Analysis of Differences in the Effects of Smoking on Variaous Semen Parameters: A Multivariable Propensity Score Matching Study

Technical Summary: Analysis of Differences in the Effects of Smoking on Various Semen Parameters

Problem Statement
Male reproductive health is increasingly recognized as a critical public health concern, with a global decline in semen quality sparking significant attention. While numerous factors influence semen quality, including genetics and pathology, unhealthy lifestyle habits—specifically smoking—are a primary modifiable risk factor. Tobacco smoke contains thousands of mutagenic compounds (e.g., nicotine, heavy metals, polycyclic aromatic hydrocarbons) known to cause structural damage to the male reproductive system, induce DNA fragmentation, and increase spermatogenic cell apoptosis.

Despite extensive research, academic consensus regarding the specific impact of smoking on distinct semen parameters remains inconsistent. Some studies suggest a "devastating" effect on all parameters, while others indicate that smoking primarily affects sperm concentration and count, leaving motility and morphology relatively unaffected. These discrepancies are often attributed to insufficient control of confounding factors in previous observational designs. Factors such as age, BMI, alcohol consumption, and sleep duration often differ between smokers and non-smokers, potentially biasing results if not rigorously controlled. This study aims to resolve these controversies by employing a robust statistical design to isolate the specific effects of smoking on four key parameters: sperm concentration, total sperm count, progressive motility, and sperm viability.

Methodology
This study utilized a cross-sectional design involving 1,355 male participants who underwent semen analysis at the Prenatal Diagnosis Unit of a tertiary hospital in Xinjiang between 2021 and 2025. Participants were categorized into a smoking group (n=626) and a non-smoking group (n=729).

• Data Collection: Demographic characteristics, lifestyle habits, and medical history were collected via an anonymous electronic questionnaire. Semen parameters were measured according to the WHO Laboratory Manual (5th edition) standards.
• Propensity Score Matching (PSM): To address confounding bias and simulate a randomized controlled trial, the study employed Propensity Score Matching. A logistic regression model calculated propensity scores based on 14 covariates, including age, residence, education, occupation, income, marital status, reproductive history, BMI, alcohol consumption, exercise, sleep duration, and environmental exposure.
• Matching Process: A 1:1 nearest-neighbor matching method was applied with a caliper of 0.25 standard deviations, resulting in 499 matched pairs (total n=998). Covariate balance was verified using Standardized Mean Differences (SMD), with an SMD < 0.1 indicating successful balancing.
• Statistical Analysis:
  • Effect Estimation: Independent-samples t-tests and multivariable linear regression were used to estimate the mean differences and independent effects of smoking on semen parameters. Effect sizes were calculated using Cohen's d.
  • Robustness and Sensitivity Analysis: To ensure the reliability of findings, the study conducted sensitivity analyses using Inverse Probability of Treatment Weighting (IPTW) and varied covariate combinations in regression models. Consistency was assessed via coefficients of variation (CV) and relative changes in Average Treatment Effects (ATT).

Key Results

• Covariate Balance: After PSM, the SMD for all covariates was reduced to < 0.1, confirming excellent balance between the smoking and non-smoking groups.
• Sperm Concentration and Total Count: Smoking was found to have a statistically significant negative impact on sperm quantity.
  • Sperm Concentration: The smoking group showed a mean concentration of 27.30 ± 25.78 × 10⁶/mL compared to 33.72 ± 35.61 × 10⁶/mL in the non-smoking group (Mean Difference: -6.42, P = 0.0049).
  • Total Sperm Count: The smoking group had a mean count of 101.89 ± 111.51 × 10⁶ compared to 129.57 ± 139.53 × 10⁶ in the non-smoking group (Mean Difference: -27.67, P < 0.001).
  • The effect sizes were small but significant (Cohen's d: -0.21 and -0.22, respectively).
• Motility and Viability: No statistically significant differences were observed between the two groups regarding progressive motility (PRNP) or sperm viability (P > 0.05).
• Robustness: Sensitivity analyses confirmed the stability of these findings. The coefficients of variation for sperm concentration and total count across different PSM methods were 0.111 and 0.087, respectively (both < 0.2), indicating high robustness. Conversely, results for motility and viability showed lower stability but remained non-significant across methods.

Significance and Claims
The authors claim that this study provides a more rigorous scientific basis for understanding the relationship between smoking and male reproductive health by effectively controlling for confounding variables through PSM. The primary contribution is the clarification that smoking's detrimental effects are parameter-specific: it significantly impairs the quantity of sperm production (concentration and total count) but does not significantly alter the quality of sperm function regarding motility or viability in this population.

The study emphasizes that these findings are highly robust, supported by consistent results across multiple matching methods and covariate combinations. The authors conclude that smoking acts as an independent negative predictor for reduced sperm concentration and total count. Consequently, the study advocates for targeted smoking cessation campaigns for men planning to conceive and suggests that future research should focus on prospective cohort designs and dose-response relationships to further validate these causal associations and explore underlying biological mechanisms. The authors modestly note that their cross-sectional design limits causal inference and that the binary definition of smoking status (without accounting for duration or intensity) represents a limitation in precision.