The Great Recession, Behavioral Health, and Self-Rated Health: An Examination of Racial/ethnic Differences in the US

Abstract

The Great Recession has been associated with racial/ethnic disparities in economic loss, alcohol-related problems and mental health in the US. In this study, we examine its effect on overall health, the role of heavy drinking and mental health, and whether these relationships vary by race/ethnicity. Using US National Alcohol Survey data collected from White, African American and Latino individuals between June 2009 and March 2010 (N=4,656), we conducted gender-stratified simultaneous path modeling to test racial/ethnic differences in hypothesized paths from recession-related hardships to overall self-rated health through current depressive symptoms and heavy drinking. Recession impacts were measured using an index of job-related, financial and housing hardships. Models accounted for demographic characteristics and heavy drinking, health conditions and alcohol-related health harms occurring prior to the Great Recession. We found that in men and women of each racial/ethnic group, more accumulated recession hardships were associated with greater depressive symptoms and more frequent heavy drinking, and depressive symptoms were associated with poorer self-rated health. Further, heavy drinking was related to poorer self-rated health in Black men and depressive symptoms in Latino men, and for Black and Latina women, prior heavy drinking was associated with current depressive symptoms. Findings highlight adverse, behavioral and overall health consequences of a severe recession for men and women of diverse racial/ethnic groups, as well as unique risks for Black and Latino men and women. Findings suggest the need for behavioral health interventions alongside multisector strategies to bolster the labor market and social safety net during severe economic downturns.

1. Introduction

In the US, the Great Recession of 2007–9 was marked by a doubling of the national unemployment rate and the highest, long-term unemployment rate in the prior 60 years ¹. While US prevalence of any alcohol use declined, the rate of frequent binge drinking (consuming 5 or more drinks on an occasion) increased, particularly among persons who lost their jobs ², similar to findings in Britain ³. Experiences of severe recession impacts have been associated with increased risk for negative drinking consequences and alcohol dependence symptoms ⁴ and poorer psychological health, including increased depression ⁵. Other reports link the Great Recession to increased alcohol-attributable hospitalizations in Taiwan ⁶, and elevated suicide rates in Europe ⁷. Recent reviews have further highlighted the negative effects of home foreclosure on health behaviors and mental and physical health ⁸, and of unemployment on self-reported health ⁹, with one review reporting robust impacts among manual workers ¹⁰.

Importantly, there is evidence of racial/ethnic disparities in exposure to and consequences of recessionary events in the US. During the Great Recession, African American and Latino households were more likely than White households to experience various recessionary hardships, including reduced hours and earnings, job loss, and housing insecurity or loss ^9,11, leading to dramatic racial/ethnic differences in relative loss of wealth ¹². African Americans were also more likely to show a subsequent increased risk for chronic mental illness than White individuals ¹³, and experiences of job or housing loss were associated with greater drinking to drunkenness, negative drinking consequences, and alcohol dependence –particularly for African American drinkers, as well as for male and middle-aged drinkers generally ^4,11,14,15. Black and Latino communities’ greater exposure to recessionary hardships and their associated impacts on behavioral health may reflect, in part, differences in employment factors (e.g., slower re-employment following layoff) and financial-buffering and coping resources such as wealth and quality healthcare ^16–18. Taken together, such findings align with the stress process model, which holds that groups with disadvantaged social statuses have both greater exposure to stressors and fewer coping resources to mitigate deleterious effects of stressors on behavioral and physical health ^19–21. The model suggests population differences in severity and duration of stressors can lead to disparities in mental health, substance use disorders, and physical health ^22,23.

In the current study, we extend prior research by examining the effects of the Great Recession on overall health, testing whether mental health and heavy drinking play a role and whether relationships differ by race/ethnicity. In keeping with the stress process model, observed associations between recessionary events and behavioral health outcomes such as heavy drinking and depressive symptoms ^4,14,24, and longitudinal effects of behavioral health on subsequent physical health ^25–29, we expected recession hardships to be associated with poorer mental health and greater heavy drinking and, in turn, poorer overall health. We also expected these relationships to be stronger among Black and Latino, as compared to White, respondents. Given documented associations between problematic alcohol use and subsequent depressive symptoms ^30,31, we accounted for prior history of heavy drinking in addition to chronic health conditions and alcohol-related health harms experienced prior to the Great Recession.

