This nationwide retrospective cohort study was based on data from the Korean National Health Insurance Service (KNHIS), a government-run program that provides universal and mandatory healthcare coverage to approximately 97% of the Korean population. ^, The NHIS conducts standardized biennial health examinations that capture a wide range of individual-level information, including demographic and socioeconomic characteristics, clinical measurements, medical history, and lifestyle behaviors. The validity of the NHIS database has been well established, and it has been widely used in large-scale epidemiological studies. ^,, Sex was analyzed according to the male/female classification recorded in the KNHIS database. Gender was not assessed in this study. In this retrospective cohort study, informed consents were waived since the data were strictly anonymized in compliance with the Personal Data Protection Act. This study was approved by the Institutional Review Board of Guro Hospital (2024GR0390), and was conducted in accordance with the tenets of the Declaration of Helsinki.

Of the 280 249 participants who underwent consecutive biennial examinations between 2009 and 2020, we excluded those with a history of major vision-impairing diseases prior to follow-up (n = 100 428), those with missing baseline covariates (n = 348), and those without smoking-status information (n = 200). The main analytic cohort consisted of participants who quit CC between 2018 and 2019, leading to a total of 179 273 participants eligible for analysis ( Figure 1 ).

Study population flow diagram. Data on adults aged ≥19 who underwent biennial health screenings between 2011 and 2012 and 2018 to 2019 and quit combustible cigarette smoking between 2018 and 2019 were collected from the Korean National Health Insurance Service. Participants with prior diagnoses of major vision-impairing diseases, individuals with missing information for covariates, and individuals with missing information for smoking status were excluded from the final study population. CC = combustible cigarette.

Smoking status was determined from self-reported questionnaires from the biennial health examinations. Participants provided information on cigarette consumption (current status, duration, amount) between 2009 and 2020, and on NNTP use (current status, duration, amount) between 2018 and 2020. Complete cessation was defined as reporting active CC use in 2011 to 2012 and quitting by the 2018 to 2019 examination, with no subsequent NNTP use. Switchers were defined as participants who reported CC use in 2011 to 2012, reported quitting by 2018 to 2019, and initiated NNTP use during 2018 to 2019.

Cumulative smoking exposure was expressed as pack-years, calculated by multiplying the number of cigarette packs smoked per day by the number of years smoked, based on self-reported survey data between 2018 and 2019.

The primary outcome was major vision-impairing eye diseases, defined as a composite of 5 conditions that substantially contribute to the global burden of visual disability and can seriously affect vision if untreated or uncorrected. This composite included cataract, glaucoma, age-related macular degeneration (AMD), diabetic retinopathy, and refractive and accommodation disorders, which have been identified by the World Health Organization as priority conditions for prevention and intervention, and by Global Burden of Disease studies as the leading causes of blindness worldwide. ^,

The individual secondary outcomes were identified based on relevant International Classification of Diseases, 10th Revision (ICD-10) diagnostic codes (Supplementary Table 1). Cataract was defined using ICD-10 codes H25 and H26, consistent with previous epidemiological studies. ^, Glaucoma was defined by ICD-10 codes H40 and H42, excluding H40.0 (glaucoma suspects), in order to increase specificity and prevent misclassification of individuals with suspected but unconfirmed glaucoma. ^, AMD was defined by ICD-10 codes H35.3, in accordance with previous studies. ^, Diabetic retinopathy was defined by ICD-10 codes E10.3, E11.3, E13.3, H36.0, consistent with previous claims-based literature. ^,, Finally, refractive and accommodation disorders were defined by ICD-10 codes H52, encompassing myopia, hyperopia, astigmatism, presbyopia, and other specified or unspecified refractive errors.

Each participant was assigned an individual index date corresponding to the date of their health examination in 2018 or 2019, to represent the point at which CC cessation or NNTP switching was assessed. This moving index ensured that exposure status reflected the actual timing of the participant’s transition, rather than a fixed calendar date. All participants were observed from their index date (start of follow-up) until the first occurrence of major vision-impairing diseases, death, or December 31, 2023, whichever came first, providing a maximum follow-up of approximately 6 years.

Categorical variables were summarized as number (%) and continuous variables as mean (SD), or median (IQR) if non-normally distributed.

Propensity score matching (PSM) was employed as a design-stage strategy to minimize confounding and emulate a pseudo-randomized comparison, thereby strengthening the causal interpretability of the estimated effects. In addition to age, covariates related to sociodemographic and clinical information were used in matching, including sex, area of residence, household income, moderate-to-vigorous physical activity (MVPA), body mass index (BMI), pack-years, Charlson comorbidity index (CCI) and alcohol consumption. One-to-one nearest neighbor matching with a caliper of 0.2 SD of the logit of the propensity score was performed. To evaluate the association in the entire analytic cohort, inverse probability of treatment weighting (IPTW) was employed with the same set of covariates as PSM, generating a weighted pseudo-population. Balance was assessed using standardized mean differences (SMD) before and after matching.

