Abstract
Concomitant with expanded legalization, cannabis is increasingly used to treat chronic pain among persons with HIV (PWH), despite equivocal benefit in research limited by small sample sizes and short duration of follow-up. To address these limitations, among a sample of PWH with pain interference enrolled in the Veterans Aging Cohort Study, we performed a target trial emulation study to compare the impact of four cannabis use strategies on pain interference. Among those receiving long-term opioid therapy (LTOT), we also explored impact of these strategies on ≥ 25% LTOT dose reduction. Among the analytic sample (N=1284), the majority were men with a mean age of 50. Approximately 31% used cannabis and 12% received LTOT at baseline. Adjusting for demographic and clinical factors, cannabis use in any of 4 longitudinal patterns was not associated with resolved pain interference over 12- to 24-month follow-up. Among 153 participants receiving LTOT at baseline, cannabis use at both baseline and follow-up was negatively associated with LTOT dose reduction compared to no use at both baseline and follow-up. These findings support other observational studies finding no association between cannabis use and improved chronic pain or LTOT reduction among PWH.
Keywords: chronic pain, cannabis use, long-term opioid therapy, target trial emulation study
Introduction
Chronic pain is highly prevalent among persons with HIV (PWH), can have major impacts on daily function, quality of life and well-being (Jessica S Merlin, Selwyn, Treisman, & Giovanniello, 2016), and can contribute to decreased adherence to antiretroviral medications (Jessica S Merlin et al., 2018). Chronic pain, including musculoskeletal and neuropathic pain that frequently accompany HIV/AIDS (Ghosh, Chandran, & Jansen, 2012), is often difficult to treat and PWH and their clinicians may struggle to find effective pain care regimens (Jessica S Merlin, Bulls, Vucovich, Edelman, & Starrels, 2016; Uebelacker et al., 2015). Furthermore, long-term opioid therapy (LTOT), a highly-prevalent treatment for chronic pain among PWH (Becker et al., 2016; E. Jennifer Edelman et al., 2013), is of increasing concern as mounting data on LTOT suggest an often unfavorable balance between benefits—which may be neglible (Chou et al., 2015)—and harms, which may include pulmonary and cardiovascular complications (Solomon et al., 2010), invasive pneumococcal infections (E Jennifer Edelman et al., 2019), opioid use disorder, and increased risk of overdose (Bohnert et al., 2011), and all-cause mortality (Gordon et al., 2020; Weisberg et al., 2015). The harms of LTOT are generally dose-dependent, and thus guidelines generally support avoiding LTOT initiation as a first-line therapy for the treatment of chronic pain, limiting dose when LTOT is used, and helping patients reduce LTOT dose when benefit no longer outweighs harm (Dowell, Haegerich, & Chou, 2016; VA/DoD, 2017).
In this context, there is increasing interest in how cannabis (a plant with hundreds of constituents, many of which are targeted for ongoing medication development), may play a role in treating chronic pain among PWH and helping reduce LTOT dose (Maharajan et al., 2020; Savage et al., 2016). To date, a small number of studies with significant limitations, including small, highly-selected samples with short follow-up periods, have shown that benefits of cannabis-based products in neuropathic pain may be outweighed by potential harms in some patients (Mücke, Phillips, Radbruch, Petzke, & Häuser, 2018). Data on cannabis to treat chronic musculoskeletal pain—also highly prevalent among PWH—and to reduce dose among those on LTOT are even more sparse (Nielsen et al., 2017; Nugent et al., 2017). In the US under the Controlled Subtances Act, cannabis is a schedule I substance, indicating a high potential for misuse and no accepted medical use, a designation that requires additional layers of approval to conduct clinical research. These legal restrictions, in addition to the complex and variable chemically active contents of herbal cannabis, have slowed efforts to execute large randomized, controlled trials (RCTs) of cannabis to treat chronic pain with adequate follow-up (Savage et al., 2016). Despite limited evidence on benefits, several states include HIV/AIDS and/or intractable chronic pain as qualifying conditions to be eligible to receive medical cannabis (National Conference of State Legislatures, 2020).
When ethical, legal or practical barriers to conducting adequately powered RCTs predominate, as is the case with cannabis research for pain, emerging literature supports the use of observational methods to emulate a large RCT (Caniglia et al., 2018; Caniglia et al., 2020; Hernán & Robins, 2016). Therefore, our objective was to use observational methods within a large, ongoing prospective cohort study of PWH to compare cannabis use or no cannabis use on the primary outcome of reduction in pain-related functional interference (from here forward termed “pain interference”), and an exploratory outcome of LTOT dose reduction among those with pain interference. Pain interference, or the degree to which pain impacts function, is recommended as a more robust measure of chronic pain than pain intensity as the latter is unidimensional and may not capture the biopsychosocial impact of the lived experience with chronic pain (Dworkin et al., 2005). As longer-term observational studies have thus far demonstrated null findings (Campbell et al., 2018; Jessica S. Merlin et al., 2019), we hypothesized that cannabis use (compared to no cannabis use) would not be associated with long-term improved pain interference or LTOT dose reduction.
