HIV-PASS (Pain and Sadness Support): Randomized controlled trial of a behavioral health intervention for interference due to pain in people living with HIV, chronic pain, and depression

Abstract

Objective:

To determine whether HIV-Pain and Sadness Support (HIV-PASS), a collaborative behavioral health intervention based on behavioral activation, is associated with decreased pain-related interference with daily activities, depression, and other outcomes in people living with HIV.

Methods:

We conducted a three-site clinical trial (n=187) in which we randomly assigned participants to receive either HIV-PASS or health education (HE) control condition. In both conditions, participants received seven intervention sessions, comprised of an initial in-person joint meeting with the participant, their HIV primary care provider and a behavioral health specialist (BHS), and six, primarily telephone-based, meetings with the BHS and participant. The intervention period lasted three months and follow-up assessments were conducted for an additional nine months.

Results:

Compared to HE, HIV-PASS was associated with significantly lower pain-related interference with daily activities at the end of month 3 (our primary outcome; b= −1.31, 95% CI = −2.28, −0.34). We did not observe other differences between groups at three months in secondary outcomes that included worst or average pain in the past week, depression symptoms, anxiety, and perceived overall mental and physical health. There were no differences between groups on any outcomes at 12 months post-enrollment.

Conclusions:

A targeted intervention can have positive effects on pain interference. At the end of intervention, effects we found were in a clinically significant range. However, effects diminished once the intervention period ended.

Trial Registration:

ClinicalTrials.gov NCT02766751

Introduction

Chronic pain is more common in people living with HIV (PLWH) than in the general population (1). Estimates of prevalence range from 25% to 85% of PLWH (1), with prevalence increasing with age (2). PLWH experience numerous types of chronic pain, including HIV-associated neuropathy (1, 2) and musculoskeletal pain both related and unrelated to HIV (1). Chronic pain interferes with social relationships, ability to work, sleep, and medication adherence (1, 2).

Depression frequently co-occurs with chronic pain (3, 4). PLWH with chronic pain are more than twice as likely to experience depression than PLWH without chronic pain (3). Depression severity prospectively predicts incidence of pain over time in PLWH (3) and increased impairment related to pain (5). The effects of depression and chronic pain are additive: data from a general (non-HIV-specific) sample show that the presence of both chronic pain and depression is associated with increased rates of disability and poorer health outcomes relative to either condition alone (6). Similarly, in PLWH, quality of life is poorest in people with both pain and depression compared to people with either condition alone (7). These data highlight the importance of developing and testing interventions that also address depression symptoms in the context of chronic pain in PLWH.

To address chronic pain, PLWH use an assortment of treatments including but not limited to pain medications, physical therapy and massage, applications of heat and cold, and psychoactive substances (8). Pharmacological interventions offer some relief, but many PLWH continue to experience substantive pain even when taking opioid medications (4, 9). There are few RCTs of non-pharmacologic treatment modalities (e.g., exercise, acupuncture, cognitive behavioral therapies) in the context of HIV, and many of the existing studies have significant methodologic concerns (e.g., small sample size, missing data) that limit conclusions about efficacy (10, 11). Thus, there is a clear need to implement methodologically sound research to study treatment strategies for coping with chronic pain in PLWH.

The HIV Medicine Association of the Infectious Disease Society of America clinical practice guidelines for managing chronic pain (12) recommend cognitive-behavioral therapy (CBT) interventions because they have demonstrated effectiveness in general populations (13). In a meta-analysis (13), the standardized mean difference (SMD) for CBT vs. treatment-as-usual was 0.32 for disability and 0.34 for distress at the end of treatment. Behavioral activation (BA) is a core component of CBT and can serve as a stand-alone intervention (14). BA for chronic pain is grounded in the fear-avoidance model of chronic pain (15) in which fear of pain, injury, or re-injury develops out of a catastrophized interpretation of pain. Fear leads to avoidance of activities associated with pain, which in turn leads to continued pain, inactivity, disability, and social dysfunction. Avoidance limits opportunities to learn that pain may not lead to feared outcomes (16). Withdrawal of positive reinforcers (through physical, psychosocial, work dysfunction) may increase depression symptoms (15). Behavioral activation targets this cycle by helping people to identify valued life goals and re-engage in corresponding meaningful and or enjoyable life activities.

