Psychological distress and alcohol use disorder during the COVID-19 era among middle- and low-income U.S. adults

Abstract

Purpose

This study examined how exposure to events during the Coronavirus Disease-19 (COVID-19) era is linked to symptoms of major depressive disorder (MDD), generalized anxiety disorder (GAD), COVID-19 era-related stress (CS), alcohol use disorder (AUD), and suicidal ideation (SI) in low and middle-income U.S adults.

Methods

A national sample of 6,607 adults (4.4% who reported testing positive for COVID-19, 25.3% testing negative, and 70.3% untested) were recruited an online platform andcompleted the Patient Health Questionnaire-2, Generalized Anxiety Disorder-2, PTSD-Checklist for DSM-5, the Alcohol Use Disorder Identification-Consumption scale, and an item assessing SI in May-June 2020. A series of multivariable analyses were conducted.

Results

In the total sample, 35.3% screened positive for current AUD, 33.6% for MDD, 33.6% for GAD, 24.6% for SI, and 20.2% for CS. Past 2-week SI (adjusted odds ratios [AORs]= 1.49–12.06), number of close friends (AORs= 1.40–2.72), history of AUD (AORs= 1.15–1.92), history of anxiety disorder (AORs= 1.07–2.63), and younger age (AORs= 0.97–0.98) were significantly associated with screening positive for MDD, GAD, CS, and AUD. COVID-19 status was not independently associated with these conditions, but the combination of testing positive for COVID-19, meeting criteria for AUD, and screening positive for MDD, GAD, or CS predicted a 96% probability for SI.

Conclusion

Predisposing factors are stronger predictors of psychological distress than personal COVID-19 infection or exposure. The additive effects of COVID-19 infection, alcohol use, and psychiatric problems in predicting SI suggest screening, monitoring, and treating these conditions in population-based prevention and treatment efforts may be important.

The period of outbreaks of the novel coronavirus disease 2019 (COVID-19), which has been referred to as the “COVID-19 era,” has given rise to a series of major life stressors, including a global pandemic, stay-at-home orders, and racially charged events. This pandemic has consequently resulted in substantial morbidity and mortality worldwide (McKee and Stuckler, 2020; Tsai and Wilson, 2020), and many have commented on panic and anxiety experienced by the general population as a result of the COVID-19 pandemic (Kickbusch et al., 2020; Torales et al., 2020). The negative effects of the pandemic has disproportionately impacted older, middle- and low-income adults in terms of food insecurity, housing instability, access to healthcare, and income disparities(Li and Mutchler, 2020; Memmott et al., 2021; Tsai et al., 2020; Wolfson and Leung, 2020). Deleterious effects of the pandemic have also been observed among low-income students and younger adults (Rudenstine et al., 2021).

While a growing number of studies have characterized the prevalence of psychological distress during the COVID-19 era (Krishnamoorthy et al., 2020; Salari et al., 2020), the majority of these studies have focused on healthcare workers rather than the general population, and scarce research has specifically examined factors associated with symptoms of major depressive disorder (MDD), generalized anxiety disorder (GAD), alcohol use disorder (AUD), and COVID-19 era-related stress (CS) during this period especially in middle and low-income populations which have been disproportionately impacted by COVID-19 (Abrams and Szefler, 2020; Blundell et al., 2020).

A burgeoning body of literature has documented that increases in depression, anxiety, and stress-related disorders such as posttraumatic stress disorder (PTSD) are common after catastrophic events, such as natural disasters and terrorist attacks (Galea et al., 2002; Kar and Bastia, 2006; Krug et al., 1998; North et al., 2004; Parker et al., 2016). More relevant to the COVID-19 pandemic, studies on two previous coronavirus outbreaks, the Severe Acute Respiratory Syndrome (SARS) and the Middle East Respiratory Syndrome (MERS), found high rates of PTSD, anxiety, and depressive symptoms (Carmassi et al., 2020; Lee et al., 2007, 2019; Mak et al., 2009; Wu et al., 2005). Greater severity of these symptoms was linked to high perceptions of life threat, low emotional support, working as a healthcare provider, and having significant others infected with the virus (Carmassi et al., 2020; Lee et al., 2007; Wu et al., 2005). A recent meta-analytic review of 72 studies that assessed psychiatric presentations of SARS, MERS, and COVID-19 (Rogers et al., 2020) reported that while the majority of patients recovered without experiencing mental illness, a substantial proportion did experience mental illness in the post-illness stage, with estimated point prevalence rates of 32.2% for PTSD, 14.9% for depression, and 14.8% for anxiety disorders. The review concluded that the impact of COVID-19 on mental health could be considerable, and that clinicians should prepare for MDD, GAD, and PTSD that can result in the aftermath of the COVID-19 pandemic. Other research has shown that to cope with these mental disorders, some individuals self-medicate with alcohol (Leeies et al., 2010) and risk becoming heavy drinkers, and scholars have portended a marked rise in alcohol abuse in the context of the COVID-19 pandemic (Clay and Parker, 2020). Finally, a systematic review focused on emerging viral disease outbreaks and suicide identified four studies that reported a slight but significant increase in suicide deaths during outbreaks, but concluded the evidence was scarce and weak (Leaune et al., 2020). Of note, the aforementioned studies focused mostly on the viral outbreaks and did not account for the other unique circumstances of the COVID-19 era, such as indirect effects of stay-at-home orders and racial strife.