To assess the risk of incident major vision-impairing diseases among CC quitters who initiated NNTP use compared to complete cessation, while considering all-cause mortality without onset as a competing event, we used the Fine–Gray subdistribution hazards model to estimate propensity score–matched subdistribution hazard ratios (SHRs) and 95% confidence intervals (CIs). Fine–Gray subdistribution hazards models were used to explicitly account for the competing risk of all-cause mortality, reducing bias that would arise if deaths were treated as noninformative censoring in a conventional Cox model. Furthermore, the Korean NHIS database is directly integrated with nationwide mortality records from Statistics Korea, enabling more rigorous competing-risk analyses across the entire cohort through complete and universal ascertainment of death events.

Two primary Fine–Gray models were evaluated: (1) a crude model with no covariate adjustment, and (2) a model adjusted for age (continuous; years) and sex (categorical; male/female). To further assess robustness and account for residual confounding, we additionally conducted a fully adjusted model covering a comprehensive list of sociodemographic, clinical, and lifestyle characteristics, including age (continuous; years), sex (categorical; male/female), household income (categorical; first, second, third, and fourth quartiles), area of residence (categorical; capital, metropolitan city, rural), BMI (continuous; kg/m²), moderate-to-vigorous physical activity (MVPA; categorical; no, 1-2 times/wk, 3-4 times/wk, and ≥5 times/wk), hypertension (categorical; no, yes), diabetes mellitus (categorical; no, yes), dyslipidemia (categorical; no, yes), alcohol consumption (categorical; none, light, moderate, heavy), history of drug abuse (categorical; no, yes), Charlson comorbidity index (CCI; categorical; <3, ≥3), and pack-years (continuous; years). Additional cause-specific risks of major vision-impairing eye diseases were estimated using the Cox proportional hazards model.

In sensitivity analyses, we excluded incident cases of major vision-impairing diseases occurring within the first 1, 2, and 3 years of follow-up to minimize potential reverse causation from preclinical or undiagnosed eye diseases. We conducted secondary sensitivity analysis using an alternative definition of major vision-impairing eye diseases that excluded refractive and accommodation disorders, which are correctable, unlike the more progressive conditions of cataract, glaucoma, AMD, and diabetic retinopathy. Cumulative incidence was reported as the number of events per 1000 person-years (PYs).

Subgroup analyses were conducted to identify potential effect modifiers. The participants were stratified by income, MVPA, obesity, hypertension, diabetes mellitus, dyslipidemia, CCI, alcohol consumption, and pack-years in the fully adjusted Fine–Gray model. Cumulative smoking exposure was categorized by pack-years as <20 and ≥20, with ≥20 pack-years considered heavy smoking exposure, consistent with prior ophthalmology research. Type 3 Wald chi-square test was employed to test for potential interactions between exposure and subgroups. Variance inflation factor (VIF) analysis was conducted to evaluate potential multicollinearity between adjustment variables. Temporal trends in CC and NNTP use by year were presented on the full cohort, expressed as the number of users divided by the total number of survey respondents who provided nonmissing information on smoking status in each year.

SHRs for individual secondary outcomes were presented. The outcomes were further divided into 3 age groups (20-39, 40-64, and ≥65 years) to identify potential associations based on age. SAS Enterprise Guide 7.1 (SAS Institute, Cary, NC, USA) was used for all data gathering, mining, and analyses. R (version 4.0.3; R Foundation for Statistical Computing, Vienna, Austria) was used to generate the cumulative incidence function.

The study included 32 316 participants after propensity score matching, with a mean age of 44.7 years (SD, 8.1 years) and 97.6% being male (Supplementary Table 2). Table 1 presents baseline characteristics of participants by exposure group. CC quitters without NNTP use had a median smoking exposure of 7.5 pack-years, while CC quitters with NNTP use had a longer median exposure of 8.4 pack-years. Table 2 shows covariate balance before and after matching; age displayed the largest baseline imbalance (SMD = −0.766), with all covariates achieving SMD < 0.05 postmatch (Supplementary Table 3 for IPTW balance).

TABLE 2

Standardized Mean Differences Before and After the Propensity Score Matching.

Variable	w/o NNTP Use vs w/ NNTP Use Among CC Quitter
	Before	After
Logit propensity score	0.820	<0.001
Age	−0.766	0.005
Sex	0.167	<0.001
Residence	0.094	0.019
Income	−0.211	0.012
MVPA	−0.054	0.014
Body mass index	0.122	0.008
Pack-years	0.050	0.020
Charlson comorbidity index	0.192	−0.015
Alcohol consumption	−0.120	0.017

Only gold members can continue reading. Log In or Register to continue

Share this:

• X
• Facebook

Related

Related posts:

1. Microperimetry as an Outcome Measure Improves Patient Eligibility and Efficacy Detection in RPGR Gene Therapy Trials
2. Posterior Segment Risk Factors for Penetrating Keratoplasty Failure
3. Clinical Utility of Ultra-Widefield Swept-Source OCT for Intraocular Tumors: Comparison With Ultrasonography, SD-OCT, and MRI
4. Comment on: Visual Field Progression in Patients With Glaucoma With Delayed Follow-Up

Stay updated, free articles. Join our Telegram channel

Join