Use of outpatient and inpatient health care services by occupation—a register study of employees in Oulu, Finland

We constructed two separate outcome variables, i.e., the use of outpatient and inpatient health care services. The use of outpatient health care was measured by the number of doctor or nurse visits in outpatient primary or specialized health care during year 2018. Finland has a three-sector healthcare system, which includes the public and private sectors as well as occupational health care. Public sector services are available to everyone, but waiting times can be long and there is usually a small customer fee. Occupational health care services are free of charge for their users and access is often fast, but these services are only available to the employed. Private sector services are available with shorter waiting times for those with the ability to pay. Being able to analyse the three sectors together is exceptional, because occupational health care is not covered in registers [8].

Data on public sector outpatient primary health care visits were obtained from the registers of the City of Oulu, while data on outpatient specialised health care were acquired from the Care Registers for Health Care (HILMO). Data on occupational health care visits were derived from the four largest occupational health care providers in Oulu, estimated to cover over 90% of the occupational health care customers [28]. Information on private sector doctor visits was obtained from the registers of Social Insurance Institution of Finland, which reimburses part of the fee charged by a private doctor. We included actual face-to-face visits to a doctor or a nurse; thus, other contact types such as phone calls or emails were not included. Also, visits related to oral health care were not included. As separate visits during the same day were inconsistently recorded in the registers of the different register holders, we approximated the number of visits by contact days with each health care provider. Finally, we calculated the total yearly number of visits (i.e. contact days) in public, occupational and private health care to form the first outcome.

Data on inpatient care were derived from the registers of the City of Oulu and the Care Registers for Health Care (HILMO), including admission dates from both primary and specialized care [29]. We measured inpatient health care as a dichotomous variable, classifying persons to those who were hospitalized at least once and those who were not hospitalized during year 2018, following the procedure used in previous studies [24, 25]. Inpatient care involving rehabilitation, oral health or childbirth was not included. Information on both outpatient and inpatient care could include also visits that took place outside the city of Oulu.

Occupation at the end of year 2017 was obtained from the registers of Statistics Finland. We classified the occupations using the Finnish version of the ISCO-08 classification [30], which consists of five hierarchical levels. We studied the differences between all occupational classes at the 1-digit level. Furthermore, to gain a more profound understanding of the occupational differences, we examined the 20 largest occupations at the 2-digit level. The classification of the 20 largest occupations was made separately for men and women. The 20 largest occupations covered 87% of employed men in the study population and 95% of women, respectively. The rest of the occupations were grouped together (results not shown).

We controlled for sociodemographic factors and two indicators of health status, which, according to earlier studies [31], may influence the use of health services. The sociodemographic factors were age, education, income and marital status. Age at the end of year 2017 was categorized into five-year age groups. Education was based on the highest completed degree or certificate. Those with primary level have completed up to 9 years of education, those with secondary education 11–12 years, and those with tertiary education at least 13 years. Income was measured as personal taxable income during the year 2017 (gross earned income, benefit income and income from capital combined, before taxes). It was classified into quartiles, separately for men and women. Marital status was classified into married, never married and separated/widowed. Income was retrieved from the registers of the Finnish Tax Administration, the other sociodemographic factors from Statistics Finland.

We measured health status through two variables available in the registers: the special reimbursement entitlements for medicine expenses and by sickness absence in 2017. The entitlements to special reimbursement for medicine expenses are available for medicines used in the treatment of severe and long-term diseases, and the information is often used as a proxy of chronic diseases [32]. Persons were classified as having at least one chronic disease if they had any entitlements in 2017. Sickness absence was measured by sickness allowance recipiency. It is available after completing a 10-day waiting period and requires a medical certificate. We added up the sickness allowance days in 2017 and categorized them into 0, 1–60 and 61–365 days. The information was obtained from the register of the Social Insurance Institution of Finland.

The distributions of the sociodemographic factors and measures of health status used in this study by occupational groups (1-digit level) and occupations (2-digit level) are presented in Table 1 for men and in Table 2 for women. Supplementary Table 1 shows the average number of outpatient care visits and the proportion of persons using inpatient care services according to the covariates. 

We used StataSE version 14 for analysis [33]. First, we calculated means for all outpatient care visits and the percentage of persons who had used inpatient care services by occupational groups (1-digit level) and by 20 more detailed occupations (2-digit level).

