Study design

This study was based on a real-world, longitudinal cohort study using a large claims database sourced from the Japan Medical Data Center Health Insurance Database (JMDC Inc., Tokyo, Japan).

Setting

The JMDC claims-database covers >14 million employees and their families aged ≤74 years between January 2005 to 2022 in Japan. It contains individual-level anonymous medical claims data (e.g. inpatient, outpatient, medications) linked with annual health check-up data (e.g. anthropometric measurements, blood test results, and lifestyle data, such as alcohol consumption, smoking status, and receptiveness for advice on improving healthy lifestyles) according to the Industrial Safety and Health Act. Information on the International Classification of Diseases 10th revision (ICD-10) status is included in this database. Additionally, WHO-ATC codes were used for drug coding. This study was approved by the Institutional Review Board of Aichi Medical University, Japan (approval no.: 2020–082).

Participants

Of a total of 4797329 insured people, we identified 1469239 individuals where claiming data could be confirmed at least once between January 2009 and December 2018. We classified the NCD diagnosis into four categories based on the following two criteria: 1) whether the diagnosed disease is generally considered serious and with a 5-year survival rate <75% (i.e. one quarter die within 5 years) (perceived severity); and 2) whether continuous smoking is associated with the progression and aggravation of the disease and whether or not smoking cessation after diagnosis is associated with improvements in disease outcomes (perceived susceptibility). Detailed criteria for the categorization of diseases are summarized in Table 1:

We extracted 1013991 records for those who were diagnosed with any of the above eight diseases and who had all been smokers at the time of their health check-up before the diagnosis.

Figure 1 shows a flow chart of the study inclusion criteria. We excluded suspected cases and those who had no health check-up data before and after diagnosis, and those aged <20 years who are forbidden to smoke by Japanese law. Additionally, we excluded patients with multiple diagnoses of these diseases, or who had a diagnosis of depression and were prescribed ≥2 of the following: selective serotonin reuptake inhibitors, serotonin and norepinephrine reuptake inhibitors, tricyclic antidepressants, or other antidepressant drugs (ATC codes: N06A4, N06A5, N06AA, N06A4, N06A5, N06AA, N05AX, N05AL). Finally, the records of 13284 patients were included in this study.

Figure 1

Flow chart of inclusion and exclusion criteria

Measures

The primary outcome of this study was to evaluate the smoking cessation rate within the first year after diagnosis. The secondary outcome was to compare changes in smoking rate every three years thereafter.

Variables

We collected demographic and health check-up data including age, sex, and readiness to adopt healthier lifestyles at pre-diagnosis as classified into a: precontemplation stage, contemplation stage, and preparation stage/action stage, using Prochaskas’ transtheoretical model¹⁹. Readiness to adopt a healthier lifestyle was routinely assessed using self-administered questionnaires at the health check-up. Smoking status data (smoking/not smoking) were collected from annual health check-ups pre-diagnosis, within one year of diagnosis (first health check-up), and every three years thereafter. A smoking habit was defined as smoking >100 cigarettes over the entire lifetime, or smoking for >6 months and in the previous month. This definition is used by the Japanese health check-up questionnaire. Participants who answered ‘yes’ to a question about smoking habits were classified as smokers, whereas those who answered ‘no’ were classified as non-smokers. Relapse was defined as a smoker who once became a non-smoker and then changed back to a smoker.

Statistical analysis

The summary statistics of participants were analyzed in four categories. To identify the impact of different diagnoses on smoking cessation, we performed multi-variable logistic regression analysis using glaucoma as the reference and adjusting for age and sex. The rate of smoking cessation accomplished was calculated for each category according to pre-diagnostic readiness for improving healthy lifestyles. To compare changes in smoking rate after diagnosis, we calculated the proportion of those who smoked at the first health check-up after the diagnosis and every three years thereafter. We also performed a sensitivity analysis for changes in smoking rate after diagnosis limited to those who participated at all follow-ups. Cochran-Armitage testing was performed to evaluate linear trends of unadjusted percentages.

Data analysis was performed with STATA ver.16 software (STATA Corp, College Station, TX). A p<0.05 (2-tailed) was considered statistically significant.

Baseline participant characteristics

The cohort comprised 11584 (87.6%) men, with 2126 (16%) aged <40 years, and 5822 (44%) aged ≥50 years. More than 40% of those participants diagnosed with asthma or appendicitis were in their 40s. More than half of the participants diagnosed with other diseases were aged ≥50 years (Table 2).

