1. Introduction
Cardiorespiratory fitness (CRF) has long been reported as a strong indicator of health, non-communicable diseases, and life expectancy [
1]. Two of the main determinants of the CRF level are current moderate-to-vigorous physical activity (MVPA) and heredity [
2,
3,
4,
5]. Of these two, only the MVPA level is modifiable, and it is the main alternative to maintain or increase CRF.
Low socioeconomic status, often assessed using education or income, has been linked to lower levels of both MVPA and CRF [
6,
7]. Along with the higher prevalence of daily smoking, poor diet, and obesity, the risk of common diseases such as cardiovascular diseases and type-2 diabetes is higher in those with low compared with high socioeconomic status [
8,
9].
We recently compared lifestyle-related risk indicators between different occupational groups, where blue-collar and low-skilled occupations had a higher prevalence of obesity, daily smoking, and poor diet [
10]. However, interestingly, the proportion of participants with low CRF differed between occupations, despite having a similar level of education. This means that the assessment of CRF in relation to occupation, compared with other measures of socioeconomics, can provide additional important aspects in relation to health and disease risk, especially as a certain level of CRF is required for the job tasks of several occupations [
11]. On the contrary, in more sedentary occupations (office workers), low CRF has been associated with a lower cognitive performance and higher sickness absenteeism [
12,
13].
While the importance of CRF for health and disease risk is well established, less is known regarding the trends in CRF over the recent decades in the general population, as well as in sub-groups of the population. Some reports indicate a negative trend in CRF in the general population [
14,
15], with a more pronounced decrease in men, younger age groups, and in those with low education. However, the trends in relation to different occupational groups are missing. Such information would be highly relevant in order to identify and target interventions towards occupations that need it the most.
The aim of this study was to examine the secular trends in CRF over the last 20 years in different occupational groups of the Swedish working population, and to forecast possible future trends.
2. Methods
Data were taken from Health Profile Institute’s database (assessed on 1 December 2020) that contained data from Health Profile Assessments (HPA) that were carried out in health services all around Sweden since the 1970s. HPA is an interdisciplinary method consisting of a questionnaire regarding lifestyle and health experiences, measurements of anthropometrics and blood pressure, a submaximal cycle test for estimation of maximal oxygen consumption (VO
2max), and dialogue with a Health Profile Coach. Participation is offered to all employees in a company or organization connected to occupational or health services and is free of charge. The HPA method has been developed and standardized throughout the years by the HPI Health Profile Institute, which is also responsible for the database. HPAs have been available in the database from 1980. However, for power reasons, we included data from January 2001 to December 2020 in the present analyses, aggregated into five-year periods (2001–2005, 2006–2010, 2011–2015, and 2016–2020). During this period, a total of 925,725 HPAs were carried out, of which 537,034 were first time HPAs and 388,691 were repeated HPAs in the same individuals (see
Figure A1 for flow chart). In participants with repeated HPAs, one valid CRF test per individual was included in each five-year period. Inclusion criteria for the present analyses were age between 18 and 65 years, data on height and body weight, educational level, and occupation. The final filtered sample included 516,122 tests, of which 413,183 were first time tests and 102,939 were repeated tests. Participants provided informed consent prior to data collection. The study was approved by the ethics board at the Stockholm Ethics Review Board (Dnr 2015/1864-31/2 and 2016/9-32) and adhered to the Declaration of Helsinki.
2.1. Assessment of Estimated VO2max
The estimated VO
2max was obtained from the Åstrand submaximal cycle ergometer test [
16]. The participants were requested to refrain from vigorous activity the day before the test, consuming a heavy meal 3 h before, smoking/snuff use 1 h before the test, and avoiding stressing during the test. The test was conducted on a calibrated cycle ergometer at an individually adapted submaximal work rate for 6 min. VO
2max was estimated from the achieved steady state pulse using a sex specific nomogram, with corresponding age-correction factors, and was expressed as the absolute (L/min) and relative (mL/min/kg) VO
2max [
16]. A low CRF was defined as <32 mL/min/kg, as it has been reported as an inflection point of increased risk of mortality [
17]. The Åstrand test has shown a low variation in the mean difference between the estimated and directly measured VO
2max (mean difference −0.07 L/min 95% CI −0.21 to 0.06) [
18].
