Abstract
Background
Antiretroviral therapy (ART) advances, aging, and comorbidities impact hospitalizations in human immunodeficiency virus (HIV)–positive populations. We examined temporal trends and patient characteristics associated with hospitalization rates and outcomes.
Methods
Among patients in the University of North Carolina Center for AIDS Research HIV Clinical Cohort receiving care during 1996–2016, we estimated annual hospitalization rates, time to inpatient mortality or live discharge, and 30-day readmission risk using bivariable Poisson, Fine-Gray, and log-binomial regression models.
Results
The 4323 included patients (29% women, 60% African American) contributed 30 007 person-years. Overall, the hospitalization rate per 100 person-years was 34.3 (95% confidence interval [CI], 32.4–36.4) with a mean annual change of −3% (95% CI, −4% to −2%). Patients who were black (vs white), older, had HIV RNA >400 copies/mL, or had CD4 count <200 cells/μL had higher hospitalization rates (all P < .05). Thirty-day readmission risk was 18.9% (95% CI, 17.7%–20.2%), stable over time (P > .05 for both 2010–2016 and 2003–2009 vs 1996–2002), and higher among black patients, those with detectable HIV RNA, and those with lower CD4 cell counts (all P < .05). Higher inpatient mortality was associated with older age and lower CD4 cell count (both P < .05).
Conclusions
Hospitalization rates decreased from 1996 to 2016, but high readmissions persisted. Older patients, those of minority race/ethnicity, and those with uncontrolled HIV experienced higher rates and worse hospitalization outcomes. These findings underscore the importance of early ART and care engagement, particularly at hospital discharge.
Keywords: HIV, hospitalization, length of stay, hospital readmission, clinical cohort
In an HIV clinical cohort in the southeastern United States, hospitalization rates have decreased since 1996 but readmission risk remains high. Patients of minority race/ethnicity or with uncontrolled HIV experienced higher rates and worse hospitalization outcomes.
(See the Editorial Commentary by Colasanti and del Rio on pages 1624–6.)
Hospitalization rates among persons living with human immunodeficiency virus (PLWH) declined sharply after the introduction of combination antiretroviral therapy (ART) in 1996, due largely to decreases in AIDS-related admissions [1, 2]. Some studies have shown a plateau of hospitalizations in the 2000s, others a persistent decrease in rates, and even an increase in some regions or for specific hospitalization causes [2–7]. Few studies have examined more recent hospitalization trends and risk factors among PLWH, although these have likely been impacted by changes in human immunodeficiency virus (HIV) care over the last decade. In 2012, United States (US) guidelines began recommending earlier ART initiation, which prevents serious AIDS and non-AIDS conditions [8]. However, the population of PLWH is aging and experiences a substantial burden of cardiovascular disease, cancer, end-stage renal disease, end-stage liver disease, and multimorbidity, which may increase hospitalization risk [9–14]. Notably, PLWH experience higher hospital mortality and readmission rates compared with HIV-negative individuals, but few studies have examined risk factors for these worse hospitalization outcomes in HIV-positive populations [15, 16]. Understanding trends and risk factors in hospitalization rates and outcomes among PLWH can help improve management strategies to prevent morbidity that requires inpatient care and may lead to poor clinical outcomes. In this study, we examined hospitalization rates, length of stay, inpatient mortality, and readmission risk between 1996 and 2016 in a clinical cohort of PLWH in the southeastern US.
METHODS
Study Population
This study was based in the University of North Carolina (UNC) Center for AIDS Research HIV Clinical Cohort (UCHCC), which prospectively collects laboratory tests, diagnoses, hospitalizations, and medication data from electronic health records and twice-yearly health record reviews, including non-UNC records obtained by UNC Hospitals. UCHCC participants in care during 1996–2016 contributed person-time from the latter of 1 January 1996 or initiation of HIV care at UNC, until death or 31 December 2016, whichever occurred earlier. Patients were censored at loss to follow-up, defined as 1.5 years with no clinic or laboratory visit, but contributed additional person-time if they reentered HIV care. Data collection and secondary analysis were both approved by the UNC Institutional Review Board, and patients provided written informed consent to participate in the UCHCC.
Study Definitions
For hospitalization rates, the 2 outcomes of interest were all-cause hospitalization and hospitalization with a CD4 count <100 cells/µL. Because discharge diagnoses were not available in this study, we used a proximal CD4 count <100 cells/µL, measured in the 9 months prior to hospitalization, as an indicator that hospitalization was likely AIDS related. The cutoff of 100 cells/µL was selected to minimize the possibility of classifying hospitalizations with other causes as AIDS related. We included hospitalizations with same-day discharge or taking place at non-UNC medical centers. For hospitalization outcomes we examined time from admission to live discharge or inpatient death as competing events, and 30-day risk of readmission. Hospitalization outcomes analyses were restricted to admissions at UNC Hospitals to ensure exact dates were available. For readmissions, index hospitalizations were defined as those with a live discharge that either were the first observed hospitalization or had no hospitalization in the prior 30 days. Readmissions taking place at non-UNC hospitals were counted if exact dates were available. In sensitivity analyses, we excluded hospitalizations among women at the time of a known delivery.
