Factor Structure and Psychometric Properties of the Connor–Davidson Resilience Scale (CD-RISC) in Individuals with Opioid Use Disorder

Abstract

Aims:

Resilience is defined as the capacity for an individual to maintain normal functioning and resist the development of psychiatric disorders in response to stress and trauma. Although previous investigators have acknowledged the important role of resilience in those with substance use disorders, this is the first study to investigate the reliability, validity, and factor structure of the Connor-Davidson Resilience Scale (CD-RISC-25) in a sample of individuals with opioid use disorder (OUD). Additionally, we explored the relationship between trait resilience and the severity of drug-related problems.

Methods:

Four hundred and three participants (22% female) with OUD completed the CD-RISC-25, Beck Depression Inventory (BDI-II), and the self-report Addiction Severity Index (ASI). Confirmatory factor analysis (CFA) tested the originally proposed 5-factor solution of the CD-RISC-25.

Results:

CFA results indicated that a 5-factor model of the CD-RISC-25 performed somewhat better than the 1-factor solution. Pearson correlation revealed a negative association between CD-RISC-25 (M=75.82, SD=15.78) and ASI drug-use composite score (M=.25, SD=−.16), r=−.148, p≤.01, and between CD-RISC-25 and BDI-II (M=11.33, SD=10.58), r=−.237, p≤.001.

Conclusions:

Albeit providing only limited support for the original 5-factor structure, our results indicate that the scale may be useful for screening individuals with OUD who have a vulnerability to stress. Consistent with prior studies, higher resilience was associated with lower depression symptoms and addiction severity, further demonstrating the CD-RISC-25 ability to predict psychiatric stability. To inform the development of more targeted interventions, future studies should examine resilience longitudinally, in addition to exploring more comprehensive approaches to measuring resilience.

INTRODUCTION

Background

Substance Use Disorders (SUDs) are commonly defined as a chronically relapsing disorder characterized by compulsive drug seeking and use, development of negative emotional states when access to the drug is denied, and inability to control intake (Uhl et al., 2019). SUDs are associated with disruptions of motivational circuits produced by a multitude of factors including inflated incentive salience and habit formation, reward deficits and excessive stress, and impaired executive function (Koob and Volkow, 2016). The economic burden of SUDs is one of the largest among health-related problems in the United States (U.S.), with an estimated 500 billion dollars annually (Miller and Hendrie, 2008). Additionally, the U.S. has experienced a significant rise in opioid-related overdose mortalities over the past three decades, primarily driven by the widespread availability of prescription opioids beginning in the mid-1990s, the rise of heroin-related overdose mortalities in the mid-2000s, and, since the mid-2010s, the proliferation of illicitly manufactured fentanyl and its analogs (Alexander et al., 2018; Paulozzi, 2006; Unick and Ciccarone, 2017). This public health crisis has mobilized researchers and healthcare providers to attempt to identify the distinct risks and protective factors that may contribute to the initiation and maintenance of non-medical opioid use. Across the scientific community, there has been growing interest in investigating individual characteristics, particularly behavioral and psychological traits, and their role in the development and maintenance of SUDs, which may contribute to the development of more personalized and precise interventions (Belcher et al., 2014; Koenig, 2012; Volkow et al., 2019).

Resilience

One individual characteristic that has recently garnered attention from researchers and clinicians is resilience. Currently, the scientific literature conceptualizes resilience as a multidimensional and dynamic construct comprised of biological and psychological features. Resilience, along with specific skills (e.g. active problem-solving), enables an individual to cope with the challenges of daily life (Campbell-Sills et al., 2006). As a state, resilience resembles a protective process, activated during stressful or traumatic events to reduce the negative and adverse effects on an individual (Johnson et al., 2011). Resilience as a trait is defined as a stable, adaptive, and intrinsic feature of human beings, characterized by an individual’s capacity to maintain normal functioning and resist the development of psychiatric disorders in response to stress and adversity (Campbell-Sills et al., 2006; Maltby et al., 2015). As such, resilience may have an important role in the susceptibility and severity of psychopathologies such as SUDs, albeit not providing invulnerability to the development of psychiatric disorders (Ingram and Price, 2010).

