1 Introduction

The leading cause of global death is cardiovascular diseases (CVD) and the most important risk factor for CVDs is high blood pressure [1, 2]. Hypertension is related to substantial global mortality and morbidity due to its prevalence and manifold consequences such as coronary heart disease and stroke [3].

Furthermore, hypertension has also been shown to be negatively associated with mental health [4, 5] and subjective well-being [6, 7]. This could be either due to the effect of higher blood pressure itself, treatment effects, or the labelling effect of being told one is hypertensive [8]. Subjective well-being (SWB), because of to the limitations of Gross Domestic Product (GDP) as an indicator of social progress, has become a key global policy goal [9, 10]. Well-being affects all areas of life: work, leisure, and social relations as well as compliance with medical treatments [11, 12]. Hence, reducing the global burden of hypertension and CVDs and promoting subjective well-being is enshrined in the UN Sustainable Development Goal 3 [13].

In this paper we study the association between hypertension and well-being. While we are not the first to study this relationship [6, 7], the existing literature suffers from small samples, limited coverage of covariates of well-being [3, 4, 6], or uses large but non-nationally representative samples [7]. A key advantage of our study is that we use survey data covering a wide range of factors associated with well-being from more than one million Americans who took part in the nationally representative Behavioral Risk Factor Surveillance System (BRFSS) survey operated by the Centers for Disease Control and Prevention (CDC). This allows us to establish a robust relationship between hypertension and well-being controlling for other factors associated with well-being and to consider mediating factors of the hypertension-SWB relationship. Furthermore, well-being as a concept has a wider scope than mental health issues such as depression or anxiety used in the literature [4, 5] and considers the overall effect of hypertension on the quality of life.

2 Methods

2.1 Study Population

We analyze data from > 1 million Americans from the nationally representative, publicly available Behavioral Risk Factor Surveillance System (BRFSS) telephone survey, which is operated by the Centers for Disease Control and Prevention (CDC) in all U.S. States, the District of Columbia, and three territories. We use data from the “Hypertension Awareness” and “Emotional Support and Life Satisfaction” module while including various covariates of well-being from other modules. The two key modules were included biannually from 2005 to 2017, while in 2011 the life satisfaction module was excluded such that we use six waves. The modules were part of the core section in 2005, 2007, 2009 and were asked to all participants while they were part of the optional section in the following waves. Research has established the reliability and validity of the BRFSS [14].

2.2 Key Variables

Our dependent variable is well-being measures by the response to the question “In general, how satisfied are you with your life?”. The responses are given on a four-point scale: very satisfied, satisfied, dissatisfied, and very dissatisfied. We follow the work by [15] and construct a dummy variable which is one if the response is satisfied or very satisfied and zero otherwise. Life satisfaction is not equivalent to subjective well-being, but it is a key factor in SWB and equally represents evaluations of life across various aspects and in general [16, 17].

The key variable of interest is hypertension. It is measured by the response to the question “Have you ever been told by a doctor, nurse, or other health professional that you have high blood pressure?”. We construct a dummy variable indicating that the response was yes or that the respondent was told she is borderline high or pre-hypertensive and zero otherwise. Following the conceptual framework by [18, 19] (see Supplementary Material for a full description) and the related literature [20,21,22], we include the standard covariates for a well-being model such as demographic and socio-economic factors (see Supplementary Material for full details of all covariates).

2.3 Statistical Analysis

We employ descriptive and regression analyses to determine the association between hypertension and well-being while controlling for the usual covariates identified from the framework by [18, 19] and the literature [20,21,22]. We estimate the conditional mean of well-being, \(WB_{i,j,t}\), for individual i in state j in survey year t:

$$ WB_{i,j,t} = \alpha + \mathop \sum \limits_{n = 1}^{N} \beta_{n} X_{i,j,t}^{n} + \varphi_{j} + \mu_{t} + \varepsilon_{i,j,t} , $$
(1)

where n indicates the different covariates. We include state, \(\varphi_{j}\), and year, \(\mu_{t}\), fixed effects to control for stable differences across states (for example different policies) and variable factors such as economic cycles and survey context effects. Finally, we use cluster-robust standard errors at the year-level.

