Trends in Comorbidities Among Human Immunodeficiency Virus–Infected Hospital Admissions in New York City from 2006–2016

Abstract

Background

Due to the advent and success of antiretroviral therapy, the number of people living and aging with human immunodeficiency virus (HIV) has grown substantially. Although people living with HIV (PLHIV) are experiencing longer life expectancies, this achievement may be undermined by increasing and disproportionate chronic disease burden among PLHIV.

Methods

This study is a retrospective analysis of adult (≥18 years) inpatient hospital discharges from a large hospital system in the New York City, New York metropolitan area, between 1 January 2006 and 31 December 2016. We aimed to investigate (1) changes in the prevalence of Charlson-defined comorbidities among PLHIV hospitalized between 2006 and 2016 and (2) changes in the unadjusted prevalence ratio (PR) of comorbidities in HIV-positive versus HIV-negative admissions over time.

Results

Of 898 139 hospital admissions from 2006–2016, 19 039 (2.1%) were HIV positive. Across all admissions during the study period, the greatest comorbidity disparities between HIV-positive and HIV-negative admissions were mild liver disease (PR, 4.9 [95% confidence interval, 4.8–5.1]), moderate or severe liver disease (PR, 2.2 [2.0–2.4]), and chronic pulmonary disease (PR, 1.8 [1.8–1.8]).

Conclusions

The prevalence and relative burden of comorbidities among hospitalized PLHIV are changing over time. Careful monitoring and intensive discharge planning may be effective strategies for addressing the evolving health needs of PLHIV.

The burden of chronic disease is significantly greater among hospitalized people living with HIV (PLHIV) compared with their HIV-negative counterparts, and the prevalence of Charlson-defined comorbidities is shifting over time. Careful monitoring is needed to address PLHIV’s evolving health needs.

Due to the advent and success of antiretroviral therapy, the number of people living and aging with human immunodeficiency virus (HIV) has grown substantially. Today, there are approximately 1.1 million people living with HIV (PLHIV) in the United States, over half of whom are aged 50 and older [1]. With early diagnosis and treatment, PLHIV can live into their 70s or beyond, depending on disease and demographic factors [2]. Current projections suggest that, by 2030, 73% of PLHIV in the United States will be aged 50 and older [3].

Although PLHIV are experiencing longer life expectancies, this achievement may be somewhat attenuated by increasing chronic disease burden. A number of studies have indicated that the burden of chronic comorbidity is increasing in this population. An assessment of Truven MarketScan data from 2003–2013 indicated that the percentage of PLHIV living with comorbidities increased over that time frame [4]. The study also found that the prevalence of several comorbidities rose over this period, including an increase in the percentage of PLHIV on Medicaid experiencing any cardiovascular event from 3% in 2003 to 6.9% in 2013, and an increase in those experiencing renal impairment from 4.8% to 10.3% during the same time period [5].

Research has established a strong connection between HIV status and multimorbidity, demonstrating that PLHIV are at greater risk for a variety of comorbidities and experience earlier onset of chronic disease than their HIV-negative counterparts [6]. For instance, analyses of the US Veterans Aging Cohort Study indicate that HIV-positive adults have a greater risk of myocardial infarction, hypertension, non–AIDS-defining cancer, end-stage renal disease, and chronic obstructive pulmonary disease than HIV-negative adults [7, 8]. People living with HIV are also more likely to experience various types of cardiovascular disease (CVD), including heart failure, stroke, and arrhythmias [9]. A meta-analysis found that PLHIV have a 1.5 to 2 times greater risk of CVD [10] and PLHIV experience sudden cardiac death at 4.5 times the rate of their HIV-negative counterparts [11]. Previous research also suggests that PLHIV have a greater risk of thyroid, liver, and endocrine disease; osteoporosis; deep vein thrombosis; hyperlipidemia; and metabolic disorders [4]. However, these trends may vary by context; a separate study conducted in British Columbia found that, while hypertension incidence doubled among PLHIV between 2002 and 2012, there was no change in the incidence of CVD and the incidence of chronic kidney and liver disease decreased [12]. A variety of mechanisms have been proposed to explain the higher number of comorbidities among PLHIV compared with HIV-negative adults, including compromised immune systems, recurring respiratory tract infections [13, 14], increased systemic and lung oxidative stress [15, 16], heightened inflammation [5, 17], and antiretroviral medication toxicity [7].

