Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Nov 1.
Published in final edited form as: J Subst Use Addict Treat. 2024 Aug 14;166:209493. doi: 10.1016/j.josat.2024.209493

Measurement Invariance of the Perceived Stress Scale Across Race, Sex, and Time, and Differential Impacts on Cocaine Use Treatment Outcomes

Melissa R Schick 1, Brian D Kiluk 1, Charla Nich 1, Donna LaPaglia 1, Angela M Haeny 1
PMCID: PMC11392621  NIHMSID: NIHMS2018140  PMID: 39151798

Abstract

Introduction:

To understand the influence of phenotypic characteristics, such as stress, on substance use treatment outcomes, measures must function equivalently across groups to allow for interpretable comparisons of effects. The present study evaluated measurement invariance of the Perceived Stress Scale (PSS) across race, sex, and time, examined its association with cocaine use disorder (CUD) treatment outcomes, and tested whether associations were moderated by race and/or sex.

Methods:

Data from four clinical trials evaluating behavioral and/or pharmacological treatments for cocaine use were combined providing a total sample of 302 participants with DSM-IV cocaine abuse/dependence (57.6% Black, 42.4% White, 43.7% females, Mage=40.22 years, SD=9.26).

Results:

Factor analyses support a two-factor model (i.e., general stress, self-efficacy to cope with stressors) that demonstrated configural, metric, and scalar invariance across race and sex and configural and metric invariance across time. End-of-treatment stress and coping were both related to treatment outcomes, but not treatment retention. Interactions between baseline and end-of-treatment stress and coping self-efficacy with race and sex predicting treatment retention and outcomes were not significant.

Conclusions:

Results support the utility of the PSS to examine between-group differences among individuals with CUD and suggest that sociodemographic groups differ in the extent to which stress and self-efficacy to cope influence treatment outcomes.

Keywords: Perceived Stress Scale, measurement invariance, Cocaine Use Disorder, treatment outcomes

1. Introduction

Rates of cocaine-involved mortality have risen over the past two decades, with approximately 16% of overdose deaths in 2021 involving cocaine (Centers for Disease Control and Prevention, 2023). These trends highlight Cocaine Use Disorder (CUD) as a significant public health concern and the need to identify factors that contribute to improved cocaine treatment outcomes. Cocaine-related outcomes differ across sociodemographic groups. Specifically, males are overrepresented among those reporting past-year cocaine use and among cocaine-related overdose deaths. On the other hand, while rates of cocaine use do not significantly differ across racial/ethnic groups, Black adults who use cocaine are more than twice as likely to die of a cocaine-related overdose relative to White adults (Cano et al., 2020). Substance use disorder treatment completion may also vary by sociodemographic group (Jordan et al., 2021; McHugh et al., 2018; Sahker et al., 2020). Research evaluating the influence of phenotypic characteristics on cocaine use treatment outcomes across and within sociodemographic groups is needed to facilitate the development of targeted interventions to improve treatment outcomes. Stress may be one such characteristic worthy of investigation.

Stress is a multifaceted and intricate concept comprising various dimensions, including biological, sociological, and psychological components (Folkman et al., 1986). Perceived stress refers to one’s self-perceived ability to manage stress well and the degree to which individuals appraise their life as overwhelming (Cohen et al., 1983; Cohen & Williamson, 1988). Extant literature has demonstrated an important role of stress, broadly defined, in substance use disorders (for a review, see Fronk et al., 2020). Many theoretical models of addiction implicate stress as a major reason for using substances. For instance, negative reinforcement models posit that the use of substances serves a regulatory function to alleviate undesirable affective states, such as stress (Koob, 2013; Koob et al., 2014). Stress, including perceived stress, has also been implicated in outcomes of substance use treatment – stressors can induce cravings directly (Fox et al., 2008; Sinha et al., 2000; Sinha et al., 2006) or may indirectly promote reactivity to other triggers for relapse (Furnari et al., 2015; Gilmore et al., 2019; Preston et al., 2018; Preston et al., 2009; Roos et al., 2020), and stress reactivity predicts relapse (Back et al., 2010; Sinha et al., 2006). Importantly, females tend to be more vulnerable to stress-related relapse (Back et al., 2005), and medications that target stress have been found to more effectively reduce cocaine cravings among females compared to males (Fox et al., 2014). Further, increased exposure to stressors and adversity has been proposed to underlie the disparate rates of cocaine-involved mortality among Black adults, compared to White adults (Fothergill et al., 2016). Thus, empirical evidence supports the notion that stress may differently impact substance use treatment outcomes for individuals from various sociodemographic groups.

However, to allow for interpretable comparisons of the effects of stress on clinical outcomes, there is a need to ensure that stress-related measures function equivalently across groups. The Perceived Stress Scale (PSS; Cohen et al., 1983) is the most common self-report measure of stress used in social science research (i.e., over 37,000 citations on the manuscript describing its development; Cohen et al., 1983) and is included as a recommended instrument under the psychosocial domain in the PhenX toolkit, a catalog of recommended measurement protocols for inclusion in human subject research. PhenX has been funded by multiple NIH agencies, and the NIH supports the recommendations of the PhenX group for assessments to include in their funded studies. The inclusion of the PSS in the PhenX toolkit may serve to further expand its use in clinical research; thus, understanding the functioning of the PSS is critical to underlie the rigor of future research. Studies examining the structure of the PSS have generally found support for a two-dimensional model reflecting general distress and perceived ability to cope with current stressors (Hewitt et al., 1992; Martin et al., 1995). Previous research has examined measurement invariance of the PSS across sex and mental health status, finding support for invariance by sex but not by mental health status, suggesting that scores cannot be compared directly between psychiatric and non-psychiatric samples (Lavoie & Douglas, 2012). Further, one study to date has examined measurement invariance of the PSS over time (Barbosa-Leiker et al., 2013), a critical step given that the PSS is often used to measure changes in stress over time, including in clinical trials (e.g., Black & Amaro, 2019; Park et al., 2021). Specifically, Barbosa-Leiker et al. (2013) found that the PSS functioned equivalently for males and females over time. While these previous studies have found support for invariance of the PSS by sex, other work has raised concern regarding potential bias – specifically, differential item functioning has been observed for certain items on the PSS (Gitchel et al., 2011). Females and members of other groups who have not historically benefitted from systemic power (e.g., members of minoritized racial and ethnic groups) experience a higher burden of stressors (Allen et al., 2020), and experience unique stressors (e.g., sexism, racism) that are entirely out of the individual’s control (Williams & Mohammed, 2009), perhaps increasing the likelihood that stressors are perceived as overwhelming and out of control. Further, differences have been found in patterns of responding to Likert-type items; specifically, White individuals have been found to be less likely to use extreme response options (Bachman & OMalley, 1984). These factors may then influence responding to measures of perceived stress, such as the PSS. However, critical gaps remain in the literature; specifically, no studies to date have examined the extent to which the PSS is invariant across race, nor have any studies examined measurement invariance of the PSS in a sample characterized by substance use.

Thus, extending previous literature, the primary aim of this study is to examine whether the PSS is invariant across race (Black versus White), sex (male versus female), race combined with sex (Black males versus Black females versus White males versus White females), and time (baseline versus end-of-treatment) using data from clinical trials that included treatment-seeking individuals with CUD. Based on previous literature (Hewitt et al., 1992; Martin et al., 1995), we expect to find support for a two-factor solution representing perceived stress and perceived ability to cope. Next, if the structure of the PSS is found to be invariant across groups, we aim to explore the association between PSS subscales identified in factor analyses and treatment retention and outcomes, and test whether these associations are moderated by racial/ethnic group and sex. If invariance is found, we hypothesize that stress will be negatively associated with treatment retention and outcomes, while coping self-efficacy will be positively associated with treatment retention and outcomes.

