Effectiveness of workers’ general health examination in Korea by health examination period and compliance: retrospective cohort study using nationwide data

“Health examination compliance” is a new variable reflecting the participation rate of periodic health examinations for several years. Using the consistency of annual participation as a health measure has never been attempted in past research. Previous research related to periodic health examinations in Korea usually analyzed single-year participation [9, 10, 15]. It is possible to assess not only single-year health effects of periodic health examinations but also its multiyear health effects when we use health examination compliance as an analysis variable. We identified that the compliant group has lower cumulative CCVD incidence than the non-compliant group (Table 4) consistently in both sexes and workplace policyholders. Moreover, we identified that the relative CCVD risk in the compliant group was statistically lower than the non-compliant group for male non-office workers (Table 5). Therefore, we suggest that health examination compliance positively affects CCVD prevention among workplace policyholders.

Further analysis compared the cumulative CCVD incidence between health examination periods (1-year vs. 2-year). We identified that the relative IHD risk of male office workers (2-year period) was statistically higher than that of male non-office workers (1-year period; Table 5). Although the analysis was limited in that the results did not show statistical significance or consistent results in both sexes and diseases, participants who received 1-year health examination showed better preventive effects than 2-year health examination for IHD in working age (40s–50s) men. There are several studies of periodic health examination effectiveness; however, a consensus remains lacking. One meta-analysis using only randomized controlled trials published from 1963 to 1999 (14 trials) revealed that periodic health examination has no beneficial effect on total mortality (RR?=?0.99, 95% CI: 0.95 to 1.03) and cardiovascular mortality (RR?=?1.03, 95% CI: 0.91 to 1.17) [1]. Another systematic review using 23 observational studies and 10 randomized controlled trials published from 1973 to 2004 also reported that periodic health examinations may be related with increased use of preventive medical service and reduced patient worry, but additional research data is needed to estimate its long term benefit [16]. Conversely, several investigations conducted in Japan showed that periodic health examinations had positive effects on total mortality (hazard ratio [HR]?=?0.74, 95% CI: 0.62 to 0.88 [6]; HR?=?0.70, 95% CI: 0.56 to 0.88 [7]; HR?=?0.83, 95% CI: 0.69 to 0.99 [8]) and cardiovascular disease mortality (HR?=?0.65, 95% CI: 0.44 to 0.95) [6] for men. The effectiveness of mass periodic health examinations is still controversial because of the difficulty of large clinical trials with periodic health examination [6]. There are no studies using health examination period as an independent variable to our knowledge despite the controversial results.

The relative CCVD risks between office and non-office workers showed subtle differences ranging from 0.93 to 1.03 (Tables 5 and 6). These results may be caused by a large study population (N?=?6,527,045) not the effect of health examinations. However, such a subtle difference might represent a meaningful result from the perspective of public health and prevention. Moreover, male non-office workers’ (1-year period) CCVD incidence gaps between the compliant and non-compliant groups were lower than office workers’ (2-year period; Table 4 for both IHD and ICVD). Thus, giving more participation chances for health examinations can narrow health effect gaps between subgroups classified by compliance.

Two perspectives are possible for why differences in effectiveness of health examinations were not identified in either ICVD or HCVD but were for IHD. One possibility is that ICVD and HCVD actually do not differ in effects by health examination period, unlike IHD. Another possibility is disease characteristics such as peak age and etiology of IHD and stroke (including both ICVD and HCVD). Although both IHD and ICVD have the same cause (arteriosclerosis), each disease shows differences in peak age and incidence as vessel ischemia from different organs (heart and brain) [17]. The peak age for IHD is the 50s–60s and 36% of IHD patients are under 45 years old [17, 18]. Conversely, ICVD occurs at a relatively older age than IHD. ICVD is rare before age 40. ICVD prevalence doubles every 10 years after age 55, so the highest prevalence (about 27%) is identified at over 80 years [19]. Therefore, a 7-year follow up period might be insufficient time for ICVD to detect effectiveness of health examinations, because ICVD occurs at relatively older ages than IHD (50s–60s). Further, HCVD’s pathophysiology itself is fundamentally different from IHD. Blood vessel rupture is the main cause of HCVD. In addition, the incidence of HCVD is 24.6 per 100,000 person-years; this value is one-tenth of IHD’s incidence (434 per 100,000 person-years) [17, 19]. HCVD’s relatively low incidence makes it difficult to draw statistically significant results, while IHD incidence analysis presented significant results.

Statistical significance was not consistent for women’s CCVD incidence by health examination period. Two perspectives are also possible for this result. One possibility is that health examination has no preventive effect for women; another possibility is the difference in disease epidemiology between the sexes. IHD occurs 10 to 20 years later in women than men and IHD occurrence in women is rare before menopause [20]. Women’s occurrence age for stroke is also later than men and the incidence rate is 33% lower than men [17, 21]. Etiology of ischemic stroke also differs between the sexes. Large vessel atherosclerotic stroke and associated coronary and peripheral artery diseases are more common in men and cardiac embolism-related stroke is more common in women [22]. Therefore, a 7-year follow up period might be insufficient time to detect IHD and stroke in women because women’s CCVD incidence is lower than men and occurrence is later than men. Further research with long-term follow up periods can determine differences of health examination effectiveness between the sexes.

