Anger and Substance Use in HIV-Positive Patients with Chronic Pain

Abstract

Chronic pain increases risk of substance use in people living with HIV (PLWH). Depression and anxiety have also been identified as risk factors for substance use among PLWH. Relatedly, other negative mood states, such as anger, may influence chronic pain among PLWH. The current cross-sectional study examined whether the distinct negative mood state of anger is associated with substance use among 187 PLWH who report chronic pain. Using negative binomial regression analyses, we found higher levels of anger were positively associated with alcohol use. Higher levels of anger were inversely associated with benzodiazepine use. No association was found between anger and marijuana use, and there were no significant interactions between anger and pain severity on substance use. Our findings suggest that anger is an independent risk factor for substance use among PLWH and chronic pain. Addressing anger may be useful when adapting behavioral therapies in the treatment of pain among PLWH.

Introduction

Human immunodeficiency virus (HIV) remains a significant public health concern, with more than 1 million people living with HIV (PLWH) in the United States in 2020 (Giroir, 2020). Chronic pain in PLWH is a common comorbid, disabling condition that is associated with a high burden of regional or widespread musculoskeletal pain (Miaskowski et al., 2011). Different attributes of pain have been measured across HIV studies, including pain intensity, pain affect, interference, location, and pattern (Dworkin et al., 2005). Estimates of pain prevalence among samples of PLWH vary from 54% to 83% in a three-month period (Parker, Stein, & Jelsma, 2014) and 25% to 80% report having experienced pain in the past week (Miaskowski et al., 2011; Uebelacker et al., 2015). In addition, PLWH and chronic pain have up to 10.5 times greater odds of functional impairment compared to PLWH without pain (Merlin et al., 2013). Contributing factors for pain in PLWH may include specific HIV-related biological factors, complex social challenges, and high degrees of medical and psychiatric concerns (Merlin et al., 2014).

There is an established bidirectional relationship between pain and substance use among PLWH (Uebelacker et al., 2015). For example, among PLWH, those with substance use disorders or engaged in illicit drug use are more likely to report pain (Knowlton, Nguyen, Robinson, Harrell, & Mitchell, 2015; Marcus, Kerns, Rosenfeld, & Breitbart, 2000). PLWH who use prescription opioids also report more severe pain compared to PLWH who do not use prescription opioids (Koeppe, Armon, Lyda, Nielsen, & Johnson, 2010), and increased pain has been associated with greater opioid use and misuse among PLWH (Hansen et al., 2011).

It is not uncommon for individuals to use more than one substance in an attempt to alleviate pain. Chronic pain in PLWH also increases the risk for other substance use beyond opioid use disorder (Cunningham, 2018). Co-use of opioids and benzodiazepines among PLWH with chronic pain is common (Parent et al., 2019; Weisberg et al., 2015). Marijuana and alcohol use have also been commonly reported among PLWH and chronic pain (Sohler et al., 2018), in part to help relieve pain (Stein et al., 2015). Indeed, PLWH commonly report marijuana use as effective for pain symptom relief (Costiniuk et al., 2019). Similarly, among PLWH, moderate to severe chronic pain has been linked to increased risky drinking over time (Tsui et al., 2014). PLWH report engaging in heavy alcohol use as a way of managing emotional stress that triggers chronic pain (Palfai et al., 2019), yet alcohol may impact sensitivity to pain (Sullivan et al., 2008). Given the pervasiveness of substance-use problems in this population, identification of psychosocial factors contributing to their development remains an important priority.

In addition to the connection between pain and substance use among PLWH, depression and anxiety have also been recognized as important psychological factors among PLWH and chronic pain (Brandt, Zvolensky, Daumas, Grover, & Gonzalez, 2016; Jones, Vogelman, Luba, Mumtaz, & Comer, 2017; Uebelacker et al., 2015). Negative mood states have been linked with higher levels of substance use disorders (Tsao, Plankey, & Young, 2012; Tsao & Soto, 2009) and opioid use (Rogers, LaRowe, Ditre, & Zvolensky, 2019). Moderate and severe symptoms of anxiety have also been identified as a risk factor for benzodiazepine use and hazardous alcohol consumption among PLWH (Mannes, Dunne, Ferguson, Cook, & Ennis, 2020; Parent et al., 2019). A diagnosis of chronic pain among PLWH has been associated with substance use, with both mood and anxiety disorders being predictors of use (Denis, Morales, Wu, Metzger, & Cheatle, 2019).

