Poorer physical health-related quality of life among aboriginals and injection drug users treated with highly active antiretroviral therapy.
Martin, Leah J. ; Houston, Stan ; Yasui, Yutaka 等
In Canada, Aboriginals and non-Aboriginals infected with HIV do not
appear to receive equal benefit from highly active antiretroviral
therapy (HAART). After starting HAART, Aboriginals have been shown to
experience lower rates of initial virological suppression, higher rates
of virological failure, (1) and higher rates of allcause (2,3) and
HIV-related2 mortality compared to non-Aboriginals. Although these
outcomes are all important indicators of treatment success, they do not
provide a complete picture of health outcomes. Improved quality of life
is another important goal of HAART, (4) and health-related quality of
life (HRQL) is of particular interest in a clinical setting. However, to
our knowledge, only one study (5) (not yet published in full at the time
of writing) has examined the impact of HAART on quality of life among
Aboriginals in Canada. Therefore, more evidence is needed to understand
the HRQL experienced by Aboriginal HIV patients.
Notwithstanding this lack of evidence, the poorer virological
outcomes experienced by Aboriginals suggest they may have an increased
risk of poor physical HRQL. Some research has found worse physical HRQL
to be associated with higher viral loads and lower CD4 cell counts;
(6,7) however, other studies have found no such associations. (8,9) In
addition, Aboriginal HIV patients may also be at increased risk of worse
HRQL due to injection drug use (IDU). In Canada, Aboriginal HIV patients
are more likely to be infected with HIV through IDU compared to
non-Aboriginals. (10) Studies have shown that IDUs have worse HRQL than
the general population (11) and, among HIV-infected individuals, than
non-IDUs. (12) Therefore, both clinical status and IDU are important
covariates to consider in understanding the relationship between
Aboriginal ethnicity and HRQL among patients receiving HAART.
Given the limited evidence available, more research is needed to
compare the HRQL of Aboriginal vs. non-Aboriginal HIV patients receiving
HAART to help improve our understanding of treatment outcomes for
Aboriginals. Therefore, the objectives of this study were to: 1)
describe the HRQL of HIV patients treated with HAART in northern
Alberta, Canada; 2) compare HRQL between non-Aboriginal non-IDUs vs.
Aboriginal IDUs, Aboriginal non-IDUs, and nonAboriginal IDUs; and 3)
assess whether any of these associations could be explained by clinical
status.
METHODS
Study patients
We assembled a cohort of patients using the Northern Alberta HIV
Program (NAHIVP) clinic database and the following eligibility criteria:
1) started HAART between 1 January 1997 and 31 December 2005 (baseline);
2) previously ART-naive; 3) [greater than or equal to] 18 years of age
(as of September 1, 2006); 4) had ever visited either of two local
hospital-based clinics during their time at NAHIVP; 5) visited a NAHIVP
clinic within the two calendar years before the study (2004-2006); and
6) were alive according to information available in the database at the
time of data extraction. Using these criteria and our enrolment period
of September 2006August 2007, the possible range of time between
starting HAART and interview was approximately 8 months-101/2 years. We
excluded patients if they started HAART [less than or equal to] 26 weeks
before delivering a baby in an effort to limit the study to patients who
started HAART for the purpose of treatment, rather than to prevent
vertical transmission of HIV; we assumed that starting HAART earlier in
pregnancy or after delivery would be for maternal indications. In
addition, we excluded patients who did not have current clinical status
available, which was assessed using a patient's most recent viral
load and CD4 cell count in the 6 months before the interview date.
Data sources
In addition to data from NAHIVP, we used viral load data from the
Alberta Provincial Public Health Laboratory to replace missing baseline
viral loads where possible. To measure HRQL, we used the Medical
Outcomes Study (MOS)-HIV questionnaire and generated mental (MHS) and
physical (PHS) health summary scores, which are standardized to have a
mean 50 (SD 10) in a standard population. (13)
Definitions
We defined Aboriginals as Treaty and non-Treaty First Nations,
Metis, and Inuit. HIV exposure categories were classified using an
exposure category hierarchy. (14) Patients were defined as IDUs if their
HIV exposure was recorded as IDU or any exposure combined with IDU;
patients with other exposure categories, including unknown or missing
exposures, were considered non-IDUs. We defined HAART as a combination
of at least three antiretrovirals, other than ritonavir, recorded as
prescribed on the same date. We excluded ritonavir under the assumption
that, during the HAART initiation period (1997-2005), ritonavir would
have been prescribed at low dosages intended to "boost" other
protease inhibitors, rather than at therapeutic levels. The HAART start
date was the first date that a HAART prescription was recorded in the
database. Baseline CD4 cell count and HIV viral load were defined as
those measures taken closest to the HAART start date, which was <6
months before, and not after starting HAART.