Statistical Analysis
Hospitalization rates were calculated as the number of hospitalizations divided by the person-time at risk, for the study period and each calendar year, among all patients and demographic subgroups. Using Poisson regression models, we estimated incidence rate ratios (IRRs) to assess calendar time trends and to compare rates by patient demographic and clinical characteristics, including sexual risk group, race/ethnicity, injection drug use (IDU) risk factor, age, having an HIV RNA load >400 copies/mL, and having a CD4 count <200 cells/µL. We selected a threshold of 200 cells/µL to identify patients at high risk of severe HIV-related morbidity. IRRs for calendar time trends were expressed as a percentage change in rate. Viral load and CD4 cell count measurements from the previous calendar year were used in regression models. CD4 cell count was only examined as a risk factor for all-cause hospitalization.
Time to live discharge or inpatient death as competing events was estimated using the Aalen-Johansen method. Subdistribution hazard ratios (sHRs) were estimated from Fine-Gray models to assess calendar time trends and compare patient characteristics. For readmissions, we used log-binomial regression to estimate risk ratios (RRs) comparing calendar periods and patient characteristics. Laboratory values for hospitalization outcomes analyses were the closest measurement in the 9 months prior to hospitalization.
CD4 count was missing for 11% of patient-years and 7% of hospitalizations, and viral load for 13% and 9%, respectively. These were excluded from models with these variables but included in other analyses. To account for patients contributing more than once to each analysis, generalized estimating equations were used in Poisson and log-binomial models, and Sandwich estimators were used in Fine-Gray models. Models comparing patient characteristics were adjusted for calendar time only. All P values were 2-sided, and P < .05 was considered statistically significant. Analyses were conducted using SAS software, version 9.4 (SAS Institute, Cary, North Carolina).
RESULTS
Patient Characteristics
Of 4323 included patients, 1250 (29%) were women, 1754 (41%) were men who have sex with men (MSM), and 587 (14%) had IDU as an HIV risk factor. There were 2615 (60%) black patients, 1323 (31%) white patients, 238 (6%) Hispanic patients, and 147 (3%) patients of other race/ethnicity. At the start of follow-up, the median calendar year was 2002 (interquartile range [IQR], 1998–2008), age 37 years (IQR, 30–45), and nadir CD4 count 168 cells/μL (IQR, 39–349). Patients contributed a median of 5.2 years of follow-up (IQR, 2.5–9.8) and a total of 30 007 person-years (PY). Demographic characteristics of the patient sample at the start of follow-up, compared to the contribution of each group by PY of follow-up, were similarly distributed, except for a higher median age across PY (Table 1). Of patient-years with available measurements, 16% were contributed by patients with a CD4 count <200 cells/µL, and 33% by patients with a viral load >400 copies/mL. From 1996 to 2016, median patient age increased from 36 years (IQR, 31–43) to 49 years (IQR, 39–57), and CD4 count from 290 cells/μL (IQR, 130–460) to 634 cells/µL (IQR, 422–843) (Table 1, both P < .01). In addition, over time more patients had an available HIV RNA measurement, and a lower proportion of these were >400 copies/mL (P < .01).
Table 1.
Characteristics of Patient-years of Follow-up and Number of Participants in Care Across Study Years, 1996–2016
| Characteristic | Total PYFU (N = 30 007) | No. of Participants in Care | |||||
|---|---|---|---|---|---|---|---|
| 1996 (n = 713) | 2001 (n = 1686) | 2006 (n = 1690) | 2011 (n = 1835) | 2016 (n = 1823) | P Valuea | ||
| Sexual risk group | < .01 | ||||||
| Heterosexual men | 8652 (29) | 227 (32) | 571 (34) | 521 (31) | 468 (26) | 423 (23) | |
| Women | 9031 (30) | 222 (31) | 543 (32) | 525 (31) | 537 (29) | 523 (29) | |
| MSM | 12 324 (41) | 264 (37) | 572 (34) | 644 (38) | 830 (45) | 877 (48) | |
| Race/ethnicity | < .01 | ||||||
| White | 9698 (32) | 272 (38) | 526 (31) | 531 (31) | 577 (31) | 572 (31) | |
| Black | 17 720 (59) | 398 (56) | 1021 (61) | 1018 (60) | 1081 (59) | 1080 (59) | |
| Hispanic | 1687 (6) | 13 (2) | 70 (4) | 88 (5) | 135 (7) | 126 (7) | |
| Other | 902 (3) | 30 (4) | 59 (4) | 53 (3) | 42 (2) | 45 (2) | |
| Injection drug useb | 3958 (13) | 138 (19) | 297 (18) | 234 (14) | 193 (11) | 152 (8) | < .01 |
| Age, y, median (IQR) | 44 (36–51) | 36 (31–43) | 40 (34–46) | 44 (37–50) | 46 (37–53) | 49 (39–57) | < .01 |
| HIV RNA >400 copies/mL | 8555 (33) | 18 (82) | 665 (51) | 544 (40) | 273 (17) | 130 (8) | < .01 |
| Missing | 3866 | 691 | 392 | 338 | 266 | 202 | |
| CD4 count <200 cells/μL | 4277 (16) | 150 (34) | 324 (25) | 255 (19) | 180 (11) | 105 (7) | < .01 |
| Missing | 3203 | 274 | 382 | 336 | 247 | 272 | |
Data are presented as no. (%) unless otherwise indicated.
