Meta-analysis of the Association Between Second-hand Smoke Exposure and Ischaemic Heart Diseases, COPD and Stroke

Florian Fischer; Alexander Kraemer


BMC Public Health. 2015;15(1202) 

In This Article


Studies of SHS Exposure and Selected Outcomes

Overall, 33 studies were included in the systematic review. The first article was published in 1988, and the most recent in 2013. Several of the articles provided information on more than one outcome. Most articles described the effect of SHS exposure on IHD (n = 20). In 12 articles stroke was investigated as an outcome and eight articles focused on COPD (Table 1).

The spatial distribution of the study locations of all studies identified by the systematic review is quite equal: nine studies were performed in Asia (mainly in China and Hong Kong), Europe (mainly in Great Britain and northern European countries), and the USA. A further five studies were located in Australia and/or New Zealand and one in South America. Half of the articles described the results of a case–control study (n = 17) and the other half used a cohort design (n = 16). In almost all studies, information on SHS exposure was based on self-reporting (n = 30), while two studies performed a cotinine assessment for measuring SHS exposure and one study used a combination of self-reporting and cotinine assessment (Table 1). Usually, never-smokers or non-smokers were studied. However, some studies did not provide any information on the smoking status of subjects or included active smokers as well as non-smokers. In these cases, smoking status was controlled for in the analyses. All but two studies[63,64] controlled for several factors.

The study samples varied between 309 599 never-smokers in a cohort study in the USA, dealing with the association between SHS exposure and IHD[65] and a case–control study with 56 female IHD patients and 136 female controls in China.[66]

Effect Sizes for SHS Exposure and Selected Outcomes

SHS and ischaemic heart disease. The RR for the single studies dealing with the association between SHS and IHD are presented in Table 2. From the 20 studies on IHD in the systematic review, five were excluded because of low methodological quality according to the quality assessment. Additionally, the Greek study from Panagiotakos et al.[67] was excluded in the meta-analysis, because the same data was used in the study by Pitsavos et al.,[50] in which the analysis was stratified by place of exposure. This led to 14 studies on the effects of SHS exposure on IHD. In 6 of these studies, information summarized for both sexes were provided (n = 24 903). The RR for the association between SHS and IHD was either stratified by sex or only observed for one sex in six studies for men (n = 8208) and nine for women (n = 111 533).

The synthesis of all the studies included in the meta-analysis results in a RR of 1.27 (95 % CI: 1.10 – 1.48) for both sexes together. The RR was much higher for women (RR = 1.50, 95 % CI: 1.31 – 1.72) than for men (RR = 1.06, 95 % CI: 0.96 – 1.19). None of the studies showed significant results for men regarding the association between SHS exposure and IHD.

The studies from McGhee et al.,[24] Pitsavos et al.[50] and Rosenlund et al.[68] had the highest impact on the synthesis, because these three studies were weighted with 88 % overall. The results of Ciruzzi et al.,[69] with a very broad confidence interval (RR = 2.04, 95 % CI: 0.99–12.52), contributed only to a small extent to the overall RR due to the weighting factor of 1.39 %. For men, the study by McElduff et al.[70] contributed most to the synthesis result (46.59 %). For women, several studies contributed to more or less the same extent to the synthesis (Table 2, Fig. 2).

Figure 2.

Forest plot–SHS and ischaemic heart disease

Cochran's Q-test revealed no heterogeneity, because the p-value was larger than 0.1 for all three subgroup syntheses. This is confirmed by the I2-statistic, which quantifies the assumption between the three different subgroup syntheses. According to the results of these tests, no heterogeneity was observed for men (I2 = 0 %), and only a small but negligible heterogeneity for the studies focusing on women (I2 = 16.00 %). I2 was highest for studies including both sexes (I2 = 30.78 %), because the RR obviously differed for men and women (Table 2).

SHS and chronic obstructive pulmonary disease. Only five studies investigating the association between SHS exposure and COPD were included in the meta-analysis, after three further studies were excluded because of low quality. Overall, 28 965 participants were included in these studies, with more than half of them (n = 15 379) being investigated in one Chinese cohort study.[71] In three studies the RRs for the association between SHS and COPD were calculated for both sexes combined (n = 21,558). Only McGhee et al.[24] provided information stratified for men (n = 3,098) and women (n = 2,503) and two further studies investigated the association between SHS and COPD in a female-only study population (Table 3).

The large study by Yin et al.[71] accounted for almost half (49.49 %) of the weighting factor for both sexes. Two further studies, by Chan-Yeung et al.[72] and McGhee et al.,[24] accounted for 25 % each for the weighting factor in the subgroup of both sexes. For the female subgroup, the weighting factors were distributed in a similar way for the three studies included, although based on different studies.

The synthesis for both sexes is based on three studies with consistent and significant results. A RR of 1.66 with a comparatively small confidence interval (95 % CI: 1.38–2.00) was calculated. Since the synthesis for men is based on only one study, the RR of 1.50 (95 % CI: 0.96–2.28) was inherited. For women, a higher RR was identified (RR = 2.17, 95 % CI: 1.48–3.18) than for men (Table 3, Fig. 3).

Figure 3.

Forest plot–SHS and COPD

The heterogeneity between studies was assessed for the subgroups of both sexes and for women. The Q-statistic and its p-value suggested no heterogeneity between study results. The I2 for both sexes was 0 % and for women it was 22.95 %, which indicates no or only small heterogeneity (Table 3).

SHS and stroke

The results for stroke are based on seven studies, after five studies were excluded due to the quality assessment. Five studies provided information combined for both sexes (n = 52,263). In four studies the analysis was stratified for sex. This leads overall to 22 905 male study participants. Two large additional studies focused only on women, which leads overall to 162 197 female study participants, which allows for investigating the association between SHS exposure and stroke. For the synthesis of all three subgroups, the study performed by McGhee et al.[24] is of particular importance due to its high weighting factor (Table 4).

The synthesis for the three stroke subgroups differs from the two outcomes for IHD and COPD described above. In this case, the RR for the association between SHS and stroke is 1.35 (95 % CI: 1.22 – 1.50) for both sexes combined. The analysis separated for sex led to a slightly higher RR for men (RR = 1.40, 95 % CI: 1.09–1.81) as well as for women (RR = 1.43, 95 % CI: 1.28–1.61) compared to the synthesis for both sexes (Table 4, Fig. 4). This is due to the fact that the studies included in the meta-analysis in which both sexes are considered in a combined effect size are not exclusively the same as those which show results for men or women separately. One study only gives results for both sexes combined[25] and two studies only give results for women.[73,74]

Figure 4.

Forest plot–SHS and stroke

The Q-statistic indicated no heterogeneity, although the p-value for the Q-statistic for men was 0.184 and therefore close to the border indicating heterogeneity. According to the I2, the studies for women (I2 = 0 %) as well as for both sexes (I2 = 2.08 %) are homogeneous. For men, a low to medium heterogeneity was observed (I2 = 37.95 %) (Table 4).