There have been a few previous trials of psychotherapeutic (CBT) interventions for pain in HIV (1, 10). One trial of a 12-week face-to-face group intervention showed pre-post treatment improvements in pain intensity, and engagement in activities, but there was no control arm and attendance at the group intervention was poor, with a mean of 4.4/ 12 sessions attended (17). Another study of 6 weeks of CBT vs. supportive psychotherapy showed no differences between groups in change in average pain severity or interference due to pain, and 16/28 participants in the CBT group dropped out prior to the end of treatment and were not included in the data analysis (18). More recent pilot work including our own (19, 20) has demonstrated better attendance rates and acceptability of behavioral interventions for chronic pain in PLWH, possibly due to the integrated, medical-psychological collaborative approach and a flexible, telehealth delivery method of the interventions. We considered these two elements to be crucial (20, 21). Collaborative, multidisciplinary care has been effective for veterans with chronic pain (22) and telehealth eliminates transportation as a barrier to care.

In the present study we randomized PLWH to a three-month behavioral intervention, HIV-PASS (Pain and Sadness Support) or to a health education (HE) intervention that controlled for time and attention. The primary outcome was pain-related interference with daily activities. Secondary outcomes included pain severity, depression symptoms, anxiety, overall mental health, and perception of overall physical health. HIV-PASS is a CBT intervention, where the key CBT component, behavioral activation, is combined with psychoeducation about chronic pain to counter catastrophic beliefs about pain. The rationale for the components of the intervention and methods of implementation are described further elsewhere (23). We hypothesized that participants randomized to HIV- PASS would have improved primary and secondary outcomes at the end of active intervention (3 months) and that these effects would be sustained through end-of-assessment (12 months).

Methods

Participants

Individuals were eligible if they: 1) were HIV+; 2) reported chronic pain—i.e., pain duration for at least six months with a mean score of 3.5 or higher on the Brief Pain Inventory-Interference Scale (BPI-I) (24); 3) had Pain severity ≥ 4 on a 0 to 10 Numeric Rating Scale indicating “worst pain in the last week”; 4) had elevated depression symptoms --i.e., a Quick Inventory of Depression Symptoms (QIDS) score of ≥ 9 (25); 5) if taking an antidepressant, were on a stable dose for at least a month; 6) were aged ≥ 18 years and 7) had access to a telephone. Individuals were ineligible if they: 1) were engaged in psychotherapy more than one time per month or in a multidisciplinary pain management program; 2) had planned surgery in the next six months; 3) had pain due to cancer; 4) had current mania or psychotic symptoms in the past year, as determined by the SCID-V (26); 5) engaged in high levels of substance use (≥ 15 days of cocaine or heroin/non-prescribed opioids or ≥ 4 days of binge drinking in the previous month); 6) endorsed suicidality that required immediate clinical attention; 7) were pregnant.

Enrollment and Randomization

We recruited via active outreach to potential participants and use of advertisements in clinics. Interested participants completed a brief phone screen. If they appeared eligible, study staff invited them to an in-person appointment to complete informed consent and the baseline assessment. This research was approved by IRBs at Butler Hospital and Miriam Hospital in Providence, RI; and at Boston Medical Center in Boston, MA. Enrollment occurred between 2017–2020 in HIV primary care clinics based in academic hospitals in Providence, Rhode Island; Boston, Massachusetts; and Minneapolis, Minnesota. The study was registered at clinicaltrials.gov (NCT02766751).

Randomization to study arms occurred with a 1:1 ratio, using stratified permuted blocks of 6, 8, and 10. Stratification variables included current opioid prescription (yes/no), current antidepressant prescription (yes/no), and site. Only the study statistician and data manager had access to the group allocation schedule. When randomizing participants, staff verified the participant’s stratum and then pressed a button within REDCap to randomize the participant and learn their allocation.

