Despite well-known benefits of regular HIV care, approximately 51% of Americans with known HIV infection are retained in HIV care.1–4 Although no uniform definition of retention in care exists, at least 1 visit to an HIV care provider every 6 months is a minimum bench mark recommended by several national organizations and used in previous studies.2–8 Prolonged gaps in care that are often associated with interruptions in treatment are also considered in the definition of retention in care. Retention in care has important individual and public health implications, as retention in care is associated with HIV viral load (VL) suppression and survival.2,6,9,10
As people live longer with HIV infection, the challenge of ensuring that HIV-infected individuals are retained in care grows. Per the National HIV/AIDS Strategy and the HIV Care Continuum Initiative, retention in care has become a priority area.11,12 In addition, the evidence-based literature focused on retention in care has grown. Several studies reveal that individuals who are young, male, racial/ethnic minorities, mentally ill, substance-using, without an AIDS diagnosis, and with insurance issues are at risk for poor retention in care.2,3,7,10,13–19 Studies have also identified reasons for dropping out of care, including feeling well, having other responsibilities, distrust of medical providers, effects of HIV medications, and depression.20–22 Additional studies found that ancillary services such as case management, outreach, support groups, and patient navigation are also associated with improved retention in care.23–36 Although these studies are important, most studies that examined reasons for poor retention and factors associated with improved retention are from a single setting or focus on a narrow group of HIV-infected individuals; few have included a comprehensive and heterogeneous HIV-infected population. In addition, few studies have explicitly examined factors associated with returning to care after gaps in care.
As part of a widespread quality improvement initiative at the New York State Department of Health's AIDS Institute, among HIV-infected patients receiving care in New York State medical facilities, we sought to identify: (1) reasons why patients had gaps in care and (2) factors associated with returning to care after gaps in care.
We identified patients who had a single visit for HIV care in one of New York State's medical facilities in 2008 and a subsequent gap in care of at least 6 months. Patients' sociodemographic and clinical characteristics, along with facilities' characteristics were obtained through electronic extraction from a state-wide reporting system and manual extraction of medical records. We report reasons for gaps in HIV care and compare patient and facility characteristics among those who did versus did not return to care after a gap in care. This study was determined to be exempt by affiliated institutional review boards.
Identification of Patients
We identified patients who had only 1 HIV visit to a single medical facility in New York State between January 2008 and December 2008 with a subsequent gap in care of at least 6 months. Patients were identified through a report from Island Peer Review Organization, a federally designated Quality Improvement Organization. This report was compiled from case lists that are submitted annually by facilities to the New York State Department of Health. This report included 60,732 patients who had at least 1 HIV visit to a physician, nurse practitioner, or physician assistant in 2008 to at least 1 of 195 medical facilities in New York that provide HIV care under the New York State Medicaid program and have signed agreements with the state to provide comprehensive HIV care services in exchange for enhanced reimbursement rates. These 195 medical facilities include hospital-based clinics, community clinics, and private practices that accept Medicaid. Care provided in medical facilities located in jails or prisons and within the Veterans Health Administration are not included in this report. From this report, we identified 3928 patients from 167 medical facilities who were reported to have a single visit at 1 medical facility, no visit at other facilities, and no visit in 2009 that was within 6 months of the visit in 2008 (eg, patients who had a single visit in December 2008 were included only if they did not have subsequent visits through May 2009—6 months later). Using these criteria, we identified patients who had 1 visit in 2008 with a subsequent gap in care of at least 6 months. Of these 167 facilities, 137 (82.0%) participated in a medical record extraction process investigating 3433 patients. Based on this record review, patients were excluded for the following reasons: 445 had more than 1 visit in 2008, 482 had a return visit within 6 months of their single visit in 2008 (eg, no gap in care of at least 6 months), 420 had no HIV primary care visit (eg, they had emergency department visits or hospitalizations only), 111 were HIV negative, 76 had missing medical records, and 34 were less than 18 years old. Seventeen facilities had no patients with a single visit in 2008 and subsequent gap in care. Thus, from 120 facilities, we identified 1865 adult HIV-infected patients with a single HIV primary care visit in 2008 and a subsequent gap in care of at least 6 months. This sample was included in the analysis.