Although depression and anxiety have been recognized as important psychological factors among PLWH and chronic pain (Fekete, Williams, & Skinta, 2018; Scott et al., 2018; Uebelacker et al., 2015), the influence of other negative mood states remains largely unknown. Anger has received less attention among PLWH and chronic pain, possibly because it has proved challenging to discretely categorize or quantify, or has been conceptualized as a component or expression of other conditions, including depression. However, hierarchical frameworks have classified anger as a separate domain from depression and anxiety (Pilkonis et al., 2011), which is consistent with a view that anger is associated with stronger action tendencies (e.g., aggression) than depression or anxiety (Cox & Harrison, 2008; Martin et al., 2016). And, anger may have greater effects on pain severity than any other negative mood states (Fernandez & Turk, 1995; Vickers & Boocock, 2005). Previous research in non-HIV specific populations demonstrates that anger is bidirectionally associated with pain – it can be a predisposition to pain, exacerbate or perpetuate pain, or be a consequence of pain (Burns, Quartana, & Bruehl, 2008; Fernandez, 2005; Greenwood, Thurston, Rumble, Waters, & Keefe, 2003). PLWH with pain express anger similarly as other populations (Fernandez, 2005; Fernandez & Turk, 1995), but may be at increased risk of developing or continuing maladaptive health habits such as the use of substances as a way of coping with the stress of pain.

According to the stress-coping model of the addictive process, substance use represents one of many coping strategies that individuals may engage in to manage life stresses (Wagner, Myers, & McIninch, 1999; Wills & Hirky, 1996). Negative mood states, such as anger, may serve as both the cause and the product of such stresses, and substance use may serve as a means of regulating them. Thus, PLWH and chronic pain with high levels of anger may be more likely to turn to substance use to alleviate their distress. While initially effective at providing short term relief of distress, continued use of substances over time to manage distress can result in a substance use problem and/or deprive the individual the opportunity to practice healthier coping strategies (Wills & Hirky, 1996).

The current study examines whether anger is associated with substance use among PLWH who report chronic pain. It is hypothesized that 1) PLWH who report higher levels of pain will have higher severity of reported anger, and that 2) controlling for pain severity, persons with greater anger levels will report higher rates of alcohol, benzodiazepine and marijuana use. The interaction of high severity of reported anger and high levels of pain is expected to produce particularly high levels of substance use.

Methods

Participants

This study is an analysis of data from a longitudinal study for 187 HIV-infected patients who have pain recruited from three HIV primary care clinics in different states (site 1 n = 116; site 2 n = 15; site 3 n = 56) from November 2016 to February 2020.

Procedures

Procedures were identical across sites and approved by the relevant Institutional Review Boards. Individuals were considered eligible for participation if they 1) were HIV+, 2) reported chronic pain (pain duration for at least six months with a mean score of 3.5 or higher on the BPI Pain Interference Scale (Cleeland & Ryan, 1994); 3) had Pain severity ≳ 4 on a Numeric Rating Scale indicating “worst pain in the last week;” (Dworkin et al., 2005); 4) had elevated depression symptoms (Quick Inventory of Depression Symptoms (QIDS) (Rush et al., 2003) score of ≥ 9 (depression severity); 5) if taking antidepressant medications, those medications were stable for at least a month; 6) were aged 18 or older; and 7) had access to a telephone. Individuals were considered ineligible if they: 1) were in engaged in psychotherapy more than one time per month or in a multidisciplinary pain management program; 2) planned surgery in the next 6 months; 3) had pain thought due to cancer; 4) had current mania, as determined by the SCID-V (First, 2015); 5) reported a history of psychotic symptoms in the previous year, as determined by the SCID-V (First, 2015); 6) engaged in hazardous substance use (15+ days of cocaine or heroin/non-prescribed opiates or 4+ days of binge drinking in the previous month); 7) endorsed suicidality that required immediate clinical attention; 8) were pregnant.