2. Methods

2.1. Data

The 2009–10 US National Alcohol Survey (NAS) was a household probability survey employing computer-assisted telephone interviews with the US adult population aged 18 or older, using a sampling frame including all 50 states and Washington, DC, with private residences selected using random-digit dialing of both landline and cellular phones. The landline sample included a main sample and an ethnic minority oversample of Latino and Black respondents. Fieldwork was undertaken by ICF Macro (Fairfax, VA). Interviews were conducted between June 2009 and March 2010, more than 16 months after the US recession officially began in December 2007.

On average, completed interviews took 55 minutes, covering a variety of topics related to alcohol use and related problems, intervention and treatment services, and individual and social factors that could be related to these. Cell phone respondents were asked a single recession item and thus were excluded from the current analysis due to a lack of data needed to create a recession index. The NAS cooperation rate ³² was 50% for the landline sample (N=6,855). Due to lack of data on prior heavy drinking for teen respondents and reduced relevance of job-related recession impacts for retirees, for this analysis we restricted the sample to respondents aged 20 to 75 years old who had data on recession impacts asked toward the end of the landline interview. After excluding 1,066 respondents due to age and 1,133 partially-complete respondents lacking recession data, the final analytic sample included 4,656 adults (2,946 women and 1,710 men).

2.2. Measures

2.2.1. Recession-related hardships

Respondents were asked if they or someone in their household had been negatively impacted by the economic recession since January 2008. Those indicating a negative impact were then asked if anyone in the household had experienced any of the following: a) hours or pay reduced at work, b) lost a job, c) trouble paying rent or mortgage, and d) lost their housing (either owned or rented). As reported in a prior NAS studies, prevalence in the full sample ranged from 32% for reduced hours/pay to 16% for job loss and (separately) trouble paying for housing, and 3.5% for housing loss ⁴. There was significant racial/ethnic variation: 12% of White, 24% of Latino, and 27% of Black respondents in the full sample experienced trouble paying for housing ¹¹. The recession impacts index was a sum of these four dichotomous indicators (range 0–4), with scores of 0 assigned to all respondents indicating their household was unaffected by the recession. In preliminary analyses, the recession index was a better predictor of self-rated health than the single item indicating severe economic loss (job or housing loss) found in prior studies to be associated with psychological distress, drinking to drunkenness, negative consequences of drinking, and alcohol dependence ^4,11,14.

2.2.2. Self-rated health

For the key outcome measure, respondents were asked “would you say your health in general is excellent, very good, good, fair or poor?” (range 1–5, coded such that higher scores indicate poorer health). This single-item measure, whose responses are typically recorded on a 4- or 5-point scale, is widely used in national surveys in the US and elsewhere. It is a holistic measure of the respondent’s self-assessed health, capturing a variety of health domains ^33–35 and robustly predicting morbidity and mortality from various conditions in population-based studies ³⁶.

2.2.3. Depressive symptoms

This was measured using eight items from the Center for Epidemiologic Studies Depression scale (CES-D; ³⁷). Respondents were asked how often during the last week they felt: a) bothered by things that usually don’t bother them, b) depressed, c) hopeful about the future (reversed), d) had restless sleep, e) happy (reversed), f) lonely, g) enjoyed life (reversed), and h) sad, with response options scored as 0 (‘rarely or none of the time’) to 3 (‘most or all of the time’). The measure used was the summation of the eight items (range: 0–24; Cronbach’s alpha = .92), shown to be highly correlated with the full CES-D (r=.93)³⁸.

2.2.4. Current heavy drinking frequency

This measure was based on a series of questions on the frequency of drinking 12 or more drinks, 8–11 drinks, and 5–7 drinks on any single occasion during the last 12 months ³⁹. Response categories included “never in last 12 months” (0), “at least once but less than monthly” (1), “at least monthly but less than weekly” (2), and “at least weekly” (3). As in national guidelines ^40,41, different thresholds for heavy drinking were chosen for each gender. For men, current heavy drinking was operationalized as the ordinal frequency of having 8 or more drinks on a single day in the past 12 months (summing frequencies reported for having 8–11 drinks or 12+ drinks in a day); for women, it was having 5 or more drinks on a single day (summing frequencies of having 5–7, 8–11, or 12+ drinks in a day).