Study management
In September 2006-August 2007, clinic staff approached eligible
patients about the study; interested patients spoke with the research
assistant, provided written consent, and self-administered the MOSHIV
questionnaire; however, the research assistant was available to help
patients and administer the questionnaire if needed. The present study
was conducted in conjunction with a more in-depth interview for
Aboriginals and IDUs; these patients were administered the MOS-HIV first
and given $10 for their participation because of the longer duration of
the second interview. The study procedures were approved by the
University of Alberta Health Research Ethics Board.
Data management and statistical analyses
We double entered the MOS-HIV questionnaires using EpiData 3.1 and
linked patients' MOS-HIV results with their NAHIVP record and
current CD4 cell counts and viral load results. We used the
"MOS-HIV Scoring Program--Scales and Summary Measures" to
calculate MOS-HIV scores; missing values were imputed using the average
score for multi-item ([greater than or equal to] 4 items) questions if
at least half the items were completed. (15) Patient characteristics
were tabulated and compared between Aboriginal IDUs, Aboriginal
non-IDUs, non-Aboriginal IDUs, and non-Aboriginal non-IDUs using [chi
square] and Fisher's exact tests for categorical variables and
Kruskal-Wallis equality of populations rank test (with ties) and
Wilcoxon rank sum test for continuous variables. In unadjusted analyses,
we compared each HRQL outcome (PHS and MHS scores) between Aboriginal
IDUs, Aboriginal non-IDUs, non-Aboriginal IDUs, and non-Aboriginal
non-IDUs, with the last category as the reference group, using linear
regression.
We developed two multivariable regression models for each HRQL
outcome. In the first model, we entered the following covariates:
indicator variables for our exposure variables of interest (Aboriginal
IDUs, Aboriginal non-IDUs, non-Aboriginal IDUs, and non-Aboriginal
non-IDUs), with the last category as the reference group; age at
interview; sex; and years of follow-up since baseline. The indicator
variables for our exposure variables of interest are equivalent to
assessing two main effect variables (Aboriginal ethnicity and IDU
status) and their interaction. In the second model, we additionally
adjusted for patients' current clinical condition by entering
current viral load (>400 copies/mL vs. [less than or equal to] 400
copies/mL) and CD4 cell count (>350 cells/[micro]L vs. [less than or
equal to] 350 cells/[micro]L) into the model. Additionally, we assessed
how using continuous variables for [log.sub.10] viral load and CD4 cell
count in our second models impacted the estimates for our exposure
variables of interest. The first models estimate the differences in the
two HRQL outcomes between Aboriginal IDUs, Aboriginal non-IDUs, and
non-Aboriginal IDUs vs. non-Aboriginal non-IDUs without accounting for
patients' current clinical status, while the second models estimate
the differences in HRQL that remain after accounting for differences in
clinical status. We assessed model assumptions (homoscedasticity,
normality of residuals, and the linearity of relationships between the
outcomes and continuous predictors) and explored unusual and influential
observations by examining residuals, leverage values, and DFBeta values
for the Aboriginal ethnicity and IDU variables.16 Data were analyzed
using SAS[R] (version 9.1; SAS Institute Inc., Cary, NC) and Stata SE
9.2 (College Station, TX).
RESULTS
Derivation of the study sample
Of the 424 potentially eligible patients identified, 119 (28%) were
approached to participate in the study. These 119 patients were not
significantly different from the 305 patients who were not approached
with respect to Aboriginal ethnicity, IDU, sex, or baseline age;
however, approached patients were more likely to have started HAART
recently (in 2003-2005) compared to those not approached (50% vs. 37%,
p=0.018). One hundred patients (84%) consented and were interviewed;
four patients were excluded for missing current CD4 cell counts or viral
load test results. The remaining 96 patients constituted the study
sample. Compared to patients who declined to participate or were
ineligible, study patients had a higher median baseline viral load
(160,000 vs. 28,000 (n=18) copies/mL, p=0.028) and tended to have a
lower median baseline CD4 cell count (170 vs. 220 (n=15) cells/uL,
p=0.066). The study patients were not more likely to be Aboriginal or
IDU, nor did they differ with respect to sex, age, or baseline calendar
year. Four of the 96 study patients were missing a total of 6 items on
the MOS-HIV.