Materials and methods
Study overview
To conduct this emulated target trial, we used data from the Veterans Aging Cohort Study (VACS), described in detail elsewhere (Justice et al., 2006). VACS is a longitudinal, multisite study of patients with and without HIV receiving care within the Veterans Health Administration of the Department of Veterans Affairs (VA) in Manhattan/Brooklyn, New York; Bronx, New York; Pittsburgh, Pennsylvania; Atlanta, Georgia; Houston, Texas; Baltimore, Maryland; Washington, D.C.; and Los Angeles, California. Data sources include patient surveys, collected approximately annually, linked to VA electronic health record data including diagnoses, laboratory, and pharmacy data, as well as administrative data. Between 2002 and 2018, there have been seven VACS survey waves with an enrollment of 3728 PWH. The institutional review boards at Yale University, VA Connecticut Healthcare System, and each participating site approved the study.
We used observational data to emulate an RCT to test the following hypothesis: among PWH with pain interference, cannabis use compared to no cannabis use would not be associated with a significant difference in resolved pain interference at 12–24 months of follow up. As an exploratory analysis in a subsample of those receiving LTOT at baseline, we examined the impact of cannabis use on LTOT dose reduction.
Study sample and observation period
We considered as eligible for inclusion in this target trial PWH who reported, during at least one study visit, pain interference, denoted by a response of moderately, quite a bit, or extremely to the survey item: “During the past 4 weeks, how much did pain interfere with your normal work (including both work outside the home and housework)?” The first study visit during which pain interference was reported served as the baseline enrollment date for each participant. We then followed participants until they completed their first follow-up survey, occurring 12–24 months post-baseline. Follow-up surveys occur annually; we allowed those missing one follow-up survey to remain in the study.
Treatment strategies
We compared the treatment strategy of cannabis use, defined as a response of less than once a month, 1–3 times a month, 1–3 times a week, 4–6 times a week, or every day to the survey item: “Please fill in the circle that best indicates how often in the past 12 months you used marijuana or hashish” to no cannabis use, defined as a response of “no use in the last year” to this item. We assessed cannabis use at baseline and follow-up to create a 4-level variable as follows: no cannabis use at baseline/no cannabis use at follow up; no cannabis use at baseline/cannabis use at follow up; cannabis use at baseline/no cannabis use at follow up; cannabis use at baseline/cannabis use at follow up. This approach allowed examination of impact of past-year cannabis use on pain interference at follow-up among those with and without prevalent cannabis use at baseline.
Primary and exploratory outcomes
The study’s primary outcome was resolved pain interference at follow-up, defined as a response of “not at all” or “a little bit” to the pain interference item listed above. Participants who reported at least moderate pain interference at follow-up were considered to have non-resolved pain interference. A priori, based on LTOT tapering studies using 25–50% reductions to indicate success (Frank et al., 2017), we defined our exploratory outcome of LTOT dose reduction as a ≥ 25% decrease in milligram (mg) morphine equivalent daily dose (MEDD) from baseline to follow-up. Our method for calculating mg MEDD, described in detail elsewhere (E. Jennifer Edelman et al., 2013), used pharmacy fill data of outpatient opioid prescriptions within the VA and excluded methadone and buprenorphine formulations used in the treatment of opioid use disorder.
Other measures
We assessed the following demographic variables at baseline: age, gender, race/ethnicity (categorized as White/non-Hispanic, Black/non-Hispanic, Hispanic (any race) and Other (multiple race and unknown)), and calendar year of baseline survey. Additionally, we assessed the following clinical variables at baseline: cigarette smoking (never, current or former) and past-year, self-reported opioid use, unhealthy alcohol use, and stimulant use from the VACS survey; chronic hepatitis C infection (yes/no) based on laboratory data and ICD 9/10 diagnoses; cancer diagnosis (yes/no) – excluding basal and squamous cell skin cancer—from the VA cancer registry; depressive symptoms (yes/no) based on Patient Health Questionnaire (PHQ)-9 score ≥ 10 from VACS survey; post-traumatic stress disorder (PTSD) based on ICD-9/10 diagnostic codes; antiretroviral medication receipt and benzodiazepine receipt (none, short- or long-term) from VA pharmacy fill and refill data; and CD4 cell count, undetectable viral load (< 500 copies/ml) and VACS Index 2.0 – a measure of disease severity correlated with mortality (Tate, Sterne, & Justice, 2019) from laboratory data.
Data analysis
We performed descriptive analysis of patients in the sample and evaluated their baseline characteristics by resolved pain interference at follow-up using a chi-square test for categorical variables and T-test or Kruskal-Wallis for continuous variables. The overall type I error rate was controlled at 0.05. Multivariable analyses used logistic regression with the primary independent variable of interest being treatment strategy and primary outcome being odds of pain interference resolution. We included all demographic variables in the multivariable model and selected clinical co-variates based on: a) biologic plausibility as a confounder, and b) non-colinearity with other variables already in the model(s). We did not include the model variables that constitute part of the VACS index (e.g., CD4 count, viral load). Among the subsample of patients receiving LTOT at baseline, we performed similar descriptive, univariate and multivariable analyses of the exploratory outcome (LTOT dose reduction ≥ 25%). All analyses were conducted in SAS version 9.4 (Cary, NC), and all tests were two-sided.
Results
Description of the analytic sample (Table 1)
Table 1.