Measurement of Resilience

One of the most frequently used measures of resilience is the Connor-Davidson Resilience Scale 25 (CD-RISC-25; Windle et al., 2011). This 25-item self-rating measure was designed to assess a series of trait characteristics that are understood to constitute trait resilience on a five-point scale (Connor and Davidson, 2003). An examination of 15 different resilience scales demonstrated that the CD-RISC-25 is a high-quality measure with sound psychometric properties (Windle et al., 2011). The authors’ primary goals in designing this scale were to develop a valid and reliable measure to compute resilience that facilitated assessment of the malleability of resilience in response to pharmacologic treatment in clinical populations, along with establishing baseline values for resilience in the general population and clinical samples (Connor and Davidson,2003).

The scale’s items, which resulted from a thorough review of the relevant literature, are believed to be related to several psychological and behavioral domains including the ability to tolerate stress and pain, the ability to adapt to change, control, patience, thinking of change as a challenge, self-efficacy, secure attachment to others, sense of humor, faith, and optimism (Connor and Davidson, 2003). The 25 items included in the final version are each rated on a 5-point scale (0 = not true at all to 4 = true nearly all of the time) and have a high internal consistency (Cronbach’s alpha = .89), A total score from 0 to 100 is generated, with higher scores reflecting a higher level of trait resilience.

The initial validation produced five factors within the CD-RISC-25 measure based on an exploratory factor analysis (EFA; Connor & Davidson, 2003). These five factors included “personal competence, high standards, and tenacity” (Factor 1), “trust in one’s instincts, tolerance of negative affect, and strengthening effects of stress” (Factor 2), “positive acceptance of change and secure relationships” (Factor 3), “control” (Factor 4), and “spiritual influences” (Factor 5). The EFA was conducted in an adult community sample (n = 577), and factors were selected based on eigenvalues greater than 1.00. Since the initial validation study, numerous investigations attempted to replicate the original factor structure proposed for the CD-RISC-25 in different samples (Faria Anjos et al., 2019; Green et al., 2014; Karairmak, 2010; Laird et al., 2019; Lamond et al., 2008; Sexton et al., 2010). While some studies found a different factor structure, all have demonstrated strong construct validity and reliability of the scale. Likewise, several studies that used the CD-RISC-25 in diverse adult psychiatric populations with trauma exposure, as well as normal healthy populations, demonstrate the scale’s general utility in examining resilience (Campbell-Sills et al., 2006; Connor et al., 2003; Green et al., 2014; Sexton et al., 2016). Although the measure has been widely implemented among a diverse set of populations, its use among individuals with SUDs merits greater exploration. Role of Resilience in the Development and severity of Substance Use Disorders Although the literature examining resilience in individuals with SUDs is limited – partly related to inconsistencies in the operationalization of resilience (i.e., resilience as a trait, outcome, or process) (Rudzinski et al., 2017) – several studies have yielded interesting results. All studies described in the following used the CD-RISC to measure resilience.

A recent study investigated the association of resilience with treatment outcomes in veterans with co-morbid post-traumatic stress disorder (PTSD) and SUDs (McGuire et al., 2018). Greater resilience was negatively associated with the severity of PTSD symptoms, substance use, and drug craving. In a large study conducted primarily with African Americans from low-income communities, individuals with higher resilience and with a history of childhood abuse and/or other trauma were found to have reduced lifetime hazardous alcohol and illicit drug use (Wingo et al., 2014). Resilience appeared to moderate SUD risk among individuals exposed to childhood abuse or other traumatic experiences.