Description of study patients
Of the 96 (72 male, 24 female) study patients, most were Caucasian
(52, 54%) or Aboriginal (34, 35%); 40 patients (42%) were IDU. At
interview, most (84%) had a viral load [less than or equal to] 400
copies/mL, but fewer than half (44%) had a CD4 cell count >350
cells/[micro]L (Table 1). Across the four exposure groups of interest,
viral loads (p=0.019), CD4 cell counts (p=0.002), PHS scores (p=0.001),
and MHS scores (p=0.031) differed significantly (Table 1). Although the
median viral load was the same for all groups (50 copies/mL), the 75th
percentile of the interquartile range was highest for Aboriginal IDUs
(44,000 copies/mL) (Table 1). Median CD4 cell counts, PHS scores, and
MHS scores were highest for non-Aboriginal non-IDUs and lower for the
other three groups. Of note, median MHS scores were higher than median
PHS scores for all groups.
Health-related quality of life
Physical Health
In our unadjusted analysis, Aboriginal IDUs, Aboriginal non-IDUs,
and non-Aboriginal IDUs had significantly lower PHS scores than
non-Aboriginal non-IDUs (Table 2). These results remained similar in our
first model, adjusting for age, sex, and follow-up years (Table 2). In
our second model, which additionally adjusted for current CD4 cell count
and viral load, compared to non-Aboriginal non-IDUs, PHS scores were
significantly lower among Aboriginal non-IDUs and non-Aboriginal IDUs;
however, PHS scores were no longer statistically significantly lower for
Aboriginal IDUs (Table 2). When we used continuous, rather than
dichotomous, variables for CD4 cell count and [log.sub.10] viral load in
our second model, the parameter estimates for our exposure variables of
interest remained similar (Aboriginal IDUs [-4.4; 95% CI -10.5, 1.7;
p=0.16], Aboriginal non-IDUs [-8.5; 95% CI -15.4, -1.6; p=0.017], and
non-Aboriginal IDUs [-6.6; 95% CI -12.8, -0.4; p=0.038]).
Mental Health
In our unadjusted analysis, non-Aboriginal IDUs had significantly
lower MHS scores than non-Aboriginal non-IDUs (Table 3). The other
exposure groups of interest (Aboriginal IDUs and Aboriginal non-IDUs)
also had lower MHS scores than non-Aboriginal nonIDUs; however these
were not statistically significant (Table 3). After adjusting for age,
sex, and follow-up years, compared to non-Aboriginal non-IDUs,
Aboriginal IDUs tended to have lower MHS scores (p=0.075) as did
non-Aboriginal IDUs (p=0.054); however, no variables reached statistical
significance in this model, and, after additionally adjusting for
current clinical status, both of these relationships lost strength and
significance (Table 3). When we used continuous, rather than
dichotomous, variables for CD4 cell count and [log.sub.10] viral load in
our second model, the parameter estimates for our exposure variables
remained similar (Aboriginal IDUs [-2.8; 95% CI -9.2, 3.6; p=0.39],
Aboriginal non-IDUs [-2.8; 95% CI -10.1, 4.5; p=0.45], and
non-Aboriginal IDUs [-3.8; 95% CI -10.4, 2.8; p=0.25]).
DISCUSSION
Aboriginal IDUs, Aboriginal non-IDUs, and non-Aboriginal IDUs
reported worse PHS scores compared to non-Aboriginal non-IDUs. These
differences appear to be partly, though not entirely, explained by
clinical status. Current CD4 cell count was the most significant
predictor of PHS in this study. Regarding mental health, compared to
non-Aboriginal non-IDUs, Aboriginal IDUs and non-Aboriginal IDUs tended
to have worse MHS, but these relationships weakened after adjusting for
current clinical status, especially for Aboriginal IDUs; Aboriginal
non-IDUs did not report statistically significantly worse MHS scores.
