Is the Emotion-Health Connection a “First-World Problem”?
- 1Department of Psychology and Social Behavior, University of California, Irvine
- 2Department of Psychology, Boston University
- 3Gallup/Clifton Strengths Institute, Omaha, Nebraska
- 4School of Business, University of Kansas
- Sarah D. Pressman, Department of Psychology and Social Behavior, University of California, Irvine, 4201 Social and Behavioral Sciences Gateway, Irvine, CA 92697-7085 E-mail: pressmas{at}uci.edu
Abstract
Emotions have been shown to play a critical role in health outcomes, but research on this topic has been limited to studies in industrialized countries, which prevents broad generalizations. This study assessed whether emotion-health connections persist across various regions, including less-developed countries, where the degree to which people’s fundamental needs are met might be a better predictor of physical well-being. Individuals from 142 countries (N = 150,048) were surveyed about their emotions, health, hunger, shelter, and threats to safety. Both positive and negative emotions exhibited unique, moderate effects on self-reported health, and together, they accounted for 46.1% of the variance. These associations were stronger than the relative impact of hunger, homelessness, and threats to safety and were not simply attributable to countries’ gross domestic products (GDPs). Furthermore, connections between positive emotion and health were stronger in low-GDP countries than in high-GDP countries. Our findings suggest that emotion matters for health around the globe and may in fact be more critical in less-developed areas.
A first-world problem is a popular meme that makes light of the inconveniences in the developed world that would likely not be considered problems by a large portion of the world’s population. The meme captures the notion that people in the first world spend ample time on concerns that are not akin to the “real” problems faced by individuals in developing countries (e.g., violence, hunger). Related to this idea of concerns restricted to the first world is the greater amount of time and money that individuals in more-developed regions devote to their emotional well-being, trying to alter their levels of happiness or sadness.
Unsurprisingly, given this focus, factors such as depression and positive affect are associated with important health and health-relevant outcomes among people in the first world. To date, however, research on the connection between emotions and physical health has been done almost entirely in developed nations, where it is plausible that more pressing predictors of health than emotions (e.g., hunger, homelessness) vary minimally among individuals. Given this, we wondered whether the close tie between psychological factors and health was a first-world problem. Specifically, is the influence of emotions on how people feel about their health a phenomenon restricted to developed nations? In other words, is this connection absent in third-world countries, where the deprivation or fulfillment of basic needs is more critical to health than one’s emotions?
A large body of research has connected emotions and health in industrialized nations. For example, depression, stress, and other negative emotions (NEs) have been consistently tied to self-reports of increased pain, fatigue, and disease (e.g., Geisser, Roth, Theisen, Robinson, & Riley, 2000; Watson, 1988), whereas positive emotions (PEs) have frequently been tied to decreased pain, fatigue, and disease (Pressman & Cohen, 2005). Although self-reported health data can be problematic, well-designed longitudinal studies and laboratory experiments have provided ample evidence that emotional states can influence human biology. Stress, for example, is tied to in vitro physiological changes, almost every known biological illness, and differential survival rates for a range of illnesses (Cohen, Janicki-Deverts, & Miller, 2007; Segerstrom & Miller, 2004). NEs predict serious health consequences ranging from heart attack to early mortality (Hemingway & Marmot, 1999; Rugulies, 2002). Moreover, dozens of studies have revealed that PEs not only alter immune and cardiovascular function (Pressman & Cohen, 2005) but also predict decreased objective symptoms of illness (Cohen, Doyle, Turner, Alper, & Skoner, 2003), improve disease survival (Moskowitz, 2003), and add years to people’s lives (Chida & Steptoe, 2008; Diener & Chan, 2011).
Although the breadth of these findings is impressive, one serious concern is their generalizability. The majority of studies on the emotion-health connection have been conducted in North America and Western Europe, with a handful carried out in other industrialized nations. This is problematic, obviously, because of the inability to generalize these findings to unsampled countries. More critically, most of these studies have relied on convenience samples of students or middle- and upper-class adults who fall into the first-world category. It is therefore possible that the emotion-health connection exists only when individuals’ basic needs are already met. It seems likely that when one faces famine, homelessness, or serious safety concerns, those experiences would be more critical correlates of health than one’s day-to-day level of happiness. It seems almost a luxury to have emotions matter to physical health. It is not that individuals in less-developed countries are never happy (Diener & Suh, 1999); it is simply that happiness may have less of an effect on their health.
