Experience of Pain among Women with Advanced HIV Disease

Abstract

We evaluated pain frequency and severity in 339 women enrolled in the Women's Interagency HIV Study (WIHS). Among these, 63% were 39 years of age or younger, 17% were white, 54% African American, and 29% Hispanic; 32% did not complete high school; 58% had a CD4 less than 200; 65% had clinical AIDS; 60% were on highly active antiretroviral therapy (HAART); and 32% had a viral load of 50,000 or more. Data were collected between 1996 and 1998. Within the past 6 months 190 (56%) women experienced pain 6 or more days and 168 (50%) women indicated pain severity scores of 4 or 5 (5-point scale). Pain frequency and pain severity were not associated with age, education, ethnicity, current therapy, or location of the WIHS site. Pain frequency and severity were related to lower CD4 count, higher depression, with a history and longer duration of smoking and use of marijuana. Severity was associated with a history of crack/cocaine or heroin use or with injection drug use as the transmission category. In the multivariate models, pain severity was related to CD4 count and depression and to current tobacco use but not to crack, cocaine, heroin, or marijuana use. Pain frequency was related to depression and to former tobacco, crack, cocaine, heroin, or marijuana use but not to current use. The long-term effects of tobacco use may be to increase pain experience but women may also smoke tobacco or use other substances to give mild pain relief. Pain is frequent and often severe among women with HIV requiring medical management.

Introduction

Studies that have examined the issue of pain among persons living with HIV/AIDS (PLWHA) show between 30% and 88% of patients experience pain.^1–6

In a cohort of approximately 500 ambulatory PLWHA Breitbart et al.² found that 62.6% reported “persistent or frequent pain of any type.” There was no correlation between intensity of pain and the purported means of acquiring HIV or whether the patient had a history of injection drug use (IDU). Pain prevalence did not differ by gender, race, or CD4 T-cell count; however, those receiving antiretroviral (non-highly active antiretroviral therapy [HAART]) medication were slightly less likely to report pain. The 117 female participants in this study who were most likely to report pain had a history of IDU, a history of at least one major opportunistic infection (OI), and had more psychological distress, hopelessness, and depression. Del Borgo et al.⁷ found pain to be more prevalent among patients with a history of IDU, however, Bernard et al.⁴ found no correlation between the presence of pain and transmission category, CD4 T-cell count and antiretroviral therapy (ART), although females were twice as likely to report pain. Rotheram-Borus⁶ found that women with advanced HIV disease had higher reported pain than men. Martin et al.⁸ found no difference in pain prevalence and number of pain sites between patients with and without a history of IDU but females with a history of IDU showed a higher prevalence of pain.

Some studies raise the question of whether the type or intensity of pain felt in female PLWHA may be different than in males^9–12 and one study found that women reported significantly greater pain intensity.¹³

The issue of pain in PLWHA is complicated by both history of substance use and by therapy that may be provided to control pain. A few studies have examined pain tolerance in substance using populations which may have particular relevance for PLWHA. Methadone maintained patients are relatively pain-intolerant and may require more opioid than other patients to obtain adequate pain relief, not because of analgesic tolerance but because of underlying relative intolerance for pain^14,15 while Liebmann et al.¹⁶ found that pain tolerance in former opioid addicts is higher than among controls.

It is important to examine a broad range of substance use among PLWHA that may have an impact on pain experience including tobacco, alcohol, cocaine, and opiods In this study, we address the prevalence and intensity of pain among a large group of women enrolled in the Women's Interagency HIV Study (WIHS). In particular we examine the history of use of a broad range of illicit and legal substances as well as demographic and disease characteristics.

Methods

WIHS is an observational and longitudinal six-center cohort study of HIV disease in women. The study methods have been described previously.¹⁷ Consent to participate in the WIHS study was obtained from each woman and an incentive payment was given to each participant. The substudy of pain was implemented at four of the six sites within the WIHS. The data were from 104 ambulatory women from the Bronx, New York; 66 from Chicago, Illinois; 68 from Washington, D.C., and 101 from Los Angeles, California. This study aimed to describe the relationship between patient characteristics including demographics, substance use and extent of HIV disease and patient self reports of pain experience. Data were collected prospectively from 1996–1998, during the regularly scheduled WIHS visit.