Interventions

The two study interventions (HIV-PASS and HE) had several features in common. Both were manualized and included seven sessions provided over a three-month period. Participants also received workbooks. Session 1 was an in-person session at the HIV clinic with the participant, their HIV care provider, and the study behavioral health specialist (BHS). Session 2 was conducted in-person without the HIV care provider. Subsequently, Sessions 3–7 occurred approximately biweekly, although some flexibility was allowed. These five sessions were conducted via telephone and lasted 30–45 minutes each. No sessions were allowed after the three-month intervention period. The BHSs (n= 22) included master’s level counselors, clinical or counseling psychology graduate students or postdoctoral fellows, or clinical psychologists. All BHSs were trained in both interventions and attended biweekly supervision with licensed psychologists who developed the interventions. Sessions 2 through 7 were audiorecorded. Senior clinical staff conducted fidelity assessments for both interventions on a) two to three sessions for each of the first two to three patients seen by a new BHS; and b) a random subsample of 10% of completed intervention sessions thereafter (not including the Session 1, which was not audiorecorded). There were two types of fidelity ratings: a) a checklist of expected session components (specific to each session); and b) a rating scale for general BHS style. We present fidelity scores as percent of the total possible score.

HIV Pain and Sadness Support (HIV-PASS).

HIV-PASS is a behavioral health intervention based primarily on behavioral activation. Main components of HIV-PASS include: 1) psychoeducation about pain and depression; 2) supportive coaching in being an informed, activated patient; 3) education about time-based pacing; and 4) behavioral activation to increase engagement in pleasant and meaningful activities.

Psychoeducation centered on differences between acute and chronic pain; discussion about how pain is not always a reliable indicator of harm, especially in the context of chronic pain; the role of avoidance in chronic pain and depression; and approaching meaningful life goals, despite pain, as an alternative to avoidance. In one session each, the BHS introduced the concept of “being an informed and activated patient” and time-based pacing; if relevant, these concepts informed participants’ individualized goals. The bulk of the treatment time was focused on behavioral activation, which included the identification of personal meaningful life goals and setting corresponding short-term goals to engage in meaningful and pleasant activities. Specific value domains in which participants set goals included social relationships, education, work, community involvement, physical health, and physical activity. Interface with the participant’s HIV care provider at the first visit allowed the provider to give the participant and BHS feedback on any medical or substance use disorder concerns and on the importance (and safety) of physical activity in the context of chronic pain. All clinical HIV-PASS session notes were sent to the HIV care provider for review. After each session, the BHS sent a letter home to participants summarizing session content. For more detail regarding HIV-PASS, see (21).

Health education.

Health education was intended to control for time and attention from the BHS, without providing content that would directly and immediately have an impact on pain or depression. In the initial session, the BHS shared a menu of ten health education topics (e.g., nutrition, sleep, preventing cancer, diabetes) and prompted the HIV care provider and participant to discuss which topics would be most beneficial for follow-up discussions with the BHS. Subsequent sessions included just the participant and the BHS. They were interactive in nature and include education on chosen topics.

Outcome Assessments

Assessment schedule and masking.

Assessments occurred at baseline, and at months 1, 2, 3, 4, 6, 9, and 12. By necessity, participants were aware of their intervention group assignment. The primary outcome was interference due to pain at three months. Planned secondary outcomes described here included interference due to pain at 12 months, and pain intensity, depression, anxiety, antiretroviral adherence, viral load, perception of global improvement, and health-related quality of life at three and 12 months.

Pain.

We measured pain interference using the Brief Pain Inventory – Interference Scale (BPI-I), which assesses the extent to which pain interferes with daily activities, including relations with others, enjoyment of life, mood, walking, general activity, and work (24). Scores range from 0–10, with 10 reflecting the most interference. We also asked about average and worst pain severity over the past week on a 0–10 numerical rating scale (NRS).

Depression and anxiety.

Depression symptom severity over the past week was measured using the Quick Inventory of Depression Symptoms – Clinician Rating scale (QIDS) (25). The QIDS yields a total score between 0–27, with 0–5 indicative of no depression, 6–10 indicative of mild depression, 11–15 indicative of moderate depression, and scores ≥ 16 indicating severe or very severe depression. Trained interviewers were masked to treatment assignment. To measure anxiety, we used the Patient Reported Outcomes Measurement Information System (PROMIS) for anxiety, short form (27). The total score was translated into a T-score (mean = 50; standard deviation = 10) using the PROMIS conversion tables.

