistent with the original analysis for three of the modular domains (intercourse, infertility and relationship with the medical profession). One item for the treatment scale failed to achieve a value
0.50 (felt frustrated because treatment is not working). However, this did reach 0.664 on the work dimension. Although all the items for the work and relationship with children dimensions obtained a value 0.50, both domains loaded on the same factor (Factor 1).
Internal consistency reliability of scales
Table III summarizes the results of the internal consistency reliability analysis for the core questionnaire compared with the initial NES postal survey. All the scales again exceeded the accepted minimum coefficient of 0.70 which is required for analysis at the group level. However, in the present analysis, four scales in the core questionnaire (pain, control and powerlessness, emotional well-being and social support) achieved an coefficient 0.90 indicating that these scales may be suitable for analysis at an individual level.
|
In addition, the same pattern of internal reliability was found in the present study for the five core scales of the questionnaire as in the NES postal survey. As summarized in Table III, the pain dimension achieved the highest
(NES survey = 0.93), followed by control and powerlessness and emotional well-being (NES = 0.89) and the social support and self-image dimensions (NES = 0.86 and 0.83).
Table III also summarizes the results of the internal consistency reliability analysis for the modular questionnaire. The results from this data set exceeded the accepted level of 0.7 for each scale suggesting that the modular dimensions are good for hypothesis testing at a group level. Four of the six scales meet the minimum coefficient (0.90) for comparison at an individual level.
Score distributions, skewness and floor and ceiling effects
Score distributions, skewness and floor and ceiling effects of the new data set are summarized in Table IV. On the core questionnaire, as with the original NES data (mean scale score = 72.7), the control and powerlessness dimension had the highest mean and therefore most negative impact upon ill health (58.5). As in the original NES data, the scales of infertility, treatment and intercourse produced the same mean pattern regarding the impact of endometriosis upon these areas of well-being (NES postal survey = 67.1, 64.0, 61.3, respectively).
|
For both the core and modular questionnaires, only the self-image scale appeared near normally distributed (–0.1) (Table IV). Overall, most of the remaining scale scores were negatively distributed (towards ill health). This was most evident for the control and powerlessness scale on the core questionnaire (–0.5) and the infertility scale on the modular questionnaire (–0.5). Positive distributions (towards good health) were only observed for scales on the modular questionnaire; relationship with children (1.6) and work (0.7) were most positively distributed, followed by medical profession (0.4).
All the response categories were used for the EHP-30 core and modular questionnaires indicating that a wide range of health status was being measured by the questionnaire. Ceiling and floor effects were evaluated. For the core questionnaire, the largest ceiling effects were seen for the self-image scale (5.1%). The self-image scale also reached the largest floor effect (13.3%). Overall, for the modular scales, slightly higher floor and ceiling effects were found with the highest ceiling effect observed for the infertility scale (17.8%) and the largest floor effect for the relationship with children scale (66.8%).
Response rates and missing data
Response rates for individual questions were analysed to assess whether any item or scale was causing particular problems to the respondents. On the whole, good response rates were achieved for each question of the core questionnaire (Table V).
|
The missing response rates for the core questionnaire items ranged from 0.2 to 1.3%. The pain dimension had the most missing responses; however, all domains of the core questionnaire had excellent response rates. As summarized in Table VI for the modular questionnaire, four items (17.4%) had a 100% response rate. The missing item rates were similar, ranging from 0.2 to 1.1%. The percentage of missing data was most evident in relation to the medical and treatment scales. Although the questions regarding sexual intercourse were personal and sensitive, there was a good response rate to each item (all the items had a >99% response rate).