Participant selection and recruitment

Any HIV-positive participant in the WIHS study who was alive as of October 1, 1996 and who ever had a CD4 T cell count less than or equal to 200 between their first and seventh WIHS visit were eligible. This criteria identified those with a immunologically defined AIDS diagnosis and allowed us to examine pain in those with more advanced disease. Eligible patients must have had a WIHS visit between visits 4 and 8. Of the WIHS study participants at these four sites, 442 met these criteria and 339 (76.7%) were enrolled in the substudy.

Measures

An interview that collected data related to pain experience was administered in English or Spanish depending on the preference of the participants. This face-to-face interview was administered by the same staff and in the same manner as the structured WIHS core interview and elicited data about the participant's pain experience over the 6 months preceding the interview. Prior to the start of enrollment, interviewers were trained via conference call to administer the interview in a consistent manner.

Pain assessment

Women were asked to estimate the frequency of their pain and the severity of their pain over the last 6 months. Regularity was assessed as no pain, infrequent pain (1–5 days), frequent pain (6–22 days), often in pain (25–120 days), and daily pain. Women were also given a 5-point visual analogue scale with extreme descriptors of “no pain” versus “extreme pain.” Patients were asked to indicate their location on this descriptor for worst pain in the last 6 months and ranged from no pain (level 1) to extreme pain (level 5). In addition, 263 medical record abstracts were examined and types of pain reported were abstracted.

Substance use

Illicit drug use history was not used as a subject selection criterion. Women were identified in terms of their history of use of tobacco, marijuana, crack, cocaine, or heroin; in each case they were categorized as never, former, or current (since last study visit, usually 6 months earlier) substance users. Alcohol use was categorized in terms of numbers of days alcohol was consumed per week and numbers of drinks per week. We further identified women whose HIV transmission risk factor included history of injection drug use or not.

Depression

We administered the Centers for Epidemiologic Studies Depression Scale (CES-D) as an index of psychological distress.^18,19 Using this scale, depression is dichotomized as not exhibiting depressive symptomotology (scores ≤15) or exhibiting depressive symptomotology (scores ≥16).

HIV disease status

Classification of disease status was based on four criteria: CD4 T-cell lymphocyte count and viral load on the day of the visit, and self report of history of AIDS defining illness, and duration of an AIDS-defining illness. HIV serostatus was confirmed using Food and Drug Administration (FDA)-approved enzyme-linked immunosorbent assay (ELISA) testing and if reactive, an FDA-approved Western blot HIV-1 confirmatory assay. Absolute CD4 T-cell count was determined by immunophenotyping and flow cytometric analysis. Laboratories performing flow cytometry testing were participating in the NIAID DAIDS Flow Cytometry Quality Assessment Program and followed Guidelines for Flow Cytometric Immunophenotyping.²⁰ The CD4 T-cell count has been shown to correlate with clinical AIDS diagnosis and disease progression in the WIHS cohort.²¹ CD4 T-cell count was stratified between those participants with CD4 greater than 200 (× 10⁶/L) and those with CD4 200 or less (× 10⁶/L) (the CDC definition for immunologic AIDS).

Plasma from individual patients were tested for the detection and quantitation of HIV RNA copies ^22,23 (viral load) using nucleic acid sequence-based amplification (NASBA), which uses an isothermic RNA amplification method. Laboratories performing the NASBA testing were participants in the NIAID AIDS Program Virology Quality Assurance HIV RNA Proficiency Program of the National Institutes of Health (NIH).²⁴ Viral load was detectable at 4000 copies or more per milliliter using the NASBA method. A stratification of viral load was used that represented approximately equivalent number of women per category: viral load less than 4000 copies per milliliter (the assay cutoff of detection), viral load 4000 to 35,000 copies per milliliter, and viral load over 35,000 copies per milliliter.

Therapy was examined in two ways. First, we categorized as HAART versus no HAART. Second, we categorized as no therapy, monotherapy, combo-therapy, protease-inhibitor (PI)–based HAART and non-nucleoside reverse transcriptase inhibitor (NNRTI)-based HAART.

