This survey was designed to obtain national estimates on the burden of MSK conditions through a household level survey. Adults aged 18 years or more comprised the study population. [29].
Sample size and sampling:
Assuming a point prevalence of MSK conditions among Bangladeshi adults of 24% [23], at 5% precision level, 280 participants were required in each reporting domain. Considering four reporting domains (rural-urban, male-female), a design effect of 1.5, and an 85% response rate, the calculated sample size for this survey was 1,978. This was finally rounded to 2,000.
The primary sampling units (PSUs) in Bangladesh constitute the sampling frame of national or subnational surveys. We used the PSUs of 2001 Census stratified in to the then seven divisions and rural and urban areas [30]. Mauza and Mahalla in rural and urban areas, respectively, were the PSUs with known boundaries. Maps with list of households of these PSUs were updated by the Bangladesh Bureau of Statistics. Total and urban rural population of the division were considered for allocating number of PSUs. Finally, 20 PSUs (8 urban and 12 rural) were selected and first consecutive100 households were included from each PSU. Households having even and odd numbers were assigned as male and female households to recruit one man and one woman, respectively, using the Kish table [31].
Field team and its training:
We employed seven field teams for seven divisions of Bangladesh. Each team consisted of one research physician (having at least one-year residency in rheumatology), one field organizer and two interviewers. The field team underwent a three-day training in Bangabandhu Sheikh Mujib Medical University before the pretest. All investigators and WHO technical team coordinated and conducted the training using a manual especially prepared for this survey. All investigators were present at the training sessions to ensure uniform understanding of procedures. After completion of the pretest, all investigators and the field had a one-day debriefing session for revising the manual and adjustment of the data collection tool. Another one-day refreshers training was done after completing one PSU by each team to minimize differences among teams.
Survey instrument and data collection:
The survey instrument was the modified COPCORD questionnaire [32]. The first part of the questionnaire aimed at detecting the respondents with musculoskeletal pain with some elaboration of the complaints. This portion was completed by the interviewers. The second part of the questionnaire had structured information for recording subjects’ history and clinical examination findings according to the COPCORD examination sheet. This was used by the research physicians for the diagnosis of conditions and detection of disability. The English version of the first part (that has been administered by the interviewers) of the questionnaire was translated to Bangla, then adapted and validated as per standard procedure [33].
Field work:
Data were collected in each PSU over a period of six days with engagement of the local community and health authority. The field organizer visited in advance and started household listing with the help of local health assistant on the first day. The field interviewers collected data (by reading out questions loudly to all participants), identified screening positive respondents, took physical measurements, and arranged interview with the research physician next five days. Two recall visits were done if the selected house was locked, selected person was not at home at the time of interviewer’s visit. They were declared non-respondents in case interview could not be done at the second recall. The research physician interviewed and examined the positive respondents for making a diagnosis. In doubtful cases, opinion of a division level investigator was taken. Investigators made at least one visit to PSUs in their respective divisions for validation of diagnosis. Erythrocyte sedimentation rate, C-reactive protein, rheumatoid factor and anti-citrullinated peptide antibody were tested in a pre-selected laboratory located nearby to aid the diagnosis. X-rays were also done as and when necessary.
Operational definitions:
Covariates:
The following variables were assessed as covariates for analysis: area of residence, sex, age, education, occupation, wealth index, body mass index (BMI). Education was categorized into four groups: no education, any primary education (completed grades 1-5), any secondary education (completed grades 6-10), and above secondary education (completed ≥ grade 11). Participants’ occupation was categorized into seven groups: home makers, laborers, business, salaried services, rickshaw/auto-rickshaw/van pullers, cultivators and others.
The wealth index was constructed using principal component analysis [29]. Asset information collected covered information on household ownership of 20 items, such as flush toilet, telephone, television, bicycle, sewing machine, bed. Each asset was assigned a weight (factor score) generated through principal components analysis. The scores were summed up for each household, individuals were ranked according to the total score of their households. The sample was then divided into four hierarchical groups from quartile one (lowest) to quartile four (highest).
Data on physical activity were collected based on self-report. First, respondents were asked the number of days they engaged in vigorous, moderate, or light physical activity throughout a typical week. Examples of vigorous, moderate and light activity were shown to the participants using showcards. Next, they were asked to estimate how many minutes per day they engaged in the activity. We then calculated metabolic equivalent tasks (MET)-minutes per week using the STEPwise Surveillance of noncommunicable disease risk factors (STEPS) protocol [29]. Finally, quintiles of MET-minutes were created, and the highest quintile was labelled as strenuous physical activity. Smoking habit was asked and recorded as current smoker, former smoker and non-smoker of any tobacco product such as cigarette, bidi and hukkah (water pipe).
History of physical trauma during last 12 months that needed medical treatment with or without residual damage, e.g., injuries due to accidents while travelling by road, trauma during occupational works while working in farming lands or factories, physical assault, etc., were obtained. Using height (meters) and weight (kilograms) measurements, we calculated BMI (weight/height2). People having BMI ≥25.0 were labelled as over-weight (this includes obese also). Random capillary blood glucose was measured. Diabetes was defined as blood glucose ≥11.1 or use of antidiabetic medication.
