ORIGINAL RESEARCH CONTRIBUTION
Long-term Follow-up After Voluntary Human Immunodeficiency Virus⁄Sexually Transmitted Infection Counseling, Point-of-service Testing, and Referral to Substance Abuse Treatment From the Emergency Department Edward Bernstein, MD, Timothy Heeren, PhD, Michael Winter, MPH, Desiree Ashong, MPH, Caleb Bliss, MPH, Guillermo Madico, MD, Beza Ayalew, and Judith Bernstein, PhD
Abstract Objectives: Public health initiatives have lowered human immunodeficiency virus (HIV) transmission risk associated with injection drug use in the United States, making sexual risk behaviors a greater source of transmission. Strategies are therefore needed to reduce these risk behaviors among all emergency department (ED) patients who use drugs, regardless of route of administration. Although recent articles have focused on the opportunity for early HIV detection and treatment through an array of ED screening and testing strategies, the effect of voluntary HIV testing and brief counseling (VT ⁄ C) on the sexual behaviors of out-of-treatment drug users over time has not yet been reported. Methods: From November 2004 to May 2008, the study screened 46,208 urban ED patients aged 18 to 54 years; 2,148 (4.6%) reported cocaine or heroin use within 30 days, 1,538 met eligibility criteria (Drug Abuse Severity Test [DAST] scores ‡3 and were either English- or Spanish-speaking), and 1,030 were enrolled. These data were obtained in the course of a randomized, controlled trial (Project SAFE) of a brief motivational intervention focused on reducing risky sexual behaviors. Although the intervention itself did not demonstrate any differential effect on the number or percentage of unprotected sexual acts, both control and intervention group participants received baseline VT ⁄ C and referral for drug treatment as part of the study protocol. This study is a report of a secondary analysis of cohort data to describe changes in sexual behaviors over time among drug users after the VT ⁄ C and referral. Results: The mean (±SD) age of enrollees was 35.8 (±8.4) years; 67% were male, 39% were non-Hispanic black or African American, 41% were white non-Hispanic, and 19% were Hispanic. Half injected drugs, and 53% met criteria for posttraumatic stress disorder (PTSD). At baseline testing, 8.8% were HIV-positive on enzyme-linked immunosorbent assay. Follow-ups were conducted at 6 and 12 months, with an attrition rate of 22%. Known HIV-positive patients accounted for 84 of 1,030 cases (8.1%), and 13 new cases were discovered: 7 of 946 at were discovered at the baseline contact (0.74%), 2 of 655 were discovered at 6 months (0.3%), and 4 of 706 (0.57%) were discovered at the 12-month contact. Twelve of the 13 returned for confirmatory testing and were actively enrolled in our infectious disease clinic. For all partners, there was a reduction in the percentage of unprotected sex acts over time (p < 0.0001), with decreases at 6 months versus baseline (odds ratio [OR] = 0.70, 95% confidence interval [CI] = 0.60 to 0.83), sustained at 12 months versus baseline (OR = 0.69, 95% CI = 0.58 to 0.82). For the outcome of percentage of sex acts while high, there was also a significant reduction over time (p < 0.0001), with a dropoff at 6 months versus baseline (OR = 0.31, 95% CI = 0.25 to 0.37) that was sustained at 12 months (OR vs. baseline 0.25, 95% CI = 0.20 to 0.30). In an adjusted model, male sex, older age, and HIV positivity predicted significant declines over time in the likelihood of unprotected sexual acts. Older age and
From the Department of Emergency Medicine (EB, DA, JB) and the Department of Infectious Disease (GM), Boston University School of Medicine, Boston Medical Center, Boston, MA; the Department of Community Health Sciences (EB, DA, BA, JB), the Department of Biostatistics (TH), and The Data Coordinating Center (MW, CB), Boston University School of Public Health, Boston, MA. Received May 2, 2011; revisions received August 5 and September 7, 2011; accepted October 1, 2011. Supported by NIDA R01 DA017061 (EB, JB) Brief Intervention to Reduce STDs in ER Drug Users; 1 ⁄ 04–12 ⁄ 08; $2.5 million. The authors have no disclosures or conflicts of interest to report. Supervising Editor: Rebecca M. Cunningham, MD. Address for correspondence and reprints: Edward Bernstein, MD; e-mail:
[email protected].