If individuals were eligible on the initial screen, they were scheduled for a baseline visit that would occur just prior to a regularly scheduled appointment with their HIV care provider. The baseline visit was scheduled within two weeks of the eligibility screen and provides the data for this analysis. After informed consent, at the baseline visit, participants met with research assistants in private spaces to complete baseline assessment questionnaires and receive randomization into one of two conditions as part of the larger, randomized trial known as the HIV, Pain, and Sadness Support intervention.

Measures

We assessed demographic characteristics, levels of depression, levels of pain, levels of anger, and days of substance use.

Demographics.

All participants were asked to provide their age, marital status, race, ethnicity, education, and employment status. HIV history was assessed with self-reported date of diagnosis.

Depression.

The Quick Inventory of Depressive Symptoms – Self Report (QIDS-SR) was used to measure depressive symptom severity (Rush et al., 2003). Nine domains were separately scored on a Likert scale representing severity from 0 to 3, with a total possible range of 0 to 27 (no depression to very severe depression).

Pain.

To assess current pain severity, participants were asked the following question: “In the past week, on a 0–10 scale, where 0 = no pain and 10 = pain as bad as you can imagine, what number describes your pain on average?” (Dworkin et al., 2005).

Anger.

Anger was assessed using PROMIS Instruments that included item banks for Anger from PROMIS (Cella et al., 2007; Stone, Broderick, Junghaenel, Schneider, & Schwartz, 2016). PROMIS Anger, based on item response theory (Cella et al., 2014), presumes that emotional distress cannot be dichotomized, and assumes a continuum from mild to disabling symptoms focusing on the subjective feeling component. The PROMIS Anger scale differs from measurements of depression and/or anxiety in that it focuses on items that were affective and cognitive manifestations of anger, and also includes indicators of behavioral activation and anger expression (Pilkonis et al., 2011). PROMIS Anger (Short Form) items assessed angry moods (anger, frustration, irritability, annoyance) over the past seven days. Each item was rated on a 5-point scale (1 = never to 5 = always) with a range in score from 5 to 25. Higher scores on anger signified greater anger severity.

Substance use.

The main outcome variables were the rates of alcohol, marijuana, and benzodiazepine use, measured separately using the Addiction Severity Index – Lite Version (ASI-Lite) (McLellan, Luborsky, Woody, & O’Brien, 1980). The substances in the analyses were the most prevalent substances used in our sample; opioid use was not explored because it was often prescribed in this sample with chronic pain. Participants reported the number of days they had used each substance in the past 30 days. A summation of the number of days used for each substance was generated.

Analytical Methods:

We present descriptive statistics to summarize the characteristics of the sample. Frequency of alcohol, benzodiazepine, and marijuana use were assessed as 30-day rates. We used negative binomial regression to estimate the adjusted associations of anger, pain, and selected covariates with rates of substance use. For each substance we also tested the linear anger by linear pain interaction. We report incident rate ratios (IRRs) and 95% confidence interval estimates to summarize associations. We used the robust Huber-White variance estimator to estimate 95% confidence intervals and for all tests of significance. Because item specific missing data reduced the effective n for the model estimation by < 5%, we used listwise deletion. We calculated Pearson’s correlation coefficients to assess the strength of association between anger and pain.

Results

Participants averaged 54.1 (± 10.2) years of age, 56.7% were male, 38.4% were White, 33.3% were Black, 28.7% identified other or mixed racial origins, and 13.4% were Latinx (Table 1). About 10.2% and 12.9% were employed full- or part-time, respectively; employment status was dichotomized in subsequent analyses to indicate employed (full or part-time) and not employed. Educationally, 23.2% had not completed high-school, 29.2% had completed high school, 28.7% reported some college, and 18.9% reported completing college degrees; this was entered as a continuous covariate in multivariate models. Mean years of living with HIV was 18.5 (± 8.94) years and the mean score on the QIDS was 14.2 (± 2.99). Using suggested cut scores for QIDS (Rush et al., 2003), 12.8%, 53.5%, 33.2%, and 0.5% of the participants would be described as having mild, moderate, severe, and very severe depressive symptomology. Additionally, 28.9%, 13.4%, 43.3%, and 14.4% of the participants would be described as having none to slight, mild, moderate, and severe anger based on suggested cut scores for PROMIS – Anger (American Psychiatric Association, 2013). Mean days of alcohol, benzodiazepine, and marijuana use were 3.42 (± 7.15), 3.60 (± 9.53), and 8.25 (± 12.47), respectively. Descriptive statistics for average pain severity and the PROMIS anger scale are also reported in Table 1. Average pain severity and anger were significantly correlated at 0.267 (p < .001).

