Validation of the WHO Disability Assessment Schedule (WHODAS 2.0) 12‐item tool against the 36‐item version for measuring functioning and disability associated with pregnancy and history of severe maternal morbidity

Abstract Objective To validate the WHO Disability Assessment Schedule 2.0 (WHODAS 2.0) 12‐item tool against the 36‐item version for measuring functioning and disability associated with pregnancy and the occurrence of maternal morbidity. Methods This is a secondary analysis of the Brazilian retrospective cohort study on long‐term repercussions of severe maternal morbidity (SMM) among women who delivered at a tertiary facility (COMMAG study). We compared WHODAS‐12 and WHODAS‐36 scores of women with and without SMM using measures of central tendency and variability, tests for instruments’ agreement (Bland‐Altman plot), confirmatory factor analysis (CFA), and Cronbach alpha coefficient for internal consistency. Results The COMMAG study enrolled 638 women up to 5 years postpartum. Although the median WHODAS‐36 and ‐12 scores for all women were statistically different (13.04 and 11.76, respectively; P<0.001), there was a strong linear correlation between them. Furthermore, the mean difference and the differences in variance analyses demonstrated agreement of total scores between the two versions. CFA demonstrated how the WHODAS‐12 questions are divided into six previously defined factors and Cronbach alpha showed good internal consistency. Conclusion WHODAS‐12 demonstrated agreement with WHODAS‐36 for total score and was a good instrument for screening functioning and disability among postpartum women, with and without SMM.


| INTRODUCTION
According to the United Nations World Report on Disability, more than 1 billion people in the world live with some form of disability, of which nearly 200 million experience considerable difficulties in functioning.
Globally, people with disabilities have poorer health outcomes, lower educational achievements, less economic participation, and higher rates of poverty than people without disabilities. 1 However, the burden of ill health associated with pregnancy-related and obstetric com- plications is yet to be completely understood because of the broad impact of short-and long-term consequences. 2 Functioning and disability among women of reproductive age is poorly studied. The use of a simple and effective tool to identify and measure disability in the postpartum period is key to improving maternal health worldwide. 2,3 The WHO has made efforts to address the problem of identifying and assessing disability and functioning by establishing an international classification system, the International Classification of Functioning, Disability and Health (ICF). 4 All standard instruments for measuring disability and health needed to be linked conceptually and operationally to the ICF to allow comparisons across different cultures and populations using these new concepts. Using the ICF's conceptualization of disability, WHO developed a new tool-the WHO Disability Assessment Schedule (WHODAS)-to measure difficulties in performing daily activities in a more simplified manner. Like ICF, the tool was designed to "assess the limitations on activity and restrictions on participation experienced by an individual, irrespective of medical diagnosis". 5 The WHODAS tool was refined to include cross-cultural measurement of health status and to respond to calls for improving the scope and cultural adaptability of the original WHODAS. Its second version (WHODAS 2.0) was presented as a general measure of functioning impairment and disability in major life domains. The WHODAS 2.0 instrument intends to measure activity function and participation in daily activities in the 30 days preceding its application. 5,6 The instrument has three versions, two of which were compared in this analysis. The complete 36-question version (WHODAS-36) was administered, and the results of the abbreviated 12-question version (WHODAS-12) were compared with those of the full version simply by extracting and analyzing the relevant subset of 12 questions. The possibility of using a shorter version of the instrument, the WHODAS-12, is appealing when planning population screening surveys; however, WHODAS-12 has neither been tested nor validated among pregnant women. For each domain of the original WHODAS-36, the 12-item version includes two sentinel items with good screening properties that identify over 90% of individuals with mild functioning impairment, based on all 36 items, in general populations. 5 WHODAS 2.0 was translated into, culturally adapted to, and validated for various languages, including Brazilian Portuguese, for a study that implemented it among postpartum women with and without severe maternal morbidity (SMM). 7,8 This retrospective cohort study included 638 women who delivered at a tertiary public hospital in Brazil. Women with SMM showed increased WHODAS-36 scores (functioning impairment) compared with women without SMM. 9 The objectives of the current analysis were to compare and validate the abbreviated WHODAS-12, using the complete WHODAS-36 as the reference, for assessing postpartum disability among women (both with and without maternal morbidity) who delivered up to 5 years before assessment.

