Prevalence and patterns of comorbidities in people with disabilities and their associated socio-demographic factors

Data source and sampling strategy

Data for this study were extracted from the Bangladesh National Household Survey on Persons with Disability (NSPD) 2021. This is a nationally representative cross-sectional survey conducted by the Bangladesh Bureau of Statistics. The survey followed a two-stage stratified random sampling method to select the respondents. In the first stage of sampling, 800 primary sampling units (PSUs) were randomly selected from the list of 293,579 PSUs. The PSU was generated by the Bangladesh Bureau of Statistics as a component of the 2011 Bangladesh National Population Census, derived from an average of 120 households19. In the second stage of sampling, 45 households were selected from each of the earlier selected PSUs through systematic random sample methods. This yields a list of 36,000 households, of them data were collected from 35,493 households. All respondents who are usual residents of the selected households were included in the survey. This covered a total of 14,659 children aged 0–4 years, 39,513 children aged 5–17 years, and 100,853 adults aged 18–95 years. Details of the sampling procedure of the surveys have been published elsewhere16.

Analysed sample

This study specifically targeted individuals aged 2 and older (excluding those under 2, as identifying the need for assistive instruments or detecting disabilities is notably challenging in that age group in Bangladesh). We focused on individuals with various types of disabilities as intended. Out of all participants in the survey, a total of 4270 individuals reported having a disability, and they were included in our analysis.

Outcome variable

Our primary objective was to investigate the occurrence of morbidities (presence of illness or health conditions) among individuals with disabilities (functional limitations or restrictions in performing activities due to impairments). Relevant data were gathered by posing two independent sets of questions. Initially, questions aimed at determining disability status were administered, followed by inquiries concerning morbidity status. To accomplish this, data collectors meticulously assessed the status of all household members, inquiring whether any reported members utilized assistive instruments, such as hearing aids, to lead a normal life. The identified respondents or their actual caregivers (for children under 18 years of age) were then subjected to a series of questions to ascertain whether the reported difficulty qualified as a disability. The Washington Group Questions for child and adult were employed for this purpose. The selected group of respondents or their actual caregivers were also presented with two additional questions to determine the presence of existing morbidities. Firstly, they were asked, “Do you (or the name of children under 18 selected for disability-related questions) have any other health or physical problems besides your disability?” Those responding affirmatively to this initial question were subsequently queried, “Which type of health-related or physical health problems do you have?” A comprehensive list of morbidities, including conditions such as blood pressure, diabetes, asthma, epilepsy, heart problems, physical/movement problems, and other health and physical issues, was provided for respondents to choose from. An option was also given to indicate if the health condition was not listed. Using these responses, we formulated a straightforward classification to discern whether a person with a disability had a morbidity or not and considered as the outcome variable.

Explanatory variables

Several explanatory variables were included. They were selected in three stages. In the first stage, we generated a list of relevant variables through compressively reviewing relevant literature covering LMICs and Bangladesh9,10,20,21,22. The availability of these selected variables was then checked in the survey during the second stage. Finally, the available variables were considered in this study and classified as individual, household, and community-level factors. Individual-level factors include the age of the respondents (children aged 2–17, adults aged 18–34, late adult aged 35–59, older individuals aged 60 or more), gender (male vs. female), respondents’ employment status (agriculture, physical workers, business, service, students, housewives, unable to work, and others), educational attainment (no education, primary, secondary or higher), religion (muslim vs. others), and marital status (married, unmarried, and widow/divorced/separated). Household wealth status (poorest, poorer, middle, richer, richest) was considered as household-level factors. The household wealth status was created by the survey authority using principal component analysis, considering several variables related to household wealth, such as ownership of a radio or television and household roof types. Place of residence (rural vs. urban) and administrative division (Barishal, Chattogram, Dhaka, Khulna, Mymensingh, Rajshahi, Rangpur, Sylhet) were included as community-level factors.

Statistical analysis

Descriptive statistics were used to explore the background characteristics of the respondents. Pearson chi-square test was used to identify the significant differences of the occurrence morbidity among person with disability with explanatory variables at the individual, household and community level. Two level (household, cluster) multilevel logistic regression models were used to explore the associations of morbidities among persons with disability with individual, household, and community level factors. The nested structure of the NSPD data, where individuals are nested within household and households are nested within clusters, necessitated the use of multilevel modelling23. Previous studies have found that multilevel modelling provides comparatively better results for such structured data than simple logistic regression models24. Two separate models were run for children aged 2–17 and adults aged 18–95 years. Each model was adjusted for the considered explanatory variables. Multicollinearity was checked before running each model. Sampling weight were considered in all analyses. Results were recorded as adjusted odds ratios (aOR) along with their 95% confidence intervals (95% CI). All statistical analyses were performed using Stata software version 14.0 (, College Station, Texas, USA). All methods are performed according to the guidelines.

Ethical approval

The survey we analysed conducted by the Bangladesh Bureau of Statistics. Before conducting the survey, they collected the ethical approval from the Bangladesh Medical Research Counsel of the Government of Bangladesh and from their own internal review broad. Informed consent was obtained from all the respondents before conducting interviews. When respondents were unable to provide informed consent, we obtained it from their legally authorized representatives, including husbands of women included in the survey. The shared us the non-identifiable individual’s data for this study based on our interest to conduct this particular research. Since we only get data in non-identifiable form, we do not need any further ethical approval.


Leave a Reply

Your email address will not be published. Required fields are marked *