Table 1.

Baseline Characteristics (n = 187).

	n (%)	Mean	SD	Median	Min	Max
Clinic/Hospital Site  Site 1  Site 2  Site 3	116 (62.0%) 15 (8.0%) 56 (30.0%)
Yrs Age		51.4	10.2	53	22	75
Sex (Male)	106 (56.7%)
Race  White  Black  Multiple/Other	71 (38.4%) 61 (33.0%) 53 (28.7%)
Employed (Yes)	43 (23.1%)
Ethnicity (Latinx)	25 (13.4%)
Employment Status  Full-Time  Part-Time  Other	19 (10.2%) 24 (12.9%) 143 (76.9%)
Education  < 8th Grade  < HS  HS  Some College  College Degree  Post Graduate Degree	6 (3.2%) 37 (20.0%) 54 (29.2%) 53 (28.7%) 32 (17.3%) 3 (1.6%)
Years HIV Positive		18.5	8.94	18.8	0.76	51.9
QIDS Depression Index		14.2	2.99	14	9	21
Average Pain Severity		6.43	1.87	6.00	2	10
PROMIS Anger		16.1	4.64	16	5	25
Days Substance Use (0–30)
Days Used Alcohol		3.42	7.15	0	0	30
# Days Used BZDs		3.60	9.53	0	0	30
# Days Used Marijuana		8.25	12.47	0	0	30

Days of alcohol use was associated positively and significantly with PROMIS Anger (IRR = 1.14, p = .001) but was not associated significantly with pain (Table 2). Males (IRR = 3.81, p < .001) and White persons (IRR = 2.36, p = .014) have significantly higher adjusted rates of alcohol use than females and persons identifying other or mixed racial backgrounds, respectively. Days of alcohol use was not associated significantly with any of the other covariates included in the model.

Table 2.

Negative Binomial Regression Models Estimating the Adjusted Association of Pain and Anger with Days of Alcohol, Benzodiazepine, and Marijuana Use.

	SUBSTANCE
	Alcohol IRR (95% CI)	Benzodiazepines IRR (95% CI)	Marijuana IRR (95% CI)
Site 1	3.34**	0.00***	0.49
	(1.27; 8.77)	(0.00; 0.00)	(0.16; 1.45)
Site 2	1.91	0.59	1.27
	(0.96; 3.79)	(0.16; 2.26)	(0.70; 2.31)
Site 3 [REFERENCE]	[1.00]	[1.00]	[1.00]
Sex (1, Male)	3.81***	0.11***	1.05
	(2.22; 6.50)	(0.32; 0.37)	(0.63; 1.75)
Years Age	1.00	1.01	0.94***
	(0.97; 1.03)	(0.92; 1.09)	(0.91; 0.97)
Race (1, White)	2.36*	13.06***	1.13
	(1.19; 4.69)	(4.82; 35.35)	(0.59; 2.17)
Latinx (1, Yes)	1.51	0.15*	1.24
	(0.64; 3.56)	(0.03; 0.69)	(0.56; 2.77)
Education	1.22	1.22	0.86
	(0.94; 1.58)	(0.75; 1.98)	(0.69; 1.09)
Employed (1, Yes)	1.09	0.17*	1.06
	(0.54; 2.19)	(0.04; 0.68)	(0.56; 2.00)
Lives w Partner (1, Yes)	0.79	4.00	1.71
	(0.35; 1.78)	(0.81; 19.82)	(0.91; 3.23)
Years HIV+	0.97	1.08	1.02
	(0.93; 1.00)	(0.99; 1.17	(0.98; 1.07)
QIDS Depression	0.90	0.77**	0.95
	(0.80; 1.00)	(0.65; 0.93)	(0.86; 1.06)
Average Pain	0.95	8.84	1.07
	(0.81; 1.12)	(0.60; 1.18)	(0.92; 1.25)
PROMIS Anger	1.14***	0.82**	1.02
	(1.03; 1.18)	(0.73; 0.93)	(0.96; 1.10)
Constant	2.86	7.24	355.01***
Observations	182	182	181