Sociodemographic and clinical characteristics of HIV-infected veterans participating in the Veterans Aging Cohort Study (VACS) who reported pain interference during at least one study visit (2002–2011).
| Variables, n (%) | Overall, (N=1284) | Pain interference not resolved at first follow-up N=688 |
Pain interference resolved at first follow-up N=596 |
p-value |
|---|---|---|---|---|
| Age, mean (SD) | 49.8 (7.8) | 50.3 (7.7) | 49.3 (8.0) | 0.03 |
| Gender: | 0.07 | |||
| Male | 1244 (96.9) | 661 (96.1) | 583 (97.8) | |
| Female | 40 (3.1) | 27 (3.9) | 13 (2.2) | |
| Race/ethnicity: | <0.01 | |||
| White/non-Hispanic | 282 (22.0) | 170 (24.7) | 112 (18.8) | |
| Black/non-Hispanic | 816 (63.6) | 403 (58.6) | 413 (69.3) | |
| Hispanic (any race) | 123 (9.6) | 71 (10.3) | 52 (8.7) | |
| Other (multiple race and unknown) | 63 (4.9) | 44 (6.4) | 19 (3.2) | |
| Cannabis at baseline, past year: | 0.40 | |||
| Never tried/none in past year | 881 (68.6) | 468 (68.0) | 413 (69.3) | |
| Less than once a month | 138 (10.8) | 76 (11.1) | 62 (10.4) | |
| 1–3 times per month | 87 (6.8) | 41 (6.0) | 46 (7.7) | |
| 1–3 times per week or more | 178 (13.9) | 103 (15.0) | 75 (12.6) | |
| Cannabis treatment strategies: | 0.37 | |||
| No use at baseline, no use at follow-up | 783 (61.0) | 420 (61.1) | 363 (60.9) | |
| No use at baseline, use at follow-up | 98 (7.6) | 48 (7.0) | 50 (8.4) | |
| Use at baseline, no use at follow-up | 141 (11.0) | 70 (10.2) | 71 (11.9) | |
| Use at baseline, use at follow-up | 262 (20.4) | 150 (21.8) | 112 (18.8) | |
| Cigarette Smoking: | 0.58 | |||
| Never | 262 (20.5) | 133 (19.4) | 129 (21.7) | |
| Current | 704 (55.0) | 383 (55.8) | 321 (54.0) | |
| Former | 315 (24.6) | 171 (24.9) | 144 (24.2) | |
| Opioid use, past year | 498 (38.8) | 279 (40.6) | 219 (36.7) | 0.16 |
| Unhealthy alcohol use, past year | 479 (37.3) | 246 (35.8) | 233 (39.1) | 0.22 |
| Stimulant use, past year | 58 (4.5) | 30 (4.4) | 28 (4.7) | 0.77 |
| Hepatitis C | 482 (37.5) | 260 (37.8) | 222 (37.3) | 0.84 |
| Cancer diagnosis | 263 (20.5) | 135 (19.6) | 128 (21.5) | 0.41 |
| Depressive symptoms | 380 (29.8) | 250 (36.7) | 130 (22.0) | <0.01 |
| PTSD | 113 (8.8) | 77 (11.2) | 36 (6.0) | <0.01 |
| Average mg morphine equivalent daily dose (MEDD) | <0.01 | |||
| none | 1131 (88.1) | 560 (81.4) | 571 (95.8) | |
| 1–50 mg MEDD | 108 (8.4) | 88 (12.8) | 20 (3.3) | |
| >50 mg MEDD | 45 (3.6) | 40 (5.7) | 5 (0.9) | |
| Benzodiazepine receipt, past year | 0.02 | |||
| No receipt | 1060 (82.6) | 549 (79.8) | 511 (85.7) | |
| Short-term | 85 (6.6) | 51 (7.4) | 34 (5.7) | |
| Long-term | 139 (10.8) | 88 (12.8) | 51 (8.6) | |
| Antiretroviral therapy receipt | 985 (76.7) | 526 (76.5) | 459 (77.0) | 0.81 |
| CD4 cell count, median IQR | 372 (224, 573) | 378 (221, 571) | 368 (227, 578) | 0.98 |
| HIV viral load, undetectable | 713 (56.1) | 388 (57.1) | 325 (54.9) | 0.44 |
| VACS index 2.0 score, median, IQR | 57 (48, 67) | 58 (48, 68) | 56 (47, 65) | 0.09 |
| Year of baseline | <0.01 | |||
| 2002–2004 | 808 (62.9) | 472 (68.6) | 336 (56.4) | |
| 2005–2007 | 243 (18.9) | 113 (16.4) | 130 (21.8) | |
| 2008–2011 | 233 (18.2) | 103 (15.0) | 130 (21.8) |
We identified 1,421 patients who met inclusion criteria and retained 1,284 who had a follow-up survey for the analytic sample (i.e., 137 patients without a follow-up survey within 12 to 24 months of their baseline asssessment were excluded from analysis). The mean age of the analytic sample was 50 years old; the vast majority were men consistent with the Veteran sample; and a majority (63.6%) self-reported Black race. Approximately 31% reported past-year cannabis use of varying degrees at baseline; most were current cigarette smokers (55%); and past-year opioid and unhealthy alcohol use were both prevalent (38.8% and 37.3%, respectively). Medical and mental health co-morbidities were relatively common, but large majorities of the cohort were not receiving LTOT or benzodiazepines (88.1 and 82.6%, respectively). A majority of the sample was receiving antiretroviral therapy and had an undetectable HIV viral load. Most entered the present study between 2002–2004 (62.9%).