Second, we used multivariate modelling to study whether the differences between occupational groups and occupations could be explained by sociodemographic factors and health status. We used multivariate modelling to estimate the gross (unadjusted) and net (adjusted) associations of occupation with health care services use. For this purpose, several models were run, each including different sets of variables. In model 1, only occupation and age were included. Next, we added sociodemographic factors (education, income and marital status) and measures for health status (special reimbursement entitlements for medicines and sickness absence) separately in models 2 and 3. In model 4 we controlled for all covariates simultaneously.

The models were run for both outcomes and with both the broader and more detailed occupational classifications, separately for men and women. We first included the categorized occupation variables in the regression models and then used the contrast command in Stata in order to recalculate the results as comparisons against the grand mean, i.e. all employees in the model [34, 35]. This procedure was chosen as no occupational class would have been a natural choice for a comparison group. Furthermore, we used two overlapping occupational classifications and we wanted to use a common reference point for them. Using the grand mean as a reference serves this purpose well.

The average number of visits to doctors or nurses in outpatient care services was modelled with negative binomial regression models; thus, the results are shown as incidence rate ratios (IRR). The method can be used for over-dispersed count data, when the conditional variance exceeds the conditional mean [36, 37]. The minimum and maximum values of the number of outpatient visits in the occupational classes are shown in Supplementary table 2. The use of inpatient care services was modelled with logistic regression models, and the results are shown as odds ratios (OR). The statistical significance of the estimates was assessed with 95% confidence intervals. As explained above, the estimates and their confidence intervals are presented in comparison to all employed men or women in the data set, respectively.

The analyses were performed separately for men and women, because of the known differences in occupational distributions [38] and in the prevalence of health service use between Finnish men and women [31].

Male employees had an average of 3.8 outpatient care visits (during 2018) (Table 3). At the 1-digit level of occupational groups, managers and professionals had the least outpatient care visits (3.3 visits) and clerical and support workers the most visits (4.4 visits). Almost all occupational groups differed from the average in either direction. In general, the number of visits was lower than average in the upper groups and higher in the lower groups. Controlling for socio-demographic factors and health status only had a small effect on the estimates. Adjusting for health status among health care professionals and personal care workers had the greatest impact.

Model 4 shows the incidence rate ratios (IRR) compared to all male employees, using occupational groups and occupations separately as independent variables, when all socio-demographic factors and health status were adjusted for. Managers and professionals had a statistically significantly lower number of visits than male employees on average. Among technicians and associate professionals, health associate professionals had a clearly higher IRR (1.16) than other occupations in this group. Clerical support workers had the highest number of visits at the 1-digit level. Among service and sales workers, personal care workers and protective service workers had the highest number of visits. Skilled agricultural, forestry and fishery workers did not differ from all employees. At the 1-digit level, craft and related trades workers and plant and machine operators and assemblers had a higher IRR, but there were some differences at the 2-digit level: the number of visits was higher especially among metal, machinery and related trades workers and stationary plant and machine operators, and lower among drivers and mobile plant operators. Elementary occupations did not differ from all employees, but within this occupational group, labourers in mining, construction, manufacturing and transport had a higher IRR (1.22).

6.8% of male employees had been in inpatient care. Between occupational groups, the proportion varied between 5.3% in elementary occupations and 8.2% among plant and machine operators and assemblers. Controlling for covariates attenuated the differences in some occupations. Controlling for health status had a greater effect, especially among health associate professionals and personal care workers. In the fully adjusted model at the 1-digit level, plant and machine operators and assemblers had higher odds (OR = 1.24) compared to all male employees. At the 2-digit level, the odds were higher than average among health associate professionals and stationary plant and machine operators.

On average, female employees had 5.7 outpatient health care visits during 2018 (Table 4). In the 1-digit level occupational groups, the number of visits varied between 4.0 (skilled agricultural, forestry and fishery workers) and 6.4 (elementary occupations). The effect of adjustments for sociodemographic factors and health status varied between occupational classes and occupations. In most occupations, sociodemographic factors had a greater effect than health status. The overall biggest effects were found among professional occupations, science and engineering associate professionals, personal care workers, and cleaners and helpers. The results from the fully adjusted model were quite similar to the results found among men. The biggest difference was the less frequent use of health care among craft and related trades workers. Other notable differences compared to men were the more common use among science and engineering associate professionals and sales workers and the less common use among personal service workers. Among clerical support workers, customer service clerks had the highest number of visits.

Among female employees, 8.4% had been in inpatient care. Occupational groups at the 1-digit level ranged between 6.8% (craft and related trades workers) and 11.6% (plant and machine operators and assemblers). Again, the results were very similar to those seen among men. The effects of controlling for sociodemographic factors and health status were for the most part small. After adjustments, the odds for health associate professionals and plant and machine operators and assemblers and for more detailed stationary plant and machine operators remained high.

