Demographic,Psychological,and Behavioral Modifiers of the Antiretroviral Treatment Access Study (ARTAS) Intervention

Abstract

The present study sought to identify demographic, structural, behavioral, and psychological subgroups for which the Antiretroviral Treatment Access Study (ARTAS) intervention had stronger or weaker effects in linking recently diagnosed HIV-positive persons to medical care. The study, carried out from 2001 to 2003, randomized 316 participants to receive either passive referral or a strengths-based linkage intervention to facilitate entry into HIV primary care. The outcome was attending at least one HIV primary care visit in each of two consecutive 6-month periods. Participants (71% male; 29% Hispanic; 57% black non-Hispanic), were recruited from sexually transmitted disease clinics, hospitals and community-based organizations in four U.S. cities. Thirteen effect modifier variables measured at baseline were examined. Subgroup differences were formally tested with interaction terms in unadjusted and adjusted log-linear regression models. Eighty-six percent (273/316) of participants had complete 12-month follow-up data. The intervention significantly improved linkage to care in 12 of 26 subgroups. In multivariate analysis of effect modification, the intervention was significantly (p < 0.05) stronger among Hispanics than other racial/ethnic groups combined, stronger among those with unstable than stable housing, and stronger among those who were not experiencing depressive symptoms compared to those who were. The ARTAS linkage intervention was successful in many but not all subgroups of persons recently diagnosed with HIV infection. For three variables, the intervention effect was significantly stronger in one subgroup compared to the counterpart subgroup. To increase its scope, the intervention may need to be tailored to the specific needs of groups that did not respond well to the intervention.

Introduction

L inking persons newly diagnosed with HIV infection to medical care has important personal and public health benefits. HIV-positive persons who enter medical care early and remain in care have the best chance for long-term suppression of the virus¹ and increased survival and quality of life.² Equally important, HIV-positive people with suppressed viral loads are less likely to transmit the virus to others through risky behaviors.^3,4

In 2005, we published the results of the Antiretroviral Treatment Access Study (ARTAS).⁵ ARTAS was a two-arm randomized controlled trial that compared a brief strengths-based case management intervention with “passive referral” in linking recently diagnosed HIV-positive persons to primary medical care. Participants in the intervention arm received up to five sessions with a case manager over 90 days to facilitate linkage. Participants in the passive referral arm (control arm) received information about HIV and local care resources, as well as a referral to a local HIV medical care provider. Results indicated that those who received the intervention were significantly more likely than those who received passive referral to have attended at least one HIV primary care visit in each of two consecutive 6-month follow-up periods.

Since the publication of ARTAS, we frequently have been asked whether the intervention had stronger effects in some subgroups of participants compared with other subgroups. In essence, the question concerns whether there are variables that significantly modify the strength of the linkage intervention. The ARTAS trial measured several demographic, behavioral, and psychological variables at baseline. A systematic analysis of subgroups of those variables may illuminate the conditions under which the intervention was especially effective. Equally important, identifying the subgroups in which the intervention was not successful may provide clues as to how the intervention may need to be adapted to do better with these subgroups.

Herein, we examine an array of demographic (e.g., age, gender, ethnicity, education), structural (housing stability), behavioral (e.g., illicit drug use, emergency room use, having assistance from family or friends in seeking care) and psychological (e.g., depressive symptoms, stage of readiness to enter care, beliefs and attitudes about HIV care and treatment) variables as potential modifiers of the ARTAS intervention. An important difference between the current investigation and the relevant literature is that prior studies^{6

–18} have examined many of these variables as correlates of engagement in care but have not examined how specific subgroups of individuals respond to a linkage intervention. Prior studies of persons diagnosed with HIV infection have shown that utilization of health care services is less likely among drug users,^18

–24 those unstably housed,^14,25
–27 those with depression,^28,29 and those who are not psychologically ready to enter care.³⁰ We examined whether these subgroups, as well as others, were less responsive to the ARTAS linkage intervention.

Methods

ARTAS was conducted from March 2001 to August 2003 in Atlanta, Georgia; Baltimore, Maryland; Los Angeles, California; and Miami, Florida. Participants were recruited from health department testing centers, sexually transmitted disease (STD) clinics, hospitals, and community-based organizations.⁵ The study received approval from the Institutional Review Boards at the Centers for Disease Control and Prevention and at each local study site. Eligibility criteria for enrollment included having been diagnosed with HIV, not having previously visited an HIV care provider more than once, not currently taking HIV antiretroviral medications, being 18 years of age or older, and being able to provide informed consent.