Relative CCVD risks in regional policyholders with 2-year health examination periods were higher than non-office workers with 1-year periods; these results were statistically significant in both sexes. However, careful attention is needed in this analysis. Selection bias by the healthy worker effect (HWE) [23] is possible between non-office workers and regional policyholders. Healthy workers have greater potential to initiate their career in better companies and continue to work for longer than unhealthy workers [24]. Therefore, unhealthy workplace policyholders are likely to be retired from their workplace and to be regional policyholders. As a result, it is possible that health status differences occurred between the two selected groups; workplace policyholders may be relatively healthier than regional policyholders. Health examination compliance may be confounded by HWE in the same manner as well. Healthy workers are more likely to get health examinations stably than unhealthy workers due to better working environment [24]. Therefore, the variable, health examination compliance, may be confounded by healthy workers’ stable participation in health examination.

The 1-year period health examination had more preventive effects on ischemic heart disease than the 2-year period. It is necessary to identify the reasons for the difference in CCVD risk by period in further studies. Although there were several reports that described the mechanism of periodic health examination effectiveness, most were about its possible benefits. Participant’s poor health habits (e.g., smoking, alcohol drinking, irregular meals, no regular exercise) might be changed through medical counseling during periodic health examinations and periodic result notifications [1]. Identifying abnormal results (e.g., high blood pressure, glucose, cholesterol) in the early stages of disease also may lead to early intervention and health management [1]. It is also possible that WGHEs had a positive effect on medical accessibility by improving the delivery of medical intervention; the more health examination opportunities they have, the more chance for medical intervention they have [16].

This study has some specific limitations. The first is HWE, mentioned in detail above. The second is inaccuracy of benefit claim records and health examination results in NHIS. In this research, we utilized ICD 10 codes from NHIS benefit claim records instead of hospital medical records. Benefit claim records request treatment charges to NHIS. Over-rated diagnosis coding is possible to avoid cutbacks in benefit claim records [10] and diagnosis can be inaccurate in some instances [10, 25]. Third, the 7-year follow up period was insufficient to successfully evaluate WGHEs. This limitation can be one reason that statistically significant results were not consistently shown in ICVD, HCVD, and women, as previously mentioned. Finally, there are possible confounding factors between office and non-office workers. Although we stratified the two subgroups by sex, age, and national insurance type, several confounders in evaluation of health examinations remain. Socioeconomic status such as income, education, and residential district, and lifestyles such as smoking, alcohol consumption, and exercise are well known determinants of health examinations [26–29]. In addition, there are several reports showing that white collar workers have greater tendency to chronic diseases than blue collar workers. Several studies reported that the risk of chronic diseases such as dyslipidemia, hypertension and metabolic syndrome were higher in white collar than blue collar workers [30, 31]; this result was consistent even though they worked in the same work place [30]. The reasons chronic disease risks differ by job style were analyzed in several studies; sedentary work of white collar workers can influence worker’s health status negatively by physical inactivity [30, 32]. WGHE results include lifestyle questionnaire data (e.g., smoking, alcohol drinking, regular meals, regular exercise, etc.), but we did not analyze the questionnaire data in this study. Other significant variables such as working hours and level of sedentary work were not available from the questionnaire data.

Because of above limitations including possible confounding factors, our study is limited in comparing office and non-office workers. However, definitions classifying office and non-office workers in our database are somewhat different from those of blue and white collar workers in the above articles. According to The Occupational Health and Safety Act, sedentary workers who work in the same territory or are exposed to similar occupational environments to manual workers (e.g., sedentary workers whose offices adjoin their firm’s factory) are classified as non-office workers, although their job style is only paperwork. As a result, our data (e.g. office and non-office worker) did not reflect past job classification categories (e.g., white and blue collar or non-manual and manual worker). This is one possible reason our study lacked some above confounding factors between white and blue collar workers. It is necessary to consider the job classifications in this study to interpret the present results.

Despite these limitations, our research has several valuable findings. First, this study was performed with a real dataset provided by NHIS, not by simulation techniques. More than 6 million participants were the target of analysis using nationwide health examination data. Initial studies usually used simulation techniques; research using real datasets were not attempted [9, 15, 33]. Since then, Yoon et al. [10] analyzed a real dataset and Jee et al. [11] adjusted a cohort study design to NHIS nationwide data. Second, we presented a 5-year continuous participation rate, not just a single-year participation rate, by defining the new concept of “health examination compliance.” Several reports presented single-year participation rates [14]; however, our research presented multiyear participation rates for the first time. In 2006, 77% of workplace policyholders participated in NGHEs [12] but the compliant group who participated in all health examination chances for 5 years (2002 to 2006) was only 24% of the total cohort group (Table 3). Single- and continuous-year participation rates need to be included in further analysis of periodic health examinations.

However, the most important point is our evaluation of WGHE effectiveness by health examination period for the first time. WGHE periods in Korea differ between office (2-year) and non-office (1-year) workers. There are non-office workers who participate in health examinations every year and office workers who participate in it biennially; thus, it is possible that their health effects differ. Previous studies only analyzed health examination effectiveness by participation and did not consider health examination periods as a descriptive variable [9–11]. For the first time, we showed that 1-year WGHE periods in non-office workers had more significant prevention effect for IHD than 2-year periods in office workers among working age (40–50s) men. However, prevention of cardio-cerebrovascular disease can be partially explained by their occupational characteristics rather than their health examination period. Our study result should influence national health policy and support the necessity of further research. Additional studies adjusting variables such as lifestyle and socioeconomic status of participants and long-term follow-ups are needed based on this study.