2.2.5. Prior heavy drinking

Four items elicited the frequency of drinking five or more (5+) drinks on one occasion during specific life decades (teens, 20s, 30s and 40s), using 5 response options: “every day or nearly every day”, “at least once a week”, “at least once a month”, “at least once a year”, and “never” ⁴². Prior heavy drinking days were calculated based on the respondent’s age and their heavy drinking during the decade immediately prior. As examples, frequency of 5+ days during the teens was used for respondents aged 20 to 29, and 5+ during the 40s was used for persons aged 50 and older. The frequency categories were recoded into number of days (range: 0–300), which was log-transformed due to skewness.

2.2.6. Prior chronic conditions and alcohol-related health harms

These measures were included as control variables. Prior chronic disease (before the recession) was an indicator variable based on a series of questions asking whether the respondent had ever been told by a doctor that they had heart disease, diabetes, a stroke or cancer, and, if so, the age they were first told about the condition. Respondents who had any of the above conditions prior to 2008 were coded as positive for prior health conditions. Prior alcohol-related health harm was based on a question about whether drinking had ever harmed the respondent’s health (including physical and mental health, accident or injury), and, if so, at what age drinking started to have harmful effects. Prior alcohol-related health harm was coded as positive for those who first experienced a health harm prior to 2008. Parental alcohol problems accounted for whether the respondent had a biological father or mother who had a drinking problem or who was an alcoholic.

2.2.7. Demographic control variables

Analyses controlled for age (20–34, 35–49, 50–75), race/ethnicity (non-Latino White, non-Latino Black, Latino, other), educational attainment (less than high school, high school diploma, some college but no degree, college graduate), family income (≤$40,000, $40,001-$80,000, >$80,000, missing), employment status (employed full or part-time, unemployed, other), and marital status (married/living with partner, never married, separated/divorced/widowed).

2.3. Analysis

Preliminary bivariate analyses described prevalence and means of major variables in the demographic subgroups of interest. We conducted path analysis using Mplus version 7.4 ⁴³ to examine effects of the Great Recession and prior heavy drinking on current health status, with recent depressive symptoms and current heavy drinking specified as potential mediators and other demographics as covariates. Poorer self-rated health status (ranging from 1–5) and the depressive symptoms scale (ranging from 0–24) were treated as continuous variables, and current heavy drinking frequency was treated as an ordinal variable. We used the weighted least squares estimator (WLSMV) with robust standard errors. Statistically non-significant paths and covariances were trimmed to conserve degrees of freedom and improve model fit, which was evaluated using the root mean square error of approximation (RMSEA), comparative fit index (CFI) and Tucker-Lewis fit index (TLI). Mediated effects were estimated using the MODEL INDIRECT subcommand.

Modeling was conducted for women and men separately due to gender differences in heavy drinking, depressive disorders, and the Great Recession’s relationship to alcohol outcomes and healthcare utilization ^4,44–46. Once an adequate model was identified for each gender subgroup, we tested racial/ethnic differences for White, Black and Latino subgroups in the effects of the recession and prior heavy drinking using multiple group analysis (respondents of other race/ethnicity were excluded due to small sample size). The identified model for each gender group was first estimated without any equality constraints across race/ethnicity groups as a baseline model. Then, we constrained each regression coefficient to be equal across groups, step by step, for each pathway of interest. For each comparison, a less-restrictive model was compared with the model with more equality constraints using the SB-scaled Chi-square difference test; overall model fit indices and model interpretability also guided specification of constraints. Non-invariant paths in the multi-group model indicated differential effects across race/ethnicity. All analyses used survey weights to adjust for sampling probability and non-response, such that data are representative of the US in 2009–2010; the sampling weights also adjusted for early drop-off (incomplete interviews) resulting in a lack of recession data and exclusion from these analyses.

3. Results

3.1. Descriptive Results

In both gender groups, a larger proportion of Black and Latino respondents reported fair or poor health than White respondents (Table 1). There also were significant racial/ethnic differences in age, education, income, employment and marital status. Compared to Black or Latino respondents, White respondents were more likely to be older and had higher education and income, and were less likely to be unemployed and never married. There were no significant racial/ethnic differences in health conditions or alcohol-related health harms prior to the Great Recession, nor in having a parent with a history of alcohol problems.

Table 1.

