The present population-based birth cohort study, for the first time, evaluated the association between PA and adjusted individual-level outpatient PHC costs related to sustained healthy weight, overweight, and obesity and weight gain in midlife. In addition, the two-part model allowed us to separately estimate the probability of incurring any outpatient PHC costs, and among those who used outpatient PHC services, the level of costs, and to compare the results according to the changes in the participants’ PA and BMI. Obesity was found to be associated with a higher probability of having outpatient PHC costs in both sexes, and among females, weight gain and sustained obesity were associated with higher outpatient PHC cost level. Contrary to our hypothesis, sustained PA reaching the current recommendations or becoming physically active did not mitigate the impact of obesity or weight gain on outpatient PHC costs. The adjusted predicted individual annual outpatient PHC costs were found to be the highest for females with stable obesity, weight gain, and stable overweight regardless of the PA category, and mostly for males as well, although only a few of the differences in adjusted predicted PHC costs between the BMI categories within a fixed PA category and between the PA categories within a fixed BMI category were found to be statistically significant. Reaching the current PA recommendations predicted PHC costs only among males with stable healthy weight, among whom those who were stable inactive had 38% higher predicted costs than those who were stable active.
Our findings of a higher probability of having healthcare costs for individuals with obesity and among females, and the association of having obesity and weight gain with healthcare cost levels are in line with the findings of previous studies (6, 9, 10). We did not find any association between having stable overweight and PHC costs, which supports the previous finding of Finkelstein et al. (10) that healthcare costs for individuals with overweight do not significantly differ from those for individuals with healthy weight.
We found that female participants had 32% higher mean individual-level outpatient PHC costs than male participants. This is in line with previous studies (21). According to our results, female participants had lower MVPA volume at age 31 than male participants. However, between ages 31 and 46, more females than males became physically active and the MVPA volume of females increased by 29 minutes/week on average, while that of males increased by only 8 minutes/week. Living with children has been reported to have a negative impact on engagement in PA (37), especially among females (38). Children growing up and easing household work between ages 31 and 46 for mothers may enable them to become physically active at middle age. Persistent PA health benefits have been reported to require sustained PA (17). Among females, the impact of increased PA on healthcare costs may not be seen within the time-period used in the present study because recent PA increase and short-lived PA may not be enough to provide healthcare cost benefits.
The predicted average annual individual-level healthcare costs in the present study were significantly lower than those in a recent Finnish study (9), which reported €2 665 total annual healthcare costs (consisting of the costs of healthcare visits, hospital stays and prescribed medicines) for individuals with obesity, and €1 799 for individuals with healthy weight or overweight in Finland. The difference is explained by the fact that the outpatient PHC costs used in the present study represented only a fraction of the total healthcare costs due to the lack of data on the costs of outpatient hospital care, inpatient services, and medications. For example, Cawley et al. (1) reported that obesity raises the costs in all major categories of healthcare, with particularly large increases in inpatient services and prescription drug expenditures. Wang et al. (20) reported pharmaceutical costs and inpatient costs, but not outpatient costs, to be significantly higher for retirees over age 65 with overweight and obesity compared to retirees with healthy weight.
The interactions between BMI, PA, and related healthcare costs may be quite complex. BMI has been reported to be associated with PA (39–41). In the past decades, studies have suggested that obesity may be a driver of physical inactivity, instead of the previous assumption that low PA leads to obesity (39, 40), or at least that the relation might be bidirectional (39, 41). Physical inactivity itself is known to be associated with higher disease burden (42), which may lead to higher additional healthcare costs in the long run if PA remains low (7), in addition to obesity-related increased healthcare costs (7, 43). Stable inactivity may also reflect poor health status, which may prevent engagement in PA. Health status is one of the strongest previously reported correlates and a suggested determinant of engagement in PA (39, 44). Obesity itself has a significant independent negative impact on health (12, 45–47). Additionally, shifting from physical inactivity to activity may increase the use of healthcare services through, for example, exercise-related injuries, pain, or other problems, especially in the early stages of physically active lifestyle (48–51), and in individuals with obesity (49, 50, 52, 53). In addition, one form of complex interaction between BMI, PA, and healthcare costs is possible curvilinear shape of the association between PA and healthcare service use and healthcare costs across BMI categories (19, 20, 22). In our previous study (54) we found curvilinear association between adulthood accelerometer-measured PA and future income. The numbers of stable active participants with obesity, overweight, and a history of weight gain were too low to enable the evaluation of the possible curvilinear associations of PA with healthcare costs in the present study.