Bivariate and multivariable associations with resolved pain interference
Slightly under half of the analytic sample (n=596, 46.4%) experienced resolved pain interference on follow-up. In bivariable analysis (Table 1), baseline past-year cannabis use was not significantly associated with resolved pain interference, nor was the 4-level cannabis treatment strategy variable. Those with resolved pain intereference were younger than those without resolved pain interefence. Race was associated with resolved pain interference: Black race participants were more likely than White participants to experience the primary outcome. The following baseline variables were also associated with a lower likelihood of resolved pain interference: depressive symptoms; PTSD diagnosis; LTOT receipt; benzodiazepine receipt; and year of study entry, with early (2002–2004) entry associated with the outcome.
On multivariable analysis (Table 2), none of the 4 cannabis treatment strategies were significantly associated with resolved pain interference. Black race was independently associated with reported resolution of pain interference, while baseline depressive symptoms, LTOT, and PTSD had negative associations with resolution of pain interference in the final model.
Table 2:
Multivariable logistic regression model of factors associated with resolved pain interference among HIV-infected veterans participating in the Veterans Aging Cohort Study (VACS) who reported pain interference during at least one study visit (n=1284)
| Variables, n (%) | Adjusted odds of resolved pain interference (AOR, CI) | p-value |
|---|---|---|
| Age (per year older) | 0.91 (0.77, 1.07) | 0.27 |
| Race/ethnicity: | <0.01 | |
| White/non-Hispanic | Ref | |
| Black/non-Hispanic | 1.52 (1.11, 2.06) | |
| Hispanic (any race) | 1.16 (0.73, 1.83) | |
| Other (multiple race and unknown) | 0.65 (0.35, 1.20) | |
| Cannabis treatment strategies: | ||
| No use at baseline, no use at follow-up | Ref | 0.42 |
| No use at baseline, use at follow-up | 1.16 (0.74, 1.83) | |
| Use at baseline, no use at follow-up | 1.32 (0.89, 1.96) | |
| Use at baseline, use at follow-up | 0.93 (0.68, 1.27) | |
| Cigarette Smoking: | 0.37 | |
| Never | Ref | |
| Current | 0.82 (0.60, 1.12) | |
| Former | 0.95 (0.66, 1.37) | |
| Unhealthy alcohol use, past year | 1.15 (0.90, 1.47) | 0.27 |
| Depressive symptoms | 0.53 (0.41, 0.70) | <0.01 |
| PTSD | 0.54 (0.34, 0.85) | <0.01 |
| Long-term opioid therapy (LTOT) receipt: | <0.01 | |
| No LTOT | Ref | |
| LTOT | 0.20 (0.12, 0.31) | |
| Benzodiazepine receipt: | 0.87 | |
| No receipt | Ref | |
| Short-term | 1.10 (0.67, 1.81) | |
| Long-term | 0.93 (0.62, 1.41) | |
| VACS index 2.0 score (5-point increments) | 0.97 (0.92, 1.01) | 0.13 |
Descriptive, bivariate and multivariable associations with LTOT reduction
The subsample with LTOT receipt (n=153) had a mean age of 51.4 and were primarily individuals of Black race, though a smaller majority than the overall analytic sample. Also, a larger majority of the subsample entered the study in the 2002–2004 period compared to the overall analytic sample. On bivariate analysis (Table 3), none of the variables was associated with a ≥25% LTOT dose decrease. In multivariable analysis (Table 4), the overall p-value for the 4-level cannabis treatment strategy variable was significant; use at baseline and follow-up was negatively associated with dose reduction among participants receiving LTOT (AOR 0.30; 95% CI 0.09–0.93), compared to the reference group (no use at baseline, no use at follow-up).
Table 3.
Sociodemographic and clinical characteristics of HIV-infected veterans participating in the Veterans Aging Cohort Study (VACS) who reported pain interference during at least one study visit, and who received long-term opioid therapy (n=153)
| Variables, n (%) | Overall, (N=153) | Less than 25% dose reduction, no dose change, or dose increase (N=119) |
At least 25% dose reduction (N=34) |
p- value |
|---|---|---|---|---|
| Age, mean (SD) | 51.4 (6.6) | 51.3 (6.8) | 51.6 (6.0) | 0.82 |
| Race/ethnicity: | 0.82 | |||
| White/non-Hispanic | 45 (29.4) | 33 (27.7) | 12 (35.3) | |
| Black/non-Hispanic | 88 (57.5) | 69 (58.0) | 19 (55.9) | |
| Hispanic (any race) | 14 (9.2) | 12 (10.1) | 2 (5.9) | |
| Other (multiple race and unknown) | 6 (3.9) | 5 (4.2) | 1 (2.9) | |
| Past year marijuana at baseline: | 0.20 | |||