2. Material and Methods

2.1. Participants and Procedures

The present study leverages data from four randomized clinical trials of participants with a primary diagnosis of DSM-IV CUD. All participants were recruited from outpatient programs in Connecticut based on the following inclusion criteria: (a) 18 years of age or older; (b) English-speaking; and (c) met DSM-IV criteria for current (past 28 days) cocaine abuse or dependence (DSM-5 was released after these trials began). Exclusion criteria included current psychosis, bipolar disorder, or suicidality, and determination that outpatient treatment was not the appropriate level of care. All participants provided written informed consent. Of the treatments under investigation in the parent study, two were 8-week protocols testing the effects of a behavioral intervention and two were 12-week protocols testing the effects of combined pharmacotherapy and a behavioral intervention (for more information regarding the parent studies and treatment protocols, see Carroll et al., 2008; Carroll et al., 2014; Carroll et al., 2018; Carroll et al., 2016). While the studies from which these data were drawn were registered on ClinicalTrials.gov (NCT03547622, NCT00350870, NCT01406899), this study’s design and its analysis were not preregistered. Data, analysis code, and research materials are available upon request through the Yale Psychotherapy Development Center by brian.kiluk@yale.edu.

2.2. Measures

Participants self-reported their demographic characteristics, including age, sex (coded for binary male or female), race and ethnicity (White, Black, Hispanic, Native American Indian, Other), educational attainment (coded dichotomously as less than a high school degree vs high school degree or higher), marital status (coded dichotomously as not married vs married/cohabitating), and employment status (coded dichotomously as not employed vs employed). For the present study, those who self-identified as non-Hispanic Black were categorized as Black. Samples of Hispanic (n = 53), American Indian (n = 2), and “other” (n = 5) racial groups were too small to include in measurement invariance models as separate subgroups and these participants were therefore excluded from analyses.

The PSS is a 10-item self-report measure assessing the degree to which situations in an individual’s life are appraised as stressful (Cohen et al., 1983). Participants rated each item on a 5-point frequency scale ranging from 0 (“never”) to 4 (“very often”). Previous work has found support for the psychometrics of the PSS in samples of healthy adults (Cohen & Williamson, 1988) and college students (Roberti et al., 2006). Study staff administered the PSS to participants at baseline and end of treatment.

Study staff used a calendar-based substance use assessment tool based on the Timeline Follow-Back to obtain daily self-reports of cocaine use for the 28 days prior to the baseline session through end-of-treatment and collected urine specimens at each study visit and at end-of-treatment to biochemically verify abstinence. Treatment retention variables reflect the percent of days participants were retained in treatment and whether participants completed the study treatment period (yes/no), and treatment outcomes variables reflect the percent of treatment days participants self-reported abstinence from cocaine, the percent of urine screens that were negative for cocaine while in treatment, and whether participants were abstinent from cocaine during the last two weeks that they were active in treatment (yes/no; derived based on a combination of self-report and urine screens).

2.3. Analytic Strategy

First, we used confirmatory factor analysis with full maximum information likelihood to investigate one- and two-factor structures of the PSS for the entire sample (N = 300). Overall model fit was assessed using the likelihood ratio test based on the chi-square value, which assesses the magnitude of difference between the observed and proposed covariance matrices (Hu & Bentler, 1999). A nonsignificant likelihood ratio test indicates good model fit. However, because the chi-square test rejects even well-fitting models (Bentler, 1990), we also examined fit indices based on the chi-square distribution, with agreement among fit indices providing evidence that at least adequate model fit has been achieved. The comparative fit index (CFI; Bentler, 1990), an incremental fit index, assessed fit relative to a null model (i.e., a model in which all variables are uncorrelated); CFI values greater than .95 indicate good model fit. The root mean square error of approximation (RMSEA: Steiger, 1990) with accompanying 90% confidence intervals assessed closeness of fit, with values less than .05 indicating good fit and less than .10 indicating adequate fit. The standardized root mean square residual (SRMR) is the standardized difference between the observed and predicted covariance matrices, with values closer to zero indicating better fit and values less than .08 considered acceptable (Hu & Bender, 1999).

Then, we used multi-group confirmatory factor analysis (MGCFA) with full maximum information likelihood to investigate measurement invariance across race, sex, and time in three iterative stages. First, we evaluated configural invariance, which assesses whether the specific items that load onto each factor are the same across groups, by constraining the factor structure of the PSS to equality across groups based on the fit indices described above. Second, we evaluated metric invariance, which assesses whether the magnitude of the factor loadings are the same across groups for each item, by constraining factor loadings to equality across groups. Metric invariance was established if changes in model fit compared to the configural model did not exceed the following cutoffs: CFI ≥ −0.01, RMSEA ≥ 0.015, or SRMR ≥ 0.03. Third, we evaluated scalar invariance, which assesses whether mean differences in the latent construct capture mean differences in the shared variance of items across groups, by constraining item intercepts to equality across groups. Scalar invariance was established if changes in model fit from the metric model did not exceed the following cutoffs: CFI ≥ −0.01 and RMSEA ≥ 0.015 (Chen, 2007) or SRMR ≥ 0.01 (Putnick & Bomstein, 2016). Confirmatory factor analysis and MGCFA were conducted in Mplus version 8.0. All subsequent analyses were conducted in SPSS version 29.0.1.0.

Next, we examined bivariate correlations as preliminary analyses to determine whether baseline and end-of-treatment perceived stress scores were associated with treatment retention (percent days in treatment and treatment completion) and treatment outcomes (percent days abstinent from cocaine, percent cocaine negative urine samples, and abstinence in the last two weeks of treatment) in the overall sample and by race and sex groups. Finally, we tested whether the association between baseline and end-of-treatment stress and/or coping indices and treatment retention and outcomes in the overall sample were moderated by race and sex (i.e., 40 moderation models total) using the PROCESS macro in SPSS. The PROCESS procedures use ordinary least squares regression and bootstrapping methodology, which confer more statistical power than do standard approaches to statistical inference and do not rely on distributional assumptions (Hayes, 2018). A Benjamini-Hochberg adjustment was used to minimize both Type I and Type II error for moderation analyses (Benjamini & Hochberg, 1995) to preserve an overall Type I error without increasing the risk for Type II error and unnecessarily reducing statistical power. For models examining race and sex as moderators, the p values for these effects were rank-ordered by size. Then, each individual p value’s Benjamini-Hochberg critical value was calculated using the formula (i/m)Q, wherein i is the individual p value’s rank, m is the total number of p values, and Q is alpha (.05). Original p values are then compared to their respective Benjamini-Hochberg critical values.

3. Results

In total, 302 Black and White adults (Mage = 40.22 years, SD = 9.26) with CUD who completed the PSS at baseline (n = 300) and/or end-of-treatment (n = 221; 73.7% of the overall sample) and were included in the present analyses. Of these, 59.9% completed treatment, and 41.4% were abstinent from cocaine in the last two weeks of treatment. Sample characteristics are summarized in Table 1. Item-level PSS scores were approximately normally distributed based on skewness < ∣0.26∣ and kurtosis < ∣0.71∣ and visual inspection of Q-Q plots. Missingness was minimal for the PSS (i.e., 1% of cases per variable on the PSS) and Little’s MCAR test suggests that data on the PSS were missing completely at random (χ2[9] = 9.25, p = .41). No participants were missing data for treatment retention variables, but 10.3% of were missing data for percent days abstinent while active in treatment, 9.9% for percent cocaine negative urines while in treatment, and 14.6% for abstinence in the last two weeks of actual treatment. Little’s MCAR test suggests that data on treatment outcome variables were also missing completely at random (χ2[9] = 13.80, p = .13), though examination of baseline differences between those who were and were not missing data on cocaine use treatment outcome variables revealed that Black participants were significantly more likely to be missing data relative to White participants (χ2(1) = 4.61, p = .03, V = .12; see Supplemental Table 1). Of note, those who did not complete treatment were also significantly more likely to be missing data (χ2(1) = 93.94, p < .001, V = .56); however, this was expected, as much of the missingness on treatment outcome variables reflects individuals who completed the baseline assessment but then did not engage in the study treatment. Missingness was handled with listwise deletion.