Another study that sought to explore the relationship between resilience and mental health problems in a sample of methadone-maintained patients found that higher resilience was independently associated with lower depression and anxiety symptoms (Jiao et al., 2017). In addition to these protective effects on psychiatric symptomology, resilience appears to be associated with the effectiveness of treatment interventions. For example, in the aforementioned study (McGuire et al., 2018) decreases in PTSD symptoms were observed among patients with higher resilience that was hypothesized by the study authors to be related to greater cognitive flexibility, enabling patients to better understand and integrate their traumatic experiences. Resilience may also facilitate the use of more adaptive coping strategies such as utilizing social support and thereby decreasing patients’ need to self-medicate with illicit drugs and alcohol (McGuire et al., 2018). For example, a recent study found evidence that resilience may serve as a compensatory strategy for managing Early Maladaptive Schema (EMS); i.e., a pervasive and dysfunctional psychological pattern developed during childhood that characterizes an individual’s understanding of themselves and their relationships (McDonnell et al., 2018). Polydrug users with opioid use disorder (OUD) had increased EMS, emotional dysregulation, and maladaptive coping, and decreased resilience compared to healthy controls (McDonnell et al., 2018). Higher levels of emotional dysregulation and maladaptive coping, and EMS were negatively associated with resilience.

Current Study

There is increasing consensus about the important contribution of resilience to maintaining normal functioning during stressful events and its protective properties when encountering adversity. However, while research consistently showed that resilience is a broad construct that involves multiple concepts of adjustment and adaptation, previous studies yielded different results regarding the factor structure of the CD-RISC-25—one of the most commonly used measures of resilience—which indicates a lack of precision in defining what constitutes the psychological dimensions of resilience. In addition, a growing body of literature highlights the mitigating role of resilience in the severity of psychopathology and specifically its association with improved outcomes in the context of mood and anxiety disorders. Yet, studies investigating associations between resilience and the development and severity of SUDs remain limited. Deciphering the role of resilience in the context of SUDs may lead to a greater understanding of these disorders and may have important implications for addressing the opioid overdose epidemic.

Thus, the primary aim of this study was to establish the reliability, validity, and factor structure of the CD-RISC-25 in individuals with OUD. As described above, the CD-RISC-25 is a frequently used resilience assessment in studies with samples with psychiatric disorders, it has been intensively validated, and shows good psychometric properties. Nonetheless, the instrument warrants further investigation in SUD populations. Additionally, we sought to explore and characterize trait resilience in a sample of people who use heroin and other opioids and met DSM-IV criteria for current opioid dependence. Lastly, this study investigated the relationship between trait resilience and the severity of substance use, related issues, and behaviors. Based on the limited available evidence, we hypothesized that resilience (CD-RISC-25 mean scores) would have an inverse relationship with addiction severity; that is, participants with higher self-reported resilience would report lower severity of drug-related problems and behaviors. Additional exploratory goals were to investigate potential sex differences in resilience, as well as differences among non-treatment seekers, those currently undergoing agonist maintenance treatment, and individuals who recently underwent drug detoxification.To our knowledge, this study is the first to examine the relationship between trait resilience and severity of drug-related problems and behaviors in a sample of individuals with OUD from an urban inner-city environment. By contributing to our understanding of the role of resilience in OUD, our findings may help to identify at-risk individuals with higher precision and aid the development of more personalized interventions to improve treatment outcomes.

METHOD

The data presented here were collected as part of a 5-year randomized clinical trial (ClinicalTrials.gov Identifier: NCT02535494; completed October 10, 2019) conducted at the New York State Psychiatric Institute (NYSPI)/Columbia University Irving Medical Center (CUIMC). The study sought to understand the risks and benefits of overdose education and naloxone distribution to individuals with OUD. For more details on the larger trial and study-related procedures see (Jones et al., 2020).

Potential participants were recruited from harm reduction service providers, Bellevue Hospital’s inpatient detoxification unit, other clinical trials in the Division on Substance Use Disorders, and through advertisements in local newspapers. Following an initial telephone screen, in-person screening procedures were conducted at the Division on Substance Use Disorders at NYSPI/CUIMC. The in-person screening visit included various questionnaire-based and clinical interviews administered by a team of research assistants, psychologists, nurses, and physicians.

