Pain and alcohol consumption among people living with HIV: examining the moderating roles of depression and social support

Abstract

Pain and heavy alcohol use are common among people living with HIV (PLWH), and influence one another, potentially exacerbating these conditions over time. This study examines the prospective association between pain and alcohol use among PLWH with a history of unhealthy drinking behaviors and/or substance use and tests whether depression and social support are moderators. A sample of 233 participants from the Boston Alcohol Research Collaborative on HIV/AIDS cohort completed measures of pain intensity (i.e., average severity of pain in the past week) and pain interference (i.e., average interference of pain in everyday life), heavy episodic drinking, number of drinks, social support and depression at baseline and 6 months later. Negative binomial regression analyses assessed whether pain at baseline predicted alcohol use at 6 months, and examined baseline social support and depression as moderators. Pain intensity was significantly associated with number of drinks (IRR= 1.80, 95% CI: 1.05, 3.08) but not number of heavy drinking days (IRR= 1.84, 95% CI: 0.83, 4.07), while pain interference was not associated with number of drinks (IRR= 1.63, 95% CI: 0.96, 2.75) nor heavy drinking days (IRR= 1.43, 95% CI: 0.64, 3.17) at six months. Neither social support, nor depression were significant moderators of the association between pain and 6-month alcohol use outcomes. Pain intensity is prospectively associated with more alcohol use, but not with heavy drinking among PLWH. We conclude that pain is an important factor to address when considering interventions to reduce alcohol use among PLWH.

1. Introduction

Heavy alcohol use is common among people living with HIV (PLWH) [1], and has been associated with a number of negative HIV-related outcomes, including problems with treatment engagement, poor antiretroviral therapy (ART) adherence and viral suppression, and increased behaviors associated with HIV-risk (e.g., sexual risk behaviors, needle sharing, and other substance use) [1, 2, 3, 4]. The impact of alcohol use on HIV outcomes is influenced by a number of significant co-occurring conditions including pain [5], increased viral replication, and poor HIV treatment outcomes [6] Further, alcohol use disorder in PLWH has shown to impact several organ systems, including the central nervous system, immune system and the musculoskeletal system, as well as affect neurological and cardiovascular functioning and accelerate disease progression. [7]. Pain is common among PLWH with prevalence exceeding 50% depending on the subpopulation [8, 9, 10] due to the nature of the disease. For example, most PLWH with chronic pain experience neuropathic pain, peripheral neuropathy musculoskeletal pain, and headache disorders, and many individuals have reported multisite pain (median of 2-5 sites of pain) [11, 12, 13]. Chronic pain in PLWH disproportionally affects older adults, women, individuals who were more medically complex, and those with a diagnosis of a psychiatric and/or substance use disorder [11]. Evidence suggests that pain and heavy alcohol use may have reciprocal influences on one another, which may contribute to the exacerbation of these conditions over time [14].

The prevalence of pain appears to be higher among heavy drinkers than in the general population [14], however, evidence that pain is associated with higher levels of alcohol use among PLWH has been equivocal [15]. There is evidence based on limited research that supports the view that pain interference (i.e., how much pain interferences in everyday life) is associated with heavy drinking status (e.g., based on National Institute on Alcohol Abuse and Alcoholism (NIAAA) criteria and Alcohol Use Disorders Identification Test (AUDIT) scores) [16, 17]. For example, Tsui et al. [16] found that at least moderate pain interference in the past month was significantly associated with past month risky drinking behaviors in PLWH who had pain compared to those without pain [16]. Ngo et al. [17] found that in PLWH, hazardous drinking was associated with greater pain interference compared to no hazardeous drinking [17]. However studies on the association between intensity of pain and heavy drinking have been decidedly more mixed. A number of cross-sectional studies have failed to find a significant association between pain intensity (i.e., average severity of pain) and hazardous drinking [17, 18, 19]. For example, Uebelacker et al. [19] failed to show a significant link between pain intensity and binge drinking in the past 30 days while Joseph et al. [18] did not find a significant association between severe pain and overall alcohol consumption [18]. However, Kuerbis et al. [20] found that greater levels of daily “worst” pain predicted heavier daily drinking (and vice versa) in older PLWH [20]. In one of the few prospective studies that examined this association among PLWH, Palfai et al. [21], did not find a significant association between moderate or greater pain intensity or pain interference and heavy drinking status one year later.