We extracted patients' sociodemographic and clinical characteristics, facility characteristics, factors associated with retention in care from previous studies, and other factors thought to be clinically relevant according to HIV medical directors from the New York State HIV Quality of Care Advisory Committee. Two types of data extraction occurred—electronic extraction from the New York State Department of Health AIDS Institute Data Application (AIDA) and manual extraction from patients' medical records.
Electronic Extraction From the New York State AIDA
The New York State AIDA is a state-wide database repository used to measure health care utilization to which medical facilities receiving support from New York State or the Ryan White Program are required to submit data. We electronically extracted data from the AIDA that included demographic, social, and clinical characteristics. Demographic characteristics included age, gender (male, female, transgender), race/ethnicity [Hispanic, black, white, other (for the latter 3 categories, patients had either a non-Hispanic ethnicity or unknown-Hispanic ethnicity with the corresponding race)], and primary language (English, Spanish, other). Social characteristics included housing status [stable (permanent owned housing, permanent rental housing), unstable (homeless in shelter, homeless in streets, transitional housing, domestic violence situation), institutionalized (skilled nursing facility or hospice, residential drug treatment program, residential group home, residential psychiatric facility, correctional facility), with relations or friends] and primary insurance [Medicaid, Medicare, AIDS Drug Assistant Program Plus (provision of free primary care services at select clinics and laboratories), private, self-pay, other]. Clinical characteristics included HIV risk category, duration of HIV infection, antiretroviral medication utilization in 2008, and CD4 count and HIV VL reported closest to the time of their visit in 2008. If data elements were missing in the AIDA, they were then manually extracted from medical records.
Manual Extraction From Medical Records
Manual medical record extraction occurred by staff employed at each medical facility. Data were entered into a central secure internet-based portal. All staff who extracted medical record data participated in a training seminar on medical record extraction.
Prepopulated data forms incorporating data extracted from the AIDA were given to staff at facilities to complete the remaining data elements. In addition to data elements that were missing from the AIDA, other information extracted from medical records included support services received in 2008 (yes/no), type of support services received (case management or social work services, treatment adherence services, harm reduction services, nutritional services, patient navigation services, other), mental health problems during or before 2008 (yes/no), type of mental health problem (depression, anxiety, other), mental health services received during or before 2008 (yes/no), history of substance use (yes/no), type of substances use history (alcohol, crack/cocaine, heroin, prescription opioids, marijuana, methamphetamine, other), medical comorbidities during or before 2008 (yes/no), and type of medical comorbidity (cancer, chronic hepatitis B virus, chronic hepatitis C virus, chronic medical conditions, opportunistic infections, other). To understand the circumstances of patients' gaps in care, medical record extraction also included documented reasons for patients' gaps in care (open text), efforts to contact them after their gap in care (yes/no), type of effort to contact patients (contact patient by phone, e-mail, in-person, or by other means; contact patient's provider), and whether patients eventually returned to care (yes/no).
For many medical record variables, an “other” option was available in which staff could enter free text. Three authors (C.C., J.B., and F.S.) independently examined these “other” free text responses and created themes to categorize them. The final categories were those in which authors reached consensus.
Facility characteristics were designated according to categories defined by the New York State Department of Health AIDS Institute. These characteristics included type of facility [Designated AIDS Center (DAC) hospital-based clinic (state-certified HIV Center of Excellence37), community health center, non-DAC hospital-based clinic, and drug treatment program with integrated HIV clinical services38], and location of facility (urban, suburban, rural).
To identify reasons for gaps in care, we performed simple frequency analysis. To identify factors associated with returning to care after a gap in care, we first excluded patients who died, relocated, or transferred care. We then categorized patients by whether they did or did not return to care after their gap in care. Next, to examine which sociodemographic and clinical factors were associated with returning to care, we conducted bivariate analyses using mixed effect logistic regression models to account for clustering effects of patients within medical facilities. Factors that were significant at P < 0.10 were then included in a multivariate mixed effect logistic regression model. Because previous studies have demonstrated differences in retention in care by gender and age, both gender and age were forced into the final multivariate model regardless of the P value on bivariate analyses.