386
ISSN 1069-6563 PII ISSN 1069-6563583
ª 2012 by the Society for Academic Emergency Medicine doi: 10.1111/j.1553-2712.2012.01314.x
ACADEMIC EMERGENCY MEDICINE • April 2012, Vol. 19, No. 4
•
www.aemj.org
387
higher baseline drug severity predicted significant decreases over time in the likelihood of sex acts while high. Conclusions: Voluntary testing and counseling for HIV or sexually transmitted infections, accompanied by referral to drug treatment, for this population of ED cocaine and heroin users was associated with reduction in unprotected sex acts and fewer sex acts while high. ACADEMIC EMERGENCY MEDICINE 2012; 19:386–395 ª 2012 by the Society for Academic Emergency Medicine
A
pproximately 1.1 million people in the United States were infected with human immunodeficiency virus (HIV) in 2006, and as many as 253,270 of them were thought to be unaware that they are infected with HIV.1 Identification of HIV infection permits prevention, a reduction in the risk of unwitting transmission of HIV to others, and early treatment to reduce morbidity and mortality. Cross-sectional analysis of 867 million emergency department (ED) visits between 1993 and 2002, a weighted national total, suggests that only 2.8 million people (0.32%) received HIV testing.2 In 2005–2006, a 1-year trial of point of service testing was implemented in EDs in three cities: Los Angeles, New York City, and Oakland.3 In these trials, 9,365 ED patients were screened, 97 new cases were identified (1.0%), and 88% of those who were newly identified were linked to HIV health care services. The Centers for Disease Control and Prevention (CDC) for many years has recognized the missed opportunity provided by the ED in reaching patients who utilize health care episodically and in 2006 recommended that HIV screening become a routine part of ED medical services, using a nontargeted, voluntary, opt-out approach for all patients aged 13 to 64 years old in settings with prevalence of undiagnosed HIV of ‡0.1%.4 Several recent reports have focused on issues related to the feasibility of ED screening and linkage to HIV treatment,3,5–14 and there have been reports of increased condom use after multisession, intensive intervention with patients in drug treatment programs.15 However, to the best of our knowledge there have been no reports to date following changes over time in risky sexual behaviors among ED patients at high risk for HIV and sexually transmitted infections (STIs), after receipt of voluntary HIV testing and brief counseling (VT ⁄ C) and referral to drug treatment in the ED. METHODS Study Design In this article we report follow-up data after baseline HIV ⁄ STI VT ⁄ C among a large convenience sample of ED patients at high risk for HIV and STI as a result of regular heroin and cocaine use in the past 30 days. We analyzed changes from baseline to 6 and 12 months in condom use and in the rate of sex while high for all participants, in aggregate and by type of partner. These cohort data were obtained in the course of a randomized, controlled trial of a single 30-minute brief motivational interview in which the behavioral intervention demonstrated no differential effect beyond VT ⁄ C and
referral for substance abuse treatment16 and thus constitute a secondary analysis. A description of the intervention and comparisons by intervention status are reported elsewhere.16 The study was approved by the Boston University Medical Campus Institutional Review Board, and oversight was provided by a data safety monitoring board. All enrollees provided informed consent. An NIH Certificate of Confidentiality was obtained to protect subject data from subpoena. Study Setting and Population Patients registered for all types of medical care from November 2004 through May 2008 at an academic, urban, Level I trauma center ED with a racially and ethnically diverse patient population were screened for cocaine and ⁄ or heroin use at bedside from 10 AM to 10 PM 7 days per week, using a Health Needs History screening survey, details of which are reported elsewhere.17–19 Eligibility criteria included: 1) Drug Abuse Severity Test (DAST) scores ‡ 3, indicating moderate or greater severity;20 2) age 18 to 54 years; and 3) Englishor Spanish-speaking. Patients were excluded from eligibility for: 1) having no reliable contact information, 2) plans to leave the area within 3 months, 3) inability to conduct consent and interview privately, 4) impaired mental status, 5) prisoner in custody, or 6) being at risk for suicide. Study Protocol After screening, detection, consent, and baseline assessment were completed, all enrollees received STI ⁄ HIV information, voluntary pretesting counseling, point-of-service testing, and referral to drug and medical treatment. Follow-up assessment took place at 6 and 12 months. Extensive tracking procedures were instituted to prevent attrition and link persons infected with gonorrhea, chlamydia, and HIV with appropriate treatment. Assessment Instruments. Enrolled patients were assessed at baseline by trained research assistants before randomization for demographic characteristics, self-reported health status, drug use patterns, sexual practices, STI history, and posttraumatic stress disorder (PTSD). PTSD was assessed using a standardized measure, the Posttraumatic Stress Disorder Checklist, Civilian Version (PCL-C), which has a sensitivity of 0.83 and specificity of 0.82.21 They also completed a 30-day Timeline Followback (TLFB) calendar describing sexual encounters day by day, starting from the day prior to enrollment. For each day they recorded sexual acts with three
388
Bernstein et al.