There are a variety of factors that may account for these mixed findings, including participant characteristics, indices of pain, and differences in measures of heavy drinking. Indeed, most studies that have explored the link between pain and alcohol use among PLWH have used categorical alcohol use measures, such as binge drinking status [19], hazardous or heavy drinking status [17, 18, 21], rather than examining quantity of standard drinks or frequency of heavy drinking episodes. Studying drinking amounts rather than the category of heavy drinking status may provide a more precise understanding of how pain may impact alcohol use among PLWH.

Another important gap in the literature has been efforts to identify potential moderators of the association between pain and alcohol use among PLWH. Depressive symptoms and social support may be particularly important in this regard given their associations with HIV-outcomes [22, 23, 24], pain [19, 25, 26, 27, 28], and alcohol use disorders [29, 30]. Previous work has suggested that depression may moderate the impact of pain on drinking and other substance use in neurosurgery and orthopedic outpatients [31] and the link between pain and functional limitations in older adults [32]. Regarding social support, greater levels of social support have been associated with a variety of positive health-related outcomes [33, 34], including lower pain intensity ratings among those with chronic pain [35, 36]. Moreover, social support has been shown to moderate the association between pain and both health-related quality of life [37] and alcohol use in older adults [38]. Despite the demonstrated importance of these factors in pain and alcohol use outcomes, there is little known about how depression and social support may impact the association between pain and alcohol use among PLWH. Finally, another limitation of the literature is that most studies have used cross-sectional designs [18]; few studies have explored the prospective impact of pain on indices of alcohol use over time among PLWH.

The aim of the current study is to examine the association between pain and alcohol use among PLWH by conducting a secondary analysis of the Boston Alcohol Research Collaborative on HIV/AIDS Cohort (Boston ARCH 4F Cohort), a sample of PLWH who have a history of substance use and/or unhealthy drinking [39]. Using a prospective design, the current study examines the impact of baseline pain intensity and pain interference on continuous measures of drinking behavior 6-months post-baseline, among a cohort of PLWH. In addition, this study examines depressive symptoms and social support as psychosocial moderators of the association between pain and heavy alcohol use among PLWH. We hypothesized that higher baseline pain intensity and interference would predict heavier alcohol use six months later, characterized by a higher number of drinks in the past two weeks and more frequent heavy drinking days in the past month. Moreover, it was hypothesized that this association would be stronger among those with lower levels of social support and higher levels of depression.

2. Methods

2.1. Procedures

Participants were recruited by phone from a prior cohort study and in-person from an adult infectious disease and HIV clinic at an urban academic medical center from 2018 to 2020 [33]. Eligibility criteria for the cohort included: (1) 18 years of age or older, (2) ability to speak English, (3) medical record confirmed HIV infection, and (4) willing and able to provide contact information for an alternative contact likely to know the participant’s whereabouts to assist with study follow-up, (5) any past 12 month use of illicit drugs, marijuana (not recommended by a healthcare provider), or nonmedical use of prescription medications (assessed using the Tobacco, Alcohol, Prescription Medication and Other Substances (TAPS) Tool [40]); and/or past 12-month alcohol use with positive AUDIT-C score (≥3 for females and ≥4 for males) [41]; and/or enrollment in the prior cohort study [39]. Exclusion criteria were: plans to leave the Boston area in less than one year, or inability to understand the research interview or provide informed consent (as determined by the judgement of study staff). For the current study, only participants with complete baseline and 6-month follow-up data were included. This study was approved by the Boston University Medical Campus Institutional Review Board. The full methods and measures used have been discussed in previous work [39, 42].

2.2. Measures

Trained research associates administered a standardized interview whereby participants reported their sociodemographics, quantity and frequency of alcohol use, pain severity and interference, social support, and depressive symptoms. Measures were completed at baseline and 6-month follow-up.

Alcohol variables were assessed using the Addiction Severity Index (ASI; [43]) and the Timeline Followback-14 day (TLFB; [44]). The first primary outcome was total quantity of standard drinks. The TLFB was used to assess the number of standard drinks consumed each day over the previous two weeks [44] which were summed to total number of drinks consumed over the past two weeks. The second primary outcome was number of heavy drinking days. The ASI evaluated the number of heavy drinking days in the past 30 days defined as days with 4 or more drinks (females) or 5 or more drinks (males). To reduce the influence of outliers, outcome values greater that 3 standardized deviations above the mean were set to 1 greater than the next highest value [45]. Both the ASI and TLFB are valid and reliable measures that have been used to assess PLWH [46, 47]. Hazardous drinking has been reported in 24% - 27% of PLWH (although definitions of hazardous drinking varied) [48, 49].

