Half of 30-Day Hospital Readmissions Among HIV-Infected Patients Are Potentially Preventable

Abstract

Thirty-day readmission rates, a widely utilized quality metric, are high among HIV-infected individuals. However, it is unknown how many 30-day readmissions are preventable, especially in HIV patients, who have been excluded from prior potentially preventable readmission analyses. We used electronic medical records to identify all readmissions within 30 days of discharge among HIV patients hospitalized at a large urban safety net hospital in 2011. Two independent reviewers assessed whether readmissions were potentially preventable using both published criteria and detailed chart review, how readmissions might have been prevented, and the phase of care deemed suboptimal (inpatient care, discharge planning, post-discharge). Of 1137 index admissions, 213 (19%) resulted in 30-day readmissions. These admissions occurred among 930 unique HIV patients, with 130 individuals (14%) experiencing 30-day readmissions. Of these 130, about half were determined to be potentially preventable using published criteria (53%) or implicit chart review (48%). Not taking antiretroviral therapy (ART) greatly increased the odds of a preventable readmission (OR 5.9, CI:2.4–14.8). Most of the preventable causes of readmission were attributed to suboptimal care during the index hospitalization. Half of 30-day readmission in HIV patients are potentially preventable. Increased focus on early ART initiation, adherence counseling, management of chronic conditions, and appropriate timing of discharge may help reduce readmissions in this vulnerable population.

Introduction

Despite major advances in HIV treatment and outcomes, hospital admissions and readmissions for HIV-infected patients remain common,^1,2 costly,³ and disproportionately impact vulnerable groups.^4

–11 With the introduction of highly active antiretroviral therapy (HAART) in 1996, overall rates of hospitalization among HIV-infected patients declined significantly, particularly for AIDS-related issues,⁹ though subsequently admissions for other causes (liver disease, non-opportunistic infections) increased^1,2,4 and demonstrated ongoing health disparities for women,¹ patients with serious mental illness,¹⁰ injection drug users^7,10 and African Americans.^1,5,8,11

In the HAART era, national HIV hospitalization costs were 3.2 billion in 2005,¹² and costs for readmissions in this population may be up to 32% higher than the index admission.¹³ Readmission rates within 30 days of discharge are 19–25% for HIV-infected patients,^14,15 which is comparable to Medicare beneficiaries,^16,17 congestive heart failure (CHF) patients,^18,19 and higher than published readmission rates for acute myocardial infarction (MI,18.9%)²⁰ and pneumonia (17.4%).²¹ Medicare's Hospital Readmissions Reduction Program (HHRP) financially penalizes hospitals with much higher than predicted 30-day readmissions for CHF, pneumonia, and MI. In the coming year, these reimbursement reduction policies are being expanded to other conditions and the magnitude of the penalties are rising. Given the policy focus on 30-day readmissions and high rates of readmissions among HIV patients, new efforts are needed to better understand and decrease readmissions in this high risk population.

Despite the widespread utilization of 30-day readmissions as a quality metric, not all readmissions are preventable. Studies measuring preventable readmissions have produced widely varying estimates, with a recent systematic review reporting that between 5–79% (median 27%) of readmissions among medical, surgical, and geriatric patients were potentially preventable.²² Several studies on preventable readmissions have used algorithms to analyze large administrative databases,^23,24 and although this allows for the analysis of large numbers of cases, this method lacks the ability to examine the detailed clinical and social reasons for readmission or to determine if and how readmissions may have been prevented. Several published studies that did perform detailed chart reviews^25
–27 were limited by lack of a clear definition of preventable readmissions or rigorous review process.

To date, there are no published data on preventability of readmissions among HIV-infected patients. In fact, HIV patients, along with trauma, burn, and oncology patients, have traditionally been excluded from preventable readmission analyses,²⁸ due “to the complexity of the patient and the low likelihood that a readmission can be prevented”.²⁹ Though HIV can be readily treated as an outpatient with excellent outcomes, admission and readmission rates remain high, especially among disadvantaged populations,⁹ suggesting that many readmissions among HIV patients are potentially preventable. In addition, inpatient admissions among HIV patients represent a unique opportunity to improve clinical outcomes, including a reduction in hospital admissions, through engagement in routine HIV care, treatment with antiretroviral therapy, and eventual virologic suppression.

This study sought to: (1) Assess the rates of 30-day readmissions for HIV in a large, urban safety net teaching hospital; (2) Determine the proportion of readmissions which were preventable using two separate methods; (3) Identify the predictors of preventable readmissions; and (4) identify when and how readmissions may have been prevented.