Antiretroviral adherence and viral load.

To measure adherence, we used the AIDS Clinical Trial Group self-report questionnaire (28). Self-report of adherence has been shown to be associated with viral suppression across studies (29). To measure viral load, we abstracted information from the medical chart on lab values collected prior to and during study participation.

Health-related Quality of Life.

We used the 36-item Medical Outcomes Study short-form health survey (SF-36) to measure health-related quality of life (30). We present results from the aggregate mental health summary score and physical health summary score (30). Scores ranged from −4.48 to 2.15 for the mental health summary score and −4.43 to 1.77 for the physical health summary score.

Perception of Global Improvement.

At month 3, we used the Patient Global Impression of Change Scales (PGIC)(31) to ask participants about percent improvement in symptoms (range = 0–100), and in overall life satisfaction (range = 0–100).

Compensation for assessments.

Participants were compensated $30 at the first study interview and the 4-month interview, $50 at the 12-month interview, and $20 at the 1, 2, 3, 6 and 9-month interviews. Participants did not receive compensation for counseling sessions.

Sample Size

Based on previous work (32–34), we determined that a minimally significant difference between groups was a standardized mean difference of 0.40. Accounting for attrition, we planned to recruit 118 participants per study arm (n=236). Using an ANCOVA repeated measures design with measurements at baseline, 1-, 2-, and 3-months, the proposed design had 1-β > 0.8 to detect a standardized difference in means as small as 0.27. However, recruitment was halted at 187 due to COVID-19. With this sample size, we estimate we have 80% power to detect a standardized effect size of 0.30.

Statistical Methods

For outcome analyses we used multiple imputation by chained equations (35) to generate 50 fully-populated data sets. Variables included in the imputation model were data collection site, variables used in stratified random assignment (use of opioids and depression), gender, age, treatment arm, and all observed instances of the outcome being evaluated. The fully-populated data sets were generated and analyzed using the mi impute and mi estimate facilities as implemented in Stata 15.1.

We used mixed-effects linear models to estimate the adjusted effect of intervention on outcomes observed at multiple time points. Covariates included the baseline instance of the outcome being evaluated, site of data collection, use of opioids, antidepressant use, gender, age, indicator variables for time, and the treatment by time interaction terms. In these models, the coefficient for the main effect of intervention gives the mean difference at the assessment specified as the reference category for time. To obtain the average marginal effects at each time period, we re-estimated the model with each assessment (1-, 2-, … 12-months) designated as the reference. We report the average adjusted marginal effect of treatment, 95% confidence intervals, and associated test statistics for each assessment. (See Supplemental Digital Content, Table S1, for the primary outcomes with all model coefficients.) All tests of significance and confidence interval estimates were based on the robust Huber-White variance estimator.

Items from the Patient Global Impression of Change were assessed only at end of active intervention. We used OLS regression to estimate the effects of intervention on these 3-month outcomes. Covariates again included site of data collection, use of opioids, depression, age, and gender.

Results

CONSORT Diagram and Baseline Characteristics

Figure 1 shows participant flow through the study. In total, we enrolled 187 participants. See Table 1 for a description of demographics and baseline instances of the outcome variables. Per CONSORT guidelines, we did not compare groups on baseline values (36).

Figure 1.

CONSORT 2010 Flow Diagram for HIV-PASS Study

Table 1.

Baseline Characteristics for Total Sample and by Intervention Arm.