2.2. Occupational Groups
Out of the 516,122 individual occupational codes, 248,467 were reported by the participants at the HPA, and they were coded according to the Swedish Standard Classification of Occupation (SSYK) [
19]. The remaining SSYK codes were derived from the national register data (Statistics Sweden: www.scb.se). These were based on SSYK reported by the employer and were matched for the year of the HPA performed. Each occupation was derived from a four-digit SSYK-code; the first digit defined the major group, the second digit defined the sub-major group, the third digit defined the minor group, and the fourth digit defined the unit group. Here, we present data on the major and sub-major levels. The major level is presented as one digit, while the sub-major is presented as a digit and a decimal digit. The occupational groups with corresponding SSYK codes included in the present analyses were as follows (1) managers; (2.1) science and engineering; (2.2) health care; (2.3) education; (2.4) other professionals; (3) associate professionals; (4) administrative and customer service; (5) service, care, and shop sales; (7) building and manufacturing; (8.1) mechanical manufacturing; (8.2) transport; and (9) elementary occupations. Furthermore, we used aggregated occupational categories for some analyses, defined as white-collar high-skilled (major groups 1–3), white-collar low-skilled (major groups 4 and 5), blue-collar high-skilled (major group 7), and blue-collar low-skilled (major groups 8 and 9). Major group 6 of agriculture and forestry was excluded due to a low number of participants (
n = 3624). A more detailed description of the occupational groups can be acquired from Vaisanen et al. [
10].
2.3. Other Measurements
The highest educational attainment at the time of the HPA was derived from Statistics Sweden (
www.scb.se, assessed on 01 December 2020) by linking the participant personal identity number, and was defined as length of education, ≤8 years, =9 years, =10 years, =12 years, =13–15 years, ≥16 years, and ≥17 years/research education, and was further aggregated for different analyses (see the Statistics section). Body mass and height were measured at the HPA in light-weight clothes, to the nearest 0.5 cm and 0.5 kg, with standardized equipment. Age was limited to 18 to 65 years and was, together with sex, derived from the HPI data.
2.4. Statistics
Weighted average was used to study the trends in both absolute (L/min) and relative (mL/min/kg) mean CRF in the 12 occupational groups, using the aggregated five-year period variable. Weights were calculated from the frequency of individuals according to sex, age-groups (18–34, 35–49, and 50–65), and education group (≤11 years, 12 years, and ≥12 years) in the Swedish population in 2019. Each value of CRF was multiplied by the assigned weight, which was summed and divided by the number of data points.
To calculate the average difference per year in CRF, three different linear regression models were used for each occupational group—a main effects model and two interaction models (the latter with year*sex and year*age-group). All of the models were controlled for year performed as a continuous variable, age-group, sex, education as a seven-pointed scale (≤8 years, =9 years, =10#x2013;11 years, =12 years, =13–15 years, ≥16 years, and ≥17 years/research education) and weight.
We calculated the potential influence of weight gain by dividing the difference between the decrease in relative and absolute CRF (in %) by the decrease in relative CRF. Contrasts were used to compare sex and age-group differences. All of the p-values were corrected with respect to the number of models using Bonferroni correction.
Direct standardization was used to study the trends in low CRF (<32 mL/min/kg) in four aggregated occupational groups (white-collar high skilled, white-collar low-skilled, blue-collar high skilled, and blue-collar low-skilled). The reference weights were calculated from the Swedish population in 2019 according to education (low ≤ 11 years or high ≥ 12 years), sex, and age-group in five-year categories. Confidence intervals were calculated with the normal approximation method.
The forecast was based on data from a weighted average calculated for two-year groups (the first two-year group of 2001–2002 and the last two-year group of 2019–2020). Several different Holt models, arima, naïve drift, and linear trend, as well as combinations of these models were tested for constant variance over time and autocorrelation. The models were tested in cross-validation, tuning the model parameters to get the lowest root mean squared error. The resulting model with the lowest root mean squared error was an ensemble model based on Holts linear exponential smoothening with a trend and a naïve model. The ensemble model was a close approximation of a linear extrapolation, which is reasonable given the approximately linear nature of the data.
All of the analyses and graphics were made using R (version 4.0.5, Austria, Vienna), with the packages Tidyverse [
20], Emmeans [
21], and Fable [
22].