Methods

We conducted a retrospective chart review study of all patients hospitalized with HIV at Parkland Health and Hospital System between January 1, 2011 and December 31, 2011 and subsequently readmitted within 30 days of discharge to the same institution. Parkland Hospital is a large urban teaching hospital that is the sole safety net provider in Dallas County. As such, Parkland provides inpatient and outpatient care to the majority of HIV-infected patients in Dallas. All admissions and readmissions were identified using the hospital's comprehensive electronic medical record. This study was approved by the Institutional Review Board of University of Texas Southwestern Medical Center.

A multidisciplinary study team consisting of clinicians, medical providers, case managers, social workers, and administrative staff reviewed each medical record and abstracted data from the first index admission and 30-day readmission occurring for each subject in 2011. Readmission was defined as an inpatient admission due to any cause to Parkland Hospital within a 30-day period from the index discharge.

Administrative staff collected data on demographics and dates of service, case managers reviewed social components such as discharge disposition and substance use, and HIV providers reviewed medical data, with at least two HIV physicians and/or nurse practitioners independently reviewing each medical record.

Data collection

A data collection tool was developed and used by all individuals abstracting data to collect the data uniformly and consistently. Covariates that were collected by administrative staff included patient demographics (age, gender, race/ethnicity, and health insurance), length of stay, and health care utilization prior to and after the index admission, and whether appointments were scheduled after discharge and if the patient attended the appointment.

Case managers reviewed whether patients were seen by a case manager (HIV and/or general) during the index admission, the location (ER, clinic, outside hospital) from which patients were initially admitted and readmitted, the patient's discharge disposition (to home, shelter, jail, nursing home, left against medical advice, etc.) and what day of the week each patient was discharged.

Clinical review was independently completed by two of six providers experienced in the care of HIV patients. They abstracted data on baseline laboratory values including CD4 count and HIV viral load at the time of index admission. If no labs were available from the index admission, CD4 count and HIV viral load were abstracted from the record for up to 90 days prior. HIV viral load was examined as a continuous variable (with log transformation) and also as a binary outcome [detectable (>20 copies/mL) versus undetectable]. Patients were defined as having an active psychiatric disorder if they underwent a psychiatric evaluation or required adjustment of psychiatric medications during the admission. Stable mental health issues where outpatient medications were continued during admission were not counted in this group. Active substance users were defined as individuals noted to be using drugs or heavy alcohol in the month prior to admission or who had a positive urine toxicology screen on admission.

Providers also recorded the medical reasons for admission and readmission; whether the admission was HIV-related (direct complication of HIV or opportunistic infection); if the HIV diagnosis was new (made within the month of initial admission), recent (diagnosed with the past 12 months) or known >12 months; if infectious diseases specialists were consulted; whether the patient had been admitted to the intensive care unit; whether the patient was prescribed ART (including inpatient initiation of treatment), and the number of medications prescribed at the time of discharge.

Assessment of preventability

Using two different methods, clinical reviewers determined whether a readmission was potentially preventable. Initial assessment was performed using published criteria, as defined by 3M Health Information Systems Potentially Preventable Readmissions and Goldfield et al. (hereafter referred to as Goldfield method),^23,30,31 wherein a readmission is considered preventable when the patient is readmitted for: (1) recurrence or continuation of the reason for the initial admission; (2) acute exacerbation of a chronic problem that was not the main reason for the initial admission; (3) complication plausibly related to the care during the first admission; (4) surgical procedure that was not done in the initial admission; or (5) complication of a surgical procedure. Readmissions for malignancies, trauma, obstetrical causes, or after leaving “against medical advice” were considered nonpreventable. Second, an in-depth implicit assessment about preventability was made after comprehensive review of the medical record to determine if the readmission was practically preventable and if so, how and during which phase (inpatient, discharge, outpatient) of care it may have been prevented.

Each phase of care was graded on several key factors: (a) Quality of inpatient care (suboptimal management of chronic condition; patient unstable at discharge or discharged too soon; complication from initial hospitalization; missed or inaccurate diagnosis); (b) Discharge process (unaddressed psychological and/or social needs; missed referrals; inadequate arrangements for supplies or medications); and (c) Follow-up outpatient care (delayed physician follow-up in clinic; lack of follow-up on pending laboratory or procedure results; patient not compliant with treatment). Readmissions could have no identifiable preventable cause or could have multiple causes. For preventability assessment, all charts were reviewed independently by two providers; disagreements were resolved by consensus in a group of six HIV providers.

Data analysis

Preventability of readmissions was calculated as a proportion of the total number of readmissions, with separate estimates using the Goldfield method²³ and implicit chart review methods. The implicit chart review determination of preventability was utilized as the primary outcome as this identified all readmissions that were practically preventable. Patient baseline characteristics and predictors were summarized by frequencies (among preventable readmissions, nonpreventable readmissions, and overall). Univariate relationships between predictors and preventable readmissions were described. A stepwise selection method was used to determine the final multivariate logistic regression model, with statistical significance defined as p<0.05. All statistical analyses were conducted using SAS, Version 9.4 (Cary, NC).