The purpose of the study reported here was to examine the extent to which the connection between emotion and health exists across the planet, using a sample of participants from 142 countries, representative of 95% of the worlds’ population. In particular, our goal was to examine whether this connection exists in all countries, including those with more-pressing daily and long-term concerns that might preclude the effects of plausibly less-consequential emotional states. We considered (a) the relative independence of PEs and NEs in their associations with self-reported health; (b) whether these associations persist when hunger, homelessness, or threats to safety are controlled for; and (c) whether these correlations vary with country-level gross domestic product (GDP) per capita.
Although self-reported health is not an ideal measure of physiological function, it has many strengths (McDowell, 2006, p. 12) and is an established predictor of objective health (e.g., future mortality) even after accounting for baseline health and comorbidity, health practices, socioeconomic factors, and other relevant covariates (Ferraro, Farmer, & Wybraniec, 1997; Idler & Benyamini, 1997). Given the impracticality of collecting biological specimens from or conducting physicals on a representative sample of the world’s population, self-reported health must be considered a limited but useful indicator of health.
Method
Participants
Participants were 150,048 individuals (52.1% female, 47.9% male; age range = 15–99 years, M = 39.39, SD = 16.91) who participated in the first wave of the Gallup World Poll. The Gallup World Poll was initiated in 2005 and annually surveys approximately 1,000 individuals from more than 142 countries, thus providing a representative sample of 95% of the world’s population. Information about the development of the survey, ethical considerations, and survey procedures used for the World Poll can be obtained online at www.gallup.com/strategicconsulting/worldpoll.aspx.
Measures
PE was measured with a series of four items that asked whether individuals had laughed or experienced enjoyment, love, or happiness during the previous day. NE was measured with a series of six items that asked whether individuals had experienced worry, sadness, stress, boredom, depression, or anger during the previous day. Self-reported health was assessed with four questions regarding whether individuals were satisfied or dissatisfied with their personal health, whether they had any health problems that prevented them from doing any of the things people their age can normally do, whether they felt well-rested, and whether they had experienced physical pain during the previous day. Responses were recoded so that higher scores indicated superior health.
To assess the degree to which participants’ basic needs were met, we asked participants whether they had not had enough money for food or shelter for their family or had gone hungry at any point during the past 12 months. Safety and security needs were assessed by asking participants whether they feel safe walking alone at night and whether, within the past 12 months, they had had money or property stolen or had been assaulted or mugged. Country-level wealth was measured in GDP per capita (in U.S. dollars), using data from the 2005 U.N. Human Development Index (United Nations, 2006).
Data analysis
A series of structural equation models were analyzed to determine the relationships among PE, NE, self-reported health, and basic needs. Latent constructs were specified using the respective items as indicators. All models were analyzed using LISREL (Version 8.80; Jöreskog & Sörbom, 2006) and maximum likelihood estimation procedures using the variance-covariance matrix. Models were evaluated using common fit indices (i.e., root mean square error of approximation, or RMSEA; comparative fit index, or CFI; standardized root-mean residual, or SRMR; and nonnormed fit index, or NNFI). Missing data (8.4%) were imputed using Markov chain Monte Carlo procedures with the PRELIS software (Jöreskog & Sörbom, 2006) prior to conducting all analyses. The primary advantage of structural equation modeling over traditional statistical techniques, such as correlating variables, is that it provides estimates of the unique associations between latent variables while controlling for measurement error.
Results
Latent structure of emotion
The latent structure of PE and NE was first examined. A measurement model specifying positive and negative emotion as oblique latent constructs demonstrated excellent fit, χ2(34, N = 150,048) = 10,787.161, p < .001 (NNFI = .975; CFI = .981; RMSEA = .046; SRMR = .029), with a latent correlation of –.54 between PE and NE. These results indicate that, across more than 150,000 individuals from more than 142 countries, positive and negative feelings were distinct constructs that exhibited a strong negative correlation with each other.