Participants were aged 21 to 65 years, met DSM-IV criteria for OUD within the past 6 months, and were in otherwise good physiological and mental health. Participants were excluded if they had an active psychiatric disorder that might have interfered with participation or made participation hazardous for them or study staff (e.g., DSM-IV psychotic disorder, active bipolar disorder with mania, or significant history of violent behavior).

Study Design and Procedures

All study procedures were approved by the NYSPI Institutional Review Board, and all participants provided written informed consent. Besides demographics and treatment status, variables of interest included depression assessed with the Beck Depression Inventory-Second Edition (BDI-II) (Beck et al., 1996), resilience assessed with the CD-RISC-25 (Connor and Davidson, 2003), and addiction severity assessed with the Addiction Severity Index (ASI) Self-Report Form (Blacken et al., 1994).

Data Analysis

Data were analyzed using Mplus Version 8.4 (Muthén and Muthén, 2019). Descriptive statistics for demographics, drug use patterns, ASI drug use composite scores, and CD-RISC scores are presented. Analysis of variance (ANOVA) was used to compare characteristics of non-treating seeking, recently detoxified and treatment-seeking individuals, and Fisher’s exact test to evaluate sex differences. Pearson correlation was used to evaluate the associations between trait resilience, severity of addiction, and depression. For all analyses, the critical level for rejection of the null hypothesis was considered to be a p-value of < 0.05. To protect against Type I error, Holmes or Bonferroni corrections were applied to all correlation analyses (0.05).

We employed confirmatory factor analysis (CFA) to test the 5-factor solution of the CD-RISC-25 previously identified by Connor and Davidson (2003). Each item was constrained to load onto one of five factors in accordance with the model described in that paper. Before conducting the CFA, we determined that all items were adequately normally distributed (skew and kurtosis values < 2.0) except item 25, which had a kurtosis value of 2.62. Therefore, we used MLM estimation to estimate the CFA model since it is robust to non-normality. We examined goodness of fit using multiple indices: the root mean square error of approximation (RMSEA), its 90% confidence interval (90% CI), the standardized root mean square residual (SRMR), the Comparative Fit Index (CFI), and the Tucker Lewis Index (TLI). Good model fit was determined based on published guidelines provided by (Hu and Bentler, 1999): RMSEA at or below .05, SRMR at or below .08, TLI at or above .95, and CFI at or above .95. Multiple fit indices were used as they assess different kinds of model fit (e.g., absolute fit, model parsimony) and can produce a more reliable evaluation when used in combination (Brown, 2015).

RESULTS

Sample Characteristics

A total of 403 participants were enrolled in the study. Socio-demographic and drug use characteristics of the sample are displayed in Table 1. Eighty-two percent were unemployed, and 85% had health insurance. Approximately one-quarter of participants (26%) reported that they had experienced at least one drug overdose at some point during their life.

Table 1.

Demographics Characteristics

	Mean (SD), % or #
Age (years)	46.48 (9.72)
Sex
Male	307 (77.1%)
Female	90 (22.6%)
Race/Ethnicity
American Indian or Alaska Native	4 (1%)
Asian	2 (0.5%)
Hispanic/Latino	115 (29.6%)
Pacific Islander	1 (0.3%)
Black/African American	169 (43.6%)
White/Caucasian	76 (19.6)
Multiracial or Other	21 (5.7%)
Opioid Use Status
Not Seeking OUD TX	186 (48.2%)
Maintained on Opioid Agonist Therapy	160 (41.5%)
Recently Detoxified from Opioids	40 (10.4%)
Duration of Heroin Use (years)	15.90 (11.52)
Current Heroin Use: Average $ Per day	$34.90 ($41.74)
CD-RISC 25 Scores (0-100)	75.82 (15.78)
ASI Composite Score (0.0-1.0)	0.2480 (0.1666)
BDI-II Scores (0-63)	11.33 (SD 10.58)