Sample characteristics, 2009-10 US National Alcohol Survey respondents aged 20-75

	Women				Men
	Total (2946)	White (1574)	Black (634)	Latina (643)	Total (1710)	White (1062)	Black (282)	Latino (311)
Fair or poor health (%)	20.6	16.8	32.6	29.7***	15.5	14.0	20.5	23.9**
Prior chronic disease (%)	19.8	20.0	25.0	15.5	18.9	20.2	14.0	15.2
Prior alcohol health harm (%)	11.7	12.3	6.8	10.8	20.4	20.7	19.3	21.4
Parental alcohol problem (%)	25.9	25.9	24.0	30.2	24.5	24.7	26.8	26.1
Age (%)
20–34	29.3	26.6	31.6	41.7***	31.0	27.0	37.5	43.5**
35–49	31.0	30.7	31.4	32.3	33.6	33.7	32.6	34.2
50–75	39.7	42.7	37.1	26.1	35.4	39.3	29.9	22.3
Race/ethnicity (%)
White	70.3	-	-	-	69.3	-	-	-
Black	10.7	-	-	-	10.7	-	-	-
Latino	12.2	-	-	-	12.8	-	-	-
Other	6.8	-	-	-	7.3	-	-	-
Education (%)
< HS grad	12.4	7.7	15.4	33.5***	13.9	8.9	20.8	36.5***
HS grad	27.4	27.8	36.1	25.0	29.6	31.0	35.3	23.7
Some college	31.9	32.9	21.5	28.8	27.8	27.5	23.9	28.6
College grad or higher	28.4	31.6	27.0	12.7	28.7	32.6	20.0	11.2
Family income (%)
≤$40,000	41.1	35.8	64.6	53.1***	40.3	34.1	57.4	59.6***
$40,001–80,000	25.0	27.3	20.5	14.2	26.2	29.9	17.4	16.9
>$80,000	20.7	24.6	8.0	12.8	24.7	27.4	13.0	12.7
Missing	13.2	12.3	6.8	19.9	8.9	8.6	12.2	10.9
Employment (%)
Employed	52.0	54.7	48.7	42 8***	67.3	69.1	55.5	71.5***
Unemployed	20.6	18.2	33.5	23.8	21.1	17.8	35.5	21.6
Other	27.3	27.1	17.8	33.4	11.6	13.1	9.0	6.9
Marital Status (%)
Married	68.2	73.6	35.8	68.1***	68.0	70.3	55.7	72.1*
Never married	15.8	11.9	36.6	17.2	23.6	21.2	35.2	22.4
Separated/divorced/widow	15.9	14.6	27.7	14.6	8.4	8.5	9.1	5.6

Weighted results, unweighted Ns.

^*p<0.05

^**p<0.01

^***p<0.001, chi-square test between White, Black and Latino for men and women separately.

Table 2 presents means and correlations of the three outcomes and two key predictors. As expected, compared to White respondents, Black and Latino respondents of both genders experienced a greater number of serious impacts due to the recession. They also tended to have less frequent prior heavy drinking, with the exception of Latino men, whose prior heavy drinking was similar to that of White men. Poorer self-rated health was more strongly correlated with recession-related hardship among women (r=0.22) than men (r=0.07). Poorer self-rated health was correlated with depressive symptoms in all subgroups. Unexpectedly, there was a significant negative correlation between current heavy drinking and poorer self-rated health among women that indicated more frequent heavy drinking was associated with better health in women overall (r = −0.058) and in the subsample of Latina women specifically (r = −0.134). The correlations correspond to mostly small and some medium (such as associations of depressive symptoms with poorer self-rated health, recession index with depressive symptoms, and prior heavy drinking with current heavy drinking) effect sizes per Cohen ⁴⁷.

Table 2.