Emotion and physical health
The associations of PE and NE with self-reported health were tested using models in which both types of emotions were specified as independent predictors. This model demonstrated good fit, χ2(74, N = 150,048) = 51,112.580, p < .001 (NNFI = .930; CFI = .943; RMSEA = .068; SRMR = .053), and indicated that both types of emotions exhibited unique, moderate associations with physical health (βs = 0.36 and −0.41, respectively). Together, they accounted for almost half (46.1%) of the variance in health. Analyses of health items revealed that all were significantly related to emotion (ps < .01). PE was most correlated with feeling well-rested (r = .42), and NE was most correlated with greater pain (r = −.35). All health factors except for pain had stronger associations with PE than with NE. Associations between emotions and self-reported health were approximately the same across all age groups (e.g., adults under 30, adults over 65).
Next, we added data on food, shelter, and safety to the model as additional predictors of health to test whether PE and NE had effects on health beyond those of basic needs. This model demonstrated excellent fit (see Fig. 1) and indicated that PE and NE both continued to exhibit moderate unique effects on health (βs = 0.34 and = −0.36, respectively) after controlling for the impact of deprivation or fulfillment of needs. Although safety did not contribute significantly to the overall equation, when it was assessed alone with health, the latent correlation was significant, r = −.156, p < .001, indicating a weak relationship that disappeared when the unique effects of all four predictors were tested in the structural equation model.
Standardized results of a structural equation model examining the unique effects of positive and negative emotions on self-reported physical health beyond the effects of deprivation or fulfillment of basic food, shelter, and safety needs. Asterisks indicate that the variance of latent variables was fixed at 1 so that results could be provided in a standardized metric.
Country-level wealth and the emotion-health link
To assess the generalizability of our results for all types of nations, we factored GDP into the associations. Specifically, we inspected the magnitude of correlations between each type of emotion and physical health in countries with high, low, and moderate levels of GDP per capita after covarying the respective effects of the other emotion. Correlations for five example countries from each GDP category are shown in Table 1. Correlations between emotions and self-reported were similar across countries with similar GDP levels. All correlations between emotion and self-reported health were significant; correlations between PE and health were particularly strong in countries with the lowest levels of GDP. For example, the correlation in Japan (a high-GDP country) was .133, compared with .337 in Sierra Leone (a low-GDP country). We found similar negative associations between NE and health, with similarly sized coefficients, across different GDP levels.
Associations of Positive and Negative Emotions With Physical Health in Countries With High, Moderate, and Low Per-Capita Gross Domestic Products
Figure 2 demonstrates the clear linear relationship across the 142 countries connecting PE to self-reported health (for brevity’s sake, we have not provided a graph of the relationship between NE and self-reported health, but it would look similar). Although there is a cluster of low-GDP countries with both low PE and poor self-reported health, GDP levels cannot explain this association. There are poor countries with PE-health correlations as high as those for countries with the highest GDPs.
The relation between positive emotion and physical health in 142 countries. Higher values on the y-axis indicate greater positive feelings, and higher values on the x-axis indicate greater self-reported physical health. Countries represented by smaller and bluer circles have a lower gross domestic product (GDP), whereas those represented by larger and greener circles have a higher GDP. Labeled countries include countries from Table 1 in addition to other examples of countries with low, moderate, and high GDPs.
Discussion
Emotions matter to health everywhere. Both PE and NE were linked to subjective ratings of health in a sample representative of the world’s population, and, contrary to our prediction, these links were stronger than the links between unmet basic needs and health, such that together, they accounted for nearly half of the variance in perceived health. Also against our expectations, results indicated that the emotion-health connection is not a first-world issue, and that the link between PE and health is in fact stronger in countries with weaker GDPs. Our findings also indicate that PE and NE are distinct phenomena, supporting previous work using smaller samples from industrialized countries (Diener & Emmons, 1984).
