Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2020 Jun 1.
Published in final edited form as: Alcohol. 2018 Oct 9;77:31–39. doi: 10.1016/j.alcohol.2018.10.003

Daily Factors Driving Daily Substance Use and Chronic Pain among Older Adults with HIV: An Exploratory Study Using Ecological Momentary Assessment

Alexis Kuerbis 1, M Carrington Reid 2, Jordan E Lake 3, Suzette Glasner-Edwards 4, Jessica Jenkins 5, Diana Liao 6, Jury Candelario 7, Alison A Moore 6
PMCID: PMC6456439  NIHMSID: NIHMS1000906  PMID: 30308287

Abstract

Background:

Adults 50 and older make up approximately 50% of persons living with HIV. Multiple co-morbidities are common among this group, including chronic pain and substance abuse, yet little is known about the daily factors that either enhance or inhibit these experiences or behaviors. This study explored daily drivers of substance use, pain, and relief from pain among older adults living with HIV utilizing ecological momentary assessment (EMA).

Method:

Participants (N=55), ages 49–71, completed seven consecutive days of daily EMA online surveys prior to treatment initiation within a randomized controlled trial. Multilevel modeling tested predictors of pain, substance use, and relief from pain by examining within- and between-person relationships.

Results:

Results revealed an associational, reciprocal relationship between daily worst pain and daily drinking, where greater worst pain ratings predicted heavier drinking and heavier drinking predicted greater daily and overall pain. Greater happiness and poorer quality of sleep predicted greater daily worst pain. Exercising and overall confidence to cope with pain without medication were associated with lower levels of daily worst pain. Finally, spending less time with a loved one over time and reporting any coping behavior were associated with relief from pain.

Conclusion:

Investigation of daily factors that drive pain and substance use behaviors among this unique population help inform which daily factors are most risky to their health and well-being. Alcohol use emerged as the only substance associated with both driving pain and responding to pain. Findings suggest key points for prevention and intervention.

Keywords: older adults, ecological momentary assessment, HIV, chronic pain, substance use, alcohol use

1. Introduction

Due to an increasing incidence of new cases and advances in medicine, people living with HIV over the age of 50 constitute a rapidly expanding group (Sangarlangkarn & Appelabaum, 2016). Of all current cases of HIV in the US, over 50% are persons 50 and older. Older adults living with HIV often present with unique complications due to high rates of multi-morbidity (Rodriguez-Penney et al., 2013; Sangarlangkarn & Appelabaum, 2016), including chronic pain and substance use (Edelman, Tetrault, & Fiellin, 2014; Jiao et al., 2016; Lum & Bruce, 2016; Parsons, Starks, Millar, Boonrai, & Marcotte, 2014). Such multi-morbidity results in, at a minimum, poor health outcomes and high health care utilization (DeLorenze, Tsai, Horberg, & Quesenberry, 2014; Jiao et al., 2016; Parsons et al., 2014). Thus, to improve health outcomes and potentially reduce need for costly intensive services, research is needed to elucidate potential targets for both prevention and treatment by understanding the interaction of these multi-morbidities on a daily level.

1.1. Chronic Pain, Substance Use and Older Adults with HIV

Empirical research on substance abuse, pain, HIV and aging typically occurs in silos, while research that examines the intersection of these factors is limited. Within the literature on HIV, higher levels of illicit drug use have been consistently observed among individuals with (versus those without) HIV (Green et al., 2010). While in the general population, drug and alcohol use tend to decrease as individuals age, these trends are less common for aging individuals with HIV (Skalski, Sikkema, Heckman, & Meade, 2013). This is particularly problematic given that medication non-adherence, including medications critical for treating HIV, is associated with substance use both across the life span, as well as specifically among those in later life (Parsons et al., 2014). Interestingly, many of these studies do not measure pain or consider it in their analytic approach.

A relationship between pain and substance use among individuals living with HIV is well-established (Uebelacker et al., 2015), yet this relationship has yet to be fully explored among individuals later in life when presumably the intersection of these factors may have increasingly dire consequences for health and well-being. In addition, evidence suggests that both psychic pain (e.g., anxiety, grief, depressed mood) and physical pain are critical to understanding substance use, well-being, and best practices for prevention and treatment among individuals living with HIV (Merlin et al., 2014).

Chronic pain, substance use, and the psychosocial/physical factors that drive both conditions on a daily basis have been underexplored among older adults. Two primary theories used to explain substance use in this population are the stress-coping model (Brennan, Schutte, & Moos, 2005; Sinha, 2001) and the self-medication model of substance abuse (Khantzian, 1999), in which pain and/or stress drives individuals to use substances to cope and/or self-medicate. Interestingly, these hypotheses have yet to be directly tested among older people living with HIV. Understanding potentially unique aspects of the intersection of pain, substance use, and psychosocial/physical factors in this population could provide critical information for heat of the moment interventions to prevent unnecessary health deterioration, reduce harm, and increase health and well-being.

1.2. Use of Ecological Momentary Assessment with Middle-aged and Older Adults

One way to understand daily relationships between pain and substance use and factors associated with both over time is through ecological momentary assessment (EMA). EMA is defined as “repeated collection of real-time data on subjects’ behavior and experience in their natural environment” (Shiffman, Stone, & Hufford, 2008, p. 3). In EMA, constructs are assessed daily (or more frequently), providing data with reduced recall bias, and it includes methods such as daily diary assessments. EMA has been used widely in empirical studies on pain (May, Junghaenel, Ono, Stone, & Schneider, 2018), many of which include older adults. While studies using EMA to assess alcohol use among older adults exist (e.g., Kuerbis et al., 2018; Sacco et al., 2015), EMA is not widely used among this population in relation to alcohol and other substance use.

To our knowledge, only one published study used EMA to examine pain and substance use among older adults with and without HIV (Paolillo et al., 2017). Importantly, this study demonstrated convergent validity of substance use reports, as well as the feasibility of using EMA across different health and mental health conditions among older adults; however, the small number of participants reporting living with HIV (n = 22) in this study limit its generalizability to people living with HIV. This investigation demonstrated that both mood and pain predicted subsequent substance use among this group, supporting the stress-coping and self-medication models of addiction. Unfortunately, the reciprocal relationship between pain and substance use, nor the impact of daily factors (e.g., exercise, spending time with a loved one) that are known to impact experience of pain, were not explored. Thus, much remains unknown about the dynamic patterns of pain and substance use among older adults with HIV.