	Total (n = 187)	HE (n = 95)	HIV-Pass (n = 92)
	Mean (± SD) or n (%) Yes	Mean (± SD) or n (%) Yes	Mean (± SD) or n (%) Yes
Demographics and Baseline Clinical Characteristics
Site
One	116 (62.0%)	60 (63.2%)	56 (60.9%)
Two	56 (30.0%)	28 (29.5%)	28 (30.4%)
Three	15 (8.0%)	7 (7.4%)	8 (8.7%)
Years of Age	51.4 (± 10.2)	51.2 (± 11.4)	51.7 (± 8.82)
Gender
Female	80 (42.8%)	37 (39.0%)	43 (46.8%)
Male	105 (56.2%)	56 (59.0%)	49 (53.3%)
Non-binary	2 (1.1%)	2 (2.1%)	0 (0.0%)
Race
American Indian/ Alaska Native	9 (4.8%)	3 (3.2%)	6 (6.6%)
Black or African American	61 (32.8%)	34 (35.8%)	27 (29.7%)
Native Hawaiian/ Pacific Islander	1 (0.5%)	0 (0.0%)	1 (1.1%)
White	72 (38.7%)	37 (39.0%)	35 (38.5%)
Other	15 (8.1%)	5 (4.3%)	10 (11.0%)
Multiracial	28 (15.1%%)	16 (16.8%)	12 (13.2%)
Ethnicity (Hispanic or Latinx)	25 (13.4%)	12 (12.6%)	13 (14.1%)
Education
< 8th Grade	6 (3.2%)	2 (2.1%)	4 (4.4%)
< High School	37 (19.9%)	18 (19.0%)	19 (20.9%)
High School	55 (29.6%)	28 (29.5%)	27 (92.7%)
Some College	53 (28.5%)	29 (30.5%)	24 (26.4%)
College Degree	32 (17.2%)	16 (16.8%)	16 (17.6%)
Master’s Degree or higher level	3 (1.6%)	2 (2.1%)	1 (1.1%)
Employment Status
Employed Full-Time	19 (10.2%)	10 (10.5%)	9 (9.8%)
Employed Part-Time	24 (12.8%)	9 (9.5%)	15 (16.3%)
Student	1 (0.5%)	1 (1.1%)	0 (0.0%)
Student and Employed	3 (1.6%)	3 (3.2%)	0 (0.0%)
Unemployed	39 (20.9%)	20 (21.1%)	19 (20.7%)
On Disability	94 (50.3%)	48 (50.5%)	46 (50.0%)
Homemaker / Full-Time Parent	1 (0.5%)	1 (1.1%)	0 (0.0%)
Retired	6 (3.2%)	3 (3.2%)	3 (3.3%)
Lives w Partner (Yes)	34 (18.2%)	12 (12.6%)	22 (23.9%)
Baseline Values of Outcome Variables
BPI-I	6.30 (± 1.56)	6.44 (± 2.15)	6.16 (± 1.98)
Worst pain severity, past week	8.21 (± 2.07)	8.27 (± 1.52)	8.14 (± 1.61)
Average pain severity, past week	6.43 (± 1.87)	6.37 (± 1.88)	6.48 (± 1.87)
QIDS	14.2 (± 2.99)	14.4 (± 2.93)	14.0 (± 3.04)
PROMIS Anxiety	24.5 (± 6.39)	24.5 (± 6.44)	24.5 (± 6.63)
ART Adherence (%)	93.3 (± 15.1)	91.5 (± 18.4)	95.1 (± 0.5)
Detectable Viral Load	33 (17.7%)	20 (21.1%)	13 (14.1%)
SF-36 Physical Health Component	−1.55 (± 1.01)	−1.67 (± 0.94)	−1.42 (± 1.07)
SF-36 Mental Health Component	−1.80 (± 1.15)	−1.83 (± 1.09)	−1.77 (± 1.21)

Note. BPI-I = Brief Pain Inventory – Interference Scale. QIDS = Quick Inventory of Depression Symptoms – Clinician Rating Scale. PROMIS = Patient Reported Outcomes Measurement Information System. ART= Antiretroviral. SF-36 = Medical Outcomes Study Short-form Health Survey.

Retention for Assessments

Retention rates for assessments were good: they ranged from 90.4% (at the 1-month assessment) to 84.5% (at the 4-month assessment). At Month 12, we conducted assessments with 87.2% of the total sample.

We used multilevel logistic regression to evaluate baseline characteristics as predictors of participant attrition during follow-up. The likelihood of having missed assessments was associated positively with age (OR = 1.07, z = 2.48, p = .013). Age was included in the multiple imputation and was also a planned covariate in the models evaluating intervention effects on outcomes. Missed assessments were not significantly associated with any of the other predictors evaluated.

