Background

Pain is prevalent among HIV-infected individuals and it worsens with progression of HIV [1–8]. Use of opioid analgesics for treatment of chronic non-cancer pain remains controversial because of the limited evidence for its efficacy and concerns about the potential for misuse [9–12]. Identifying individuals at risk for opioid analgesic misuse is challenging [13].

Predictors of opioid analgesic misuse and pain, such as illicit substance use and depression, are also associated with decreased ARV adherence [14–21]. Pain is associated with decreased medication adherence among individuals with chronic medical conditions other than HIV infection [22, 23]. Prior studies among HIV-infected individuals have not demonstrated a significant association between pain and ARV adherence; however, these studies were small, cross sectional, had a narrow sampling of participants (e.g. from a single methadone or neurology clinic), and did not adjust for other variables known to be associated with decreased adherence, such as substance abuse and depression [19, 24, 25].

We found no studies examining the association of either taking prescribed, or misusing, opioid analgesics with medication adherence among HIV or non-HIV infected individuals. Opioid-related substance use disorders are associated with decreased ARV adherence, and treatment of opioid use disorders with methadone or buprenorphine improves ARV adherence among HIV-infected individuals [26–28].

We examined whether pain, opioid analgesic use, and opioid analgesic misuse were associated with self-reported ARV adherence in a cohort of HIV-infected indigent adults. We hypothesized that increased pain severity and the misuse of opioid analgesics would be associated with incomplete antiretroviral adherence, while appropriately using prescribed opioid analgesics would be associated with optimal adherence.

Materials and Methods

Study Population and Design

Pain study participants were enrolled from the Research in Access to Care in the Homeless (REACH) cohort, which consisted of individuals recruited using probability sampling from homeless shelters, free meal programs, and single room occupancy hotels who tested positive for HIV [29]. We attempted to recruit all REACH cohort members who came for a quarterly REACH follow-up interview from September 2007 thru June 2008 (n = 337) into the Pain Study, regardless of current pain status or opioid analgesic use. Of REACH cohort members active at the time, 87.8 % (296) participated in the Pain Study. All participants reporting an ARV regimen at any visit were included in our analysis (n = 258). For the purposes of our analysis, follow up begins with the first visit at which subjects reported being prescribed ARVs and continues through all subsequent visits.

REACH study visits involved a 45-min structured interview that assessed demographics, health status, depression, HIV medication use and adherence, recent illicit substance use, alcohol use, housing status, and recent incarceration. At baseline, and 7 quarterly follow-up visits, participants completed the Pain Study questionnaire, a 45-min structured interview conducted by trained interviewers about pain and use and misuse of analgesic medications. To minimize under-reporting of stigmatized behavior, participants self-administered questions about opioid analgesic misuse behavior using Audio Computer-Assisted Self Interview (ACASI) technology [30–33]

All study procedures took place at the Tenderloin Clinical Research Center (TCRC), a University-affiliated Clinical Research Center associated with the University of California, San Francisco (UCSF) Clinical and Translational Science Institute. All participants provided written and informed consent prior to participation. We gave each participant modest reimbursement for their participation. We received a Certificate of Confidentiality from the National Institute on Drug Abuse (NIDA).

Measurement of Adherence

We assessed adherence at each study visit by self-report of percentage of prescribed doses taken in the past 7 days. For participants not reporting taking ARVs at subsequent visits, we categorized them as having zero adherence. We further categorized adherence as ≥90 % versus <90 % adherence; 90–95 % adherence is the optimal minimal adherence for virologic control [34–36].

Measurement of Pain

At each interview, we assessed worst pain severity during the prior week using a 0–10 numeric rating scale, based on the modified Brief Pain Inventory [37–39]. We categorized responses of 1–4 as mild pain, 5–6 as moderate pain, and ≥7 as severe pain [40, 41].

Opioid Analgesic Prescriptions

At each quarterly interview, we asked participants if they took opioid analgesics prescribed by a health care provider in the past 90 days. If they did, we asked them to identify the opioid analgesic, dose, and schedule from a picture of pills representing all opioid analgesics on the market. Medications that were formulated for sustained release, such as oxycodone controlled release and morphine sulfate controlled release, were classified as long-acting; other opioid analgesics were classified as short-acting. We converted the total dose of medication to oral morphine equivalent doses and reported the participants’ total average daily dose [42, 43].