| MATERIALS AND METHODS
This is a secondary analysis of the Brazilian retrospective cohort study, known as COMMAG, on the long-term repercussions of SMM on women who delivered at a tertiary maternity unit (between July 2008 and June 2012). 9 The methods have previously been published. 9,10 Briefly, WHODAS 2.0 was applied to a cohort of women with and without the diagnosis of SMM (potentially life-threatening conditions and maternal near-miss incidents), according to WHO standard definition and criteria. 11 Score calculations for the analysis used the WHODAS "item-response-theory" (IRT) based scoring. 5 After obtaining individual informed consent, face-to-face interviews were carried out by healthcare professionals specially trained for the study. All women meeting the SMM criteria who delivered during the study period were invited to participate and a control group (1:1 rate) was also selected. For each woman who experienced SMM ("exposed" group), a woman without SMM ("nonexposed group"), irrespective of other less-severe morbidities, and who delivered the same year and at the same institution, was recruited. barriers and obstacles to such interactions, and other problems, such as maintenance of personal dignity. 6 Response options for every question are: no difficulty, little, moderate, severe, and extreme difficulty.
To study WHODAS-12, the questions common to WHODAS-36 were labelled from S1 to S12 (Table 1) and these comprise two corresponding questions from each WHODAS-36 domain (sentinel key questions). 6 It should be noted that if the woman does not work or study, the number of questions in the versions reduces to 11 and 32, respectively (this impacts the analysis discussion as some questions will have missing values). Both the full and short versions of WHODAS 2.0 generate an overall score ranging from 0 to 100 (0=no disability; 100=full disability).
To perform the comparison between WHODAS-36 and WHODAS-12 scores, we analyzed the overall median, mean, and standard deviation. Differences in medians were tested using the nonparametric Wilcoxon signed rank test because of skewed distributions. In addition to the measures of central tendency, the Pearson correlation coefficient was used to address linear correlation between WHODAS-36 and WHODAS-12 total scores through the different dispersion values. Then, to evaluate the mean and variance of the difference between WHODAS-12 and WHODAS-36 scores, we used the Bland-Altman plot with Pitman test of difference in variance. Both methods were used to evaluate agreement between the short and full versions of WHODAS 2.0. 12,13 These analyses were repeated to compare each domain across the two tools. The score for each domain was proportionally converted to a score also ranging from 0 to 100.
Additionally, we compared the 5th, 10th, 25th, 50th, 75th, 90th, and 95th percentiles (with their respective 95% confidence intervals) of total score for both groups (with and without morbidity), using McNemar test to evaluate statistical difference between WHODAS-12 and WHODAS-36 through the different cut-off points for each percentile. A P value of 0.05 or below was considered statistically significant. Finally, we performed a confirmatory factor analysis (CFA) and Participating women were initially approached and interviewed by phone using the computer assisted telephone interview (CATI) unit, at which time they were invited to the hospital for a visit. During this visit, additional evaluations were performed, including an assessment of the corresponding child.

| RESULTS
The COMMAG study enrolled 638 women, 323 without and 315 with severe maternal morbidity. Their general characteristics are published elsewhere. 9 In total, 631 women completed the entire WHODAS-36 instrument, which in turn contains all the WHODAS-12 questions.
Missing information in any domain limits the calculation of the total scores and therefore seven women were excluded from the analysis.
The results presented compare the 36-item and 12-item scores for all 631 women with complete data; the differences between scores in women with morbidity or without morbidity are explained in further detail elsewhere in this Supplement. 14 In comparing the two instruments, the median of WHODAS-36 and WHODAS-12 total scores as well as those for each domain were assessed. The values for the total scores for the 631 women were 13.04 and 11.76 (P<0.001), respectively (    Figure 5 shows the percentile scores of the total sample, and a breakdown of scores by women's history of SMM using a boxplot. Although there was a statistically significant difference (P=0.003) when comparing the 36-and 12-item scores in the 25th percentile for women with SMM and the 50th percentile for women without SMM, we considered these differences very small and not clinically relevant.
The CFA with six factors showed that both questions from

| DISCUSSION
Our validation study indicated that WHODAS-12 is a good substitute for WHODAS-36. We found a very high correlation between the total scores of WHODAS-12 and WHODAS-36, but relatively poor agreement at the sublevel of specific domains. Reassuringly, the agreement between the versions did not seem to vary significantly according to