The development of disorders such as MDD, GAD, PTSD, and AUD depends on individual variability in a host of risk factors (Dobson and Dozois, 2008; Rapee, 2001; Tsai et al., 2017). First, there are sociodemographic factors that may increase risk for developing mental disorders (Kessler et al., 2012; Skapinakis et al., 2013). Second, there are physical and mental health factors, such as good physical health, which may help promote psychological resilience (Fox, 1999; Penedo and Dahn, 2005); and prior mental health conditions that can increase risk for the development of mental disorders (Brewin et al., 2000; Reiss, 1997; Rudolph et al., 2009; Xue et al., 2015). Third, psychosocial factors, such as the extent to which one perceives threat and control in a given situation, as well as the amount of social support that is available, can determine risk for mental disorders (Berkman, 1995; Holbrook et al., 2001; Pinto et al., 2015; Tsai et al., 2012; Uchino, 2009). It may also be important to differentiate internalizing (e.g., depression and anxiety) from externalizing pathology (e.g., alcohol abuse) in its effect on overall functioning and suicide risk (Achenbach et al., 2016; Soto‐Sanz et al., 2019).

Aims of the study

In the current study, we analyzed data from a nationally sample of low and middle-income U.S. adults during the COVID-19 era to: 1) compare the prevalence of positive screens for mental illness and alcohol use disorder; 2) identify sociodemographic, clinical, and psychosocial factors associated with positive screens for mental illness and alcohol use disorder; and 3) examine the interactive effects of testing positive for COVID-19 with internalizing and externalizing psychopathology on risk for suicidal ideation.

Materials and methods

A national sample of 6607 low and middle-income U.S. adults was recruited in May-June 2020 to examine health and social well-being during the COVID-19 pandemic. Eligibility for the study were adults who were at least 22 years old, lived in the U.S., and reported an annual personal gross income of $75,000 or less. Participants were recruited through Amazon Mechanical Turk (MTurk), an online labor market with over 500,000 participants across 200 countries that has become a popular method for conducting surveys and online interventions. To ensure data quality, only participants who had completed ≥50 approved previous Human Intelligence Tasks (HITs) and had an HIT approval rating ≥50% were invited. Cross-sample investigations have demonstrated that data obtained from MTurk is the same level of quality or higher than data collected from traditional subject pools such as community samples, college students, and professional panels (Kees et al., 2017; Mason and Suri, 2012).

A total of 9760 individuals initially agreed to participate in the study and 6762 (69.3%) met the eligibility criteria. An additional 155 workers were removed due to failing a validity check, i.e., they failed three items from the validity scales from the Minnesota Multiphasic Personality Inventory-2 (MMPI-2), which included items: “It would be better if almost all laws were thrown away,” (“yes” response was a validity failure) “Sometimes when I am not feeling well I am irritable,” (“no” was validity failure) and “Once in a while I put off until tomorrow what I ought to do today” (“no” was validity failure). The final study sample consisted of 6607 participants (67.7% of eligible sample) from all 50 U.S. states and the District of Columbia. The recruitment period lasted from May 20-June 8, 2020. To maximize generalizability of our findings, we used raking procedures to create sample weights representative of the U.S. population who was 22 years or older with annual personal income of $75,000 or below, which were the study inclusion criteria. We used data from 2018 American Community Survey to compute poststratification weights, so that inferential analyses yielded estimates comparable to the target population with respect to age, sex, race, ethnicity, and geographic region. All participants provided informed consent and study procedures were approved by the institutional review board at the University of Texas Health Science Center at Houston.

Measures

Sociodemographic information was assessed using a sociodemographic questionnaire. Veteran status was defined as “ever served on active duty in the U.S. military” and history of homelessness was defined as: “did not have a stable night-time residence (such as staying on streets, in shelters, cars, etc.).”

COVID-19 testing and infection status was assessed by asking participants whether they have been tested for COVID-19 and what the outcome was (i.e., positive, negative, not tested). They were also asked whether anyone close to them (e.g., friends, family) had tested positive for COVID-19.

Social connectedness was assessed with the Medical Outcomes Study (MOS) Social Support Survey-Short Form (Holden et al., 2014) and a question about the number of close friends and relatives that participants have.

Physical health status was assessed by asking participants whether they have ever been diagnosed with any of 22 different medical conditions (e.g., cancer, heart disease, arthritis) and the total number of medical conditions was summed (Thomas et al., 2017).

Psychiatric history was assessed by asking participants whether they have ever been diagnosed with any of 9 psychiatric or substance use disorders. Current mental health and substance use was assessed with validated measures, including the Patient Health Questionnaire-2 (PHQ-2; Kroenke et al., 2003), the Generalized Anxiety Disorder-2 (GAD-2; Plummer et al., 2016), and the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C; Bush et al., 1998). Recent suicidal ideation (SI) was assessed with an item from the Mini-International Neuropsychiatric Interview (MINI; Sheehan et al., 1998) that asked participants whether they considered “hurting yourself, felt suicidal, or wish that you were dead” over the last 2 weeks. Responses were dichotomized into “Not at all” versus “Several days/More than half the days/Nearly every day.” Participants were also asked whether they used any illicit drugs in the past month. For this study, Cronbach's α= 0.83 for the PHQ-2, α= 0.84 for the GAD-2, and α= 0.74 for AUDIT-C.

Given that we assessed for COVID-19 testing/infection status and that Criterion A of the DSM-5 criteria for PTSD excludes natural illness, including viral infections, as a qualifying trauma,(American Psychiatric Association, 2013) we administered the PTSD Checklist for DSM-5 (PCL-5; Weathers et al., 2013) to assess more general COVID-19 era-related stress symptoms (CS). Participants were asked to refer to the COVID-19 era as an index stressor event (i.e., “thinking about your experience with COVID-19 and the current situation”) and to rate the degree to which they experienced each of the 20 symptoms over the past month on a scale of 0 (not at all) to 4 (extremely). Internal consistency of the COVID-19-indexed PCL-5 was excellent (Cronbach's α = 0.98). A supplementary item was administered to assess distress and dysfunction related to symptoms: “Did these reactions cause you distress or result in a failure to fulfill obligations at home, work, or school?” using the same 0–4 scale and one-month time frame. For this study, a positive screen for CS was determined by PCL-5 responses that included endorsement of at least one item from Criterion B, 1 item from Criterion C, 2 items from Criterion D, 2 items from Criterion E, and endorsement of Criterion G (American Psychiatric Association, 2013). Items rated at 2 (“Moderately”) or higher were considered indicative of positive symptom endorsement (National Center for Posttraumatic Stress Disorder, 2014).