Questionnaire data were collected from participants at enrollment (baseline) and again at approximately 6 and 12 months after the baseline assessment. The data were collected with an Audio Computer-Assisted Self-Interview (ACASI). The questions were available in both English and Spanish.

The primary outcome variable was self-reported attendance at an HIV primary care provider (i.e., physician, nurse practitioner, or physician assistant). Being “in care” was defined as having attended at least one HIV primary care visit in each of two consecutive 6-month follow-up periods. All participants completing 6- and 12-month interviews signed releases permitting medical records of clinic attendance to be abstracted. These data were used to assess the validity of the respondent's self-reports of entering HIV primary care. Of the 177 participants who reported that they received care between baseline and 6 months, 164 (93%) could be confirmed by medical records. Of the 192 participants who reported that they received care between 6 and 12 months, 165 (86%) were confirmed through medical records.

The variables examined as potential effect modifiers were measured at baseline (before the ARTAS intervention). Table 1 displays the variables along with their response options and cut-points that defined the subgroups used in the analysis.

The ARTAS linkage case management intervention has been described in previous publications.^{5,20,30
–32} The intervention provided intensive, short-term assistance to facilitate the process of linking HIV-positive individuals to HIV primary medical care. The intervention used a strengths-based approach that calls on clients to identify internal strengths and abilities and to develop a personal plan to acquire needed resources. Case management sessions focused on building a relationship with the client, identifying and addressing client strengths and needs, identifying strategies to overcome personal or system barriers to health care, and encouraging contact with an HIV medical care provider. The intervention consisted of up to five sessions with the case manager over a 90-day period. The ARTAS linkage case managers did not have access to participants' ACASI responses.

Statistical analysis

For purpose of analysis, each of the effect modifier variables was divided into two subgroups. We examined the subgroup-specific relative risk (RR) of being successfully linked to HIV primary care among participants in the intervention arm compared with those in the control arm. A higher RR value indicates a stronger intervention effect. We quantified the importance of subgroup-specific differences in the RR by testing the interaction between trial arm and a modifier variable. The interaction effect yields information on whether the RR in one subgroup is significantly (p < 0.05) larger than the RR in the other subgroup. Each effect modifier candidate was first examined separately in a univariate analysis. The univariate model included two main effect terms (trial arm and modifier variable) and a single two-way interaction term (trial arm x modifier variable). Interaction terms that reached a significance level of p < 0.20 in the univariate analyses were simultaneously included in a multivariate model along with corresponding main-effect terms. Thus, in the multivariate model, the finding for each interaction term was adjusted for the presence of the other interaction terms. These analyses were done using a log-linear analysis to estimate relative risks directly. The log-linear model was fit in PROC GENMOD using Statistical Analysis System (SAS) software (SAS version 9.1, SAS Institute, Inc., Cary NC).

Results

A total of 316 persons were enrolled and completed the baseline assessment; 273 (86%) of these participants had complete outcome data at the 12-month follow-up and are the focus of the analysis. Seventy-four percent of the 273 participants had not visited an HIV care provider before enrolling in the study; 26% had one prior visit (the distribution was highly similar in each arm). Seventy-seven percent had been diagnosed with HIV infection no longer than six months prior to enrollment. Most (71%) of the participants were men, 63% were between the ages of 18–39, 57% were non-Hispanic black, 8% non-Hispanic white, 29% Hispanic, and 6% other race/ethnicity. Table 2 shows the baseline characteristics on the 13 effect modifier variables among participants overall and for each trial arm. None of these variables differed significantly by arm.

Table 1.

Effect Modification Variables and Binary Categories Used in the Analysis, ARTAS Study, 2001–2003