Occupational groups differed in their use of outpatient and inpatient health care services among working-age employees. However, there were also differences within occupational groups, examined at a more detailed level of occupations. Controlling for covariates explained only part of the differences, more among women than among men. After adjusting for sociodemographic factors and health status, the use of outpatient care services was most frequent among clerical support workers. The biggest difference between men and women was observed among craft and related trades workers, among whom the use of outpatient care services was high among men but low among women. The use of inpatient care services was highest among plant and machine operators and assemblers. At a more detailed level, both adjusted outpatient and inpatient care were more common among health associate professionals and stationary plant and machine operators, among men as well as among women. In addition, outpatient care was common among such groups as business and administration associate professionals, male personal care workers and female sales workers.

Taking into account both primary and specialized care, the use of outpatient care services seemed to be more common in occupations lower in the socio-economic hierarchy, especially among men. However, there was also variation within the 2-digit level occupations in the same occupational group. Earlier studies on outpatient health care services use by occupational class have mostly studied primary and specialized care separately. These studies have found, on one hand, contradictory results concerning the use of primary care and, on the other hand, more frequent use of specialized care in the higher socio-economic classes. Differences in results concerning primary care may be explained by differences in, for example, study populations, occupational class classifications or health care systems. To our knowledge, there are no prior studies on outpatient health care services use at the level of more detailed occupations.

Concerning inpatient care, our study showed that the use of inpatient care services was most common among stationary plant and machine operators as in a previous study [24]. We also found that health associate professionals had a higher use of inpatient care services, a result that was not obtained in a study by Kaila-Kangas et al. [24] but was instead supported by a study by Varje et al. [25]. However, earlier results [24] indicating a more common use of inpatient care services among male labourers in mining, construction, manufacturing and transport and among female craft and related trades workers and cleaners and helpers, were not supported in our study. In recent years, the number of outpatient visits in specialist care has increased while the number of inpatient care periods has decreased. Inpatient days have also decreased in primary care [39]. This is of significance when comparing the results of this study with previous studies, but does not give rise to concern as to the comparison between the different occupations.

In outpatient care, adjustment for socio-economic factors by and large had a greater effect among women on the estimates than did health status, whereas among men the effects of all adjustments were very small. In inpatient care, adjusting for health status had a greater effect on the estimates concerning the occupational groups. Sociodemographic factors may be assumed to be of lesser importance in terms of inpatient care, since data on hospitalizations include only the most severe cases of illness and exclude less serious cases not requiring hospitalization. Outpatient care visits indicate milder morbidity – accordingly, non-health-related factors can be expected to play a greater role in the use of these services. At the 2-digit level, especially among health associate professionals and male personal care workers, the adjustment of health status had greater effect to estimates than that of sociodemographic factors.

Higher socio-economic groups may be better informed about health and health care and more able to navigate the health care system and communicate with health care professionals [40]. This may increase health care services use in managers and professionals. However, according to our results, their use of health care services was at an average or below average level. A better ability to identify diseases that require care may explain the high use of services among health associate professionals and male personal care workers. Furthermore, high interests in taking care of one’s health and positive attitudes towards health services may explain some differences. Women use health care more frequently than men, and men working in female dominated sectors such as health care may adopt that habit [13]. Respectively, working in a male dominated sector can explain the lower use among female craft and related trades workers and science and engineering professionals.

Differences in outpatient care visits may also be related to sickness certification requirements. In manual occupations, certification is typically required earlier, sometimes even from the first day of illness, than in non-manual occupations [41]. A stricter sickness certificate requirement may increase their number of outpatient visits compared to non-manual workers. In addition, the frequency of mandatory health check-ups may affect differences between occupations. The check-ups are mandatory e.g. for those working in jobs which pose a particular risk of illness or accident [42], and such occupations are mostly manual [10‐12].

The Finnish three-sector system is comprised of occupational, private and public health care. First, occupational health care plays a major role in employee health care in Finland [39]. It is the only sector that is free of charge for the patient at the point of delivery and provides easy access to outpatient care. However, the availability and coverage of occupational health care varies across organizations according to the occupational health care contract [14]. There are shortcomings in occupational health care in small enterprises, which are typical in construction, land transport, accommodation and food service activities [15]. This may partly explain the lower use of outpatient care services among female personal service workers and male drivers and mobile plant operators. Second, private sector services are accessible with high co-payments but access to specialist services is easy. Use of private health care services is more common among upper non-manual employees [16] and may thus increase the total number of outpatient care visits in this group. Third, those without access to occupational health care and a low ability to pay turn to the public sector, where patients’ co-payments are low, but access may be difficult and waiting times long [39]. Further studies are needed to better understand the effect of the Finnish three-sector health care system on occupational differences in health care services use.