Results

Of 1137 index HIV admissions during 2011, 213 (19%) resulted in 30-day readmissions. These admissions represent 930 unique individual HIV patients hospitalized during the study year, of which 137 individuals (15%) were readmitted within 30 days. Of these 137 patients evaluated, 7 were excluded as they did not constitute true readmissions (patients were transferred to other floors within the hospital, such as psychiatry prior to “readmission”). Of the 130 remaining, 79 had one readmission, 27 had two readmissions, 8 had 3 readmissions, and 16 had 4 or more readmissions during the study period. Patients with multiple readmissions fell into three general categories: patients with advanced AIDS and recurrent AIDS-defining illnesses, patients with chronic conditions with exacerbations (CHF, chronic obstructive pulmonary disease) and patients with cancer who were receiving chemotherapy. Preventability was assessed for the initial readmission only.

By the Goldfield method, 69/130 (53%) of initial readmissions were determined to be potentially preventable. By the implicit chart review method, 62/130 (48%) were classified as preventable readmissions, including 57 of the 69 potentially preventable readmissions identified by the Goldfield method. There was initial disagreement on 7% of assessments of preventability, which was resolved by group consensus. Figure 1 depicts the proportion of patients in each category for preventable and nonpreventable readmissions as determined by implicit chart review, with the 12 which were preventable by the Goldfield method but nonpreventable by chart review classified as “not practically preventable”.

FIG. 1.

Flowchart of potentially preventable and nonpreventable readmissions.

Baseline characteristics of the cases are described in Table 1, indicating that about two-thirds of readmitted patients were male, 70% were either African American or Hispanic, and very few had private insurance. Over half of patients had AIDS by CD4 criteria and 45% were not prescribed ART at the time of the index admission. Mean length of stay for the index admission was 5.8 days. One-third of patients were seen by an infectious diseases specialist during their stay, and they were prescribed a mean of 9.5 medications at discharge. Over half (58%) had attended at least one HIV clinic visit in the past 12 months, with 76/78 attending an appointment within the preceding 6 months, and 32% attended a post-hospitalization appointment in an outpatient HIV clinic.

Table 1.

Characteristics of HIV-Infected Patients with 30-Day Readmissions by Preventability

	Preventable (n=62)	Not preventable (n=68)	Total (n=130)
Demographics
Age (median, range, years)	45.0 (25–69)	41.8 (20–73)	43.4 (20–73)
Gender
Male	44 (71%)	42 (62%)	86 (66%)
Female	18 (29%)	26 (39%)	44 (34%)
Race
White, non-Hispanic	17 (27%)	22 (33%)	39 (30%)
Black, non-Hispanic	35 (56%)	30 (44%)	65 (50%)
Hispanic	10 (16%)	16 (24%)	26 (20%)
Financial classification
Private	0 (0%)	2 (3%)	2 (2%)
Medicare	8 (13%)	6 (9%)	14 (11%)
Medicaid	16 (26%)	22 (32%)	38 (29%)
Ryan White	23 (37%)	20 (29%)	43 (33%)
PHP/jail	9 (14%)	9 (13%)	18 (14%)
Self-pay	6 (10%)	9 (13%)	15 (12%)
Clinical factors
Smoker	25 (40%)	20 (29%)	45 (35%)
Active substance or alcohol use	31 (49%)	25 (37%)	56 (43%)
Active psych disorder	11 (18%)	14 (21%)	25 (19%)
CD4 at initial admit (mean, range, cells/mm³)	248 (1–1895)	287 (4–1101)	269 (1–1895)
CD4 category (cells/mm³)
<50	17 (27%)	17 (25%)	34 (26%)
51–200	22 (35%)	15 (22%)	37 (28%)
201–500	15 (24%)	21 (31%)	36 (28%)
>500	8 (13%)	15 (22%)	23 (18%)
CD4<200 and not on HAART	22 (35%)	8 (12%)	30 (23%)
On HAART
During first admit	24 (39%)	47 (69%)	71 (55%)
During second admit	33 (53%)	51 (75%)	84 (65%)
HIV viral load at initial admit (mean/median, range, copies/mL)	390,956	395,657	393,434 (<20–7.6M)
Undetectable (VL<400)	19 (31%)	29 (43%)	48 (37%)
On HAART and VL<400	19/33 (57%)	29/56 (52%)	48/89 (54%)
Admission/readmission details
Length of stay (mean, range, days)	5.5 (1–23)	6.0 (1–51)	5.8 (1–51)
Days to readmission (mean, range)	11.5 (0.6–30)	12.4 (0–30)	12.0 (0–30)
Admit to internal medicine service	52 (83%)	48 (72%)	100 (77%)
ICU stay	12 (19%)	8 (12%)	20 (15%)
Seen by inpatient case management	43 (68%)	48 (72%)	91 (70%)
HIV case management	40 (63%)	44 (66%)	84 (64%)
HIV-related admission	28 (45%)	27 (40%)	55 (43%)
New HIV diagnosis	8 (13%)	7 (10%)	15 (12%)
ID consult	20 (32%)	23 (34%)	43 (33%)
Discharge/ health utilization
Clinic visit within past year	37 (60%)	39 (57%)	76 (58%)
#ED visits 2011 (mean, range)	3.0 (2–8)	3.6 (2–9)	3.3 (2–9)
Discharge disposition
Home	46 (74%)	52 (76%)	99 (76%)
Unstable (homeless, shelter)	11 (18%)	7 (10%)	18 (14%)
Structured (nursing home, jail)	4 (6%)	5 (7%)	9 (7%)
AMA	1 (2%)	4 (6%)	5 (4%)
Total medications at discharge (mean, range)	9.3 (2–25)	9.6 (0–26)	9.5 (0–26)
Post-discharge appointment scheduled	37 (59%)	32 (48%)	69 (53%)
Post-discharge appointment kept (% of scheduled)	23 (62%)	19 (59%)	42 (61%)
Weekend discharge	19 (30%)	14 (21%)	33 (25%)
Weekend readmission	7 (11%)	11 (16%)	18 (14%)