1.3. The Current Study

As part of a pilot investigation testing the preliminary effects of a combined cognitive-behavioral therapy, tai chi, and motivational text messaging intervention to simultaneously reduce chronic pain and substance use for older adults with HIV, this study used pre-treatment data from seven consecutive days of daily diary assessment (DDA) to explore the intersection of psychosocial factors, physical factors, pain, and substance use on a daily basis. We specifically sought to assess: 1) whether daily rating of pain predicted use of alcohol, marijuana, and other substances; 2) which daily factors were associated with rating of daily worst pain; and 3) which daily factors were associated with reported relief from pain.

2. Method

Data were collected during the week prior to treatment initiation in a pilot randomized controlled trial with older adults with HIV and chronic pain who screened positive for at risk substance use. All procedures were approved by the Institutional Review Board at the University of California at Los Angeles. Detailed procedures of the pilot study are reviewed elsewhere (Moore et al., under review).

2.1. Participants

Participants (N=55) were recruited from a community-based agency in Southern California serving individuals living with HIV. After screening, eligible participants were enrolled in the study.

2.1.1. Inclusion criteria.

Inclusion criteria were: 1) age ≥50 years (3 persons were inadvertently enrolled who were aged 49 at the time of study entry); 2) able to read and understand English; 3) HIV+; 4) self-reported pain of ≥3 months duration on most days (not due to an active malignancy); 5) at-risk substance use, e.g., consuming ≥5 drinks on more than one occasion in the past three months and/or use of any substances (e.g., cocaine, amphetamine/methamphetamine, marijuana, opiates/heroin) ≥1 weekly, without a prescription or more than prescribed, in the past three months (National Institute on Alcohol Abuse and Alcoholism, 2013; National Institute on Drug Abuse, 2012); 6) self-reported ability to participate in a low intensity exercise program; 7) owned a phone with internet access and text message capability; and 8) not currently enrolled in substance abuse treatment.

2.2. Procedures

After providing written consent, participants completed a series of global self-report assessments. Participants were informed of their condition assignment (cognitive behavioral therapy + Tai Chi + motivational text messaging (CBT+TC+TXT), Support Group (SG), or Assessment Only (AO)) at the end of this assessment visit and trained on DDA. No data from the treatment period are included in the present analysis.

2.2.1. Daily diary assessment (DDA) procedures.

Participants were asked to complete a once daily online survey for all seven days prior to treatment initiation. Each day, participants received a text message prompt to complete the survey. Participants who did not have a smartphone with reliable internet access were instructed to use a computer to complete the survey. Participants chose the specific timing of the prompts to align with their schedules for optimal response rates. A majority of participants (68.6%) elected to complete the survey in the evening, with 21.6% and 9.8% completing the surveys in the morning and afternoon, respectively. Surveys were completed in about 5 to 10 minutes. Participants were compensated $1 for each day they completed the survey and given a $3 bonus for completing all seven days.

2.3. Measures

2.3.1. Sociodemographics.

Age, gender, educational and occupational information, race and ethnicity were collected via a self-report demographic questionnaire.

2.3.2. Daily-level assessments.

Primary predictors and outcome variables were assessed via the daily online surveys.

2.3.2.1. Pain.

Pain was assessed using two items. Any pain in the last 24 hours was assessed by asking “In the past 24 hours, did you experience any pain?” The response set was dichotomous, with 1 “yes” and 0 “no. If the participant responded no, they skipped the questions about coping with pain and relief from pain in the past 24 hours. Participants who answered “yes” were asked to “rate the pain they experienced at its WORST in the last 24 hours” (worst pain). The response set on these items ranged from 0 “no pain at all” to 10 “Pain as bad as you can imagine.” These questions were taken from the Brief Pain Inventory (BPI) (Cleeland & Ryan, 1994).

2.3.2.2. Alcohol and substance use.

Four items assessed participants’ substance use. One item asked participants to report how many standard drinks they had consumed in the last 24 hours. Another asked whether the participant had used marijuana in any form in the last 24 hours, with a response set consisting of 0 “no”, 1 “Yes, I smoked or vaped marijuana”, 2 “Yes, I ate something with marijuana in it”, and 3 “Yes, I both smoked and ate marijuana”. The third item asked “In the last 24 hours, have you taken any pain medications that were NOT prescribed to you (such as Percocet, oxycodone, codeine, hydrocodone, etc.)?” Response items included 0 “No”, 1 “Yes, I took them for pain”, 2 “Yes, I took them for fun/recreation”, and 3 “Yes, I took them for both pain and fun/recreation”. The fourth item asked “In the last 24 hours, have you taken any mood altering substance, such as a benzodiazepine (e.g., Xanax, Ativan, Valium), heroin, methamphetamine, cocaine, or K2 (or Spice)?” The response set was identical to the above item on pain medication. Due to the few days in which an opioid pain medication or other drugs were reported, these two variables were combined to create a binary other drug outcome variable.

2.3.2.3. Coping with Pain.

Also adapted from the BPI, one item asked, “When you were feeling pain at its WORST in the past 24 hours, which of the following did you do in response?” Participants could choose from a set of 10 responses and provide their own. Participants could also select more than one. Responses included options such as “Nothing”, “Took an over the counter medication”, “Took a medication prescribed by your doctor,” “Took a medication NOT prescribed by your doctor”, and “Performed a deep relaxation exercise.”

2.3.2.4. Relief from pain.

Another item taken from the BPI asked “How much relief did doing the [coping activity/activities] provide?” The response set ranged from 0% “No relief” to 100% “Complete relief.”

2.3.2.5. Mood and emotional state.

Several items were used to assess overall mood and emotional state. Each item asked the extent to which participants felt a certain way over the last 24 hours, with the response set ranging on these items from 0 “Not at all” to 10 “Extremely.” The constructs measured were stress, boredom, anger, loneliness, sadness, happiness, agitation, and calmness. These items were used in previous studies and their predictive validity is established (Armeli, Feinn, Tennen, & Kranzler, 2006; Kuerbis et al., 2018; Morgenstern et al., 2016).