Data analysis

Data analyses proceeded in five steps. First, participants were divided into those who reported they had tested positive for COVID-19 (COVID-19+), those who reported they had tested negative (COVID-19-), and those who reported they had been untested. Second, these groups were compared on sociodemographic, social, and clinical characteristics using analysis of variance (ANOVA) and chi-square tests. Post-hoc pairwise group comparisons were conducted with Tukey's Honestly Significant Difference (HSD) test and Chi-square tests. Third, hierarchical logistic regressions were conducted to identify characteristics associated with MDD, GAD, CS, and AUD. In these regressions, four blocks of variables were sequentially entered: the first block included COVID-19 era-related variables, the second block included sociodemographic variables, the third block included past psychiatric history variables, and the fourth block included current clinical and psychosocial variables. Odds ratios (ORs) with 95% confidence intervals were calculated for each variable along with total Nagelkerke R² and change in R² values for each block. Fourth, all these same independent variables were entered simultaneously to identify variables that were mostly significantly associated with MDD, GAD, CS, and AUD net of other variables. Multicollinearity analysis revealed that none of the independent variables had correlations greater than 0.7 and tolerance values were below 0.1 indicating no collinearity (Hair et al., 2014). Fifth, predicted probability analyses were conducted to examine the extent to which COVID-19 status, externalizing psychopathology (i.e., MDD, GAD, CS), and internalizing psychopathology (i.e., AUD) interacted to predict SI. Based on the derived models, the predicted probability of suicidal ideation was calculated as a function of testing positive for COVID-19, scoring above the cut off on the AUDIT-C for AUD, and meeting above criteria for MDD, GAD, or CS. Poststratification weights were applied to all inferential analyses to permit generalizability of results to the U.S. adult low and middle-income population.

Results

Bivariate analyses

In the total sample (n = 6607), 4.4% were in the COVID-19+ group, 25.3% in the COVID-19- group, and 70.3% in the untested group. As shown in Table 1 , there were significant bivariate differences on nearly every demographic, clinical, and psychosocial variable assessed. Not surprisingly, compared to the COVID-19- and untested groups, the COVID-19+ group was more likely to have somebody close to them test positive for COVID-19. The COVID-19+ group also reported that they perceived COVID-19 as less of a threat and rated the government's response to COVID-19 as more adequate than the COVID-19- and untested groups.

Table 1.

Demographic, clinical, and psychosocial characteristics of low and middle-income adults who tested positive, negative, or were untested for COVID-19 (N = 6607).