Means and Correlations of Outcomes with Predictor Variables

			Correlation
	Mean	SE	Poorer Self-rated health	Depressive symptoms	Recession index	Current heavy drinking
Women Overall
Poorer self-rated health	2.57	0.03
Depressive symptoms	4.83	0.15	0.463***
Recession index	0.73	0.03	0.217***	0.305***
Current heavy drinking	1.20	0.02	−0.058**	−0.025	0.077***
Prior heavy drinking	1.00	0.05	0.042*	0.097***	0.102***	0.389***
Women: White
Poorer self-rated health	2.47	0.04
Depressive symptoms	4.69	0.18	0.476***
Recession index	0.64	0.04	0.221***	0.304***
Current heavy drinking	1.23	0.02	−0.019	−0.029	0.067**
Prior heavy drinking	1.06	0.06	0.060*	0.102***	0.122***	0.411***
Women: Black
Poorer self-rated health	2.89	0.09
Depressive symptoms	5.56	0.47	0.483***
Recession index	0.85	0.09	0.205***	0.369***
Current heavy drinking	1.14	0.05	−0.017	0.156***	0.230***
Prior heavy drinking	0.77	0.10	0.073	0 191***	0.078	0.308***
Women: Latina
Poorer self-rated health	2.83	0.07
Depressive symptoms	5.26	0.31	0.353***
Recession index	1.04	0.07	0.206***	0.260***
Current heavy drinking	1.13	0.03	−0.134***	−0.047	0.005
Prior heavy drinking	0.77	0.10	−0.012	0.021	0.045	0.209***
Men Overall
Poorer self-rated health	2.42	0.04
Depressive symptoms	4.03	0.17	0.404***
Recession index	0.71	0.04	0.068**	0.209***
Current heavy drinking	1.28	0.03	−0.025	0.056*	0.091***
Prior heavy drinking	1.99	0.07	0.050*	0.121***	0.015	0.377***
Men: White
Poorer self-rated health	2.36	0.05
Depressive symptoms	3.92	0.20	0.448***
Recession index	0.64	0.04	0.051	0.222***
Current heavy drinking	1.32	0.03	−0.048	0.031	0.090**
Prior heavy drinking	2.11	0.08	0.059	0.137***	0.031	0.433***
Men: Black
Poorer self-rated health	2.53	0.11
Depressive symptoms	4.34	0.45	0.314***
Recession index	0.98	0.15	0.021	0.256***
Current heavy drinking	1.19	0.07	0.102	0.190**	0.257***
Prior heavy drinking	1.37	0.19	0.011	0.058	0.052	0.079
Men: Latino
Poorer self-rated health	2.74	0.10
Depressive symptoms	4.66	0.39	0.375***
Recession index	0.97	0.11	0.187**	0.250***
Current heavy drinking	1.25	0.06	0.071	0.169**	0.054
Prior heavy drinking	2.02	0.18	0.087	0.086	−0.016	0.316***

Weighted results.

^*p<0.05

^**p<0.01

^***p<0.001.

“Other” race/ethnicity was dropped due to the small sample size.

3.2. Results from Path Models

Tables 3 and 4 show full path models for the overall samples of men and women, respectively. Excellent model fit was achieved for both gender subgroups (see table notes). Figures 1 and 2 present overall coefficients along with results of gender-specific analyses testing differences across White, Black, and Latino subgroups.

Table 3.

Results from Path Models for the Total Male Sample

	Depressive Symptoms			Current Heavy Drinking Frequency			Poorer self-rated health
	Coef	SE	P	Coef	SE	P	Coef	SE	P
Current heavy drinking	-	-	-	-	-	-	−0.017	0.046	0.705
Depressive symptoms	-	-	-	-	-	-	0.080	0.007	<0.001
Recession index	0.701	0.154	<0.001	0.139	0.060	0.020	−0.027	0.035	0.443
Prior heavy drinking	0.205	0.082	0.012	0.368	0.041	<0.001	−0.007	0.025	0.772
Prior chronic disease							0.590	0.083	<0.001
Prior alcohol health harm	1.160	0.370	0.002				0.326	0.079	<0.001
Age (ref 20–34)
35–49				−0.632	0.165	<0.001	0.216	0.112	0.055
50–75				−0.727	0.190	<0.001	0.195	0.117	0.094
Race/ethnicity (ref White and others)
Black				−0.201	0.218	0.357	0.051	0.108	0.633
Latino				−0.330	0.180	0.067	0.241	0.106	0.023
Education (ref < HS grad)
HS grad	−0.992	0.498	0.046	−0.630	0.216	0.004	−0.070	0.115	0.544
Some college	−1.618	0.524	0.002	−0.388	0.215	0.071	−0.196	0.122	0.109
College grad or higher	−1.355	0.542	0.012	−0.367	0.225	0.103	−0.395	0.124	0.002
Family income (ref ≤ $40,000)
$40,001–80,000	−0.358	0.403	0.374	0.268	0.172	0.119	−0.211	0.089	0.017
>$80,000	−0.973	0.475	0.041	0.453	0.202	0.025	−0.202	0.106	0.057
Missing	−0.526	0.626	0.401	0.299	0.299	0.317	−0.156	0.121	0.197
Employment (ref employed)
Unemployed	1.187	0.378	0.002	0.057	0.180	0.753	0.162	0.088	0.065
Others	−0.610	0.458	0.183	−0.469	0.277	0.091	0.017	0.096	0.859
Marital Status (ref married)
Never married	0.396	0.439	0.366	0.451	0.171	0.008
Separate/divorced/Widowed	2.077	0.443	<0.001	0.624	0.213	0.003
R-square		0.144			0.454			0.288