2.3.2.6. Sleep quality.

One item adapted from the Pittsburgh Sleep Quality Index (Buysse, Reynolds, Monk, Berman, & Kupfer, 1989) asked participants to “rate your sleep quality last night overall.” The response set for this item ranged from 0 “No sleep or extremely poor quality” to 10 “Best, most restful sleep.”

2.3.2.7. Exercise.

One dichotomous item asked participants “In the last 24 hours, outside of what you may have received in this study, did you do any form of exercise (e.g., go for walk, lift weights, tai chi)?” Response set was yes or no.

2.3.2.8. Time with loved one.

One dichotomous item asked participants “Did you spend time with someone you cared about in the 24 hours?”

2.3.2.9. Pain-related fear.

Two items adapted from the Tampa Scale for Kinesiophobia (Kori, Miller, & Todd, 1990) and the Pain Catastrophizing Scale (Sullivan, Bishop, & Pivik, 1995; Vlaeyen, de Jong, Geilen, Heuts, & van Breukelen, 2001) comprised a short scale to measure pain-related fear. Items included “When I feel pain, I sometimes think something dreadful will happen” and “When I’m in pain, I keep thinking about how badly I want the pain to stop.” The response set for each ranged from 0 “Totally disagree” to 10 “Totally agree.” These items together yielded Cronbach’s alpha=.72.

2.3.2.10. Confidence to cope with pain.

One item asked “How confident are you that you will be able to cope effectively with any pain you might have today?” The response set on these items ranged from 0 “not at all” to 10 “extremely.” This variable was adjusted prospectively for analyses so that it was aligned with the next day’s reports of substance use and pain.

2.3.2.11. Confidence to cope with pain without medication or a mood altering substance.

One item asked participants “How confident are you that you will be able to cope effectively with any pain you might have today without using a prescribed medication or a mood altering substance?” The response set on these items ranged from 0 “not at all” to 10 “extremely.” This variable was adjusted prospectively for analyses so that it was aligned with the next day’s reports of substance use and pain.

2.3.2.12. Weekday/weekend.

A covariate designating whether a particular assessment day was a weekday (0) or weekend (1) was calculated from the date that they survey was completed.

2.4. Analytic Plan

Descriptive statistics were generated to explore potential group differences and identify additional covariates. For all of the models, daily ratings of each of the predictors were averaged to create estimates of person-level (level 1, between-person or grand mean) averages and daily-level values were averaged within person (level 2, within-person mean) for each construct. Person-level averages were used as covariates in their respective models to isolate the within-person (i.e., daily) changes in the predictors and their impact on subsequent drinking from the between-person changes (Bolger & Laurenceau, 2013).

For each of the six outcome variables (number of daily drinks, marijuana use, other drug use, any pain in the last 24 hours, worst pain, and relief from pain), multilevel models (MLMs) with daily ratings (level 1) nested within persons (level 2) were estimated. MLMs account for the non-independence of observations due to nesting, are robust to missing data, and can include random terms to model individual variability (Gibbons, Hedeker, & DuToit, 2010; Raudenbush & Bryk, 2002; Singer & Willett, 2003). Models were initially constructed to allow for random intercepts and slopes to account for individual variability or random effects, which provided good model fit only for worst pain and relief from pain. For number of daily drinks, only intercepts varied randomly across participants. An unstructured variance-covariance matrix was specified for models focusing on these three outcomes, and all MLMs utilized restricted maximum likelihood estimation.

For all of the binary outcome variables (marijuana use, other drug use, and any pain in the last 24 hours), random intercepts and slopes did not vary significantly. Therefore models examining these outcomes were constructed using generalized estimating equations (GEE, Liang & Zeger, 1986; Zeger & Liang, 1986). GEE shares similar strengths with MLM, such as its ability to account for correlation of observations in repeated measures designs (Stokes et al., 2000). For these models, a binomial distribution with logit link function was specified with an exchangeable working correlation matrix, which provided good model fit for each of this set of outcome variables (Stokes et al., 2000).

First, age, gender, education, employment, condition, time (count of day of DDA), and a variable indicating whether a day was a weekday or not were tested independently (alone for each outcome) as covariates. All but time and the weekday variable were insignificant (p > .05). Both time and weekday were significant and retained as covariates for all the models (Bolger & Laurenceau, 2013). Given the difference in DDA compliance rates across condition, condition was also retained as a covariate.

Next, models tested whether daily pain predicted substance use (alcohol, marijuana use, and other drug use, independently). Then, models independently tested which daily factors (substance use; coping with pain; level of relief from using coping strategy; overall health; mood; sleep; exercise; time spent with a loved one; and pain-related fear) were associated with rating of daily worst pain. Predictors significant at the p < .05 level were put into a final model. Lastly, the same model building procedure was used to test which of the aforementioned factors were independently associated with reported relief from pain. In all models, an interaction term (time X focal predictor) was added to the model to test for changes in focal predictors’ impact on outcomes over time (e.g., boredom X time). Where non-significant, interaction terms were removed. All models were tested using SAS 9.4 (SAS Institute Inc., 2002–2012).

3. Results

3.1. Sample Description and Condition Differences

Table 1 shows basic demographics across condition. Mean age was 55, and a majority of the participants were cisgender male, Black, non-Latino, and reported being unemployed or disabled. A majority had received schooling beyond high school.

Table 1.

Baseline Characteristics of Study Sample by Condition

CBT+TC+ TXT (N=18) M (SD) or % Support Group (N=19) M (SD) or % Assessment Only (N=18) M (SD) or % Total (N=55)
Demographics
    Age (range 49–71) 53.3 (4.8) 55.2 (5.0) 56.8 (6.0) 55.1 (5.4)
    Gender
        Male 77.7 73.7 77.7 76.4
        Female 11.1 15.8 16.7 14.5
        Transgender 11.2 10.5 5.6 9.1
    Race/Ethnicity
        Hispanic/Latino 38.9 31.6 22.2 30.9
        Black, Non-Hispanic 50.0 57.9 50.0 52.7
        Other, Non-Hispanic 11.1 10.5 27.8 16.4
    Education
        High School/Equivalent or Less 38.9 47.4 27.8 38.2
        Beyond High School 61.1 52.6 72.2 61.8
    Employment
        Working Full or Part Time 11.1 5.6 11.1 9.3
        Unemployed or Disabled 88.9 94.4 88.9 90.7
Open in a new tab

Note: Table is a reproduction of a table in (Moore et al., under review). There were no significant differences across conditions. CBT+TC+TXT= Arm consisting of CBT, Tai Chi, and text messaging support messages.