	COVID-19 Positive (n = 354) P	COVID-19 Negative (n = 1819) N	COVID-19 Untested (n = 4434) U	Test of difference	Pairwise comparisona
	Weighted Mean/SD or Raw n/Weighted%b	Weighted Mean/SD or Raw n/Weighted%	Weighted Mean/SD or Raw n/Weighted%	F, $\chi$
Demographic characteristics
Age	39.5 (12.0)	45.0 (15.0)	50.5 (18.5)	F(2, 9997)= 155.37**	U>N>P
Gender Male Female	251 (67.7) 103 (32.3)	926 (50.5) 889 (49.5)	1795 (38.2) 2618 (61.8)	$\chi \left(2\right)$= 232.45**	P>N>U>N>P
Race/Ethnicity White non-Hispanic Black non-Hispanic White Hispanic Black Hispanic Asian/Pacific Islander NA/ANc Other	133 (31.7) 54 (10.7) 74 (30.4) 58 (20.5) 19 (3.9) 15 (2.1) 1 (0.7)	1040 (54.3) 239 (11.0) 230 (17.2) 120 (8.8) 111 (3.1) 49 (1.0) 30 (4.5)	3154 (71.5) 328 (6.0) 374 (9.7) 93 (2.2) 355 (3.9) 44 (0.4) 86 (6.3)	$\chi \left(12\right)$= 916.93**	U>N>P N>P>U P>N>U P>N>U/P>N P>N>U>N>P
Education High school and below Some college Associate/Bachelors Advanced degree	14 (4.8) 28 (10.5) 147 (40.7) 165 (43.9)	135 (8.5) 335 (20.2) 920 (47.8) 429 (23.5)	402 (8.8) 939 (23.4) 2288 (49.3) 805 (18.5)	$\chi \left(6\right)$= 188.70**	U>N>P U>N>P U>N>P>N>U
Student status Not a student Part-time Full-time	109 (30.9) 63 (18.1) 182 (51.0)	1303 (74.8) 163 (7.8) 353 (17.4)	3598 (87.3) 295 (4.5) 541 (8.1)	$\chi \left(4\right)$= 524.37**	U>N>P>N>U P>N>U
Marital status Single D/S/Wd Married/LWPe	51 (10.0) 14 (4.1) 289 (85.8)	507 (21.7) 129 (10.8) 1183 (67.4)	1607 (26.9) 481 (25.3) 2346 (47.8)	$\chi \left(6\right)$= 563.82**	U>N>P U>N>P>N>U
Number of minors in household None One Two Three Four or more	57 (15.6) 155 (45.5) 101 (27.5) 31 (9.6) 10 (1.8)	774 (47.3) 568 (27.7) 371 (19.4) 74 (4.1) 32 (1.5)	2747 (70.7) 849 (15.2) 587 (10.1) 175 (2.8) 76 (1.2)	$\chi \left(8\right)$= 878.88**	U>N>P>N>U P>N>U P>N>U P>N>U
Work Status Full-time Half-time Unemployed R/D/Of Self-employed	292 (80.4) 37 (10.0) 6 (1.1) 11 (5.7) 8 (2.7)	1204 (60.8) 247 (13.3) 157 (6.8) 94 (12.0) 117 (7.2)	2245 (38.9) 600 (12.3) 715 (11.7) 453 (25.8) 421 (11.3)	$\chi \left(8\right)$= 665.31**	P>N>U N>U>P U>N>P U>N>P U>N>P
Personal income	$26,392.31 (25,810.40)	$36,472.68 (21,940.03)	$34,600.11 (21,452.38)	F(2,9997)= 40.27**	N>U>P
State of residence Northeast Midwest South West	76 (19.2) 45 (13.3) 130 (36.6) 103 (30.9)	341 (16.7) 300 (17.8) 692 (38.3) 486 (27.2)	820 (16.7) 935 (22.1) 1696 (38.0) 983 (23.2)	$\chi \left(6\right)$= 49.96**	P>N/U>N>P N>U>P>N>U
Veteran status	243 (69.4)	239 (15.6)	334 (11.8)	$\chi \left(2\right)$= 1058.61**	P>N>U
Social status
# of close friends	2.45 (0.41)g	1.65 (0.57)	1.37 (0.40)	F(2, 9997)= 1521.67**	P>N>U
Medical Outcomes Study Social Support Survey	22.8 (4.0)	21.3 (5.8)	21.1 (6.5)	F(2, 9997)= 14.72**	P>N>U
COVID-19 questionnaire
Have someone close test positive for COVID-19	302 (85.2)	420 (21.0)	888 (16.9)	$\chi \left(2\right)$= 1160.88**	P>N>U
How much of a threat?h	2.8 (0.8)	3.2 (0.9)	3.3 (0.9)	F(2, 9997)= 67.82**	U>N>P
Opinion about government responsei	3.0 (1.1)	2.7 (1.3)	2.3 (1.3)	F(2, 9997)= 146.74**	P>N>U
Physical and mental health conditions
# of medical conditions	3.6 (4.7)	2.0 (2.6)	2.0 (2.1)	F(2, 9997)= 93.98**	P>U>N
History of psychiatric disorders Schizophrenia-spectrum PTSD Alcohol use disorder Bipolar disorder Anxiety disorder Major depression Drug use disorder Traumatic brain injury	82 (22.0) 107 (30.4) 152 (42.9) 117 (32.6) 111 (32.2) 87 (26.8) 54 (15.5) 27 (8.4)	82 (4.6) 196 (10.8) 368 (19.4) 170 (9.6) 544 (28.2) 308 (15.7) 133 (6.5) 45 (2.2)	96 (1.8) 381 (6.8) 267 (5.5) 261 (4.9) 1387 (24.0) 718 (13.0) 140 (2.5) 61 (1.0)	$\chi \left(2\right)$ = 522.79** 298.46** 878.65** 498.44** 28.07 69.00** 225.62** 151.44**	P>N>U P>N>U P>N>U P>N>U P>N>U P>N>U P>N>U P>N>U
Positive screen for COVID-19 era-related stress symptoms	269 (75.5)	661 (32.4)	757 (12.3)	$\chi \left(2\right)$= 1336.10**	P>N>U
Positive screen for major depression	313 (88.4)	887 (45.6)	1445 (25.8)	$\chi \left(2\right)$= 945.32**	P>N>U
Positive screen for generalized anxiety disorder	306 (87.4)	876 (43.6)	1506 (26.7)	$\chi \left(2\right)$= 831.78**	P>N>U
Past 2-week suicidal ideation	322 (90.8)	804 (38.8)	938 (15.4)	$\chi \left(2\right)$= 1627.66**	P>N>U
Positive screen for AUD	274 (76.9)	798 (40.3)	1534 (31.0)	$\chi \left(2\right)$= 416.83***	P>N>U
Any illicit drug use in past month	266 (75.3)	430 (22.1)	779 (18.0)	$\chi \left(2\right)$= 803.17**	P>N>U

Note: *p< .05, **p<.001.

With respect to psychiatric history, relative to the COVID-19- and untested groups, the COVID-19+ group was more likely to have been diagnosed with a range of psychiatric disorders including schizophrenia-spectrum disorders, PTSD, bipolar disorder, anxiety disorder, major depression, alcohol and drug use disorders, and traumatic brain injury.

In terms of current psychiatric status, the COVID-19+ group was more likely to screen positive for MDD, GAD, CS, AUD, illicit drug use, and past 2-week SI than both the COVID-19- and untested groups. In the total sample, 2645 (33.6%) screened positive for MDD; 2688 (33.6%) for GAD; 1687 (20.2%) for C-PTSD; 2606 (35.3%) for AUD; and 2064 (24.6%) for SI. Compared to the COVID-19- group, the COVID-19+ group was more likely to screen positive for MDD (Δ42.8%), GAD (Δ43.8%), CS (Δ43.1%), SI (Δ52.0%), AUD (Δ36.6%), and any illicit drug use (Δ75.3). Among MDD, GAD, and CS, in the total sample, 47.3% screened positive for any of these three mental health conditions, 32.0% screened positive for two or more conditions, and 15.0% screened positive for all three conditions. Comparing the COVID-19 groups, 95.2% of the COVID-19+ group screened positive for any of the three mental health conditions, 87.4% screened positive for two or more, and 62.7% screened positive for all three conditions; this was significantly higher than the 54.3% in the COVID-19- group who screened positive for any of the three conditions, the 40.5% who screened positive for two or more, and 22.2% who screened positive for all three (all p's<0.001).