No difference in mean resilience scores was observed among non-treatment seekers (M = 75.83, SD = 16.53), those currently undergoing opioid agonist maintenance treatment (M = 75.20, SD = 15.26), and individuals who were recently detoxified from opioids (M = 76.75, SD = 15.05); F (2, 297) = .143, p = .867, and CD-RISC-25 scores did not significantly differ between males (M = 75.79, SD = 15.95) and females (M 75.77, SD = 15.63); t (302) = .11, p = .803. Moreover, there were no differences in ASI drug use composite scores between males (M = .244, SD =.1683) and females (M = .262, SD = .1648); t (389) = −.898, p = .703. Likewise, there was no significant difference in BDI-II scores between males (M = 10.83, SD = 10.12) and females (M= 12.59, SD = 11.33); t (369) = −1.37, p = .112. And no statistically significant differences were found among racial/ethnic groups (Hispanic, Black/African American, White/Caucasian, Multiracial) in CD-RISC-25 scores, F(3,288) = 2.29, p = .078.

Pearson correlation revealed a significant negative association between CD-RISC-25 scores and ASI drug use composite scores, r = −.148, p ≤ .01. In addition, a significant negative association was observed between CD-RISC-25 scores and BDI-II scores, r = −.237, p ≤ .001, as well as between ASI drug use composite scores and BDI-II scores r = −.309, p ≤ .00.

Confirmatory Factor Analysis

CFA was performed using the MLM estimation method testing the five-factor model. In addition, a one-factor model was tested, also using MLM estimation. Both models converged without errors. Model fit statistics are presented in Table 2 for both the five-factor model (Model 2) and the one-factor model (Model 1). Based on the Hu and Bender guidelines (Hu and Bentler, 1999), both models were within the good range for RMSEA and SRMR, and both were close to .95 for the CFI and TLI, with the five-factor model performing somewhat better. Chi-square statistics were highly significant for both models (Model 1: χ²(275) = 637.00, p < .001; Model 2: χ²(263) = 457.68, p < .001), but these are often problematic with larger samples (Kline, 2016). Based on these fit statistics both models are an adequate fit to the data.

Table 2.

Summary of Fit Indices for Model 1 (one-factor solution) and Model 2 (five-factor solution)

					RMSEA 90% CI
Model	CFI	TLI	SRMR	RMSEA	Lower	Upper
1	0.862	0.850	0.056	0.068	0.061	0.075
2	0.926	0.915	0.046	0.051	0.043	0.059

We then computed the Satorra-Bentler scaled chi-square difference test to directly compare the two models (Satorra and Bentler, 2010). We found that the five-factor model had a significantly better fit than the one-factor model using this method, χ² (10) = 54.29, p < .001. Standardized factor loadings, standard errors, and R² values of the five-factor solution along with Cronbach’s alpha coefficients for each factor are presented in Table 3. All loadings were significant at the p < .001 level. However, R² values for all five factors are weak (< .50), a sign that the original structure is not consistent with the current data.

Table 3.

Standardized factor loadings (Estimate), standard errors (SE), and R² values of the five factor solution, and Cronbach’s alpha coefficients for each factor

Factor	Indicator	Estimate*	SE	R2	Cronbach’s α
Factor 1					.868
	CD_11	0.992	0.081	0.578
	CD_10	0.839	0.079	0.437
	CD_23	0.920	0.104	0.301
	CD_16	1.040	0.107	0.391
	CD_17	1.065	0.089	0.554
	CD_25	0.836	0.072	0.513
	CD_12	0.981	0.086	0.498
	CD_24	1.000	0.000	0.514
Factor 2					.748
	CD_19	1.281	0.197	0.348
	CD_18	1.167	0.192	0.256
	CD_20	1.000	0.000	0.204
	CD_06	1.053	0.177	0.248
	CD_07	1.464	0.213	0.421
	CD_15	0.998	0.177	0.237
	CD_14	1.427	0.209	0.451
Factor 3					.696
	CD_05	1.402	0.171	0.473
	CD_01	1.000	0.000	0.284
	CD_02	0.623	0.164	0.075
	CD_08	1.506	0.182	0.498
	CD_04	1.564	0.190	0.482
Factor 4					.621
	CD_21	1.035	0.140	0.378
	CD_22	1.000	0.000	0.275
	CD_13	1.112	0.145	0.421
Factor 5					.570
	CD_03	1.000	0.000	0.361
	CD_09	0.930	0.183	0.444

^*All p < .001

Internal consistency was highest for the first two factors, but Factor 5 showed less than adequate reliability, and Factors 3 and 4 were only in the adequate range. Several of the factors showed quite high intercorrelations (rs > .90), suggesting there is some redundancy in the factors.