Smoking cessation rate at the first health check-up after diagnosis

Smoking cessation rates after diagnosis, in descending order, were Category I, II, III, and IV. The smoking cessation rate after diagnosis in Category I exceeded 59% (Table 3) whereas it ranged 31–40% in Category II. For Category III, the smoking cessation rate ranged 16–23%, and 9–13% for Category IV. The adjusted odds ratio for Category I was almost as great as multiplication for Categories II and III. Thus, the adjusted odds ratios for smoking cessation were 14.2 (95% CI: 11.4–17.8) to 14.8 (95% CI: 12.5–17.4) in Category I; 4.5 (95% CI: 3.8–5.4) to 6.6 (95% CI: 5.4–8.0) in Category II; and 1.9 (95% CI: 1.7–2.1) to 2.8 (95% CI: 2.2–3.7) in Category III (Table 3).

Trends in smoking cessation probability according to readiness to improve the lifestyle

Of the individuals whose diagnosis placed them in Category I, >50% of those who were precontemplating lifestyle improvement and smoking cessation before their diagnosis succeeded in quitting. Smoking cessation rates for those at the precontemplation stage, contemplation stage, and preparation or action stage before the diagnosis were 58.7%, 61.0% and 58.7%, respectively, for patients with an AMI diagnosis, and 65.2%, 54.0% and 59.9% for lung cancer patients (Table 4). In contrast, this was the case for <15% of those diagnosed with appendicitis and glaucoma (Category IV) or asthma (Category III). This rate significantly increased according to the position on the readiness scale for lifestyle improvement. Smoking cessation rates for patients at the precontemplation stage, contemplation stage, and preparation or action stage were 9.1%, 11.9% and 16.6%, respectively, for appendicitis patients, 6.9%, 9.8% and 10.5% for glaucoma patients, and 19.5%, 22.1% and 26.2% for asthma patients (p for trend: 0.022, 0.001, and 0.037, respectively) (Table 4). In contrast, participants whose NCDs were of perceived high severity (Category I and II) showed no significant increase in smoking cessation rates according to their level of readiness for lifestyle improvement.

Changes in smoking rate after diagnosis

Figure 2 shows changes in smoking rate at the first health check-up and every three years thereafter. In Category I, the smoking rate after diagnosis mostly remained the same or had increased a little three years after the diagnosis (40.4% to 43.4% for AMI patients, 40.7% to 45.3% for lung cancer patients), neither being statistically significant (p for trend for AMI and lung cancer 0.20 and 0.24, respectively). Similarly, in Category II, smoking rates after diagnosis did not change in a linear manner. Percentages of patients continuing to smoke after a diagnosis of colorectal or stomach cancer were 68.2% before to 65.1% after (p for trend = 0.20), and 59.8% to 57.5% (p for trend = 0.52), respectively. On the other hand, in Category III and IV, almost 80% of participants had continued to smoke after their diagnosis, and smoking rates declined gradually thereafter. Percentages of patients continuing to smoke after a diagnosis of asthma and TIA were 84.0% to 75.2% (p for trend <0.01) and 77.4% to 74.4 % (p for trend = 0.248); for appendicitis and glaucoma these values were 87.4% to 81.1% (p for trend = 0.004) and 90.7% to 91.0% (p for trend <0.001); respectively.

Figure 2

Changes in smoking rates after diagnosis: all participants

We conducted sensitivity analyses limited to those who participated in all follow-ups (Supplementary file Figure). In Category I, almost 40% of participants continued to smoke after diagnosis, whereas in Category III and Category IV, this was almost 80%.

Strengths and limitations

A strength of this study is that it is longitudinal, using a large sample consisting of insurance records from the Japanese population, thus allowing for sensitivity analyses that strengthen the robustness of the results. Another strength is that because our study using real-world data included a very large number of people, we were able to evaluate outcomes according to different NCD diagnoses. Our study also has some limitations. First, we used medical claims data. ICD-10 codes were used for all diagnoses. We excluded suspected cases, but there may be differences in the recorded ICD-10 codes and actual diagnoses. Second, we did not directly assess perceived severity or perceived susceptibility at NCD diagnosis at the individual level but defined these items based on the NCD diagnosis. This might increase misclassification in the four categories, which would underestimate the differences of smoking cessation rates among the categories. Third, our study had limited representability. It included only health check-up data of Japanese employees and all study participants were aged <75 years. Additionally, due to differences in medical technology and perceptions of diseases diagnosis, further studies are needed in different countries. Fourth, other potential confounders such as smoking-related factors (number of cigarettes smoked per day, duration of current smoking, and nicotine dependence score) and socio-economic factors should ideally have been considered. Nevertheless, our primary aims were to compare the smoking cessation rates after diagnosis of several different diseases, so the impact of this limitation is likely to be small because diagnosis of disease would include these potential factors.