Table 1.

Sample Demographic and Descriptive Characteristics

Overall
Sample
(N = 302)		 Black
(n = 128)		 White
(n = 174)		 Males
(n = 170)		 Females
(n = 132)
Age, M (SD) a* 40.22 (9.26) 44.55 (6.92) 37.04 (9.49) 40.42 (9.24) 39.96 (9.31)
Sex, n (%)*
 Male 170 (56.3%) 78 (60.9%) 92 (52.9%) - -
 Female 132 (43.7%) 50 (39.1%) 82 (47.1%) - -
Race, n (%)†
 Black 128 (42.4%) - - 78 (45.9%) 50 (37.9%)
 White 174 (57.6%) - - 92 (54.1%) 82 (62.1%)
High School Degree or Higher, n (%)*† 241 (79.8%) 96 (75.0%) 145 (83.3%) 144 (84.7%) 97 (73.5%)
Married or Cohabitating, n (%) 78 (25.8%) 38 (29.7%) 40 (23.0%) 44 (25.9%) 34 (25.8%)
Employed, n (%) 92 (30.5%) 41 (32.0%) 51 (29.3%) 57 (33.5%) 35 (26.5%)
Pretreatment PSS Scores, M (SD)
 General Stress b* 1.78 (0.81) 1.60 (0.85) 1.91 (0.76) 1.74 (0.78) 1.83 (0.85)
 Self-Efficacy to Cope with Stressors b 1.97 (0.75) 1.88 (0.85) 2.03 (0.65) 1.94 (0.74) 2.01 (0.75)
Post-Treatment PSS Scores, M (SD)
 General Stress b*† 1.61 (0.86) 1.43 (0.85) 1.74 (0.84) 1.49 (0.84) 1.76 (0.86)
 Self-Efficacy to Cope with Stressors b 2.25 (0.90) 2.32 (1.01) 2.19 (0.82) 2.32 (0.89) 2.15 (0.92)
Baseline Cocaine Use Frequency (# days in past 30), M(SD) c 13.84 (8.67) 13.55 (9.22) 14.05 (8.26) 13.47 (9.33) 14.31 (9.10)
Treatment Retention
 % Treatment Days Completed, M (SD) d* 70.12 (39.27) 63.43 (41.44) 75.04 (36.95) 68.70 (39.95) 71.95 (38.45)
 Treatment Completed, n (%)* 181 (59.9%) 69 (53.9%) 112 (64.4%) 98 (57.6%) 83 (62.9%)
Treatment Outcomes
 % Treatment Days Abstinent from Cocaine, M (SD) d† 70.03 (26.23) 73.04 (26.66) 67.95 (25.81) 76.03 (22.13) 62.71 (28.95)
 % Cocaine Negative Urine Samples While in Treatment, M (SD) d 32.15 (36.15) 35.21 (39.07) 30.02 (33.92) 35.17 (37.26) 28.16 (34.38)
 Abstinent in Last 2 Weeks of Treatment, n (%) 125 (41.4%) 49 (38.3%) 76 (43.7%) 72 (42.4%) 53 (40.2%)
Open in a new tab

Note. a Range = 18 – 62; b Range = 0 – 4; c Range = 0 - 28; dRange = 0–100; * denotes significant differences between Black and White participants at the level p < .05; † denotes significant difference between male and female participants at the level p < .05

3.1. Confirmatory Factor Analysis

The two-factor model reflecting stress and self-efficacy to cope with stressors provided good fit to the data, χ2(34) = 78.69, p < .001, CFI = 0.96, RMSEA = .07, 90% CI [.05, .09], SRMR = 0.05. The one-factor model did not fit the data well, χ2(35) = 365.86, p < .001, CFI = 0.68, RMSEA = .18, 90% CI [.16, .19], SRMR = 0.14, and was therefore rejected. Thus, the solution modeling two distinct factors was retained1. The factor correlation between the stress and self-efficacy to cope with stressors factors was .14 (p = .049). See Table 2 for factor loadings derived from the confirmatory factor analysis assuming a two-factor solution.

Table 2.

Latent factor structure and standardized factor loadings from Confirmatory Factor Analysis for the 2-factor Perceived Stress Scale (n = 300)

In the past month, how often have you… General
Stress		 Self-Efficacy to Cope with
Stressors
1. …been upset because of something that happened unexpectedly? .65
2. …felt that you were unable to control the important things in your life. .74
3. …felt nervous and stressed? .76
6. …found that you could not cope with all the things that you had to do? .70
9. …been angered because of things that happened that were outside of your control? .68
10. …felt difficulties were piling up so high that you could not overcome them? .74
4. …felt confident about your ability to handle your personal problems? .58
5. …felt that things were going your way? .80
7. …been able to control irritations in your life? .57
8. …felt that you were on top of things? .73
Open in a new tab

3.2. Multi-Group Confirmatory Factor Analyses

Configural, metric, and scalar invariance of the two-factor model across race, sex, and time was examined using MGCFA. Results for all models are included in Table 3.

Table 3.

Multigroup Confirmatory Factor Analysis by Race, Sex, and Time

Model
Invariance		 χ2 (df) CFI RMSEA
(90CI)		 SRMR Model
comparison		 Δχ2 (Δ
df)		 ΔCFI ΔRMSEA ΔSRMR Decision
Race (Black [n = 127] versus White [n = 173])
 M1a:Configural 118.53 (68)* .950 .070 (.049, .091) .057 - - - - - -
 M2a: Metric 130.71 (76)* .946 .069 (.049, .089) .066 Mia 12.18 (8) .004 .001 .009 Accept
 M3 a: Scalar 144.36 (84)* .940 .069 (.050, .088) .073 M2a 13.66 (8) .006 .000 .007 Accept
Sex (Males [n = 169] versus Females [n = 131])
 M1b:Configural 117.90 (68)* .952 .070 (.048,.091) .054 - - - - - -
 M2b: Metric 119.71 (76)* .958 .062 (.040, .082) .056 M1b 1.81 (8) .006 .008 .002 Accept
 M3b: Scalar 128.59 (84)* .957 .059 (.038, .079) .057 M2b 8.88 (8) .001 .003 .001 Accept
Time (Baseline [n = 300] versus End-of-Treatment [n = 221])
 M1d:Configural 138.19 (68)* .965 .063 (.048, .078) .050 - - - - - -
 M2d: Metric 159.15 (76)* .958 .065 (.051, .079) .059 Mid 20.96 (8)* .007 .002 .009 Accept
 M3d: Scalar 198.81 (84)* .943 .072 (.059, .085) .059 M2d 39.65 (8)* .015 .007 .000 Reject
Open in a new tab

Note. *p< 01

3.2.1. Race: Black versus White

Configural invariance was found across the Black (n = 127) and White (n = 173) groups (χ2[68] = 118.53, p < .001, CFI = 0.95, RMSEA = .07, 90% CI [.05, .09], SRMR = .06). The model fit indices for the metric invariance model did not substantially differ from the configural model (Δχ2 = 12.18, p = .14, ΔCFI = .004, ΔRMSEA = .001, ΔSRMR = .009), nor did fit indices for the scalar invariance model substantially differ from the metric model (Δχ2= 13.66, p = .09, ΔCFI = .006, ΔRMSEA = .000, ΔSRMR = .007).