Intervention Fidelity and Engagement

We conducted fidelity ratings on a total of 69 HIV-PASS sessions and 55 HE sessions. For HIV-PASS, the average intervention component checklist score was 94% (SD=7%), and the general style score was 94% (SD = 7%). For HE, the average intervention component checklist score was 97% (SD=8%), and the general style score was 93% (SD = 10%). Out of a possible 7 sessions, participants attended an average of 5.9 (SD= 1.8) HIVPASS sessions and 5.7 (SD =1.9) HE sessions.

Primary Outcome: Pain Interference with daily activities

We present the average marginal effects (b-values) of intervention on BPI-I scores at each assessment in Table 2. Participants randomized to HIV-PASS had significantly lower adjusted mean BPI scores (i.e., less pain-related interference) than those randomized to HE at Month 2 and Month 3. That is, at Month 3, the estimated adjusted mean BPI-I score was 4.19 for HIV-PASS and 5.50 for HE, representing a difference of 1.31 points on this 0–10 scale. Although calculation of effect sizes is challenging given the complexities of our data analytic strategy, this might reasonably be characterized as a small-medium sized effect (33). Between group differences at post-intervention assessments (i.e., Month 4 and beyond) were substantively small and not significantly different. See Figure 2 for a depiction of estimated mean scores over time.

Table 2.

Adjusted Mean Effect of Intervention on Outcomes at each Assessment Point.

	ESTIMATED AVERAGE MARGINAL EFFECTS
Assessment	b	95% CI	z	p-value
BPI-I (Primary Outcome)
Month 1	−0.48	(−1.29; 0.34)	−1.14	.25
Month 2	−1.02	(−1.93; −0.11)	−2.20	.028
Month 3 (End of Intervention)	−1.31	(−2.28 −0.34)	−2.66	.008
Month 4	−0.16	(−1.16; 0.84)	−0.32	.75
Month 6	−0.27	(−1.25; 0.71)	−0.54	.59
Month 9	0.16	(−0.80; 1.12)	0.33	.74
Month 12	−0.57	(−1.53; 0.40)	−1.15	.25
Average Pain Severity – Past Week
Month 1	−0.04	(−0.67; 0.60)	−0.11	.91
Month 2	−0.81	(−1.53; −0.09)	−2.20	.028
Month 3 (End of Intervention)	−0.59	(−1.36; 0.17)	−1.53	.13
Month 4	0.31	(−0.51; 1.12)	0.74	.46
Month 6	−0.01	(−0.79; 0.76)	−0.04	.97
Month 9	−0.22	(−1.00; 0.56)	−0.56	.58
Month 12	0.41	(−0.39; 1.21)	1.01	.31
Worst Pain Severity—Past Week
Month 1	−0.25	(−0.96; 0.47)	−0.68	.50
Month 2	−0.50	(−1.34; 0.34)	−1.16	.25
Month 3 (End of Intervention)	−0.19	(−1.07; 0.70)	−0.42	.67
Month 4	0.24	(−0.72; 1.21)	0.49	.63
Month 6	−0.05	(−0.98; 0.89)	−0.10	.92
Month 9	−0.24	(−1.13; 0.67)	−0.52	.60
Month 12	0.28	(−0.64; 1.21)	0.60	.55
Depression Symptom Severity (QIDS)
Month 1	0.06	(−1.24; 1.38)	0.09	.93
Month 2	0.53	(−0.88; 1.95)	0.74	.46
Month 3 (End of Intervention)	0.15	(−1.32: 1.61)	0.20	.84
Month 4	−0.04	(−1.50; 1.42)	−0.05	.96
Month 6	0.89	(−0.55; 2.33)	1.21	.23
Month 9	−0.47	(−2.06; 1.10)	−0.59	.56
Month 12	−0.51	(−2.18; 1.15)	−0.61	.55
PROMIS Anxiety
Month 1	−0.29	(−2.11; 1.53)	−0.32	.75
Month 2	0.97	(−1.18; 3.13)	0.89	.38
Month 3 (End of Intervention)	0.15	(−2.00; 2.30)	0.14	.89
Month 4	1.37	(−0.78; 3.54)	1.25	.21
Month 6	1.77	(−0.42; 3.97)	1.58	.11
Month 9	1.74	(−0.46; 3.96)	1.55	.12
Month 12	0.29	(−2.01; 2.60)	0.25	.80
SF-36 Physical Health Component Summary
Month 1	−0.03	(−0.27; 0.20)	−0.27	.79
Month 2	0.05	(−0.24; 0.35)	0.35	.73
Month 3 (End of Intervention)	0.12	(−0.15; 0.40)	0.87	.38
Month 4	−0.01	(−0.31; 0.28)	−0.10	.92
Month 6	0.08	(−0.22; 0.39)	0.54	.59
Month 9	0.12	(−0.17; 0.42)	0.82	.42
Month 12	0.54	(−0.26; 0.37)	0.34	.74
SF-36 Mental Health Component Summary
Month 1	−0.08	(−0.45; 0.28)	−0.46	.65
Month 2	−0.16	(−0.56; 0.25)	−0.75	.45
Month 3 (End of Intervention)	−0.11	(−0.52; 0.31)	−0.51	.61
Month 4	−0.37	(−0.78; 0.04)	−1.81	.070
Month 6	−0.23	(−0.57; 0.31)	−0.57	.57
Month 9	−0.08	(−0.52; 0.35)	−0.38	.70
Month 12	0.15	(−0.29; 0.59)	0.66	.51