Opioid Analgesic Misuse

We created a global variable of opioid analgesic misuse using participant ACASI responses to a series of yes/no questions about past-90-day opioid analgesic behavior. For each quarterly interview, we coded participants as reporting opioid analgesic misuse if they endorsed one or more of the following: using opioid analgesics to get high, altering the prescribed route of administration for transdermal or oral medications, selling of opioids, borrowing of opioids, exchanging opioids for sex or illicit drugs, attempting to forge a prescription for opioid analgesics, stealing opioids from another person or a pharmacy, hospital or clinic, and buying opioids without a prescription.

Depression

We assessed depressive symptoms using the Beck Depression Inventory (BDI) and categorized scores of 0–13 as mild depression, 14–28 as moderate depression, and above 28 as severe depression [44]. In the multivariate model, we analyzed BDI score as a continuous variable.

Illicit Substance and Alcohol Use

We assessed current use of illicit drugs and alcohol at each quarterly interview. We categorized participants as current illicit substance users (versus not) based on self-reported use of cocaine, methamphetamine, and/or heroin in the past 90 days. We asked participants the average number of drinks per day on a typical day of drinking, in which one drink was defined as a 12 oz. can of beer, a 5 oz. glass of wine, or a 1.5 oz. drink of liquor. We defined problem drinking among participants if they consumed an average of >2 drinks per day (men) >1 drink per day (women) during the previous 30 days [45].

Health Status

We classified health as poor or fair versus good, very good or excellent based on participant self-report.

Statistical Analysis

Perceived general health and BDI score were missing for 7 and 5 observations, respectively. No other predictor variable had more than one missing observation. A total of 1,661 interviews were available for the 258 subjects. The mean number of interviews was 6.44 and 140 (54.3 %) subjects received all 8 pain interviews.

Odds ratios were calculated using Generalized Estimating Equations (GEE) with logit link. The exchangeable, m-dependent (with m = 2, 4 and 6), and auto-regressive structures were evaluated using the quasilikelihood under independence criterion (QIC) statistic, and the auto-regressive structure was selected based on this criterion. Predictors with a bivariate p value of 0.20 or less were included in multivariate models, which were then reduced using backward elimination until only predictors with p-value of 0.05 or less remained. We performed the analysis using SAS software (version 9.2).

Results

Of the 296 participants initially enrolled in the Pain Study, 258 participants reported taking ARVs at any visit during the Pain Study. At the baseline interview, participants were predominantly male (73.3 %) and African-American or White (39.9 % and 39.5 %, respectively), with a mean age of 48 (Table 1). A quarter of participants (27.1 %) reported moderate or severe depression, and 37.6 % rated their health status as poor or fair. Participants reported cocaine use (22.5 %), methamphetamine use (16.7 %), problem alcohol use (7.0 %), and heroin use (5.4 %) in the preceding 90 days.

Table 1 Characteristics of study participants
Full size table

At the baseline interview, most participants reported experiencing moderate (34.1 %) or severe (47.7 %) pain in the preceding week. Over half of participants (53.1 %) reported receipt of opioid analgesics prescribed by a health care provider in the past 90 days. Of those taking prescribed opioid analgesics, 34.7 % took more than 100 mg/day of oral morphine equivalent dose. One fifth of participants (20.5 %) reported opioid analgesic misuse in the preceding 90 days. Most participants (78.3 %) reported ≥90 % adherence to ARV therapy (Table 1). Subsequent visits in which participants who initially reported taking ARVs reported no longer receiving ARVs occurred for 103 (6.2 %) visits.