Overall, the variables assessed in our study were not as strongly
associated with MHS as they were with PHS.
Several factors may account for the differences in PHS scores we
observed. First, the poorer clinical status of Aboriginals and IDUs
appears to be partially responsible, especially for Aboriginal IDUs and
non-Aboriginal IDUs. This is demonstrated by our second model that
controlled for CD4 cell count and viral load: the difference in PHS for
these two groups weakened and became statistically non-significant.
Although the lack of statistical significance may be due in part to a
lack of power in our study, these findings support the importance of
clinical status as a determinant of PHS, which suggests that these
observed inequalities in physical HRQL may be diminished by helping to
improve patients' clinical status, for example, through improved
adherence to HAART.
Second, we did not assess current IDU or non-injection drug or
alcohol use at the time of the study, which could have contributed to
the poorer physical HRQL we observed among Aboriginals and IDUs. For
example, Aboriginals have been shown to have higher rates of alcohol
dependence/abuse compared to non-Aboriginals, (17) which may also be
true for the Aboriginal patients in our study, and HRQL has been shown
to be worse among individuals with alcohol dependence and HIV combined
compared to those with either condition separately. (18) Therefore,
future studies should measure ongoing injection and non-injection drug
and alcohol use to assess how it may impact HRQL in this population.
Clinical status (especially CD4 cell count >350 cells/uL) also
appeared to be an important independent predictor of PHS. Research
reports discrepant results with respect to the association between
clinical status and physical HRQL. Some studies have found no
significant relationship, (8,9) while others have found higher CD4 cell
counts associated with better physical HRQL. (6,7) In our study,
patients may have been aware of their clinical status before completing
the MOS-HIV and this knowledge could have influenced patients'
perceptions about their HRQL. However, it makes intuitive sense that
better clinical status would be associated with better HRQL.
In contrast to our results for PHS, although Aboriginal IDUs and
non-Aboriginal IDUs tended to have worse MHS scores than non-Aboriginal
non-IDUs, these relationships were not statistically significant and
weakened after adjusting for current clinical status. This suggests that
these differences are at least partly attributable to poorer clinical
status. On the other hand, this non-significance could also be due in
part to a lack of statistical power caused by our small sample size.
Although the results for MHS scores were not statistically significant,
the estimated MHS scores for Aboriginal IDUs, Aboriginal non-IDUs, and
non-Aboriginal IDUs were 3-6 units lower than for non-Aboriginal
non-IDUs in the multivariable models. Since the MHS scores are
standardized to a population with a mean of 50 and a SD of 10, these
differences correspond to approximately one third to one half the SD,
which could be interpreted as small to medium effect sizes. (19) Given
the dearth of research in this area and the over-representation of
Aboriginals in Canada's HIV epidemic, (10) more research is needed
to understand mental HRQL in this population.
Limitations
This study has several limitations. First, we had a small sample
size. Although we achieved an excellent response rate, with 84% of
patients approached by clinic staff agreeing to participate in the
study, only 28% of potentially eligible patients were approached,
creating a low rate of enrollment. One possible reason for this was the
busy clinic environment within which the study was conducted, which made
patient enrollment challenging. Nevertheless, despite the small sample
size of our study, these results provide an initial description of the
relationships investigated as a basis for future research. Second, a
clinical database, which has inherent limitations, was used to identify
eligible patients and provide data on exposure and confounding
variables. Data quality can be affected by data entry errors and
omissions. Using these data, we could not be certain of the date
patients started HAART or whether they were ART-naive when starting
HAART. Misclassifications may have occurred; for example, we may have
misclassified patients by categorizing individuals with unknown or
missing exposure categories as non-IDUs. Furthermore, we lacked data on
potentially important confounders, including use of and adherence to
HAART at the time of the interview, education, income, employment,
medical and psychiatric comorbid conditions, active substance use,
unstable housing, and social support which may influence HRQL. Future
studies should assess these variables, which could provide a more
comprehensive understanding of quality of life, and might help to better
explain the relationships investigated in this study. Last, the MOS-HIV
instrument showed some psychometric weaknesses in this patient
population. Several of the domains demonstrated ceiling and floor
effects. Further, among Aboriginals and IDUs, one of the MOS-HIV domains
(role functioning) had a Cronbach alpha score <0.70, which suggests
that, for these two groups, the internal consistency reliability for
this domain was less than satisfactory. Comparing different cultural
groups to each other does pose some difficulties. For example,
Aboriginals and non-Aboriginals may perceive quality of life
differently; therefore, these two groups may have responded to the
MOS-HIV questions differently. This may be a possible explanation for
the differences we observed in HRQL between Aboriginals and
non-Aboriginals.
