Abstract
Background/Aim: Smoking has been reported to be a risk factor for a variety of diseases. In Japan, the Brief Job Stress Questionnaire (BJSQ) has been administered by the Ministry of Health, Labour and Welfare since December 2015, but few reports have focused on its relationship with smoking. We investigated the current situation of smokers among staff of Kagoshima University who underwent a medical check-up.
Patients and Methods: Of 2,478 people who underwent a medical check-up in May and June 2021, we targeted 2,237 workers who reported whether they smoked. We examined results of the medical check-up and BJSQ and the background of smokers (n=139, 6.2%). We compared smokers and non-smokers (n=2,098) using propensity score matching (PSM) for sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours at a 1:1 ratio.
Results: The results showed that white blood cell count (p=0.044), platelet count (p<0.001), glutamyl transferase (p=0.023), and triglyceride (p=0.027) were significantly higher among current smokers in comparison with current non-smokers. Smokers reported significantly more stress than non-smokers in terms of social support (p=0.027).
Conclusion: As a result of PSM, several blood test items related to non-communicable diseases (lifestyle-related diseases) showed high values in current smokers, and these individuals reported significantly more stress than non-smokers. According to the emphasis on group analysis in the Total Health Promotion Plan revised in 2020, our findings can be helpful in enhancing smoking cessation programs in the workplace.
Keywords: Brief Job Stress Questionnaire, group analysis, propensity score matching, smoking
Smoking has been reported to be a risk factor for a variety of cancers, such as lung and pharyngeal cancer, respiratory disease, heart disease, cerebrovascular disease, emphysema, periodontal disease, low birth weight infants, abortion, and sudden infant death syndrome. In Japan, the Third Stage Cancer Control Promotion Basic Plan of March 2018 set the goal of decreasing the proportion of adult smokers to 12% by 2022. However, that target had not been met as of 2019, with the percentage of smokers at 16.7%. Moreover, it is widely acknowledged among adult smokers that increases in smoking are often precipitated by stressful events (1). Far from acting as an aid for mood control, nicotine dependency seems to exacerbate stress. This is confirmed in the daily mood patterns described by smokers, with normal moods during smoking and worsening moods between cigarettes. Thus, the apparent relaxant effect of smoking only reflects the reversal of tension and irritability that develop during nicotine depletion (2). The Brief Job Stress Questionnaire (BJSQ) has been administered by the Ministry of Health, Labour and Welfare since December 2015, and various studies have been conducted among workers using the BJSQ (3); however, few reports have focused on its relationship with smoking. Additionally, tobacco or nicotine use is known to influence glucose and lipid homeostasis with important clinical implications (4), but few reports have so far been made on the relationship between lifestyle-related diseases and tobacco. We aimed to analyse the relationship between smoking status and data of the BJSQ or lifestyle-related diseases obtained from medical check-ups at Kagoshima University.
In Japan, the Total Health Promotion Plan was revised in 2020, emphasizing group analysis and recommending initiatives that are more suited to the actual conditions in each workplace. Propensity score matching (PSM) analysis is useful in population analysis. In this facility, non-smoking policies in the workplace have been established since October 2017, and no smoking during working hours has been enforced from January 2020. Along these lines, group analysis is recommended to encourage smokers to quit smoking. In the present study, we have examined whether we could obtain the significant information of the smokers among staff of Kagoshima University who underwent a medical check-up by using PSM, one of the procedures to reduce interferences of confounding factors.
Patients and Methods
Of 2,478 workers who underwent a medical check-up at Kagoshima University in May and June 2021, we targeted 2,237 people. According to the results of their health examination and BJSQ performed at the same time, we investigated the background of smokers (n=139, 6.2%) in comparison with non-smokers (n=2,098) using the PSM method. This was performed after matching participants’ background according to sex and age at a ratio of 1:1. The procedure was also performed by matching participants’ background according to sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours at a 1:1 ratio. Liver function and triglyceride levels are affected by drinking habits. Cholesterol and triglyceride levels also depend on whether medication for dyslipidaemia is being administered. Additionally, the amount of stress felt could differ depending on differences in overtime work.
The BJSQ has been used in previous studies, as well as in workplaces across Japan by the Ministry of Health, Labour and Welfare in guiding the Stress Check Program. Participants are required to answer questions on the BJSQ using a four-point Likert scale. The BJSQ comprises several related questions, and scores on the individual questions are summed to produce a total for each category. The total score for each category indicates high stress with a higher number of points (simple total score). The questionnaire content is broadly divided into three components: Job Stressors, Mental and Psychological Stress Reactions, and Social Support (1).