Although the rising burden of chronic disease among PLHIV has been acknowledged [4–6, 12], further research is needed to clarify which comorbidities will pose the greatest threat to health in the coming years, as well as whether disparities in comorbidity burden are worsening or are being alleviated in PLHIV relative to the general population. Such an investigation is critical for developing treatment guidelines that address the myriad of challenges that PLHIV experience and enable PLHIV to implement manageable symptom-management strategies. These findings will also have important economic implications, as research suggests that the increased prevalence of comorbidities among patients with HIV has resulted in higher direct costs of medical care compared with the general population [18]. This study adds to the literature by reporting vital statistics on comorbidity burden among PLHIV using data from a large hospital system in New York City from 2006–2016. Specifically, we aimed to assess the following: (1) changes in the prevalence of comorbidities among PLHIV hospitalized between 2006 and 2016 and (2) changes in the relative risk of comorbidities in HIV-positive versus HIV-negative admissions over time.

METHODS

Sample

Data for this study were drawn from an administrative database created for a study on hospital-acquired infections in the New York City area. This study is a retrospective analysis of adult (≥18 years) inpatient hospital discharges between 1 January 2006 and 31 December 2016 from a large hospital system in the New York City, New York metropolitan area. The hospital system includes a community hospital and 2 tertiary/quaternary care hospitals with more than 2000 beds and over 100 000 inpatient discharges annually in an ethnically diverse population. This study was approved and monitored by the Columbia University Irving Medical Center Institutional Review Board.

Measures

Admissions characteristics examined included inpatient admission and discharge dates, age, race, gender, and the status of the 17 comorbidities (binary: positive vs negative) included in the Charlson/Deyo comorbidity index. Results were stratified by HIV status, and the primary outcome explored was yearly prevalence of all Charlson/Deyo-defined comorbidities, excluding HIV.

The Charlson/Deyo comorbidity index is a weighted score originally developed to predict 1-year mortality risk for an individual patient [19]. The index is composed of 17 medical conditions, all of which have been linked to mortality risk. With the exception of HIV, we chose to include all Charlson/Deyo comorbidities as outcomes in the present analyses, given strongly established links to mortality and the widespread use of the Charlson/Deyo index as a predictor of mortality. Each admission was coded as positive or negative for each of the 17 comorbidities using enhanced International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), coding algorithms as outlined by Quan et al [20]. The present-on-admission (POA) indicator is collected in New York State discharge data to distinguish existing comorbidities from complications that occurred during hospital admission [21, 22]. Only comorbidities identified by the POA indicator were used to identify the comorbidities in this study.

Analysis

We performed secondary data analysis of 898 139 adult (age ≥18 years) inpatient hospital discharges between 1 January 2006 and 31 December 2016. We first assessed sample characteristics and prevalence of all Charlson/Deyo comorbidities in the full sample of admissions (2006–2016). Bivariate analyses were then performed to determine the unadjusted prevalence ratio (PR) of comorbidities among HIV-positive admissions relative to HIV-negative admissions by year. The unadjusted PR of each comorbidity in HIV-positive admissions relative to HIV-negative admissions was then generated separately for each year. Wald confidence intervals (CIs) were also calculated for the unadjusted PR [23, 24]. We explored monotonic time trends in yearly comorbidity prevalence from 2006–2016, by HIV status, using the nonparametric Mann-Kendall trend test.

Next, we calculated the unadjusted PR of each comorbidity among HIV-positive and HIV-negative admissions by year. The Mann-Kendall trend test (α = .05) was again applied to explore whether there was a monotonic trend in disparities of disease over time. To avoid inflated type II error in planned tests, we did not perform multiple testing adjustment [25]. Analyses were conducted in R Studio version 1.2.1335 (The R Project for Statistical Computing) and SAS version 9.4 (SAS Institute).