Robust standard errors are in parentheses

^***p<.001,

^**p<.01,

^*p<.05

The linear anger by linear pain interaction was not statistically associated with the rate of alcohol use (IRR = 0.99, p = .479), rate of benzodiazepine use (IRR = 0.93, p = .172), or rate of marijuana use (IRR = 1.01, p = .758). The interaction of anger by pain was not included in any of the models we report.

Anger was inversely and significantly (IRR = 0.82, p = .002) associated with days of benzodiazepine use (Table 2). Males had significantly lower adjusted mean rates of benzodiazepine use than females (IRR = 0.11, = p < .001). White persons had significantly higher adjusted mean rates of benzodiazepine use (IRR = 13.06, p < .001) and persons who are Latinx had significantly lower expected rates of benzodiazepine use than non-Latinx persons (IRR = 0.15, p = .015). Employed persons had significantly lower rates of benzodiazepine (IRR = 0.17, p = .013) than those not employed. The rate of benzodiazepine use was also associated inversely and significantly associated with the QIDS depression index scores (IRR = 0.77, p = .005).

Neither pain (IRR = 1.07, p = .381) nor anger (IRR = 1.02, p = .486) was significantly associated with frequency of marijuana use when adjusting for other covariates included in the negative binomial regression model (Table 2). Frequency of marijuana use was inversely and significantly associated with age (IRR = 0.94, p < .001) but was not associated significantly with any other variables included in the model.

Discussion

This is the first study to investigate the association of pain severity, anger, and substance use in PLWH. Consistent with previous HIV-pain conceptual and empirical work supporting a bio-psychosocial approach (Merlin et al., 2014; Scott et al., 2018), our results showed that PLWH reporting higher levels of pain also report higher levels of anger. Our second hypothesis, that PLWH with greater levels of anger report higher rates of substance use, was partially supported. Results indicated that anger was significantly associated with alcohol and inversely related to benzodiazepine use, but not associated with marijuana use, even while controlling for depression, pain severity, and other covariates. That is, the relationship between anger and substance use was not accounted for simply by severity of depression symptoms. These results are in line with previous research on other negative mood states such as depression and anxiety (Basu, Chwastiak, & Bruce, 2005; Chaudhury, Bakhla, & Saini, 2016) suggesting that PLWH may be using alcohol as a way to cope with the stress expressed as anger (Wagner et al., 1999; Wills & Hirky, 1996). The relationship between anger and substance use was not moderated by average pain severity in our results, despite several studies showing a strong interaction between pain, mood, and substance use among PLWH (Merlin et al., 2013; Merlin et al., 2012; Merlin et al., 2014).

Although investigating anger among PLWH who report chronic pain is novel, there is an extensive literature regarding pain-related effects of anger in the context of chronic pain (Bruehl, Chung, & Burns, 2006; Burns et al., 2008). There is reason to suspect that the strong connection between anger and pain is not completely reducible to the effects of other negative mood states such as depression (Berkowitz, 1990; Berkowitz, 1993). Expression of angry feelings has been found to be the strongest predictor of reports of pain even when depression was concurrently examined (Kerns, Rosenberg, & Jacob, 1994). Thus, anger among PLWH and pain is an important concept to consider given the substantial stress associated with managing chronic health conditions. Indeed, anger may be more potent than other negative affective states in exacerbating the experience of pain, and/or leading to subsequent substance use to manage the anger or pain.