Note. The b-value reflects the estimated mean difference between HIV-PASS and HE arms. BPI-I = Brief Pain Inventory – Interference Scale. QIDS = Quick Inventory of Depression Symptoms – Clinician Rating Scale. PROMIS = Patient Reported Outcomes Measurement Information System. SF-36 = Medical Outcomes Study Short-form Health Survey.

Figure 2:

Estimated Adjusted Average Marginal Mean Brief Pain Inventory – Interference Scores (BPI-I) by Intervention Arm and Assessment.

Note. Baseline mean was observed and not estimated by the mixed-effects linear model.

Secondary Outcomes

Participant randomized to HIV-Pass also had significantly lower adjusted mean average pain ratings at Month 2 but not Month 3 (see Table 2). As documented in Table 2, there were no differences between intervention groups on any other secondary outcome variables at end-of-intervention or any other timepoint. See Figure 3 for a depiction of change in depression symptom severity scores over time. Depression symptoms decreased in both groups over the first 3 months of the intervention; there was no differential change between groups.

Figure 3:

Estimated Adjusted Average Marginal Mean Quick Inventory of Depressive Symptomatology (QIDS) Scores by Intervention Arm and Assessment.

Note. Baseline mean was observed and not estimated by the mixed-effects linear model.

We did not analyze differences in antiretroviral adherence or viral load over time. Adherence was very high at baseline, and viral load was typically undetectable, and therefore we did not expect any differences over time due to ceiling/floor effects.

Finally, we examined percent improvement in symptoms and overall life satisfaction at 3 months. We did not find group differences in perceived improvement in symptoms (b = 1.71; 95%CI −7.26; 10.68, t = 0.38, p = .71) or perceived improvement in overall life satisfaction (b = −0.89; 95%CI −9.53; 7.84, t = −0.19, p = .85).

Discussion

This study demonstrated that HIV-PASS had a significant impact on the designated primary outcome, pain-related interference with daily activities, at the end of the intervention period (3 months). The mean difference between groups was 1.3 points on the BPI-I at the end of intervention. The Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT) (33) determined that an improvement of 1 point on this scale is a minimally clinically meaningful improvement, and that 1–2 points differentiate people who are satisfied or report improvement from treatment from those who are not satisfied or do not report improvement (although we did not see this in this study). The effect we observed is likely as good as the effect observed in pharmacologic treatments for pain. For example, in a review of treatments for chronic low back pain, Qaseem et al. (37) reported that NSAIDs, strong opioids, and duloxetine may be associated with a small improvement in function (i.e., a standardized mean difference between medication and placebo of 0.2 – 0.5, or a small-to-medium sized effect), whereas long-acting opioids, tricyclic antidepressants, and SSRIs were not associated with any improvement in function. Further, we note that we observed BPI-I differences between groups using a robust comparison condition – one that controlled for time and attention and was well-attended by participants.