Covariance Current Heavy Drinking and Depressive Symptoms: −0.109 (0.197)

Model Fit: RMSEA = 0.000 (90% CI: 0.000–0.020), CFI = 1.000, TLI = 1.052

Table 4.

Results from Path Models for the Total Female Sample

	Depressive symptoms			Current Heavy Drinking Frequency			Poorer self-rated health
	Coef	SE	P	Coef	SE	P	Coef	SE	P
Current heavy drinking	-	-	-	-	-	-	−0.074	0.042	0.078
Depressive symptoms	-	-	-	-	-	-	0.065	0.005	<0.001
Recession index	1.112	0.114	<0.001	0.119	0.056	0.032	0.083	0.025	0.001
Prior heavy drinking	0.204	0.086	0.017	0.324	0.036	0.000	0.043	0.024	0.068
Parental alcohol problem				0.298	0.127	0.019	0.195	0.064	0.002
Prior chronic disease	1.135	0.298	<0.001				0.486	0.056	<0.001
Prior alcohol health harm	0.913	0.367	0.013				0.236	0.085	0.005
Age (ref 20–34)
35–49				−0.578	0.147	<0.001
50–75				−0.687	0.157	<0.001
Race/ethnicity (ref White and others)
Black				−0.390	0.187	0.037
Latina				−0.225	0.157	0.153
Education (ref < HS grad)
HS grad	−0.910	0.399	0.023	0.075	0.223	0.735	−0.400	0.090	<0.001
Some college	−1.485	0.420	<0.001	0.332	0.224	0.138	−0.465	0.095	<0.001
College grad or higher	−1.945	0.466	<0.001	0.457	0.237	0.054	−0.650	0.102	<0.001
Family income (ref ≤ $40,000)
$40,001–80,000	−0.898	0.370	0.015	0.357	0.157	0.023	−0.219	0.073	0.003
>$80,000	−0.974	0.421	0.021	0.306	0.182	0.093	−0.350	0.084	0.000
Missing	−0.346	0.406	0.394	0.122	0.211	0.562	−0.072	0.078	0.357
Employment (ref employed)
Unemployed	2.678	0.333	<0.001	−0.187	0.149	0.211	0.356	0.069	<0.001
Others	0.316	0.327	0.334	−0.277	0.153	0.070	0.129	0.065	0.047
Marital Status (ref married)
Never married	0.560	0.443	0.206	0.460	0.185	0.013
Separate/divorced/Widowed	0.782	0.299	0.009	−0.121	0.150	0.418
R-square		0.221			0.417			0.371

Covariance Current Heavy Drinking and Depressive Symptoms: −0.318 (0.190)

Model Fit: RMSEA = 0.004 (90% CI: 0.000–0.019), CFI = 0.999, TLI = 0.996

Figure 1.

Coefficients and racial/ethnic differences from the men’s overall model

Figure 2.

Coefficients and racial/ethnic differences from the women’s overall model

For all three racial/ethnic subgroups of men, recession-related hardships were positively associated with both depressive symptoms and current heavy drinking, and depressive symptoms, in turn, were associated with poorer self-rated health. Current heavy drinking was associated with poorer self-rated health only for Black men, not White or Latino men, nor men overall. Prior heavy drinking was positively associated with depressive symptoms and current heavy drinking for all groups of men.