•
categories of partners (main, casual, and exchange sex), specifying circumstances for each event (condom use and sex while high). The TLFB provided self-report data to construct the outcome variables: number and percentage of unprotected sex episodes in past 30 days and the number and percentage of sex episodes while high, with all, main, casual, and exchange sex partners. This instrument is the criterion standard for recording drug use over a 6-month period, has excellent convergent validity with urine samples and prospective diary methods,22,23 and is commonly used to record sexual behaviors.15,24 Biologic Markers. Testing was conducted for bacterial STIs and HIV. Urine specimens were processed by the Boston Medical Center Microbiology Laboratory for gonorrhea and chlamydia using APTIMA Combo-2 (Gen-Probe Diagnostics, Inc., San Diego, CA) transcription-mediated amplification according to manufacturer’s instructions (Gen-Probe), with results available in 3 to 5 business days. Oral mucosal transudate (OMT) samples for HIV (Orasure, Smith Kline Beecham, London, UK) were processed by the Massachusetts Department of Public Health Laboratory, with results available in 1 to 2 weeks. The OMT specimen was tested for HIV antibodies by enzyme-linked immunosorbent assay (ELISA), confirmed by Western blot. Medical records were reviewed to verify an incident case. Definition of Outcomes. Unprotected sex (with different partner types) was defined as the number and proportion of total sex acts unprotected by a condom. Sex while high (with different partner types) was defined as the number and proportion of total sex acts while high, regardless of self-reported condom use, because of the potential for reduced validity of selfreport of specific details while high. Follow-up Procedures. Sexual behaviors were reassessed via TLFB, at the 6- and 12-month visits. Selfreport of bacterial STI and HIV testing was also obtained at these follow-up visits. To minimize attrition, participants received written and telephone reminders, including e-mail and text messages, at intervals prior to appointments, using standard methods for contacting friends, family members, caseworkers, and agencies as necessary to follow-up on disconnected phone numbers and collateral reports of imprisonment and death.25,26 Data Analysis Self-reported sexual risk behaviors were measured in terms of the number and proportion of vaginal and anal sex acts protected by condom use and by number and percentage of sex acts while high, derived from 30-day TLFB data. Unadjusted changes in the number of sex acts from baseline to the 12-month follow-up were analyzed through the Wilcoxon signed rank test (to account for the highly nonnormal distribution of these variables) and changes in the percentage using condoms during their last sex act from baseline to 12 months were analyzed through McNemar’s paired sample chi-square test. Because of the highly nonnormal distribution of
LONG-TERM FOLLOW-UP AFTER VOLUNTARY HIV ⁄ STI COUNSELING
the per-subject percentage of unprotected sex acts and percentage of sex acts while high, these variables were categorized into 100, 1 to 99, and 0%, and proportional odds logistic regression models for longitudinal data were used to analyze associations with these risk behaviors over time (from baseline to 6 to 12 months). The proportional odds model is appropriate for ordinal categorical data and describes the association with being in a higher risk category through odds ratios (ORs). The longitudinal regression models accounted for the correlation from using repeated observations on the same subject, using generalized estimating equations (GEE) multinomial regression models with empirical standard errors.27 Multivariable models to control for potential confounding included the following covariates, selected based on evidence from the literature: sex, race (white, black or African American, Hispanic, other), age, DAST score as a measure of drug severity, PTSD score (>50, £50), HIV status, intravenous drug use as a measure of HIV risk, and study time point (baseline, 6 months, or 12 months). These variables were selected to account for potential confounders of sexual behavior based on existing literature and the theory of reasoned action and planned behavior.28 Power for the sample size was determined within the primary study, and variables selected for entry in a regression analysis were limited to be consistent with the number of enrollees available for secondary analysis. Covariates were tested for multicollinearity by examining correlations at baseline. Multivariable models also included an indicator variable for intervention group (VT ⁄ C intervention vs. control), and group-by-time interaction terms, which modeled separate changes in risk behavior over time for the VT ⁄ C intervention and control groups. All data analysis was conducted using SAS ⁄ STAT software, Version 9, of the SAS System for Windows (SAS Institute, Cary, NC). RESULTS We screened 46,208 patients registered for ED care for cocaine and ⁄ or heroin. Figure 1 shows the study flow in detail. Briefly, 2,148 patients screened positive for use in the past 30 days, and 1,538 (72% of positives) met eligibility criteria. Among those eligible, 354 (23%) refused enrollment and 154 (10%) were unable to complete the consent process because of time or medical care restraints. In all, 1,030 eligible patients (67%) were enrolled. Of the 1,030 persons who enrolled in the study, 802 enrollees (78%) had either 6-month or 12-month follow-up data. Baseline Characteristics Demographic characteristics are presented in Table 1. The mean (±SD) age of enrollees was 35.8 (±8.4) years, and the majority of enrollees were male. A substantial number of the females (17%) were pregnant at time of enrollment, while six enrollees were transgender. The sample was racially and ethnically diverse. Educational levels were high for an inner-city population, but socioeconomic status was discordant with educational attainment, with high rates of homelessness and unemployment.