3.2.2. Sex: Males versus Females

Configural invariance was found across males (n = 169) and females (n = 131; χ2[68] = 117.90, p < .001, CFI = 0.95, RMSEA = .07, 90%CI [.05, .09], SRMR = .05). The model fit indices for the metric invariance model did not substantially differ from the configural model (Δχ2 = 1.81, p = .99, ΔCFI = .006, ΔRMSEA = .008, ΔSRMR = .002), nor did fit indices for the scalar invariance model substantially differ from the metric model (Δχ2 = 8.88, p = .35, ΔCFI = .001, ΔRMSEA = .003, ΔSRMR = .001).

3.2.3. Time: Baseline versus End-of-Treatment

Configural invariance was found between baseline (n = 300) and end-of-treatment (n = 221; χ2[68] = 138.19, p < .001, CFI = 0.97, RMSEA = .06, 90%CI[.05, .08], SRMR = .05). The model fit indices for the metric invariance model did not substantially differ from the configural model (Δχ2 = 20.96, p = .007, ΔCFI = .007, ΔRMSEA = .002, ΔSRMR = .009). Model fit indices did significantly differ when comparing the metric and scalar models (Δχ2 = 39.65, p < .001, ΔCFI = .015, ΔRMSEA = .007, ΔSRMR = .000), suggesting that the PSS demonstrated scalar non-invariance over time.

For between-group comparisons regarding baseline and end-of-treatment stress and coping, see Supplemental Table 2 for comparisons by race and Supplemental Table 3 for comparisons by sex.

3.3. Stress and Coping Self-Efficacy in Relation to Treatment Retention and Outcomes

Preliminary bivariate correlations between baseline and end-of-treatment stress and coping self-efficacy and treatment retention and outcomes are presented in Supplemental Table 4. Baseline stress and coping self-efficacy were not significantly correlated with treatment retention or outcomes in the overall sample, nor was baseline coping self-efficacy significantly related to treatment retention or outcomes in any subsamples. Baseline stress was significantly correlated with percent cocaine negative urine samples while in treatment for men only (r = . 16, p = .04). Stress scores at end-of-treatment were significantly negatively correlated with being abstinent from cocaine in the last two weeks of treatment (r = −.15, p = .03) in the overall sample and particularly Black participants (r = −.22, p = .048) and with percent study days abstinent from cocaine among White participants (r = −.20, p = .03). Finally, coping self-efficacy scores at end-of-treatment were significantly positively correlated with percent of study treatment days abstinent from cocaine in the overall sample (r = .37, p < .001) and among Black participants (r = .31, p = .003), White participants (r = −.22, p = .03), and males (r = .34, p < .001), and with percent of cocaine negative urine samples while in treatment in the overall sample (r = .24, p < .001) and among Black participants r = .30, p = .005) and males (r = .29, p = .001). Stress and coping self-efficacy scores at end-of-treatment were not significantly related to treatment retention in the overall sample or any subsamples, nor were they related to any treatment outcomes for females. After applying a Benjamini-Hochberg correction, no interactions between baseline or end-of-treatment stress or coping and sex or race were statistically significant; moderation analyses by race are summarized in Table 4 and by sex are summarized in Table 5.

Table 4.

Main and Interactive Effects of Observed Stress and Coping Self-Efficacy and Race on Treatment Retention and Outcomes

IV: Baseline Stress IV: End-of-Treatment Stress IV: Baseline Coping IV: End-of-Treatment Coping
Construct b (SE) 95% CI b (SE) 95% CI b (SE) 95% CI b (SE) 95% CI
Outcome: Percent days in treatment
 IV −5.74 (8.80) −23.05, 11.58 −5.62 (7.51) −20.43, 9.19 −9.99 (9.95) −29.56, 9.59 −6.65 (7.42) −21.28, 7.98
 Race −12.27 (4.63)* −21.38, −3.16 −15.71 (4.24)** −24.08, −7.35 −11.54 (4.57)* −20.53, −2.54 −16.02 (4.17)** −24.23, −7.81
 IV X Race 1.51 (5.64) −9.58, 12.60 3.71 (4.93) −6.02, 13.43 5.07 (6.10) −6.93, 17.06 5.54 (4.59) −3.52, 14.59
Outcome: Completed treatment
 IV −0.35 (0.47) −1.28, 0.58 −0.29 (0.64) −1.54, 0.96 −0.52 (0.54) −1.57, 0.53 −0.74 (0.65) −2.02, 0.53
 Race −0.48 (0.25) −0.95, 0.01 −0.79 (0.34)* −1.46, −0.13 −0.43 (0.24) −0.90, 0.04 −0.83 (0.33)* −1.48, −0.18
 IV X Race 0.15 (0.30) −0.44, 0.73 0.24 (0.39) −0.53, 1.00 0.35 (0.33) −0.29, 0.98 0.50 (0.38) −0.24, 1.25
Outcome: Percent study days abstinent from cocaine
 IV −9.78 (5.96) −21.51, 1.95 −14.22 (6.12)* −26.28, −2.16 −13.25 (6.71)* −26.46, −0.04 1.09 (1.47) −1.80, 3.98
 Race 6.50 (3.25)* 0.10, 12.89 4.86 (3.49) −2.02, 11.74 6.00 (3.20) −0.30, 12.30 3.97 (3.35) −2.64, 10.58
 IV X Race 6.73 (3.88) −0.90, 14.36 8.86 (4.06)* 0.85, 16.86 7.52 (4.18) −0.72, 15.75 0.45 (0.91) −1.34, 2.24
Outcome: Percent cocaine negative urine samples
 IV −1.57 (8.41) −18.13, 14.99 −18.05 (8.85)* −35.49, −0.61 −17.10 (9.50) −35.80, 1.59 0.73 (2.13) −3.47, 4.92
 Race 6.88 (4.56) −2.09, 15.86 6.93 (5.06) −3.05, 16.92 5.75 (4.47) −3.05, 14.54 5.44 (4.88) −4.17, 15.05
 IV X Race 3.99 (5.48) −6.79, 14.78 11.67 (5.88)* 0.08, 23.25 12.54 (5.90)* 0.92, 24.16 1.04 (1.32) −1.56, 3.64
Outcome: Abstinence in last two weeks of treatment
 IV −0.19 (0.48) −1.12, 0.74 −0.02 (0.09) −0.19, 0.15 0.01 (0.56) −1.08, 1.10 0.62 (0.51) −0.38, 1.63
 Race −0.06 (0.26) −0.57, 0.45 −0.42 (0.30) −1.00, 0.17 0.02 (0.26) −0.49, 0.53 −0.32 (0.29) −0.88, 0.25
 IV X Race 0.04 (0.31) −0.57, 0.66 −0.03 (0.06) −0.15, 0.08 0.15 (0.35) −0.53, 0.83 −0.21 (0.32) −0.83, 0.41
Open in a new tab

Note. *Significant at the level p < .05 prior to applying Benjamini-Hochberg correction; **Remained statistically significant after applying Benjamini-Hochberg correction, ***p < .001

Table 5.

