|
nif:isString
|
-
The Research Ethics Review Committee of the World Health Organization and the Ethics Committee of the Ministry of Health of Zanzibar approved this study. Participants were informed orally about this study and also given a detailed information sheet. Only those who gave written consent were interviewed. No compensation was offered for the interview. Interview data sheets did not bear the names of respondents and all data were anonymized before analysis.
The east African archipelago of Zanzibar belongs to the United Republic of Tanzania and is inhabited by ∼1.2 million people who are predominantly Muslim. Kiswahili is the main language, but English is also widely used. The archipelago is located ∼60 km off the coast of mainland Tanzania and consists of two major islands—Unguja in the south and Pemba in the north—and several islets; it can be reached from the coast by ferry or air within 20 minutes to 2 hours. Zanzibar has been regularly affected by cholera; the first cases in recent times were detected in 1978 [36], [37]. This study was conducted in the peri-urban Shehia of Chumbuni (population ∼11,000) in Unguja and the rural Shehia of Mwambe (∼8,000) in Pemba. Both Shehias (administrative term for community in Zanzibar) were among the core areas of a mass vaccination campaign that was conducted in early 2009 by the Ministry of Health of Zanzibar (MoH) with support from the WHO. The sample for this study was drawn from these two Shehias because they had been studied in a pre-vaccination survey in 2008 [23], [24]. The peri-urban site, an unplanned, slum-like extension of the capital, mainly consists of brick houses and is characterized by a high population density; the rural site consists of hamlets and most people live in mud houses. More details of both sites have been reported elsewhere [24].
The mass vaccination campaign aimed to vaccinate ∼50,000 inhabitants with Dukoral®, a two-dose OCV containing killed V. cholerae O1 bacteria and recombinant cholera toxin B subunit protein [38]. Dukoral® was the only OCV pre-qualified by the WHO at the time of vaccination. It requires a cold chain for storage and safe water (∼1.5 dl per dose) for its administration. It was offered without charge in two rounds from January 17 to 26 and February 7 to 16, 2009, to residents aged two years or older from six Shehias from Unguja and Pemba that had been identified as recent cholera hotspots. Nine vaccination posts were set up on each island that operated daily for at least eight hours and were staffed with local healthcare workers and villagers. Information activities for the campaign started with a meeting with district officials on December 23, 2008, followed by three meetings to inform leaders, Shehia committee members and mobilizers from each community (January 5 and 10, 2009) and general community residents (January 15, 2009) (MoH, Health Promotion Unit, OCV Social Mobilization Report, February 20, 2009). A refresher meeting in the communities followed shortly before the second round on February 5, 2009. Social mobilization used posters, leaflets, street banners and T-shirts to disseminate information on the OCV campaign and to reinforce general hygiene and sanitation messages in the six Shehias. Messages were continuously broadcast on national TV and radio from the first until the last day of the campaign. The local press was also briefed and newspaper articles reported from the campaign to promote participation. The campaign was officially launched by the Minister of Health who drank the vaccine publicly at the Chumbuni Primary Healthcare Unit (Zanzibar Today, January 18, 2009). Mobilizer teams were formed for each Shehia and delivered information from house to house and by megaphone. Each team consisted of five to six community residents representing also women's groups, youth, religious groups and members of the opposition party. Key messages highlighted not only the importance of vaccination for cholera prevention, but also promoted hygiene messages to prevent other diarrheal diseases, and explained administration of the OCV, its characteristic features and potential for mild side effects.
This was a cross-sectional interview survey based on a case-control design to identify factors associated with vaccine uptake among vaccinated and unvaccinated community members targeted for the mass vaccination campaign. In addition, unvaccinated participants were also interviewed about barriers to uptake for site- and gender-related comparative analysis. Data were collected in June and July 2009, six months after the mass vaccination campaign. The sampling frame for this study was derived from the census database that had been compiled by the International Vaccine Institute shortly before the mass vaccination campaign implementation in early 2009 [39]. Names, age, sex, OCV vaccination status and a unique house identification number were extracted for both study Shehias. Respondents' houses in Chumbuni were located with the help of aerial photographs indicating house numbers; houses in Mwambe were located with the help of local assistants. Approximately 380 adults, based on a sample size of 330 [40] with 15% compensation for missing data, were identified following a stratified random sampling procedure. After exclusion of respondents who had been interviewed before the vaccination for the baseline study [24], all respondents aged 18 years and older were selected. Second, peri-urban and rural respondents were separated and groups of women and men created among them. Third, of the approximately 95 women and 95 men required per site, 50% were selected from those who had received two doses of the OCV, 40% from those who had not received a single dose and 10% from those with one dose only. Only residents who were physically and mentally fit to stand an interview of approximately one hour duration were included in the sample. Women who had not taken the vaccine because of pregnancy during the mass vaccination campaign were not interviewed.