CONCLUSIONS
In summary, we observed a difference in physical HRQL between the
groups investigated (Aboriginal IDUs, Aboriginal non-IDUs, and
non-Aboriginal IDUs vs. non-Aboriginal non-IDUs). These associations for
physical HRQL appear to be partially, though not completely, explained
by poorer clinical status, especially for Aboriginal IDUs, which
suggests that the observed inequalities may be diminished by improving
patients' clinical status; for example, through improving adherence
to HAART. Improving clinical status may help to alleviate these observed
inequalities; however, broader issues of social and economic
marginalization need to be addressed to achieve overall improvements in
these patients' lives.
Previous Presentations: This work was presented in part at the 18th
Annual Canadian Conference on HIV/AIDS Research, Vancouver, BC, April
23-26, 2009 (published abstract: Can j Infect Dis Med Microbiol
2009;20(Suppl B):40B) and was included as a chapter in L.J.
Martin's PhD thesis.
Received: June 5, 2012 Accepted: October 5, 2012
Acknowledgements: This study was funded by the Alberta Heritage
Foundation for Medical Research (AHFMR) Health Research Fund. L.J.
Martin was supported by an AHFMR full-time studentship and a Canadian
Institutes for Health Research Doctoral Research Award. We thank B.E.
Lee from the Provincial Public Health Laboratory for providing viral
load data; the NAHIVP clinic staff for helping to enrol patients and
assisting with study management; J. MacDonald for administrative
assistance; J. Kelecevic for interviewing patients and entering data;
and J. Bietz for entering data.
Conflict of Interest: None to declare.
REFERENCES
(1.) Martin LJ, Houston S, Yasui Y, Wild TC, Saunders LD. Rates of
initial virological suppression and subsequent virological failure after
initiating highly active antiretroviral therapy: The impact of
Aboriginal ethnicity and injection drug use. Curr HIV Res
2010;8(8):649-58.
(2.) Martin LJ, Houston S, Yasui Y, Wild TC, Saunders LD. All-cause
and HIV-related mortality rates among HIV-infected patients after
initiating highly active antiretroviral therapy: The impact of
Aboriginal ethnicity and injection drug use. Can J Public Health
2011;102(2):90-96.
(3.) Lima VD, Kretz P, Palepu A, Bonner S, Kerr T, Moore D, et al.
Aboriginal status is a prognostic factor for mortality among
antiretroviral naive HIV-positive individuals first initiating HAART.
AIDS Res Ther 2006;3:14.
(4.) Dybul M, Fauci AS, Bartlett JG, Kaplan JE, Pau AK. Guidelines
for using antiretroviral agents among HIV-infected adults and
adolescents. Recommendations of the Panel on Clinical Practices for
Treatment of HIV. MMWR Recomm Rep 2002;51(RR-7):1-55.
(5.) Brandson EK, Fernandes KA, Palmer A, Clement K, Olson B, Lima
VD, et al. HAART-full of life: Variations in quality of life among
Aboriginal and non-Aboriginal peoples ever on antiretroviral therapy in
BC. XVII International AIDS Conference. 2008 Aug 3-8; Mexico City,
Mexico. Abstract no. THPE0818. Available at:
http://www.iasociety.org/Default.aspx?pageId=11&abstractId=200718980
(Accessed March 6, 2012).
(6.) Call SA, Klapow JC, Stewart KE, Westfall AO, Mallinger AP,
Demasi RA, et al. Health-related quality of life and virologic outcomes
in an HIV clinic. Qual Life Res 2000;9(9):977-85.
(7.) Preau M, Protopopescu C, Spire B, Sobel A, Dellamonica P,
Moatti JP, et al. Health related quality of life among both current and
former injection drug users who are HIV-infected. Drug Alcohol Depend
2007;86(2-3):175-82.