Similarly, for all three racial/ethnic groups of women, recession hardships were positively associated with depressive symptoms and current heavy drinking, and depressive symptoms were related to poorer self-rated health. Current heavy drinking was not associated with poorer self-rated health in the overall women’s model, nor for White women. It was inversely associated with poorer self-rated health in Black and Latina women, indicating that more frequent heavy drinking was related to better self-rated health. As found for men, prior heavy drinking was positively related to both depressive symptoms and current heavy drinking in women overall, but multi-group analysis revealed prior heavy drinking was only related to depressive symptoms in Black and Latina women.

3.2.1. Indirect effects

For men, the overall path analysis showed a significant indirect pathway from recessionary hardships to poorer self-rated health through depressive symptoms (estimate=0.056, standard error (SE) = 0.013) which was confirmed by multi-group analysis to be invariant across racial/ethnic groups of men (estimate (SE) = 0.062 (0.013)) for all subgroups). While we found racial/ethnic differences in the association of current heavy drinking with poorer self-rated health, there were no significant indirect effects of the recession on poorer self-rated health via current heavy drinking for any male subgroup. The overall model also indicated a significant pathway from prior heavy drinking to poorer self-rated health through depressive symptoms for men (estimate (SE) = 0.016 (0.007)), confirmed to be invariant across racial/ethnic groups (estimate (SE) = 0.014 (0.006) for all subgroups). In addition, multi-group analysis indicated a significant indirect effect of prior heavy drinking on poorer self-rated health through current heavy drinking in Black men only (estimate (SE) = 0.052 (0.019)).

Among women, and as seen for men, overall path analysis (Table 4) revealed a significant indirect pathway from recession hardships to poorer self-rated health through depressive symptoms (estimate (SE) = 0.072 (0.009)), but not through current heavy drinking; this pathway through depression did not vary across racial/ethnic groups (see Figure 1). In contrast to results for men, a significant direct effect of recession hardships on poorer self-rated health was found across female subgroups (Figure 2). With regard to alcohol (and as seen among men), there was a significant indirect effect of prior heavy drinking on poorer self-rated health through depressive symptoms that was invariant across Black and Latina women (estimate (SE) = 0.029 (0.009) for both groups), but not found for White women. Finally, we observed a surprising, significant indirect pathway from prior heavy drinking to self-rated health through current heavy drinking that was invariant for Black and Latina women (estimate (SE) = −0.034 (0.014) for both groups), but not found for White women; it suggests a pathway from prior heavy drinking to current heavy drinking was associated with better health status for these groups.

4. Discussion

To our knowledge, this is the first study to examine US racial/ethnic differences in behavioral and overall health outcomes associated with accumulated, economic hardships experienced by men and women during the Great Recession. Findings indicate certain recession-related associations and indirect pathways are shared by all three racial/ethnic groups, but some paths appear to be unique to Black and Latino subgroups. As expected based on the stress process model and in keeping with prior reviews ^48,49 and longitudinal findings ^24,50,51, recession impacts were associated with both poorer mental health and more frequent heavy drinking. Further, through negative effects on mental health, recessionary hardships were indirectly related to poorer health, consistent with findings from a longitudinal study of home foreclosure ⁵².

Unexpectedly, associations between recession impacts, behavioral health and self-rated health did not differ significantly across racial/ethnic groups. This may be due, in part, to controlling for education and family income. According to the stress process model, socioeconomic position influences resources available for coping with stressors. Controlling for heavy drinking, alcohol-related health harm and health conditions prior to the recession might further reduce racial/ethnic variability in the recession’s effects on drinking and overall health. It is very important, however, to view our finding of similar associations across racial/ethnic groups in the context of these groups’ differential exposure to severe economic impacts during the 2007–9 recession. Considering this, our study suggests the population-level health impacts of the Great Recession were worse for Black and Latino communities due to disparities in exposure to recessionary hardships.

Observed racial/ethnic differences in paths are notable. We found two pathways to poorer self-rated health, including one involving recessionary events and mental health that was shared by all subgroups. A second pathway involved on-going heavy drinking, but only for Black men, which was surprising as their current heavy drinking frequency was similar to or less than that of White and Latino men. One possible explanation is that as Black men age, they may be more vulnerable to alcohol-related health consequences despite similar (or less) heavy drinking than other groups⁵³ due to greater chronic, cumulative stress that can profoundly harm health ^54,55. Another finding, observed only for Latino men, was that current heavy drinking and depressive symptoms remained strongly correlated even when accounting for all paths and covariates in the model. This might reflect exposure to factors not examined here, such as acculturative stress, anti-immigrant discrimination and cultural drinking norms, which can shape Latino men’s mental health or drinking behaviors ^22,56. To the extent that heavy drinking exacerbates depressive symptoms in Latino men, as found in general population and other samples ^57–59, heavy drinking could indirectly increase Latino men’s risk for poorer self-rated health during a severe recession.