Unlike previous studies of behavioral interventions for chronic pain (17, 18, 38), we found that attendance at sessions was good, with participants attending, on average, almost 6 of the 7 sessions. We designed HIV-PASS with acceptability and feasibility in mind, and it included several features that might have contributed to high attendance rates: explicit endorsement from the HIV care provider (in session 1), use of an initial in-person meeting at the HIV clinic at the time of a regularly scheduled medical appointment (so participants would have an opportunity to meet the BHSs), and follow-up phone calls (participants would not have to travel to the clinic). BHSs used continued outreach to participants when they missed a session, which undoubtedly also increased the number of sessions completed.

We conducted assessments for nine months following the end of the intervention period, hypothesizing that the HIV-PASS group would continue to show improved pain interference relative to HE. However, inspection of Figure 2 suggests that level of pain interference in the HIV-PASS group was similar to that observed in the HE group post-intervention, with both groups somewhat improved compared to baseline. We do not know whether a longer intervention period (e.g., six months) would maintain or even increase improvements for the HIV-PASS group. Alternatively, it is possible that ongoing support is important for consolidating and maintaining gains in this population which experiences numerous stressors including social, financial, physical health, mental health, stigma, and substance use concerns. Similar arguments have been made that ongoing treatment is needed for substance use disorder, another chronic disease (39). Such ongoing support may not need to be intense; it is possible that continuing to have two phone calls per month (similar to months 2 and 3 of HIV-PASS) would be sufficient, with options for more intensive care if individuals experience exacerbations. In addition, efforts to systematically address the social determinants of health on a population level may also be needed to see meaningful and sustained improvements in the pain experience of PLWH.

Supporting this suggestion, previous research has shown that higher levels of interference due to pain are associated with a) financial strain amongst PLWH (5), and b) community-level social metrics such as income, education level, and housing type in a sample of people presenting for orthopedic treatment (40).

We also did not see the positive pain-related interference results generalize to secondary outcomes. In particular, behavioral activation is an efficacious intervention for depression (41), and we expected to see a positive impact of HIV-PASS on depression symptoms relative to HE. We ruled out the explanation that HIV-PASS (or HE) was not delivered as intended, as we documented high levels of intervention fidelity in both groups, as well as good attendance at sessions. As depicted in Figure 3, we observed that average depression scores decreased in both groups from baseline to Month 1, and then remained at these lower levels for the duration of the assessment period. Possible explanations include a regression to the mean that could occur for any participant, or that the BHS contact in the HE condition served as a type of social support and had an antidepressant effect. Further, as symptoms of depression remained present post-intervention, such symptoms may have meant that the differential reductions in pain interference seen at month 3 were fragile and could not be maintained without more support.

Limitations of this study include the fact the fact we could not conduct masked assessment of the primary outcome (BPI-I) because it was assessed via self-report. Because we wanted to maximize external validity, there was considerable heterogeneity in types of pain participants experienced. Strengths of this study include recruitment of an ethnically and racially diverse sample at three sites, an extended follow-up assessment period, strong intervention participation, and a positive outcome in light of a robust control intervention.

In sum, HIV-PASS appears to reduce interference with life activities due to pain. It is acceptable to participants, and behavioral health specialists integrated into HIV primary care practices can deliver it by telephone with fidelity. However, it appears to be effective only during the active intervention period, with effects diminishing thereafter. Future research might examine the impact of a longer intervention period (i.e., six months), and how HIVPASS might be used in combination with other treatment strategies, and/or programs that address social determinants of health, to increase the magnitude and durability of its effect. In addition, it will be useful to examine moderators of treatment effect, as well as the extent to which participants actively met the goals they set, and whether this type of BA engagement predicts outcome (42).

Acronyms used in text:

HIV-PASS

HIV-Pain and Sadness Support

HE

health education

BHS

behavioral health specialist

CI

Confidence interval