Assessing public support for extending smoke-free policies beyond enclosed public places and workplaces: protocol for a systematic review and meta-analysis 65 3 Outcomes and prioritisation Data will be extracted for each estimate of public support by the spaces that they cover (e.g. playgrounds, private cars, multi-unit housing). If weighted and unweighted estimates are presented, we will extract estimates that are weighted to most adequately reflect the general population. If multiple estimates are presented that relate to public support, we will extract the estimate that covers the most general spaces. For example, we will prioritise “it should be illegal to smoke in all playgrounds” above “it should be illegal to smoke in this specific playground”. If public support is asked in general and specifically related to children, we will extract both estimates. For example, we will extract “it should be illegal to smoke in private cars” and “it should be illegal to smoke in private cars when minors are present”. Risk of bias assessment We will assess risk of bias for each study using the Mixed Methods Appraisal Tool (MMAT) for descriptive studies. The MMAT 2018 version was developed based on criteria from 18 existing critical appraisal tools and input from over 50 international experts. The following five elements will be assessed: relevance of the sampling strategy, representativeness of the target population, appropriateness of the outcome measurements, risk of non-response bias, and appropriateness of the statistical techniques. Each of the elements will be categorised by using the answer categories yes, no or can’t tell, following MMAT criteria. Results of the risk-of-bias analysis will be presented in tables.30 Data synthesis Obtaining comparable data is essential to facilitate meta-analysis, thus homogenisation of the outcome data is needed. Public support will be analysed as proportional data, i.e. proportion of the population supporting a particular smoke-free policy. When results are fairly normal distributed the raw proportions will be analysed, if not logit transformations will be applied31The outcome estimates will be reversed if studies report on the proportion not in favour of the smoke-free policies. Often Likert-scale type questions are used to assess support, if studies report percentages per answer option instead of total support, the answer categories above neutral (i.e. indicating a positive response) will be combined.
RkJQdWJsaXNoZXIy MTk4NDMw