To account for missing data, we imputed data 5 times applying the fully conditional specification approach using SAS v9.3 PROC MI39. We applied mixed-effects logistic regression models to each imputed data set. The final parameter estimates (odds ratios) of each variable were computed by applying Rubin pooling method for multiple imputation data analysis using PROC MIANALYZE.40 We conducted sensitivity analyses by repeating the steps outlined above with the original data without imputation. The final model using the imputed data set was similar to the final model using the original data set without imputation. Therefore, we report the findings from the imputed data set.
We constructed additional multivariate models to explore how types of mental health problems, types of substance use histories, and type of support services were associated with returning to care, as previous studies have shown these factors to be associated with retention in care.10,17,23–36 To further explore how mental health problems were associated with returning to care, we created 2 different models in which all variables were the same as the base multivariate model described above except for the variable representing mental health problems. For example, in these 2 models, instead of including a variable that represented any mental health problem, 1 model included a variable representing depression (depression, no depression but other mental health problem, no mental health problem), and 1 model included a variable representing anxiety (anxiety, no anxiety but other mental health problem, no mental health problem). We repeated this process creating 4 different models for substance use history (exploring a history of alcohol, heroin, crack/cocaine, and marijuana use) and 1 different model for support services received (exploring the receipt of case management).
Of 1865 patients who had a single HIV primary care visit at 1 medical facility in New York, no visit at other New York facilities, and a subsequent gap in care of at least 6 months, the mean age was 43 years, and most were men (63.2%), black (51.4%), lived in stable housing (65.0%), and had Medicaid (53.9%) (Table 1). Most (61.2%) were diagnosed with HIV infection more than 5 years before their visit in 2008. Many patients had CD4 counts greater than 500 cells per cubic millimeter (34.4%) and undetectable VL (45.0%). The majority of patients (62.8%) had been prescribed antiretroviral medications at some time during 2008. Nearly one-third (29.2%) had a mental health problem, with depression most common. Nearly half (41.7%) had a history of substance use, with alcohol or crack/cocaine histories most common. Almost half (43.7%) had a medical comorbidity. Over half (55.1%) received supportive services, with case management as the most commonly received service. The majority of patients had a single visit in 2008 in a New York State DAC, and at a facility located in an urban area.
Over half (55.9%) of the patients had an unknown reason for their gap in care (Table 2). Of those with known reasons, reasons varied considerably with no clear predominant reason for gaps in care. Medical facilities made efforts to contact 67.7% of patients who had a gap in care, with the most common effort through phone calls (70.4%) or e-mail (63.0%). After a gap in care, excluding those who transferred care, relocated, or expired, over half (54.6%) returned to care. Of those who returned to care, the median duration of their gap was 336 days (interquartile range, 264–420; range, 184–932 days).
In bivariate analysis, those who did (vs. did not) return to care significantly differed by race/ethnicity, language, housing status, insurance, HIV risk category, duration of HIV infection, CD4 count, HIV VL, use of antiretroviral medications, mental health problems, substance use histories, and receipt of supportive services (Table 3). In addition, significant differences in the location and type of medical facility existed between those who did versus did not return to care. In the multivariate analysis, those who did (vs. did not) return to care were more likely to have stable housing, longer duration of HIV infection, high CD4 count, suppressed VL, have been prescribed antiretroviral medications, have received care in a DAC and a medical facility located in a non-urban area, and had a facility make efforts to contact them. Those who returned to care were also less likely to be uninsured and have mental health problems or substance use histories.