| Never tried/none in past year | 99 (64.7) | 72 (60.5) | 27 (79.4) | |
| Less than once a month | 23 (15.0) | 19 (16.0) | 4 (11.8) | |
| 1–3 times per month | 8 (5.2) | 7 (5.9) | 1 (2.9) | |
| 1–3 times per week or more | 23 (15.0) | 21 (17.7) | 2 (5.9) | |
| Cannabis treatment strategies: | 0.11 | |||
| No use at baseline, no use at follow-up | 93 (60.8) | 69 (58.0) | 24 (70.6) | |
| No use at baseline, use at follow-up | 6 (3.9) | 3 (2.5) | 3 (8.8) | |
| Use at baseline, no use at follow-up | 18 (11.8) | 16 (13.5) | 2 (5.9) | |
| Use at baseline, use at follow-up | 36 (23.5) | 31 (26.1) | 5 (14.7) | |
| Cigarette smoking: | 0.15 | |||
| Never | 30 (19.6) | 27 (22.7) | 3 (8.8) | |
| Current | 82 (53.6) | 63 (52.9) | 19 (55.9) | |
| Former | 41 (26.8) | 29 (24.4) | 12 (35.3) | |
| Opioid use, past year | 74 (48.4) | 61 (51.3) | 13 (38.2) | 0.18 |
| Unhealthy alcohol use, past year | 52 (34.0) | 38 (31.9) | 14 (41.2) | 0.32 |
| Stimulant use, past year | 3 (2.0) | 2 (1.7) | 1 (2.9) | 0.53 |
| Long-term opioid therapy receipt: | 0.20 | |||
| 1–50 mg morphine eq daily dose (MEDD) | 108 (70.6) | 87 (73.1) | 21 (61.8) | |
| >50 mg MEDD | 45 (29.4) | 32 (26.9) | 13 (38.2) | |
| Benzodiazepine receipt: | 0.89 | |||
| No receipt | 98 (64.1) | 75 (63.0) | 23 (67.7) | |
| Short-term | 20 (13.1) | 16 (13.5) | 4 (11.8) | |
| Long-term | 35 (22.9) | 28 (23.5) | 7 (20.6) | |
| Hepatitis C | 60 (39.2) | 48 (40.3) | 12 (35.3) | 0.60 |
| Cancer diagnosis | 33 (21.6) | 22 (18.5) | 11 (32.4) | 0.08 |
| Depressive symptoms | 52 (32.4) | 41 (35.0) | 11 (32.4) | 0.77 |
| PTSD | 12 (7.8) | 9 (7.6) | 3 (8.8) | 0.73 |
| Antiretroviral therapy receipt | 134 (87.6) | 106 (89.1) | 28 (82.4) | 0.38 |
| VACS index 2.0 score, median, IQR | 59 (50, 70) | 60 (49.5, 72) | 56 (52, 64) | 0.44 |
| Year of baseline | 0.93 | |||
| 2002–2004 | 125 (81.7) | 96 (80.7) | 29 (85.3) | |
| 2005–2007 | 18 (11.8) | 15 (12.6) | 3 (8.8) | |
| 2008–2011 | 10 (6.5) | 8 (6.7) | 2 (5.9) |
Table 4:
Multivariable model of at least 25% dose reduction at follow-up among HIV-infected veterans participating in the Veterans Aging Cohort Study (VACS) who reported pain interference during at least one study visit and who were prescribed long-term opioid therapy (n=153)
| Variables, n (%) | Adjusted odds of ≥25% dose reduction (AOR, 95% CI) | p-value |
|---|---|---|
| Age, 10-year increments | 0.72 (0.34, 1.55) | 0.40 |
| Race: | 0.60 | |
| White/non-Hispanic | Ref | |
| Black/non-Hispanic | 0.58 (0.21, 1.57) | |
| Hispanic (any race) | 0.44 (0.08, 2.49) | |
| Other (multiple race and unknown) | 0.32 (0.03, 3.71) | |
| Cannabis treatment strategies: | ||
| No use at baseline, no use at follow-up | Ref | 0.05 |
| No use at baseline, use at follow-up | 3.03 (0.48, 19.12) | |
| Use at baseline, no use at follow-up | 0.27 (0.05, 1.39) | |
| Use at baseline, use at follow-up | 0.30 (0.09, 0.93) | |
| Cigarette smoking: | 0.13 | |
| Never | Ref | |
| Current | 3.03 (0.75, 12.25) | |
| Former | 4.76 (1.05, 21.57) | |
| Unhealthy alcohol use, past year | 1.99 (0.80, 4.95) | 0.14 |
| Depressive symptoms | 0.66 (0.24, 1.79) | 0.41 |
| PTSD | 1.24 (0.28, 5.52) | 0.78 |
| Benzodiazepine receipt | 0.60 | |
| No receipt | Ref | |
| Short-term | 0.49 (0.12, 1.96) | |
| Long-term | 0.86 (0.28, 2.64) | |
| VACS index 2.0 score (5-point increments) | 0.96 (0.81, 1.14) | 0.64 |
Discussion
In this large, prospective, target trial emulation study comparing cannabis use to no cannabis use on the outcome of resolved pain interference among PWH with pain, we found no association with cannabis use and improved pain, controlling for a wide array of potential confounders. This finding held true for individuals with and without cannabis use at baseline (i.e., among those with prevalent cannabis use and those initiating use during the study period). In an exploratory subanalysis examining LTOT dose reduction in a small subsample, those with baseline cannabis use who also reported cannabis use at follow-up were less likely to experience significant LTOT dose reduction. This study supported our hypothesis that cannabis use would not be associated with improved pain interference in long-term follow up. We interpret our exploratory findings on LTOT reduction with caution given the small sample size; nonetheless, our findings do not lend support for cannabis use as an effective strategy to promote opioid dose reduction and, in fact, we observed opposite associations.