Overall, participants reported experiencing at least some pain five or more days (M=5.5, SD=2.7) during the week. Of the days they completed the survey, they reported drinking an average of 41.3% of days, ranging from 0 days to 7 days, and drank less than two drinks (M=1.2, SD=1.9) per day. Participants reported using marijuana three days out of the week (M=2.9, SD=3.4). Other drugs were used about two days (M=1.8, SD=3.1) out of the week. Table 2 shows descriptive statistics for the seven days (pre-treatment) of DDA by group. Individuals in CBT+TC+TXT demonstrated greater DDA compliance and more alcohol use than the other two conditions. In addition, a greater proportion of participants in CBT+TC+TXT reported prescription medication use as a means of coping with pain. Finally, those participants in SG reported a significantly greater proportion of days using marijuana for pain only compared to the other two groups who reported using marijuana for both pain and recreation.

Table 2.

Descriptive Statistics of Daily Variables over Seven Days

Range CBT+TC+TXT M (SD) or % Support Group M (SD) Assessment Only M (SD) F or x2or Fisher’s Exact Statistic p-value
(N=126 possible responses) (N=133 possible responses) (N=126 possible responses)
Compliancea 0 -- 100 71.4 60.2 43.7 11.2 < .001
    Overall mean % days completed (out of 126 (133 for Support) responses possible) 71.4 60.2 43.7
    Mean days completed 5.1 (3.2) 4.2 (3.4) 3.1 (3.5)
(N=90 completed responses) (N=80 completed responses) (N=55 completed responses)
Pain
Experienced pain in last 24 hours -- 84.4 78.8 69.1 4.8 .09
Worst pain 0 – 10 7.2 (1.9) 6.8 (1.9) 7.2 (1.8) 1.0 .37
Substance Use
Alcohol (# of drinks per day) 0 – 12 1.7 (2.3) .78 (1.4) 1.1 (1.4) 5.4 < .01
Used marijuana in past 24 hours -- 38.7 37.7 48.1 1.6 .44
Reason used marijuana 30.0 < .001
    Fun/recreation -- 20.6 0.0 12.0
    Pain -- 23.5 75.9 36.0
    Both pain and recreation -- 55.9 13.8 52.0
    Neither -- 0.0 10.3 0.0
Used other drugs in past 24 hours -- 29.9 27.3 19.2 2.0 .38
Pain coping techniques
    Nothing -- 3.7 .16
    OTC Med -- 25.3 31.7 39.5 2.4 .30
    Prescribed Med -- 58.7 36.5 42.1 7.2 .03
    Non-prescribed Med -- 10.7 9.5 0.0 4.2 .12
    Relaxation exercise -- 9.3 12.7 5.3 1.5 .47
    Exercised -- 24.0 20.6 13.2 1.8 .40
    Marijuana, not prescribed -- 18.7 9.5 15.8 2.3 .32
    Alcohol -- 22.7 9.5 21.1 4.5 .11
    Rested -- 32.0 41.3 31.6 1.6 .46
    Other illicit drug -- 6.7 3.2 15.8 1.6 .45
Relief from pain 0 – 10 5.5 (2.6) 5.3 (2.4) 5.2 (2.3) .24 .79
Mood/emotional state
    Bored 0 – 10 3.1 (2.9) 3.2 (2.9) 2.8 (2.9) .24 .79
    Stressed 0 – 10 4.9 (3.3) 3.8 (2.8) 4.4 (3.0) 2.6 .08
    Anger 0 – 10 3.3 (3.0) 3.1 (3.1) 3.2 (3.1) .10 .95
    Lonely 0 – 10 3.3 (3.4) 3.3 (3.2) 3.0 (3.0) .20 .82
    Sad 0 – 10 2.8 (2.6) 3.2 (3.0) 2.0 (2.4) 2.7 .07
    Happy 0 – 10 4.9 (3.0) 4.5 (3.3) 4.9 (2.9) .43 .65
    Agitated 0 – 10 3.7 (3.0) 3.5 (3.1) 3.6 (2.8) .07 .94
    Calm 0 – 10 4.8 (2.8) 4.2 (3.6) 3.9 (2.5) 1.6 .20
Quality sleep 0 – 10 6.2 (5.9) 5.5 (2.6) 5.8 (1.9) 2.0 .14
Exercised in last 24 hours -- 40.9 53.2 49.1 2.6 .27
Spent time with loved one -- 55.8 66.2 69.2 3.1 .21
Fear of pain 0 – 20 11.3 (5.4) 11.1 (4.5) 11.4 (4.1) .07 .93
Confidence to cope 0 – 10 5.7 (2.8) 5.0 (2.9) 5.1 (2.0) 1.7 .18
Confidence to cope w/o Meds 0 – 10 4.2 (3.0) 4.3 (3.2) 3.8 (2.4) .50 .61
Open in a new tab
a

Significant differences were between the two treatment arms and the control. There were no significant differences between the two treatment arms.

Note: CBT+TC+TXT= Arm consisting of CBT, Tai Chi, and text messaging support messages.

3.2. Daily Pain Predicting Daily Substance Use

Any pain in the last 24 hours was not a significant predictor of same day substance use (alcohol, marijuana, or other drug use) (Table 3). Daily rating of worst pain in the last 24 hours was a significant predictor of only alcohol use (Table 3), such that for every one unit increase in worst pain beyond an individual’s personal mean, alcohol use increased by a quarter drink.

Table 3.

Results of Independent Models Testing Pain Predicting Substance Use

Alcohol Use (# of Drinks) Marijuana Use Other Drug Use
Independent Predictors B SE p B SE p B SE p
Pain in last 24 Hoursa .21 .27 .46 .34 .30 .25 .37 .30 .21
Pain 24*Time .05 .27 .68 .16 .17 .34 .02 .18 .91
Daily Worst Paina .23 .10 .02 .04 .04 .32 −.11 .16 .50
Person-average Worst Paina .09 .15 .56 .29 .19 .13 −.34 .17 .05
Daily Worst Pain*Time −.07 .06 .24 .01 .03 .59 .01 .05 .78
Open in a new tab
a

Parameter estimates are those without the interaction term, given that it was not significant.