Abbreviations: HIV, human immunodeficiency virus; IQR, interquartile range; MSM, men who have sex with men; PYFU, person-years of follow-up.
a P values compare calendar years 1996, 2001, 2006, 2011, and 2016, estimated using χ 2 tests for categorical variables and Kruskal-Wallis tests for continuous variables.
bInjection drug use as a risk factor for HIV acquisition.
Trends in Hospitalization Rates
During follow-up, patients experienced a median of 1 (IQR, 0–3) hospitalizations and a total of 10 310 hospitalizations. The proximal CD4 count was <100 cells/µL in 28% of hospitalizations, and these are considered likely AIDS related. Over the study period, the hospitalization rate per 100 PY was 34.3 (confidence interval [CI], 32.4–36.4) for all-cause hospitalizations and 9.5 (95% CI, 8.6–10.4) for hospitalizations with CD4 count <100 cells/µL. The all-cause hospitalization rate per 100 PY decreased from 50.1 (95% CI, 42.8–58.6) in 1996 to 22.7 (95% CI, 19.5–26.5) in 2016 (mean change per year, −3% [95% CI, −4% to −2%]; Figure 1A). For hospitalizations with CD4 count <100 cells/µL, the rate per 100 PY decreased from 21.0 (95% CI, 16.0–27.5) in 1996 to 2.6 (95% CI, 1.6–4.3) in 2016 (mean change per year, −7% [95% CI, −8% to −6%]; Figure 1A).
Figure 1.
Annual hospitalization rates and mean rate change, 1996–2016, for all-cause hospitalizations (blue) and hospitalizations with CD4 count <100 cells/µL (red) among all patients (A), black patients (B), white patients (C), women (D), men who have sex with men (E), heterosexual men (F), patients aged <50 years (G), and patients aged ≥50 years (H). Mean annual rate change was estimated using Poisson regression models. In the absence of discharge diagnosis data, a CD4 count <100 cells/µL, measured in the 9 months prior to hospitalization, was used as an indicator that hospitalization was likely AIDS related. Abbreviation: CI, confidence interval.
Declines in rates over time were similar when stratifying by race/ethnicity, sexual risk group, and patient age (Figure 1B–H). Among heterosexual men and MSM combined (not shown), the mean change per year was −4% (95% CI, −5% to −3%) for all-cause hospitalizations and −10% (95% CI, −12% to −9%) for hospitalizations with CD4 count <100 cells/μL. Among Hispanic patients (not shown), the mean change per year was −6% (95% CI, −9% to −3%) for all-cause hospitalizations and −12% (95% CI, −17% to −6%) for hospitalizations with CD4 count <100 cells/μL. See Supplementary Table 1 for 1996 and 2016 stratified rates.
Risk Factors for Hospitalization
Adjusting for calendar year, MSM had lower hospitalization rates than heterosexual men with an IRR of 0.73 (95% CI, .63–.84) for all-cause hospitalizations and a similar estimate for hospitalizations with CD4 count <100 cells/µL, while women did not have significantly different rates compared to heterosexual men (Table 2). Comparing women to all men, the IRR was 1.12 (95% CI, .99–1.26) for all-cause hospitalizations, and 0.99 (95% CI, .80–1.22) for hospitalizations with CD4 count <100 cells/µL. Compared to white patients, black patients had higher hospitalization rates, with an IRR of 1.54 (95% CI, 1.36–1.75) for all-cause hospitalizations and 2.24 (95% CI, 1.77–2.84) for hospitalizations with CD4 count <100 cells/µL. Hispanic ethnicity was associated with higher rates only for hospitalizations with CD4 count <100 cells/µL (IRR, 1.89 [95% CI, 1.24–2.86]). Older age was associated with higher rates of all-cause hospitalization, but with lower rates of hospitalization with CD4 count <100 cells/µL. IDU risk factor was associated with higher rates of all-cause hospitalization only. Patients with a recent viral load >400 copies/mL, compared to those with <400 copies/mL, had higher rates for all-cause hospitalizations (IRR, 2.00 [95% CI, 1.79–2.24]) and hospitalizations with CD4 count <100 cells/µL (IRR, 11.70 [95% CI, 9.07–15.09]). Patients with a recent CD4 count <200 cells/µL, compared to those with ≥200 cells/µL, had an IRR of 3.32 (95% CI, 2.97–3.70) for all-cause hospitalizations.
Table 2.
