UK Population Sampling
Adults aged 15 years or more were interviewed at home using IPSOS Mori's Capibus survey during a five-day period at the end of May 2011. This uses a two-stage method, the primary sampling units (PSU) being amalgamations of the 2001 UK Census Output Areas, each of which has approximately 125 addresses. The primary sampling units are randomly selected and within them randomly selected secondary sampling units are contiguous zones made up of Output Areas, usually two of them. This yields a random sample of 2,007 anonymous respondents, representative of the population at a national and regional level.
The standard tool was supplemented by a malaria-specific questionnaire covering knowledge of transmission, clinical features, severity and outcomes, and methods of prevention. Information on past travel and malaria destinations and willingness to pay for malaria prevention was also captured.
Airport departure lounge sampling was undertaken by trained interviewers through a face-to-face survey conducted by the Civil Aviation Authority (CAA). Five-hundred British nationals travelling to sub-Saharan Africa or other malaria-endemic countries, as defined in the UK's Advisory committee on Malaria Prevention (ACMP) guidelines, were interviewed prior to boarding their flight between June and August 2011. A multistage sampling design was employed on those who were randomly selected for interview and completed an anonymous CAA passenger questionnaire with added questions on malaria, similar to those used in the IPSOS survey, supplemented with additional information on chemoprophylaxis use for the journey, source(s) of pre-travel health advice and household income.
The data were cleaned and imported into R where descriptive tables were created. Analysis of the IPSOS data used the survey weights provided. For each of the two surveys, a malaria knowledge score was based on four questions relating to the mode of transmission, symptoms, severity, and curability of malaria. For the first two questions, multiple options were offered and negative points were assigned for incorrect answers. For the other two questions, a single option was chosen from a list, each option being assigned zero or a positive number of points. The score for each question was scaled to 25, resulting in a maximum knowledge score of 100.
Sources of pre-travel advice were categorized according to whether this was from a health source, non-health source (internet or friend) or no pre-travel advice. Within each survey, differences in knowledge score were calculated between subgroups. This was done using parametric methods (t test and regression) because initial comparisons with bootstrap gave very similar results, suggesting that they were sufficiently robust. Where 95% confidence intervals for a difference overlapped between the two surveys, a meta-analysis combined estimate was also calculated.
Malar J. 2013;12(461) © 2013 BioMed Central, Ltd.