Positive respondent:
A subject was considered a positive respondent if he/she reported occurrence of pain at muscles, bones, joints, or any part of the body (musculoskeletal symptom) during the preceding seven days. Subjects who did not report pain on those seven days but were taking prescribed medicines for relieving pain, e.g., non-steroidal anti-inflammatory drugs or steroids, were also included. The respondents in whom musculoskeletal pain appeared, developed, or disappeared in the preceding seven days were also labeled as a positive respondent.
MSK conditions:
All positive respondents were interviewed and thoroughly examined by the research physicians. Internationally accepted criteria [34-38] were used with adaptations whenever necessary. For conditions with no internationally accepted criteria and epidemiological definition, the clinical judgment of the research physician was used. In case uncertainty, opinion was taken from the investigators during their routine visit to respective PSUs. Following criteria were used for diagnosis of the MSK conditions:
- Rheumatoid arthritis: 2010 American College of Rheumatology (ACR)/European League Against Rheumatism (EULAR) Classification Criteria [39];
- Spondyloarthritis (axial and peripheral): Ankylosing Spondylitis Assessment Study (ASAS) criteria [40];
- Ankysosing spondylitis: Modified New York Criteria 1984 [41];
- Psoriatic arthritis: Classification Criteria for Psoriatic Arthritis (CASPAR) criteria [42];
- Knee osteoarthritis: ACR clinical classification criteria for knee osteoarthritis (OA) [43];
- Systemic Lupus Erythematosus: ACR Revised Criteria for the Classification of Systemic Lupus Erythematosus 1997 Systemic Lupus Erythematosus [44];
- Soft tissue rheumatism: Commonly included subacromial bursitis, epicondylitis, trochanteric bursitis, anserine bursitis, and fibromyalgia [45];
Considering the limitations of investigations in the field situation, the differentiation between non-specific low back pain and lumbar spondylosis was not possible in many cases. Therefore, we have pooled these two together to report he prevalence. These are reported as low back pain.
Disability and work loss:
Disability was scored with a validated Bangla version of the Health Assessment Questionnaire (B-HAQ) [46]. This tool assesses the subjects’ level of functional ability and included questions of fine movements of the upper extremity, locomotor activities of the lower extremity, and activities that involve extremities. The B-HAQ included 20 items referring to basic activities of daily living, grouped into eight categories of functioning, viz., dressing and grooming, arising, eating, walking, hygiene, reach, grip and activities. Each category contained two or three specific component questions. Respondents are asked to rate the degree of difficulty they experienced in carrying out each activity on a 4-point rating scale: 0 (without any difficulty), 1 (with some difficulty), 2 (with much difficulty), and 3 (unable to do). The highest response in each category was divided by 8 to create a B-HAQ Disability index (B-HAQ-DI), yielding a total disability score of 0–3, where zero is no disability and 3 is severe disability [47]. Any one scoring ≥0.8 for B-HAQ-DI was categorized as having disability according to Quintana R et al [49].
The recall period for determining work loss was 12 months. We have asked the participants whether they had to stop their usual occupational work, paid or unpaid (such as home makers), due to MSK conditions or related pain. Then the duration of such work loss (in days) was asked and recorded
Statistical analysis:
The data were entered into Excel spreadsheet and transferred to EpiInfo (version 7) for analysis. Missing values were identified to confirm the denominators, and consistency were checked.
All quantitative variables such as age, years of education, body mass index (BMI), B-HAQ-DI score were categorized before analysis. Alfa was set at 5% for considering statistical significance. Therefore 95% confidence intervals (CI) were calculated for all prevalence estimates such as MSK conditions, disabilities and related work loss. Results were presented for four reporting domains: rural and urban residential locations, and sex groups. Univariate logistic regression analysis was done for 11 candidate variables (age, sex, education, wealth quartiles, urban residence, smoking, strenuous physical activity, occupation, overweight, history of physical trauma, and diabetes) to get odds ratios (ORs) with their 95% CIs for MSK conditions combined (yes/no). Tri-variate logistic regression analysis was done for nine candidate variables (education, wealth quartiles, urban residence, smoking, strenuous physical activity, occupation, over-weight, history of physical trauma, and diabetes) to obtain age and sex adjusted odds ratios and their 95% confidence intervals of MSK conditions combined.
ETHICS APPROVAL AND CONSENT TO PARTICIPATE
Ethical guidelines as outlined by the Declaration of Helsinki were followed throughout the study [49]. Ethical clearance was obtained from the Institutional Review Board of Bangabandhu Sheikh Mujib Medical University. Concurrence has been obtained from the local health authorities and elected representatives of the local government prior to data collection. Written (or thumb impression if unable to write) consent was obtained from the respondents in Bangla as per Institutional Review Board guidelines.