Factors Associated With Increased Hospitalization Rates
| Characteristic | All-cause Hospitalizations | Hospitalizations With CD4 Count <100 Cells/μLa |
|---|---|---|
| IRR (95% CI)b | IRR (95% CI)b | |
| Sexual risk group | ||
| Heterosexual men | 1 (ref) | 1 (ref) |
| Women | 0.94 (.83–1.08) | 0.85 (.68–1.06) |
| MSM | 0.73 (.63–.84) | 0.74 (.59–.93) |
| Race/ethnicity | ||
| White | 1 (ref) | 1 (ref) |
| Black | 1.54 (1.36–1.75) | 2.24 (1.77–2.84) |
| Hispanic | 1.00 (.79–1.28) | 1.89 (1.24–2.86) |
| Other | 1.41 (.99–1.99) | 1.38 (.81–2.36) |
| Injection drug usec | 1.54 (1.34–1.77) | 1.15 (.90–1.46) |
| Aged | 1.07 (1.02–1.13) | 0.81 (.74–.87) |
| HIV RNA >400 copies/mLe,f | 2.00 (1.79–2.24) | 11.70 (9.07–15.09) |
| CD4 count <200 cells/μLe,g | 3.32 (2.97–3.70) | NAh |
Abbreviations: CI, confidence interval; HIV, human immunodeficiency virus; IRR, incidence rate ratio; MSM, men who have sex with men; NA, not applicable; ref, reference.
aIn the absence of discharge diagnosis data, a CD4 count <100 cells/µL, measured up to 9 months prior and including hospital admission date, was used as an indicator that hospitalization was likely AIDS related.
bIRRs, 95% CIs, and P values from Poisson regression models including only 1 characteristic and calendar year.
cInjection drug use as a risk factor for HIV acquisition.
dPer 10-year increase.
eLagged by a year.
fExcludes 3866 person-years missing an HIV RNA measurement.
gExcludes 3203 person-years missing a CD4 cell count measurement.
hCD4 count was not examined as a risk factor for likely AIDS-associated hospitalizations, as CD4 cell count was used to define this outcome.
Inpatient Length of Stay and Mortality
Outcomes were examined for 7501 admissions among 2006 patients, including 630 (31%) women, 703 (35%) MSM, and 1263 (63%) black patients (Table 3). Overall, 1.7% of hospitalizations resulted in death, and the average duration of stay, irrespective of discharge status, was 7 days (range, 1–216). In time-to-event analyses with competing risks, the median time to live discharge was 5 days (IQR, 3–8), with 1.5% (95% CI, 1.2%–1.8%) inpatient mortality in the first 30 days of hospitalization. Time to live discharge decreased over the study period, with an sHR of 1.09 (95% CI, 1.01–1.17) comparing the years 2010–2016 to 1996–2002, signifying a 9% increase in the discharge rate (Table 3). Adjusting for calendar period, time to live discharge was longer for Hispanic vs white patients, with an sHR of 0.84 (95% CI, .73–.97). Black patients did not have a significantly different time to discharge compared to white patients. Women had a shorter time to live discharge than heterosexual men (sHR, 1.08 [95% CI, 1.01–1.16]). Comparing women to all men, the sHR was 1.05 (95% CI, .98–1.12]). Inpatient mortality did not vary by calendar period, racial/ethnic group, or sexual risk group. Older age was not associated with time to live discharge, but for a 10-year age increase, the sHR for inpatient mortality was 1.18 (95% CI, 1.01–1.38), signifying an 18% increase in the inpatient mortality rate. Having a CD4 count <200 cells/µL at admission was associated with longer time to live discharge (sHR, 0.72 [95% CI, .68–.77]) and higher inpatient death (sHR 2.06 [95% CI, 1.41–3.01]). Patients with a viral load >400 copies/mL at admission had longer time to live discharge (sHR, 0.84 [95% CI, .79–.89]), but no association with inpatient mortality.
Table 3.