(8.) Burgoyne RW, Saunders DS. Quality of life among urban Canadian
HIV/AIDS clinic outpatients. Int JSTD AIDS 2001;12(8):505-12.
(9.) Badia X, Podzamczer D, Garcia M, Lopez-Lavid CC, Consiglio E.
A randomized study comparing instruments for measuring health-related
quality of life in HIV-infected patients. AIDS 1999;13(13):1727-35.
(10.) Public Health Agency of Canada. HIV/AIDS Epi Updates,
November 2007. Ottawa, ON: Centre for Infectious Disease Prevention and
Control. Surveillance and Risk Assessment Division, PHAC, 2007.
(11.) Dalgard O, Egeland A, Skaug K, Vilimas K, Steen T.
Health-related quality of life in active injecting drug users with and
without chronic hepatitis C virus infection. Hepatology
2004;39(1):74-80.
(12.) Perez IR, Bano JR, Lopez Ruz MA, Jiminez AA, Prados MC, Liano
JP, et al. Health-related quality of life of patients with HIV: Impact
of sociodemographic, clinical and psychosocial factors. Qual Life Res
2005;14:1301-10.
(13.) Wu AW, Revicki DA, Jacobson D, Malitz FE. Evidence for
reliability, validity and usefulness of the Medical Outcomes Study HIV
Health Survey (MOS-HIV). Qual Life Res 1997;6(6):481-93.
(14.) Public Health Agency of Canada. HIV and AIDS in Canada.
Surveillance Report to December 31, 2006. Ottawa: Centre for Infectious
Disease Prevention and Control, Surveillance and Risk Assessment
Division, PHAC, 2007.
(15.) Wu AW. MOS-HIV Health Survey Scoring Guidelines. Waltham, MA:
Medical Outcomes Trust, 1997.
(16.) Dohoo I, Martin W, Stryhn H. Veterinary Epidemiologic
Research, 2nd ed. Charlottetown, PE: VER Inc., 2009.
(17.) Clarke DE, Colantonio A, Rhodes AE, Escobar M. Pathways to
suicidality across ethnic groups in Canadian adults: The possible role
of social stress. Psychol Med 2008;38(3):419-31.
(18.) Rosenbloom MJ, Sullivan EV, Sassoon SA, O'Reilly A, Fama
R, Kemper CA, et al. Alcoholism, HIV infection, and their comorbidity:
Factors affecting self-rated health-related quality of life. J Stud
Alcohol Drugs 2007;68(1):115-25.
(19.) Cohen J. Statistical Power Analysis for the Behavioral
Sciences, 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates, 1988.
Leah J. Martin, PhD, [1] Stan Houston, MD, FRCPC, [1,2] Yutaka
Yasui, PhD, [1] T. Cameron Wild, PhD, [3] L. Duncan Saunders, MBBCh, PhD
[1] (QUANTITATIVE RESEARCH)
Author Affiliations
[1.] School of Public Health, University of Alberta, Edmonton, AB
[2.] Division of Infectious Diseases, Department of Medicine,
University of Alberta, Edmonton, AB
[3.] Centre for Health Promotion Studies, School of Public Health,
University of Alberta, Edmonton, AB
Correspondence: Dr. Leah J. Martin, 4-057 Edmonton Clinic Health
Academy, 11405 87 Avenue, Edmonton, AB T6G 1C9, Tel: 780-492-7700,
E-mail: leah.martin@ualberta.ca
Table 1. Patient Characteristics by Ethnicity and Injection
Drug Use Exposure (N=96)
Characteristic Aboriginal Aboriginal
IDUs Non-IDUs
(n=21) (n=13)
Sex, no. (%)
Male 13 (62) 8 (62)
Female 8 (38) 5 (38)
Age at interview in years, 41.1 43.9
median (IQR) (39.0-47.1) (39.0-50.7)
HIV viral load at interview, 50 50
copies/mL, median (IQR) (50-44,000) (50-71)
>400 7 (33) 2 (15)
[less than or equal to]400 14 (67) 11 (85)
CD4 cell count at interview, 280 270
cells/[micro]L, median (IQR) (170-390) (190-350)
>350 7 (33) 3 (23)
[less than or equal to]350 14 (67) 10 (77)
Years since starting HAART,
median (IQR), total 4.0 (1.7-6.1) 6.1 (3.2-8.0)
PHS 89.3 70.7
Median (IQR) 43.8 (36.6-50.7) 39.9 (37.8-44.5)
Mean (SD) 44.4 (10.5) 40.6 (9.9)
MHS