RESULTS

Of 898 139 hospital admissions from 2006–2016, 19 039 (2.1%) were HIV positive. The average age of HIV-positive admissions was 48.5 (SD = 11.7) years, while the average age of HIV-negative admissions was 57.7 (20.0) years. Among HIV-positive admissions, 65% were male, 12% were white, and 26% were Black. Among HIV-negative admissions, 44% were male, 31% were white, and 10% were Black. Admissions positive for HIV were coded with an average of 1.3 (SD = 1.2) comorbidities, while HIV-negative admissions were coded with an average of 1.0 (SD = 1.2) comorbidities over the entire study period. The mortality rate was 2.3% among HIV-positive admissions and 2.2% among HIV-negative admissions (P = .4124). A complete description of the sample is reported in Table 1.

Table 1.

Sample Demographic Characteristics and Hospital Admission Attributes, Length of Stay, and Readmissions Outcomes

	All Admissions (N = 898 139)	HIV-Positive Admissions (n = 19 039)	HIV-Negative Admissions (n = 879 100)	P a
Sociodemographic factors
Sex, n (%)				<.0001
Male	399 098 (44)	12 612 (66)	386 486 (44)
Female	499 024 (56)	6427 (34)	492 597 (56)
Age in years, mean (SD)	57.5 (19.9)	48.5 (11.7)	57.7 (20.0)	<.0001
Race, n (%)				<.0001
White non-Hispanic	276 097 (31)	2316 (12)	273 781 (31)
Black non-Hispanic	91 056 (10)	5044 (26)	86 012 (10)
Hispanic	48 952 (5)	1387 (7)	47 565 (5)
Asian/Pacific Islander	25 674 (3)	89 (0)	25 585 (3)
Other	135 850 (15)	4760 (25)	131 090 (15)
Unknown	280 586 (31)	4907 (26)	275 679 (31)
Declined	39 924 (4)	536 (3)	39 388 (4)
Insurance plan, n (%)				<.0001
Commercial	172 527 (19)	1368 (7)	171 159 (19)
Medicare	401 325 (45)	5662 (30)	395 663 (45)
Medicaid	207 290 (23)	10 944 (57)	196 346 (22)
Blue Cross	93 587 (10)	680 (4)	92 907 (11)
Self-pay	14 533 (2)	221 (1)	14 312 (2)
All other	7598 (1)	141 (1)	7457 (1)
Missing	1279 (0)	23 (0)	1256 (0)
Mean (SD) number of comorbidities	1.0 (1.2)	1.3 (1.2)	1.0 (1.2)	<.0001
Hospital admission attributes
ICU stay, n (%)	111 194 (12.4)	2069 (10.9)	109 125 (12.4)	<.0001
Mean (SD) length of stay, days	7.5 (10.0)	8.9 (11.1)	7.5 (10.0)	<.0001
Mean (SD) number of prior hospitalizations	1.7 (3.8)	3.2 (5.5)	1.7 (3.8)	<.0001
Death associated with admission, n (%)	19 839 (2.21)	437 (2.30)	19 402 (2.21)	.4124
Admission year, n (%)				<.0001
2006	77 400 (9)	1946 (10)	75 454 (9)
2007	77 476 (9)	1952 (10)	75 524 (9)
2008	74 966 (8)	1790 (9)	73 176 (8)
2009	81 290 (9)	1925 (10)	79 365 (9)
2010	84 060 (9)	1862 (10)	82 198 (9)
2011	84 054 (9)	1708 (9)	82 346 (9)
2012	82 350 (9)	1831 (10)	80 519 (9)
2013	82 126 (9)	1830 (10)	80 296 (9)
2014	73 515 (8)	1491 (8)	72 024 (8)
2015	90 208 (10)	1641 (9)	88 567 (10)
2016	90 694 (10)	1063 (6)	89 631 (10)

N = 1 027 129 admissions. Abbreviations: HIV, human immunodeficiency virus; ICU, intensive care unit.

^aBased on chi-square tests of independence for categorical variables and 2-sample t tests for continuous variables.