In exploratory models, specific types of mental health problems and types of substance use histories were also significantly associated with returning to care. Compared with those with no mental health problem, patients with depression were less likely to return to care [adjusted odds ratio (AOR) = 0.68; 95% confidence interval (CI): 0.50 to 0.93], whereas those with anxiety were not (AOR = 0.92; 95% CI: 0.59 to 1.44). Compared with those with no substance use history, patients with a history of heroin use (AOR = 0.53; 95% CI: 0.32 to 0.86), crack/cocaine use (AOR = 0.60; 95% CI: 0.41 to 0.87), and alcohol use (AOR = 0.67; 95% CI: 0.47 to 0.94) were less likely to return to care, whereas patients with a history of marijuana use were not (AOR = 0.80; 95% CI: 0.53 to 1.21). Finally, compared with those who did not receive supportive services, there were no differences in whether patients returned to care by receipt of case management (AOR = 0.95; 95% CI: 0.72 to 1.24).
Of 1865 HIV-positive adult New Yorkers who had a single visit to an HIV primary care provider in 2008 followed by a subsequent gap in care of at least 6 months, most had indicators of well-controlled HIV infection (high CD4 count and suppressed VL), and nearly half had a substance use history or medical comorbidity. Over half received supportive services. Most had unknown reasons for gaps in care, and in those with known reasons, no clear predominant reasons for gaps in care emerged. For the majority of patients, medical facilities attempted to contact them, and over half of the patients returned to care.
With our study sample drawn from New York State, which includes a heterogeneous patient population from a spectrum of geographic locations (urban, suburban, and rural), our findings extend those of previous studies. We found similar sociodemographic and clinical characteristics among patients not retained in care as those reported in previous studies.2,3,7,10,13–18 For example, a high proportion of patients who were black, had high CD4 counts, and had substance use histories or mental health problems had gaps in care.
Few US studies have explicitly examined reasons for gaps in care, and to the best of our knowledge, all these studies interviewed patients to determine reasons. Reasons included feeling well, having other responsibilities or competing priorities, distrust of or negative experiences with medical providers, effects of HIV medications, depression, lack of insurance, and not wanting to think about HIV.20–22 We used different methods to determine reasons for gaps in care—we extracted this information from medical records. Although our methods undoubtedly contributed to our finding that approximately half of the patients had unknown reasons for gaps in care, examining gaps in care from medical records can contribute meaningful findings. For example, studies that derive data from patient interviews are likely to under-report certain reasons for gaps in care, such as incarceration, relocating residence, and transferring care because patients who have these reasons may be least likely to be included in samples of those interviewed. In contrast, from medical record extraction, this type of information is available; however, other more nuanced reasons for gaps in care may be under-reported in medical records. Given these differences based on methods used in studies, it is important that studies report on reasons for gaps in care using a variety of methods, including both patient interviews and medical record reviews.
In our study, of those with known reasons for gaps in care, no predominant reason emerged. The most common reasons for gaps in care included transferring care, relocating residence, and death. However, it is not known if those who transferred care or relocated actually established care in another facility outside of New York. Other common reasons for gaps in care included incarceration, “life stressors” (eg, work, school, housing issues, caring for others), and that the HIV provider was not the patient's primary care physician. It is no surprise that life stressors may be barriers to achieving optimal HIV care utilization, as HIV-positive individuals tend to be marginalized in many ways, which can lead to an additional level of burden that may take priority over seeking medical care. This theme is consistent with other studies.21,41,42 Taken together, the lack of consistent emerging themes in our and others' studies to explain patients' gaps in care leaves medical providers and policy makers uncertain on how to mitigate or prevent gaps in care. Because of the variety of reasons for gaps in care, individualized problem solving or case management to address specific challenges of remaining in care seem to be necessary. Further exploration into reasons for or causes of gaps in care, and ultimately how to prevent these gaps in care is warranted.