Multivariable analyses

Table 2 shows results of hierarchical logistic regression analyses including four blocks of variables associated with positive screens for MDD, GAD, CS, and AUD. These results were quite similar across these four conditions. For MDD, GAD, and CS, the first block of variables (COVID-19-related variables) explained 13–21% of the variance, the second block (sociodemographic variables) an additional 9–12% variance, the third block (past psychiatric history) an additional 5–11% variance, and the fourth block (current clinical and psychosocial variables) explained an additional 3–17% variance. Together, these blocks of variables explained 22–54% of the variance in MDD, GAD, CS, and AUD.

Table 2.

Hierarchical logistic regression with factors associated with positive screens for major depressive disorder (MDD), generalized anxiety disorder (GAD), COVID-19-era related-stress symptoms (CS), and alcohol use disorder (AUD) among low and middle-income adults.

	MDD Odds ratio (95% CI)	GAD Odds ratio (95% CI)	CS Odds ratio (95% CI)	AUD Odds ratio (95% CI)
First block: COVID-19-related variables
COVID-19 testing status Negative Positive Untested	Ref 5.19 [3.80, 7.09]*** 0.44 [0.40, 0.49]***	Ref 5.54 [4.10, 7.50]*** 0.49 [0.45, 0.54]***	Ref 3.71 [2.89, 4.75]*** 0.32 [0.29, 0.36]***	Ref 3.30 [2.58, 4.22]*** 0.70 [0.63, 0.77]***
Any friends/family COVID-19+	2.24 [2.01, 2.50]***	2.15 [1.93, 2.39]***	2.42 [2.14, 2.73]***	1.75 [1.58, 1.95]***
How much of a threat?j	0.89 [0.84, 0.94]***	0.98 [0.93, 1.04]	1.06 [0.99, 1.13]	0.89 [0.84, 0.94]***
Opinion about government responsek	1.13 [1.08, 1.17]***	1.06 [1.02, 1.10]**	1.29 [1.23, 1.35]***	1.05 [1.01, 1.08]*
Δ R2	.16	.13	.21	.07
F or $\chi$2	1228.19***	1019.20***	1440.01***	550.32***
Total Nagelkerke R2	.16	.13	.21	.07
Second block: Sociodemographic
Age	0.98 [0.97, 0.98]***	0.97 [0.97, 0.97]***	0.97 [0.96, 0.97]***	0.98 [0.98, 0.99]***
Male	1.08 [0.98, 1.20]	0.87 [0.78, 0.96]*	1.17 [1.04, 1.32]*	0.94 [0.85, 1.03]
Race/Ethnicity White Black Asian Other	Ref 1.36 [1.17, 1.56]*** 0.82 [0.64, 1.05] 0.84 [0.69, 1.02]	Ref 1.39 [1.21, 1.61]*** 0.77 [0.60, 0.98]* 0.66 [0.54, 0.80]***	Ref 1.62 [1.38, 1.89]*** 1.06 [0.80, 1.39] 0.52 [0.39, 0.69]***	Ref 1.03 [0.90, 1.18] 0.56 [0.44, 0.72]*** 0.84 [0.70, 1.02]
Education level	1.11 [1.05, 1.17]**	1.09 [1.03, 1.15]*	1.15 [1.07, 1.23]***	1.06 [1.01, 1.12]*
Student status	1.35 [1.26, 1.45]***	1.43 [1.33, 1.53]***	1.52 [1.41, 1.64]***	1.29 [1.20, 1.38]***
Marital Status Married/Living with partner Single D/S/Wl	Ref 0.92 [0.81, 1.04] 1.25 [1.08, 1.44]**	Ref 0.69 [0.61, 0.79]*** 1.03 [0.89, 1.19]	Ref 0.62 [0.53, 0.73]*** 1.07 [0.88, 1.31]	Ref 0.82 [0.73, 0.92]** 0.95 [0.83, 1.08]
Number of minors in household	1.25 [1.18, 1.31]***	1.19 [1.13, 1.25]***	1.24 [1.17, 1.32]***	1.12 [1.06, 1.18]***
Employment Status Working half-time/full-time Self-employed Not working	Ref 0.62 [0.52, 0.74]*** 0.69 [0.61, 0.79]***	Ref 0.66 [0.55, 0.79]*** 0.71 [0.63, 0.81]***	Ref 0.62 [0.49, 0.78]*** 0.59 [0.50, 0.69]***	Ref 1.29 [1.10, 1.50]** 0.74 [0.66, 0.84]***
Annual personal income	1.00 [1.00, 1.00]***	1.00 [1.00, 1.00]***	1.00 [1.00, 1.00]***	1.00 [1.00, 1.00]***
Veteran status	2.08 [1.79, 2.41]***	2.06 [1.77, 2.39]***	1.73 [1.46, 2.04]***	2.65 [2.31, 3.03]***
Δ R2 since first block	.12	.14	.14	.09
F or $\chi$2	2240.74***	2165.19***	2522.91***	1291.47***
Total R2or Nagelkerke R2	.28	.27	.35	.17
Third block: Psychiatric history
History of schizophrenia-spectrum disorder	2.60 [1.78, 3.78]***	2.57 [1.77, 3.73]***	1.26 [0.92, 1.73]	1.00 [0.76, 1.33]
History of PTSD	1.59 [1.33, 1.91]***	1.75 [1.46, 2.10]***	1.84 [1.52, 2.22]***	1.04 [0.88, 1.23]
History of bipolar disorder	1.68 [1.36, 2.07]***	2.28 [1.84, 2.82]***	1.99 [1.62, 2.45]***	1.05 [0.86, 1.26]
History of anxiety disorder	1.80 [1.61, 2.02]***	2.52 [2.25, 2.82]***	1.67 [1.47, 1.91]***	1.03 [0.93, 1.15]
History of major depression	2.33 [2.02, 2.68]***	1.72 [1.50, 1.98]***	1.11 [0.94, 1.30]	0.97 [0.85, 1.12]
History of AUD	3.88 [3.23, 4.65]***	3.47 [2.90, 4.15]***	2.24 [1.89, 2.67]***	2.47 [2.10, 2.91]***
History of drug use disorder	1.48 [1.11, 1.97]*	1.73 [1.30, 2.30]***	1.42 [1.08, 1.85]*	1.41 [1.10, 1.81]*
Δ R2 since second block	.09	.11	.05	.02
F or $\chi$2	3095.39***	3168.20***	2911.68***	1457.71***
Total R2or Nagelkerke R2	.37	.38	.40	.19
Fourth block: Current psychosocial status
# of close friends	1.40 [1.18, 1.66]***	1.95 [1.65, 2.30]***	2.72 [2.30, 3.22]***	1.56 [1.35, 1.79]***
Medical Outcomes Study Social Support Survey	0.93 [0.92, 0.94]***	0.95 [0.94, 0.96]***	0.97 [0.96, 0.98]***	1.00 [1.00, 1.01]
# of medical conditions	1.13 [1.09, 1.17]***	1.08 [1.04, 1.12]***	0.97 [0.94, 1.01]	0.95 [0.93, 0.98]***
Any illicit drug use in past month	1.25 [1.07, 1.45]*	1.17 [1.00, 1.35]	1.30 [1.11, 1.53]**	1.94 [1.72, 2.19]***
Any past 2-week suicidal ideation	12.06 [10.25, 14.19]***	9.01 [7.69, 10.55]***	5.38 [4.58, 6.32]***	1.49 [1.30, 1.71]***
Δ R2 since third block	.17	.13	.10	.03
F or $\chi$2	4858.54***	4576.91***	3826.94***	1751.34***
Total R2or Nagelkerke R2	.54	.51	.50	.22