DISCUSSION

To our knowledge, this is the first study to evaluate the underlying factor structure of the 25-item CD-RISC-25 measure in a sample of individuals with OUD. The data demonstrate adequate reliability and validity of the measure within samples of individuals who were either not currently seeking treatment, recently detoxified from opioids, or receiving opioid agonist maintenance pharmacotherapies.

Overall, average resilience scores in our sample across all groups (non-treatment seekers, those currently undergoing agonist maintenance treatment, and recently detoxified individuals), were higher compared to other psychiatric populations (Davidson et al., 2005; Obbarius et al., 2018; Torgalsbøen, 2012). Consistent with prior studies, we found that higher resilience was associated with lower depression symptoms, further demonstrating the ability of the CD-RISC 25 to predict psychiatric stability (Min et al., 2013; Smith, 2009). The data also supported our hypothesis of addiction severity being negatively associated with resilience. However, this association was relatively weak in our study. This may be related to issues in measuring addiction severity. The ASI drug use composite score may not fully represent the impairments in psychosocial and behavioral functioning related to SUDs (Poudel et al., 2016; Quednow, 2020). Although ASI composite scores have shown strong associations with the diagnostic criteria of SUDs, these disorders are complex by definition with several neurobiological and psychological subdomains which have variable clinical manifestations affecting severity and overall burden of the disorder (Martin et al., 2008; Rikoon et al., 2006). Furthermore, prior studies have proposed an ASI drug use composite score of ≥0.4 as severe (Hoots et al., 2020; Rikoon et al., 2006). Thus, the relatively low observed mean ASI drug use composite score of .248 in our sample may have contributed to the weak association between resilience and addiction severity. Future studies may benefit from the use of more sensitive instruments, such as the Addiction Severity Assessment Tool (Butler et al., 2005), to assess the severity of drug-related problems.

Our findings provide limited support for the underlying five-factor structure proposed in the initial CD-RISC-25 validation (Connor et al., 2003). Although the confirmatory factor analysis indicated a better model fit of the 5-factor solution as compared to a 1-factor solution, the majority of the R² values for the five factors were weak, indicating that the structure described in the initial validation study may not be consistent with the current data. Our findings demonstrated the most robust internal consistency for the first two factors, however, factor 5 showed less reliability, which is in concordance with the initial validation study and subsequent investigations (Connor and Davidson, 2003; Green et al., 2014; Solano et al., 2016). The fifth factor (spiritual influences) was supported by only two items, contrary to the generally accepted guidelines which typically require factors to be represented by at least three to five measured items (MacCallum et al., 1996).

Additionally, several factors showed high intercorrelations, which suggests some redundancy in the five-factor structure. The high intercorrelations may be a result of population differences, such as differences in psychopathology patterns in clinical populations (Campbell-Sills et al., 2006; Campbell-Sills and Stein, 2007; Connor and Davidson, 2003; Moss et al.,2015). For example, SUDs are often characterized by emotional dysregulation, dysfunctional and maladaptive interpersonal relationships, and increased negative feelings, problems with cognitive reappraisal, and executive functioning impairments (Ramey and Regier, 2019; Shorey et al., 2014; Thorberg and Lyvers, 2010; Wilcox et al., 2016). All of these domains appear to have a relationship with resilience. However, as the primary study did not systematically assess psychiatric comorbidities (only depression), the authors were not able to further explore this hypothesis.