Median (IQR) 48.3 (41.9-55.7) 51.1 (37.4-54.7)
Mean (SD) 47.7 (10.5) 47.5 (10.2)
Characteristic Non-Aboriginal
IDUs
(n=19)
Sex, no. (%)
Male 15 (79)
Female 4 (21)
Age at interview in years, 47.2
median (IQR) (39.1-49.8)
HIV viral load at interview, 50
copies/mL, median (IQR) (50-140)
>400 3 (16)
[less than or equal to]400 16 (84)
CD4 cell count at interview, 250
cells/[micro]L, median (IQR) (160-330)
>350 4 (21)
[less than or equal to]350 15 (79)
Years since starting HAART,
median (IQR), total 7.1 (3.8-8.9) 117.6
PHS
Median (IQR) 40.9 (29.6-53.6)
Mean (SD) 41.5 (12.4)
MHS
Median (IQR) 47.2 (36.5-56.8)
Mean (SD) 45.6 (12.8)
Characteristic Non-Aboriginal p-value
Non-IDUs
(n=43)
Sex, no. (%) 0.17
Male 36 (84)
Female 7 (16)
Age at interview in years, 46.0 0.39
median (IQR) (41.2-53.0)
HIV viral load at interview, 50 0.019
copies/mL, median (IQR) (50-50)
>400 3 (7.0) 0.058 *
[less than or equal to]400 40 (93)
CD4 cell count at interview, 440 0.002 *
cells/[micro]L, median (IQR) (280-580)
>350 28 (65) 0.002
[less than or equal to]350 15 (35)
Years since starting HAART,
median (IQR), total 3.4 (2.3-5.7) 185.6 0.11
PHS
Median (IQR) 55.4 (44.9-60.5) 0.001
Mean (SD) 51.7 (11.1)
MHS
Median (IQR) 56.1 (47.1-60.9) 0.031
Mean (SD) 53.0 (11.3)
* Fisher's exact test.
IDU=injection drug use, IQR=interquartile range, HAART=highly
active antiretroviral therapy, SD=standard deviation.
Table 2. Unadjusted and Adjusted Models Describing Physical
Health (PHS) Summary Scores (N=96)
Variable Unadjusted Model *
Parameter
Estimate 95% CI p-value
Aboriginal ethnicity and
IDU exposure status
Aboriginal IDUs -7.3 -13.1, -1.4 0.016
Aboriginal non-IDUs -11.0 -18.0, -4.1 0.002
Non-Aboriginal IDUs -10.1 -16.2, -4.1 0.001
Non-Aboriginal 0.0 - -
non-IDUs (ref)
Age at interview - - -
(years) ([dagger])
Sex
Female - - -
Male (ref) - - -
Follow-up yearst - - -
CD4 cell count
>350 cells/LiL - - -
[less than or equal - - -
to]350 cells/LiL
HIV viral load at interview
[less than or equal - - -
to]400 copies/mL
>400 copies/mL - - -
Variable Multivariable Model
(Excluding Clinical Variables) *
Parameter
Estimate 95% CI p-value
Aboriginal ethnicity and
IDU exposure status
Aboriginal IDUs -8.2 -14.3, -2.2 0.008
Aboriginal non-IDUs -11.4 -18.5, -4.2 0.002
Non-Aboriginal IDUs -10.2 -16.5, -3.9 0.002
Non-Aboriginal 0.0 - -
non-IDUs (ref)
Age at interview -0.2 -0.5, 0.03 0.082
(years) ([dagger])
Sex
Female 0.09 -5.3, 5.5 0.97
Male (ref) 0.0 - -
Follow-up yearst -0.2 -1.0, 0.7 0.68
CD4 cell count
>350 cells/LiL - - -
[less than or equal - - -
to]350 cells/LiL
HIV viral load at interview
[less than or equal - - -
to]400 copies/mL
>400 copies/mL - - -
Variable Multivariable Model
(Including Clinical Variables) *
Parameter
Estimate 95% CI p-value
Aboriginal ethnicity and
IDU exposure status
Aboriginal IDUs -4.7 -10.7, 1.3 0.12
Aboriginal non-IDUs -7.9 -14.9, -0.92 0.027
Non-Aboriginal IDUs -6.3 -12.6, -0.06 0.048
Non-Aboriginal 0.0 - -
non-IDUs (ref)
Age at interview -0.3 -0.5, 0.007 0.057
(years) ([dagger])
Sex
Female 0.6 -4.6, 5.8 0.82
Male (ref) 0.0 - -
Follow-up yearst -0.3 -1.1, 0.5 0.44
CD4 cell count
>350 cells/LiL 7.2 2.4, 11.9 0.003
[less than or equal 0.0 - -
to]350 cells/LiL
HIV viral load at interview
[less than or equal 5.2 -1.0, 11.4 0.099
to]400 copies/mL
>400 copies/mL 0.0 - -
* The intercepts for the unadjusted model, the multivariable model
excluding clinical variables, and the multivariable model including
clinical variables were: 51.7, 59.2, and 50.3, respectively.