Our study adds to the literature by shedding light on factors associated with returning to care after a gap in care. Although a few studies have examined returning to care after the employment of a public health case worker or retention specialist,21,43 we are not aware of studies that have examined a variety of factors associated with returning to care. We found that those who did (vs. did not) return to care were more likely to have indicators of psychosocial and medical stability. Many of our findings are consistent with other studies' findings in terms of factors associated with retention in care. These findings are particularly problematic, as it is likely that patients who are less likely to return to care—those with unstable social and medical characteristics (mental health problems, substance use, and poorly controlled HIV)—are likely to be at highest risk of transmitting HIV to others. Similar to other studies, our findings point to the importance of addressing comprehensive psychosocial care of HIV-infected patients, including addressing social issues (eg, housing, insurance), integration of care (eg, mental health and substance abuse treatment services integrated into HIV care), joint planning, problem solving, and shared decision making among providers and patients. In particular, studies demonstrating the benefits of strength-based case management and managed problem solving provide support for an individualized approach that is tailored to patients' needs and situations.30,44
Although we found that supportive services were not associated with returning to care, efforts to contact patients were associated with returning to care. A substantial body of literature consisting mostly of observational studies, with a few randomized trials, provides evidence that support services such as case management, outreach, support groups, and patient navigation are associated with improved retention in care.23–36 Our findings differ from these studies. However, in our study, except for case management which was received by the vast majority of patients and may be variably defined by different facilities, other support services were not commonly received, possibly limiting our ability to detect differences between those who did and did not receive these services. In addition, our study specifically examined patients with gaps in care, while many other studies examined patients “at risk” for poor retention in care. Possibly, supportive services help keep patients who are at risk of dropping out of care in care, but support services alone may not be sufficient to bring patients who have had a gap in care back into care. Despite these findings, after a gap in care, efforts made to contact patients were associated with returning to care. These data suggest that developing strategies to identify and contact patients who have a gap in care can lead to successful return to care. Further examination of the types of efforts (phone calls, letters, home visits) to contact patients is warranted to best use services and resources.
Our study has limitations. The medical facilities and their HIV-infected patients that we included in our study do not represent all medical facilities or all HIV-infected patients receiving care in New York State. In addition, on medical record review, we found errors in the initial report of patients who were reported to have a single visit in 2008, and removed them from our sample. It is possible that the opposite error may have occurred (in which patients may have had a single visit in 2008, but were reported to have more than 1 visit). We relied on data extracted from the New York State AIDS Institute Data Application—an electronic database designed to measure health services utilization—and from manual medical record extraction. As mentioned above, because we did not survey patients directly, it is no surprise that our data regarding reasons for gaps in care differed from previous studies that interviewed patients. Using medical records to determine reasons for gaps in care may lead to “unknown” reasons for gaps in care for a substantial number of patients. In addition, medical record data are likely unable to capture nuanced information about reasons for gaps in care or returning to care. Because of our study design, causality cannot be determined. For example, one cannot determine whether case management was ineffective in facilitating returning to care, or if those who received case management did so because they were the most likely to not return to care. To address missing data, we relied on imputed data for our final models. However, when we conducted analyses with and without imputed data, we found similar results. Although our study focuses on care received in 2008–2009, the issue of poor retention in care remains important today. Unfortunately, few new policies or interventions have been implemented that substantially improve retention in care at the population level. Therefore, we are confident that our findings still contribute to the field. Finally, because our data capture HIV care in New York, our findings may not be generalizable to other states.
In summary, HIV-infected patients who had a single HIV primary care visit to a New York facility in 2008 with a subsequent gap in care had similar sociodemographic and clinical characteristics to patients not retained in care in previous studies. Just over half of the patients with a gap in their HIV care returned to care, and no major theme emerged regarding reasons for gaps in care. Compared with those who did not return to care, those who did return to care were more likely to have indicators of medical and psychosocial stability. Additionally, efforts to contact patients after their gap in care, and not supportive services, were associated with returning to care. Our findings emphasize that stabilizing patients' psychosocial factors and contacting patients after a gap in care are key strategies to retain HIV-infected New Yorkers in care.
The authors thank Jillian Brown, Fareesa Islam, and Jeremy Collado for their contributions to this project.