Our primary findings are similar to a related study by Merlin et al. that included a smaller sample of PWH (N=433) and examined changes in pain and opioid therapy as a function of changes in cannabis use (Jessica S. Merlin et al., 2019). As in the present study, Merlin et al. found no evidence that cannabis use in PWH was associated with improved pain outcomes or reduced opioid prescribing. Our study supported Merlin et al.’s findings and extended them by including a larger sample and longer-duration follow-up and, most importantly, using a more robust and more clinically relevant pain outcome (i.e., pain interference compared to the 0–10 numeric rating scale of pain intensity). Consistent with our findings, a large observational study in a general sample of Australians similarly did not find an association between cannabis use and improved pain outcomes (Campbell et al., 2018). While some small, short-duration trials and cross-sectional studies (Sohler et al., 2018) have demonstrated or suggested analgesic or opioid-sparing benefits of cannabis (Nugent et al., 2017), to date, longer follow-up observational studies, such as this one, have not demonstrated benefit. This pattern of findings, also seen in the history of opioid analgesic studies (Hamilton & Gage, 2018), underscores the need for rigorous, controlled studies with large sample sizes, relevant outcome ascertainment, and adequate follow-up before clinical guidance can be generated.
Other interesting findings emerged in our work that deserve comment. First, Black race was associated with resolved pain interference at follow-up. Prior research has demonstrated that Black patients are less likely to have pain assessed, less likely to have pain treated in general, and less likely to receive specific treatments such as LTOT (Anderson, Green, & Payne, 2009; Becker et al., 2016; Green et al., 2003). While we do not know whether these patterns hold true in this sample, it is possible, perhaps even probable, that they do; thus, the finding that Black participants were more likely to experience resolved pain interference suggests better-developed pain self-management strategies in this group; however, this hypothesis needs more detailed exploration. Second, LTOT at baseline was associated with a lower likelihood of resolved pain interference at follow-up, a finding that reinforces findings that LTOT may not improve pain in the long-term compared to non-opioid therapies (Krebs et al., 2018). Finally, our findings that depressive symptoms and PTSD were independently associated with persistent pain interference corroborates literature demonstrating the challenge of successfully treating co-occuring mental illness and chronic pain (Bair, Wu, Damush, Sutherland, & Kroenke, 2008).
The factors related to LTOT dose reduction (or lack thereof) in our study deserve comment as well. We found that using cannabis at baseline and follow-up was negatively associated with dose reduction among PWH prescribed LTOT. This may be a reflection of prescriber behavior. Some prescribers believe cannabis use is an indication to discontinue LTOT while others do not (Cooke, Knight, & Miaskowski, 2019). A patient prescribed LTOT at baseline while using cannabis may be prescribed them by a prescriber willing to continue LTOT unchanged in the setting of ongoing cannabis use. A corollary point to this line of reasoning is that observed opioid-reduction effects of cannabis in clinical research in general may not be because cannabis is reducing pain—and thus reduced need for opioids—but in fact prescribers lowering doses because of safety concerns regarding co-use of cannabis and LTOT (Cooke et al., 2019; Gaither et al., 2018).
Our study has important limitations. First, we excluded patients who did not respond to a follow-up survey within 24 months of their first study visit during which pain interference was reported. If characteristics of those lost to follow-up differed from those who remained in the study, especially with respect to important confounders, it could have biased our results. However, given that less than 10% of the eligible sample failed to meet this inclusion criterion, we expect the magntitude of this selection bias, if present, to be small. Second, our treatment strategy (cannabis vs. no cannabis use) data was based on self-report without biologic verification meaning a possibility of self-report or social desirability bias. Morever, we were not able to differentiate prescribed medical cannabis use from recretional cannabis use. Third, while we had long-term follow up, we did not have interim data, leaving open the possibility of variable individual patterns of cannabis use that were not captured. Fourth, the LTOT-receiving subsample was unexpectedly small, restricting model stability and increasing likelihood of a type II error. Finally, we studied mostly male U.S. Veterans and only included LTOT received in the VA; thus, our findings may not generalize to other samples and we may have misclassified LTOT non-recipients. Large-scale studies in other countries where access to cannabis in various formulations and potencies may be different (Krcevski-Skvarc, 2018) than the present study are needed. Participants in a recent qualitative study in Canada—where cannabis has been legal since 2018—reported analgesic effects of cannabis (Chayama et al., 2021). Also of note was participants’ report of cannabis substitution for benzodiazepines and gabapentanoids – a compelling finding in need of further study among PWH.
In conclusion, our large target trial emulation study with up to 24 months follow up examining the impact of cannabis use on pain and LTOT outcomes found no association between cannabis use and improved pain or reduced LTOT. Future research with a larger sample to assess opioid dose-lowering effects and interim measures of cannabis use and pain are needed.
Acknowledgements
This work was supported by the National Institutes of Health [5U01AA020790-10 and R01 DA040471] and VA Health Services Research & Development [COR 19-489]. The views expressed in this article are those of the author(s) and do not necessarily represent the views of the VA or the United States government.
Footnotes
Declaration of interest statement
The authors have no conflicts of interest to report.