In a GEE analysis adjusting for age, education, homelessness, self-reported health status, depression, and substance use, neither severe pain (unadjusted odds ratio (OR) 1.37, CI 1.02–1.85) nor receiving prescribed opioid analgesics (OR 1.40 CI 0.99–1.97) was associated with incomplete ARV adherence (Table 2). Opioid analgesic misuse was independently associated with an increased odds of incomplete adherence [Adjusted odds ratio (AOR) 1.47, CI 1.06–2.03], as was illicit substance use (AOR 2.10, CI 1.53–2.87) (Table 2). Other variables associated with increased odds of incomplete adherence in the adjusted analysis were homelessness for more than one consecutive year (AOR 1.67, CI 1.07–2.59), depression (AOR 1.02, CI 1.01–1.04) and poor or fair self-rated health (AOR 1.49, CI 1.14–1.97).

Table 2 Bivariate and multivariate analysis of factors associated with suboptimal participant adherence to ARVs
Full size table

Discussion

In a cohort of indigent, HIV-infected adults on ARV therapy, we found a high prevalence of optimal ARV adherence. Neither pain nor receipt of prescribed opioid analgesics was associated with adherence to antiretroviral therapy. Opioid analgesic misuse was associated with a 47 % increased odds of incomplete adherence.

Use of prescribed opioid analgesics was not associated with ARV adherence, but misuse of opioid analgesics was significantly associated with incomplete adherence after adjustment for other substance use disorders. Opioid analgesic misuse likely represents a similar, but distinct, phenomenon as illicit substance use, in which individuals’ priorities are focused on obtaining and using the desired substance at the expense of social, occupational, and personal obligations, including health maintenance [46]. HIV-infected adults with co-morbid substance abuse disorders are less likely to follow-up and remain in care, and are more likely to have ARV treatment interruptions [47, 48].

Individuals misusing opioid analgesics may benefit from interventions that are successful in improving adherence among HIV-infected adults with a history of substance use, such as opioid substitution therapy and directly observed therapy for ARVs [49–52]. Case management can assist with coordination of social, mental health, and medical services, as well as linkage to substance abuse treatment and adherence to ARVs [52, 53]. Health care provider knowledge of opioid analgesic misuse should trigger an assessment of ARV adherence among HIV-infected individuals, and vice versa, with appropriate referral to services.

In our unadjusted analysis, severe pain was associated with decreased ARV adherence, but this association was not significant in the multivariate analysis. We have confirmed the negative findings of prior studies demonstrating no association between pain and ARV adherence with a larger cohort of HIV-infected individuals followed longitudinally while considering many potential confounders, such as illicit substance use and depression [19, 24, 25, 40]. Substance use and depression are associated with both pain and decreased medication adherence [54–56]. These factors likely confound the association of pain and medication adherence and were not controlled for in prior studies that demonstrated an association between pain and medication adherence among non-HIV-infected individuals [22, 23].

Receiving prescribed long acting opioid analgesics was not associated with decreased adherence. Opioids have not consistently demonstrated efficacy in reducing pain among HIV-infected adults, and there is no clear data that use of opioid analgesics improves self-efficacy or functional status in HIV-infected and uninfected adults [9, 57, 58]. Many providers believe that prescribing of opioids will improve patients’ quality of life [59]. Treating pain with opioid analgesics in an effort to promote ARV adherence is not supported by our findings.

Participants who were homeless for at least 1 year were more likely to have decreased ARV adherence, consistent with prior studies looking at homelessness and ARV adherence [28, 60]. The association between poor or fair self-reported health status and decreased ARV adherence is consistent with prior findings demonstrating that lower self-rated health is associated with lower CD4 cell counts and later ARV initiation [61].

Our study has several limitations. We collected misuse and adherence data by self-report, though we minimized bias and consequent underreporting of opioid misuse behaviors with the use of ACASI software. Prior studies within this sample found that self-report generally overestimates adherence, but correlates with electronic pill count monitoring and unannounced pill count measures [34, 62–64].

Our finding that opioid analgesic misuse increases the odds of incomplete ARV adherence warrants further exploration. Research into engagement in care, barriers to adherence, and self-efficacy among participants misusing opioid analgesics may further elucidate the association between opioid analgesic misuse and ARV adherence. In patients with pain and incomplete ARV adherence, treatment of pain may not improve adherence, but evaluation and management of co-occurring substance use disorders and depression might.