Pain intensity and pain interference, the primary independent variables, were assessed with the Brief Pain Inventory (BPI; [50]) using a “past week” timeframe. Pain intensity was assessed using one item regarding how the individual would describe their average pain in the past week on a 0 to 10 scale with 0 indicating “no pain” and 10 indicating “pain as bad as you can imagine.” Pain interference was rated on a series of 7 items, each scored 0 to 10 where 0 indicated pain “does not interfere” and 10 indicated pain “completely interferes” with daily life (e.g., general activity, mood, walking ability, sleep, etc.) over the past week the mean rating of the 7 items constituted the interference score. Both pain intensity and interference were modeled as dichotomous variables, comparing “none/mild” to “moderate to severe.” Moderate to severe pain intensity was defined as rating of ≥ 4 vs. ≤ 3 indicating no/mild pain intensity [51, 52], whereas pain interference was defined as an average pain interference score of ≥ 2 (in contrast to ≤ 1 which indicated no or mild pain interference) [53, 54].

To assess social support, we used a modified measure of the Medical Outcomes Study Social Support Survey (MOS-SSS; [55]), an 8-item questionnaire assessing frequency of social support. Participants rated the availability of types of social support (e.g., “how often is someone available to help you if you were confined to bed”) on a 5-point scale ranging from “none of the time” to “all of the time” during their baseline visit. Scores were calculated as the average of the two subscales (emotional and instrumental social support) on the measure and transformed to a 0 to 100 scale, with higher scores representing greater social support [55]. In our analysis, we created a dichotomous variable (High/Low) based on a median split. The MOS-SSS has shown to be a valid and reliable measure of social support in PLWH [56], and in the current study, the measure demonstrated excellent internal consistency (α = .92).

Depression was defined as Center for Epidemiologic Studies Depression Scale, 10-item version (CES-D-10; [57]). Participants rated how often they felt symptoms of depression in the past week on a 4-point scale ranging from “rarely or none of the time (less than 1 day)” to “most or all of the time (5-7 days)” during their baseline visit. Clinically significant depression is indicated by a score of 10 or greater [57]. In our analysis, we created a dichotomous variable based on CES-D-10 scores ≥ 10 (depression) versus scores < 10 (no depression). The CES-D has demonstrated strong validity and reliability in PLWH [58, 59].

Measures of HIV disease severity (CD4 count and HIV viral load) were abstracted from the electronic medical record.

2.3. Statistical Analysis

First, we examined descriptive statistics for sociodemographics, pain variables, and alcohol use (see Table 1). Continuous variables were screened for normality and evidence of skewness and kurtosis. Separate negative binomial regression analyses were conducted to assess whether pain intensity and pain interference respectively at baseline predicted alcohol use (i.e., number of heavy drinking days in the past month and number of drinks in the past two weeks) at 6 months. Covariates for these analyses included age, sex, past month non-prescribed or illicit drug use (illicit cocaine, opioid, sedative or marijuana use) (Addiction Severity Index) and the corresponding alcohol consumption variable at baseline. Further, we also conducted exploratory analyses to examine whether severity level of pain intensity and pain interference was linked with alcohol outcomes. We created three groups for each pain variable to denote clinically significant levels of pain intensity and pain interference consistent with previous research (e.g., Hoffman et al. [54], Shi et al. [60]). These analyses effectively distinguish between moderate and severe levels for intensity and interference [retaining the original “no pain/mild pain” category]. Negative binomial analyses for the pain intensity and interference variables were conducted on alcohol outcomes that examined levels of moderate and severe intensity contrasted to the no/mild categories. Next, we examined social support and depression as moderators in subsequent negative binomial models. Due to a left skewed distribution of the data, we created a dichotomous variable of high and low social support based on the median split to facilitate interpretability of outcomes and to not make assumptions about the linear associations in the data. We confirmed the appropriateness of over-dispersion assumptions by assessing the dispersion parameter and its 95% confidence interval in the negative binomial model. All analysis was performed in SPSS Version 26 [61].