Note: *p<.05, **p<.01, ***p<.001.

All variables were then entered simultaneously in logistic regression analyses to identify variables that were independently associated with MDD, GAD, CS, and AUD (see Table 3 ). The results revealed that the variables that were most consistently and strongly associated with all four conditions were past 2-week SI (adjusted odds ratios [AORs]= 1.49–12.06), number of close friends (AORs= 1.40–2.72), history of AUD (AORs= 1.15–1.92), history of anxiety disorder (AORs= 1.07–2.63), and younger age (AORs= 0.97–0.98). For MDD and GAD, a history of schizophrenia-spectrum disorder was also associated with increased risk for these conditions (AORs= 1.35–1.43). For GAD and CS, a history of PTSD (AORs= 1.43–1.80) or bipolar disorder (AORs= 1.53–1.55) were associated with increased risk for these two conditions. For AUD, being a veteran (AOR= 2.31) and any recent illicit drug use (AOR= 1.94) was associated with increased risk for this condition. Across MDD, GAD, CS, and AUD, COVID-19+ status was not significantly associated with increased risk for these four conditions after controlling for all other variables (AORs= 1.25–1.32).

Table 3.

Logistic regression with simultaneous entry of variables associated with positive screens for major depressive disorder (MDD), generalized anxiety disorder (GAD), COVID-19-era related stress symptoms (CS), and alcohol use disorder (AUD) among low and middle-income adults.