3 Results

3.1 Sample Characteristics

We present (weighted) descriptive statistics in Table 1. In our sample, we find that 29.2% ± 45.5% of respondents have hypertension and 94.5% ± 22.9% report that they have high well-being (satisfied or very satisfied with general life). In our sample, 49.8% ± 50% are male, most people are between 35 and 54 years of age (40.9%), have a High School (and some College) education, and are married or with a partner (66%). The majority is of good general health (84.4% ± 36.3%) and 4% ± 19.5% report having had a heart attack, 43.5% ± 49.6% have at least one child, and 76.4% ± 42.4% report that they exercised. High social capital is reported by 78.9% ± 40.8%, 62.9% ± 48.3% are employed, almost half have an annual household income above $50,000 per year (47.6%), and 85.5% ± 35.2% have health care coverage.

Table 1 Descriptive statistics

Finally, results suggest that there are significant differences between males and females in our key variables: Males are more likely to have high well-being (difference is 0.4%, p<.001) but also are more likely to have hypertension (difference is 2.9%, p<.001) according to t-tests allowing for unequal variances.

3.2 Regression Results

Table 2 presents our estimated results of the association between hypertension and well-being in the BRFSS. We find that hypertension is negatively associated with well-being (< 0.001). The association, however, is quantitatively small as being hypertensive reduces the likelihood of high well-being by 0.1%. Nevertheless, it is comparable to the negative well-being association of having had a heart attack (Table S.1 in the Supplementary Material). We also confirm this by computing the Shapley values, which allows us to quantify the relative importance of each variable [23, 24]. We find, again, that the relative contribution of hypertension is small but comparable to the relative importance of having had a heart attack, more important than gender. Finally, we also used machine learning to address the question of relevance: using Lasso regression, we find that hypertension is always (in the full sample and for men and women separately) selected by the algorithm from a large set of potential factors. The subsequent regression results confirm the results from this section.

Table 2 Regression results

The associations between well-being and covariates (presented in Table S.1) are in line with the consensus in the literature [20,21,22]. In short, high income, good health, having social contacts and support, and education are all positively associated with high well-being while age has a U-shaped association with well-being.

Furthermore, we find that this negative association only holds for males but not females. For males, being hypertensive reduces the likelihood of high well-being by 0.3% (p<.001) while there is no significant association found for females. For men, the association is of similar importance than having children.

4 Discussion

We add to the literature on the association between hypertension and well-being. Using the BRFSS study covering more than one million Americans and a wide range of well-being covariates, we find that hypertension is associated with lower well-being. However, we provide novel findings that this association is only found for males but not females.

We considered various extensions of our findings and discuss them as exploratory findings. First, the association between hypertension and well-being is increasing in age, especially for males. Second, we study how exercise mediates the relation [25]. We find that females with hypertension who exercise are more likely to have high well-being compared to females with hypertension who do not exercise but there is no such association for males. Finally, related to well-being is mental health and we find that the share of days over the last month with reported mental health problems is higher amongst those respondents with hypertension, especially for males.

Practically, our results suggest that the diagnosis of hypertension should not only trigger medical treatment but also involve a careful psychological management to reduce negative well-being effects and the related, potential negative interference with the adherence to the treatment plan [12]. We also note that research about the hypertension-SWB relationship across countries is needed.

Several limitations should be highlighted. First, our results should not be interpreted as causative since we lack an exogenous instrument to examine directionality. Second, we acknowledge that there will be other factors (e.g., treatment information) correlated with well-being and hypertension which are not included in our study. Finally, we lack information about when the respondent was told about the hypertension which might be important as well-being effects might be higher directly after the diagnosis and become smaller over time.