The protocol for this systematic review has been registered in the International Prospective Registry of Systematic Reviews (PROSPERO) under the registration number CRD42023444710.

Relevant studies were retrieved from electronic databases, such as PubMed/MEDLINE, Science Direct, Cochrane Library, CINAHL, African Journals Online, HINARI, and Sematic Scholar. Google and Google Scholar searches were also used to search for relevant articles. Besides, gray literatures were also identified from different university’s digital libraries and published articles. The following key terms were used to search the studies: “non-compliance,” “compliance,” “smoke-free law,” “smoking-ban,” “smoking ban,” “smoke-free legislation,” “associated factors,” “factors,” “determinant factors,” “factors associated,” “public institution,” “home,” “restaurant,” “cafe,” “hospitals,” “schools,” “bar,” “bar and restaurant” and “public place.” All key terms were combined using the Boolean operators “AND” or “OR” as appropriate (Supplementary material S1). The search was carried out up to November 23, 2023 by four authors independently (CD, MG, AKG, and AA). Those studies searched from selected databases were transferred to Endnote version 20 and duplicate files were excluded. The process of selecting these studies was following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines (39) (Supplementary material S2).

Population: This meta-analysis includes global studies conducted on smoke-free law in public places.

Studies without full text, qualitative studies, irretrievable studies, letters-to-the editor, studies with poor methodological quality, and studies that did not report the outcome of interest were excluded from the meta-analysis.

The primary outcome of the study was to estimate the pooled prevalence of non-compliance with smoke-free law in public places, determined by dividing the number of public places/smokers by the total sample size and multiplying by 100.

Three authors (CD, AA, and MG) independently extracted all the necessary data using a standard data extraction template. The data extraction template consisted of various study details, such as author’s name, country, publication year, type of public place, study design, methods of data collection, response rate, and prevalence. Three reviewers (CD, AA, and AKG) screened the relevant articles for inclusion after duplicate files were removed. The quality of each article was evaluated using the Joana Brigg Institute (JBI) critical appraisal checklist for cross-sectional studies (40) (Supplementary material S3). Two authors (CD and AA) independently assessed the quality of each article, with scores measured on a scale of 100%. A quality score of greater than 50% was used to include articles for further analysis (41, 42). In the case of any discrepancies encountered during the quality assessment, the mean score was computed from the evaluations of all reviewers.

The level of heterogeneity among the included studies was statistically evaluated using the Higgs I² test, with values of 25, 50, and 75% indicating low, moderate, and high heterogeneity, respectively (43). Because high heterogeneity was observed among the included studies (I² = 100%, p < 0.000), DerSimonian and Liard (44) method of random-effects model was used to estimate the pooled prevalence of non-compliance with smoke-free law in public places.

Forest plot was used to present the pooled prevalence of non-compliance. A sensitivity analysis was performed to assess the influence of a single study on the pooled prevalence estimates. Subgroup analysis was also conducted based on various study characteristics, such as year of publication (before 2020 or 2020 and after), sample size (small- < 1000 or large- ≥1000) and methods of data collection (observation, self-administration or review document). Besides, publication bias was assessed using a funnel plot and Egger’s test with a p-value less than 0.05 suggesting a publication bias (45). Moreover, meta-regression analysis was also carried out considering variables, such as year of publication, sample size and method of data collection to the outcome variable.

A total of 2013 articles were identified from an international database. Using the Endnote reference manager, 587 duplicate articles were excluded; while 1,389 were excluded as they do not meet the inclusion criteria based on their titles and abstracts. Besides, 20 articles were excluded based on the quality of the assessment and the outcomes of the studies. Finally, 23 full-text articles were eligible for this meta-analysis and the processes of selecting these studies followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guideline (Figure 1).

In this meta-analysis, 25,573,329 study participants were involved. Five of the included studies focused on healthcare facilities (19, 34–37), sixteen studies (14–17, 21–27, 29–31, 33, 38), one study on education sector (38), and one study on food and drinking service establishment (32). All of the included studies followed cross-sectional study design. From the included studies, the highest non-compliance with smoke-free law was found to be 98.2% in food and drinking establishments (32) and the lowest non-compliance was 10.3% in healthcare facilities (36). Regarding the study country, four studies were conducted in India (29, 31, 35, 37), two in China (17, 30), two in Russia (28, 38), one in United states (15), one in Guatemala (24) one in Indonesia (16), two in Ethiopia (14, 36), two in South Africa (23, 32), one in Australia (19), one in Portugal (34), one in European Union (27), one in Bangladesh (26), one in Vietnam (33), one in Turkey (22), one in Pakistan (21), and one in Nepal (25). Among the included the studies, the sample size ranged between 25,475,032 (15) and 40 (37) (Table 1).

Regarding methods of data collection, thirteen studies collected data using observational methods (14, 16, 21, 25, 26, 29, 30, 32–37), five studies using review report (15, 23, 28, 31, 38), three studies using self-administered questionnaire (17, 19, 27), one study using both observational and interview (22), and one study collected using data measuring (24) (Table 1).

We found that the pooled prevalence of non-compliance with the smoke-free law is 48.02% (95% CI: 33.87–62.17) (Figure 2). Among the types of public places, the highest pooled prevalence of non-compliance was observed in hotels (59.4%) followed by homes (56.8%) (Table 2).

Funnel plot revealed the presence of significant publication bias (Figure 3). The Egger test statistics also revealed the presence of statistically significant publication bias (p = 0.002). To determine the sources of this bias, a trim and fill analysis was conducted, revealing notable variation in the newly estimated pooled odds ratio, denoted as the adjusted point estimate [OR = 3.56, (95% CI: 3.19–3.94)], when compared to the initial or observed point estimate [OR = 4.36, (95% CI: 4.35–4.37)] (Figure 4).

The pooled prevalence of non-compliance with smoke-free law was found to be higher among studies conducted after 2020 (58.4, 95% CI: 36.16–78.56) than studies conducted before 2020 (44.4, 95% CI: 27.55–61.18) (Figure 5). When categorized based on the sample size, the highest prevalence of non-compliance with smoke-free law was reported among studies with a small sample size (53.4, 95% CI: 39.91–68.82) as compared to studies with a high sample size (42.2, 95% CI: 21.58–62.74) (Figure 6). Regarding the methods of data collection, the highest pooled non-compliance was observed among studies conducted by review report (50.3, 95% CI: 29.05–71.51) followed by observational studies (48.1, 95%CI: 33.87–62.25) (Figure 7).

A univariate meta-regression model was carried out to pinpoint the source of heterogeneity by considering methods of data collection, year of publication, and sample size as factors. However, none of these variables demonstrated statistical significance (Table 3). Furthermore, a sensitivity analysis was performed to evaluate the impact of individual studies on the overall pooled estimate of non-compliance. The results also indicated no single study effect (Figure 8).