	MDD Odds ratio (95% CI)	GAD Odds ratio (95% CI)	CS Odds ratio (95% CI)	AUD Odds ratio (95% CI)
COVID-19 testing status Negative Positive Untested	Ref 1.29 [0.87, 1.91] 0.82 [0.72, 0.94]**	Red 1.32 [0.91, 1.91] 1.01 [0.88, 1.15]	Ref 1.25 [0.92, 1.69] 0.66 [0.57, 0.76]***	Ref 1.28 [0.97, 1.69] 1.07 [0.96, 1.19]
Any friends/family COVID-19+	1.22 [1.05, 1.41]*	1.07 [0.93, 1.24]	1.19 [1.01, 1.39]*	1.18 [1.04, 1.32]*
How much of a threat?m	1.14 [1.05, 1.22]**	1.31 [1.22, 1.41]***	1.48 [1.36, 1.60]***	1.00 [0.94, 1.06]
Opinion about government responsen	1.05 [1.00, 1.10]	0.94 [0.89, 0.99]*	1.13 [1.07, 1.19]***	0.99 [0.95, 1.03]
Age	0.98 [0.98, 0.99]***	0.98 [0.97, 0.98]***	0.97 [0.97, 0.98]***	0.98 [0.98, 0.99]***
Male	0.81 [0.72, 0.92]**	0.63 [0.55, 0.71]***	0.88 [0.76, 1.02]	0.80 [0.72, 0.88]***
Race/Ethnicity White Black Asian Other	Ref 0.99 [0.83, 1.19] 1.02 [0.76, 1.37] 0.96 [0.76, 1.21]	Ref 1.03 [0.87, 1.23] 0.96 [0.72, 1.27] 0.75 [0.59, 0.95]*	Ref 1.15 [0.96, 1.38] 1.23 [0.90, 1.67] 0.64 [0.47, 0.87]**	Ref 0.82 [0.71, 0.96]* 0.57 [0.44, 0.74]*** 0.97 [0.80, 1.18]
Education level	1.02 [0.95, 1.10]	1.01 [0.94, 1.08]	1.02 [0.94, 1.11]	0.99 [0.94, 1.05]
Student status	1.07 [0.98, 1.18]	1.13 [1.03, 1.24]*	1.19 [1.09, 1.30]***	1.11 [1.03, 1.19]*
Marital Status Married/Living with partner Single D/S/Wo	Ref 0.97 [0.83, 1.13] 1.09 [0.92, 1.30]	Ref 0.75 [0.65, 0.87]*** 0.96 [0.81, 1.13]	Ref 0.77 [0.64, 0.92]** 1.25 [1.00, 1.56]	Ref 0.94 [0.82, 1.06] 0.96 [0.83, 0.10]
Number of minors in household	1.09 [1.02, 1.16]*	1.01 [0.95, 1.08]	1.08 [1.00, 1.16]	1.06 [1.00, 1.11]
Employment Status Working half-time/full-time Self-employed Not working	Ref 0.71 [0.57, 0.88]** 0.73 [0.63, 0.85]***	Ref 0.80 [0.65, 0.99]* 0.81 [0.70, 0.94]*	Ref 0.76 [0.58, 0.99]* 0.77 [0.64, 0.93]*	Ref 1.48 [1.26, 1.73]*** 0.84 [0.74, 0.95]*
Annual personal income	1.00 [1.00, 1.00]***	1.00 [1.00, 1.00]**	1.00 [1.00, 1.00]	1.00 [1.00, 1.00]***
Veteran status	1.24 [1.02, 1.51]*	1.23 [1.01, 1.49]*	0.80 [0.64, 1.00]*	2.31 [2.00, 2.68]***
History of schizophrenia-spectrum disorder	1.43 [0.94, 2.17]	1.35 [0.89, 1.98]	0.89 [0.64, 1.22]	0.84 [0.62, 1.14]
History of PTSD	1.20 [0.98, 1.47]	1.43 [1.18, 1.75]***	1.80 [1.47, 2.20]***	1.04 [0.87, 1.24]
History of bipolar disorder	1.00 [0.79, 1.28]	1.55 [1.21, 1.95]***	1.53 [1.23, 1.91]***	0.90 [0.731.09]
History of anxiety disorder	1.64 [1.43, 1.87]***	2.63 [2.31, 2.99]***	1.83 [1.57, 2.12]***	1.07 [0.96, 1.20]
History of major depression	1.71 [1.46, 2.01]***	1.24 [1.06, 1.45]*	0.90 [0.75, 1.07]	0.95 [0.82, 1.09]
History of AUD	1.78 [1.43, 2.21]***	1.54 [1.25, 1.89]***	1.15 [0.95, 1.39]	1.92 [1.62, 2.29]***
History of drug use disorder	1.19 [0.86, 1.63]	1.45 [1.06, 1.99]*	1.29 [0.97, 1.72]	1.30 [1.01, 1.68]*
# of close friends	1.40 [1.18, 1.66]***	1.95 [1.65, 2.30]***	2.72 [2.30, 3.22]***	1.56 [1.35, 1.79]***
Medical Outcomes Study Social Support Survey	0.93 [0.92, 0.94]***	0.95 [0.94, 0.96]***	0.97 [0.96, 0.98]***	1.00 [1.00, 1.01]
# of medical conditions	1.13 [1.09, 1.17]***	1.08 [1.04, 1.12]***	0.97 [0.94, 1.01]	0.95 [0.93, 0.98]***
Any illicit drug use in past month	1.25 [1.07, 1.45]*	1.17 [1.00, 1.35]*	1.30 [1.11, 1.53]**	1.94 [1.72, 2.19]***
Any past 2-week suicidal ideation	12.06 [10.25, 14.19]***	9.01 [7.69, 10.55]***	5.38 [4.58, 6.32]***	1.49 [1.30, 1.71]***
F or $\chi$2	4858.54***	4576.91***	3826.94***	1751.34***
Total R2or Nagelkerke R2	.53	.51	.50	.22

Note: *p<.05, **p<.01, ***p<.001.

a P= COVID-19 positive; N= COVID-19 negative; U= COVID untested.

b Values shown are weighted means and standard deviations (SD) for continuous variables, or raw counts and weighted% for categorical variables.

b NA/AN= Native American/Alaskan Native.

d D/S/W= Divorced/separated/widowed.

e LWP= Living with partner.

f R/D/O= Retired, disabled, others.

g Values for # of close friends shown are log-transformed, and actual means (standard deviations) for the groups are: COVID-19 Positive= 342.0 (140.9), COVID-19 Negative= 103.4 (152.4), and COVID-19 Untested= 35.1 (91.1).

h Threat was rated on a 4-point scale from 1 (Not a threat) to 4 (A great threat).

i Government response was rated on a 5-point scale from 1 (Great underreaction) to 5 (Great overreaction).

j Threat was rated on a 4-point scale from 1 (Not a threat) to 4 (A great threat).

k Government response was rated on a 5-point scale from 1 (Great underreaction) to 5 (Great overreaction).

l D/S/W= Divorced/separated/widowed.

m Threat was rated on a 4-point scale from 1 (Not a threat) to 4 (A great threat).

n Government response was rated on a 5-point scale from 1 (Great underreaction) to 5 (Great overreaction).

o D/S/W= Divorced/separated/widowed.

Predicted probabilities

Fig. 1 displays the predicted probabilities for SI among participants with combinations of three different factors: COVID-19 positive test results; positive screens for MDD, GAD, or CS; and positive screen for AUD. Testing positive for COVID-19 and screening positive for MDD, GAD, and/or CS was associated with a higher predicted probability for SI than either factor alone. Together with AUD, participants who had all three factors had a 96% predicted probability of SI.

Fig. 1.

Predicted probability of suicidal ideation as a function of COVID-19, alcohol use disorder (AUD), and major depressive disorder (MDD), generalized anxiety disorder (GAD), and COVID-19 era-related symptoms (CS).