Both outpatient and inpatient health care was common among male and female health associate professionals and stationary plant and machine operators. For interpretation of the findings, it may be instructive to compare our results to those on mortality [2‐4]. The high use of health care services might indicate an overuse of health services; however, it could also indicate genuine demand not fully captured by the health status variables used in our study. Health associate professionals may be better at disease identification and may have a positive attitude towards health care and knowledge of how to use it. Their lower mortality [3] may be another indicator of the same phenomenon. However, they may have some work-related risks such as threat of violence [43], which increases the need of health care services. Stationary plant and machine operators have high risks for chemical and biological hazards [10], occupational diseases [11], work accidents [12] and mortality [2, 4]. They seem to underuse rather than overuse health care services. The severity of symptoms and the influence on working ability can vary by occupation, even if the symptoms are the same. For example, physical disability makes it more difficult to work as a stationary plant operator than as a clerical support worker. The explanations of better ability, attitude and knowledge may also apply to male personal care workers, while the explanations suggesting higher occupational risks may also pertain to male building and related trades workers, to metal, machinery and related trades workers, and to labourers in mining, construction, manufacturing and transport, who have also found to have high mortality [4]. Use of outpatient care services was common also in some other occupations. Female customer service workers have high mortality [4]. Male protective services workers and female sales workers both had high risk of disability retirement but not mortality [3]. The frequent use of outpatient care services among male and female business and administration associate professionals needs further study. The connection between health care use and mortality is thus less clear.

Managers and professionals used health care services less than the average employee. Lower use of health care services might indicate not only better health and a lower need of care but also an underuse of care. However, due to better access to occupational and private health care, better self-rated health [5], a higher probability to work sick and a smaller need of sickness absence certification [41], the underuse does not seem probable. Underuse is more likely among unskilled workers [44]. Among manual workers, male drivers and mobile plant operators, female personal service workers and female craft and related trades workers probably have low access to occupational health care [15] and are thus dependent on the less accessible public health care.

The key strength of our study was the ability to use individual-level register-based data on all inhabitants in one city. Our study included both outpatient and inpatient care. We were able to cover all three sectors of Finnish outpatient health care: public, private and occupational health care. The last of these is rarely covered in registers [8], which means that previous studies have usually not been able to use register-based data on occupational health care. Because of our access to register sources, we were also able to use more precise information on occupation than many other studies, and adjust for some important correlates of health, which are likely to confound the association between occupation and use of health care services. The register data has practically no missing data and do not suffer from a self-report bias. Thus, they are held to be more reliable than self-reported data on health care services use and socio-economic indicators.

There were also some limitations. The number of health care contacts had to be calculated as separate visits days in outpatient care, because it was not possible to measure each separate visit reliably if there was more than one visit during a single day. Furthermore, as there are few health-related covariates in register data, we were able to utilize somewhat restricted proxy variables for health status. Our data did not include, for example, information on self-reported health or health behaviours, which could affect the use of health services in different occupational groups. The data did not include the causes for visits so we could not estimate how many of the visits were due to a health check-up or common cold and which visits may have been related to a more serious illness or injury. We could not take into account all demand factors such as earlier experience, knowledge, attitudes and preferences, and the supply side, including access to and coverage of occupational health care. Further, we could not take health selection into account. For example, an individual may have previously been in a more hazardous occupation but has switched occupations due to illness. Overall, as our register-based data set included only a restricted set of control variables related to socio-economic status and health, we could not totally explain the occupational-class differences in outpatient and inpatient health care use with our data.

Another limitation is that our data were derived from one city only. The employment rate and the average educational level in Oulu were higher than across all of Finland in year 2018. The share of employees working in industry was somewhat lower and the share of those working in the public sector higher than the country average [27]. The supply of occupational and private sector health care varies by region [39]. Since Oulu is a large city, its population has more opportunities to use occupational health care and private sector health services than inhabitants of more sparsely populated areas. Even though there are some differences in the population composition as compared to the country as a whole, the health care system is fundamentally similar in the whole country. Thus, the results are estimated to be generalizable to the whole of Finland and may provide insights to other countries with roughly similar systems.