 |
Tobacco and Cancer Programme International Union Against Cancer 3, Rue du Conseil-Général, 1205 Geneva, Switzerland Tel: (4122) 809 1830, Fax: (4122) 809 1810 E-mail: tobacco-control@globalink.org | Tobacco and Cancer Programme |
 |
Objective: To assess the magnitude of the tobacco problem.
Uses: Strengthening of advocacy positions
Identification of high risk sub-groups
Evaluation of a tobacco control programme
Trend in tobacco use behaviour.
Order and brief description of steps essential for planning and conducting a prevalence survey.
Population: Identify the target population
Generally this is the adult population of a country or a region. Or, depending on requirements, it may be school children, specific occupational group, etc.
Define inclusion and exclusion criteria
For example, age range, residence requirement in the region, etc.
Tobacco use or no tobacco use must not be a criterion. Avoid ill-defined
populations such as "hospital patients and visitors",
"weekly market or shopping mall customers", etc.
Sampling
If a sample is to be selected, decide on the sampling procedure
for getting an unbiased, representative sample. Avoid self-selection,
that is volunteers who come forward with information and answers.
Sample size
There is no need to use large numbers for a prevalence survey
on tobacco use. Large numbers are generally required in surveys
on the association between tobacco use and diseases.
One way of determining sample size is the desired precision of the estimate of prevalence in the smallest sub-group. For example, if the sample size is 100 the prevalence estimate is expected to be at least within 10% and if it is 400 then at least within 5% (plus or minus) of its true value.
After determining the desired sample size in the sub-group that is expected to be of the smallest size, work backwards and estimate what total sample size would give you the desired number in the sub-group. Sample size depends a lot on the resources and other practical considerations.
Questionnaire: Develop the questionnaire
A prevalence survey on tobacco use ought to contain at minimum the following items of information: age, sex, level of education, types of tobacco use (cigarette smoking, other smoking, smokeless tobacco), the age of starting tobacco use and the frequency of tobacco use per day. Depending on the objectives it would be desirable to include other items of information such as indicators of socio-economic status, reasons for starting the tobacco use, perceived association with diseases, attitude towards tobacco control measures, attempts at quitting the habit, etc.
Instructions: Prepare an instructions manual in which every question is explained.
Use international standardized criteria, e.g. those given by the WHO and the UICC as far as possible. Pretest the questionnaire on a few (10-20) individuals and, if necessary, revise.
Methodology: Decide the methodology for administering the questionnaire and conducting the survey. Common methods are sending the questionnaire by post to each individual in the sample, getting the questionnaires filled by individuals at a time, for example, school children in a classroom setting, or eliciting information through
appointed interviewers using a specific procedure, e.g. house-to- house. The last method may be the only valid option if part of the population is illiterate.
Pretest the proposed methodology in a pilot survey on a small sample.
Explore the possibility of checking the reliability of the answers to the questionnaire by using a biochemical test (e.g. urinary or salivary cotinine) to independently confirm tobacco use status at least for a small sub-sample (10% or more of the sample).
Non-response: During the conduct of the prevalence survey, pay a great deal of attention to the problem of non-response, as this greatly reduces the accuracy of the survey. Reduce it as much as possible, for example, by sending reminders, changing the time of the interview, repeating the visits etc.
Editing: Recheck all questionnaires. Prepare a coding manual and code all the questionnaires for inputting into computer, if this is possible. Input data preferably in the form of a rectangular file, that is one record per person, the same number of columns and a specific single item of information in the same specific column for all individuals.
Data analysis: Prevalence of tobacco use is usually expressed as the percentage of users among the whole population. It is worked out by dividing the number of smokers by the number of people interviewed and multiplying this figure by 100.
The prevalence in a particular sub-group, for example among women aged 25-34 years, is defined as (women in the age-group 25-34 years who use tobacco divided by the number of women interviewed in the age group 25-34 years) x 100.
The term "tobacco user" may be replaced by "cigarette smoker" to obtain prevalence of cigarette smoking or by any other type of tobacco use.
Reporting: Write a report or a paper and get it published and disseminated. A painstakingly conducted survey is of little use unless the results are properly disseminated. Pay particular attention to an accurate and detailed description of methodology. Describe and report on any unusual kind of tobacco use that has been encountered.
While discussing sub-group analysis, give enough data so that readers may reconstruct your argument by themselves. For example if you comment on the age-distribution of smoking, along with age-specific prevalence figures, provide at least the numerator, or the denominator. An interested reader can then calculate the remaining number, if required, e.g. for a test of significance. It is sufficient to provide prevalence figures up to one decimal place.
Mailing list: Prepare a mailing list of interested individuals and organizations and send a copy of the report or paper to each one. This list could include:
March 1993
|
Tobacco and Cancer Programme International Union Against Cancer 3, Rue du Conseil-Général, 1205 Geneva, Switzerland Tel: (4122) 809 1830, Fax: (4122) 809 1810 E-mail: tobacco-control@globalink.org | Tobacco and Cancer Programme |
|