Hospitalization Characteristics and Factors Associated With Hospitalization Outcomes
| Characteristic at Admission | No. (%) (N = 7501 Admissions) | Live Discharge | Inpatient Mortality | 30-day Readmission |
|---|---|---|---|---|
| sHR (95% CI)a | sHR (95% CI)a | RR (95% CI)b | ||
| Sexual risk group | ||||
| Heterosexual men | 2633 (35) | 1 (ref) | 1 (ref) | 1 (ref) |
| Women | 2352 (31) | 1.08 (1.01–1.16) | 0.68 (.44–1.03) | 0.92 (.79–1.08) |
| MSM | 2516 (34) | 1.06 (.99–1.15) | 0.74 (.49–1.11) | 0.92 (.78–1.08) |
| Race/ethnicity | ||||
| White | 1778 (24) | 1 (ref) | 1 (ref) | 1 (ref) |
| Black | 5116 (68) | 0.97 (.90–1.04) | 0.88 (.59–1.30) | 1.18 (1.01–1.37) |
| Hispanic | 359 (5) | 0.84 (.73–.97) | 1.08 (.51–2.32) | 1.15 (.85–1.56) |
| Other | 248 (3) | 0.83 (.71–.97) | 1.29 (.50–3.33) | 1.09 (.78–1.51) |
| Injection drug usec | 1463 (20) | 0.98 (.91–1.05) | 1.23 (.81–1.88) | 1.05 (.88–1.26) |
| Age, yd | 43 (36–50)e | 0.99 (.97–1.02) | 1.18 (1.01–1.38) | 0.98 (.93–1.04) |
| Calendar period | ||||
| 1996–2002 | 2530 (34) | 1 (ref) | 1 (ref) | 1 (ref) |
| 2003–2009 | 2760 (37) | 1.07 (1.00–1.14) | 0.88 (.60–1.29) | 1.09 (.95–1.25) |
| 2010–2016 | 2211 (29) | 1.09 (1.01–1.17) | 0.67 (.43–1.04) | 1.06 (.91–1.24) |
| CD4 count <200 cells/μLf | 3393 (49) | 0.72 (.68–.77) | 2.06 (1.41–3.01) | 1.48 (1.29–1.70) |
| HIV RNA >400 copies/mLf | 3644 (43) | 0.84 (.79–.89) | 0.91 (.61–1.36) | 1.17 (1.03–1.33) |
Abbreviations: CI, confidence interval; HIV, human immunodeficiency virus; MSM, men who have sex with men; ref, reference; RR, risk ratio; sHR, subdistribution hazard ratio.
asHRs, 95% CIs, and P values from proportional hazards models estimating time from hospital admission to live discharge with inpatient mortality as a competing risk. Models include only 1 characteristic and calendar period. An sHR <1 for discharge signifies longer hospitalization duration and sHR >1 for mortality signifies shorter time to inpatient death.
bRRs, 95% CIs, and P values from log-binomial models including only 1 characteristic and calendar period. Readmission risk was estimated only for index hospitalizations (n = 5759), defined as the first observed hospitalization or any hospital admission without a hospitalization in the prior 30 days.
cInjection drug use as a risk factor for HIV acquisition.
dIn models, per 10-year increase.
eReported as median (interquartile range).
fClosest laboratory measurement in the 9 months prior to and including hospital admission date. Excludes 533 (7%) hospitalizations missing CD4 cell count and 685 (9%) missing HIV RNA.
Thirty-day Readmission
Of 5759 index hospitalizations, 18.9% (95% CI, 17.7%–20.2%) resulted in readmission within 30 days, one-third of which occurred within 7 days. Adjusting for calendar period, readmission risk was higher for black than white patients (RR, 1.18 [95% CI, 1.01–1.37]; Table 3). Patients with a CD4 count <200 cells/µL or viral load >400 copies/mL at index hospitalization also had higher readmission risk, with RRs of 1.48 (95% CI, 1.29–1.70) and 1.17 (95% CI, 1.03–1.33), respectively. There was no association between calendar period or any other patient characteristic and 30-day readmission risk.
Sensitivity Analysis
After excluding 127 hospitalizations around the time of a known delivery, calendar time trends in all-cause hospitalizations among women were identical to the main findings, and the IRR for all-cause hospitalizations was 0.91 (95% CI, .79–1.04) comparing women to heterosexual men. Estimates for time to discharge were similar but less precise than main findings, with a sHR of 1.06 (95% CI, .99–1.14) comparing women to heterosexual men. Estimates comparing inpatient mortality and readmission risk were similar to the main findings.
DISCUSSION
In the UCHCC, between 1996 and 2016, hospitalization rates decreased 3% annually and 55% overall, driven by an 88% decrease in hospitalizations with proximal CD4 count <100 cells/µL, considered to be likely AIDS related. Trends were similar across demographic and risk groups. Adjusting for calendar period, patients of minority race/ethnicity, and those with uncontrolled HIV experienced higher hospitalization rates, longer stays, and higher 30-day readmission risk. Older age was associated with higher all-cause hospitalization rates and higher inpatient mortality.
After the introduction of combination ART in 1996, hospitalization rates per 100 PY in the US and Canada declined to 15–40 in 2001, and further to 17–27 in 2008, with substantial variation between populations [1, 2, 4, 6, 17, 18]. In the UCHCC, rates continued to decrease through 2016, reaching 23 admissions per 100 PY. In prior studies, declines were driven by a substantial decrease in hospitalizations for AIDS-defining illnesses, consistent with decreasing rates of hospitalizations with CD4 count <100 cells/µL in our study [1, 6]. More recent studies have generally found decreasing or stable hospitalization rates for non-AIDS causes, with a few reporting increasing rates due to cancer, cardiovascular disease, and renal disease [6, 7, 19, 20]. Aging and greater comorbid burden among PLWH could be expected to increase non-AIDS hospitalizations [9–14], though heightened awareness of these issues and better preventive care could be mitigating factors. In our study, all-cause hospitalizations decreased even among patients aged ≥50 years, suggesting UCHCC patients, despite age-associated comorbidities, are receiving preventive care and outpatient management of chronic conditions. However, the 2016 hospitalization rate is twice that of the general population, and close to rates among Medicare beneficiaries [21, 22], possibly due to higher prevalences of frailty and age-associated comorbidities among PLWH compared to age-matched HIV-negative persons [23, 24].
