Introduction
Health-related quality of life (HRQOL) describes a patient’s perception of how his or her health status affects physical, psychological, and social functioning and well-being [
1,
2]. The focus for health care has become increasingly aimed at HRQOL as well as the quantity of life bestowed by clinical treatments [
3]. This is especially true among chronic disease populations where cure remains elusive. In particular, cancer care frequently measures HRQOL as an important outcome [
4,
5]. Even for diseases where cures are more routine, the impact on HRQOL of the patient has been shown to be related to treatment outcome [
4,
6,
7]. Recent studies showed that HRQOL can be prognostic of survival and other treatment outcomes as well as useful in identification of otherwise undetected clinical problems [
4,
7].
Along with many HRQOL items, perceived general health and depressive symptoms have received increased attention for potential clinical relevance. Perceived general health (i.e., self-assessed health) has been recognized as a valuable clinical tool, as it captures both current health status and subtle changes in health [
8]. Recent studies reported that overall perception of general health can be used as a quick tool for identifying patients at high risk of imminent death and hospitalization [
9]. Depressive symptoms are also known to have a potentially large influence on overall clinical outcomes including survival [
10]. Patients with depressive symptoms are less likely to seek treatment for medical conditions and to adhere to treatment recommendations [
11‐
13], and thus have potentially worse clinical outcomes.
There are gender differences in both the frequency and manner in which patients report clinical symptoms and treatment-related side effects [
4]. These reporting differences may affect self-reported HRQOL. Men have been observed to communicate their needs less than female counterparts and risk failing to acknowledge existing medical problems until the window of opportunity for effective intervention has expired [
14,
15]. It is less studied whether predictors of HRQOL are the same among men and women, and how the impacts of these predictors are influenced by gender.
Most research on HRQOL predictors focus on patients with individual diseases. Some predictors include socioeconomic status, age, gender, and comorbid health burden. A study based on pediatric patients with diabetes showed that HRQOL is influenced not only by disease-related factors but by the complex of non-disease related determinants such as gender and socioeconomic status [
16]. In terms of determinants not directly related to a specific disease symptom, arthritis-related pain is most prevalent among patients aged between 45 and 64, blacks and Hispanics, and with less than a high school education [
17]. A study using patients with Type 1 diabetes and coeliac disease showed an impact of multiple comorbidities on HRQOL [
18]. Those with both conditions had significantly lower HRQOL than patients with only Type 1 diabetes alone. The impact of multiple comorbidities on HRQOL is well supported by numerous studies [
19‐
22]. For instance, a recent study based on Medicare beneficiaries showed that the majority of chronic conditions, including cancer, were associated with decrements in HRQO, with substantial impact of the cumulative effects of comorbid conditions [
19]. However, in broad patient populations without specific illnesses, the predictors of HRQOL are relatively less known. Furthermore, it is not yet quantified the relative influence of HRQOL of each determinant such as comorbid health burden, age, gender, lifestyle issues, and body mass index (BMI) in the general population.
Using patients with a wide range of medical conditions, this study was to evaluate the impact of commonly collected health and lifestyle determinants for HRQOL deficits. More importantly, we aimed to quantify the relative influence of each determinant when all of these determinants are simultaneously considered. As a secondary aim, we assessed gender as a potential modifier of the contributors.
Discussion
Using a broad patient population enrolled into the Mayo Clinic Biobank, we found that prevalence of deficits in perceived general health (10 % with poor or fair health) and depressive symptoms (5 % with PHQ-2 score 3 or higher) was relatively low. The Center for Disease Control’s 2010 Behavioral Risk Factor Surveillance System Survey (
http://apps.nccd.cdc.gov/HRQOL/) reported 16 % for subjects with fair or poor self-rated health, with higher percentages for older subjects (>24 % among subjects aged over 65 years). For depressive symptoms, a recent study reported that roughly 11 % of the subjects in the primary care population had PHQ-2 scores 3 or higher [
31]. We also found that the greatest contributor to HRQOL deficits was disease burden, especially for perceived general health (relative influence of 63 %). For depressive symptoms, disease burden and age had similar influence on the risk of deficits. The impact of disease burden on deficits of HRQOL likely reflects the impact of disease symptoms (like pain, shortness of breath) upon HRQOL. It may also reflect the relationship between disease burden and functional decline [
32]. After accounting for disease burden and age, demographic characteristics showed minimal influence, regardless of individual strong association. While the importance of disease burden on HRQOL deficits has been supported in studies on patients with chronic diseases and/or more general population [
21,
33‐
36], our study quantified the relative influence of disease burden when other commonly collected determinants such as demographic characteristics were simultaneously considered.
We observed that subjects with deficits in perceived general health were slightly older than those without deficits. However, age was inversely associated with risk of deficits in depressive symptoms. Major depressive symptoms have been reported as fairly uncommon in older adults [
37,
38]. The incidence of depression tends to peak at age 30 and decrease thereafter with a small increase at age 50 [
37,
38]. We also observed that lower level of education, ever smoking, and no alcohol consumption were associated with higher risk of HRQOL deficits. These findings are supported by others using more general populations [
39‐
42]. Lower educational level can be a surrogate for lower socioeconomic status. The deficit in HRQOL may reflect the relationship between lower income and HRQOL. In the MCB, very few of our participants reported excessive drinking (86 % with one drink only per day), thus our population is primarily one of moderate drinkers vs. non-drinkers. Moderate drinking has been reported to have health benefits and may partially explain our finding [
43,
44].
Employment status was observed to be associated with risk of HRQOL deficits, with different effect by gender. The association of currently not working for pay was stronger among males compared to females, even after adjusting for disease burden. This observation suggests that psychological stress related to unemployment may be higher for males than females and thus impact quality of life. It may also reflect the functional ability to continue to work with those with functional disabilities opting out of employment.
Obesity did impact deficit of HRQOL and did show some gender differences. For females, obesity is strongly associated with higher rate of HRQOL deficits, although there is no difference between normal and overweight females. Once adjusting for disease burden, obesity is no longer associated with HRQOL. Although overweight and obesity are socially undesirable, especially in females, its psychological impact may be minimal, considering roughly 70 % of the MCB participants were at least overweight. For males, overweight is negatively associated with the risk of deficits in HRQOL. The association remained after adjusting for disease burden. Such a finding is supported by several recent studies showed that being overweight is linked to better clinical outcomes, including survival [
45‐
48].
There are some limitations to the study. First, there may be survival bias due to the use of prevalent diseases. Second, the MCB participants do not fully represent all patients seen at Mayo Clinic, as it does not include those who are healthiest (because they did not visit their primary care providers and thus were not invited) and the sickest (inability or refusal to participate). Third, comorbid health conditions are based on self-report, and thus there is potential for recall bias with self-report. Fourth, patients may have underlying health reasons of minimizing less desirable lifestyle attributes like alcohol or smoking. In addition, participants with no current alcohol consumption may include those who drank excessively before but now do not drink any longer. Lastly, a significant proportion (roughly 40 %) of the MCB participants is residents of Olmsted County, MN, where the Mayo Clinic Rochester is located. These residents are largely white and well educated which may limit some of the generalizability of the study to different populations.
Competing interests
The authors declare that they have no financial or non-financial competing interests.
Authors’ contributions
ER, PYT, JEO, JRC, JAS contributed for the design of the study, interpretation of the results, and preparation of the manuscript. ER, MAH, PJN performed statistical analyses. JP and SJB provided critical revision of the manuscript. All authors approved the final manuscript.