Over 70% of HIV-positive admissions had a comorbidity compared with 56.0% of HIV-negative admissions. The most common comorbidities among HIV-positive admissions over the entire study period were chronic pulmonary diseases (25.3%), mild liver disease (19.8%), and renal disease (18.8%). Among HIV-negative admissions, the most common comorbidities were diabetes without chronic complication (15.3%), chronic pulmonary disease (14.0%), and congestive heart failure (13.5%) (Table 2).

Table 2.

Prevalence of Charlson-Deyo Defined Comorbidities, HIV-Positive and HIV-Negative Admissions

Diagnosis	Prevalence in HIV-Positive Sample (n = 19 039), %	Prevalence in HIV-Negative Sample (n = 879 100), %	Prevalence Ratio (95% CI)	P a	Rank in HIV-Positive Sample	Rank in HIV-Negative Sample
Chronic pulmonary diseases	25.3	14.0	1.8 (1.8–1.8)	<.0001	1	2
Mild liver disease	19.8	4.0	4.9 (4.8–5.1)	<.0001	2	9
Renal disease	18.8	13.4	1.4 (1.4–1.5)	<.0001	3	4
Any malignancy, including lymphoma and leukemia, except for malignant neoplasm of skin	13.5	11.1	1.2 (1.2–1.3)	<.0001	4	5
Diabetes without chronic complication	11.4	15.3	.8 (.7–.8)	<.0001	5	1
Congestive heart failure	11.3	13.5	.8 (.8–.9)	<.0001	6	3
Cerebrovascular disease	4.2	6.6	.6 (.6–.7)	<.0001	7	6
Diabetes with chronic complication	3.3	3.9	.9 (.8–.9)	<.0001	8	10
Peripheral vascular disease	3.2	6.0	.5 (.5–.6)	<.0001	9	7
Moderate or severe liver disease	2.6	1.2	2.2 (2–2.4)	<.0001	10	16
Metastatic solid tumor	2.5	4.3	.6 (.5–.6)	<.0001	11	8
Myocardial infarction	1.9	3.4	.6 (.5–.6)	<.0001	12	11
Hemiplegia or paraplegia	1.7	1.5	1.1 (1–1.3)	.028	13	14
Peptic ulcer disease	1.2	1.4	.9 (.8–1)	.0367	14	15
Rheumatic disease	0.8	2.1	.4 (.3–.5)	<.0001	15	12
Dementia	0.5	1.5	.3 (.3–.4)	<.0001	16	13

Abbreviations: CI, confidence interval; HIV, human immunodeficiency virus.

^aBased on chi-square tests of independence.

Among HIV-positive admissions, monotonic time trends were observed for 8 comorbidities at α = .05: renal disease, congestive heart failure, diabetes with chronic complications, peripheral vascular disease, all malignancies, cerebrovascular disease, chronic pulmonary diseases, and dementia (Table 3, Figure 1A). Several comorbidities more than doubled in prevalence over the 2006–2016 time period (Figure 1A). The prevalence of renal disease increased from 12.0% in 2006 to 26.6% in 2016, of congestive health failure from 8.3% to 17.4%, of diabetes with chronic complications from 2.1% to 5.7%, of peripheral vascular disease from 2.0% to 4.5%, and of dementia from 0.1% to 1.98%.

Table 3.

Average Change Over 11 Years (2006–2016) in (1) Comorbidity Prevalence Among Human Immunodeficiency Virus (HIV)–Positive and HIV-Negative Admissions Separately and (2) Prevalence Ratio Between HIV-Positive and HIV-Negative Admissions

	HIV-Positive Admissions		HIV-Negative Admissions		Prevalence Ratio
Disease	Average Yearly Change in Prevalence, %	P a	Average Yearly Change in Prevalence, %	P a	Average Change in Disparity Over Time, %	P a
Any malignancy	7	.010*	2	.052	5	.016*
Cerebrovascular disease	8	.016*	2	.052	7	.024*
Chronic pulmonary diseases	6	.016*	3	.001*	3	.243
Congestive heart failure	9	.004*	1	.312	8	.020*
Dementia	82	.034*	18	.697	57	.016*
Diabetes with chronic complication	15	.004*	4	.029*	11	.024*
Diabetes without chronic complication	5	.484	6	.697	0	.052
Hemiplegia or paraplegia	8	.061	12	.004*	3	.243
Metastatic solid tumor	10	.086	0	.061	10	.016*
Mild liver disease	3	.243	3	.118	−1	.016*
Moderate or severe liver disease	−2	.697	2	.118	−3	.815
Myocardial infarction	24	.586	7	.309	14	.139
Peptic ulcer disease	11	.073	2	.029*	8	.073
Peripheral vascular disease	12	.004*	1	.139	10	.001*
Renal disease	9	<.001*	2	.006*	7	.004*
Rheumatic disease	60	.9380	8	.061	32	.586

