Study area, design and period
A population-based cross-sectional study was conducted in Dabat district, northwest Ethiopia from July to October, 2012. The district had an estimated population of 145,458 living in 27 rural and 3 urban kebeles (the smallest administrative unit in Ethiopia). Like the rest of the districts in the northern part of the country, the livelihood of the community largely depended on subsistence agriculture. Only two health centers delivered the directly observed chemotherapy (DOTS) service [9]. There was an on-going TB surveillance project at Dabat Health and Demographic Surveillance System (HDSS) hosted by the University of Gondar. The HDSS covered 10 randomly selected kebeles with a total population of about 46,165 [10]. This study was conducted on the non-HDSS kebeles where routine government health services, case detection, and treatment of TB under the National Tuberculosis and Leprosy Control Programs were available [2] (see map, Figure 1).
Participants and data collection
All people aged ≥15 years and were living in the study area permanently (Government registered members) were considered as the source population and were included in the study if they were determined to have had cough for at least two weeks. Such people who were seriously sick and were on anti-TB treatment already were excluded from the study.
A pre-tested and structured interview questionnaire was used to collect data from the study participants to assess their health-seeking behavior for TB. The questionnaire contained detailed information on socio-demographic, behavioral, and environmental factors which were believed to affect the health-seeking behavior of the study participants (see Additional file 1). Data collectors performed house-to-house visits and asked the representatives of each household whether a person aged ≥15 years and had cough for at least two weeks at the time of the interview was present or not. Then data on health-seeking behavior for TB and associated factors were collected from each person who had with cough for at least two weeks. At the end of the interview, health education was given to the households with the intention of reducing stigma towards TB.
Data quality assurance
The training of data collectors and supervisors emphasized issues such as data collection instruments, field methods, inclusion–exclusion criteria, and record keeping. The principal investigator and supervisors coordinated the interview process, spot-checked and reviewed the completed questionnaires on a daily basis to ensure the completeness and consistency of the data collected. They also conducted random quality checks by re-interviewing about 10% of the respondents. The interview questionnaire was pre-tested on 25 respondents who had characteristics nearly similar to people in Dabat district in order to identify potential problem areas, unanticipated interpretations, and cultural objections to any of the questions. Based on the pre-test results, the questionnaire was adjusted contextually. The entered data by the principal investigator and another independent individual were compared to check for any variations in results.
Sample size calculation
Sample size was calculated on the Open-EPI sample size calculator software using a 5% level of significance, 4.2% prevalence of cough [2], the total adult population (aged ≥15 years) of 43,128, a 2% margin of error, a design effect of 2, and a non response rate of 10%. The final calculated sample size was 843.
Sampling technique
We purposively considered 20 non-HDSS kebeles for sampling selection with the intention of excluding the effect of interventions by the TB surveillance project at the HDSS sites. Then 10 kebeles were randomly selected by the lottery method and the required sample sizes were allocated using probability proportional to size formula. Interviewees in the chosen kebeles were identified by house-to-house visits.
Data management and statistical analysis
Data entry and cleaning carried out using the Epi Info Version 2002 statistical software, were analyzed on SPSS software package version 16.0. Descriptive statistics, such as frequency distribution, mean, and percentage were employed for most variables. Forward stepwise binary logistic regression analysis was done to assess the relative importance of the explanatory variables on the dependent variable (appropriate health seeking behavior). The odds ratio (OR) with a 95% confidence interval (CI) was used to test the statistical significance of variables.
Operational definitions
Health-seeking behavior
Any action undertaken by individuals who perceive themselves to have health problems or to be ill for the purpose of finding an appropriate remedy [11].
Appropriate health-seeking behavior
The intention to seek diagnosis for TB at a medical facility or service in the event of a chronic cough [6]. In this study, those who mentioned to have visited health posts, health center and the hospital had appropriate health seeking behavior. Those who mentioned to have visited private clinics, pharmacies, Holy water, traditional healers and other sources had inappropriate health seeking behavior.
TB suspect
A person who had cough for at least two weeks [5].
Knowledge of TB
Those study participants who mentioned germs as causes of TB had good knowledge of TB [6].
Stigma towards TB
Eleven questions with 4 responses each (strongly disagree, disagree, agree and strongly agree) were asked to assess stigma. A total stigma score for TB was created by summing up the scores of all questions. Individuals who scored above the mean value were categorized as having high stigma towards TB [12].
Income
In this study, income referred to monthly real per capita income of the participants. Employed workers were asked their monthly salary, where-as farmers were asked the annual amount of cereal harvested and changed to Birr which was then divided by the months of the year. For the analysis, we used 500 Birr, which is the average urban and rural monthly real per capita total consumption expenditure set by the Federal Ministry of Finance and Economic Development of Ethiopia [13].
Family size
In this study we used 5 (the average fertility rate of Ethiopia) as an average household size [14].
Marital status
For analysis purposes, marital status was grouped into Single and Married. Single included never married, divorced, widowed, and separated.
Ethical considerations
The study protocol was reviewed and approved by the Institutional Review Board of the University of Gondar via the Institute of Public Health. Government officials at various levels and community leaders were consulted and permission was obtained prior to data collection. Study participants were interviewed after informed written consent was obtained. Informed written consent regarding eligible participants below 18 years was obtained from parents or legal guardians. Individual records were coded and accessed only by the research staff (Additional file 1). Immediate referrals were arranged to the nearest health facility for those who had cough for at least two weeks.