Despite substantial decreases over the study period in all patient groups, disparities persist in rates of hospitalization with CD4 count <100 cells/µL, which were more frequent among black and Hispanic patients compared to white patients, and among patients with detectable viral loads. These findings, consistent with prior studies, may reflect late diagnosis, challenges to care engagement, and unequal healthcare access [1, 5, 6, 18, 20, 25, 26]. In contrast, older age and being MSM were associated with lower rates of hospitalization with CD4 count <100 cells/µL, possibly due to better treatment adherence, clinic attendance, and access to care [27, 28]. A history of IDU was associated with higher rates of all-cause hospitalizations but not hospitalizations with CD4 count <100 cells/µL, consistent with studies showing IDU is associated with higher rates of non-AIDS hospitalizations [3, 5, 6, 26]. This could reflect poor care engagement or be a marker for ongoing IDU, which may lead to hospital admissions for non-AIDS infections, overdose, psychiatric conditions, or complications of hepatitis C coinfection [17, 29]. Prior studies have reported higher hospitalization rates in women [1, 6, 26]. In our study, women had a small increase in rates compared to men that was not statistically significant.
Inpatient mortality was low overall as previously reported in similar settings, and close to general population estimates [15, 18, 30, 31], but it did not improve over time. It is possible that sicker patients with AIDS at the beginning of our study period were discharged to hospice care, leading to similar inpatient mortality across calendar periods [32]. Only older age and low CD4 cell count were associated with inpatient death, likely reflecting more severe condition at admission. Hospital length of stay decreased slightly over the study period, whereas 30-day readmission risk remained stable at almost 20%, consistent with other reports [16, 33, 34]. In one study, readmission risk was 9 percentage points higher for PLWH than for HIV-negative patients [16]. We did not find that patients with IDU risk factor experienced worse hospitalization outcomes, unlike studies showing drug use was associated with longer stays and readmissions, a discrepancy that could be due to using a historical rather than current measure of IDU [4, 33, 35]. Markers of uncontrolled HIV, CD4 count <200 cells/µL and viral load >400 copies/mL, were associated with longer stays and higher readmission risk in our study as in others [5, 30, 35–37]. Patients with these characteristics may face several challenges to care engagement, such as inadequate health insurance, housing and food insecurity, substance use, and mental health disorders. One study found that low CD4 cell count and lack of ART were associated with a readmission being preventable and that among the most likely causes of readmission were poor chronic condition management and not having psychological or social needs met during the index hospitalization [34].
This study’s strengths include 2 decades of follow-up from a large clinical cohort in the Southeast, the US region most impacted by HIV. Our results may not be generalizable to other settings. North Carolina did not expand Medicaid coverage under the Patient Protection and Affordable Care Act, which, compared to other states, might have impacted access to care and emergency department utilization in later study years. An additional strength is the capture of hospitalizations at external medical centers, meaning rates and readmissions are less likely to be underestimated than in studies limited to single hospital systems. A UCHCC analysis found that missing external hospitalizations could underestimate rates by up to 9 admissions per 100 PY, and a New York State study found that 33% of readmissions occurred at a different hospital [33, 38]. Another strength of our study is the competing risk approach used to examine length of stay and inpatient mortality, shown to be less biased than traditional methods [39]. However, an important limitation is the lack of data on discharge diagnoses. Relying on low CD4 cell counts as proxies for AIDS-related diagnoses may underestimate the burden of AIDS hospitalizations. Our sensitivity analysis may not have excluded all delivery-related hospitalizations among women, which could bias trend estimates and comparisons to men. Finally, we did not have data on substance use, mental health disorders, or socioeconomic status, which are important risk factors for hospitalization and readmission [1, 33, 40]. Future studies should examine these factors as well as hospitalization causes over time.
CONCLUSIONS
From 1996 to 2016, hospitalization rates decreased in the UCHCC, driven by a decline in hospitalizations with a very low CD4 cell count. These findings are likely attributable to effective ART and improved care management of HIV and comorbid conditions and occurred despite the increasing age of the population of PLWH. In addition, although hospital length of stay decreased, readmission risk remained high. Black patients and those with uncontrolled HIV experienced higher incidences of hospitalization and readmission. Improving care access and engagement should be prioritized to manage HIV and other chronic diseases in the outpatient setting and prevent both hospitalizations and readmissions.
Supplementary Data
Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Financial support. This study was funded by the University of North Carolina at Chapel Hill Center for AIDS Research, a National Institutes of Health–funded program (grant number P30 AI50410). Traineeship for T. D.-M. was provided by the National Institute of Allergy and Infectious Diseases (grant number T32 AI007001).
