PMC:4144696 / 4936-15741 JSONTXT

{"project":"2_test","denotations":[{"id":"25146939-24467862-60562896","span":{"begin":1957,"end":1959},"obj":"24467862"},{"id":"25146939-24400999-60562897","span":{"begin":6554,"end":6555},"obj":"24400999"},{"id":"25146939-9055211-60562898","span":{"begin":7247,"end":7249},"obj":"9055211"}],"text":"Methods\n\nEthics\nThe Nepal Health Research Council, the Institutional Review Committee of Kathmandu University School of Medical Sciences, Dhulikhel Hospital (IRC-KUSMS), and The Central Regional Committee for Health and Research Ethics in Norway all approved the study protocol.\nProspective participants who could do so read the written information approved by the ethics committees describing the nature and purpose of the study; to those who were illiterate, the interviewers read the information in the presence of family members. All had opportunity to ask questions. Before interviews commenced, consent was given in writing when possible but otherwise by fingerprint in accordance with requirements of IRC-KUSMS. All participants knew they were free to discontinue their interviews at any time.\nPersonal identification details of participants were separated from the completed questionnaires and stored in a locked room in the Department of Epidemiology at the Kathmandu University School of Medical Sciences. No information relating to identifiable individuals was disseminated beyond the researchers immediately involved in the study. All data were protected in accordance with Norwegian data-protection legislation.\n\nStudy design\nThis was a cross-sectional, population-based survey using face-to-face structured interviews administered by trained interviewers. It was conducted by unannounced visits to households, with multistage random cluster sampling.\n\nPopulation of interest\nThe participants were adults aged 18–65 years who were Nepali speaking and living in Nepal. Excluded were immigrants (defined as those who had stayed for \u003c6 months in the household and locality) and those who were deaf or otherwise unable to participate through major physical or mental health conditions.\n\nSample size\nWe estimated a total sample size of 2,100, assuming a headache prevalence ≥10% and an absolute margin of error of 1.3% with 95% confidence interval [13].\n\nSampling\nTo ensure an adequate representation from the country as a whole, we used a multistage random sampling technique described below. We divided the sample in the proportions 8:44:48 (numerically 170, 930, 1,000) between Mountain, Hill and Terai physiographic divisions according to their relative populations [14], and within each division equally between the five regions.\nFurthermore, we aimed to recruit 25% of participants from above 2,000 m.\n\nSelection of districts\nThe sampling procedure included each of the three physiographic divisions (Mountain, Hill and Terai) and five developmental regions (Far Western, Mid-Western, Western, Central and Eastern) dividing the country (Figure  1). By randomly selecting one district in each region of each division, we sampled 15 districts out the total of 75, spread across the country. We made one purposive change to the random selection, in the central developmental region of the Hill division, replacing Sidhuli by Ramechap because we needed one more cluster above 2,000 m.\nFigure 1 Flow chart of sites selection of the study.\n\nSelection of village development committees (VDCs) or municipalities\nEach district consists of several village development committees (VDCs) in rural areas or municipalities in urban areas. We selected one of these local units from each district by a lottery sampling technique. Before doing this in the five districts in the Mountain division, VDCs or municipalities below 2,000 m were excluded so that sampling was done only from those above 2,000 m (representative of the mountainous terrain). In the Hill division, we applied an element of convenience sampling: two VDCs were selected purposively because access was very difficult to districts above 2,000 m in the selected districts (Ramechap and Rukum).\n\nSelection of clusters\nEach VDC or municipality is built up of wards, these being collections of houses. In VDCs there are always nine wards, but in a municipality there may be more. We selected one ward per VDC or municipality, randomly by the lottery method, the households therein forming the clusters for final sampling (Figure  1).\n\nSelection of households\nWe defined a household as a group of people living together and sharing a common kitchen, and considered this as the sampling unit. A household in Nepal consists on average of 4.4 people [15]. A residential map of the cluster was drawn, excluding non-residential properties (empty or locked houses, store houses or animal sheds) with the help of the Female Community Health Volunteers (FCHVs) (local women working to improve community participation and enhance the outreach of health services at community level in Nepal). To select the first household we went to the center of the cluster and randomly selected a direction by the \"spin the bottle\" method. We then selected every household along the line in this direction until the projected sample size was achieved. If there were not enough households in the ward, the sampling continued in the same direction into the next adjoining ward of the same community.