Related to the independent associations of PE and NE with health were the differential associations between these variables and health according to different levels of GDP among participants’ home countries. Although the NE-health link was consistent across countries, the PE-health link was strongest in countries with low GDP. When we statistically removed the effect of GDP, the NE-health link remained significant, with the PE-health link remaining significant and growing in magnitude from rich to poor countries. Separate analyses showed that, when controlling for GDP, people in the United States exhibit the same NE-health link as people in Malawi (a low GDP country) (−.196 and −.194, respectively), but people from Malawi exhibit a stronger link between positive emotion and health than do people from the United States (.295 vs. .166, respectively).
In past research conducted with samples from industrialized nations, emotions have been connected to health. However, the magnitude of these associations has typically been moderate to small. We hypothesize that this pattern may be due to medical interventions downgrading the impact of emotions on health. For example, an unhappy adult with resultant hypertension in an industrialized country can take blood- pressure-lowering medication. Most Malawaians cannot. As far as the graded relation of self-reported health with PE only, the explanation is unclear. This pattern does suggest, as discussed previously, some independence of PE and NE, and it is consistent with hypotheses that it may be the absence of PE that is critical to objective health (Pressman & Cohen, 2005), whereas NE is more relevant to perceptions of health (Cohen et al., 1995; Watson & Pennebaker, 1989), which may vary less than objective health across countries.
The health care needs of typical Americans and Malawians are as disparate as their life expectancies. Nevertheless, there are healthy and sick people in both countries, and these people’s health status is associated with their emotional experiences more than it is associated with the fulfillment of their needs and their income. What is common among the healthy happy and the sad sick are the physiological changes known to co-occur with emotional factors (see reviews by Cohen et al., 2007; Herbert & Cohen, 1993; Pressman & Cohen, 2005). Given the numerous studies on this topic, it is hard to dismiss the data from this study suggesting that emotions and health are linked across the world.
A limitation of this research is that self-reported health was used as a proxy for objective health. Self-reported health is a valid predictor of future health (Ferraro et al., 1997), as well as other important outcomes having to do with self-care, such as medical compliance (Mechanic, 1972). Although judgments of physical states previously have been shown to be influenced by emotional states (Watson & Pennebaker, 1989), studies have shown that even after accounting for the overlap between mood and health perceptions, objective health outcomes are still tied to self-reported health (Idler & Benyamini, 1997). Furthermore, self-reported health has been found in many studies to outperform indicators of chronic illness (e.g., Miilunpalo, Vuori, Oja, Pasanen, & Urponen, 1997) and common risk factors such as smoking, alcohol consumption, and marital status (e.g., Schoenfeld, Malmrose, Blazer, Gold, & Seeman, 1994) as a predictor of mortality. Given that there are symptoms of poor health that cannot be explained by physiological markers of illness, poor health as indexed by a self-report measure of health using a multi-item scale may reflect something physiologically untestable that, judging from the data in this study, is associated with emotion around the planet.
The other limitation of this research is the cross-sectional nature of the data. It is well established that poor health influences emotional states and that emotions predict health (e.g., Cohen & Rodriguez, 1995; Lehrer, Isenberg, & Hochron, 1993). Given such previous findings, in addition to the highly subjective nature of the health measure used in our research, we cannot speak to the issue of causality, and it is very likely that the relationship between emotions and health is bidirectional.
Finally, it was surprising that the measure of safety was not associated with health in the full structural equation model. There was a weak association in a preliminary confirmatory-factor-analysis model, indicating that emotion may be the pathway by which safety affects health. That being said, the size of the effect of safety was still approximately one third that of the emotion variables, which indicates that differences in the deprivation or fulfillment basic needs such as safety are not the entire reason for these emotion-related differences.
Our findings suggest that positive and negative emotions are inextricably linked to reports of health among both rich and poor people in the global community. Given these findings, health professionals around the world should begin considering the importance of the emotions of their patients and research subjects. At the global level, surveys are frequently used to assess GDPs and people’s basic needs but disregard these people’s day-to-day feelings. Given the strong and prevalent correlations revealed by our research, this factor can no longer be ignored.
Article Notes
-
Declaration of Conflicting Interests The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.
- Received February 15, 2012.
- Accepted June 4, 2012.
- © The Author(s) 2013
OnlineFirst Version of Record
Other journals by APS