Potential conflicts of interest. J. J. E. has received research grants from Janssen, Gilead Sciences, and ViiV Healthcare and has served as a consultant to Merck & Co, Janssen, Gilead Sciences, and ViiV Healthcare. D. A. W. has received research grants from Gilead Sciences and Merck & Co and has served as a consultant to Gilead Sciences, Janssen, ViiV Healthcare, and Merck & Co. All other authors report no potential conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
References
- 1. Buchacz K, Baker RK, Moorman AC, et al. Rates of hospitalizations and associated diagnoses in a large multisite cohort of HIV patients in the United States, 1994–2005. AIDS 2008; 22:1345–54. [DOI] [PubMed] [Google Scholar]
- 2. Krentz HB, Dean S, Gill MJ. Longitudinal assessment (1995–2003) of hospitalizations of HIV-infected patients within a geographical population in Canada. HIV Med 2006; 7:457–66. [DOI] [PubMed] [Google Scholar]
- 3. Gebo KA, Diener-West M, Moore RD. Hospitalization rates in an urban cohort after the introduction of highly active antiretroviral therapy. J Acquir Immune Defic Syndr 2001; 27:143–52. [DOI] [PubMed] [Google Scholar]
- 4. Yehia BR, Fleishman JA, Hicks PL, Ridore M, Moore RD, Gebo KA. Inpatient health services utilization among HIV-infected adult patients in care 2002–2007. J Acquir Immune Defic Syndr 2010; 53:397–404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Fleishman JA, Gebo KA, Reilly ED, et al. Hospital and outpatient health services utilization among HIV-infected adults in care 2000–2002. Med Care 2005; 43(9 Suppl):iii40–52. [DOI] [PubMed] [Google Scholar]
- 6. Berry SA, Fleishman JA, Moore RD, Gebo KA. Trends in reasons for hospitalization in a multisite United States cohort of persons living with HIV, 2001–2008. J Acquir Immune Defic Syndr 2012; 59:368–75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Bellino S, Borghetti A, Lombardi F, et al. Trends of hospitalisations rates in a cohort of HIV-infected persons followed in an Italian hospital from 1998 to 2016. Epidemiol Infect 2019; 147:e89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Lundgren JD, Babiker AG, Gordin F, et al. Initiation of antiretroviral therapy in early asymptomatic HIV infection. N Engl J Med 2015; 373:795–807. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Centers for Disease Control and Prevention. HIV among people aged 50 and over Available at: https://www.cdc.gov/hiv/group/age/olderamericans/index.html. Accessed 12 June 2019.
- 10. Drozd DR, Kitahata MM, Althoff KN, et al. Increased risk of myocardial infarction in HIV-infected individuals in North America compared with the general population. J Acquir Immune Defic Syndr 2017; 75:568–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Silverberg MJ, Lau B, Achenbach CJ, et al. North American AIDS Cohort Collaboration on Research and Design of the International Epidemiologic Databases to Evaluate AIDS Cumulative incidence of cancer among persons with HIV in North America: a cohort study. Ann Intern Med 2015; 163:507–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Abraham AG, Althoff KN, Jing Y, et al. North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) of the International Epidemiologic Databases to Evaluate AIDS (IeDEA) End-stage renal disease among HIV-infected adults in North America. Clin Infect Dis 2015; 60:941–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Klein MB, Althoff KN, Jing Y, et al. North American AIDS Cohort Collaboration on Research and Design of IeDEA; North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) of IeDEA Risk of end-stage liver disease in HIV-viral hepatitis coinfected persons in North America from the early to modern antiretroviral therapy eras. Clin Infect Dis 2016; 63:1160–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Wong C, Gange SJ, Moore RD, et al. North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) Multimorbidity among persons living with human immunodeficiency virus in the United States. Clin Infect Dis 2018; 66:1230–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Kim JH, Psevdos G Jr, Gonzalez E, Singh S, Kilayko MC, Sharp V. All-cause mortality in hospitalized HIV-infected patients at an acute tertiary care hospital with a comprehensive outpatient HIV care program in New York City in the era of highly active antiretroviral therapy (HAART). Infection 2013; 41:545–51. [DOI] [PubMed] [Google Scholar]
- 16. Berry SA, Fleishman JA, Moore RD, Gebo KA. Thirty-day hospital readmissions for adults with and without HIV infection. HIV Med 2016; 17:167–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Betz ME, Gebo KA, Barber E, et al. Patterns of diagnoses in hospital admissions in a multistate cohort of HIV-positive adults in 2001. Med Care 2005; 43(9 Suppl):iii3–14. [DOI] [PubMed] [Google Scholar]
- 18. Bachhuber MA, Southern WN. Hospitalization rates of people living with HIV in the United States, 2009. Public Health Rep 2014; 129:178–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Engsig FN, Hansen AB, Gerstoft J, Kronborg G, Larsen CS, Obel N. Inpatient admissions and outpatient visits in persons with and without HIV infection in Denmark, 1995–2007. AIDS 2010; 24:457–61. [DOI] [PubMed] [Google Scholar]
- 20. Rentsch C, Tate JP, Akgun KM, et al. Alcohol-related diagnoses and all-cause hospitalization among HIV-infected and uninfected patients: a longitudinal analysis of United States veterans from 1997 to 2011. AIDS Behav 2016; 20:555–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Krumholz HM, Nuti SV, Downing NS, Normand SL, Wang Y. Mortality, hospitalizations, and expenditures for the medicare population aged 65 years or older, 1999–2013. JAMA 2015; 314:355–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Healthcare Cost and Utilization Project. HCUP fast stats Available at: https://www.hcup-us.ahrq.gov/faststats/NationalTrendsServlet. Accessed 12 August 2019.