3.3. Daily Predictors of Daily Worst Pain

Table 4 lists the independent predictors of daily worst pain. Independent models revealed greater number of daily drinks, less daily happiness, more daily stress, poorer quality sleep, no exercise, and lower confidence to cope with pain without medication as significant independent predictors of daily worst pain. These variables were then entered into a final model. Daily stress was rendered insignificant and removed from the final model. Results of the final reduced model (Table 4) revealed more daily and person level number of drinks, less daily happiness but more overall happiness, poorer quality sleep overall, no exercise, and lower confidence to cope without medication or other substances emerged as significant predictors of daily worst pain.

Table 4.

Parameter Estimates of Significant Independent and Final Multilevel Models of Focal Predictors of Daily Worst Pain

Independent Models Final Reduced Model
Predictors B SE p B SE p
Daily number of drinks .16 .07 .02 .48 .12 <.001
Person average drinks .23 .15 .12 .32 .12 <.01
Daily drinks × Time −.05 .04 .18
Marijuana Use .43 .36 .29
Marijuana Use × Time .04 .13 .77
Other Drug Use .38 .30 .23
Other Drug Use × Time .01 .14 .95
Daily sad .13 .07 .05
Person average sad .16 .09 .09
Daily sad × Time −.02 .04 .66
Daily happy −.12 .06 .04 −.21 .09 .03
Person average happy −.08 .08 .033 .33 .09 <.001
Daily happy × Time −.04 .03 .17
Daily loneliness .08 .05 .11
Person average loneliness .12 .09 .17
Daily loneliness × Time −.01 .03 .69
Daily stressed .11 .05 .03
Person average stressed .10 .09 .27
Daily stressed × Time −.01 .03 .66
Daily boredom .02 .05 .68
Person average boredom .15 .10 .13
Daily boredom × Time 0.0 .02 .84
Daily anger .07 .05 .19
Person average anger .06 .09 .51
Daily anger × Time .01 .03 .83
Daily agitated .04 .06 .47
Person average agitated .07 .10 .45
Daily agitated × Time −.03 .03 .37
Daily calm .00 .04 .96
Person average calm −.09 .11 .39
Daily calm × Time .02 .02 .24
Daily Quality Sleep −.08 .06 .19 −.03 .08 .77
Person Quality Sleep −.35 .11 <.01 −.29 .09 <.01
Daily Quality Sleep × Time −.05 .04 .22
Exercise −.66 .23 <.01 −.67 .25 .02
Exercise × Time −.05 .11 .61
Daily Fear of Pain .06 .03 .06
Person Average Fear of Pain .03 .06 .59
Daily Fear of Pain × Time .01 .02 .67
Daily Confidence to Cope .19 .10 .06
Person Average Confidence to Cope −.20 .10 .05
Daily Confidence to Cope × Time .14 .07 .05
Daily Confidence to Cope w/o Meds −.02 .06 .79 −.06 .06 .27
Person Average Confidence to Cope w/o Meds −.26 .10 <.01 −.34 .10 <.01
Daily Confidence to Cope w/o Meds × Time .05 .04 .13
Open in a new tab

Note: In cases where the interaction term is not statistically significant, the parameter estimates shown here are those without the interaction term.

3.4. Daily Predictors of Daily Relief from Pain

Table 5 shows the results of the independent and final models for relief from pain. Independent models revealed that person level greater quality sleep, spending less time with a loved one across time, decreasing fear of pain across time, and doing anything to cope with pain significantly predicted relief from pain. When all of these variables were entered into a final model together, only spending less time with a loved one across time and doing anything to cope with pain were associated with relief from pain (Table 5). Figure 1 demonstrates the relationship between relief of pain and time with a loved one across the seven days, revealing that relief from pain is more variable without time spent with a loved one, demonstrating more peaks than valleys, whereas for those with a loved one, relief is more consistent and steady across days.

Table 5.

Parameter Estimates of Significant Independent and Final Multilevel Models of Predictors of Relief from Pain

Independent Models Final Reduced Model
Predictors B SE p B SE p
Daily number of drinks .05 .10 .59
Person average drinks −.31 .21 .13
Daily drinks × Time −.03 .06 .64
Marijuana Use .42 .51 .46
Marijuana Use × Time .05 .18 .77
Other Drug Use .66 .41 .14
Other Drug Use × Time −.12 .18 .50
Daily Quality Sleep .06 .09 .47
Person Quality Sleep .40 .16 .02
Daily Quality Sleep × Time −.04 .05 .41
Exercise .23 .32 .48
Exercise × Time .05 .15 .76
Loved one −.07 .37 .86 .14 .35 .69
Loved one × Time −.34 .15 .03 −.39 .14 <.01
Daily Fear of Pain .00 .04 .96
Person Average Fear of Pain .07 .08 .35
Daily Fear of Pain × Time −.06 .02 <.01
Doing Nothing to Cope −2.06 .58 <.01 −2.3 .58 <.01
Doing Nothing to Cope × Time .42 .24 .08
OTC Medication .07 .39 .85
OTC Medication × Time .06 .17 .72
Prescribed Medication .76 .41 .08
Prescribed Medication × Time .06 .15 .69
Non-prescribed Medication .66 .57 .29
Non-prescribed Medication ×
Time		 −.28 .35 .43
Relax .19 .50 .71
Relax × Time −.16 .25 .54
Coping by exercising −.32 .42 .47
Coping by exercising × Time .10 .19 .60
Coping by using marijuana .42 .52 .44
Coping by using marijuana × Time .26 .23 .26
Coping by using alcohol .44 .39 .27
Coping by using alcohol × Time −.12 .20 .54
Rest .70 .36 .06
Rest × Time .16 .15 .29
Cope by using illicit drugs 1.69 .71 .10
Cope by using illicit drugs × Time −.36 .33 .27
Open in a new tab

Note: B = parameter estimate; SE = standard error. Daily variables were centered at the individual person-mean. All others were centered at the grand mean. Covariates time, condition, and weekday were entered into all models.