AMA, Against Medical Advice; ED, Emergency Department; HAART, highly active antiretroviral therapy; ICU, Intensive Care Unit; ID, infectious diseases; M, Million; PHP, Parkland Health Plus (charity care); TCU, Transitional Care Unit.

Univariate and multivariate associations between predictors and preventable readmission are presented in Table 2. Older age, smoking, active substance use, CD4<200 cells/μL, not receiving ART (antiretroviral therapy), higher HIV viral load at admission, having a detectable HIV viral load, admission to the internal medicine service, being discharged on a weekend day, and having a post-hospitalization follow-up appointment scheduled were positively associated with preventable readmission. However, in multivariate analyses, the four factors remaining in the model were having Medicaid insurance (less likely to have preventable readmission, (OR 0.5, p=0.14) and those positively associated with preventable readmission: having had an appointment in the past year (OR 1.8, p=0.14), attending a post-hospitalization follow-up appointment (OR 2.1, p=0.06) and not being treated with ART during either admission (OR 5.9, p<0.01).

Table 2.

Univariate and Multivariate Logistic Regression Analyses of Preventable Readmissions Among HIV-Infected Patients

	Unadjusted OR	p Value	Adjusted OR	p Value
Demographics
Age (median, range)	1.03 (1.0–1.1)	0.09
Male gender	1.5 (0.73–3.2)	0.27
Race
White, non-Hispanic	Ref
Black, non-Hispanic	1.5 (0.7–3.4)	0.31
Hispanic	0.8 (0.3–2.2)	0.68
Financial classification
Medicaid vs. other	0.9 (0.4–2.0)	0.76	0.5 (0.2–1.2)	0.14
Clinical factors
Smoker	1.6 (0.8–3.4)	0.19
Active substance use (alcohol/drug)	1.7 (0.9–3.5)	0.13
Active psych disorder	0.8 (0.3–2.0)	0.68
CD4<200 at initial admit	1.9 (0.9–3.8)	0.07
No HAART during either admit	3.8 (1.7–8.5)	<0.01^*	5.9 (2.4–14.8)	<0.01^*
CD4<200 and no HAART Rx	4.1 (1.7–10.2)	<0.01^*
HIV viral load at initial admit	1.1 (1.0–1.1)	0.04
Undetectable viral load	0.6 (0.3–1.3)	0.18
Admission/readmission details
Length of stay	1.1 (0.7–1.6)	0.58
Time to readmission	1.0 (0.9–1.0)	0.51
ICU stay	1.8 (0.7–4.7)	0.23
Admitted to medicine service vs. other	2.2 (0.9–5.1)	0.07
Seen by inpatient case management	0.9 (0.4–2.0)	0.87
HIV-related admission	1.3 (0.6–2.5)	0.53
New HIV diagnosis	1.3 (0.4–3.8)	0.64
ID consult	0.9 (0.4–1.9)	0.85
Discharge/ healthcare utilization
Clinic visit within past year	1.1 (0.5–2.2)	0.79	1.8 (0.8–4.3)	0.14
# ED visits 2011	0.9 (0.5–1.9)	0.81
Discharged home	0.9 (0.4–2.0)	0.76
≥10 medications at discharge	1.3 (0.5–2.1)	0.93
Post-discharge appointment scheduled	1.7 (0.8–3.3)	0.15
Post-discharge appointment kept	1.5 (0.7–3.2)	0.27	2.1 (1.0–4.4)	0.06
Weekend discharge	1.7 (0.8–3.8)	0.19
Weekend readmission	0.66 (0.24–1.8)	0.42

ED, Emergency Department; HAART Rx, Highly active antiretroviral therapy prescription; ICU, Intensive care unit; ID, Infectious Diseases.