Overview of statistical analyses

We examined the relationship of the pain scales with the brief assessments from the RAND Quality of Life scale, which asks three questions: “How much body pain have you generally had during the past 4 weeks?” “During the past 4 weeks how much did bodily pain interfere with normal work (including work outside the house and housework)” “Overall, how would you rate your quality of life?” In order to provide a validation of the pain scales, we correlated the two pain scales with these items form the RAND assessment. These assessments are highly correlated, in all cases with p < 0.001 (correlations ranging from 0.47 to 0.62 with the RAND pain questions and from −0.22 to −0.35 with the RAND quality-of-life question). We chose to use the 6-month assessments because these cover the same time period assessments as on the WIHS questionnaire.

Pain was collapsed into dichotomous variables so that category frequencies were sufficient for analysis using logistic regression. We used a series of univariate analyses that predicted severity and persistence of pain, calculating odds ratios (ORs) and 95% confidence intervals (CIs; i.e., p ≤ 0.05) using maximum likelihood estimates from logistic regression models. We examined the relationship of severe and persistent pain with demographic characteristics, depression, tobacco, marijuana, crack/cocaine or heroin use, disease characteristics, and HAART use. We used multivariate logistic regression, adjusted for the variables that were significant on univariate analysis, to evaluate the risk for pain associated with these covariates separately for each substance use variable due to the high level of covariation between the substance use variables.

Results

Of the 442 eligible patients who were present at one or more visits between visits 4 and 8, we collected data from 339 (76.7%). We compared characteristics of those who were eligible and enrolled to those who were eligible but not enrolled to determine potential biases in the surveyed sample. Those enrolled were not significantly different from those not enrolled on ethnicity, age, CD4 count, viral load, and site of data collection. However, those who were enrolled were more likely to be on HAART (60% versus 12%), more likely to have an AIDS-defining illness (65% versus 49%), and for those with clinical AIDS, to have a longer duration of an AIDS-defining illness (21 versus 12 months; all p < 0.01). Those enrolled were also less depressed as measured by the CES-D (mean 15.6 versus 18.4; p < 0.01).

Among these 339, 63% were 39 years of age or younger, 17% were white, 54% African American, and 29% Hispanic; 32% did not complete high school, 31% completed high school, and 37% had some college or more; 58% had a CD4 count under 200 and 65% had clinical AIDS; 60% were on HAART; and 32% had a viral load of 50,000 or more.

Pain reports

Women were asked to estimate the frequency and severity of their pain over the last 6 months. While 16.5% indicated no pain, 27.4% indicated infrequent pain (1–5 days), 25.7% indicated frequent pain (6–22 days), 20.4% indicated they were often in pain (25–120 days), and 10% indicated pain more than 120 days or daily. Pain frequency was dichotomized as less than 6 days of pain in the past 6 months as compared to 6 or more days of pain in the last 6 months. In terms of the severity of their worst pain in the last 6 months, 15.6% indicated “no pain” (level 1) 12.7%, 22.1%, and 15.0% were at levels 2, 3, and 4, respectively, and 34.5% at level 5, (extreme pain). Pain severity was dichotomized with the levels 1–3 considered low pain and 4–5 considered high pain. After dichotomizing, 190 (56%) women experienced pain 6 or more days in the past 6 months and 168 (50%) women indicated pain severity scores of 4 or 5.

Over the same period of time, various types of pain were abstracted from 263 medical records: including Peripheral neuropathy, 22 (8.4%); herpetic or postherpetic neuralgia, 10 (3.8%); headache, 55 (21.1%); musculoskeletal, 49 (18.8%); abdominal pain, 46 (17.6%); menstrual cramps, 2 (0.8%); ear, nose, throat, neck, 13 (4.9%); esophageal, 3 (1.4%); skin/infection, 7 (2.7%); chest pain, 10 (3.8%); miscellaneous (bladder, kidney, sickle cell), 3 (1.4%); and unspecified, 6 (3.0%). While not all records were abstracted, this subset does provide an indication of the types of pain reported.

Relationship between pain assessment and sociodemographic and disease indicators

Pain frequency and pain severity were not associated with age, education, ethnicity, or location of the WIHS site (p >0.05 for all). Lower CD4 was highly correlated with increased pain severity over the past 6 months OR = 2.22, p = 0.0005) and with increased pain frequency over the past 6 months (OR = 1.84, p = 0.007). Increased viral load, dichotomized at over and under 50,000, was also related to increase pain frequency (OR = 1.88, p = 0.01) but not with increased pain severity (OR = 1.46 p = 0.11).