- 23. Schouten J, Wit FW, Stolte IG, et al. AGEhIV Cohort Study Group Cross-sectional comparison of the prevalence of age-associated comorbidities and their risk factors between HIV-infected and uninfected individuals: the AGEhIV cohort study. Clin Infect Dis 2014; 59:1787–97. [DOI] [PubMed] [Google Scholar]
- 24. Kooij KW, Wit FW, Schouten J, et al. AGEhIV Cohort Study Group HIV infection is independently associated with frailty in middle-aged HIV type 1-infected individuals compared with similar but uninfected controls. AIDS 2016; 30:241–50. [DOI] [PubMed] [Google Scholar]
- 25. Dennis AM, Napravnik S, Seña AC, Eron JJ. Late entry to HIV care among Latinos compared with non-Latinos in a southeastern US cohort. Clin Infect Dis 2011; 53:480–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Lazar R, Kersanske L, Xia Q, Daskalakis D, Braunstein SL. Hospitalization rates among people with HIV/AIDS in New York City, 2013. Clin Infect Dis 2017; 65:469–76. [DOI] [PubMed] [Google Scholar]
- 27. Pence BW, Bengtson AM, Boswell S, et al. Who will show? Predicting missed visits among patients in routine HIV primary care in the United States. AIDS Behav 2019; 23:418–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Kleeberger CA, Phair JP, Strathdee SA, Detels R, Kingsley L, Jacobson LP. Determinants of heterogeneous adherence to HIV-antiretroviral therapies in the Multicenter AIDS Cohort Study. J Acquir Immune Defic Syndr 2001; 26:82–92. [DOI] [PubMed] [Google Scholar]
- 29. Gebo KA, Diener-West M, Moore RD. Hospitalization rates differ by hepatitis C satus in an urban HIV cohort. J Acquir Immune Defic Syndr 2003; 34:165–73. [DOI] [PubMed] [Google Scholar]
- 30. Falster K, Wand H, Donovan B, et al. Hospitalizations in a cohort of HIV patients in Australia, 1999–2007. AIDS 2010; 24:1329–39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Hall MJ, Levant S, DeFrances CJ.. Trends in inpatient hospital deaths: National Hospital Discharge Survey, 2000–2010. NCHS data brief, no 118. Hyattsville, MD: National Center for Health Statistics, 2013. [PubMed] [Google Scholar]
- 32. Kelly JJ, Chu SY, Buehler JW. AIDS deaths shift from hospital to home. AIDS Mortality Project Group. Am J Public Health 1993; 83:1433–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Feller DJ, Akiyama MJ, Gordon P, Agins BD. Readmissions in HIV-infected inpatients: a large cohort analysis. J Acquir Immune Defic Syndr 2016; 71:407–12. [DOI] [PubMed] [Google Scholar]
- 34. Nijhawan AE, Kitchell E, Etherton SS, Duarte P, Halm EA, Jain MK. Half of 30-day hospital readmissions among HIV-infected patients are potentially preventable. AIDS Patient Care STDS 2015; 29:465–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. HIV Research Network. Hospital and outpatient health services utilization among HIV-infected patients in care in 1999. J Acquir Immune Defic Syndr 2002; 30:21–6. [DOI] [PubMed] [Google Scholar]
- 36. Berry SA, Fleishman JA, Yehia BR, et al. HIV Research Network Thirty-day hospital readmission rate among adults living with HIV. AIDS 2013; 27:2059–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Nijhawan AE, Clark C, Kaplan R, Moore B, Halm EA, Amarasingham R. 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; 61:349–58. [DOI] [PubMed] [Google Scholar]
- 38. Davy-Mendez T, Napravnik S, Zakharova O, Wohl DA, Farel CE, Eron JJ. Estimating bias in hospitalization rates due to missing hospitalization data. In: Society for Epidemiologic Research Annual Meeting, Baltimore, MD, 2018. [Google Scholar]
- 39. Keene CM, Dondorp A, Crawley J, Ohuma EO, Mukaka M. A competing-risk approach for modeling length of stay in severe malaria patients in South-East Asia and the implications for planning of hospital services. Clin Infect Dis 2018; 67:1053–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Nijhawan AE, Metsch LR, Zhang S, et al. Clinical and sociobehavioral prediction model of 30-day hospital readmissions among people with HIV and substance use disorder: beyond electronic health record data. J Acquir Immune Defic Syndr 2019; 80:330–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.