Of all readmissions, 67 (51%) were admitted with a diagnosis in the same category, including 55% who had an AIDS-defining illness. Figure 2 demonstrates the proportion of individuals with an index diagnosis who had a preventable readmission, starting with the category with the largest number of readmissions down to the category with the least.

FIG. 2.

Potentially preventable and nonpreventable readmissions by admitting diagnosis.

The phase of care (inpatient stay, discharge process, after discharge) analysis is presented in Table 3. Of the 62 preventable readmissions, 74% had an identifiable preventable cause for readmission during the initial inpatient stay, 29% during discharge, and 40% during the follow-up period. The most common reasons for preventable readmission were patient noncompliance, poor management of a chronic condition, complication from the initial hospitalization, or patient was discharged too soon.

Table 3.

Phases of Care and Causes of Potentially Preventable Readmissions

	Preventable readmissions (n=62)
Inpatient care
Poor management of chronic condition	19 (31%)
Complication of hospitalization	17 (27%)
Discharge too soon	16 (26%)
Missed/wrong diagnosis	13 (21%)
Any of the above	46 (74%)
Discharge process
Unaddressed psych/social needs	14 (22%)
Missed referrals	1 (1%)
Supplies/medication access	4 (6%)
Any of the above	18 (29%)
Follow-up
Delayed outpatient medical follow-up	3 (4%)
Missed lab result	2 (3%)
Noncompliance	23 (37%)
Any of the above	25 (40%)
Any phase of care process potentially related to readmission	62 (100%)

Discussion

In this study of patients with HIV hospitalized in a large, urban safety net teaching hospital, we found that 19% of all admissions resulted in readmissions (or 14% of individual patients experienced readmissions) within 30 days of discharge. Of the 130 readmitted individuals, we determined that approximately half were determined to have had a potentially preventable readmission using published criteria (53%) or implicit chart review (48%). To our knowledge, this is the first publication to address potentially preventable readmission rates in the HIV population. The rate of readmissions is comparable to prior studies of 30-day readmissions among patients living with HIV.^14,15 The rate of preventable readmissions is high when compared to a meta-analysis of studies including multiple medical conditions, with a median preventability rate of 27%,²² though it is similar to other studies involving detailed chart reviews.^32,33 The chart review method of defining potentially preventable readmissions found a somewhat lower rate of preventability than when the Goldfield/3M definition of preventable readmissions was used (48% vs. 53%), which is consistent with other studies comparing these methods.³⁰

Previous studies on preventable admissions have excluded patients with HIV,³⁰ and HIV is an excluded condition (along with malignancies, trauma, burns) in a 3M software product used by many state governments to assess potentially preventable readmissions.^28,29,31 However, our results suggest that many patients with HIV do have potentially preventable readmissions and ought not to be excluded from such analyses.

Not receiving ART was the main predictor of preventable readmissions in our multivariate analysis. ART is critical to HIV care, and has been found to significantly reduce morbidity and mortality in HIV patients.³⁴ Treatment with ART is associated with a lower rate of hospitalization^2,35,36 for both HIV-related and non-HIV-related admissions.³⁷ Similarly, not receiving ART when indicated has been associated with hospital readmission.³⁸ Other studies have found disparities in time to ART initiation, with longer time to ART in blacks, uninsured, heterosexual women, MSM, and persons without AIDS.³⁹

Unlike treatment for other chronic conditions, such as CHF, cirrhosis, or chronic obstructive pulmonary disease, HIV treatment can lead to significant reversal of the main underlying illness. In fact, due to improvement in tolerability and recognition of long-term consequences of viral replication, ART is recommended for virtually all patients with HIV.⁴⁰ Early initiation of ART in the setting of opportunistic infections improves survival;⁴¹ therefore, many patients have indications for ART initiation in the inpatient setting. In this study population, there were low rates of ART initiation, including in patients with AIDS-defining illnesses (18/30, 60%), which may in part be related to low rates of infectious diseases consultation. Other potential barriers to ART initiation are concerns about immune reconstitution inflammatory syndrome, drug interactions, and poor adherence to treatment,⁴² as well as heterogeneity in provider attitudes towards ART initiation in the inpatient setting.