Table 1.

Baseline sociodemographic, alcohol use, and pain characteristics among a cohort of people living with HIV (N=233)

Characteristics	N (%)
Age in years, Mean (SD)	52.4 (10.3)
Sex
Female	74 (31.8)
Male	159 (68.2)
Race/ethnicity
Black, non-Hispanic	125 (53.6)
Hispanic or Latino	46 (19.7)
White, non-Hispanic	43 (18.5)
Multi-racial/other	19 (8.2)
Unhoused	30 (12.9)
Unmarried or not living with a partner	173 (74.2)
Unemployed	174 (74.7)
Illicit drug use in the past montha	152 (65.2)
Cannabis/marijuana	117 (50.2)
Cocaine	54 (23.1)
Sedatives/tranquilizers/ and/or other miscellaneous drug use	29 (12.4)
Heroin	23 (9.8)
Other prescription opioids	15 (6.4)
CD4 Count < 200	17 (7.5)
HIV Viral Load < 200	195 (87.4)
Number of heavy drinking days in the past month, Mean (SD)b
Baseline	2.9 (6.1)
6 months	2.2 (5.5)
Number of drinks in the past 2 weeks, Mean (SD)c
Baseline	17.8 (31.0)
6 months	12.2 (23.5)
Pain intensityd
No to minimal pain intensity	123 (52.8)
Moderate to severe pain intensity	110 (47.2)
Pain interferenced
No to minimal pain interference	122 (52.4)
Moderate to severe pain interference	111 (47.6)
Depressione	126 (54.1)
High social supportf	117 (50.2)
SD=Standard Deviation

^aPast month illicit drug use was assessed by the ASI

^bHeavy drinking days were assessed by the ASI and defined as at least 4 drinks (for females) or 5 drinks (for males) per occasion

^cNumber of drinks was examined by the TLFB

^dPain intensity and pain interference assessed by the BPI

^eClinically significant depression defined as a score of at least 10 on the CESD

^fHigh and low social support are based on a median split assessed by the mMOS-SS

3. Results

3.1. Sociodemographic characteristics

A total of 233 participants, aged 24-71 years with a mean age of 52 years, were included (See Table 1). A majority of the sample was male (68.2%), unemployed (74.7%), and not married or not living with a partner (74.2%). The sample also identified as 53.6% Black, non-Hispanic, 19.7% Hispanic or Latino, 18.5% White, non-Hispanic, and 8.2% multi-racial/other. Approximately half the population reported moderate to severe pain intensity (47.2%) and pain interference (47.6%) compared to no to minimal pain intensity (52.8%) and pain interference (52.4%). On average, participants consumed approximately 17.8 drinks (standard deviation [SD]=31.0, range: 0-207) in the past two weeks at baseline and 12.2 drinks (SD=23.5, range: 0-104) at 6 months. Participants had on average 2.9 heavy drinking days (SD=6.1) in the past month at baseline and 2.2 heavy drinking days (SD=5.5) at 6 months. A total of 18 participants were excluded from data analysis due to missing data, including 17 participant who did not complete 6-month follow up pain and alcohol questionnaires and 1 participant who did not complete baseline pain questionnaires. Correlations between all variables of interest are presented in Table 2. Please see Kim et al. [39] and Lucero et al. [5] for further descriptive information about this sample.

Table 2.

Correlational matrix for variables of interest (N=233)

Variable	Sex	Age	BDU	IT	INT	HDD	6HDD	BTD	6TD
Sex	-
Age	−.027	-
BDU	−.121	−.241***	-
IT	.112	.145*	.058	-
INT	.106	.123	.047	.837***	-
HDD	.055	−.069	.104	.052	.039	-
6HDD	.055	−.043	.080	.096	.069	.434***	-
BTD	−.070	−.106	.144*	.100	.087	.642***	.369***	-
6TD	−.022	−.104	.113	.124	.073	.471***	.837***	.473***	-

Note: *p<.05. **p<.01. ***p<.001; BDU= Baseline drug use; IT = pain intensity; INT = Pain interference; HDD= Baseline heavy drinking day; 6HDD = 6 month heavy drinking days; BTD = baseline total number of drinks; 6TD= 6 month total number of drinks

3.2. Effects of Pain at Baseline on Alcohol Use at 6 Months

Pain intensity was significantly associated with a greater number total number of drinks (IRR = 1.80, 95% CI: 1.05, 3.08, p = 0.03) but not heavy drinking days (IRR = 1.84, 95% CI: 0.83, 4.07, p = 0.13) at 6 months. Pain interference was not significantly associated with total number of drinks (IRR = 1.63, 95% CI: 0.96, 2.75, p = 0.07), nor was it associated with the number of heavy drinking days (IRR = 1.43, 95% CI: 0.64, 3.17, p = 0.39) at 6 months. Full model effects are presented in Table 3 and Table 4.¹

Table 3.