3. Results
The characteristics of the participants are shown in
Table 1. There were a greater number of participants with HPA in the last 10 years (
n = 327,213) compared with the first 10 years (
n = 188,909) of the study period. The proportion of women decreased from 52% to 35% from the first to the last five-year period, while the age-group constitutions were evenly distributed throughout time. The proportion of participants with a high education increased, whereas the participants with a low education decreased. Body weight increased by 4.2 kg (6.1%) in women and 3.6 kg (4.3%) in men between 2001 and 2020. The proportion of blue-collar high-skilled participants increased over time, whereas the proportion of white-collar low-skilled participants decreased.
The weighted trend values in the absolute and relative CRF are shown in
Figure 1a,b, respectively. The largest decreases in both the absolute and relative CRF between the first and last five-year period were seen in administrative and customer service (absolute VO
2max 2.8 L/min to 2.6 L/min,
p < 0.001, −10.1% decrease, and relative VO
2max 38.4 mL/min/kg to 34.9 mL/min/kg,
p < 0.001, −9.4% decrease). In addition, a large decrease was seen in mechanical manufacturing (absolute VO
2max 3.0 L/kg to 2.8 L/kg,
p < 0.001, −6.5% decrease, and relative VO
2max 37.1 mL/min/kg to 34.4 mL/min/kg,
p < 0.001, −7.8% decrease). A large decrease in absolute CRF was seen in science and engineering (3.1 L/min to 2.9 L/min,
p < 0.001, −5.4% decrease). Large decreases in the relative CRF were seen in transport (35.1 mL/min/kg to 33.1 mL/min/kg,
p < 0.001, −6.1% decrease) and building and manufacturing (36.7 mL/min/kg to 35.9 mL/min/kg,
p < 0.001, −2.4% decrease). Only marginal changes were present in the absolute CRF for associate professionals (+0.8%,
p < 0.001) and service, care, and shop sales (−0.3%,
p < 0.001), and in relative CRF for managers (
p = 1.000) and health care (
p < 0.001). For relative CRF, it was for managers (−0.7%,
p = 0.036) and health care (−0.3%,
p = 1.000).
The larger decrease in relative compared with absolute CRF in seven of the occupational groups may have been attributed to simultaneous weight gain (as weight influences relative CRF). Approximately one third of the decrease in relative CRF may be explained by weight for education (34% due to weight gain) and other professionals (29%), while similar calculations revealed a larger contribution in some occupational groups (service care and shop sales, 94%, transport, 43%, and elementary occupations, 65%), and less for mechanical manufacturing (17%).
Regression models presenting change per year, adjusted for age, sex, weight, and education, are shown in
Figure 2 and
Table A1,
Table A2 and
Table A3. The main effect model (detailed in
Table A1) shows no change in CRF per year in managers (adjusted β = 0.000 (−0.017 to 0.018),
p = 1) and an increase for health care (adjusted β = 0.046 (0.014 to 0.077),
p = 0.180) in CRF, while transport (β = −0.162 (−0.190 to −0.134),
p < 0.001) stood out with the largest negative change per year. While white-collar occupational groups had a large range in change point estimates (β = −0.105 to 0.045) in different occupational groups, blue-collar occupational groups were more homogenous (β = −0.162 to −0.103).
For other professionals, associate professionals, administrative and customer service, and transport, there were significant interactions between men and women (adjusted
p < 0.001), with a more pronounced decrease per year in men compared with women for all (
Table A4).
There was a clear distinction in change in CRF per year when comparing the youngest and the oldest age-groups, where the younger group had a larger decrease in all individual occupational groups,
p < 0.001 (
Table A5). This was specifically pronounced in education, other professional, administrative and customer service, and transport. For managers, science and engineering, health care, other professionals, and associate professionals, the oldest age-group had an increase in CRF.
Figure 3 shows the change in proportion of low CRF (<32 mL/min/kg) over the study period in relation to the aggregation of occupational groups. All of the aggregated groups had an increase in the proportion with a low CRF. White-collar high-skilled occupations had the lowest proportion of low CRF at all timepoints and the lowest increase over the time period.
In the forecast analysis, the trend of decreasing CRF continued in all occupational groups into the foreseeable future (
Figure 4). The forecast model predicts a relatively large decrease in blue-collar low-skilled, −10% (from 33.3 mL/min/kg, to 30.3 mL/min/kg), and white-collar low-skilled, −9% (from 34.2 mL/min/kg, to 31.4 mL/min/kg), until 2040. The forecast predicts a smaller decrease in both white-collar high-skilled, −5% (from 36.4 mL/min/kg, to 34.8 mL/min/kg), and blue-collar high-skilled, −3% (from 35.6 mL/min/kg, to 34.6 mL/min/kg).