References
- Anderson KO, Green CR, & Payne R. (2009). Racial and ethnic disparities in pain: causes and consequences of unequal care. The Journal of Pain, 10(12), 1187–1204 [DOI] [PubMed] [Google Scholar]
- Bair MJ, Wu J, Damush TM, Sutherland JM, & Kroenke K. (2008). Association of depression and anxiety alone and in combination with chronic musculoskeletal pain in primary care patients. Psychosomatic medicine, 70(8), 890. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Becker WC, Gordon K, Edelman EJ, Kerns RD, Crystal S, Dziura JD, . . . Justice AC (2016). Trends in any and high-dose opioid analgesic receipt among aging patients with and without HIV. AIDS and Behavior, 20(3), 679–686 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bohnert AS, Valenstein M, Bair MJ, Ganoczy D, McCarthy JF, Ilgen MA, & Blow FC (2011). Association between opioid prescribing patterns and opioid overdose-related deaths. Jama, 305(13), 1315–1321 [DOI] [PubMed] [Google Scholar]
- Campbell G, Hall WD, Peacock A, Lintzeris N, Bruno R, Larance B, . . . Mattick RP (2018). Effect of cannabis use in people with chronic non-cancer pain prescribed opioids: findings from a 4-year prospective cohort study. The Lancet Public Health, 3(7), e341–e350 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Caniglia EC, Rebecca Z, Jacobson DL, Diseko M, Mayondi G, Lockman S, . . . Hernán MA (2018). Emulating a target trial of antiretroviral therapy regimens started before conception and risk of adverse birth outcomes. AIDS (London, England), 32(1), 113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Caniglia EC, Stevens ER, Khan M, Young KE, Ban K, Marshall BD, . . . Edelman EJ (2020). Does Reducing Drinking in Patients with Unhealthy Alcohol Use Improve Pain Interference, Use of Other Substances, and Psychiatric Symptoms? Alcoholism: Clinical and Experimental Research [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chayama KL, Valleriani J, Ng C, Haines-Saah R, Capler R, Milloy MJ, . . . McNeil R. (2021). The role of cannabis in pain management among people living with HIV who use drugs: A qualitative study. Drug Alcohol Rev. doi: 10.1111/dar.13294 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chou R, Turner JA, Devine EB, Hansen RN, Sullivan SD, Blazina I, . . . Deyo RA (2015). The effectiveness and risks of long-term opioid therapy for chronic pain: a systematic review for a National Institutes of Health Pathways to Prevention Workshop. Annals of internal medicine, 162(4), 276–286 [DOI] [PubMed] [Google Scholar]
- Cooke AC, Knight KR, & Miaskowski C. (2019). Patients’ and clinicians’ perspectives of co-use of cannabis and opioids for chronic non-cancer pain management in primary care. International Journal of Drug Policy, 63, 23–28 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dowell D, Haegerich TM, & Chou R. (2016). CDC guideline for prescribing opioids for chronic pain—United States, 2016. Jama, 315(15), 1624–1645 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Dworkin RH, Turk DC, Farrar JT, Haythornthwaite JA, Jensen MP, Katz NP, . . . Bellamy N. (2005). Core outcome measures for chronic pain clinical trials: IMMPACT recommendations. Pain, 113(1), 9–19 [DOI] [PubMed] [Google Scholar]
- Edelman EJ, Gordon K, Becker WC, Goulet JL, Skanderson M, Gaither JR, . . . Fiellin DA (2013). Receipt of opioid analgesics by HIV-infected and uninfected patients. Journal of general internal medicine, 28(1), 82–90. doi: 10.1007/s11606-012-2189-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- Edelman EJ, Gordon KS, Crothers K, Akgün K, Bryant KJ, Becker WC, . . . Marshall BD (2019). Association of Prescribed Opioids With Increased Risk of Community-Acquired Pneumonia Among Patients With and Without HIV. JAMA internal medicine [DOI] [PMC free article] [PubMed] [Google Scholar]
- Frank JW, Lovejoy TI, Becker WC, Morasco BJ, Koenig CJ, Hoffecker L, . . . Krebs EE (2017). Patient outcomes in dose reduction or discontinuation of long-term opioid therapy: a systematic review. Annals of internal medicine, 167(3), 181–191 [DOI] [PubMed] [Google Scholar]
- Gaither JR, Gordon K, Crystal S, Edelman EJ, Kerns RD, Justice AC, . . . Becker WC (2018). Racial disparities in discontinuation of long-term opioid therapy following illicit drug use among black and white patients. Drug and alcohol dependence, 192, 371–376 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ghosh S, Chandran A, & Jansen JP (2012). Epidemiology of HIV-related neuropathy: a systematic literature review. AIDS research and human retroviruses, 28(1), 36–48 [DOI] [PubMed] [Google Scholar]
- Gordon K, Manhapra A, Crystal S, Dziura J, Edelman E, Skanderson M, . . . Becker W. (2020). All-cause mortality among males living with and without HIV initiating long-term opioid therapy, and its association with opioid dose, opioid interruption and other factors. Drug and Alcohol Dependence, 216, 108291 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Green CR, Anderson KO, Baker TA, Campbell LC, Decker S, Fillingim RB, . . . Tait RC (2003). The unequal burden of pain: confronting racial and ethnic disparities in pain. Pain medicine, 4(3), 277–294 [DOI] [PubMed] [Google Scholar]
