This is a population-based, cross-sectional study using Korea National Health and Nutrition Examination Survey (KNHANES) data from 2016 to 2020.

KNHANES uses a nationally representative sample based on a complex, stratified, multistage cluster sampling method that includes geographical area, gender, and age to provide a representation of the Korean population that is freely available to the public. KNHANES uses a series of cross-sectional national surveys that have been conducted by the Korean Centers for Disease Control and Prevention. These consist of a health survey, a health examination, and a nutrition survey [26].

The participants of this study were shift workers aged 19 years and older. Shift workers were defined as those who work at night or are nonday workers. This included both evening work, night work, day/night regular shift work, and irregular shift work. The exclusion criteria were as follows: (1) those who were younger than 19 years, (2) those who were nonworkers, and (3) those with missing data on any of the chronic diseases. The detailed sample selection flow is shown in Multimedia Appendix 1, Figure S1.

This study was exempted from ethical review and approval by the institutional review board of Chung-Ang University, HCH’s institution (1041078-20230518-HR-139), as this study was a secondary analysis of preexisting data. The primary data (from KNHANES) were anonymous, and informed consent was obtained prior to the data collection.

Data were analyzed using Stata (version 16.1; StataCorp LLC) and JASP (version 0.17.1; Eric-Jan Wagenmakers). Additionally, the KNHANES analytic guidelines were followed to adjust the data for complex sampling designs to estimate the population-level statistics for Korea. We conducted a complete case analysis without any imputation following the analytic guidelines [26]. All statistical tests were based on point estimation using a 2-sided P value (<.05) and interval estimation using 95% CIs. Specific analyses was conducted as follows: first, descriptive statistics were used, including weighted N, N, weighted percentage, weighted mean, and SE to present the population characteristics and their multimorbidity characteristics. Second, univariate analyses using the survey-corrected Rao-Scott χ² test and adjusted Wald test were conducted to reveal the differing characteristics between the multimorbidity and the nonmultimorbidity shift worker groups. Third, survey-corrected logistic regression analyses were used to identify factors influencing multimorbidity among shift workers. In the final model, variables that were statistically significant in the univariate analysis were included. We additionally conducted a subgroup analysis based on age, using 50 years as the threshold. Lastly, network analysis was used for exploring the patterns of multimorbidity clusters among shift workers. The network model was graphically represented by nodes (circles representing each morbidity) and edges (lines connecting the nodes). The structural importance of multimorbidity patterns was analyzed using node centrality measures, including closeness, betweenness, strength, and expected influence. Clustering and naming of subgroups within multimorbidity groups were determined through discussion among the researchers based on prior studies [7,34].

The final selected shift worker sample size was 1704 and the weighted population size was 2,697,228. The weighted percentage of male workers (56.7%) was slightly more than female workers, and the mean age of the participants was 41.93 (SE 0.42; SD 13.74) years, with an age range of 19 to 80 years (Table 1). The majority of the participants had household incomes in the third (n=538, 32.9%) and fourth (n=561, 33.4%) quartiles. As for educational attainment levels, 87.2% (n=1419) of participants had a high school education or higher. Additionally, there were more married (n=1209, 64.6%) than single participants. Participants were divided by shift work type: 920 (53.4%) in evening work, 191 (12.2%) in night work, 454 (27.3%) in regular shift work, and 139 (7.1%) in irregular shift work. The proportion of nonregular workers was high at 63.7% (n=821), and the average number of working hours per week was 38.01 (SD 17.67).

The mean BMI was 24.20 (SE 0.11; SD 3.47) kg/m², and overweight and obese participants (those with a BMI of 23.0 kg/m² or higher) accounted for 60.3% of the total. Nonsmokers were the most common at 51.9% (n=954), and low-risk drinking was engaged in by the majority of the workers at 67.2% (n=969). The proportion of workers engaged in aerobic physical activity was 50.05% (n=808), which was similar to the proportion of participants not engaged in aerobic physical activity. The average sleep time was 7.34 (SD 1.33) hours, and workers who showed a poor sleep pattern accounted for the majority at 55.9% (n=940).

Table 2 shows the multimorbidity characteristics of the shift workers. Hypertension (n=287, 14.2%) was the most common of the 13 chronic conditions, followed by dyslipidemia (n=235, 11%), diabetes (n=107, 5.3%), and arthritis (n=126, 5.1%). The multimorbidity prevalence in shift workers was found to be 13.7% (n=302) and the number of chronic conditions ranged from 0 to 6.

The differences between the multimorbidity group and the nonmultimorbidity group are presented in Table 3. The average age of the multimorbidity group was 57.21 (SE 0.69; SD 10.47) years, which was statistically significantly higher than that of the nonmultimorbidity group (mean 39.51, SE 0.42, SD 12.69 years). The multimorbidity group had lower household income and educational level, a lower number of unmarried people, higher irregular work level, higher BMI, fewer current smokers, less physical activity, and a higher proportion of participants with poor sleep compared to the nonmultimorbidity group, with all differences being statistically significant.