Discussion

Using data from a national sample of low and middle-income U.S. adults, we found that the majority did not screen positive for MDD, GAD, CS, or AUD. However, nearly half (47.3%) of participants screened positive for MDD, GAD, or CS and there were major differences in rates of positive screens based on COVID-19 status. Among participants who reported COVID-19 infection, 88% screened positive for MDD, 87% for GAD, 77% for AUD, and 76% for CS, which were all much higher than those who tested negative for COVID-19 and those who were untested. However, after adjusting for other sociodemographic, clinical, and psychosocial characteristics, testing positive for COVID-19, as well as having a close friend or family member who tested positive were no longer associated with MDD, GAD, or CS. Instead, having a history of mental health problems, a greater number of close friends, and younger age were each independently associated with greater likelihood of screening positive for MDD, GAD, CS, and AUD. Together, these findings suggest that many low and middle-income Americans experienced significant psychological distress in the first five months of the COVID-19 pandemic in the U.S. They further suggest that predisposing factors are much stronger predictors of psychological distress than personal COVID-19 infection or having a significant other with COVID-19 infection.

Having a psychiatric history predisposed participants to screen positive for MDD, GAD, or CS which is consistent with the “stress sensitization” model of psychopathology, which posits that in response to repeated exposure to mental illness episodes and external stress, individuals become sensitized to stress over time so that the level of stress that triggers future mental illness episodes becomes increasingly lower (McLaughlin et al., 2010). It may be notable that there were mostly similar factors associated with MDD, GAD, and CS, which may be expected given that they share similar symptomatology (Gros et al., 2012; Møller et al., 2020), consistent with the possibility that general personal vulnerability to psychopathology could account for the association of pre-existing psychiatric illness with post-COVID-19 psychopathology. We found that 15% of our sample screened positive for all three conditions, underscoring a substantial proportion that experienced a high level of psychological distress. Of note, internalizing psychopathology (MDD, GAD, CS) interacted with externalizing behaviors (AUD) among adults who tested positive for COVID-19, such that adults with all three factors—COVID-19+ test, psychological distress, and positive screen for alcohol use disorder had a markedly high predicted probability of SI of 96%. In light of widespread concerns about rising rates of suicide during the COVID-19 era (Gunnell et al., 2020; Kawohl and Nordt, 2020; Sher, 2020), our findings highlight individual factors associated with suicide risk, as well as how interactions among these factors are linked to increased risk of suicidal thinking in the general adult population.

There are three key practice and policy implications of this study to conclude. First, we found high rates of psychological distress during the beginning of the COVID-19 era (i.e., May-June 2020), which may continue during the pandemic and may be particularly elevated for individuals with psychiatric histories. Healthcare providers and public health professionals should be aware of these increased vulnerabilities for mental health problems during the pandemic. Mental health screening and access to mental health and substance abuse treatment services perhaps may be facilitated at point-of-care for COVID-19 testing and results, or during contact tracing activities. Second, adequate tele-mental health and virtual care services may need to be developed and made available to treat individuals with psychological distress as the pandemic continues and social distancing measures remain in place. Policies that supply technology resources and support providers to deliver services virtually may need to be further enacted. Third, COVID-19 infection status, psychological distress, and alcohol use disorder should not be considered in isolation but rather the combination of these factors may have additive effects that are important to consider in a clinical context, especially as it pertains to potential risk for suicide. To the extent that research and policy in these areas can be synergized, a more holistic approach may be taken to understand the full sequalae of the pandemic. Taken together, the study findings point to important mental health practice and policy issues to consider when planning for the needs of low- and middle-income adults during the COVID-19 era and future pandemics.

Several study limitations are worth noting. First, COVID-19 test status was based on self-report and symptoms were not assessed clinically. Second, this was a cross-sectional study and so no inferences could be made about the directionality or causality of associations found. Third, all of the self-report symptom measures in this study collected data on current symptoms rather than on incident (new) symptoms after the start of the pandemic, and thus any current symptoms that were already present before the pandemic may be unrelated to the pandemic etiologically. They could be of importance because they may interfere with ability to cope with and function in the pandemic and warrant clinical intervention. This methodological issue may explain why COVID-19 exposures were not independently associated with psychiatric symptoms in regression models, but predisposing and other personal characteristics were. Fourth, we did not collect detailed data on specific COVID-19 era-related events that participants experienced, which may be differentially associated with psychological distress. Fifth, screening measures were used to assess MDD, GAD, CS, and AUD, which are not diagnostic instruments and should therefore not assumed to yield prevalence estimates of psychiatric disorders or relied on to guide treatment. However, like our study, most other COVID-19 mental health studies have also used self-report measures to assess psychological symptoms.

To date, there has been little guidance or consensus in the field on how to best assess COVID-19-related mental health syndromes, particularly PTSD, given the exclusion of naturally-occurring medical illnesses from categorization as trauma and hence from the diagnosis of PTSD according to DSM-5 criteria. Further research is needed to determine how to interpret these findings, and to address the diagnostic and nosological uncertainties surrounding trauma and PTSD in studying psychiatric consequences of the COVID-19 pandemic. Specifically, studies are needed to resolve whether psychiatric syndromes following exposure to COVID-19 are distinct or indistinguishable from those associated with traumatic events such as physical assaults or severe accidents to help inform decisions of how to best categorize and measure COVID-19-related PTSD and related syndromes.

These limitations notwithstanding, there were several strengths of the study, including the national diverse sample, the assessment of a broad range of psychosocial and mental health variables, and findings that may contribute to efforts to address negative mental health consequences in the aftermath of the COVID-19 pandemic. Further research is needed to replicate these results in other samples; identify biopsychosocial vulnerabilities and long-term mental health and substance abuse outcomes; and evaluate the efficacy of psychological treatments for COVID-19 era-related mental health and substance abuse conditions.

Authors contributions

J. Tsai designed the study, supported the data collection, and wrote the manuscript. E. Elbogen helped design the study and write the manuscript. M. Huang assisted with data collection and the data analyses. C. North provided important insights and helped write the manuscript. R. Pietrzak helped interpret the results and write the study.

Declaration of Competiting Interest

None of the authors report any conflicts of interest with this work.