Figure 1.

Figure 1.

Open in a new tab

Interaction of time by time spent with loved one predicting relief from pain. The higher the number, the greater the relief from pain.

4. Discussion

This study aimed to explore and identify daily factors that predicted substance use, pain presence and level, and degree of pain relief among older adults living with HIV using EMA. Findings revealed that daily alcohol use was associated with daily reports of pain. While temporal order and thus causality cannot be determined, this associational relationship may point to a self-medication model of substance use. Alcohol may also exacerbate pre-existing pain. This correlational relationship is consistent with empirical literature that finds a strong positive association between chronic pain and heavy alcohol use in a general population of older adults (Brennan et al., 2005), and it provides evidence that this association exists on a daily level among older adults living with HIV. Interestingly, there was no relationship between use of other substances and reported pain or pain relief. A previous study that used EMA with substance using older adults with and without HIV found that pain was predictive of substance use earlier in the day (Paolillo et al., 2017); however this was not differentiated by type of substance (e.g., alcohol or drug use).

Worst pain appears to be driven by lower levels of happiness, poorer quality sleep, lack of exercise, and low confidence to cope with pain without medication or another mood altering substance. These findings are consistent with previous studies with other populations that show a relationship between increased pain and negative mood (Merlin et al., 2014), sleep disturbance (Aǧargün et al., 1999; Cappelleri et al., 2009), lack of exercise (Dzierzewski et al., 2014; Geneen et al., 2017), and low self-efficacy coping with pain (Jensen, Turner, & Romano, 1991). These findings provide important information about potential avenues for intervention—primarily improving quality of sleep, encouraging exercise, and increasing confidence to cope with pain without medication—all behaviors which can be relatively easily addressed with cognitive behavioral therapy (CBT). Given that CBT is already an empirically supported treatment for both chronic pain and substance use (Lunde, Inger, Nordhus, & Pallesen, 2009; Monti, Kadden, Rohsenow, Cooney, & Abrams, 2002; Rotgers, 2003), these findings provide further support for broadening the use of CBT to target a wider variety of behaviors with this group to reduce both pain and substance use—and to adapt these interventions to potentially intervene on a daily level, using mobile technology (Kuerbis, Mulliken, Muench, Moore, & Gardner, 2017).

Analyses did not reveal any significant predictors of daily relief from pain, with the exception of doing nothing to cope with pain and spending time with a loved one, both of which reduced rather than increased relief. While results suggest that any method of coping with pain is better than nothing, no one type of coping method emerged as a significant predictor of relief in this sample—suggesting they were a perfect target group for intervention. In contrast to extant empirical literature on pain (Hughes et al., 2014; Roberts, Klatzkin, & Mechlin, 2015), spending more time with a loved one reduced pain relief. This relationship may have emerged due to the variability of relief when participants were without a loved one. Relief experienced with a loved one is more stable—albeit slightly lower than without. There is no clear explanation for why this might be, particularly in light of the measurement of time with a loved one. There were no detailed questions about the quality, pleasantness, length of time spent, or the activities shared with the loved one. Further research is needed to elucidate this relationship.

Limitations of this study require that findings be interpreted with caution. Given the small sample, generalizability is limited to older adults living with chronic pain, substance use, and HIV. The lack of adjustment for gender in the inclusion criteria also limits generalization to women who binge drink at higher intensity. In addition, data were collected once a day for seven consecutive days—limiting data collection to retrospective recall of the past 24 hours. It may be that daily factors driving pain, substance use, and relief from pain may have differed if participants were assessed several times a day, as in Paolillo et al. (2017), or over a longer period. Even with these limitations and a small sample, power was adequate to discover important relationships between daily factors of interest.

Compliance with the EMA was a challenge for this study. While all phones used in the study had internet access, several participants did not own a smartphone—limiting the ability to collect data easily. Internet speed was quite slow on such devices, and compliance suffered as a result. In addition, publicly funded phones or phones provided by social service providers are often shared within families or social networks, limiting the access to the participant at times. Future EMA research with this population needs to include provision of smartphones for improved compliance rates. In addition to potential challenges with devices themselves, individuals in AO specifically had dramatically lower rates of EMA compliance. One potential explanation is that those participants knew their condition assignment during the seven day assessment period. As a result, knowing they would not receive treatment, they may have lost motivation to participate with EMA. Other studies using EMA have demonstrated that when blind to condition, compliance rates during periods prior to treatment initiation are comparable (e.g., Kuerbis, Armeli, Muench, & Morgenstern, 2013). Despite statistically significantly different compliance rates across condition, a sensitivity analysis was performed using only the two treatment groups. Results of all the findings reported here were identical. In addition, we used compliance as an additional covariate, again with identical results.

Among those days when participants did respond, it is possible that they learned that specific skip pattern responses made it possible to complete a shorter survey. Given the main skip pattern was related to the presence or absence of pain and that there was ample data on pain, this was not likely a factor in reducing the quality of the data. Additionally, any physical limitations that participants may have experienced (e.g., a tremor) that made a touch screen more difficult to provide precise answers (e.g., clicked the wrong button for a category) may have affected the data (Kuerbis et al., 2017). Given consistency of data across time, we are confident that this level of human error remained relatively low.

Despite these challenges to data collection, participants reported a surprising amount of positive feedback about the EMA during their debriefing interviews. Several participants reported that EMA helped them to “reflect on” their day, raised awareness about their own behavior and how they were feeling, and helped them to “stay focused”. While this is a positive sign for potential intervention using mobile technology with this group, it may also be indicative that assessment reactivity may have influenced the ability to determine the “true” daily factors driving pain, substance use, and relief from pain.

Future research on substance using older adults living with chronic pain and HIV should include a wide variety of data collection methods and much larger samples to obtain the most accurate and detailed data with the greatest ability for generalizability. Mobile technology should continue to be utilized across the life span and across a variety of health conditions and problematic behaviors, as it can provide unique information about when and how providers can intervene to improve health.

Highlights.

  • Ecological momentary assessment measured daily pain and substance use.

  • An associational relationship between daily alcohol use and pain was demonstrated.

  • Findings point to key targets for psychosocial interventions.