This was a case study that included a cross-sectional study at a single facility among workers who underwent a medical check-up in May and June 2021. During the study period, a total of 2,478 workers underwent a medical check-up; of these, 2,237 were included in the analysis. We excluded workers who did not undergo a medical examination and those whose smoking status was unknown because it was not reported (Figure 1).
Figure 1. Flow chart of data analysis in the comparison of clinical features between current smokers and current non-smokers undergoing a medical check-up, following propensity score matching for sex and age (Model 1) and for sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours (Model 2).
We conducted a comparison of items on the BJSQ together with findings of the medical check-up, physical findings, and results of blood tests among current smokers and current non-smokers. We used PSM for sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours. We also used PSM for sex and age to compare the clinical features of current smokers and non-smokers who underwent a medical check-up. The findings of the medical check-up were judged as follows: ‘no abnormality’, ‘slightly outside the standard range but does not interfere with daily life’, ‘requires follow-up’, ‘needs treatment’, ‘requires close inspection’, and ‘continue treatment’. Physical findings included body mass index, systolic and diastolic blood pressure, and abdominal circumference.
This study was conducted after receiving approval from the Ethics Committee at Kagoshima University Hospital (institutional review board. 210290).
Statistical analysis. Continuous variables are expressed as mean±standard deviation. Comparisons between continuous variables were analysed using the t-test, and discontinuous variables were analysed using Fisher’s exact test. A p-value of <0.05 was considered statistically significant.
To reduce the effect of treatment selection bias and potential confounding in this observational study, we adjusted for significant differences in the baseline characteristics of patients using PSM. Patients were matched for sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours (5). Before PSM, we confirmed multicollinearity among sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours.
The data were analysed using EZR version 1.54 (Saitama Medical Center, Jichi Medical University, Saitama, Japan) (6), which is a graphical user interface for R version 2.13.0 (The R Foundation for Statistical Computing, Vienna, Austria).
Results
Characteristics of current smokers. Background information regarding the characteristics of medical check-up examinees is shown in Table I. Among current smokers, 110 were men and 29 were women; the average participant age was 41.8 years. Among smokers, 51 did not drink, and 43 drank daily. Few participants had any comorbidities; those identified among examinees included hypertension (n=15, 10.8%), diabetes mellitus (n=3, 2.2%), hyperlipidaemia (n=8, 5.8%), renal dysfunction (n=5, 3.6%), hyperuricemia (n=6, 4.3%), liver disease (n=4, 2.9%), and anaemia (n=8, 5.8%). However, 47 examinees (33.8%) had body mass index 25 kg/m2 or higher, and 128 (92.1%) had some type of abnormality according to the guidance for comprehensive medical judgment. Among them, abnormality in peripheral blood test items (n=79, 56.8%), dyslipidaemia (n=73, 52.5%), and abnormal blood pressure (i.e., hypertension or hypotension, n=56, 40.3%) were most common. In this group, 17.3% (n=24) of participants reported having no intention to improve their lifestyle.
Table I. Background information of medical check-up examinees.
*Comprehensive judgement: 1, no abnormality in the range of this test; 2, slightly outside the standard range, but does not interfere with daily life; 3, attention needed in daily life and requires follow-up; 4, needs treatment; 5, needs close examination; 6, continue treatment. #Excludes unknown cohorts.
Comparison of health diagnosis factors regarding the presence or absence of smoking after propensity score matching (PSM). We compared smokers and non-smokers among medical check-up examinees regarding physical and mental items, after performing PSM (1:1) for clinical features using sex and age. There was no significant difference in terms of the presence or absence of metabolic syndrome and abnormalities in the medical check-up. However, in the current smoker group, there were significantly more abnormalities according to comprehensive judgment; here, ‘abnormal’ was defined as all judgments apart from ‘no abnormality’ and ‘slightly outside the standard range but does not interfere with daily life’. The statistical power of the other explanatory variables, which showed no significant difference, was generally low (0.067-0.51) (Table II).
Table II. Comparison of metabolic syndrome and abnormalities according to medical check-ups with respect to the presence or absence of smoking, after propensity score matching for sex and age (univariate analysis).