Table 4 presents the factors affecting multimorbidity in shift workers. Examining the relationship between each influencing factor and the 2 variables (crude odds ratio [OR]) of multimorbidity showed that age, household income, education, marital status, regularity of work, BMI, smoking, physical activity, and poor sleep quality were all factors influencing multimorbidity. In the final adjusted logistic model, factors affecting multimorbidity were age (OR 1.106, 95% CI 1.069-1.144; P<.001) third-quartile (OR 0.302, 95% CI 0.119-0.768; P=.01) vs first-quartile household income, fourth-quartile (OR 0.366, 95% CI 0.142-0.942; P=.04) vs first-quartile household income, nonregular work (OR 1.804, 95% CI 1.008-3.228; P=.047), and obese BMI (OR 2.152, 95% CI 1.155-4.010; P=.02) vs normal BMI. That is, when the age increased by 1 year, the risk of multimorbidity increased by 1.11 times, and the risk of multimorbidity in the third-quartile and fourth-quartile household income groups decreased by 69.8% and 63.4%, respectively, compared to the first-quartile household income group. The risk of multimorbidity in the nonregular work group increased by 1.8 times compared with the regular work group, and the risk of multimorbidity in the obese group increased by 2.15 times compared to the normal BMI group. Subgroup analysis showed that age, regular work, and obese BMI were associated with multimorbidity in the subgroup younger than 50 years, while household income, education, marital status, and obese BMI were associated in the subgroup aged 50 years and older (Multimedia Appendix 2, Table S1).

As presented in Figure 1, the network analysis revealed that chronic diseases were clustered into three groups: (1) cardiometabolic multimorbidity (hypertension, dyslipidemia, diabetes, CHD, and stroke), (2) musculoskeletal multimorbidity (arthritis and osteoporosis), and (3) unclassified diseases (depression, CLD, thyroid disease, asthma, cancer, and CKD). A centrality plot of multimorbidity patterns in shift workers shows hypertension, dyslipidemia, arthritis, diabetes, and osteoporosis by strength and expected influence in order (Multimedia Appendix 3, Figure S2).

This study aimed to identify patterns of multimorbidity and examine the factors associated with multimorbidity among shift workers in Korea. Among the 1704 shift workers included in the study, approximately 14% had multimorbidity. A recent meta-analysis of multimorbidity in the worldwide adult population found that there was a prevalence of approximately 37% [15]. The difference in prevalence could be due to several factors. First, the mean age of our sample was relatively low at 41.9 (SD 13.7) years, compared to 56.9 years in the meta-analysis. It has been recognized that older age is associated with an increased number of chronic diseases [35]. Furthermore, women are more prone to developing chronic diseases than men [36,37], and a higher proportion of our sample was male.

It is recognized that shift work is associated with adverse health outcomes. Previous research reported that shift work increases the risk of cardiovascular incidence and mortality, cancer, and stroke [1,7,38-40]. Approximately 14% of the Korean population is involved in shift work, and although some shift work is inevitable and may provide economic benefits, policies are needed to modify and manage shift schedules. Socioeconomic and behavioral factors are also known to increase the risk for multimorbidity. Consistent with previous research, we found that shift workers in the low-income and obese groups had increased risks of developing multimorbidity. The relationship between low income and the development of chronic diseases and multimorbidity may be explained by less physical activity and lower fruit and vegetable consumption in the low-income group [41,42]. Furthermore, obesity is a major risk factor associated with developing several chronic diseases, including diabetes, heart disease, asthma, arthritis, and depression [43]. We found that a high proportion of Korean workers are involved in nonregular work (63.7%), and having regular work was a significant factor influencing multimorbidity. Many nonregular workers are paid low wages (less than US $10 per hour), are employed on 2-year fixed-term contracts, and may be required to extend their employment or ask to terminate the employment every 2 years. Those who work less than a certain number of hours per week are not covered by social insurance and are not entitled to weekly holidays or paid annual leave, which could be important factors contributing to poor health outcomes among nonregular workers in Korea [44]. Labor laws may require reform so that nonregular workers are guaranteed stability in their jobs, social insurance coverage, and proper wages to enable a minimum standard of living. However, longitudinal and cohort studies are needed to assess and analyze the relationships between shift work, socioeconomic and behavioral factors, and multimorbidity. Further, the subgroup analysis revealed that education and marital status were significant factors influencing multimorbidity in shift workers aged 50 years and older. Consistent with previous research, the risk of multimorbidity increased when workers were less educated and when they were single [45,46]. A deeper understanding of these relationships may require further analysis to identify the subgroups at greatest risk for multimorbidity and develop targeted interventions, as there could be interactions between socioeconomic and demographic factors.