Acknowledgments

This study was supported by the National Institutes of Health grant R21DA038163. Authors were also supported by K24AG053462 (AAM), K23AI110532 (JEL), P30AG028748 (AAM and AK), P30AG021684 (AAM). The authors have no competing interests to declare.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Aǧargün MY, Tekeoǧlu I, Güneş A, Adak B, Kara H, & Ercan M (1999). Sleep quality and pain threshold in patients with fibromyalgia. Comprehensive Psychiatry, 40(3), 226–228. doi: 10.1016/S0010-440X(99)90008-1 [DOI] [PubMed] [Google Scholar]
  2. Armeli S, Feinn R, Tennen H, & Kranzler H (2006). The effects of naltrexone on alcohol consumption and affect reactivity to daily interpersonal events among heavy drinkers. Experimental and Clinical Psychopharmacology, 14(2), 199–208. [DOI] [PubMed] [Google Scholar]
  3. Bolger N, & Laurenceau JP (2013). Intensive longitudinal methods: An introduction to diary and experience sampling research. New York: The Guilford Press. [Google Scholar]
  4. Brennan PL, Schutte KK, & Moos RH (2005). Pain and use of alcohol to manage pain: Prevalence and 3-year outcomes among older problem and non-problem drinkers. Addiction, 100(6), 777–786. doi: 10.111/j..1360-0443.2005.0a074.x [DOI] [PubMed] [Google Scholar]
  5. Buysse DJ, Reynolds CF 3rd, Monk TH, Berman SR, & Kupfer DJ (1989). The Pittsburgh Sleep Quality Index: a new instrument for psychiatric practice and research. Psychiatry Res, 28(2), 193–213. [DOI] [PubMed] [Google Scholar]
  6. Cappelleri JC, Bushmakin AG, McDermott AM, Sadosky AB, Petrie CD, & Martin S (2009). Psychometric properties of a single-item scale to assess sleep quality among individuals with fibromyalgia. Health and Quality of Life Outcomes, 7, 54–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cleeland CS, & Ryan KM (1994). Pain assessment: Global use of the brief pain inventory. Annals Academy of Medicine Singapore, 23(2), 129–138. [PubMed] [Google Scholar]
  8. DeLorenze GN, Tsai AL, Horberg MA, & Quesenberry CP Jr. (2014). Cost of Care for HIV-Infected Patients with Co-Occurring Substance Use Disorder or Psychiatric Disease: Report from a Large, Integrated Health Plan. AIDS Res Treat, 2014, 570546. doi: 10.1155/2014/570546 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Dzierzewski JM, Buman MP, Giacobbi PR Jr., Roberts BL, Aiken-Morgan AT, Marsiske M, & McCrae CS (2014). Exercise and sleep in community-dwelling older adults: evidence for a reciprocal relationship. J Sleep Res, 23(1), 61–68. doi: 10.1111/jsr.12078 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Edelman EJ, Tetrault JM, & Fiellin DA (2014). Substance use in older HIV-infected patients. Curr Opin HIV AIDS, 9(4), 317–324. doi: 10.1097/coh.0000000000000069 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Geneen LJ, Moore RA, Clarke C, Martin D, Colvin LA, & Smith BH (2017). Physical activity and exercise for chronic pain in adults: an overview of Cochrane Reviews. The Cochrane Database of Systematic Reviews(4), CD011279. doi: 10.1002/14651858.CD011279.pub3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Gibbons RD, Hedeker D, & DuToit S (2010). Advances in analysis of longitudinal data. Annu Rev Clin Psychol, 6, 79–107. doi: 10.1146/annurev.clinpsy.032408.153550 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Green TC, Kershaw T, Lin H, Heimer R, Goulet JL, Kraemer KL, . . . Justice AC (2010). Patterns of drug use and abuse among aging adults with and without HIV: a latent class analysis of a US Veteran cohort. Drug Alcohol Depend, 110(3), 208–220. doi: 10.1016/j.drugalcdep.2010.02.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Hughes S, Jaremka LM, Alfano CM, Glaser R, Povoski SP, Lipari AM, … Kiecolt-Glaser JK (2014). Social support predicts inflammation, pain, and depressive symptoms: longitudinal relationships among breast cancer survivors. Psychoneuroendocrinology, 42, 38–44. doi: 10.1016/j.psyneuen.2013.12.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Jensen MP, Turner JA, & Romano JM (1991). Self-efficacy and outcome expectancies: Relationship to chronic pain coping strategies and adjustment. Pain, 44(3), 263–269. [DOI] [PubMed] [Google Scholar]
  16. Jiao JM, So E, Jebakumar J, George MC, Simpson DM, & Robinson-Papp J (2016). Chronic pain disorders in HIV primary care: clinical characteristics and association with healthcare utilization. Pain, 157(4), 931–937. doi: 10.1097/j.pain.0000000000000462 [DOI] [PubMed] [Google Scholar]
  17. Khantzian EJ (1999). Treating addiction as a human process. Northvale, NJ: Jason Aronson Inc. [Google Scholar]
  18. Kori S, Miller R, & Todd D (1990). Kinesophobia: A new view of chronic pain behaviour. Pain Management, Jan-Feb, 35–43. [Google Scholar]
  19. Kuerbis A, Armeli S, Muench F, & Morgenstern J (2013). Motivation and self-efficacy in the context of moderated drinking: Global self-report and ecological momentary assessment. Psychology of Addictive Behaviors, 27(4), 934–943. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kuerbis A, Mulliken A, Muench F, Moore AA, & Gardner D (2017). Older adults and mobile technology: Toward a better understanding of utilization and applications to behavioral health. Mental Health and Addiction Research, 2(2), 1–11. doi: 10.15761/MHAR.1000136 [DOI] [Google Scholar]
  21. Kuerbis A, Treloar Padovano H, Shao S, Houser J, Muench F, & Morgenstern J (2018). Comparing daily drivers of problem drinking among older and younger adults: An electronic daily diary study using smartphones. Drug & Alcohol Dependence, 183, 240–246. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Liang K, & Zeger SL (1986). Longitudinal analysis using generalized linear models. Biometrika, 73(1), 13–22. [Google Scholar]
  23. Lum PJ, & Bruce RD (2016). Comorbid substance use among persons with HIV and chronic pain Chronic Pain and HIV (pp. 78–96): John Wiley & Sons, Ltd. [Google Scholar]