Furthermore, the CD-RISC-25 was not established by a universally accepted theory for resilience (Green et al., 2014), which may hamper confirmation of a meaningful factor structure. The results of our investigation together with prior studies that yielded mixed results regarding the initially proposed five-factor structure (Campbell-Sills and Stein, 2007; Connor and Davidson, 2003; Karairmak, 2010; Lamond et al., 2008; Sexton et al., 2016; Tan et al., 2019) indicate a need to refine the CD-RISC-25, based on a more robust theoretical conceptualization of resilience, to achieve a more stable factor structure.

A limitation of the current study is the recruitment of research volunteers exclusively from the New York City metropolitan area, which may limit generalizability of findings beyond urban environments. The sex distribution of our sample roughly reflects the ratio of men to women with OUD in the US (McHugh, 2020). In addition, our sample is representative of the NYC population and many similar urban communities. Nonetheless, further research is tasked with examining resilience in general and the usefulness of this measure in particular within different OUD populations who may be exposed to different kinds of stress and adversity (e.g., rural living Caucasians and Native Americans/Alaskan Natives, individuals from different socioeconomic backgrounds, and inpatients), in addition to other SUD populations. As mentioned above, our sample reported relatively low addiction severity along with relatively high resilience, indicating greater levels of psychosocial functioning as compared to other SUD samples. This may be suggestive of a selection bias related to the study design. The primary study required research volunteers to participate in the investigation for a period of 12 months, requiring a certain level of commitment and psychosocial functioning. Therefore, individuals with higher addiction severity and/or lower resilience may have been screened out, and our results may not be generalizable to less selective samples. Further, we employed a cross-sectional study design to estimate the factor structure of the CD-RISC-25 among individuals with OUD, resilience, and addiction severity; thus, data may only represent a snapshot of an individual’s impairment and functioning, respectively, at the time of measurement administration. Finally, we utilized a self-report measure to assess the construct of resilience; as such, bias arising from social desirability, recall period, or selective recall cannot be ruled out. Future studies would benefit from integrating self-report with other methods such as behavioral models to objectively measure resilience.

To inform the development of more targeted interventions, future studies should examine resilience longitudinally to determine whether the trait is stable over time and assess its malleability, in addition to exploring other approaches to measuring resilience. Although our results indicate good psychometric properties of the CD-RISC-25 and provide some support for the 5-factor structure, more research is needed to understand which additional factors may be associated with adaptive responses to stress and adversity in this population. A more comprehensive investigation of the psychological, behavioral, and biological underpinnings of resilience may help to further refine the measurement of this construct, and thereby help achieve greater parsimony and a more stable factor structure. As the CD-RISC aims to measure trait resilience, the instrument may benefit from integration with established psychological trait theories, such as the five-factor model of personality, to improve its overall theoretical construct and predictive power. Moreover, to achieve a universally accepted model of resilience, it is important to develop behavioral models of resilience that utilize approaches from cognitive and affective neuroscience as well as behavioral economics. This approach would a) allow the falsifiability of the theory, and b) help identify brain networks associated with resilience. Additionally, the use of laboratory models and greater integration with existing psychobiological constructs may improve the examination of associations between resilience, behaviors such as drug use and seeking, and treatment outcomes. Finally, as resilience is important in mitigating psychological adversity, how resilience may be fortified using behavioral intervention deserves further study.

In summary, the present study indicates that the CD-RISC-25, as an economic measure of resilience, may be useful for screening individuals with OUD who have a vulnerability to stress. Additionally, our findings suggest that there is a relationship between resilience and the severity of SUDs. Even though the latter may require more sophisticated and comprehensive measurement approaches, further deciphering the role of resilience may present an opportunity to tailor interventions to the patients’ specific needs, which may help reduce their psychological and behavioral burden and improve treatment outcomes.

Highlights.

• CD-RISC-25 demonstrated adequate reliability and validity in individuals with OUD

• The study provided limited support for the original 5-factor structure

• Resilience in the sample was higher compared to other psychiatric populations

• Resilience was associated with lower addiction severity and depression symptoms