Findings highlight the need for routine, multidisciplinary assessment of inpatients with HIV, including establishing policies for early ART initiation or re-initiation. Having an undetectable viral load tended to be protective against a preventable readmission [OR 0.6, (0.30–1.3), p=0.18], though this was not statistically significant. Patients with an undetectable viral load were less frequently admitted with AIDS related infections and more likely to have other causes of admission (endocrine, cardiovascular, obstetrical, trauma).

The associations between healthcare utilization and preventable readmissions reflect the complex relationship that exists between inpatient and outpatient care visits. In our study, attending an outpatient HIV visit in the year prior to admission was associated with preventable readmissions, though this was not statistically significant. Prior studies have found that outpatient HIV clinic utilization is associated with inpatient admission in a “J” shaped relationship⁴³—where patients with infrequent clinic visits or very high numbers of clinic visits were more likely to be admitted when compared to those with a moderate number of clinic visits. This may reflect that those with inadequate engagement in care and medically ill individuals who require many visits are more likely to be admitted and may have preventable readmissions.

In addition, we found that those who attended their post-hospitalization clinic visit were more likely to have a preventable readmission (borderline significant), which may be due to multiple patients who were directly readmitted from clinic. Overall these findings highlight the challenge of measuring appropriate engagement and retention in HIV care, which may be defined by metrics such as gaps in care, missed visits, or number of visits over time, though currently no gold standard exists.⁴⁴

Of note, of the patients with an appointment in the past year, 97% of these attended a clinic appointment in the past 6 months, which is an HIV/AIDS Bureau core performance measure. In addition, post-hospitalization follow-up visits may prevent some readmissions but can also result in early readmission, which could reflect preventable factors which occurred during the inpatient stay or discharge process, such as a patient being discharged too soon or unaddressed psychosocial needs.

We found that care during the inpatient stay, including poor management of a chronic condition, complications of hospitalization, missed diagnoses, and being discharged too soon, was most likely to contribute to preventable causes of readmission. In addition, unaddressed psychosocial needs during the discharge process and noncompliance in the follow-up period were other times where preventable causes of readmission were commonly identified. These findings suggest that an involved inpatient medical and psychosocial assessment, including a focus on the transition of care and adherence to ART, may impact readmissions in this population. Some prior studies have identified similar quality issues during the inpatient stay as contributing to readmissions,^45,46 whereas others found that preventable causes of readmission were present during the index stay, discharge process and follow-up care.³²

Our study has several limitations. First, our study was conducted at a single safety net hospital, therefore our findings may not apply to other settings. Second, due to the detailed nature of the data collection, including the determination of preventable causes of readmission, the sample size for this study was relatively small. Third, as this was a chart review, we were limited by what information was documented and available in the electronic medical record in order to determine if a readmission was preventable or not. Also, we were not able to capture events that were not documented, such as if ART was offered but refused by the patient. However, Parkland Hospital is an integrated delivery system of inpatient and outpatient care, therefore we had access to both inpatient and outpatient records within this system. Lastly, we were only able measure and analyze readmissions to Parkland Hospital and therefore likely underestimated the overall readmission rate, though the main focus of this study was preventability rather than determining a point estimate of readmissions.

In summary, we have demonstrated that nearly half of hospital readmissions among HIV patients are potentially preventable. Although this population is medically and often socially complex, this is not a reason to exclude these patients from analyses or from efforts to reduce readmissions. After adjusting for multiple clinical and social confounders, we found that the strongest predictor of a potentially preventable readmission was not receiving ART. Given the mounting evidence for universal treatment with ART and the benefits of early ART in the setting of acute opportunistic infections, a new approach to inpatient HIV management is indicated.

Our findings suggest that patient-centered interventions that include comprehensive diagnosis and treatment during the inpatient stay, evaluation of the need for and timing of ART, assessment of psychosocial needs, adherence counseling, and focused preparation for the transition to outpatient care may help reduce preventable readmissions. Ultimately, in order to address health disparities and improve HIV outcomes in this vulnerable population, we need to increase efforts to engage HIV patients starting during hospitalization.

Footnotes

Acknowledgments

We would like to acknowledge the HIV High Performance team at Parkland Hospital for their assistance with this project.

Funding sources: KL2TR001103 (National Center for Advancing Translation Sciences); K23AI112477 (National Institute of Allergy and Infectious Diseases) (Nijhawan); 1R24HS022418 (Agency for Healthcare Research and Quality) (Halm).

Author Disclosure Statement

No conflicting financial interests exist.

References

Bachhuber

, Southern

. Hospitalization rates of people living with HIV in the United States, 2009. Public Health Rep, 2014; 129:178–186.