Self-report of history of an AIDS-defining illness was not associated with pain severity (OR = 1.44, p = 0.11) but approached significance with increased pain frequency (OR =1.5, p = 0.08). Both pain frequency and pain severity also approached significance with duration of an AIDS-defining illness (OR = 1.18, p = 0.06; OR = 1.18, p = 0.07, respectively).

Most women were on antiretroviral therapy at the time of the interview; 42 (12.4%) women were on no therapy, 15 (4.4%) were on monotherapy, 79 (23.4%) were on combination therapy, and 174 (51.4%) were on PI-based highly active antiretroviral therapy (HAART), 5 (1.4%) on NNRTI-based HAART, and 23 (6.8%) received both PI- and NNRTI-based HAART. Using another categorization 136 (40.1%) were not on HAART and 203 (59.9%) were on HAART. Regardless of the categorization, there was no relationship with pain severity or frequency.

Relationship between pain assessment and history of substance use

The prevalence of past and current substance use is shown in Table 1. In terms of HIV transmission category, those with more severe pain were more likely to have a history of IDU as their HIV transmission category (OR = 1.83, p = 0.01) but this was not significantly related to pain frequency.

Table 1.

Univariate Analysis of Demographic, HIV Disease Status, Depression, and Substance Use on Pain Severity and Pain Frequency

	Severity of paina			Frequency of painb
Variable	n	Univariate OR	p value	Univariate OR	p value
Age (years)
<39	213	1.00		1.00
≥40	126	1.47 (0.9–2.3)	0.09	1.07 (0.7–1.7)	0.75
Ethnicity
White	56	1.00		1.00
African American	182	1.27 (0.8–2.0)	0.30	1.38 (0.9–2.2)	0.16
Hispanic	100	0.85 (0.5–1.4)	0.52	1.13 (0.7–1.8)	0.69
Education
Did not complete high school	107	1.0		1.00
Completed high school	106	1.23 (0.7–2.1)	0.45	0.91 (0.5–1.5)	0.74
Some college or more	126	1.55 (0.9–2.6)	0.10	1.1 (0.6–1.8)	0.77
Test for Trend		1.24 (0.96–1.61)	0.10	1.09 (0.80–1.35)	0.76
Risk category at diagnosis
Not IDU	249	1.0		1.0
IDU	96	1.83 (1.13–2.97)	0.01	1.29 (0.80–2.09)	0.30
Viral load (copies/mL)
<50000	219	1.00		1.00
≥50000	107	1.46 (0.9–2.3)	0.11	1.88 (1.2–3.1)	0.01
CD4 T-lymphocyte (× 109/L)
≥200	134	1.00		1.00
<200	195	2.22 (1.4–3.5)	0.0005	1.84 (1.2–2.9)	0.007
Clinical AIDS
No	119	1.0		1.0
Yes	220	1.44 (0.92–2.25)	0.11	1.50 (0.9–2.3)	0.08
Duration of clinical AIDS
No AIDs	119	1.0		1.0
0–18 months	55	1.24 (0.66–2.36)	0.51	1.53 (0.8–2.9)	0.20
19–25 months	73	1.32 (0.74–2.38)	0.35	1.17 (0.6–2.1)	0.61
>25 months	92	1.68 (0.97–2.90)	0.07	1.82 (1.0–3.2)	0.04
Test for trend		1.18 (0.99–1.41)	0.07	1.18 (0.99–1.1)	0.06
CES-D depression
≤15	172	1.0		1.0
>15	167	2.52 (1.62–3.90)	<0.0001	2.45 (1.58–3.81)	<0.0001
HAART
No	136	1.0		1.0
Yes	203	0.80 (0.52–1.23)	0.31	1.12 (0.72–1.73)	0.62
Antiretroviral therapy
No treatment	42	1.0		1.0
Mono	15	0.55 (0.17–1.83)	0.33	1.02 (0.31–3.40)	0.97
Combo	79	0.99 (0.47–2.09)	0.97	0.70 (0.33–1.49)	0.35
PI-based HAART	174	0.70 (0.36–1.38)	0.31	0.82 (0.41–1.62)	0.56
NNRTI Based HAART	5	0.21 (0.02–2.01)	0.17	2.72 (0.28–26.5)	0.39
HAART with PI and NNRTI	23	1.55 (0.54–4.43)	0.41	1.93 (0.63–5.88)	0.25
Smoking
Never	110	1.00		1.00
Former	65	1.69 (0.91–3.12)	0.09	2.17 (1.16–4.06)	0.02
Current	158	2.22 (1.36–3.63)	0.001	1.79 (1.10–2.91)	0.02
Marijuana or hash use
Never	104	1.0		1.00
Former	185	1.72 (1.06–2.79)	0.03	1.99 (1.22–3.22)	0.006
Current	48	2.13 (1.07–4.67)	0.03	1.55 (0.78–3.09)	0.21
Crack/cocaine/heroin use
Never	128	1.0		1.00
Former	167	1.30 (0.82–2.05)	0.27	1.77 (1.11–2.81)	0.02
Current	42	2.31 (1.12–4.74)	0.02	1.61 (0.80–3.27)	0.18
Days drink per week
None	59	1.0		1.0
<1 day/wk	82	1.26 (0.65–2.44)	0.50	1.22 (0.63–2.37)	0.56
1–7 days/wk	196	0.93 (0.53–1.66)	0.81	1.83 (0.67–2.11)	0.57
# Drinks/week (continuous)		1.03 (0.83–1.27)	0.82	0.91 (0.74–1.12)	0.38

