Association between cigarette smoking and the risk of dysmenorrhea: A meta-analysis of observational studies.

Dysmenorrhea is the most common gynaecological disorder worldwide, with a prevalence ranging from 50% to 90% according to different countries [ 1 – 4 ]. The pelvic or lower abdominal pain is one of the most common symptoms of dysmenorrhea, which usually lasts for three days from the beginning of menstruation. Dysmenorrhea is classified as primary or secondary [ 5 ]. Primary dysmenorrhea (PD) refers to painful menses or cramps in the lower abdomen before and/or during menstruation without an identifiable organic pathology [ 6 ]. In contrast, secondary dysmenorrhea refers to menstrual pain resulting from anatomic and/or evident pelvic pathology, such as endometriosis [ 7 ]. The discomfort brought about by dymenorrhea has negatively affected women’s quality of life and their performance in everyday activities. In addition, the prevalence of dysmenorrhea causes heavy medical burden, such as serious health costs [ 8 ]. Therefore, dysmenorrhea is an urgent public health problem. Nevertheless, about more than 10% of adolescent girls have severe dysmenorrhea, which is the main reason for school and work absenteeism among the adolescent girls [ 9 ]. However, only a few women seek medical help for this disorder. Despite the preceding evidence on the prevalence of dysmenorrhea, factors influencing dysmenorrhea over the reproductive life span are not fully understood. In spite of that, hitherto the condition of dysmenorrhea correlates with a number of factors, such as family history of dysmenorrhea, age of menarche onset, caffeine consumption, dietary habits, exercise, cigarette smoking, and some psychological or gynaecological factors [ 10 ]. Cigarette smoking has been reported in literature as one of the modifiable risk factors for dysmenorrhea. Although women who smoke reported a range of more adverse reproductive outcomes than their non-smoking counterparts, the relationship between smoking and dysmenorrhea is still heterogeneous [ 11 , 12 ]. That is, some studies found that smokers were more likely to experience dysmenorrhea than non-smokers, whereas other studies reported that smokers were less likely to experience dysmenorrhea than non-smokers, and another study showed that smoking had no effect on dysmenorrhea. For instance, a study conducted in Turkey reported that smokers had a 1.6-fold higher risk of dysmenorrhea than non-smokers [ 13 ]. This was consistent with an earlier study, which found that women smokers experienced more severe dysmenorrhea, whose degree varied with the number of cigarettes smoked every day [ 14 ]. Furthermore, Parazzini et al. reported that the risk of dysmenorrhea increased with the duration of smoking and the number of cigarettes smoked every day [ 15 ]. In contrast, Andersch et al. reported that smokers had significantly less dysmenorrhea than non-smokers [ 16 ]. In addition, other three studies (two conducted in Turkey and one conducted in Japan) did not find an association between smoking and dysmenorrhea among women [ 17 – 19 ]. In this regard, the aim of this study was to quantitatively synthesize the previous findings on the association between cigarette smoking and dysmenorrhea and give further suggestions on interventions for dysmenorrhea.