A relationship between pain and alcohol use was not found; however, there was a positive relationship between anger and alcohol use. In addition to the stress-coping model of the addictive process (Wagner et al., 1999; Wills & Hirky, 1996), investigators have examined people’s motives to drink alcohol in order to cope with distress or escape from negative feelings (Cooper, Frone, Russell, & Mudar, 1995), and viewed anger as a motivator for alcohol consumption (Hester & Miller, 1995; Marlatt & Donovan, 2008). Some studies have shown that males are more likely than females to report drinking to cope with distress (Nolen-Hoeksema & Harrell, 2002). Individuals who reported higher anger drank more frequently, and the association between angry emotions and drinking might be stronger for men than for women (Leibsohn, Oetting, & Deffenbacher, 1994). Controlling for depression, pain severity and other variables in the model, our results suggest that anger is a risk factor for alcohol use among PLWH, and that males are 3.8 times more likely than females to consume alcohol. Anger in this instance may serve as both the cause and the product of stressors from pain, and alcohol use may serve as a means of regulating these stressors. PLWH experiencing chronic pain with high levels of anger may be more likely to drink alcohol to alleviate their distress.

Contrary to our hypothesis, our results show an inverse association between anger and benzodiazepine use, and no association between pain and benzodiazepine use. Research has shown that chronic pain patients report benzodiazepine use despite its limited benefit in alleviating pain (King & Strain, 1990); no evidence supports the use of benzodiazepines for the treatment of chronic pain among HIV-infected individuals (Merlin, 2015). Current literature on the relationship between anger and benzodiazepine use is mixed, with one systematic review suggesting a “moderate” association between benzodiazepine use and anger (Albrecht et al., 2014). The circumstances under which anger might follow the use of benzodiazepines and the underlying mechanisms remain unclear. According to the stress-coping model, benzodiazepines may be used to alleviate distress. Our results suggest that 1) PLWH may be using benzodiazepines to reduce their anger and/or 2) PLWH may be using benzodiazepines to alleviate their depression. Both anger and depression are important considerations when evaluating the risk for benzodiazepine use among PLWH.

Although not statistically significant, our results were consistent with previous research indicating a positive relationship between pain severity and marijuana use among PLWH to help relieve pain (Sohler et al., 2018; Stein et al., 2015). There does not appear to be a significant relationship between anger and marijuana use, or depression and marijuana use in our study. Anger appears to operate differently than other negative mood states, such as depression or anxiety, where symptoms have been consistently associated with increased marijuana use (Hayatbakhsh et al., 2007). Marijuana use has been associated with younger age and considering the average age of 51 in our sample, marijuana use may not be as commonly used as alcohol to cope with anger.

There are several limitations to the current study. The cross-sectional study design is unable to determine causality, with respect to the reported effects between anger, depression, pain, and substance use. The current study enrolled patients with a pain severity score of at least 4 out of 10 and had elevated depressive symptoms (QIDS score of ≥ 9), which may produce a floor effect, underestimating the role of pain in the relationship between anger and substance use. The inclusion criteria limit our ability to generalize the results to PLWH without pain and depressed mood. Third, we assessed anger over past 7 days while substance use was assessed over past 30 days. Lastly, we did not have a comparison PLWH group without pain available.

In sum, we found a positive association between anger and alcohol use, and an inverse association between anger and benzodiazepine use, while controlling for depression. This study provides further evidence that anger in PLWH constitutes a distinct negative affect category, independent from depression. The associations may be underestimated due to high baseline levels of chronic pain in our sample. Future longitudinal studies are needed to understand if substance use is contributing to anger, or if PLWH are angry due to their pain and are using substances as a way to deal with the pain. Our findings suggest that anger is an important factor to consider when evaluating the risk for substance use among PLWH and chronic pain. Awareness of anger may aid in the assessment of substance use in this population, and indeed anger may serve as a promising target for treatment interventions. Cognitive-behavioral therapy in the treatment of pain in PLWH (Evans, Fishman, Spielman, & Haley, 2003; Huggins, Bonn-Miller, Oser, Sorrell, & Trafton, 2012), adapted to include elements targeting anger specifically (Deffenbacher, 2011), may ameliorate the effects of pain-related stressors among PLWH and decrease the impact of chronic pain.