Effect modification variables (assessed at baseline)	Binary analysis categories
Gender	Male vs. female
Age	18–39 vs. 40+ (The ARTAS trial results,⁵ found individuals age 40+ were significantly more likely than younger persons to be linked to HIV care.)
Race/ethnicity	Hispanic vs. non-Hispanic white, non-Hispanic black, or other (Asian, Pacific Islander) (The ARTAS trial results⁵ found Hispanic individuals were significantly more likely than others to be linked to HIV care.)
Education	< High school graduate vs. high school graduate or greater
Time from since being diagnosed HIV positive to study entry	≤ 6 months vs. > 6 months
Housing	Unstable vs. stable
Unstable housing included participants who reported that they were in temporary housing (a shelter), on the street, or in an institution (a mental health facility, a drug treatment facility, a half-way house or group home, hospital or nursing home).
Stable housing included participants who reported that they were living in a house or apartment alone or with a spouse, partner, friend, family, or roommate.
Usual source of medical care	Emergency department or no usual source of care vs. other sources
Reported that they used emergency room or said no usual source of care versus any of the following other sources: private doctor's office/private clinic, public/county health clinic, veterans administration clinic, health maintenance organization, company or school health clinic.
Symptoms of depression (CES-D)	Symptoms of depression vs. no symptoms of depression
Participants with symptoms of depression are those scoring 16 or more on the 20-item CES-D scale (scored from 0 to 3). This cut-point has traditionally been used as the “caseness” level indicating possible depression.
Recent drug use	Yes vs. no
Using crack cocaine or injecting an illicit drug in the past 30 days.
Help from others in obtaining HIV medical care	Some help from at least one person vs. no help from anyone
Since testing HIV positive, have family, friends or non-ARTAS counselors, social workers, case managers or other professionals been helping you get into HIV care? Responses were tangible forms of help (explained HIV; explained where to go; helped with transportation; offered to accompany to clinic; helped with clinic paperwork).
Stages of readiness to enter HIV care	Precontemplation or contemplation vs. preparation or action
Participants were asked: “Since testing HIV positive, which of the following best describes your current feelings?” Participants selected one response.
Precontemplation
I am not thinking about HIV medical care at this time.
HIV medical care is important to me but I am not ready to start care yet.
Contemplation
I have thought about starting HIV medical care but I have not yet tried to find a doctor or clinic.
I have found a doctor or clinic for HIV medical care but have not yet tried to make an appointment.
Preparation
I have tried to obtain HIV care from a doctor or clinic but have not been successful yet.
I have an appointment for HIV care with a doctor or clinic but have not been there yet.
Action
I have already gone to a doctor or clinic for HIV care once.
Attitudes and beliefs about HIV treatment and medical care	A median split on the distribution of the overall score on the 12-item scale, with the “positive” category including scores above the median and the “negative” category including scores below the median
A 12-item scalea reflecting attitudes and beliefs about HIV treatment and medical care. Each item was scored on a 4-point Likert scale.
Felt well	Yes vs. no
This was a response to a question asking participants whether “I felt well” was a reason they had not yet made or kept an appointment for HIV care at time of baseline survey.

Example items from the scale: HIV can kill you if left untreated; I do not need medical care and HIV medicines until I get very sick; missing doses of my HIV medicine will not harm my health; I do not want to start taking HIV medicines because they will do me more harm than good; the government hasn't tested the new HIV medicines long enough to be sure they are safe.

CES-D, Center for Epidemiologic Studies Depression Scale.

Table 2.

Baseline Characteristics for Effect Modifier Variables among Participants Overall and by Intervention vs. Control Arm, ARTAS Study, 2001–2003

Effect modification variables	Overall sample (N = 273) n (%)	Intervention arm (N = 137) n (%)	Control arm (N = 136) n (%)	p value comparing arms
Gender
Male	194 (71.1)	95 (69.9)	99 (72.3)	0.66
Female	79 (28.9)	41 (30.1)	38 (27.7)
Age
40+	101 (37.0)	49 (36.0)	52 (38.0)	0.74
18–39	172 (63.0)	87 (64.0)	85 (62.0)
Ethnicity
Hispanic	80 (29.3)	42 (30.9)	38 (27.7)	0.57
all other ethnicities	193 (70.7)	94 (69.1)	99 (72.3)
Education
High school grad or greater	126 (46.2)	67 (49.3)	59 (43.1)	0.30
< High school graduate	147 (53.8)	69 (50.7)	78 (56.9)
Time from HIV+ diagnosis to study entry
≤ 6 months	210 (76.9)	101 (74.3)	109 (79.6)	0.30
> 6 months	63 (23.1)	35 (25.7)	28 (20.4)
Housing
Stable	211 (77.3)	102 (75.0)	109 (79.6)	0.37
Unstable	62 (22.7)	34 (25.0)	28 (20.4)
Usual source of care
Emergency department or no usual source	40 (14.7)	21 (15.4)	19 (13.9)	0.71
All others	233 (85.3)	115 (84.6)	118 (86.1)
Depressive symptomsa
No	139 (52.9)	74 (55.2)	65 (50.4)	0.43
Yes	124 (47.2)	60 (44.8)	64 (49.6)
Recent drugb use
No	202 (74.0)	104 (76.5)	98 (71.5)	0.35
Yes	71 (26.0)	32 (23.5)	39 (28.5)
Help received from others in obtaining medical care for HIV
Some help	53 (19.4)	22 (16.2)	31 (22.6)	0.18
No help from anyone	220 (80.6)	114 (83.8)	106 (77.4)
Stages of readiness to enter HIV care
Preparation or action	143 (52.4)	67 (49.3)	76 (55.5)	0.30
Precontemplation or contemplation	130 (47.6)	69 (50.7)	61 (44.5)
Beliefs and attitudes about HIV treatment and medical carec
Positive	135 (49.6)	65 (47.8)	70 (51.5)	0.54
Negative	137 (50.4)	71 (52.2)	66 (48.5)
Felt well
Yes	74 (27.1)	36 (26.5)	38 (27.7)	0.81
No	199 (72.9)	100 (73.5)	99 (72.3)