Crum-Cianflone

, Grandits

, Echols

, et al. Trends and causes of hospitalizations among HIV-infected persons during the late HAART era: What is the impact of CD4 counts and HAART use?. J Acquir Immune Defic Syndr, 2010; 54:248–257.

Hellinger

. HIV patients in the HCUP database: A study of hospital utilization and costs. Inquiry, 2004; 41:95–105.

Gebo

, Diener-West

, Moore

. Hospitalization rates in an urban cohort after the introduction of highly active antiretroviral therapy. J Acquir Immune Defic Syndr, 2001; 27:143–152.

Fleishman

, Hellinger

. Recent trends in HIV-related inpatient admissions 1996–2000: A 7-state study. J Acquir Immune Defic Syndr, 2003; 34:102–110.

Fielden

, Rusch

, Levy

, et al. Predicting hospitalization among HIV-infected antiretroviral naive patients starting HAART: Determining clinical markers and exploring social pathways. AIDS Care, 2008; 20:297–303.

Weber

, Yip

, O'Shaughnessy

, et al. Determinants of hospital admission among HIV-positive people in British Columbia. Can Med Assoc J, 2000; 162:783–786.

Oramasionwu

, Hunter

, Skinner

, et al. Black race as a predictor of poor health outcomes among a national cohort of HIV/AIDS patients admitted to US hospitals: A cohort study. BMC Infect Dis, 2009; 9:127.

Buchacz

, Baker

, Moorman

, et al. Rates of hospitalizations and associated diagnoses in a large multisite cohort of HIV patients in the United States, 1994–2005. AIDS, 2008; 22:1345–1354.

10.

Himelhoch

, Chander

, Fleishman

, et al. Access to HAART and utilization of inpatient medical hospital services among HIV-infected patients with co-occurring serious mental illness and injection drug use. Gen Hosp Psychiat, 2007; 29:518–525.

11.

Kerr

, Stephens

, Gibson

, et al. Risk factors associated with inpatient hospital utilization in HIV-positive individuals and relationship to HIV care engagement. J Acquir Immune Defic Syndr, 2012; 60:173–182.

12.

Zhao

, Encinosa

, Hellinger

. HIV Hospitalizations in 1998 and 2005: Statistical Brief #41. Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville, MD, 2006.

13.

Kanel

, Elster

, Vrbin

. PRHI Readmission Brief, Brief II: Patterns of Hospital Admission and Readmission Among HIV-Positive Patients in Southwestern Pennsylvania. 2010. Available at: http://www.chqpr.org/downloads/PRHI_ReadmissionBrief_HIV_July2010.pdf (Last accessed August 27, 2013 ).

14.

Nijhawan

, Clark

, Kaplan

, et al. An electronic medical record-based model to predict 30-day risk of readmission and death among HIV-infected inpatients. J Acquir Immune Defic Syndr, 2012; 61:349–358.

15.

Berry

, Fleishman

, Yehia

, et al. Thirty-day hospital readmission rate among adults living with HIV. AIDS, 2013; 27:2059–2068.

16.

Jencks

, Williams

, Coleman

. Rehospitalizations among patients in the Medicare fee-for-service program. New Eng J Med, 2009; 360:1418–1428.

17.

Joynt

, Orav

, Jha

. Thirty-day readmission rates for Medicare beneficiaries by race and site of care. JAMA, 2011; 305:675–681.

18.

Krumholz

, Parent

, Tu

, et al. Readmission after hospitalization for congestive heart failure among Medicare beneficiaries. Arch Intern Med, 1997; 157:99–104.

19.

Amarasingham

, Moore

, Tabak

, et al. An automated model to identify heart failure patients at risk for 30-day readmission or death using electronic medical record data. Med Care, 2010; 48:981–988.

20.

Krumholz

, Lin

, Drye

, et al. An administrative claims measure suitable for profiling hospital performance based on 30-day all-cause readmission rates among patients with acute myocardial infarction. Circ Cardiovasc Qual Outcomes, 2011; 4:243–252.

21.

Lindenauer

, Normand

, Drye

, et al. Development, validation, and results of a measure of 30-day readmission following hospitalization for pneumonia. J Hosp Med, 2011; 6:142–150.

22.

van Walraven

, Bennett

, Jennings

, et al. Proportion of hospital readmissions deemed avoidable: A systematic review. Can Med Assoc J, 2011; 183:E391–E402.

23.

Goldfield

, McCullough

, Hughes

, et al. Identifying potentially preventable readmissions. Health Care Financ Rev, 2008; 30:75–91.

24.

Friedman

, Basu

. The rate and cost of hospital readmissions for preventable conditions. Med Care Res Rev, 2004; 61:225–240.