*Fisher’s exact test. **Objective variables include abnormal highs and abnormal lows. ‡‘Yes’ includes metabolic syndrome reserve. †‘No’ includes 2 (slightly outside the standard range but does not interfere with daily life) in judgement during medical examination. §Data exclude missing values. CI: Confidence interval.
In a 1:1 post-PSM study of clinical features with sex and age in Model 1, white blood cell count (p=0.030), platelet count (p=0.037), and triglyceride (p=0.012) were significantly increased among current smokers as compared with non-smokers. Smokers also reported significantly more stress than non-smokers in terms of social support, which included three subscales: social support from superiors, co-workers, and family members and/or friends (p=0.040). The statistical power of the other explanatory variables, which showed no significant difference, was generally low (0.027-0.39) (Table III). Before PSM, there was no multicollinearity between the independent variables sex and age; each variance inflation factor was 1.016 and 1.016, respectively).
Table III. Comparison of health diagnosis factors with respect to the presence or absence of smoking after propensity score matching (univariate analysis).
Model 1: propensity score matching for sex and age; Model 2: propensity score matching for sex, age, drinking habits, medication for dyslipidemia, and overtime working hours. *Mann-Whitney U-test. **Mean±standard deviation, median (minimum value-max value).
Alcohol consumption and hyperlipidaemia are thought to affect triglyceride levels in health check-ups. Moreover, overtime working hours are likely to differ depending on the type of job performed. Therefore, PSM was conducted for sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours (1:1 in Model 2). The results showed that white blood cell count (p=0.044), platelet count (p<0.001), glutamyl transferase (GT; p=0.023), and triglyceride (p=0.027) were significantly increased in current smokers compared with non-smokers. Smokers reported significantly more stress than non-smokers in terms of social support, which included the three subscales of superiors, co-workers, and family members and/or friends (p=0.027). The statistical power of the other explanatory variables, which showed no significant difference, was generally low (0.033-0.71) (Table III). Before PSM, there was no multicollinearity between the independent variables sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours; each variance inflation factor was 1.12, 1.16, 1.085, 1.081, and 1.11, respectively.
Discussion
We examined relevant information of smokers from among the staff of Kagoshima University who had undergone a medical check-up and used PSM to reduce interference from confounding factors. PSM analysis revealed the relationship between smoking status and several items obtained in health check-ups, such as white blood cell count, platelet count, GT, and triglycerides. Additionally, smokers reported being significantly more stressed than non-smokers in terms of social support, according to responses on the BJSQ.
Current cigarette smokers are known to have significantly higher white cell counts than never smokers. Increased levels of inflammatory markers may partly reflect elevated inflammatory cytokines, such as interleukin-6 and tumour necrosis factor-α, which are major regulators of the reactant plasma protein component of the inflammatory response. Interleukin-6 levels are increased in smokers (7-9). Previous reports also clearly show that smokers have higher platelet counts than non-smokers. Platelets in smokers have been reported to show increased aggregability and to be more prone to spontaneous aggregation (10).
A previous report showed that cigarette smoking exerted no effect on GT in teetotallers, but there was a statistically significant effect of smoking among participants with greater alcohol consumption intensity. Moreover, smokers had significantly higher GT levels in the presence of non-alcoholic fatty liver disease (11). In our study, GT was significantly higher in smokers than in non-smokers when matched using PSM according to age, sex, drinking habits, medication for dyslipidaemia, and overtime working hours.
Triglycerides are known to be elevated in smokers. Previous studies have reported that smokers have higher levels of serum triglycerides and blood glucose concentrations, and levels of high-density lipoprotein cholesterol are lower than those in non-smokers (12). Nicotine-induced sympathetic nerve activation has been suggested to promote catecholamine release, increase blood free fatty acids, and increase low-density lipoprotein production, as one mechanism by which smoking alters serum lipids (13). Indeed, our results showed a significant increase in triglycerides among smokers in post-PSM analysis for sex, age, drinking habits, dyslipidaemia medication, and overtime working hours, similar to those previous reports.