1. Vital signs: HIV prevention through care and treatment–United States. MMWR Morb Mortal Wkly Rep. 2011;60:1618–1623.
2. Tripathi A, Youmans E, Gibson JJ, et al.. The impact of retention in early HIV medical care on viro-immunological parameters and survival: a statewide study. AIDS Res Hum Retroviruses. 2011;27:751–758.
3. Torian LV, Wiewel EW. Continuity of HIV-related medical care, New York City, 2005-2009: do patients who initiate care stay in care? AIDS Patient Care STDS. 2011;25:79–88.
4. Marks G, Gardner LI, Craw J, et al.. Entry and retention in medical care among HIV-diagnosed persons: a meta-analysis. AIDS. 2010;24:2665–2678.
5. Mugavero MJ, Davila JA, Nevin CR, et al.. From access to engagement: measuring retention in outpatient HIV clinical care. AIDS Patient Care STDS. 2010;24:607–613.
6. Mugavero MJ, Westfall AO, Zinski A, et al.. Measuring retention in HIV care: the elusive gold standard. J Acquir Immune Defic Syndr. 2012;61:574–580.
7. Mugavero MJ, Amico KR, Horn T, et al.. The state of engagement in HIV care in the United States: from cascade to continuum to Control. Clin Infect Dis. 2013;57:1164–1171.
8. Horstmann E, Brown J, Islam F, et al.. Retaining HIV-infected patients in care: where are we? where do we go from here? Clin Infect Dis. 2010;50:752–761.
9. Mugavero MJ, Lin HY, Willig JH, et al.. Missed visits and mortality among patients establishing initial outpatient HIV treatment. Clin Infect Dis. 2009;48:248–256.
10. Giordano TP, Hartman C, Gifford AL, et al.. Predictors of retention in HIV care among a national cohort of US veterans. HIV Clin Trials. 2009;10:299–305.
13. Rebeiro P, Althoff KN, Buchacz K, et al.. Retention among North American HIV-infected persons in clinical care, 2000-2008. J Acquir Immune Defic Syndr. 2013;62:356–362.
14. Mugavero MJ, Lin HY, Allison JJ, et al.. Failure to establish HIV care: characterizing the “no show” phenomenon. Clin Infect Dis. 2007;45:127–130.
15. Cunningham CO, Sohler NL, Wong MD, et al.. Utilization of health care services in hard-to-reach marginalized HIV-infected individuals. AIDS Patient Care STDS. 2007;21:177–186.
16. Arici C, Ripamonti D, Maggiolo F, et al.. Factors associated with the failure of HIV-positive persons to return for scheduled medical visits. HIV Clin Trials. 2002;3:52–57.
17. Pecoraro A, Royer-Malvestuto C, Rosenwasser B, et al.. Factors contributing to dropping out from and returning to HIV treatment in an inner city primary care HIV clinic in the United States. AIDS Care. 2013;25:1399–1406.
18. Meyerson BE, Klinkenberg WD, Perkins DR, et al.. Who's in and who's out: use of primary medical care among people living with HIV. Am J Public Health. 2007;97:744–749.
19. Bertolli J, Garland PM, Valverde EE, et al.. Missed connections: HIV-infected people never in care. Public Health Rep. 2013;128:117–126.
20. Magnus M, Herwehe J, Murtaza-Rossini M, et al.. Linking and retaining HIV patients in care: the importance of provider attitudes and behaviors. AIDS Patient Care STDS. 2013;27:297–303.
21. Udeagu CC, Webster TR, Bocour A, et al.. Lost - or just not following up?: public health effort to re-engage HIV-infected persons lost to follow-up into HIV medical care: 108(120). AIDS. 2013;27:2271–2279.
22. Christopoulos KA, Massey AD, Lopez AM, et al.. “Taking a half day at a time:” patient perspectives and the HIV engagement in care continuum. AIDS Patient Care STDS. 2013;27:223–230.
23. Liau A, Crepaz N, Lyles CM, et al.. Interventions to promote linkage to and utilization of HIV medical care among HIV-diagnosed persons: a qualitative systematic review, 1996-2011. AIDS Behav. 2013;17:1941–1962.