*P < .05.

^a P values from Mann-Kendall monotonic time trend test.

Figure 1.

Prevalence of comorbidities among (A) HIV-positive admissions and (B) HIV-negative admissions, by year. Only comorbidities showing a significant time trend from 2006–2016 (α = .05) are included. Abbreviation: HIV, human immunodeficiency virus.

Among HIV-negative admissions, monotonic time trends were observed for 5 comorbidities at α = .05: chronic pulmonary disease, hemiplegia or paraplegia, renal disease, peptic ulcer disease, and diabetes with chronic complication (Table 3, Figure 1B). However, the magnitude of changes in comorbidity prevalence was smaller relative to the HIV-positive sample. The prevalence of chronic pulmonary diseases increased from 11.7% to 15.7% over the study period, of plegias (paralysis) from 0.9% to 1.9%, of renal disease from 10.98% to 13.66%, of diabetes with chronic complications from 3.09% to 4.52%, and of peptic ulcer disease from 1.05% to 1.27% (Figure 1B).

Across all admissions during the study period, the greatest comorbidity disparities between HIV-positive and HIV-negative admissions were mild liver disease (PR = 4.9; 95% CI, 4.8–5.1), moderate or severe liver disease (PR = 2.2; 95% CI, 2.0–2.4), and chronic pulmonary diseases (PR = 1.8; 95% CI, 1.8–1.8) (Table 2). When comparing the relative burden of disease in HIV-positive and HIV-negative admissions over time, 9 comorbidities exhibited time trends, indicating that the disparity between HIV-positive and HIV-negative admissions changed from 2006 to 2016 (Table 3). With the exception of mild liver disease, the prevalence ratio of comorbidities for HIV-positive admissions worsened over the study period: for comorbidities with a lower prevalence among HIV-positive admissions in 2006, the prevalence in HIV-positive admissions approached or surpassed that of HIV-negative admissions over the study period. For comorbidities with a higher prevalence among HIV-positive admissions, the disparity continued to widen.

DISCUSSION

We found that the burden of chronic disease is increasing over time among hospitalized PLWH. These findings are aligned with previous literature indicating that comorbidities are significantly more prevalent among PLWH compared with their HIV-negative counterparts. Overall, we found evidence that a greater number of comorbidities increased over time among HIV-infected admissions. Additionally, the pace at which they increased was more rapid among HIV-positive admissions compared with their HIV-negative counterparts. Among HIV-positive admissions, we identified substantial and approximately 2-fold increases in renal disease, congestive health failure, diabetes with chronic complications, and peripheral vascular disease over the 11-year study period.

Our finding that chronic pulmonary diseases and renal disease are among the most prevalent comorbidities among individuals with HIV has been reported in previous research [4, 8, 26]. Several studies have also shared our finding that renal disease and peripheral vascular disease have increased over time and disproportionately among PLHIV. For example, Gallant et al [4] found that the percentage of PLHIV with renal impairment or any cardiovascular event, including peripheral vascular disease, nearly doubled between 2003 and 2013. Likewise, in a cross-sectional analysis of HIV-positive patients and HIV-negative controls in Amsterdam, Schouten et al [27] found a significantly higher prevalence of impaired renal function and peripheral arterial disease among HIV-positive patients.