ACADEMIC EMERGENCY MEDICINE • April 2012, Vol. 19, No. 4
Total Screened
•
www.aemj.org
389
Not eligible (N=610): In custody, not medically cleared, moving out of area or unable to give contact info, or unable to give informed consent
Detected N=2148
N=46,208 44,060 non-detected
Not Enrolled (N=508) due to: Refused (N=354) Missed / Incomplete* (N=154) *Missed = no opportunity to explain study Incomplete = no opportunity to complete
Eligible N=1538 (72%)
Enrolled N=1030 (67%)
6 Month Follow-up 655/956 (68.5%) n = 655 Followed n = 301 No follow-up, window closed# n = 7 Deaths between enrollment and 6 mo. n = 67 Incarceraons^
12 Month Follow-up 706/925 (76.3%) n = 706 Followed n = 219 No follow-up, window closed# n = 14 New deaths (21 total) n = 84 Incarceraons^
Figure 1. Cohort enrollment and follow-up. #Window is defined as follows: for 6-month follow-up = 2 months allowed after the 6-month follow-up date (and 2 weeks before). For 12-month follow-up, 3 months allowed after the 12-month follow-up date (and 2 weeks before). ^Incarcerated for more than 50% of the follow-up window.
Medical ⁄ Mental Health and Substance Use Status. Access to primary care was limited and health status was low; although 91% stated that they had public or private insurance coverage, 36% used the ED routinely, and only 61% had an identified primary care provider. Exposure to trauma and violence was common and occurred early in life. A third of the enrollees reported that they were currently taking psychiatric medications and ⁄ or had been in a psychiatric hospital in the past year, and 30% reported that in the past 30 days they nearly always or always had nothing to look forward to. The majority reported polydrug use. The median DAST score for enrollees also indicated a high level of severity (8 ⁄ 10). Half of the enrollees reported injecting drugs in the past 30 days, and 44% reported daily injection drug use, while 39% reported sharing needles. Test Results. We tested all enrollees for HIV, rather than rely solely on self-report. At enrollment 91 (8.8%) tested positive on the Orasure ELISA test. There were seven new HIV+ conversions; six were successfully transferred to infectious disease clinic, and one refused confirmatory testing or treatment. The baseline incidence of bacterial STIs was 18 of 1,030 (1.7%). Contraception as a rationale for condom use. For a large percentage of the sample, prevention of pregnancy was not a reason for condom usage. Among female enrollees, 56 (16.6%) reported being pregnant at
time of enrollment. We asked both men and women about pregnancy plans; 34.9% said ‘‘it would be ok to get pregnant now (or get a partner pregnant),’’ and 15.5% currently planned or were trying to get pregnant or impregnate a female partner. Sexual Risk Behaviors. The majority of participants were sexually active and heterosexual. For those who were sexually active in the past year, the mean (±SD) number of sex partners of either sex was 2.7 (±11.7) partners. A total of 524 of 768 of the enrollees who had sex in past 30 days had a main partner, and main partner sex accounted for 64% of total sex acts, with a mean (±SD) unprotected rate of 81.1% (±37.4%). The majority of these acts occurred while high (55%). A total of 254 of 768 enrollees who had sex in past 30 days had a casual partner, accounting for 23.8% of total sex acts, with 52.2% unprotected and 72.0% occurring while high. In response to a series of questions about exchange sex, 26% of the sample reported ever receiving money for sex and 27% ever receiving drugs for sex. A total of 135 of 768 enrollees who had sex in last 30 days had exchange sex partners, accounting for 12.2% of the total sex acts and 40.1% of unprotected sex acts, with 84.4% occurring while high. Study Attrition Of the 1,030 persons who enrolled in the study, 802 (78%) had either 6-month or 12-month follow-up data.
390
Bernstein et al.
•
Table 1 Demographic Characteristics at Baseline Baseline Characteristics* Age at enrollment (yr), mean (±SD) Sex Male Female Transgender Race Black ⁄ African American non-Hispanic White non-Hispanic Hispanic Other U.S. born (missing = 1) Primary language English Spanish Portuguese Creole Haitian Creole Other Education less than high school Homeless (missing = 1) Work Full-time (35+ hours) Part-time (