Main and Interactive Effects of Observed Stress and Coping Self-Efficacy and Sex on Treatment Retention and Outcomes

IV: Baseline Stress IV: End-of-Treatment Stress IV: Baseline Coping IV: End-of-Treatment Coping
Construct b (SE) 95% CI b (SE) 95% CI b (SE) 95% CI b (SE) 95% CI
Outcome: Percent days in treatment
 IV 5.66 (8.73) −11.52, 22.84 2.02 (7.76) −13.27, 17.31 12.22 (9.33) −6.14, 30.58 3.46 (7.34) −11.00, 17.93
 Sex 3.19 (4.58) −5.82, 12.20 1.48 (4.36) −7.11, 10.07 3.16 (4.57) −5.83, 12.15 2.02 (4.32) −6.50, 10.54
 IV X −5.31 (5.60) −16.33, 5.70 −0.54 (5.06) −10.52, 9.44 −9.48 (6.12) −21.53, 2.57 −1.43 (4.78) −10.85, 7.99
 Sex
Outcome: Completed treatment
 IV 0.21 (0.45) −0.68, 1.10 0.27 (0.58) −0.88, 1.41 0.59 (0.49) −0.37, 1.55 0.25 (0.55) −0.83, 1.33
 Sex 0.24 (0.24) −0.23, 0.71 0.12 (0.33) −0.53, 0.78 0.23 (0.24) −0.25, 0.70 0.17 (0.33) −0.48, 0.81
 IV X −0.20 (0.30) −0.78, 0.38 −0.08 (0.39) −0.84, 0.68 −0.38 (0.32) −1.02, 0.26 −0.13 (0.36) −0.84, 0.59
 Sex
Outcome: Percent study days abstinent from cocaine
 IV 0.09 (5.91) −11.55, 11.73 1.90 (6.11) −10.15, 13.95 0.76 (6.40) −11.84, 13.35 11.37 (5.48)* 0.57, 22.18
 Sex −12.04 (3.10)** −18.15, −5.93 −10.98 (3.42)** −17.73, −4.23 −11.89 (3.11)** −18.00, −5.77 −10.18 (3.28)** −16.64, −3.72
 IV X −0.35 (3.74) −7.72, 7.01 −2.03 (3.97) −9.85, 5.80 −1.66 (4.11) −9.74, 6.42 −3.20 (3.58) −10.25, 3.85
 Sex
Outcome: Percent cocaine negative urine samples
 IV 16.33 (8.32) −0.05, 32.71 6.56 (8.94) −11.07, 24.19 18.21 (9.10)* 0.32, 36.10 18.58 (8.06)* 2.69, 34.47
 Sex −7.20 (4.42) −15.90, 1.51 −9.30 (5.02) −19.20, 0.60 −7.04 (4.42) −15.75, 1.67 −8.37 (4.83) −17.90, 1.16
 IV X −8.59 (5.33) −19.09, 1.91 −5.38 (5.81) −16.83, 6.07 −11.17 (5.89) −22.76, 0.42 −6.56 (5.26) −16.93, 3.82
 Sex
Outcome: Abstinence in last two weeks of treatment
 IV 0.51 (0.48) −0.43, 1.45 −0.08 (0.09) −0.25, 0.09 0.41 (0.54) −0.65, 1.47 0.13 (0.48) −0.82, 1.07
 Sex −0.16 (0.25) −0.66, 0.33 −0.01 (0.29) −0.58, 0.56 −0.19 (0.25) −0.69, 0.31 −0.06 (0.29) −0.62, 0.51
 IV X −0.43 (0.31) −1.03, 0.18 0.01 (0.06) −0.10, 0.12 −0.11 (0.35) −0.79, 0.57 0.11 (0.32) −0.51, 0.73
 Sex
Open in a new tab

Note. *Significant at the level p < .05 prior to applying Benjamini-Hochberg correction; **Remained statistically significant after applying Benjamini-Hochberg correction

4. Discussion

Although empirical evidence supports the notion that perceived stress may impact substance use treatment outcomes, there is a need to demonstrate that measures of perceived stress function similarly across sociodemographic groups before making between-group comparisons. This study is the first to evaluate the measurement invariance of the PSS across race, despite it being the most widely used measure of perceived stress and a recommended instrument for inclusion in NIH-funded human subject research, including in substance use-focused clinical trials. Consistent with previous literature (Barbosa-Leiker et al., 2013; Hewitt et al., 1992; Martin et al., 1995), our findings support a two-factor solution reflecting general stress and self-efficacy to cope with stressors across race (Black versus White) and sex (males versus females) among adults with a primary CUD, providing evidence that the PSS can be used to make meaningful between-group comparisons. Our finding regarding invariance of the PSS across sex is consistent with previous research (Barbosa-Leiker et al., 2013; Milfont & Fischer, 2010; Widaman & Reise, 1997), while our findings regarding race are novel. Given disparities across groups in substance use and treatment outcomes observed in the literature (Cano et al., 2020; Saloner & Cook, 2013), comparisons of perceived stress and coping self-efficacy may prove important to inform needed adaptations or improvements to interventions in order to reduce these disparities.

We also found support for configural and metric, but not scalar, invariance over time (baseline versus end-of-treatment) for the PSS, suggesting that differences in mean responses from baseline to end-of-treatment may be due to measurement bias (specifically, differences in intercepts) rather than reflecting true differences. Previous work has found more robust support for the invariance of the PSS over time – namely, prior studies have found support for scalar invariance as well as configural and metric invariance (Barbosa-Leiker et al., 2013). However, a critical distinction between these studies is that Barbosa-Leiker et al. (2013) used data from a longitudinal study that was observational in nature; thus, scores would not necessarily be expected to systematically change over time. Previous work using clinical trial data to test invariance of a measure of depressive symptoms similarly found lack of scalar invariance over time and suggest that changes in item intercepts from pre- to post-treatment may represent a response shift, wherein individuals are better able to respond about their emotional state following treatment (Fokkema et al., 2013). Overall, further work is needed to more thoroughly consider the properties of the PSS over time.

A secondary aim of this work was to explore the association of PSS subscales to cocaine use treatment retention and outcomes and to test whether these associations were moderated by race and sex. When examining the bivariate associations of stress and coping self-efficacy with treatment outcomes across race and sex, we observed some notable between-group differences, though effect sizes were generally fairly small. Specifically, baseline stress was significantly correlated with percent of cocaine negative urines for male (but not female) participants, while end-of-treatment stress was significantly correlated with likelihood of abstinence in the last two weeks of treatment for Black (but not White) participants and with percent study days abstinent from cocaine for White (but not Black) participants. End-of-treatment coping self-efficacy was correlated with percent study treatment days abstinent from cocaine for male (but not female) participants and with percent cocaine negative urine samples for Black (but not White) participants and male (but not female) participants. These findings regarding the differential nature of associations between stress and coping and treatment outcomes highlight the need for specificity in future research when considering what precise index of treatment outcome is being examined, and to consider what other factors may influence treatment outcomes in conjunction with stress and coping. Yet, the reason for these differential effects is not clear. The lack of association of stress and coping with each outcome for female participants may suggest that females experienced additional barriers to engagement in treatment or stressors in treatment that were not captured by the PSS but that influenced their treatment outcomes (e.g., childcare needs; Brogly et al., 2018). However, when examining the interaction of stress and coping with race and sex predicting treatment outcomes, we found no significant results. It may be that we were underpowered to detect such effects, as larger sample sizes tend to be required for testing interaction terms. Further work is needed to more deeply examine these associations and understand these differential findings, including using larger samples.