The search strategy yielded a total of 3,671 studies, of which 24 were selected as eligible for this meta-analysis following a rigorous screening process ( Fig 1 ). Characteristics of the 24 eligible studies are displayed in Table 1 . Among these studies, the diversity of participants’ characteristics was considerable. Altogether, these studies investigated 27,091 female participants, among whom 11,731 (43.3%) had dysmenorrhea. In addition, among the 24 eligible studies, there were 4 case-control studies, 2 cohort studies and 18 cross-sectional studies. Moreover, 10 studies were conducted in Europe, 9 in Asia, 3 in North America and 2 in Oceania. The sample sizes of the eligible studies ranged from 258 to 9,067; while age of the participants varied between 15 and 59 years. Considering the individual eligible studies, the prevalence of dysmenorrhea varied between 20.1% and 91.8%. Also, 8 studies investigated cigarette smoking and the risk of primary dysmenorrhea, whereas others explored cigarette smoking and the risk of nonspecific dysmenorrhea. Many of these studies defined smokers as smoking at least one cigarette per day or they described the current smoking status of the participants, while few did not provide a clear definition. Furthermore, some studies defined dysmenorrhea as cramps or abdominal pain or backache one day before and/or the first day of menstruation, while other studies considered dysmenorrhea to be an activity-limiting pain that requires medication. Almost all studies used a self-administered questionnaire to measure the outcomes of dysmenorrhea with respect to smoking status. Additionally, 2 case-control studies and 2 prospective cohort studies were of high quality, whereas other 2 case-control studies were of medium quality according to the assessment tool of NOS ( S1 and S2 Tables). Also, among the 18 cross-sectional studies, 5 studies were of high quality, 10 studies were of medium quality, and 3 studies were of low quality, following the evaluation with the AHRQ tool ( S3 Table ). *data are presented as range or mean±SD; NA: not available The relationship between cigarette smoking and dysmenorrhea among females is shown in Fig 2 . Using random-effects model, the results of this meta-analysis showed that, smokers were 1.45 times more likely to develop dysmenorrhea than non-smokers (OR = 1.45, 95% CI: 1.30–1.61). This result was associated with medium heterogeneity ( I 2 = 41.0%, p = 0.02). Moreover, individuals classified as currently smoking were 1.50 times more likely to develop dysmenorrhea than those who were classified as never smoking (OR = 1.50, 95% CI: 1.33–1.70) and, again, medium heterogeneity was associated with this result ( I 2 = 48.0%, p = 0.005). Also, being a former smoker was 1.31 times more likely to develop dysmenorrhea than being a never smoker (OR = 1.31, 95% CI: 1.18–1.46), and no heterogeneity was associated with this result ( I 2 = 0%, p = 0.59). The results are shown in Fig 3 and Fig 4 . The results of sensitivity analysis indicated that inclusion and exclusion of Wood`s study [ 20 ] significantly affected the magnitude of the heterogeneity associated with the pooled OR. That is, after excluding this study, the pooled OR was reduced but still significant (OR = 1.39, 95% CI: 1.26–1.54), and with reduced heterogeneity ( I 2 = 28%). Besides, the pooled results were still significant after excluding one study at a time. Furthermore, results of subgroup analyses are shown in Table 2 . As regards prospective cohort studies, the pooled OR (OR = 1.35, 95% CI: 1. 09–1.67) was lower than that of case-control studies (OR = 2.03, 95% CI: 1.15–3.57) or cross-sectional studies (OR = 1.43, 95% CI: 1.26–1.62) ( S1 Fig ). Comparable pooled ORs were observed among studies conducted in Europe (OR = 1.45, 95% CI: 1.17–1.78), Asia (OR = 1.47, 95% CI: 1.19–1.81), North America (OR = 1.34, 95% CI: 1.12–1.61) and Oceania (OR = 1.64, 1.00–2.68) ( S2 Fig ) Considering type of dysmenorrhea, smoking increased the risk of primary dysmenorrhea by 56% among the females (OR = 1.56, 95% CI: 1.31–1.86) ( S3 Fig ). The pooled OR of studies with high methodological quality (OR = 1.34, 95% CI: 1.23–1.45) was lower than that of studies with medium methodological quality (OR = 1.45, 95% CI: 1.19–1.77) and those with low methodological quality (OR = 1.36, 95% CI: 0.81–2.28) ( S4 Fig ). According to the results of the Begg’s test ( z = 1.66; p = 0.097) and Egger’s test ( t = 1.97, p = 0.061), there was no significant publication bias. However, the fact that the preceding results were closer to being significant, there might be moderate or non-obvious publication bias as shown by the funnel plot ( S5 and S6 Figs).

Dysmenorrhea has been an important public health problem and has a significant negative impact on the affected females. The evidence from this meta-analysis indicated a significant association between cigarette smoking (both current and former smoking) and dysmenorrhea. Therefore, the adverse effects of smoking provide further support for dysmenorrhea prevention plans and emphasize on the need for health intervention programs for females in the future.