|
Item total correlation
Item total correlation was evaluated, as summarized in Table VII for the core questionnaire and Table VIII for the modular questionnaire. All the item scale correlations (corrected for overlap) exceeded 0.40 and therefore indicated good item internal consistency for the core questionnaire (range 0.67–0.93). In addition, each item in the core questionnaire correlated more strongly with its parent dimension (corrected for overlap), compared with the other dimensions of the core instrument, thereby verifying the content of the scales produced from the NES postal survey. The same pattern was observed for the modular questionnaire. All the item internal consistency coefficients exceeded 0.40 with scale total correlations ranging from 0.80 to 0.97.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The aim of this study was to test the data quality, scaling assumptions and scoring algorithms underlying the EHP-30 questionnaire in a new and different sample of women with endometriosis. Six criteria were used to evaluate the data quality of the EHP-30 questionnaire: secondary factor analysis, internal reliability consistency, descriptive statistics of the data, data completeness including the levels of scales and individual items missing data, floor and ceiling effects and item total correlation (corrected for overlap).
In general, it appeared that the quality of the data and scaling assumptions of the questionnaire were acceptable. The results produced from the factor analysis on this new sample, compared with the original analysis, verified the structure of the dimensions for the core and modular questionnaires. However, on both the core and modular questionnaires two domains loaded on one factor. For the core questionnaire, it may suggest that the negative cognitions accompanying endometriosis (i.e. feeling a lack of control and powerlessness) are most strongly associated with the pain experienced by the woman. This is perhaps not surprising given that pain has been identified as the cardinal symptom of the condition (Maclaverty and Shaw, 1992).
The high rate of data completeness which was achieved for the items and the total amount of dimension scores which could be calculated indicate that the EHP-30 questionnaire was acceptable to the respondents and understandable.
The psychometric properties of the EHP-30 were also validated. For each scale on the core and modular questionnaires, internal consistency reliability coefficients exceeded 0.70 indicating the appropriateness of the scales for analysis at the group level. Eight of the 11 scales (73%) met the minimum value of 0.90, which suggests that these scales would be suitable for an analysis of patients at an individual level, although further research would be needed to establish the suitability of the EHP-30 in clinical assessments. Item total correlation was also demonstrated as the minimum accepted correlation coefficient of 0.40 was exceeded by a substantial margin for all items on the core and modular scales.
It has been argued that including questions, which are of a sensitive and difficult nature such as those that relate to socially disapproved behaviour, socially taboo topics or highly personal or private matters, may reduce the response rate for a questionnaire or produce incomplete or dishonest answers (Oppenheim, 1992). Evaluating the levels of missing data was therefore particularly relevant to the EHP-30, as one whole module contains questions of a sensitive and personal nature relating to sexual intercourse. Very few missing data were found for this dimension, and for all of the questions, a >99% response rate was achieved. This result was in accordance with other studies which have included questions about sexual intercourse in questionnaires. For example, Stead et al. (1999) reported that the inclusion of a sexual activity questionnaire in gynaecological clinical trials was not seen as intrusive by the participants. Compliance rates >80% were achieved, and there were few missing data indicating the appropriateness of using questionnaires to assess sexual function in women with gynaecological conditions.
A substantially higher response rate was achieved in the present study compared with the original NES survey. One explanation may be that an introductory letter was sent out before the questionnaire was mailed to the subjects. This has been reported to increase the response rate as the questionnaire is seen as less intrusive (Streiner and Norman, 1995; Edwards et al., 2002). Different methods of administration were also used in this postal survey, which may have increased the response rate. A personalized covering letter was used, whereas a standard letter addressed ‘to the occupant’ was used in the NES survey. Also an 87-item questionnaire was administered in the original survey, whereas a 53-item questionnaire was used in this study. However, although some research has shown a negative effect in postal surveys when longer questionnaires are used (Edwards et al., 2002; Dillman, 1978), there is little evidence that the length of a questionnaire is an important variable once someone has decided to complete it (Streiner and Norman, 1995).