Interestingly, and contrary to expectations, we did not find any effect of stress or coping self-efficacy on treatment retention in the overall sample or in any sociodemographic subgroups. This finding contrasts with previous literature reporting that higher levels of perceived stress upon admission are associated with shorter treatment retention (though, it is worth noting that some individuals included in this study were enrolled in methadone maintenance prior to beginning the study treatment, so baseline PSS scores are not necessarily reflective of true baseline stress and coping; Ware et al., 2023), and that individuals with greater hypothalamic-pituitary-adrenal (HPA) axis dysregulation (reflecting an individual’s inability to respond adaptively to stress) are more likely to drop out of treatment, and more likely to do so after a shorter number of days in treatment (Daughters et al., 2009). This disparity may reflect differences in the types of treatment under investigation (i.e., residential substance use treatment in previous literature versus outpatient treatment in the current study); it may be that individuals undergoing residential treatment experience heightened stressors related to being away from their homes and taking time away from work and other responsibilities relative to individuals undergoing outpatient treatment. Further, individuals in residential substance use treatment tend to have greater clinical severity (Tiet et al., 2007) and may therefore experience greater HPA-axis dysfunction and less adaptive response to stressors (such as early termination of treatment). Examination of objective indices of stress may reveal differing associations with indices of outpatient treatment retention; further investigation is needed.

Findings from the present study should be considered within the context of its limitations. First, there was variability in the sample with respect to treatments under investigation (i.e., medication versus psychosocial treatment) and other co-occurring conditions among participants that may have influenced outcomes. Second, while the Perceived Stress Scale is the most commonly used measure of perceived stress in social science research, these findings may not extend to other measures of perceived stress (e.g., the Rhode Island Stress and Coping Inventory; Fava et al., 1998), measures reflecting specific stressful situations (e.g., the Daily Hassles Scale; DeLongis et al., 1982), or types of stressors that may be particularly relevant to specific sociodemographic groups (e.g., race-based traumatic stress; Carter et al., 2013). Third, some subgroups had a relatively small sample size for testing measurement invariance (e.g., 127 Black participants and 131 female participants). Previous research has found that samples smaller than n = 100 per group tend to over-reject correct models using alternative fit indices (e.g., RMSEA; Chen et al., 2008); our group sizes were greater than this threshold. While simulation studies have found larger standard errors with n = 100 per group, this is more pronounced with a higher number of factors (whereas we are testing two-factor models in the present study). Thus, these analyses should be replicated with larger sample sizes; specifically, n = 200 or greater per group is preferable (Meade & Bauer, 2007; Meade et al., 2008). Fourth, while individuals from minoritized backgrounds are more likely to experience psychosocial stressors that increase their risk for substance use (e.g., racial/ethnic or sex discrimination), we did not account for these in the present study, nor did we account for other factors which likely influence treatment retention and outcomes (e.g., social support, coping skills, relapse prevention skills). Future research could examine how the substance use treatment retention and outcomes are influenced by the combination of general perceived stress and stress caused by experiencing discrimination, and by the interactions between stress and other predictors. Fifth, while the present study found evidence for measurement equivalence of the PSS across race and sex, we did not consider whether other important psychometric properties of the Perceived Stress Scale are influenced by sociodemographic characteristics among individuals who use substances (e.g., construct validity, differential item functioning). Previous work using population-based studies has found that many commonly used measures of stress, including the Perceived Stress Scale, demonstrate poor discriminant validity (Schmidt et al., 2020). Additional work in this area is needed. Finally, data were missing on cocaine use treatment outcome variables for 10-15% of participants, and missingness was more pronounced for Black participants. This increased missingness may reflect the additional barriers to treatment engagement experienced by racially minoritized individuals (Lam et al., 2022) and may have implications for whom the observed associations between PSS scores and treatment outcomes apply. Additional work is needed to further explore this finding.

5. Conclusion

Despite these limitations, our results provide support for measurement invariance of a two-factor PSS across Black and White males and females with CUD. Further, we found associations between stress and coping self-efficacy and treatment outcomes that differed based on race and sex. Additional work is needed to explore the underlying other role of factors that may influence treatment outcomes, such as discrimination (Matsuzaka & Knapp, 2020; Mays et al., 2017) and other structural barriers that may influence treatment outcomes and which have been found to differentially impact individuals from different sociodemographic backgrounds (Holzhauer, 2020; Naz et al., 2019).

Supplementary Material

1

Highlights.

  • Two factors reflecting stress and self-efficacy to cope with stress had best fit.

  • The PSS was invariant across race and sex, and partially invariant across time.

  • End-of-treatment stress and coping and changes in coping influenced outcomes.

Funding Source:

Data used in this manuscript were drawn from studies supported by National Institute on Drug Abuse grants R01DA019078 and R37DA015969. Work on this manuscript was supported by National Institutes of Health grants T32DA019426, K23059609, K02AA027300, K23AA028515, and R25DA035163.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Conflicts of Interest: The authors report no conflicts of interest.

1

We tested additional models using WLSMV estimation given the ordinal nature of the PSS response options. These models had similar model fit for the two-factor solution (χ2[34] = 99.89, p < .001, CFI = 0.97, RMSEA = 0.08, 90%CI [.06, 0.10]) and significantly worse model fit for the one-factor solution (χ2[35] = 1147.32, p < .001, CFI = 0.52, RMSEA = 0.33, 90%CI [.31, 0.34]). Based on the slightly lower RMSEA and χ2 value, we chose to retain the two-factor model using ML estimation.