The most common chronic disease in this study was hypertension, followed by dyslipidemia, diabetes, and arthritis. This is similar to the findings of a previous meta-analysis of multimorbidity in the United States, United Kingdom, and Spain [20]. Additionally, we performed a network analysis and found 2 distinct groups: hypertension, dyslipidemia, diabetes, CHD, and stroke clustered as one group; arthritis and osteoporosis clustered as another group. Previous research found that the coexistence of hypertension and at least 1 other comorbidity was most common among patients with multimorbidity [47]. Similar to our findings, Yang et al [7] found that shift workers with hypertension in the United Kingdom were more prone to cardiometabolic multimorbidity, including diabetes, coronary artery disease, and stroke. However, the only known multimorbidity cluster is cardiovascular multimorbidity, and the evidence on multimorbidity in other chronic diseases and their clusters in shift workers remains limited. In the general population, cardiovascular diseases and metabolic diseases tend to cluster together, and osteoarthritis tended to cluster together in an Australian sample, a finding that is similar to ours [48]. Some chronic diseases are more likely to cluster together, as seen in our study and the studies in the United Kingdom and Australia. Some chronic diseases are more likely to cluster together as they may share similar pathophysiological pathways [24]. Also, shift work is known to disrupt circadian rhythms and sleep and is associated with overweight and with blood glucose levels, which are important factors for developing cardiometabolic diseases [49,50]. Although it is controversial, some have suggested that patients with arthritis may have greater bone loss [51]. Considering the relationship between arthritis and osteoporosis and that the incidence of both conditions increases with age [52,53], and that the mean age of our participants with multimorbidity was 57 (SD 10.47) years, the co-occurrence of these chronic disease may be understandable. Identifying these disease clusters can be important, as we can target and tailor interventions for specific groups. However, caution is needed when interpreting and comparing clusters of multimorbidity or patterns across studies, as researchers use different statistical methods to analysis clusters or patterns and may include a different list of chronic diseases in their analyses. Further research is needed to better understand and assess multimorbidity clusters and their trajectories and patterns over time among shift workers using unified and valid analytic methods with the most prevalent chronic diseases in shift workers.

Most chronic care models and guidelines focus on the treatment and management of individual chronic diseases [54]. Multimorbidity requires more complex care that prioritizes what is most important for each patient. Several types of interventions for multimorbidity have been implemented in the past, but the evidence to deliver specific interventions remains limited. Furthermore, interventions specific for shift workers with multimorbidity are relatively unknown, despite shift workers being at greater risk for developing multimorbidity than nonshift workers. Nevertheless, occupational or organization-based interventions may benefit shift workers. These potentially include policy development regarding shift schedules, early screening diagnostic interventions, and diet and physical activity programs. However, additional research remains necessary to confirm the multimorbidity clusters in shift workers and to confirm the differences between socioeconomic and behavioral factors to enable the development of specific intervention programs for each cluster group.

To our knowledge, this study is the first to assess multimorbidity and its associated factors in Korean shift workers. We analyzed a 2016-2020 series of national population-based study data (KNHANES) using network analysis to reveal multimorbidity patterns. However, this study has some limitations. First, KNHANES is a cross-sectional study; thus, we cannot assume temporal relationships between variables. We also could not assess the long-term patterns of multimorbidity development and the effect of socioeconomic behavioral factors on multimorbidity. Therefore, we recommend longitudinal studies on multimorbidity among shift workers in the future. Second, the data were based on self-reports from participants. Thus, recall bias cannot be excluded. Some objective data, for example, medical records to identify multimorbidity, may be needed in a future study. Third, the specific patterns of shift work were not specified in the current data set. For example, we do not know if the shift workers had rotating shifts, only night shifts, or how many night or evening shifts were included in their work schedules. Further studies should specify the shift work schedules; a subgroup analysis may be beneficial. Fourth, we also recommend subgroup analysis of different work sectors, since working conditions and shift schedules may differ. For example, the health care and manufacturing industries may use more complex shift schedules and require more specific interventions based on subgroup analysis. Fifth, network analysis provides only a graphical presentation of multimorbidity patterns and does not allow for statistical analyses, such as regression, to identify factors associated with each multimorbidity pattern. Future studies may consider using other statistical analyses, such as latent class analysis, if researchers are interested in exploring factors associated with multimorbidity patterns in shift workers. Finally, the study results may have excluded some confounding variables, such as family support and use of health care, that may have influenced multimorbidity.

Multimorbidity is a crucial factor influencing premature death, poor health and quality of life, and use of health care. Our findings indicate that approximately 14% of Korean shift workers have multimorbidity and that several socioeconomic and behavioral factors are associated with multimorbidity. This suggests that policy development regarding work schedule modification is necessary. Furthermore, screening and tailored intervention programs at the organizational level may benefit efforts to prevent and monitor multimorbidity among shift workers. However, we also suggest a future longitudinal study to assess and confirm multimorbidity patterns among shift workers.