24. Higa DH, Marks G, Crepaz N, et al.. Interventions to improve retention in HIV primary care: a systematic review of U.S. studies. Curr HIV/AIDS Rep. 2012;9:313–325.
25. Bradford JB, Coleman S, Cunningham W. HIV system navigation: an emerging model to improve HIV care access. AIDS Patient Care STDS. 2007;21:S49–S58.
26. Cabral HJ, Tobias C, Rajabiun S, et al.. Outreach program contacts: do they increase the likelihood of engagement and retention in HIV primary care for hard-to-reach patients? AIDS Patient Care STDS. 2007;21(suppl 1):S59–S67.
27. Cunningham CO, Shapiro S, Berg KM, et al.. An evaluation of a medical outreach program targeting unstably housed HIV-infected individuals. J Health Care Poor Underserved. 2005;16:127–138.
28. Cunningham CO, Sohler NL, Berg KM, et al.. Type of substance use and access to HIV-related health care. AIDS Patient Care STDS. 2006;20:399–407.
29. Cunningham CO, Sanchez JP, Li X, et al.. Medical and support service utilization in a medical program targeting marginalized HIV-infected individuals. J Health Care Poor Underserved. 2008;19:981–990.
30. Gardner LI, Metsch LR, Anderson-Mahoney P, et al.. Efficacy of a brief case management intervention to link recently diagnosed HIV-infected persons to care. AIDS. 2005;19:423–431.
31. Harris SK, Samples CL, Keenan PM, et al.. Outreach, mental health, and case management services: can they help to retain HIV-positive and at-risk youth and young adults in care? Matern Child Health J. 2003;7:205–218.
32. Katz MH, Cunningham WE, Fleishman JA, et al.. Effect of case management on unmet needs and utilization of medical care and medications among HIV-infected persons. Ann Intern Med. 2001;135:557–565.
33. Kushel MB, Colfax G, Ragland K, et al.. Case management is associated with improved antiretroviral adherence and CD4+ cell counts in homeless and marginally housed individuals with HIV infection. Clin Infect Dis. 2006;43:234–242.
34. Lo W, MacGovern T, Bradford J. Association of ancillary services with primary care utilization and retention for patients with HIV/AIDS. AIDS Care. 2002;14(suppl 1):S45–S57.
35. Magnus M, Schmidt N, Kirkhart K, et al.. Association between ancillary services and clinical and behavioral outcomes among HIV-infected women. AIDS Patient Care STDS. 2001;15:137–145.
36. Messeri PA, Abramson DM, Aidala AA, et al.. The impact of ancillary HIV services on engagement in medical care in New York City. AIDS Care. 2002;14(suppl 1):S15–S29.
38. Rothman J, Rudnick D, Slifer M, et al.. Co-located substance use treatment and HIV prevention and primary care services, New York State, 1990-2002: a model for effective service delivery to a high-risk population. J Urban Health. 2007;84:226–242.
39. van Buuren S. Multiple imputation of discrete and continuous data by fully conditional specification. Stat Methods Med Res. 2007;16:219–242.
40. Rubin DB. Multiple Imputation for Nonresponse in Surveys. New York, NY: J. Wiley & Sons; 1987.
41. French T, Tesoriero J, Agins B. Changes in stress, substance use and medication beliefs are associated with changes in adherence to HIV antiretroviral therapy. AIDS Behav. 2011;15:1416–1428.
42. French T, Weiss L, Waters M, et al.. Correlation of a brief perceived stress measure with nonadherence to antiretroviral therapy over time. J Acquir Immune Defic Syndr. 2005;38:590–597.
43. Sitapati AM, Limneos J, Bonet-Vazquez M, et al.. Retention: building a patient-centered medical home in HIV primary care through PUFF (Patients Unable to Follow-up Found). J Health Care Poor Underserved. 2012;23(suppl 3):81–95.
44. Gross R, Bellamy SL, Chapman J, et al.. Managed problem solving for antiretroviral therapy adherence: a randomized trial. JAMA Intern Med. 2013;173:300–306.