We found that the mortality rate of HIV-positive hospital admissions was only 0.1% higher than that of HIV-negative admissions. This is far smaller than the difference of 2.2% reported in a recent comparison of hospitalization trends and comorbidities among PLHIV and the overall hospitalized population in Illinois [28]. While this study utilized a different methodology than the study in Illinois, the mortality rate of HIV-positive and HIV-negative hospital admissions in New York City may differ less than in other contexts due to strong citywide access to HIV testing, early diagnosis, and treatment. New York City has the largest number of HIV-positive residents but was the only major US city to reach a 90% threshold for diagnosis, treatment, and suppression in 2017 [29].

While our overall results have been reported in other populations, our investigation revealed a few critical insights. Since our analyses included a control group, we were able to determine that the accumulation of these conditions was not just a part of the routine aging process but that hospitalized PLHIV experienced a substantially higher risk of acquiring certain chronic conditions than their HIV-negative counterparts. Further, our analysis examined trends in comorbidity burden and disparities among HIV-positive hospital admissions over a decade. This longitudinal analysis indicates that the burden of comorbidities among PLHIV is increasing and that failure to address this increasing burden will result in widening health disparities in a marginalized population.

Limitations

While we report observed trends in comorbidity prevalence and burden among HIV-positive hospital admissions in New York City, a primary limitation of our study was lack of available data to explain the behavioral drivers of trends in comorbidity prevalence and changes in disparities between HIV-positive and HIV-negative individuals over time. For example, PLHIV are more likely to have a history of smoking and drug use, 2 behaviors that increase one’s risk of developing chronic disease [30]. Further research is needed to identify drivers of these trends as well as interventions that might alleviate them. In addition, PLHIV who acquire diseases, such as liver disease or chronic pulmonary diseases, may experience more symptoms than those who have those diseases but do not have HIV, resulting in more hospitalizations. This would make the inpatient prevalence of liver disease or chronic pulmonary diseases higher among PLHIV than people without HIV, even if the overall prevalence of these comorbidities does not differ.

There are further limitations that might arise from the dataset used in our analyses. First, the unit of analysis was admissions rather than patients, which might result in counting the same patient several times. If HIV-positive individuals with a greater number of comorbidities were more likely to be admitted (and counted) multiple times, the difference in comorbidity prevalence between HIV-positive and HIV-negative persons might be inflated. However, this would likely not alter our interpretation of increasing or decreasing disparity over time, assuming that this source of bias remained constant over the study period.

Second, comorbidities may be identified at higher rates in PLHIV than their HIV-negative counterparts due to detection bias, as PLHIV undergo more medical examinations and receive more thorough screenings [27]. This might result in overestimating the relative prevalence of comorbidities in PLHIV compared with their HIV-negative counterparts.

Third, diagnoses based on ICD-9 codes may also underestimate incidence, as time and resource constraints on healthcare staff are a barrier to complete health-data capture; this has resulted in high specificity but moderate sensitivity for a variety of conditions [31]. However, this would not bias our findings unless this under-capture were systematically higher or lower for HIV-positive patients, or if the degree of under-capture changed over time.

Finally, it is important to consider potential impacts of mortality selection on the adjusted prevalence of comorbidities that generally emerge in late life. For example, Aung [32] previously reported that the burden of dementia risk is higher in HIV-positive people than in the general population and is attributable to other diseases like hypertension and CVD. These differences may be obscured in our sample if PLHIV are more likely to die before developing these late-life comorbidities, and should be monitored as the lifespan of PLHIV continues to increase.

Conclusions and Relevance

The prevalence and relative burden of comorbidities among hospitalized PLHIV are changing over time. Careful monitoring and intensive discharge planning may be effective strategies for addressing the evolving health needs of PLHIV. Given the increased risk for comorbidities such as liver disease and chronic pulmonary diseases among PLHIV, prompt diagnosis, continuous monitoring, and extensive management of these conditions may help improve quality of life and mortality among PLWH.

Notes

Financial support. This work was supported by the Agency for Healthcare Research and Quality (grant number R01HS024915) and the Small Research Project from the National Health, Lung, and Blood Institute–funded PRIDE-CVD grant number R25HL105446 to Dr Mohamed Boutjdir at SUNY Health Science University, Brooklyn, New York.

Potential conflicts of interest. The authors: No reported 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.