References

  1. Allen JO, Watkins DC, Mezuk B, Chatters L, & Johnson-Lawrence V (2020). Mechanisms of racial health sisparities: Relationships between coping and psychological and physiological stress responses. Ethnicity & Disease, 30(4), 563–574. 10.18865/ed.30.4.563 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Bachman JG, & O'Malley PM (1984). Yea-saying, nay-saying, and going to extremes: Black-white differences in response styles. Public Opinion Quarterly, 48(2), 491–509. 10.1086/268845 [DOI] [Google Scholar]
  3. Back SE, Brady KT, Jackson JL, Salstrom S, & Zinzow H (2005). Gender differences in stress reactivity among cocaine-dependent individuals. Psychopharmacology, 180(1), 169–176. 10.1007/s00213-004-2129-7 [DOI] [PubMed] [Google Scholar]
  4. Back SE, Hartwell K, Desantis SM, Saladin M, McRae-Clark AL, Price KL, Moran-Santa Maria MM, Baker NL, Spratt E, Kreek MJ, & Brady KT (2010). Reactivity to laboratory stress provocation predicts relapse to cocaine. Drug and Alcohol Dependence, 106(1), 21–27. 10.1016/j.drugalcdep.2009.07.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Barbosa-Leiker C, Kostick M, Lei M, McPherson S, Roper V, Hoekstra T, & Wright B (2013). Measurement invariance of the Perceived Stress Scale and latent mean differences across gender and time. Stress and Health, 29(3), 253–260. 10.1002/smi.2463 [DOI] [PubMed] [Google Scholar]
  6. Benjamini Y, & Hochberg Y (1995). Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal Statistical Society: Series B (Methodological), 57(1), 289–300. 10.1111/j.2517-6161.1995.tb02031.x [DOI] [Google Scholar]
  7. Bentler PM (1990). Comparative fit indexes in structural models. Psychological Bulletin, 107(2), 238–246. 10.1037/0033-2909.107.2.238 [DOI] [PubMed] [Google Scholar]
  8. Black DS, & Amaro H (2019). Moment-by-Moment in Women's Recovery (MMWR): Mindfulness-based intervention effects on residential substance use disorder treatment retention in a randomized controlled trial. Behaviour Research and Therapy, 120, 103437. 10.1016/j.brat.2019.103437 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Brogly SB, Link K, & Newman A (2018). Barriers to treatment for substance use disorders among women with children. Canadian Journal of Addiction, 9(3), 18–22. 10.1097/cxa.0000000000000025 [DOI] [Google Scholar]
  10. Cano M, Oh S, Salas-Wright CP, & Vaughn MG (2020). Cocaine use and overdose mortality in the United States: Evidence from two national data sources, 2002-2018. Drug and Alcohol Dependence, 214, 108148. 10.1016/j.drugalcdep.2020.108148 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Carroll KM, Ball SA, Martino S, Nich C, Babuscio TA, Nuro KF, Gordon MA, Portnoy GA, & Rounsaville BJ (2008). Computer-assisted delivery of cognitive-behavioral therapy for addiction: a randomized trial of CBT4CBT. American Journal of Psychiatry, 165(7), 881–888. 10.n76/appi.ajp.2008.07111835 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Carroll KM, Kiluk BD, Nich C, Gordon MA, Portnoy GA, Marino DR, & Ball SA (2014). Computer-assisted delivery of cognitive-behavioral therapy: efficacy and durability of CBT4CBT among cocaine-dependent individuals maintained on methadone. American Journal of Psychiatry, 171(4), 436–444. 10.1176/appi.ajp.2013.13070987 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Carroll KM, Nich C, DeVito EE, Shi JM, & Sofuoglu M (2018). Galantamine and computerized cognitive behavioral therapy for cocaine dependence: a randomized clinical trial. The Journal of Clinical Psychiatry, 79(1), 7588. 10.4088/JCP.17m11669 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Carroll KM, Nich C, Petry NM, Eagan DA, Shi JM, & Ball SA (2016). A randomized factorial trial of disulfiram and contingency management to enhance cognitive behavioral therapy for cocaine dependence. Drug and Alcohol Dependence, 160, 135–142. 10.1016/j.drugalcdep.2015.12.036 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Carter RT, Mazzula S, Victoria R, Vazquez R, Hall S, Smith S, Sant-Barket S, Forsyth J, Bazelais K, & Williams B (2013). Initial development of the Race-Based Traumatic Stress Symptom Scale: Assessing the emotional impact of racism. Psychological Trauma: Theory, Research, Practice, and Policy, 5(1), 1–9. 10.1037/a0025911 [DOI] [Google Scholar]
  16. Centers for Disease Control and Prevention. (2023). Multiple causes of death 1999-2021. Retrieved from https://wonder.cdc.gov/wonder/help/mcd.html
  17. Chen F, Curran PJ, Bollen KA, Kirby J, & Paxton P (2008). An empirical evaluation of the use of fixed cutoff points in RMSEA test statistic in structural equation models. Sociological Methods & Research, 36(4), 462–494. 10.1177/0049124108314720 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Chen FF (2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling: a Multidisciplinary Journal, 14(3), 464–504. 10.1080/10705510701301834 [DOI] [Google Scholar]
  19. Cohen S, Kamarck T, & Mermelstein R (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24(4), 386–396. 10.2307/2136404 [DOI] [PubMed] [Google Scholar]
  20. Cohen S, & Williamson G (1988). Perceived stress in a probability sample in the United States. In Spacapan S & Oskamp S (Eds.), The Social Psychology of Health (pp. 31–67). Oxford. [Google Scholar]
  21. Daughters SB, Richards JM, Gorka SM, & Sinha R (2009). HPA axis response to psychological stress and treatment retention in residential substance abuse treatment: A prospective study. Drug and Alcohol Dependence, 105(3), 202–208. 10.1016/j.drugalcdep.2009.06.026 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. DeLongis A, Coyne JC, Dakof G, Folkman S, & Lazarus RS (1982). Relationship of daily hassles, uplifts, and major life events to health status. Health Psychology, 1(2), 119–136. 10.1037/0278-6133.1.2.119 [DOI] [Google Scholar]
  23. Fava JL, Ruggiero L, & Grimley DM (1998). The development and structural confirmation of the Rhode Island Stress and Coping Inventory. Journal of Behavioral Medicine, 21(6), 601–611. 10.1023/a:1018752813896 [DOI] [PubMed] [Google Scholar]
  24. Fokkema M, Smits N, Kelderman H, & Cuijpers P (2013). Response shifts in mental health interventions: an illustration of longitudinal measurement invariance. Psychological Assessment, 25(2), 520–531. 10.1037/a0031669 [DOI] [PubMed] [Google Scholar]
  25. Folkman S, Lazarus RS, Dunkel-Schetter C, DeLongis A, & Gruen RJ (1986). Dynamics of a stressful encounter: Cognitive appraisal, coping, and encounter outcomes. Journal of Personality and Social Psychology, 50(5), 992–1003. 10.1037//0022-3514.50.5.992 [DOI] [PubMed] [Google Scholar]
  26. Fothergill K, Ensminger ME, Doherty EE, Juon H-S, & Green KM (2016). Pathways from early childhood adversity to later adult drug use and psychological distress: A prospective study of a cohort of African Americans. Journal of Health and Social Behavior, 57(2), 223–239. 10.1177/0022146516646808 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Fox HC, Hong K-IA, Siedlarz K, & Sinha R (2008). Enhanced sensitivity to stress and drug/alcohol craving in abstinent cocaine-dependent individuals compared to social drinkers. Neuropsychopharmacology, 33(4), 796–805. 10.1038/sj.npp.1301470 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Fox HC, Morgan PT, & Sinha R (2014). Sex differences in guanfacine effects on drug craving and stress arousal in cocaine-dependent individuals. Neuropsychopharmacology, 39(6), 1527–1537. 10.1038/npp.2014.1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Fronk GE, Sant'Ana SJ, Kaye JT, & Curtin JJ (2020). Stress allostasis in substance use disorders: promise, progress, and emerging priorities in clinical research. Annual Review of Clinical Psychology, 16, 401–430. 10.1146/annurev-clinpsy-102419-125016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Furnari M, Epstein DH, Phillips KA, Jobes ML, Kowalczyk WJ, Vahabzadeh M, Lin J-L, & Preston KL (2015). Some of the people, some of the time: field evidence for associations and dissociations between stress and drug use. Psychopharmacology, 232(19), 3529–3537. 10.1007/s00213-015-3998-7 [DOI] [PubMed] [Google Scholar]
  31. Gilmore AK, Guille C, Baker NL, Brady KT, Hahn CK, Davis CM, McCauley JL, & Back SE (2019). Gender differences in subjective stress and neuroendocrine response to a stress task among individuals with opioid dependence: A pilot study. Addictive Behaviors, 92, 148–154. 10.1016/j.addbeh.2018.12.022 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Gitchel WD, Roessler RT, & Turner RC (2011). Gender effect according to item directionality on the perceived stress scale for adults with multiple sclerosis. Rehabilitation Counseling Bulletin, 55(1), 20–28. 10.1177/0034355211404567 [DOI] [Google Scholar]