([dagger]) Centred variables at 18 years for age at interview
and 1 year for follow-up years. IDU=injection drug use.
Table 3. Unadjusted and Adjusted Models Describing Mental
Health (MHS) Summary Scores (N=96)
Variable Unadjusted *
Parameter 95% CI p-value
Estimate
Aboriginal ethnicity and
IDU exposure status
Aboriginal IDUs -5.3 -11.3, 0.7 0.080
Aboriginal non-IDUs -5.5 -12.6, 1.6 0.13
Non-Aboriginal IDUs -7.5 -13.7, -1.3 0.019
Non-Aboriginal 0.0 - -
non-IDUs (ref)
Age at interview - - -
(years) ([dagge])
Sex
Female - - -
Male (ref) - - -
Follow-up years ([dagger]) - - -
CD4 cell count
>350 cells/[micro]L - - -
[less than or equal to] - - -
350 cells/[micro]L (ref)
HIV viral load at interview
[less than or equal - - -
to]400 copies/mL
>400 copies/mL (ref) - - -
Variable Multivariable Model
(Excluding Clinical Variables) *
Parameter 95% CI p-value
Estimate
Aboriginal ethnicity and
IDU exposure status
Aboriginal IDUs -5.6 -11.7, 0.6 0.075
Aboriginal non-IDUs -4.7 -12.0, 2.5 0.20
Non-Aboriginal IDUs -6.3 -12.7, 0.1 0.054
Non-Aboriginal 0.0 - -
non-IDUs (ref)
Age at interview -0.1 -0.4, 0.2 0.44
(years) ([dagge])
Sex
Female -0.9 -6.4, 4.6 0.74
Male (ref) 0.0 - -
Follow-up years ([dagger]) -0.7 -1.5, 0.1 0.10
CD4 cell count
>350 cells/[micro]L - - -
[less than or equal to] - - -
350 cells/[micro]L (ref)
HIV viral load at interview
[less than or equal - - -
to]400 copies/mL
>400 copies/mL (ref) - - -
Variable Multivariable Model
(Including Clinical Variables) *
Parameter 95% CI p-value
Estimate
Aboriginal ethnicity and
IDU exposure status
Aboriginal IDUs -3.4 -9.8, 3.1 0.30
Aboriginal non-IDUs -3.1 -10.5, 4.4 0.42
Non-Aboriginal IDUs -4.5 -11.2, 2.2 0.19
Non-Aboriginal 0.0 - -
non-IDUs (ref)
Age at interview -0.1 -0.4, 0.2 0.39
(years) ([dagge])
Sex
Female -1.0 -6.6, 4.5 0.71
Male (ref) 0.0 - -
Follow-up years ([dagger]) -0.8 -1.6, 0.06 0.068
CD4 cell count
>350 cells/[micro]L 2.7 -2.4, 7.8 0.29
[less than or equal to] 0.0 - -
350 cells/[micro]L (ref)
HIV viral load at interview
[less than or equal 5.2 -1.4, 11.9 0.12
to]400 copies/mL
>400 copies/mL (ref) 0.0 - -
* The intercepts for the unadjusted model, the multivariable model
excluding clinical variables, and the multivariable model including
clinical variables were: 53.0, 58.6, and 52.6, respectively.
([dagger]) Centred variables at 18 years for age at interview and 1
year for follow-up years. IDU=injection drug use.