Results of negative binominal regression analysis examining the prospective association between baseline pain and total number of drinks in the past two weeks at the 6-month timepoint among a cohort of people living with HIV (N=233)

	Number of drinks in the past two weeks
	Adjusted analysis*			Base analysis**
	IRR	95% CI	p-Value	IRR	95% CI	p-Value
Pain intensity	1.80	1.05, 3.08	0.03	1.72	1.00, 2.99	0.05
Sex	1.00	0.52, 1.92	1.00
Age	0.97	0.95, 1.00	0.02
Drug use	1.52	0.75, 3.07	0.24
Baseline drinks	1.04	1.02, 1.05	0.01***
Pain interference	1.63	0.96, 2.75	0.07	1.62	0.93, 2.83	0.09
Sex	1.03	0.54, 2.00	0.92
Age	0.98	0.95, 1.00	0.03
Drug use	1.56	0.77, 3.15	0.22
Baseline drinks	1.04	1.02, 1.05	0.01***

IRR: incidence rate ratio; CI: confidence interval

^*Adjusted for the following covariates: baseline number of drinks, age, sex, unhoused status, past month drug use

^**Base analyses refer to analyses adjusting for baseline number of drinks only

^***p<.001

Table 4.

Results of negative binominal regression analysis examining the association between baseline pain and past-month number of heavy drinking days (HDD)at the 6-month timepoint among a cohort of people living with HIV (N=233)

	Heavy drinking days in the past month
	Adjusted analysis*			Base analysis**
	IRR	95% CI	p-Value	IRR	95% CI	p-Value
Pain intensity	1.84	0.83, 4.07	0.13	1.11	0.61, 2.04	0.74
Sex	1.06	0.46, 2.43	0.89
Age	1.00	0.97, 1.03	0.95
Drug use	1.32	0.44, 3.98	0.62
Baseline HDD	1.15	1.09, 1.22	0.01***
Pain interference	1.43	0.64, 3.17	0.39	1.53	0.72, 3.23	0.27
Age	1.00	0.97, 1.03	0.91
Sex	1.10	0.49, 2.49	0.81
Drug use	1.44	0.48, 4.33	0.52
Baseline HDD	1.15	1.09, 1.21	0.01***

IRR: incidence rate ratio; CI: confidence interval

^*Adjusted for the following covariates: baseline number of heavy drinking days, age, sex, unhoused status, past month drug use

^**Base analyses refer to analyses adjusting for baseline number of heavy drinking days only

^***p<.001

In the exploratory analyses that examined associations between level of pain and alcohol outcomes, those with severe pain interference consumed significantly more drinks than those with no/mild pain. No other comparisons were statistically significant. For levels of pain intensity, results of total drinks analyses were moderate IRR = 1.64 [0.87, 3.12], severe IRR = 1.96 [0.93, 4.15] and for heavy drinking days results showed moderate IRR = 1.41 [0.53, 3.75] and severe IRR = 2.32 [0.91, 5.87]. For levels of pain interference, results of total drinks analyses were moderate IRR = 1.22 [0.66, 2.49] and severe IRR = 2.14 [1.02, 1.05] and for heavy drinking days moderate: IRR = 0.64, [0.29, 1.40], severe: IRR = 2.34 [0.87, 6.28].