  33. Hayes AF (2018). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach (2nd ed.). Guilford Press. [Google Scholar]
  34. Hewitt PL, Flett GL, & Mosher SW (1992). The Perceived Stress Scale: Factor structure and relation to depression symptoms in a psychiatric sample. Journal of Psychopathology and Behavioral Assessment, 14(3), 247–257. [Google Scholar]
  35. Holzhauer C. (2020). Sex and gender effects in recovery from Alcohol Use Disorder. Alcohol Research: Current Reviews, 40(3). 10.35946/arcr.v40.3.03 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Hu LT, & Bentler PM (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 1–55. 10.1080/10705519909540118 [DOI] [Google Scholar]
  37. Jordan A, Nich C, Babuscio T, Quainoo S, & Carroll K (2021). To the Editor: Our response to “Substance use improvement depends on Race/Ethnicity: Outpatient treatment disparities observed in a large US national sample”. Drug and Alcohol Dependence, 223, 108670. 10.1016/j.drugalcdep.2021.108670 [DOI] [PubMed] [Google Scholar]
  38. Koob GF (2013). Negative reinforcement in drug addiction: the darkness within. Current Opinions in Neurobiology, 23(4), 559–563. 10.1016/j.conb.2013.03.011 [DOI] [PubMed] [Google Scholar]
  39. Koob GF, Buck CL, Cohen A, Edwards S, Park PE, Schlosburg JE, Schmeichel B, Vendruscolo LF, Wade CL, Whitfield TW, & George O (2014). Addiction as a stress surfeit disorder. Neuropharmacology, 76, 370–382. 10.1016/j.neuropharm.2013.05.024 [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Lam PC, Simpson D, John DA, Rodriguez M, Bridgman-Packer D, Cruz AG, O’Neill MA, & Lewis-Fernéndez R (2022). Differential engagement by race/ethnicity in experimental trials of mental health treatment interventions. The Journal of Clinical Psychiatry, 83(6). 10.4088/jcp.21r14343 [DOI] [PubMed] [Google Scholar]
  41. Lavoie JA, & Douglas KS (2012). The Perceived Stress Scale: Evaluating configural, metric and scalar invariance across mental health status and gender. Journal of Psychopathology and Behavioral Assessment, 34(1), 48–57. 10.1007/s10862-011-9266-1 [DOI] [Google Scholar]
  42. Martin RA, Kazarian SS, & Breiter HJ (1995). Perceived stress, life events, dysfunctional attitudes, and depression in adolescent psychiatric inpatients. Journal of Psychopathology and Behavioral Assessment, 17(1), 81–95. [Google Scholar]
  43. Matsuzaka S, & Knapp M (2020). Anti-racism and substance use treatment: Addiction does not discriminate, but do we? Journal of Ethnicity in Substance Abuse, 19(4), 567–593. 10.1080/15332640.2018.1548323 [DOI] [PubMed] [Google Scholar]
  44. Mays VM, Jones AL, Delany-Brumsey A, Coles C, & Cochran SD (2017). Perceived discrimination in health care and mental health/substance abuse treatment among Blacks, Latinos, and Whites. Medical Care, 55(2), 173–181. 10.1097/mlr.0000000000000638 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. McHugh RK, Votaw VR, Sugarman DE, & Greenfield SF (2018). Sex and gender differences in substance use disorders. Clinical Psychology Review, 66, 12–23. 10.1016/j.cpr.2017.10.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Meade AW, & Bauer DJ (2007). Power and precision in confirmatory factor analytic tests of measurement invariance. Structural Equation Modeling: a Multidisciplinary Journal, 14(4), 611–635. 10.1080/10705510701575461 [DOI] [Google Scholar]
  47. Meade AW, Johnson EC, & Braddy PW (2008). Power and sensitivity of alternative fit indices in tests of measurement invariance. Journal of Applied Psychology, 93(3), 568–592. 10.1037/0021-9010.93.3.568 [DOI] [PubMed] [Google Scholar]
  48. Milfont TL, & Fischer R (2010). Testing measurement invariance across groups: Applications in cross-cultural research. International Journal of Psychological Research, 3(1), 111–130. 10.21500/20112084.857 [DOI] [Google Scholar]
  49. Naz S, Gregory R, & Bahu M (2019). Addressing issues of race, ethnicity and culture in CBT to support therapists and service managers to deliver culturally competent therapy and reduce inequalities in mental health provision for BAME service users. The Cognitive Behaviour Therapist, 12. 10.1017/s1754470x19000060 [DOI] [Google Scholar]
  50. Park CL, Finkelstein-Fox L, Sacco SJ, Braun TD, & Lazar S (2021). How does yoga reduce stress? A clinical trial testing psychological mechanisms. Stress and Health, 37(1), 116–126. 10.1002/smi.2977 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Preston KL, Kowalczyk WJ, Phillips KA, Jobes ML, Vahabzadeh M, Lin J-L, Mezghanni M, & Epstein DH (2018). Exacerbated craving in the presence of stress and drug cues in drug-dependent patients. Neuropsychopharmacology, 43(4), 859–867. 10.1038/npp.2017.275 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Preston KL, Vahabzadeh M, Schmittner J, Lin J-L, Gorelick DA, & Epstein DH (2009). Cocaine craving and use during daily life. Psychopharmacology, 207(2), 291–301. 10.1007/s00213-009-1655-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Putnick DL, & Bornstein MH (2016). Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Developmental Review, 41, 71–90. 10.1016/j.dr.2016.06.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Roberti JW, Harrington LN, & Storch EA (2006). Further psychometric support for the 10-item version of the perceived stress scale. Journal of College Counseling, 9(2), 135–147. 10.1002/j.2161-1882.2006.tb00100.x [DOI] [Google Scholar]
  55. Roos CR, Kiluk BD, McHugh RK, & Carroll KM (2020). Evaluating a longitudinal mediation model of perceived stress, depressive symptoms, and substance use treatment outcomes. Psychology of Addictive Behaviors, 34(6), 660–668. 10.1037/adb0000581 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Sahker E, Pro G, Sakata M, & Furukawa TA (2020). Substance use improvement depends on Race/Ethnicity: Outpatient treatment disparities observed in a large US national sample. Drug and Alcohol Dependence, 213, 108087. 10.1016/j.drugalcdep.2020.108087 [DOI] [PubMed] [Google Scholar]
  57. Saloner B, & Cook BL (2013). Blacks and Hispanics are less likely than whites to complete addiction treatment, largely due to socioeconomic factors. Health affairs, 32(1), 135–145. 10.1377/hlthaff.2011.0983 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Schmidt K, Enge S, & Miller R (2020). Reconsidering the construct validity of self-reported chronic stress: A multidimensional item response theory approach. Psychological Assessment, 32(11), 997–1014. 10.1037/pas0000829 [DOI] [PubMed] [Google Scholar]
  59. Sinha R, Fuse T, Aubin L-R, & O'Malley SS (2000). Psychological stress, drug-related cues and cocaine craving. Psychopharmacology, 152(2), 140–148. 10.1007/s002130000499 [DOI] [PubMed] [Google Scholar]
  60. Sinha R, Garcia M, Paliwal P, Kreek MJ, & Rounsaville BJ (2006). Stress-induced cocaine craving and hypothalamic-pituitary-adrenal responses are predictive of cocaine relapse outcomes. Archives of General Psychiatry, 63(3), 324. 10.1001/archpsyc.63.3.324 [DOI] [PubMed] [Google Scholar]
  61. Steiger JH (1990). Structural model evaluation and modification: An interval estimation approach. Multivariate Behavioral Research, 25(2), 173–180. 10.1207/s15327906mbr2502_4 [DOI] [PubMed] [Google Scholar]
  62. Tiet QQ, Ilgen MA, Byrnes HF, Harris AHS, & Finney JW (2007). Treatment setting and baseline substance use severity interact to predict patients' outcomes. Addiction, 102(3), 432–440. 10.1111/j.1360-0443.2006.01717.x [DOI] [PubMed] [Google Scholar]
  63. Ware OD, Sacco P, Cagle JG, Frey JJ, Wagner FA, Wimberly AS, Gyebi-Foster B, Diaz M, Peters K, & Zemore SE (2023). Higher perceived stress during admission is associated with shorter retention in short-term residential substance use disorder treatment. Addictive Behaviors Reports, 100502. 10.1016/j.abrep.2023.100502 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Widaman KF, & Reise SP (1997). Exploring the measurement invariance of psychological instruments: Applications in the substance use domain. In Bryant KJ, Windle M, & West SG (Eds.), The science of prevention: Methodological advances from alcohol and substance abuse research (pp. 281–324). American Psychological Association. 10.1037/10222-009 [DOI] [Google Scholar]
  65. Williams DR, & Mohammed SA (2009). Discrimination and racial disparities in health: evidence and needed research. Journal of Behavioral Medicine, 32(1), 20–47. 10.1007/s10865-008-9185-0 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2018140-supplement-1.docx (22.9KB, docx)

RESOURCES