Several differences observed for women (as compared to men) and racial/ethnic female subgroups are notable. First, effects of recession hardships on poorer self-rated health were only partially explained by depressive symptoms. Other mental health consequences of the recession, such as increased anxiety, might contribute to poorer self-rated health, which suggests the value of using a more comprehensive measure of mental health in future studies. Second, prior heavy drinking was strongly and directly related to Black and Latina (but not White) women’s depressive symptoms, suggesting Black and Latina women with a history of heavy drinking may be especially at risk for depressive symptoms associated with severe recession impacts. Finally, controlling for recession impacts, behavioral health and demographics, both Black and Latina women’s current heavy drinking was unexpectedly associated with better self-rated health. This finding could reflect a variant of the “sick quitter” phenomenon ⁶⁰, whereby drinkers with health conditions aggravated by alcohol give up drinking. In this case, Black and Latina women with health conditions may be more likely than White women to avoid heavy drinking, given indications of less permissive drinking norms among Black and Latina women ⁶¹.

The racial/ethnic differences observed here raise questions for further study. They also suggest the application of a stress process approach in alcohol disparities research could benefit from consideration of cultural drinking norms and attitudes, as noted by others ⁶².

4.1. Study Strengths and Limitations

Several strengths distinguish this study from prior research. First, our measure of recession-related impacts was based on a variety of household experiences rather than state or county-level unemployment rates. Second, we assessed differential population impacts by comparing racial/ethnic groups, stratified by gender. To our knowledge, this is the first US study to examine intersections of race/ethnicity and gender in relation to health effects of the 2007–9 recession using national survey data. Finally, our analysis accounted for potentially important alcohol and health confounders occurring prior to the recession.

Along with these strengths, several limitations should be recognized. First is the use of cross-sectional data, which preclude causal inference. Despite the lack of prospective, longitudinal data on depressive symptoms, heavy drinking and self-rated health, our cross-sectional results are highly consistent with those from rare longitudinal studies of the Great Recession described above. Our use of only landline data is another limitation. Prevalence of general negative effects of the recession (based on a single item) was slightly higher in the NAS cell phone vs. landline sample (57% vs. 51%); however, associations of this item with drinking outcomes were consistent across the two groups. Other data limitations include use of an abbreviated version of the CES-D, which limits comparability with other studies, although the sensitivity and specificity are similar to the full CES-D ⁶³, as well as lack of data on depressive symptoms prior to the recession and alternative mental health indicators post-recession. Finally, although self-reported health is a widely-used measure, use of a single item tends to limit test-retest reliability; however, dichotomizing the measure (as we have done) might help mitigate this ⁶⁴. Additionally, some research suggests subjective health ratings can underestimate poor health in disadvantaged groups and minimize health disparities ^65,66. Our findings for self-reported health therefore may not fully capture racial/ethnic disparities. Nonetheless, programmatic and policy implications of the current results remain relevant.

4.2. Conclusions

This study suggests that experiencing severe recessionary events is directly associated with increased heavy drinking and indirectly associated with poorer overall health through negative effects on mental health. These findings have implications for the present economic crisis during the COVID-19 pandemic. As in the Great Recession, communities of color have been disproportionately impacted by job loss ^67,68. Interventions that bolster the labor market and increase re-employment, together with social safety net programs that reduce income loss and alleviate food and housing insecurity, could help cushion long-term population impacts ⁶⁹. Special efforts focused on the most severely impacted communities might mitigate against widening economic and health disparities.

HIGHLIGHTS.

• Recession hardships were associated with depressive symptoms and, in turn, poorer health

• Recession hardships were also associated with more frequent, heavy drinking

• These relationships were found for all race/ethnicity-by-gender subgroups

• Racial/ethnic minorities showed additional paths to behavioral and health outcomes

• Structural and behavioral health interventions are needed during severe recessions