A more plausible explanation for the discrepancy in response rates is that those patients in the present study who did not return the questionnaire after 2–3 weeks were sent another questionnaire. In comparison, none of the women in the original study could be recontacted because, to respect the confidentiality of its members, the NES would not provide the names and addresses of women to whom the questionnaire was sent. The importance of following-up non-responders in postal surveys is well documented (Dillman, 1978; de Vaus, 1999; Edwards et al., 2002). Most postal surveys require follow-up mailings, as initial response rates for this method of data collection are too low (Oppenheim, 1992; Edwards et al., 2002). Overall, the results from our study suggest that if appropriate guidelines for administering postal questionnaires are followed, then a good response rate can be obtained with the EHP-30 in postal surveys.
Floor and ceiling effects have been found to be a problem for some existing health status instruments. In particular, most people who complete the Nottingham Health Profile (NHP) questionnaire score 0 for most, and sometimes all, of the six dimensions because it was designed to detect the severe end of illness (Kind and Carr-Hill, 1987). Consequently, it does not reflect the health states for respondents with mild-to-moderate disease. For both the core and modular parts of the EHP-30 questionnaire, all the response categories were used indicating that a broad range of health states, relevant to women with endometriosis, are being measured by the items.
Overall, low ceiling and floor effects for the core questionnaire and scales were found. However, these were higher for the modular scales; floor effects for the relationship with children scale were particularly high, perhaps indicating that this scale may have limitations for reporting an improvement or deterioration in health state in future assessments.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The authors thank Pfizer Inc. for funding this research with an educational grant. Source of funding was Educational grant, Pharmacia Corporation, USA.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Bindman AB, Keane D, Lurie N. (1990) Measuring health changes among severely ill patients: the floor phenomenon. Med Care 28:1142–1152.[Web of Science][Medline]
Bjorner JB, Damsgaard MT, Watt T, Groenvold M. (1998) Tests of data quality, scaling assumptions, and reliability of the Danish SF-36. J Clin Epidemiol 5:1001–1011.
Bodner C, Garratt AM, Ratcliffe J, Macdonald LM, Penney GC. (1997) Measuring health-related quality of life outcomes in women with endometriosis: results of the gynaecology audit project in Scotland. Health Bull (Edinb) 55:109–117.[Medline]
Burry KA. (1992) Nafarelin in the management of endometriosis: quality of life assessment. Am J Obstet Gynecol 166:735–739.[Web of Science][Medline]
Colwell H, Mathias SD, Pasta DJ, Henning JM, Steege JF. (1998) A health-related quality of life instrument for symptomatic patients with endometriosis: a validation study. Am J Obstet Gynecol 179:47–55.[CrossRef][Web of Science][Medline]
Cronbach L. (1951) Coefficent alpha and the internal structure of tests. Psychometrika 16:297–334.[CrossRef][Web of Science]
Dillman DA. (1978) Mail and Telephone Surveys: The Total Design Method(Wiley, New York).
Edwards P, Roberts I, Clarke M, DiGuiseppi C, Pratap S, Reinhard Wentz R, Kwan I. (2002) Increasing response rates to postal questionnaires: systematic review. BMJ 324:1183.
Gandek B, Ware JE, Aaronson NK, Alonso J, Apolone G, Bjorner J, Brazier J, Bullinger M, Fukuhara S, Kaasa S, et al. (1998) Tests of data quality, scaling assumptions, and reliability of the SF-36 in eleven countries: results from the IQOLA Project. International Quality of Life Assessment. J Clin Epidemiol 51:1149–1158.[CrossRef][Web of Science][Medline]
Garry R, Clayton R, Hawe J. (2000) The effect of endometriosis and its radical laparoscopic excision on quality of life indicators. Br J Obstet Gynaecol 107:44–54.
Jenkinson C, Fitzpatrick R, Swash M, Levvy G. (2001) ALSAQ User Manual: Amyotrophic Lateral Sclerosis Assessment Questionnaire(Health Services Research Unit, University of Oxford, Oxford).