Stress has been found to be a significant risk factor for cigarette smoking (14). Azagma stated that among smokers, light tobacco users are the group most vulnerable to stress. A possible reason for the differential effects of job strain between light and heavy smokers may be varying degree of sensitization to tobacco use among these groups (15). Childs et al. (16) investigated the effects of acute psychosocial stress on cigarette craving, the subjective effects of smoking, and smoking behaviour in daily smokers. The authors found that stress significantly increased cigarette craving but did not increase smoking. This result is supported by previous evidence that acute psychosocial stress increases the desire to smoke (16). We found that among BJSQ items in the post-PSM analysis matched for sex, age, drinking habits, medication for dyslipidaemia, and overtime working hours, smokers reported being significantly more stressed than non-smokers in terms of receiving support from their bosses, colleagues at work, spouse, friends, and family.
In our research, we used PSM in the analyses, which was devised by Rosenbaum and Rubin in 1983 (17). PSM is among the most popular approaches to deal with causal inference in clinical and epidemiologic research (18). PSM is known as a statistical technique used to adjust for covariates and estimate causal effects in observational studies that are difficult to randomize and prone to various confounders. Moreover, PSM is a useful method for analysing cross-sectional data (19). When the target variable involves two factors, there are two methods of analysis: multivariate analysis and PSM. In multivariate analysis, the number of samples in the smaller objective variable must be at least 10 times the number of covariates, whereas in PSM there is no such condition; even if there are matching participants between groups, a comparison group can be formed (20). Whereas multivariate analysis requires consideration of multicollinearity, PSM does not have this requirement. The PSM method increases the ability to simply compare the effects of independent variables (21).
The present study had several limitations. First, this was a cross-sectional study at a single facility. We therefore performed 1:1 PSM with only 256 examinees for comparison of health diagnosis factors regarding the presence or absence of smoking with the items sex, age, drinking habits, dyslipidaemia medication, and overtime working hours; the statistical power of nearly all items was low. However, the statistical power was generally high for explanatory variables with significant differences (0.39-0.95). Second, we could not divide the non-smoker group into never smokers and past smokers. We also could not measure the amount of smoking, e.g., the number of cigarettes smoked per day or the Brinkman Index. This is because we did not ask about workers’ smoking history at the time of the health check-up; these are not included among the usual health check-up items. Third, we compared liver function abnormalities between smokers and non-smokers, but we could not confirm whether participants had fatty liver. Instead, medication for dyslipidaemia and the presence or absence of alcohol consumption were matched using PSM for smokers and non-smokers. As a result, smokers had higher triglyceride and GT levels than non-smokers. Finally, whether the results of this study will lead to future smoking cessation among study participants remains unclear. Therefore, assessment of whether the present findings led to an increase in smoking cessation among workers at this facility should be carried out.
In conclusion, we demonstrated that smokers were significantly more likely than non-smokers to have some type of health abnormality. Similar to the findings of previous reports, white blood cell count, platelet count, GT, and triglyceride were increased among smokers in our study. By making the results of this analysis available in the workplace as a group analysis, smokers may better understand the physical effects of smoking. Regarding stress, improvement of the work environment should be considered when providing guidance on smoking cessation.
Conflicts of Interest
The Authors declare that there are no conflicts of interest in relation to this study.
Authors’ Contributions
YK: study concept and design, analysis of data, and writing of the manuscript; MU and SK: analysis of data and approval of the manuscript; MH: study concept and design, critical review, and approval of the manuscript.
Acknowledgements
The Authors would like to thank Edanz (https://jp.edanz.com/ac) for English language editing.