Participants classified as having depressive symptoms are those who scored 16 or higher on the CES-D scale.

use in the past 30 days of crack cocaine or injection drugs.

Median split of index comprised of 12 items rated on a 4-point Likert scale. The “positive” category includes scores above the median and the “negative” category includes scores below the median.

Participants in control arm received passive referral only.

CES-D, Center for Epidemiologic Studies Depression Scale.

Overall, 154 of the 273 participants (56%) reported at least one visit to an HIV medical care provider in each of two consecutive six-month periods. There was a significant intervention effect. Sixty-four percent of participants who received the linkage intervention compared with 49% of participants who received passive referral visited an HIV primary care clinician at least once in each of two consecutive six-month periods (RR 1.41, 95% confidence interval [CI] 1.1, 1.6, p = 0.006).

As seen in Table 3, the intervention significantly improved linkage to care (the 95% confidence interval did not include an RR value of 1.00) in 12 of the 26 subgroups examined here: males; Hispanic persons; those with lesser education; those within 6 months of their HIV diagnosis; the unstably housed; those who reported no usual source or care or usually used the ER; those without depressive symptoms; those who did not use crack cocaine or inject drugs in the past 30 days; those who had not received help from others (e.g., family, friends, others) in obtaining care for HIV; those in the preparation or action stages; those with positive beliefs and attitudes about HIV treatment and care; and those who did not respond that feeling well was a reason for not seeking HIV care. In 10 other subgroups (Table 3), the intervention improved linkage somewhat but fell short of statistical significance. There were only four subgroups in which the linkage rate was virtually identical in the intervention and control arms (those with higher education, those who had symptoms of depression, those who used crack cocaine or injected drugs recently, and those who had help from others in obtaining care).

Table 3.

Relative Risk of Being Successfully Linked to HIV Primary Care a among Participants in the Intervention vs. Control Arm within Subgroups of Effect Modification Variables, ARTAS Study, 2001–2003