Semi-structured interviews based on the Explanatory Model Interview Catalogue (EMIC) are the principal instrument for cultural epidemiological studies and elicit locally valid features of illness-related experience (operationalized as categories of distress), meaning (perceived causes) and behavior (help seeking) [25], [35]. An EMIC interview for study of diarrhea-free community residents was developed based on the pre-vaccination survey [24] (see supporting information, Text S1). A ten-day workshop was conducted shortly before the survey to train field workers and pilot the EMIC interview in Shehias adjacent to the study communities. After recording relevant socio-demographic characteristics, interviews began with the telling of a brief story in easily understandable terms, making use of a clinical vignette that described a cholera patient with cardinal somatic symptoms. To study socio-cultural features of cholera-like illness, respondents were asked a series of open and closed questions. These elicited patterns of distress (i.e., respondents' opinions on what additional physical symptoms the cholera patient described in the vignette might suffer from, and how the illness might impact him/her socially, emotionally and financially), perceived causes (i.e., what causes the illness may be attributed to) and help-seeking behavior (i.e., what would usually be done at the patient's home for self treatment and what sources of help would be consulted outside the household). Respondents who did not swallow two doses of the OCV during the mass vaccination campaign were queried about their reasons against vaccination by specifically inquiring about barriers related to logistical, social and system-relevant and medical aspects.
Quantitative data were recorded by interviewers on data sheets, double entered in Epi Info 3.5.1 (CDC, Atlanta, GA, USA) by data entry clerks and cleaned for statistical analysis in SAS 9.2 (SAS Institute, Cary, NC, USA). Qualitative data were written down during the interview by note takers in Kiswahili (or in English in a few cases). After translation into English, narratives were typed in a pre-coded word processor template that reflected interview items; this procedure followed the pre-vaccination survey [23]. This enabled automatic importation of entire interviews with codes into the qualitative data analysis software MAXQDA 10 (VERBI Software, Consult. Sozialforschung. GmbH, Marburg, Germany). For integrated analysis of quantitative and qualitative data, quantitative variables (see below) were imported into MAXQDA 10; this made it possible to retrieve narrative segments based on analytically relevant findings or statistical relationships.
Multivariable analysis of factors of uptake: Socio-demographic characteristics were coded as numeric or categorical variables. Categories of socio-cultural features of cholera-like illness were assigned a value of 2 if they were mentioned spontaneously and a value of 1 if they were mentioned only after probing. Those among the reported categories that were identified as single most troubling (among patterns of distress), most important (perceived causes) or most helpful (help seeking) were given an additional value of 3. A cumulative prominence was then calculated for each category ranging from 0 to 5. This approach based on the ranked prominence of responses has been widely used in analytic cultural epidemiological studies, which have examined how socio-cultural features of illness affect health behavior [32], [34]. To identify social and cultural factors explaining OCV uptake, a multivariable logistic regression model was calculated. The outcome variable (i.e., OCV vaccination status) was obtained from mass vaccination campaign data that had been compiled electronically during the campaign [39]. Based on the recommended schedule for Dukoral® requiring two doses for full protection, respondents who had received two doses were coded as 1 (“vaccinated”) and those who had received only one or no dose were coded as 0 (“unvaccinated”). The regression analysis included interaction with site as suggested by site-specific findings from the pre-vaccination survey [24] and because OCV uptake was higher in the rural than in the peri-urban site (58.8% vs. 40.8%, p = 0.001). Only explanatory variables reported by 5–95% were considered for analysis. Following the approach taken in the pre-vaccination survey [23], variables whose univariable association with OCV uptake had a p<0.2 were identified first. Second, multivariable regression models related to patterns of distress, perceived causes and help seeking were run by considering only variables that were retained in the first step. Each of these sub-models was adjusted for socio-demographic characteristics. To calculate the final model, only those variables which were retained with a p<0.2 in these sub-models were considered. Interaction between each explanatory variable and site (rural vs. peri-urban site at baseline) was tested in sub-models; only interaction terms retained with a p<0.1 in sub-models were used in the final model. The final model reports adjusted odds ratios with 95% confidence intervals and p values. In case of significant interaction with site, site-specific estimates are presented.
Descriptive analysis of barriers to uptake: Coding and calculation of variables related to barriers followed the approach used for socio-cultural features of illness. Unvaccinated respondents' spontaneous and probed answers for each barrier and the barrier they identified as most important were recorded. Thematically similar barriers were subsumed under groups of logistical, medical and social/system-related barriers. The non-parametric Wilcoxon test was used for identifying statistically significant differences of prominence between both sites and between genders.
|
![[This page as RDF]](http://data.gesis.org/somesci/static/rdf-icon.gif){kind=icon}