3.3. Social Support and Depression as Moderators of the Association Between Pain and Alcohol Use

Social support did not significantly moderate the association between pain intensity and number of drinks (IRR = 1.11, 95% CI: 0.42, 2.94, p = 0.84) nor heavy drinking days (IRR = 0.67, 95% CI: 0.14, 3.32, p = 0.63). Social support was also not a significant moderator of pain interference and number of drinks (IRR = 1.23, 95% CI: 0.45, 3.31, p = 0.69) or with heavy drinking days (IRR = 0.85, 95% CI: 0.17, 4.28, p = 0.85).²

Similarly, depression did not moderate the relationship between pain intensity and number of drinks in the past two weeks (IRR = 1.87, 95% CI: 0.70, 4.95, p = 0.21) nor heavy drinking days (IRR = 0.80, 95% CI: 0.16, 4.01, p = 0.78) at 6 months. Depression was also not a significant moderator of pain interference and number of drinks in the past two weeks (IRR = 2.09, 95% CI: 0.78, 5.61, p = 0.15) nor heavy drinking days (IRR = 1.44, 95% CI: 0.30, 6.86, p = 0.65) at 6 months. Full model effects for social support and depression are presented in Table 5.

Table 5.

Negative binominal regression analyses examining the interactions between social support and depression on the prospective relationship of pain and alcohol use among a cohort of people living with HIV (N=233)

	Heavy drinking days (HDD) in the past month			Number of drinks in the past two weeks
	IRR	95% CI	p-Value	IRR	95% CI	p-Value
Pain intensity (PI)
PI X Social support	0.67	0.14, 3.32	0.63	1.11	0.42, 2.94	0.84
Sex	1.06	0.50, 2.28	0.88	0.90	0.51, 1.60	0.72
Age	1.00	0.97, 1.03	0.95	0.97	0.95, 0.99	0.01
Drug use	1.34	0.48, 3.70	0.58	1.69	0.94, 3.05	0.08
Baseline drinks/HDD	1.15	1.09, 1.21	0.01***	1.04	1.03, 1.06	0.01***
PI X Depression	0.80	0.16, 4.01	0.78	1.87	0.70, 4.95	0.21
Sex	1.18	0.51, 2.74	0.70	0.95	0.50, 1.80	0.87
Age	1.00	0.97, 1.03	0.86	0.98	0.96, 1.00	0.06
Drug use	1.74	0.66, 4.60	0.27	1.70	0.87, 3.29	0.12
Baseline drinks/HDD	1.16	1.10, 1.22	0.01***	1.04	1.03, 1.05	0.01***
Pain interference (INT)
INT X Social support	0.85	0.17, 4.28	0.85	1.23	0.45, 3.31	0.69
Sex	1.09	0.51, 2.32	0.83	0.92	0.51, 1.63	0.77
Age	1.00	0.97, 1.03	0.86	0.97	0.95, 0.99	0.01
Drug use	1.48	0.53, 4.13	0.46	1.77	0.98, 3.17	0.06
Baseline drinks/HDD	1.15	1.09, 1.21	0.01***	1.04	1.03, 1.05	0.01***
INT X Depression	1.44	0.30, 6.86	0.65	2.09	0.78, 5.61	0.15
Sex	1.10	0.48, 2.55	0.82	0.98	0.52, 1.87	0.95
Age	1.00	0.97, 1.03	0.86	0.98	0.96, 1.00	0.06
Drug use	1.83	0.69, 4.87	0.23	1.70	0.87, 3.32	0.12
Baseline drinks/HDD	1.16	1.10, 1.22	0.01***	1.04	1.03, 1.05	0.01***

IRR: incidence rate ratio; CI: confidence interval

^***p<.001

4. Discussion

Our findings supported the hypothesis that pain intensity would be prospectively associated with more alcohol consumption as indicated by a greater number of drinks at the 6-month follow-up, but it did not predict frequency of heavy drinking episodes. Pain interference was not significantly associated with either quantity of alcohol consumed or frequency of heavy drinking days 6 months later. However, exploratory analyses suggested that a severe level of pain interference was associated with total quantity consumed suggesting that level of severity of pain interference may be relevant to consider in future work. Moderational analyses did not support hypotheses that social support and depression would moderate associations between pain and alcohol use outcomes as none of these analyses indicated significant interactions.