Jenkinson C, Hobart J, Tarani C, Fitzpatrick R, Swash M. The ALS-HPS Steering Group. (2002) Use of the short form health survey (SF-36) in patients with amyotrophic lateral sclerosis: tests of data quality, score reliability, response rate and scaling assumptions. J Neurol 249:178–183.[CrossRef][Web of Science][Medline]
Jones G, Kennedy S, Barnard A, Wong J, Jenkinson C. (2001) Development of an endometriosis quality-of-life instrument. The Endometriosis Health Profile-30. Obstet Gynecol 98:258–264.[CrossRef][Web of Science][Medline]
Jones G, Kennedy S, Jenkinson C. (2002) Health-related quality of life measurement in women with common benign gynecological conditions: a systematic review. Am J Obstet Gynecol 187:501–511.[CrossRef][Web of Science][Medline]
Jones GL, Jenkinson C, Kennedy SH. (2004) Evaluating the sensitivity to change of the Endometriosis Health Profile Questionnaire. The EHP-30. Qual Life Res 13:705–713.[CrossRef][Web of Science][Medline]
Kennedy S. (1997) Is there a genetic basis to endometriosis? Semin Reprod Endocrinol 15:309–318.[Medline]
Kind P and Carr-Hill R. (1987) The Nottingham Health Profile: a useful tool for epidemiologists? Soc Sci Med 25:905–910.[CrossRef][Web of Science][Medline]
Kline P. (2000) A Psychometrics Primer(Free Association Books, London).
Kosinski M, Keller SD, Hatoum HT, Kong SX, Ware JE. (1999) The SF-36 Health Survey as a generic outcome measure in clinical trials of patients with osteoarthritis and rheumatoid arthritis: tests of data quality, scaling assumptions and score reliability. Med Care 37:Suppl 5, MS10–22.[Web of Science][Medline]
Low WY and Edelmann RJ. (1990) Psychosocial aspects of endometriosis: a review. J Psychosom Obstet Gynaecol 12:3–12.
Low YL, Edelmann RJ, Sutton C. (1993) A psychological profile of endometriosis patients in comparison to patients with pelvic pain of other origins. J Psychosom Res 37:111–116.[CrossRef][Web of Science][Medline]
Maclaverty CM and Shaw RW. (1992) Pelvic pain and endometriosis. In Shaw RW (Ed.). Endometriosis: Current Understanding and Management(Blackwell Science, Oxford) pp. 112–146.
McHorney CA, Ware JE, Lu JF, Sherbourne CD. (1994) The MOS 36-item short-form health survey (SF-36): III. Tests of data quality, scaling assumptions, and reliability across diverse patient groups. Med Care 32:40–66.[Web of Science][Medline]
Nunnally JC. (1978) Psychometric Theory(Tata McGraw-Hill, New Delhi).
Olive DL and Schwartz LB. (1993) Endometriosis. N Engl J Med 328:1759–1769.
Oppenheim AN. (1992) Questionnaire Design, Interviewing and Attitude Measurement New edition (Continuum Press, London).
Rowan K. (1994) Global questions and scores. In Jenkinson C (Ed.). Measuring Health and Medical Outcomes(UCL Press, London) pp. 54–76.
Stead ML, Crocombe WD, Fallowfield LJ, Selby P, Perren TJ, Garry R, Brown JM. (1999) Sexual activity questionnaires in clinical trials: acceptability to patients with gynaecological disorders. Br J Obstet Gynaecol 106:50–54.[Web of Science][Medline]
Streiner DL and Norman G. (1995) Health Measurement Scales: A Practical Guide to Their Development and Use 2nd edn. (Oxford University Press, Oxford).
de Vaus DA. (1999) Surveys in Social Research(UCL Press, London).
Ware JE, Brook RH, Williams KN, Stewart AL, Davies-Avery A. (1980) Conceptualisation and measurement of health for adults in the Health Insurance Study. Vol. 1: Model of Health and Methodology(RAND Corporation, California).
Submitted on March 31, 2006; resubmitted on May 11, 2006; accepted on May 18, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||