25.

Haines-Wood

, Gilmore

, Beringer

. Re-admission of elderly patients after in-patient rehabilitation. Ulster Med J, 1996; 65:142–144.

26.

Madigan

, Schott

, Matthews

. Rehospitalization among home healthcare patients: Results of a prospective study. Home Healthcare Nurse, 2001; 19:298–305.

27.

Vinson

, Rich

, Sperry

, et al. Early readmission of elderly patients with congestive heart failure. J Am Geriatr Soc, 1990; 38:1290–1295.

28.

THHSC.. Potentially Preventable Readmissions in the Texas Medicaid Population, State Fiscal Year 2011 2012. Available at: http://www.hhsc.state.tx.us/reports/2012/PPR-Readmissions-FY2011.pdf (Last accessed August 10, 2013 ).

29.

Department of Healthcare and Family Services Illinois.. Potentially Preventable Readmissions Policy. Available at: http://www2.illinois.gov/hfs/SiteCollectionDocuments/PPR_Overview.pdf (Last accessed September 16, 2014 ).

30.

Jackson

, Fireman

, Feigenbaum

, et al. Manual and automated methods for identifying potentially preventable readmissions: A comparison in a large healthcare system. BMC Med Inform Decision Making, 2014; 14:28.

31.

3M Health Information Systems.. Potentially Preventable Readmissions- A Methodology Overview. Education Session for Texas Hospitals 2011. Available at: http://www.tmhp.com/Workshop_Materials/Potentially%20Preventable%20Readmissions%20%28PPR%29%20Reports/Texas%20PPR%20Methodolgy%20Overview.pdf (Last accessed September 16, 2014 ).

32.

Feigenbaum

, Neuwirth

, Trowbridge

, et al. Factors contributing to all-cause 30-day readmissions: A structured case series across 18 hospitals. Med Care, 2012; 50:599–605.

33.

Williams

, Fitton

. Factors affecting early unplanned readmission of elderly patients to hospital. BMJ, 1988; 297:784–787.

34.

Palella

Jr , Delaney

, Moorman

, et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV Outpatient Study Investigators. N Engl J Med, 1998; 338:853–860.

35.

Baum

, Morris

, Gibbons

, et al. Reduction in human immunodeficiency virus patient hospitalizations and nontraumatic mortality after adoption of highly active antiretroviral therapy. Military Med, 1999; 164:609–612.

36.

Sax

, Meyers

, Mugavero

, et al. Adherence to antiretroviral treatment and correlation with risk of hospitalization among commercially insured HIV patients in the United States. PLoS One, 2012; 7:e31591.

37.

Floris-Moore

, Lo

, Klein

, et al. Gender and hospitalization patterns among HIV-infected drug users before and after the availability of highly active antiretroviral therapy. J Acquir Immune Defic Syndr, 2003; 34:331–337.

38.

Nosyk

, Sun

, Li

, et al. Highly active antiretroviral therapy and hospital readmission: Comparison of a matched cohort. BMC Infect Dis, 2006; 6:146.

39.

Adedinsewo

, Wei

, Robertson

, et al. Timing of antiretroviral therapy initiation in a nationally representative sample of HIV-infected adults receiving medical care in the United States. AIDS Patient Care STDS, 2014; 28:613–621.

40.

Panel on Antiretroviral Guidelines for Adults and Adolescents.. Guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents. Available at: http://aidsinfo.nih.gov/contentfiles/lvguidelines/AdultandAdolescentGL.pdf (Last accessed September 13, 2014 ).

41.

Zolopa

, Andersen

, Powderly

, et al. Early antiretroviral therapy reduces AIDS progression/death in individuals with acute opportunistic infections: A multicenter randomized strategy trial. PLoS One, 2009; 4:e5575.

42.

Grant

, Zolopa

. When to start ART in the setting of acute AIDS-related opportunistic infections: The time is now! Curr HIV/AIDS Rep, 2012; 9:251–258.

43.

Fleishman

, Moore

, Conviser

, et al. Associations between outpatient and inpatient service use among persons with HIV infection: A positive or negative relationship?. Health Serv Res, 2008; 43:76–95.

44.

Mugavero

, Westfall

, Zinski

, et al. Measuring retention in HIV care: The elusive gold standard. J Acquir Immune Defic Syndr, 2012; 61:574–580.

45.

Balla

, Malnick

, Schattner

. Early readmissions to the department of medicine as a screening tool for monitoring quality of care problems. Medicine, 2008; 87:294–300.

46.

Scott

, Shohag

, Ahmed

. Quality of care factors associated with unplanned readmissions of older medical patients: A case-control study. Int Med J, 2014; 44:161–170.