^aPain severity is dichotomized on a 5-point scale with 1 = no pain to 5 = extreme pain; levels 1–3 are categorized as low pain and levels 4–5 as high pain.

^bPain frequency was dichotomized as no pain or pain 5 days or fewer in the past 6 months compared to pain experienced 6 days or more in the prior 6 months.

OR, odds ratio; IDU, injection drug use; CES-D, Centers for Epidemiologic Studies Depression Scale; HAART, highly active antiretroviral therapy; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.

Using past and current drug use histories, participants could be categorized as never, former or current users of tobacco, marijuana/hashish or crack/cocaine, or heroin. In the univariate analyses, by comparing former and current substance users against never substance users, significant relationships were between pain and tobacco, marijuana, and crack cocaine/heroin use. Pain frequency was related to current tobacco use and marginally to former tobacco use, to former and current marijuana use, and to current crack, cocaine or heroin. Pain severity was related to former and current tobacco use, former marijuana use and former crack/cocaine or heroin. While not every comparison was significant the data consistently indicated that former or current substance use was related to both more severe and more frequent pain. Neither pain severity nor pain frequency was significantly related to alcohol use.

Because of the high level of covariation between the substance use variables, we conducted our multivariate analysis separately for tobacco, marijuana/hashish, and crack/cocaine/heroin. These results are shown in Tables 2 and 3. Pain severity was significantly related to CD4 T-cell count and to depression in all three multivariate models. Pain severity was also related to current tobacco use but not to current or former marijuana use or crack, cocaine or heroin use. Pain frequency was highly related to depression in all models. Pain frequency was related to former and marginally to current tobacco use as well as to former use of marijuana, crack, cocaine, or heroin use.

Table 2.

Multivariate Analysis of Severity of Pain Using Three Models with Specific Substance Use Variables^a

Variable	n	Multivariate OR	p value for trend	Multivariate OR	p value for trend	Multivariate OR	p value for trend
Age (years)
<39	203	1.00		1.00		1.00
≥40	126	1.33 (0.81–2.19)	0.25	1.35 (0.82–2.21)	0.24	1.35 (0.82–2.20)	0.24
Risk category at diagnosis
Not IDU	240	1.00		1.00		1.00
IDU	96	1.31 (0.74–2.33)	0.36	1.51 (0.87–2.60)	0.14	2.68 (0.93–3.02)	0.09
CD4 T-lymphocyte (× 109/L) (NIO1 visit)
≥200	134	1.00		1.00		1.00
<200	195	2.08 (1.22–3.51)	0.007	2.12 (1.25–3.60)	0.006	2.07 (1.22–3.50)	0.007
Viral load (copies/mL) (NIO1 visit)
<50000	219	1.00		1.00		1.00
≥50000	107	1.06 (0.61–1.87)	0.83	1.03 (0.58–1.81)	0.92	1.10 (0.63–1.92)	0.75
Duration of clinical AIDS(NIO1 visit)
No AIDS	119	1.0		1.0		1.0
0–18 months	55	0.71 (0.34–1.49)	0.37	0.74 (0.36–1.55)	0.43	0.72 (0.35–1.49)	0.38
19–25 months	73	1.06 (0.56–2.03)	0.86	1.05 (0.55–2.00)	0.88	1.04 (0.55–1.97)	0.91
>25 months	92	1.17 (0.64–2.15)	0.62	1.28 (0.70–2.37)	0.42	1.26 (0.69–2.31)	0.46
CES-D depression (NIO1 visit)
≤15	172	1.0		1.0		1.0
>15	167	2.38 (1.47–3.84)	0.0004	2.34 (1.45–3.77)	0.0005	2.29 (1.42–3.69)	0.0007
Smoking
Never	110	1.00
Former	65	1.59 (0.80–3.15)	0.18
Current	158	1.84 (1.03–3.30)	0.04
Marijuana or hash use
Never	104			1.00
Former	185			1.42 (0.8 2–2.45)	0.21
Current	48			1.70 (0.80–3.93)	0.17
Crack/cocaine/heroin use
Never	128					1.00
Former	167					0.91 (0.53–1.58)	0.74
Current	42					1.12 (0.49–2.56)	0.78