- Hamilton I, & Gage SH (2018). A painful lesson: are we repeating previous mistakes in pain management? The Lancet Public Health, 3(7), e309–e310 [DOI] [PubMed] [Google Scholar]
- Hernán MA, & Robins JM (2016). Using big data to emulate a target trial when a randomized trial is not available. American journal of epidemiology, 183(8), 758–764 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Justice AC, Dombrowski E, Conigliaro J, Fultz SL, Gibson D, Madenwald T, . . . Rimland D. (2006). Veterans aging cohort study (VACS): overview and description. Medical care, 44(8 Suppl 2), S13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krcevski-Skvarc N. 2018. Availability and approval of cannabis-based medicines for chronic pain management and palliative/supportive care in Europe: A survey of the status in the chapters of the European Pain Federation. European journal of pain, 22(3), 440–454. doi: 10.1002/ejp.1147 [DOI] [PubMed] [Google Scholar]
- Krebs EE, Gravely A, Nugent S, Jensen AC, DeRonne B, Goldsmith ES, . . . Noorbaloochi S. (2018). Effect of opioid vs nonopioid medications on pain-related function in patients with chronic back pain or hip or knee osteoarthritis pain: the SPACE randomized clinical trial. Jama, 319(9), 872–882 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Maharajan MK, Yong YJ, Yip HY, Woon SS, Yeap KM, Yap KY, . . . Yap KX (2020). Medical cannabis for chronic pain: can it make a difference in pain management? Journal of anesthesia, 1–9 [DOI] [PubMed] [Google Scholar]
- Merlin JS, Bulls HW, Vucovich LA, Edelman EJ, & Starrels JL (2016). Pharmacologic and non-pharmacologic treatments for chronic pain in individuals with HIV: a systematic review. AIDS care, 28(12), 1506–1515 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Merlin JS, Long D, Becker WC, Cachay ER, Christopolous KA, Claborn KR, . . . Liebschutz JM (2019). Marijuana Use Is Not Associated With Changes in Opioid Prescriptions or Pain Severity Among People Living With HIV and Chronic Pain. Journal of acquired immune deficiency syndromes (1999), 81(2), 231–237. doi: 10.1097/QAI.0000000000001998 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Merlin JS, Long D, Becker WC, Cachay ER, Christopoulos KA, Claborn K, . . . Kertesz SG (2018). The association of chronic pain and long-term opioid therapy with HIV treatment outcomes. Journal of acquired immune deficiency syndromes (1999), 79(1), 77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Merlin JS, Selwyn PA, Treisman GJ, & Giovanniello AG (2016). Chronic Pain and HIV: Wiley Online Library. [Google Scholar]
- Mücke M, Phillips T, Radbruch L, Petzke F, & Häuser W. (2018). Cannabis-based medicines for chronic neuropathic pain in adults. Cochrane Database of Systematic Reviews(3) [DOI] [PMC free article] [PubMed] [Google Scholar]
- National Conference of State Legislatures. (2020). State Medical Marijuana Laws. Retrieved from https://www.ncsl.org/research/health/state-medical-marijuana-laws.aspx
- Nielsen S, Sabioni P, Trigo JM, Ware MA, Betz-Stablein BD, Murnion B, . . . Smith A. (2017). Opioid-sparing effect of cannabinoids: a systematic review and meta-analysis. Neuropsychopharmacology, 42(9), 1752–1765 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Nugent SM, Morasco BJ, O’Neil ME, Freeman M, Low A, Kondo K, . . . Paynter R. (2017). The effects of cannabis among adults with chronic pain and an overview of general harms: a systematic review. Annals of internal medicine, 167(5), 319–331 [DOI] [PubMed] [Google Scholar]
- Savage SR, Romero-Sandoval A, Schatman M, Wallace M, Fanciullo G, McCarberg B, & Ware M. (2016). Cannabis in pain treatment: clinical and research considerations. The Journal of Pain, 17(6), 654–668 [DOI] [PubMed] [Google Scholar]
- Sohler NL, Starrels JL, Khalid L, Bachhuber MA, Arnsten JH, Nahvi S, . . . Cunningham CO (2018). Cannabis Use is Associated with Lower Odds of Prescription Opioid Analgesic Use Among HIV-Infected Individuals with Chronic Pain. Subst Use Misuse, 53(10), 1602–1607. doi: 10.1080/10826084.2017.1416408 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Solomon DH, Rassen JA, Glynn RJ, Garneau K, Levin R, Lee J, & Schneeweiss S. (2010). The comparative safety of opioids for nonmalignant pain in older adults. Archives of internal medicine, 170(22), 1979–1986 [DOI] [PubMed] [Google Scholar]
- Tate JP, Sterne JAC, & Justice AC (2019). Albumin, white blood cell count, and body mass index improve discrimination of mortality in HIV-positive individuals. AIDS, 33(5), 903–912. doi: 10.1097/qad.0000000000002140 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Uebelacker LA, Weisberg RB, Herman DS, Bailey GL, Pinkston-Camp MM, & Stein MD (2015). Chronic pain in HIV-infected patients: relationship to depression, substance use, and mental health and pain treatment. Pain medicine, 16(10), 1870–1881 [DOI] [PMC free article] [PubMed] [Google Scholar]
- VA/DoD. (2017). Clinical Practice Guideline for Opioid Therapy for Chronic Pain(version 3.0) [Google Scholar]
- Weisberg DF, Gordon KS, Barry DT, Becker WC, Crystal S, Edelman EJ, . . . Kerns RD (2015). Long-term prescription opioids and/or benzodiazepines and mortality among HIV-infected and uninfected patients. Journal of acquired immune deficiency syndromes (1999), 69(2), 223. [DOI] [PMC free article] [PubMed] [Google Scholar]