	n/N (%) linked to HIV primary care N = 273
Effect modification variables (measured at baseline)	Intervention arm	Control arm	Unadjusted subgroup-specific RR b (95% CI)	p value of interaction unadjusted	Adjusted subgroup-specific RR b (95% CI) N = 263	p value of interaction adjusted
Gender
Male	65/95 (68.4)	49/99 (49.5)	1.38 (1.09, 1.76)	0.537	—
Female	22/41 (53.7)	18/38 (47.4)	1.13 (0.71, 1.71)
Age
40+	34/49 (69.4)	26/52 (50.0)	1.39 (1.00, 1.92)	0.741	—
18–39	53/67 (60.9)	41/85 (48.2)	1.26 (0.96, 1.67)
Ethnicity
Hispanic	39/42 (92.9)	23/38 (60.5)	1.53 (1.17, 2.00)	0.157	2.16 (1.40, 3.35)	0.015
all other ethnicities	48/94 (51.1)	44/99 (44.4)	1.15 (0.85, 1.54)		1.30 (0.88, 1.90)
Education
High school grad or greater	42/67 (62.7)	36/59 (61.0)	1.03 (0.78, 1.35)	0.031	1.43 (0.94, 2.16)	0.124
< High school graduate	45/69 (65.2)	31/78 (39.7)	1.64 (1.19, 2.27)		1.97 (1.31, 2.96)
Time from HIV+ diagnosis to study entry
≤ 6 months	73/101 (72.3)	59/109 (54.1)	1.34 (1.08, 1.65)	0.90	—
> 6 months	14/35 (40.0)	8/28 (28.6)	1.40 (0.69, 2.85)
Housing
Stable	66/102 (64.7)	60/109 (55.1)	1.18 (0.94, 1.47)	0.046	1.10 (0.87, 1.39)	0.010
Unstable	21/34 (61.8)	7/28 (25.0)	2.47 (1.23, 4.95)		2.54 (1.30, 4.97)
Usual source of care
Emergency department or no usual source	59/89 (64.0)	37/84 (44.0)	1.45 (1.09, 1.93)	0.250	—
All others	30/47 (63.8)	30/53 (56.6)	1.13 (0.82, 1.56)
Depressive symptomsc
No	53/74 (71.6)	30/65 (46.2)	1.55 (1.15, 2.09)	0.052	2.30 (1.51, 3.51)	0.002
Yes	33/60 (55.0)	35/64 (54.7)	1.01 (0.73, 1.38)		1.22 (0.82, 1.81)
Recent drugd use
No	75/104 (72.1)	53/98 (54.1)	1.33 (1.07, 1.66)	0.462	—
Yes	12/32 (37.5)	14/39 (35.9)	1.04 (0.57, 1.93)
Help received from others in obtaining medical care for HIV
Some help	16/22 (72.7)	22/31 (71.0)	1.02 (0.73, 1.44)	0.102	1.47 (0.95, 2.27)	0.233
No help from anyone	71/114 (62.3)	45/106 (42.6)	1.47 (1.13, 1.91)		1.91 (1.29, 2.82)
Stages of readiness to enter HIV care
Preparation or action	50/67 (74.6)	42/76 (55.3)	1.35 (1.06, 1.72)	0.890	—
Precontemplation or contemplation	37/69 (53.6)	25/61 (41.0)	1.31 (0.90, 1.90)
Beliefs and attitudes about HIV treatment and medical caree
Positive	48/65 (73.9)	37/70 (52.9)	1.40 (1.07, 1.82)	0.507	—
Negative	39/71 (54.9)	30/66 (45.5)	1.21 (0.86, 1.69)
Felt well
Yes	17/36 (47.2)	12/38 (31.6)	1.50 (0.84, 2.68)	0.589	—
No	70/100 (70.0)	55/99 (55.6)	1.26 (1.01, 1.57)

Attended at least one HIV primary care visit in each of two consecutive 6-month follow-up periods.

A higher relative risk (RR) value indicates a stronger intervention effect.

Participants classified as having depressive symptoms are those who scored 16 or higher on the CES-D scale.

use in the past 30 days of crack cocaine or injection drugs.

Median split of index comprising 12 items rated on a 4-point Likert scale. The “positive” category includes scores above the median and the “negative” category includes scores below the median.

Participants in control arm received passive referral only.

CES-D, Center for Epidemiologic Studies Depression Scale.

Formal tests of effect modification were conducted with interaction analyses. Five variables showed some signs of effect modification in that their interaction terms reached p < 0.20 in unadjusted analysis (Table 3).

Specifically, the linkage intervention tended to be stronger among Hispanics than among the other ethnic groups combined (RR = 1.53 versus 1.15, p = 0.157), stronger in the group of participants with less than a high school education compared with those with more education (RR = 1.64 versus 1.03, p = 0.031), stronger among those with unstable than stable housing (RR = 2.47 versus 1.18, p = 0.046), stronger among those without depressive symptoms compared with those with symptoms of depression (RR = 1.55 versus 1.01, p =0.052), and stronger among those who had not received help from others in obtaining HIV medical care compared with those who had help from others (RR = 1.47 versus 1.02, p = 0.102).

These five variables were examined in a multivariate log-linear model that included six main-effect terms (counting the trial arm variable) and five two-way interaction terms. Three interactions were significant (p < 0.05; last column of Table 3). The intervention effect was significantly stronger among Hispanics than other ethnic groups (adjusted RR = 2.16 versus 1.30, p = 0.015), stronger among those with unstable than stable housing (adjusted RR = 2.54 versus 1.10, p = 0.01), and stronger among those who were not experiencing depressive symptoms compared with those who were (adjusted RR =2.30 versus 1.22, p = 0.002). Furthermore, of these three variables with statistically significant effect modifications (interactions), all maintained relative differences between the subgroup-specific RRs of 40% to 47%. For education and “help from others,” the relative differences between subgroup RRs fell below 30%.