^aPain severity is dichotomized on a 5 point scale with 1 = no pain to 5 = extreme pain; levels 1–3 are categorized as low pain and levels 4–5 as high pain.

Each substance use variable is tested in a separate model.

OR, odds ratio; IDU, injection drug use; CES-D, Centers for Epidemiologic Studies Depression Scale.

Table 3.

Multivariate Analysis of Frequency of Pain Using Three Models with Specific Substance Use Variables^a

Variable	n	Multivariate OR	p value for trend	Multivariate OR	p value for trend	Multivariate OR	p value for trend
CD4 T-lymphocyte (× 109/L) (NIO1 visit)
≥200	134	1.00		1.00		1.00
<200	195	1.46 (0.87–2.47)	0.15	1.50 (0.89–2.51)	0.13	1.43 (0.85–2.40)	0.17
Viral load (copies/mL) (NIO1 visit)
<50000	219	1.00		1.00		1.00
≥50000	107	1.60 (0.91–2.81)	0.10	1.52 (0.86–2.66)	0.15	1.58 (0.90–2.77)	0.11
Duration of clinical AIDS (NIO1 visit)
No AIDS	119	1.0		1.0		1.0
0–18 months	55	1.06 (0.52–2.17)	0.88	1.05 (0.52–2.14)	0.89	1.09 (0.53–2.21)	0.82
19–25 months	73	0.94 (0.50–1.78)	0.85	0.91 (0.48–1.71)	0.76	0.97 (0.52–1.83)	0.93
>25 months	92	1.63 (0.88–3.02)	0.12	1.69 (0.92–3.11)	0.09	1.75 (0.95–3.21)	0.07
CES-D depression (NIO1 visit)
≤15	172	1.0		1.0		1.0
>15	167	2.51 (1.55–4.06)	0.0002	2.40 (1.49–3.86)	0.0003	2.46 (1.52–3.98)	0.0002
Smoking
Never	110	1.00
Former	65	2.58 (1.30–5.13)	0.007
Current	158	1.56 (0.92–2.64)	0.10
Marijuana or hash use
Never	104			1.00
Former	185			1.76 (1.05–2.97)	0.033
Current	48			1.41 (0.67–2.95)	0.36
Crack/cocaine/heroin use
Never	128					1.00
Former	167					1.67 (1.01–2.76)	0.05
Current	42					0.99 (0.46–2.11)	0.97

^aPain Frequency was dichotomized as no pain or pain 5 days or fewer in the past 6 months as compared to pain experienced 6 days or more in the prior six months.

Each substance use variable is tested in a separate model.

OR, odds ratio; CES-D, Centers for Epidemiologic Studies Depression Scale.

Discussion

There is little prior research on the pain experience among women living with HIV/AIDS. These data suggest that pain is a significant problem both in terms of intensity and persistence of pain. Of the 339 women, 190 (56%) experienced pain six or more days in the past 6 months and 168 (50%) indicated their worst pain severity scores of 4 or 5 with 5 indicating extreme pain. These results are consistent with reports indicating a wide range of pain prevalence from 30% to 88%.^5,11 Our data demonstrated little relationship between pain and demographic characteristics including age, ethnicity, and education.