Previous studies have shown mixed results regarding the association between pain and alcohol use among PLWH as some have suggested that higher levels of pain may be associated with heavy drinking [16, 17, 20, 62], while others have not [18, 19]. Research in adults with chronic pain suggests that using alcohol to cope with pain leads to greater alcohol consumption, and those who perceive pain as a barrier to alcohol cessation were more likely to engage in hazardous drinking and be alcohol dependent [63, 64]. The current study supports the view that clinically significant pain (i.e., moderate or greater) is prospectively associated with higher consumption, even though rates of heavy drinking episodes and frequency of drinking did not seem to be significantly impacted by pain. Given the conflicting evidence in the literature, the current study adds to an important line of research on pain and alcohol use among the HIV community in a number of ways. First, it is one of the few studies that have explored the prospective association between pain and alcohol use among PLWH over time. With few exceptions [21] most studies have used cross-sectional designs [18, 65] rather than designs to test the predictive value of pain. Second, the current study offers a more precise measure of alcohol use by using continuous measures of drinking quantity and heavy drinking frequency. As noted above, most of the previous work exploring the association between pain and alcohol use among PLWH has used broad categorical measures of drinking status, such as binge drinking status [19], hazardous drinking status [17, 18], or frequency of drinking measured as categories with Likert scales [38]. Given risks to HIV outcomes associated with conditions that frequently co-occur with alcohol consumption among PLWH [1, 2, 3, 4, 5], better understanding the nature of this association is important for developing approaches to HIV care.

A third contribution is the examination of potential moderators of the link between pain and alcohol use. Depression and social support did not significantly moderate associations between pain variables and alcohol consumption outcomes, such that the effect of pain intensity and interference on drinking did not differ significantly in those with greater social support or those with lower levels of depression. Previous studies have shown that depression and social support moderate the association between pain and alcohol/substance use (prescription opioid misuse) in general populations [31, 38]. Given that depressive symptoms have been associated with negative outcomes, like higher levels of prescription opioid misuse [31], and social support with positive outcomes, such as protecting against stressors on poor health outcomes [66] and lower levels of alcohol consumption [38], it was surprising that depression and social support did not affect these associations. This suggests that the strength of the association between pain and alcohol would not vary by depression or social support status. Previous studies that have found evidence of moderation utilized different measures of pain, depression and social support [31, 38]. However, none of these studies was conducted among PLWH. To our knowledge, this study is the first to assess depression and social support as moderators of pain and alcohol use in PLWH. There may be other social determinants of health, not examined in this study, that may affect this association, such as stigma and discrimination [67, 68].

There are limitations to this study that should be noted. First, alcohol consumption was assessed using retrospective self-report measures, which may be influenced by social desirability and/or recall bias [69]. Alcohol measures completed at the 6-month outcome assessed heavy drinking over the previous month and number of drinks in the previous two weeks, therefore they did not account for drinking patterns within the full period of time following baseline pain assessment. Due to the nature of the data, we cannot ascertain whether the days with more pain were associated with a greater number of heavy drinking days and number of drinks. Future research assessing these associations, utilizing daily diaries or ecological momentary assessment are needed [20]. Second, it should be acknowledged that even though the sample is from a moderately large HIV cohort study, it was not specifically designed or powered to address the issues examined in this secondary analysis. Further, the study included a sample of PLWH with a substance use history or history of unhealthy drinking from an urban, HIV clinic setting in the northeast United States (U.S.). Although rates of pain intensity and pain interference for this sample are consistent with the range of pain rates in other studies among PLWH [70-74], it remains unclear whether these associations would be observed among a sample that was not specifically selected based on substance/alcohol use history or among other HIV populations such as those not engaged in HIV care, or those in rural or other geographic areas. Further, future work should consider the reciprocal influences of pain and alcohol use in the context of other substances as individuals may use substances such as cannabis and opioids as alternative to alcohol in order to manage pain. Future studies may also consider exploring levels of severity and interference of pain and alcohol use at different cutoffs (mild, moderate and severe). Finally, this study examined the prospective relationship between pain intensity and pain interference at baseline with alcohol use 6 months later. Levels of pain may fluctuate throughout a 6 month period which may affect alcohol consumption months later. Future studies examining pain and alcohol use at multiple time points, including 3 months, 9 months and 12 months may better explain the relationship between pain and future alcohol use in PLWH.

Despite these limitations, this study provides one of the only prospective studies of PLWH in the U.S. that examines the influence of both pain intensity and pain interference on drinking. It supports the importance of addressing pain when considering strategies to reduce alcohol consumption among PLWH.

Highlights.

• Pain intensity was associated with number of alcoholic drinks, but not number of heavy drinking days 6 months later in people living with HIV

• Pain interference was not associated with number of alcoholic drinks, nor number of heavy drinking days 6 months later

• Social support and depression are not significant moderators between baseline pain and alcohol use 6 months later.