Discussion

We found that the ARTAS intervention significantly improved linkage to care (compared with passive referral) in 12 of 26 subgroups of persons recently diagnosed with HIV infection. In 10 other subgroups, the intervention improved linkage somewhat but fell short of statistical significance. There were only four subgroups in which the intervention fell completely flat. Taken as a whole, the intervention had relatively broad reach.

Our formal multivariate analysis of effect modification identified three variables in which the intervention was significantly more likely to affect the behavior of persons in one subgroup compared with persons in the counterpart subgroup. Those variables included (1) symptoms of depression, (2) housing stability, and (3) race/ethnicity.

Of the psychological variables we examined, participants without signs of depression had a much stronger response to the intervention than did participants with possible depression (based on their scores on the Center for Epidemiologic Studies Depression Scale [CES-D] scale). This suggests that positive psychological states of participants may play a role in the success of the ARTAS intervention, which used a strengths-based approach that attempted to empower participants and heighten their self-efficacy to take the steps necessary to enter care. Participants who were experiencing depression or psychological distress may not have been as receptive to that kind of intervention.

The other two psychological variables we examined were not observed to be effect modifiers. The intervention effect was significant but only slightly stronger among participants who had positive (above the median) versus less positive (below the median) attitudes about HIV treatment and care. The distribution of scores on the attitude index from which the two subgroups were formed was highly skewed; relatively few participants (<10% of the overall sample) had negative attitudes (i.e., a sum score ≤24 out of a total of 48). Had there been more variability, this attitude variable might have produced stronger effect modification. Participants' stages of psychological readiness to enter care also did not modify the intervention effect. The effect was approximately equally strong among those who reported at baseline that they were in the “action” or “preparatory” stages compared with those in the “precontemplation” or “contemplation” stages at baseline. It is possible that the intervention increased participants' readiness to enter care among the initial precontemplators and contemplators which, in turn, promoted entry into care, but we could not assess this possibility.

The intervention had a stronger effect (either significant or a trend) in subgroups with an unmet need or barrier—unstable housing, low education, lack of assistance from others in obtaining care. Among participants in these three subgroups, the intervention increased the percentage linked to HIV primary care compared with participants in these subgroups who were in the control arm. We do not have data to describe the intervention process with these subgroups, but the success of the intervention may have hinged on the receptiveness of the client, the motivation or concrete assistance of the linkage case manager, or other interpersonal factors. We should note, however, that drug users (used crack cocaine or injected drugs in past 30 days) were not helped by the intervention. This may have been due to lack of responsiveness of drug-using participants, the design of the intervention (e.g., brief duration), or both aspects. Our findings for drug users are consistent with other studies suggesting that drug users may be less responsive than nonusers to helping situations.^19

–24 In future research, it may be useful to collect intervention process data from clients and linkage case managers to help identify interpersonal and other factors contributing to successful and unsuccessful intervention effects.

Hispanic participants had the highest marginal rate of linkage to care (78%) and highest rate in the intervention arm (92%) of all the subgroups in this study. We believe that some of the high marginal rate and elevated Hispanic intervention effect may be due to the colocation of the HIV testing site and HIV care. Many of the Hispanics were recruited in a Hispanic section of Los Angeles and received their HIV-positive test results at a nearby HIV care facility. Other unmeasured variables also may have contributed to the Hispanics' response to the intervention (e.g., cultural attitudes about the importance of caring for oneself). Colocation of HIV testing and HIV care facilities was also found to contribute to significantly higher rates of linkage to care in the ARTAS demonstration study²⁰ that was performed after the ARTAS randomized trial.

Caution is warranted in interpreting the findings. Our analysis was largely exploratory in nature. We did not have a priori hypotheses as to which subgroups might have the strongest response to the ARTAS linkage intervention. Furthermore, we had a relatively small sample for the analysis. The total analytic sample for the ARTAS intervention trial was 273 participants. Half of these were allocated to the intervention arm and half to the control arm, and within each arm we divided participants by the subgroups in order to examine for effect modification. The small cell sizes, and the conservative confidence intervals from the log-linear models, may have reduced statistical power to detect significant differences in the strength of the intervention by subgroup.