\n\nSelection of participants\nIn each household, the head, or a person who could provide reliable information, was asked to list all the family members within the age range 18–65 years. From this list, one was randomly invited to participate by the sealed envelopes method (this is a method of avoiding bias in selecting participants: the sealed envelope contains the numbers of family members, from which the head of the household picks one randomly). No replacements were permitted within a household for participants who withheld consent or failed to be available for interview.\n\nRecruitment and training of interviewers\nWe advertised in a national daily paper to find interviewers from throughout the country, selecting 47 of 150 applicants with backgrounds in health care, fluent in English and Nepali and with prior experience of population-based surveys. We divided them into eight teams according to their residence (knowledge of local language and customs) and appointed one member of each, with best skills, as team leader.\nWe provided a 4-day training programme at Dhulikhel Hospital on important aspects of the MDBs, on the principles of epidemiology and on the specific methods and requirements of the study. We observed them during simulated field interviews and gave feedback.\n\nStudy instruments\nFor headache diagnosis and headache-attributed burden estimation, we used the Headache-Attributed Restriction, Disability, Social Handicap and Impaired Participation (HARDSHIP) questionnaire developed by LTB [6].This modular instrument included demographic enquiry and headache questions for all participants, then, for those reporting headache in the preceding year, diagnostic questions based on ICHD-3 beta followed by enquiries into the multiple components of burden [16]. We translated this into Nepali language following the LTB translation protocol for hybrid documents [17]. We administered the translated version in a pre-pilot study to 20 patients presenting at the Psychiatry Outpatient Department in Dhulikhel Hospital to verify cultural acceptability and inoffensiveness.\nWe imported Nepali-translated and culturally-validated versions of the Hospital Anxiety and Depression Scale (HADS) [18] and Eysenck Personality Questionnaire-Neuroticism (EPQ-N), revised version [19], into HARDSHIP (Additional file 1) to assess psychiatric morbidity in terms of caseness (a dichotomous measure) and symptom severity (a continuous measure).\nWe used a digital device (3BM1-3® by Microlife) to record blood pressure, with the participant sitting on the floor because many Nepalese would not have a chair in their homes. We used a simple portable scale to measure weight with shoes and outdoor clothes removed, and a measuring tape for height and waist circumference.\nWe used a portable altimeter (SAL 7030® by Sunoh) to establish the altitude of each household.\n\nPilot study\nThe pilot study was a full test, in the field, of the survey methods and instruments. We convenience-selected Kavre district, with three clusters (Dhulikhel Municipality, Panchkhal and Kavre Bhanjyang), and set the sample size at 10% of that of the main study (ie, n = 210). Review during its progress and on completion led to modifications of the questionnaire and allowed assessment of the time required for data collection.\n\nMain survey\n\nEngagement with participants\nWe ensured that each interviewer team included at least one member from the locality, and the assistance of FCHVs was sought in engaging with the community (Risal et al. submission of manuscript in Journal Headache and Pain).\nSelected households were visited unannounced (cold-calling). Since most people were occupied with agriculture, most homes were locked during the day time. Interviewers therefore visited in the very early mornings and in the evenings. The survey was also continued during weekends (Saturday only in Nepal) and holidays. If the selected respondent was present and consented, the interview took place immediately. If he or she was not present at that time, a convenient appointment was made for a second visit, and, if necessary, a third. After three failed attempts, the selected person was registered as a non-participant.\n\nQuality assurance\nWe prepared a working manual for use by interviewer teams. In the field, completed questionnaires were reviewed by the team leader for completeness, accuracy and legibility at the end of each day. The team leader looked specifically for use of Arabic symbols for numbers (having different meanings in Nepali), mismatched age and gender of participants and differences between the list given by the head of household and the information from the participants. When minor mistakes were seen, the team leader corrected them after discussion with the interviewer; major mistakes or missing data missing were rectified by revisiting the household.\nEvery day while in the field, each team leader was required to contact the Nepali researchers (AR and KM) and report difficulties or queries. In addition, AR and KM made surprise visits to one of the two assigned sites of each team during data collection to ascertain that this was being done in accordance with expectations. We reviewed a random 10% sample of completed questionnaires and also cross-checked the collected data by revisiting 10% of households during the visit.\n\nData management\nAll completed questionnaires were safely stored by the team leader in a plastic-coated box at the end of each day. After completion of data collection, they were sealed in double plastic bags and brought back to Dhulikhel hospital.\nThe data were entered by AR and KM into IBM SPSS Statistics version 20. Double entry of all data was made, with inconsistencies reconciled by reference to the original documents."}