Consistent with prior studies, pain severity and frequency were significantly related or approached significance to indicators of HIV disease progression in the univariate analysis including lower CD4 T-cell count, clinical AIDS, duration of clinical AIDS, and viral load. In the multivariate analysis of pain severity, lower CD4 T-cell count was related to increased pain, although other markers of HIV disease became nonsignificant in the multivariate analysis possibly due to covariation between the disease progression indicators. The relationship between pain and progressive HIV disease could be related to the effects of the virus on the nervous system, to development of intercurrent opportunistic infections and/or cancer, or possibly even to treatments for HIV infection.^25,26 Our data indicated that the deterioration of the immune system as evidenced by lower CD4 T-cell counts as well as the diagnosis and duration of clinical AIDS were important factors related to the experience of pain, however, we found no relationship between antiretroviral therapy and pain. However, it should also be noted that these data were collected in the early era of effective HAART therapy and viral control may have been recent and of short duration. Further, medications differed in dosing interval and pill burden and potentially were less robust and may have had more side effects than the regimens of today. Our study is thus limited because the data were collected at a time when protease inhibitors were being newly introduced and HAART has changed dramatically since these data were collected and these changes in therapy may have an impact on manifestations of pain.

As expected, pain severity and frequency were also related to depression. Nearly half of the women in this study experienced depressive symptoms ≥ 16 on the CES-D scale. The relationship between chronic pain and depression have been well documented in individuals with chronic disease.^27–29 The direction of this relationship may be reciprocal, those who experience pain are likely to become depressed and those who are depressed may be more likely to report pain experience. While not all patients with pain are depressed, both pain and depression may result from manifestations of disease that disturb other aspects of life function such as mobility, vitality, social interactions, and fear of an uncertain future. Depression is measured in terms of experience of sadness, discouragement, lack of motivation and activity; all of these may be associated with the experience of pain as well. The relationships between disease progression, depression, and pain are clearly plausible and have been repeatedly documented. This constellation of pain and depression may require aggressive clinical management as the disease progresses.

The analysis of substance use and the experience of pain yielded interesting results. IDU as the likely cause of HIV infection was related to pain severity but not to pain frequency in the univariate analysis consistent with several previous studies.^13,30 In the univariate analysis, current or former use of tobacco, marijuana, and crack-cocaine or heroin were all related to increased risk for the experience of severe pain and frequent pain. In multivariate analyses, current tobacco use continued to be related to pain severity and former tobacco use to pain frequency. Former marijuana use was related to pain frequency. Former crack-cocaine or heroin use was related to pain frequency. Tsao et al.³¹ also showed a relationship between pain and illicit substance use history and suggested that illicit substance use increases illness burden and subsequent pain. However, specific drugs were not individually examined in that study and tobacco and alcohol use were not examined at all. Nevertheless, the concept that substance use leads to greater health problems and adverse events and subsequently to pain is plausible. Alternatively, it may be that women with a long history of tobacco, marijuana, and other substance use have less ability to tolerate pain. Persons with a history of substance use may be more likely to exhibit hypervigilance to pain and to report more pain symptoms.³¹

Tsao et al.³² also showed that HIV-infected patients with pain were more likely to utilize complementary and alternative medicine including unproven medications purported to treat their pain, further those with a history of IDU were more likely to use unproven medications when in pain and perhaps they may also seek various substances to treat their pain. Due to the long history of substance use in many women with HIV, substance use may have predated the diagnosis of HIV; the pain experienced since the diagnosis of HIV could lead to a return to or continuation of substance use.

The association of tobacco use and pain among persons with HIV has not been extensively explored. The level of current tobacco use is very high (47%) in our cohort, and was associated with greater experience of pain. Conclusions cannot be drawn about the direction of causality. It may be that smokers experience more pain or it may be that patients who experience pain find some relief in smoking tobacco or marijuana. A few studies have shown that smoking may increase the pain threshold and tolerance of pain particularly in men.^34,35 However many studies have shown no effect of smoking on pain^36,37 and one study has shown that while there is reduced pain to acute stressors this differs by gender and the specific characteristic of the stressor.³⁸ If smoking increases pain tolerance then patients may use it as a means of self medication. On the other hand, one study of smoking cessation in PLWHA found that smoking cessation and particularly the length of smoking abstinence was significantly associated with lower levels of HIV-related symptom burden.³⁹ The symptom burden index did include pain as a component but analysis for each specific component of symptom burden was not reported. Another study also showed that smoking was associated with reduced measures of health-related quality of life (HRQL), which also included pain as a component of the measure but again the specific relationship with pain was not reported.⁴⁰