Despite the small sample, we were able to show that many subgroups of recently diagnosed HIV-positive persons responded well to the ARTAS linkage intervention. We also identified, in a preliminary way, subgroups that did not respond as well to the intervention (e.g., those who showed signs of depression, drug users). This is perhaps the most useful data from our analysis, because it identifies individuals who may need more than just a brief strengths-based intervention to link them to care. These individuals may need more intensive case management and referral to social, mental health, and drug treatment services. It was encouraging that the ARTAS intervention had strong effects in disadvantaged groups (e.g., those with low education) and those with barriers to care entry in their lives (e.g., the unstably housed, those who did not have others besides the interventionist helping them enter care). Future research that attempts to refine the ARTAS model and to understand the dynamics of successful case manager–client relationships may help further broaden the reach of the ARTAS intervention and thereby increase the number of HIV-positive persons who enter care soon after their diagnosis.

The ARTAS Study Group

The ARTAS study group consists of Dr. Lytt I. Gardner, Dr. Scott D. Holmberg, and Christopher S. Krawczyk (the Centers for Disease Control and Prevention, Atlanta, GA); Dr. Carlos del Rio, Sonya Green, Maribel Barragan, Dr. Wayne Duffus, Dr. Michael Leonard, Christine O'Daniels, Catherine Abrams, Felicia Berry, Valerie Hunter, and Howard Pope (Emory University School of Medicine, Division of Infectious Diseases, Atlanta, GA); Dr. Steffanie Strathdee, Dr. Anita Loughlin, Steven Huettner, Matt Woolf, Ovedia Burt, and Janet Reaves (Johns Hopkins Bloomberg School of Public Health, Baltimore, MD); Dr. Pamela Anderson-Mahoney, Dr. Peter Kerndt, Bobby Gatson, Brandon Schmidt, Norma Perez, Amy Chan, Lawrence Fernandez, Jr., Jenny Kotlerman, Dr. Ellen Rudy, and Stella Gutierrez (Health Research Association, Los Angeles, CA); Dr. Lisa Metsch, Dr. Clyde McCoy, Dr. Gordon Dickinson, Dr. Toye Brewer, Eduardo Valverde, Brad Wohler-Torres, Lauren Gooden, Dr. Wei Zhao, Faye Yeomans, and Yolanda Davis Camacho (University of Miami School of Medicine, Miami, FL); Dr. Harvey A. Siegel, Richard C. Rapp, Teri L. Rust, and Dr. Jichuan Wang (Wright State University School of Medicine, Dayton, OH).

Footnotes

Acknowledgments

The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.

Author Disclosure Statement

No competing financial interests exist.

References

Giordano

, Gifford

, White

et al. Retention in care: A challenge to survival with HIV infection. Clin Infect Dis, 2007; 44:1493–1499.

Bucciardini

, Fragola

, Massella

et al. Health-related quality of life outcomes in HIV-infected patients starting different combination regimens in a randomized multinational trial: The INITIO-QoL substudy. AIDS Res Hum Retroviruses, 2007; 23:1215–1222.

Quinn

, Wawer

, Sewankambo

. et al. Viral load and heterosexual transmission of human immunodeficiency virus type 1. Rakai project study group. N Engl J Med, 2000; 342:921–929.

Tovanabutra

, Robison

, Wongtrakul

et al. Male viral load and heterosexual transmission of HIV-1 subtype E in northern Thailand. J Acquire Immune Defic Syndr, 2002; 29:275–283.

Gardner

, Metsch

, Mahoney

et al. Efficacy of a brief case management intervention to link recently diagnosed HIV-infected persons to care. AIDS, 2005; 19:423–431.

Molitor

, Waltermeyer

, Mendoza

et al. Locating and linking to medical care HIV-positive persons without a history of care: Findings from the California Bridge Project. AIDS Care, 2006; 18:456–459.

Mugavero

, Lin

, Allison

et al. Failure to establish HIV care: Characterizing the “no show” phenomenon. Clin Infect Dis, 2007; 45:127–130.

Giordano

, Visnegarwala

, White

et al. Patients referred to an urban HIV clinic frequently fail to establish care: Factors predicting failure. AIDS Care, 2005; 17:773–783.

Israelski

, Gore-Felton

, Power

, Wood

, Koopman

. Sociodemographic characteristics associated with medical appointment adherence among HIV-seropositive patients seeking treatment in a county outpatient facility. Prev Med, 2001; 33:470–475.

10.

Palacio

, Shiboski

, Yelin

, Hessol

, Greenblatt

. Access to and utilization of primary care sevices among HIV-infected women. J Acquire Immune Defic Syndr, 1999; 21:293–300.

11.