This meta-analysis was conducted in accordance with the guidelines for observation study protocols (Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines). A systematic search for literature was conducted in the electronic databases of PubMed, Medline, Web of Science, and EMBASE to locate related studies, published not later than November 2019, using the following keywords: “smoking” or “cigarette smoking” or “tobacco” or “nicotine” or “smoke” or “risk factors” in combination with “dysmenorrhea” or “menstruation pain” or “menstruation disorder” or “painful menstruation” or “menstrual painful” or “menstrual dysfunction”. Besides, reference lists of eligible studies were searched manually for relevant studies. Studies were considered eligible for inclusion in this meta-analysis if they fulfilled the following inclusion criteria: (a) observational studies published in English, which evaluated the main outcome as the relationship between cigarette smoking and dysmenorrhea, or the influencing factors of dysmenorrhea including cigarette smoking; (b) studies assessed the existence of dysmenorrhea in the smoking and non-smoking groups; and (c) studies presented the sample sizes, number of cases, and odds ratios (ORs) with 95% confidence intervals (CIs). The exclusion criteria of studies were as follows: (a) reviews, studies not on humans, abstracts, and comments; (b) studies on subjects experiencing passive smoking; (c) studies with inadequate information for this meta-analysis; and (d) duplicates of already included studies. Two authors (Lu-Lu-Qin, Zhao Hu) independently selected eligible studies by screening the titles, abstracts and full texts according to the eligibility criteria. Any discrepancies on the selected studies were resolved by consensus with another author (Bang-An Luo), who reviewed again the different selections and made decisions or agreements based on the eligibility criteria. Then, the following data were extracted from all the studies which met the eligibility criteria: (a) name of first author; (b) publication year; (c) location of the study; (d) sample size; (e) study design; (f) age; (g) type of dysmenorrhea; (h) prevalence of dysmenorrhea; (i) definition of exposure; and (j) outcome. The main variables for assessing the relationship between cigarette smoking and dysmenorrhea were as follows: (a) number of smokers (current and ever); (b) number of females who smoked cigarettes and had dysmenorrhea; (c) number of non-smokers; and (d) number of females who did not smoke and had dysmenorrhea. Extracted data were recorded in standardized tables and confirmed by two investigators separately (Lulu Qin, Zhao Hu). The methodological quality of cross-sectional studies was assessed using the tool, the Agency for Healthcare Research and Quality (AHRQ). Studies assessed using this tool are classified as high quality (if they scored 8–11), moderate quality (if they scored 4–7) or low quality (if they scored 0–3). The methodological quality of cohort or case-control studies was assessed using the New-Ottawa Scale (NOS). Studies assessed using the NOS are classified as high quality (if they scored 7–9), moderate quality (if they scored 4–6) or low quality (if they scored 0–3). The ORs together with their corresponding 95% CIs were calculated using the Mantel-Haenszel procedure for the comparisons of dichotomous outcomes between the smoking and non-smoking groups. The heterogeneity among the results of the included studies was evaluated using the I 2 statistic tests. A random-effects model was chosen to calculate the pooled ORs, when there was substantially moderate or high heterogeneity among the included studies. Besides, sensitivity analysis was performed by excluding one study at a time in order to assess the consistency of the pooled estimate of the ORs. Also, the possibility of publication bias was assessed using a funnel plot and the Begg’s and Egger’s tests (significance level, p <0.05). Furthermore, subgroup analysis was conducted to explore potential heterogeneity according to study design, study location, type of dysmenorrhea and study quality. Only two studies reported the number of cigarettes smoked; thus, we did not analyse the dose-response association between the degree of smoking exposure and dysmenorrhea due to insufficient data. Most statistical analyses were performed in RevMan Software (Version 5.3, Cochrane Collaboration, London, UK). The publication bias test and subgroup analysis were conducted in STATA version 12.0 (StataCorp, College Station, TX).

(DOC) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (TIF) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file.