The CKB is an ongoing population-based prospective cohort study, and details of the study design, methods and participant characteristics have been published previously [31, 32]. Briefly, the study recruited 512,715 Chinese adults (including 302,522 females) aged 30–79 years from five urban and five rural areas in China between June 2004 and July 2008. Extensive data collection was conducted at baseline via interviewer-administered questionnaires, which gathered information on socio-demographic, lifestyle and environmental factors (including duration, amount and frequency of tobacco smoking), medical history, and reproductive history for females. Trained staff collected blood samples and performed physical measurements (e.g. height, weight, lung function) following standardised protocols. Three resurveys were conducted in random samples of 4–5% of the surviving participants in 2008, 2013–2014, and 2020–2021. Ethical approvals were obtained from the Oxford Tropical Research Ethics Committee at the University of Oxford (Oxford, United Kingdom) and the Ethical Review Committee of the Chinese Centre for Disease Control and Prevention (CDC), Beijing, China. All participants provided written informed consent upon recruitment.

Self-reported information on female reproductive history was collected at baseline, including age at menarche, numbers of pregnancies, spontaneous or induced abortions, stillbirths and live births, age at birth and breastfeeding duration for each live birth, OC use, duration of use, OC starting age, menopausal status, age at menopause (for postmenopausal females), and history of hysterectomy and oophorectomy. For each post-menopausal female, the reproductive period was calculated as the duration between age at menarche and age at menopause.

Monitoring participants’ vital status was conducted by regularly screening official residential records and death certificates available from the regional CDC. Cancer incidence was ascertained through linkages to established cancer registries and national health insurance databases (covering ~ 98% of the study participants) using their unique national identification numbers. All events were coded by trained staff blinded to baseline data following the International Classification of Diseases, 10th Revision (ICD-10) [33]. The primary outcome of interest in the present study was incident lung cancer (ICD-10: C33-C34).

Among 302,522 females recruited at baseline, after excluding ever smokers (defined as occasional smokers who had not completely stopped smoking for at least 6 months before baseline, those who had smoked ≥ 100 cigarettes but had quit smoking by choice for ≥ 6 months before baseline and regular smokers, n = 15,330); participants with a prior history of any cancer (n = 1,518); and participants who reported having been ever oral contraceptive users but reported 0 months as duration of use (n = 301), 285,373 females never-smokers (individuals who had smoked < 100 cigarettes during their lifetime [34]) remained. To reduce any influence of extreme values, females in the top and bottom 0.1% of age at menarche, age at first live birth and age at menopause were further excluded from the analysis (n = 2,171). In addition, the top 0.1% (n = 644) were excluded for the number of pregnancies and breastfeeding per child. There were no missing data in the remaining variables included in the analysis except for family history of any cancer (n = 10,098), which was assigned into a separate category in the subsequent analyses. Cohen’s kappa (κ) or Spearman correlations were used to assess the agreement between reported exposures at baseline and subsequent resurveys.

The categorisation of the exposure variables was performed in accordance with previous studies [35, 36], where appropriate, with some regrouping done based on the frequency distribution of specific variables as follows: age at menarche (< 13, 13–14, 15–16, > 16 years), number of pregnancies (never pregnant, 1–2, 3–4, > 4), parity defined as the total number of live births and stillbirths (nulliparous, 1, 2, 3–4, > 4 births), age at first birth (< 20, 20–22, 23- 25, > 25 years), average breastfeeding duration per child (never breastfed, < 7, 7–12, > 12 months), OC use (ever, never), duration of use (never users, ≤ 5 and > 5 years), age at starting using OC (never users, ≤ 25 and > 25 years old), menopausal status (pre- and perimenopausal, postmenopausal), age at menopause (< 43, 43–52, > 52 years), total reproductive period (< 30, 30- 35, > 35 years), history of oophorectomy (yes, no) and history of hysterectomy (yes, no).

Crude incidence rates of lung cancer were calculated by categories of each reproductive factor. Cox proportional hazard regression was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) of incident lung cancer by each reproductive factor reported at baseline with age as the underlying time scale, and individuals were considered at risk from the age at study entry. The models were sequentially adjusted for the following covariates: age (years), study area (10 areas), occupation (agriculture and related workers, factory workers, administrative/technical/service workers, retired/unemployed, housewives, others), highest attained education (no formal schooling, primary school, middle school, high school, technical school/college/university), log-height and log-weight, physical activity calculated in metabolic equivalent of tasks (METs-h/day), family history of any cancer (yes, no, missing), personal history of lung disease including TB, COPD and asthma (yes/no) alcohol drinking (never/ever regular drinkers), frequency of environmental tobacco smoking exposure (never/almost never, < once/week, 1-2days/week, 3-5days/week, daily/almost daily), exposure to heating fuel during winter (no heating, coal, wood, central heating/gas/electricity/others) and cooking fuel (never/no cooking facility, coal, wood, gas/electricity/others). To assess for trend, categorical variables were entered into the model as numeric.

Additional analyses were conducted by further adjusting for age at menarche, the number of pregnancies, OC use, age at menopause, and history of oophorectomy or hysterectomy. The analysis of age at first birth and breastfeeding per child was conducted among parous females only (n = 279,107), while the analysis of age at menopause and reproductive period was restricted to postmenopausal females at the study baseline (n = 143,890). Breastfeeding was additionally grouped into three groups: never breastfed, ≤ 12 and > 12 months, to ensure comparability with subsequent subgroup analysis.

Subgroup analyses and likelihood ratio tests for interaction were conducted for child-bearing factors and OC use among subgroups defined by menopausal status, year of birth (before 1950/in or after 1950, a cut-off selected to investigate the impact of the one-child policy, assuming that females who were born after 1950 have reached childbearing age by the time of the policy implementation), and area (rural/urban). To maintain a sufficient number of cases in each category, never-pregnant and nulliparous females were excluded from the subgroup analysis, parity was regrouped into (1–2, 3–4, > 4 births), and average breastfeeding per child regrouped into (≤ 12, > 12months).

Sensitivity analyses were conducted by restricting the analysis to never-alcohol drinkers, never-oral contraceptive users, and those with no prior history of major lung diseases at baseline (TB, COPD and asthma). All analyses were performed using Stata/SE software version 16.1 (StataCorp, College Station, TX), and figures were plotted using R software version 3.3.2. We considered p-values < 0.05 as evidence of an association.