Samet

, Freedberg

, Savetsky

, Sullivan

, Padmanabhan

, Stein

. Discontinuation from HIV medical care: Squandering treatment opportunities. J Health Care Poor Underserved, 2003; 14:244–255.

12.

Shapiro

, Morton

, McCaffrey

et al. Variations in the care of HIV-infected adults in the United States. JAMA, 1999; 281:2305–2315.

13.

Arno

, Bonuck

, Green

et al. The impact of housing status on health care utilization among persons with HIV disease. J Health Care Poor Undeserved, 1996; 7:36–49.

14.

Aidala

, Lee

, Abramson

, Messeri

, Siegler

. Housing need, housing assistance and connection to HIV medical care. AIDS Behav, 2007; 11,Suppl 2:S101–S115.

15.

Knowlton

, Hua

, Latkin

. Social support networks and medical service use among HIV-positive injection drug users: Implications to intervention. AIDS Care, 2005; 17:479–492.

16.

Tobias

, Cunningham

, Cabral

et al. Living with HIV but without medical care: Barriers to engagement. AIDS Patient Care STDs, 2007; 21:426–434.

17.

Cunningham

, Sohler

, Wong

et al. Utilization of health care services in hard-to-reach marginalized HIV-infected individuals. AIDS Patient Care STDs, 2007; 21:177–186.

18.

Wilkinson

, Zhao

, Arnsten

et al. Longitudinal correlates of health care-seeking behaviors among HIV-seropositive injection drug users. J Acquire Immune Defic Syndr, 2007; 46,Suppl 2:S120–S126.

19.

Redko

, Rapp

, Elms

, Snyder

, Carlson

. Understanding the working alliance between persons with substance abuse problems and strengths-based case managers. J Psychoactive Drugs, 2007; 39:241–250.

20.

Craw

, Gardner

, Marks

. et al. Brief strengths-based case management promotes entry into HIV medical care. J Acquire Immune Defic Syndr, 2008; 47:597–606.

21.

Strathdee

, Palepu

, Cornelisse

et al. Barriers to use of free antiretroviral therapy in injection drug users. JAMA, 1998; 280:547–549.

22.

Cunningham

, Sohler

, Berg

, Shapiro

, Heller

. Type of substance use and access to HIV-related health care. AIDS Patient Care STDs, 2006; 20:399–407.

23.

Sohler

, Wong

, Cunningham

, Drainoni

, Cunningham

. Type and pattern of illicit drug use and access to health care services for HIV-infected people. AIDS Patient Care STDs, 2007; 21,Suppl 1:S68–S76.

24.

Mizuno

, Wilkinson

, Santibanez

. et al. Correlates of health care utilization among HIV-seropositive injection drug users. AIDS Care, 2006; 18:417–425.

25.

Rumptz

, Tobias

, Rajabiun

et al. Factors associated with engaging socially marginalized HIV-positive persons in primary care. AIDS Patient Care STDs, 2007; 21,Suppl 1:S30–S39.

26.

Smith

, Rapkin

, Winkel

, Springer

, Chhabra

, Feldman

. Housing status and health care service utilization among low-income persons with HIV/AIDS. J Gen Intern Med, 2000; 15:731–738.

27.

Gordon

, McGinnis

. Associations between alcohol use and homelessness with healthcare utilization among human immuno virus-infected veterans. Med Care, 2006; 44,Suppl 2:S37–S43.

28.

Basta

, Shacham

, Reece

. Psychological distress and engagement in HIV-related services among individuals seeking health care. AIDS Care, 2008; 20:969–976.

29.

Bodenlos

, Brothe

, Whitehead

, Konkle-Parker

, Jones

, Brantley

. Attitudes toward health care providers and appointment attendance in HIV/AIDS patients. J Assoc Nurses AIDS Care, 2007; 18:65–73.

30.

Gardner

, Metsch

, Marks

et al. Psychological and behavioral correlates of entering care for HIV infection: The Antiretroviral Treatment Access (ARTAS) Study. AIDS Patient Care STDs, 2007; 21:418–425.

31.

Metsch

, Peyeyra

, Messinger

et al. HIV transmission risk behaviors among HIV-infected persons who are successfully linked to care. Clin Infect Dis, 2008; 47:577–584.

32.

Brewer

, Zhao

, Peyeyra

et al. Initiating HIV care: Attitudes and perceptions of HIV positive crack cocaine users. AIDS Behav, 2007; 11:897–904.