References
- 1.Tashiro S, Kato K, Kitazawa M, Fujihara K, Kodama S, Tashiro M, Matsuda K, Otsuka M, Sato K, Sone H. Differences in occupational stress by smoking intensity and gender in cross-sectional study of 59 355 Japanese employees using the Brief Job Stress Questionnaire (BJSQ): the Niigata Wellness Study. BMJ Open. 2022;12(4):e055577. doi: 10.1136/bmjopen-2021-055577. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Parrott A. Does cigarette smoking cause stress. American Psychologist. 2017;54(10):817–820. doi: 10.1037/0003-066x.54.10.817. [DOI] [PubMed] [Google Scholar]
- 3.Okita S, Daitoku S, Abe M, Arimura E, Setoyama H, Koriyama C, Ushikai M, Kawaguchi H, Horiuchi M. Potential predictors of susceptibility to occupational stress in Japanese novice nurses - a pilot study. Environmental Health and Preventive Medicine. 2020;22(1):20. doi: 10.1186/s12199-017-0641-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Tweed J, Hsia S, Lutfy K, Friedman T. The endocrine effects of nicotine and cigarette smoke. Trends in Endocrinology & Metabolism. 2019;23(7):334–342. doi: 10.1016/j.tem.2012.03.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Joffe M, Rosenbaum P. Invited commentary: propensity scores. American Journal of Epidemiology. 2017;150(4):327–333. doi: 10.1093/oxfordjournals.aje.a010011. [DOI] [PubMed] [Google Scholar]
- 6.Kanda Y. Investigation of the freely available easy-to-use software ‘EZR’ for medical statistics. Bone Marrow Transplantation. 2021;48(3):452–458. doi: 10.1038/bmt.2012.244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Tappia P, Troughton K, Langley-Evans S, Grimble R. Cigarette smoking influences cytokine production and antioxidant defences. Clinical Science. 2021;88(4):485–489. doi: 10.1042/cs0880485. [DOI] [PubMed] [Google Scholar]
- 8.Lowe G. Why do smokers have higher plasma fibrinogen levels than non-smokers. Clin Sci (Lond) 2001;101(2):209–210. doi: 10.1042/cs1010209. [DOI] [PubMed] [Google Scholar]
- 9.Woodward M, Rumley A, Tunstall-Pedoe H, Lowe G. Associations of blood rheology and interleukin-6 with cardiovascular risk factors and prevalent cardiovascular disease. British Journal of Haematology. 2020;104(2):246–257. doi: 10.1046/j.1365-2141.1999.01158.x. [DOI] [PubMed] [Google Scholar]
- 10.Launay J, Del Pino M, Chironi G, Callebert J, Peoc’h K, Mégnien J, Mallet J, Simon A, Rendu F. Smoking induces long-lasting effects through a monoamine-oxidase epigenetic regulation. PLoS ONE. 2020;4(11):e7959. doi: 10.1371/journal.pone.0007959. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Oni ET, Figueredo V, Aneni E, Veladar E, McEvoy JW, Blaha MJ, Blumenthal RS, Conceicao RD, Carvalho JA, Santos RD. Non-alcoholic fatty liver disease modifies serum gamma-glutamyl transferase in cigarette smokers. J Clin Med Res. 2020;12(8):472. doi: 10.14740/jocmr3932. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Niemiec P, Nowak T, Iwanicki T, Gorczynska-Kosiorz S, Balcerzyk A, Krauze J, Grzeszczak W, Wiecha M, Zak I. The rs2516839 polymorphism of the USF1 gene may modulate serum triglyceride levels in response to cigarette smoking. International Journal of Molecular Sciences. 2020;16(12):13203–13216. doi: 10.3390/ijms160613203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Benowitz NL. Pharmacologic aspects of cigarette smoking and nicotine addiction. N Engl J Med. 1988;319(20):1318–1330. doi: 10.1056/NEJM198811173192005. [DOI] [PubMed] [Google Scholar]
- 14.Lawless M, Harrison K, Grandits G, Eberly L, Allen S. Perceived stress and smoking-related behaviors and symptomatology in male and female smokers. Addictive Behaviors. 2018;51:80–83. doi: 10.1016/j.addbeh.2015.07.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Azagba S, Sharaf M. The effect of job stress on smoking and alcohol consumption. Health Economics Review. 2019;1(1):15. doi: 10.1186/2191-1991-1-15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Childs E, De wit H. Effects of acute psychosocial stress on cigarette craving and smoking. Nicotine & Tobacco Research. 2019;12(4):449–453. doi: 10.1093/ntr/ntp214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Rosenbaum P, Rubin D. The central role of the propensity score in observational studies for causal effects. Biometrika. 2017;70(1):41–55. doi: 10.1093/biomet/70.1.41. [DOI] [Google Scholar]
- 18.Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399–424. doi: 10.1080/00273171.2011.568786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mason C, Sabariego C, Thắng ĐM, Weber J. Can propensity score matching be applied to cross-sectional data to evaluate community-based rehabilitation? Results of a survey implementing the who’s community-based rehabilitation indicators in Vietnam. BMJ open. 2019;9(1):e022544. doi: 10.1136/bmjopen-2018-022544. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Miles J, Shevlin M. Applying regression and correlation: A guide for students and researchers. Sage. 2001 [Google Scholar]
- 21.Shiba K, Kawahara T. Using propensity scores for causal inference: pitfalls and tips. Journal of Epidemiology. 2021;31(8):457–463. doi: 10.2188/jea.JE20210145. [DOI] [PMC free article] [PubMed] [Google Scholar]