  24. Lunde LH, Inger H, Nordhus IH, & Pallesen S (2009). The effectiveness of cognitive and behavioural treatment of chronic pain in the elderly: a quantitative review. J Clin Psychol Med Settings, 16, 254–262. [DOI] [PubMed] [Google Scholar]
  25. May M, Junghaenel DU, Ono M, Stone AA, & Schneider S (2018). Ecological Momentary Assessment Methodology in Chronic Pain Research: a Systematic Review. J Pain. doi: 10.1016/j.jpain.2018.01.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Merlin JS, Walcott M, Ritchie C, Herbey I, Kertesz SG, Chamot E, . . . Turan JM (2014). ‘Two pains together’: patient perspectives on psychological aspects of chronic pain while living with HIV. PLOS ONE, 9(11), e111765. doi: 10.1371/journal.pone.0111765 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Monti PM, Kadden R, Rohsenow DJ, Cooney NL, & Abrams D (2002). Treating alcohol dependence: A coping skills training guide. NY: The Guilford Press. [Google Scholar]
  28. Moore AA, Lake JE, Glasner-Edwards S, Karlamanga AS, Kuerbis A, Preciado D, . . . Reid MC (under review). Establishing the feasibility, acceptability and preliminary efficacy of a multi-component behavioral intervention to reduce pain and substance use and improve physical performance in older adults with HIV. [DOI] [PMC free article] [PubMed]
  29. Morgenstern J, Kuerbis A, Houser J, Muench FJ, Shao S, & Treloar H (2016). Withinperson associations between daily motivation and self-efficacy and drinking among problem drinkers in treatment. Psychology of Addictive Behaviors, 30(6), 630–638. doi: 10.1037/adb0000204 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. National Institute on Alcohol Abuse and Alcoholism. (2013). Rethinking drinking. Retrieved from Bethesda, MD: [Google Scholar]
  31. National Institute on Drug Abuse. (2012). Screening for Drug Use in General Medical Settings: Resource Guide. Retrieved from Bethesda, MD: https://www.drugabuse.gov/publications/resource-guide/preface [Google Scholar]
  32. Paolillo EW, Obermeit LC, Tang B, Depp CA, Vaida F, Moore DJ, & Moore RC (2017). Smartphone-based ecological momentary assessment (EMA) of alcohol and cannabis use in older adults with and without HIV infection. Addictive Behaviors. doi: 10.1016/j.addbeh.2017.10.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Parsons JT, Starks TJ, Millar BM, Boonrai K, & Marcotte D (2014). Patterns of Substance Use among HIV-Positive Adults Over 50: Implications for Treatment and Medication Adherence. Drug and Alcohol Dependence, 139, 33–40. doi: 10.1016/j.drugalcdep.2014.02.704 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Raudenbush SW, & Bryk AS (2002). Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA: Sage Publications. [Google Scholar]
  35. Roberts MH, Klatzkin RR, & Mechlin B (2015). Social support attenuates physiological stress responses and experimental pain sensitivity to cold pressor pain. Ann Behav Med, 49(4), 557–569. doi: 10.1007/s12160-015-9686-3 [DOI] [PubMed] [Google Scholar]
  36. Rodriguez-Penney AT, Iudicello JE, Riggs PK, Doyle K, Ellis RJ, Letendre SL, . . . Woods SP (2013). Co-Morbidities in Persons Infected with HIV: Increased Burden with Older Age and Negative Effects on Health-Related Quality of Life. AIDS Patient Care and STDs, 27(1), 5–16. doi: 10.1089/apc.2012.0329 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Rotgers F (2003). Cognitive-behavioral theories of substance abuse In Rotgers F, Morgenstern J, & Walters ST (Eds.), Treating substance abuse: Theory and technique (2nd ed., pp. 166–189). New York: The Guilford Press. [Google Scholar]
  38. Sacco P, Burruss K, Smith CA, Kuerbis A, Harrington D, Moore AA, & Resnick B (2015). Drinking behavior among older adults at a continuing care retirement community: Affective and motivational influences. Aging & Mental Health, 19(3), 279–289. doi: 10.1080/13607863.2014.933307 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Sangarlangkarn A, & Appelabaum JS (2016). Caring for older adults with the Human Immuno-deficiency Virus. Journal of American Geriatrics Society, 64, 2322–2329. [DOI] [PubMed] [Google Scholar]
  40. SAS Institute Inc. (2002–2012). SAS software, Version 13.1 for Windows. Cary, NC: Author. [Google Scholar]
  41. Shiffman S, Stone AA, & Hufford MR (2008). Ecological momentary assessment. Annual Review of Clinical Psychology, 4, 1–32. doi: 10.1146/annurev.clinpsy.3.022806.091415 [DOI] [PubMed] [Google Scholar]
  42. Singer JD, & Willett JB (2003). Applied longitudinal data analysis: Modeling change and event occurrence. New York: Oxford University Press. [Google Scholar]
  43. Sinha R (2001). How does stress increase risk of drug abuse and relapse? Psychopharmacology (Berl), 158(4), 343–359. doi: 10.1007/s002130100917 [DOI] [PubMed] [Google Scholar]
  44. Skalski LM, Sikkema KJ, Heckman TG, & Meade CS (2013). Coping styles and illicit drug use in older adults with HIV/AIDS. Psychology of Addictive Behaviors, 27(4), 1050–1058. doi: 10.1037/a0031044 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Sullivan MJL, Bishop SR, & Pivik J (1995). The pain catastrophising scale: Development and validation. Psychological Assessment, 7(2), 524–532. [Google Scholar]
  46. 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 Med, 16(10), 1870–1881. doi: 10.1111/pme.12799 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Vlaeyen JW, de Jong J, Geilen M, Heuts PH, & van Breukelen G (2001). Graded exposure in vivo in the treatment of pain-related fear: A replicated single-case experimental design in four patients with chronic low back pain. Behav Res Ther, 39(2), 151–166. [DOI] [PubMed] [Google Scholar]
  48. Zeger SL, & Liang K (1986). Longitudinal data analysis for discrete and continuous outcomes. Biometrics, 42, 121–130. [PubMed] [Google Scholar]

RESOURCES