A large study in Germany showed that tobacco smoking is related to analgesic drug.⁴¹ Similarly, Palmer et al.⁴² found in a large British population sample that former and current smokers report more back, neck, and limb pain than never smokers. Furthermore, in a review of 38 studies, 18 of 26 studies in men and 18 of 20 studies in women showed an association between smoking and back pain.⁴³ While there may be methodological concerns about the studies linking smoking and pain, particularly back pain, there are also plausible biologic mechanisms that could account for this finding including diminished bone mineral content resulting in osteoporosis and microfractures of the vertebrae, reduced blood flow to the vertebrae, and excessive coughing resulting in straining of the spine and promoting disc bulging and herniation. Furthermore, current smoking may be used to provide temporary relief from pain similar to the use of analgesics.

Similarly, marijuana may also be effective as an analgesic agent in patients with pain. However few studies have been conducted to confirm the efficacy of marijuana, particularly smoked marijuana, for relief of pain in humans.⁴⁴ Noyes et al.⁴⁵ found that THC administered as an injection was superior to placebo for pain reduction. However, Campbell et al.⁴⁴ showed that canabinoids were not more effective in controlling pain than codeine. However, because marijuana is available and familiar, it may be a preferred choice for self treatment by people with pain. In a study of 252 HIV-infected individuals, 28% believed that smoked marijuana would provide pain relief. Those with moderate to severe pain were somewhat more likely to use marijuana than those with no pain although this finding was not statistically significant.⁴⁶ Another study on pain among trauma or surgery patients showed that 15% used cannabis for pain relief, and of these 78% reported at least moderate relief of pain.⁴⁷ A small pilot trial indicated a reduction in painful peripheral neuropathy after use of smoked marijuana.⁴⁸

In line with past examinations of this topic those with pain and a history of substance use, particularly IDU, may be less likely to receive therapy for their pain, suggesting that the pain experience may partially be due to a failure in the appropriate management of pain, a concern that his been raised by Breitbart et al.⁴⁹ and others.^12,50 In this study, while 96 women were classified as IDU transmission only 18 were current IDU and it is less likely that former IDU would be denied pain therapy and the 18 current users are a sufficiently small number that it would be unlikely to alter the results. Nevertheless this concern deserves further exploration.

Studies of pain are hampered by the subjective measurement biases that are inherent in large clinical cohort studies. Self-report of pain can be susceptible to other mood disturbances as well as to problems of under and over reporting. Our data suggest that depression is highly related to pain frequency and severity although the causal direction is not clear and may be recursive. Furthermore, the data show that lower CD4 T-cell count is related to greater pain severity. Our data suggest that women with HIV who are long-term smokers experience greater pain than nonsmokers. However, the relationship between substance use (including tobacco use) and pain among persons with HIV needs further study to determine more specific characteristics of the pain and the biological basis of the pain. Furthermore, we need to understand whether current smoking is providing some level of pain relief despite the continued risk incurred. Clearly, pain is a considerable problem among women with HIV. Nonetheless, the pain experience is poorly understood and not extensively investigated. It is important to understand how pain control for women with HIV disease should be managed.

Acknowledgments

Data in this manuscript were collected by the Women's Interagency HIV Study (WIHS) Collaborative Study Group with centers (Principal Investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, New York (Howard Minkoff ); Washington, D.C. Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); Data Coordinating Center (Stephen Gange). This WIHS substudy was conducted at four of the six WIHS sites. The WIHS is funded by the National Institute of Allergy and Infectious Diseases (UO1-AI-35004, UO1-AI-31834, UO1-AI-34994, UO1-AI-34989, UO1-AI-34993, and UO1-AI-42590) and by the National Institute of Child Health and Human Development (UO1-HD-32632). The study is cofunded by the National Cancer Institute, the National Institute on Drug Abuse, and the National Institute on Deafness and Other Communication Disorders. Funding is also provided by the National Center for Research Resources (UCSF-CTSI Gant Number UL1 RR024131) The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health.

Author Disclosure Statement

No competing financial interests exist.