OBJECTIVE: Methods for meta-analysis of studies with individual participant data and continuous exposure variables are well described in the statistical literature but are not widely used in clinical and epidemiological research. The purpose of this case study is to make the methods more accessible. STUDY DESIGN AND SETTING: A two-stage process is demonstrated. Response curves are estimated separately for each study using fractional polynomials. The study-specific curves are then averaged pointwise over all studies at each value of the exposure. The averaging can be implemented using fixed effects or random effects methods. RESULTS: The methodology is illustrated using samples of real data with continuous outcome and exposure data and several covariates. The sample data set, segments of Stata and R code, and outputs are provided to enable replication of the results. CONCLUSION: These methods and tools can be adapted to other situations, including for time-to-event or categorical outcomes, different ways of modelling exposure-outcome curves, and different strategies for covariate adjustment.

OBJECTIVE: Methods for meta-analysis of studies with individual participant data and continuous exposure variables are well described in the statistical literature but are not widely used in clinical and epidemiological research. The purpose of this case study is to make the methods more accessible. STUDY DESIGN AND SETTING: A two-stage process is demonstrated. Response curves are estimated separately for each study using fractional polynomials. The study-specific curves are then averaged pointwise over all studies at each value of the exposure. The averaging can be implemented using fixed effects or random effects methods. RESULTS: The methodology is illustrated using samples of real data with continuous outcome and exposure data and several covariates. The sample data set, segments of Stata and R code, and outputs are provided to enable replication of the results. CONCLUSION: These methods and tools can be adapted to other situations, including for time-to-event or categorical outcomes, different ways of modelling exposure-outcome curves, and different strategies for covariate adjustment.

BACKGROUND: Cigarette smoking is associated with earlier menopause, but the impact of being a former smoker and any dose-response relationships on the degree of smoking and age at menopause have been less clear. If the toxic impact of cigarette smoking on ovarian function is irreversible, we hypothesized that even former smokers might experience earlier menopause, and variations in intensity, duration, cumulative dose, and age at start/quit of smoking might have varying impacts on the risk of experiencing earlier menopause. METHODS AND FINDINGS: A total of 207,231 and 27,580 postmenopausal women were included in the cross-sectional and prospective analyses, respectively. They were from 17 studies in 7 countries (Australia, Denmark, France, Japan, Sweden, United Kingdom, United States) that contributed data to the International collaboration for a Life course Approach to reproductive health and Chronic disease Events (InterLACE). Information on smoking status, cigarettes smoked per day (intensity), smoking duration, pack-years (cumulative dose), age started, and years since quitting smoking was collected at baseline. We used multinomial logistic regression models to estimate multivariable relative risk ratios (RRRs) and 95% confidence intervals (CIs) for the associations between each smoking measure and categorised age at menopause ( < 40 (premature), 40-44 (early), 45-49, 50-51 (reference), and > /=52 years). The association with current and former smokers was analysed separately. Sensitivity analyses and two-step meta-analyses were also conducted to test the results. The Bayesian information criterion (BIC) was used to compare the fit of the models of smoking measures. Overall, 1.9% and 7.3% of women experienced premature and early menopause, respectively. Compared with never smokers, current smokers had around twice the risk of experiencing premature (RRR 2.05; 95% CI 1.73-2.44) (p < 0.001) and early menopause (1.80; 1.66-1.95) (p < 0.001). The corresponding RRRs in former smokers were attenuated to 1.13 (1.04-1.23; p = 0.006) and 1.15 (1.05-1.27; p = 0.005). In both current and former smokers, dose-response relationships were observed, i.e., higher intensity, longer duration, higher cumulative dose, earlier age at start smoking, and shorter time since quitting smoking were significantly associated with higher risk of premature and early menopause, as well as earlier menopause at 45-49 years. Duration of smoking was a strong predictor of age at natural menopause. Among current smokers with duration of 15-20 years, the risk was markedly higher for premature (15.58; 11.29-19.86; p < 0.001) and early (6.55; 5.04-8.52; p < 0.001) menopause. Also, current smokers with 11-15 pack-years had over 4-fold (4.35; 2.78-5.92; p < 0.001) and 3-fold (3.01; 2.15-4.21; p < 0.001) risk of premature and early menopause, respectively. Smokers who had quit smoking for more than 10 years had similar risk as never smokers (1.04; 0.98-1.10; p = 0.176). A limitation of the study is the measurement errors that may have arisen due to recall bias. CONCLUSIONS: The probability of earlier menopause is positively associated with intensity, duration, cumulative dose, and earlier initiation of smoking. Smoking duration is a much stronger predictor of premature and early menopause than others. Our findings highlight the clear benefits for women of early smoking cessation to lower their excess risk of earlier menopause.

OBJECTIVES: The International Collaboration for a Life Course Approach to Reproductive Health and Chronic Disease Events (InterLACE) project is a global research collaboration that aims to advance understanding of women's reproductive health in relation to chronic disease risk by pooling individual participant data from several cohort and cross-sectional studies. The aim of this paper is to describe the characteristics of contributing studies and to present the distribution of demographic and reproductive factors and chronic disease outcomes in InterLACE. STUDY DESIGN: InterLACE is an individual-level pooled study of 20 observational studies (12 of which are longitudinal) from ten countries. Variables were harmonized across studies to create a new and systematic synthesis of life-course data. MAIN OUTCOME MEASURES: Harmonized data were derived in three domains: 1) socio-demographic and lifestyle factors, 2) female reproductive characteristics, and 3) chronic disease outcomes (cardiovascular disease (CVD) and diabetes). RESULTS: InterLACE pooled data from 229,054 mid-aged women. Overall, 76% of the women were Caucasian and 22% Japanese; other ethnicities (of 300 or more participants) included Hispanic/Latin American (0.2%), Chinese (0.2%), Middle Eastern (0.3%), African/black (0.5%), and Other (1.0%). The median age at baseline was 47 years (Inter-quartile range (IQR): 41-53), and that at the last follow-up was 56 years (IQR: 48-64). Regarding reproductive characteristics, half of the women (49.8%) had their first menstruation (menarche) at 12-13 years of age. The distribution of menopausal status and the prevalence of chronic disease varied considerably among studies. At baseline, most women (57%) were pre- or peri-menopausal, 20% reported a natural menopause (range 0.8-55.6%) and the remainder had surgery or were taking hormones. By the end of follow-up, the prevalence rates of CVD and diabetes were 7.2% (range 0.9-24.6%) and 5.1% (range 1.3-13.2%), respectively. CONCLUSIONS: The scale and heterogeneity of InterLACE data provide an opportunity to strengthen evidence concerning the relationships between reproductive health through life and subsequent risks of chronic disease, including cross-cultural comparisons.