The present study had several strengths. The unselected population-based data represented both sexes and individuals from all sectors of the economy, occupational statuses, and education levels. Additionally, the participants in the present study were born around the same year in the same geographical area in Finland and were still living in Finland at the time of the study; thus, the risk of bias arising from the effect of age, race, and culture on PHC costs was low. The fact that all the participants in the present study were living in Finland also made the healthcare cost calculations reliable and made the obtained cost values comparable to each other. Furthermore, using individual-level data, we were able to capture the outpatient PHC costs of all health conditions leading to outpatient PHC service use, whether associated with obesity or not, so the possible outpatient PHC costs from PA-related injuries and diseases were also included. We also had a longitudinal perspective, as opposed to the previous cross-sectional studies (19–21).
Despite the foregoing, the present study also had some limitations. For example, the self-reported data concerning PA and healthcare service use in the year before age 46 might have led to recall bias. In addition, the data on self-reported PA used in the study reflected leisure time PA and did not include, for example, work-related PA. The standard unit costs reported by the Finnish Institute for Health and Welfare (30) that were used in the calculations also did not reveal the real healthcare service costs. Thus, the calculated and predicted individual-level outpatient PHC costs in the present study were underestimations. Additionally, we had no information about the participants’ reasons for healthcare service use, and we do not know if they had some health-related reasons for decrease or increase PA. For example, some disease or injury between ages 31 and 46 could have affected both PHC costs and PA. However, the effects of these factors on the main study results and conclusions were limited because the study aimed to compare the individual-level PHC costs according to the changes in PA and BMI, not to reveal the exact cost level. The classification of the participants as physically inactive or physically active was based on PA recommendations (17). The shift from physical inactivity to physically active or vice versa could have been caused by a change in MVPA as minimal as 1 min/week, because individuals with ≤ 149 minutes/week MVPA were categorized as physically inactive and those with ≥ 150 minutes/week MVPA as physically active. In addition, we did not have any information on the possible fluctuations in BMI and PA between ages 32 and 45. Moreover, in some subgroups, the numbers of participants were quite low, only 7 at the lowest. The number of individuals with weight loss between ages 31 and 46 among NFBC1966 cohort members was so low that we were not able to statistically analyze the associations of weight loss and PA categories with PHC costs. It is also well known that dietary factors and other healthy life habits (i.e. PA), BMI, and education level may interact with each other. Individual-level healthcare expenditures have also been found to increase with age (21, 55), obesity duration has been reported to increase functional limitations and disability in the long run (56, 57), and persistent PA health benefits have been reported to require continuing PA (17). Thus, the relatively young (46 years) age cohort in the present study might not be optimal for revealing the longitudinal associations of obesity and PA with healthcare costs. The complex interactions between BMI, PA, and healthcare costs could not be recognized in the study protocol that we used, and we were not able to distinguish the effects of different components of health behavior on healthcare costs.
As conclusion, this study shows that obesity is associated with a higher probability of having outpatient primary healthcare costs, and among females, obesity and weight gain are also associated with higher outpatient PHC cost levels. Reaching the current PA recommendations did not mitigate the impact of obesity and weight gain on outpatient PHC costs in midlife. Reaching the current PA recommendations alone is insufficient when aiming to decrease obesity-related PHC costs in midlife. Long-term effects of PA on outpatient PHC costs should be evaluated in future studies.