Abstract
Objective:
People with HIV infection experience excessive mortality compared to their non-infected counterparts. It is unclear whether the impact of HIV infection on mortality varies by comorbidities or whether sex difference exists in this relationship. This study assessed the effect of newly diagnosed HIV infection on overall mortality among Medicare beneficiaries for both disabled and older adults (≥65 years old) based on their original entitlement.
Methods:
We constructed a retrospective matched cohort using a 5% nationally representative sample of Medicare beneficiaries between 1996 and 2015. People with incident HIV diagnoses were individually matched to up to three controls based on demographics. Cox proportional hazards models adjusted for baseline demographics and comorbidities were used to assess the effect of HIV status on survival among four disabled groups by sex strata. Within each stratum, interactions between comorbidity variables and HIV status were examined.
Results:
People with HIV, especially older females, had a higher prevalence of baseline comorbidities than controls. HIV-mortality association varied according to sex in older adults (p=0.004). Comorbidity-HIV interactions were more pronounced in disabled groups (p<0.0001). People with HIV with more chronic conditions had a less pronounced increase in the risk of death than those with fewer conditions, compared to uninfected controls.
Conclusion:
Medicare enrollees with newly diagnosed HIV had more prevalent baseline comorbidities and were at higher risk of death than people without HIV. HIV infection has a more pronounced effect among those with fewer comorbidities. Sex differences in HIV-mortality association exist among older Medicare enrollees.
Keywords: HIV, Medicare, comorbidity, survival, retrospective cohort
Introduction
The introduction of combination antiretroviral therapy (ART) in the mid-1990s has significantly increased survival among people with HIV (PWH) in the United States (U.S.) [1]. However, PWH still experience excessive mortality compared with their uninfected counterparts [2–5]. One Danish HIV cohort study between 1996 and 2014 estimated that the median survival time from 50 years of age was at least eight years shorter among PWH than among individually age- and gender-matched uninfected controls [6].
Well-established risk factors for mortality among PWH include old age at diagnosis [7], being male [7,8], being Black [9], injection drug use transmission [7,8], AIDS [7], low CD4 count [7,8], and detectable viral replication [7]. Since ART is now more effective and better tolerated, other age-related comorbidities, such as non-AIDS-related cancer, cardiovascular disease, chronic renal disease, cirrhosis, hepatitis C co-infection, and chronic kidney disease, have become more prevalent in HIV patients [10,11] and may adversely impact survival [4,5,12]. However, data are limited on this topic.
To date, only one large study used Medicare data to extensively examine comorbidities and mortality in people ≥65 years with HIV [11]. They concluded that older beneficiaries with HIV have significantly higher risks for comorbidities and mortality compared to older people without HIV. However, this study was limited because it included only individuals ≥65 years and excluded younger beneficiaries who qualified for Medicare coverage based on disability. Its mortality analyses also removed anyone newly diagnosed with HIV, which comprises 22% of all HIV infections among older adults [11]. Many beneficiaries with HIV were likely diagnosed before enrollment and the length of their HIV infection and previous ART likely changed their viral loads and drug toxicity, and thereby the severity of adverse outcomes and mortality. An analysis of newly diagnosed cases (i.e., incidence cohort) would alleviate these problems. Furthermore, the extended survival model in this study treated the comorbidities as risk factors/confounders. No study has evaluated the role of comorbidities as an effect modifier in the relationship between HIV and mortality. Management of comorbidities may impact the treatment of HIV, and the effect of HIV infection may differ between those with and without comorbidities, warranting further investigation.
Furthermore, sex differences in comorbidities and mortality among PWH are not well understood [13,14]. Prior studies demonstrated that the non-AIDS comorbidity burden was higher among women with HIV compared to their HIV uninfected counterparts, especially in older age groups [14,15], and that sex differences in comorbidities exist between older women and men with HIV [11,16]. However, these studies did not examine the sex differences in the impact of HIV infection on mortality or the role of comorbidities in this relationship.
Moreover, there is a higher prevalence of HIV infection among enrollees who qualify for Medicare because of disability [17], many of whom aged into the program. However, there are limited data in the literature describing the characteristics and mortality of this population and how they compare to their HIV uninfected counterparts.
In this study, we assessed the effect of newly diagnosed HIV infection on overall mortality among Medicare beneficiaries using carefully matched controls without HIV infection for both disabled and older adults. We also examined sex differences in baseline comorbidities and the impact of HIV infection on mortality; and explored the interaction effects between HIV and baseline comorbidities on survival.
Methods
Data source
This retrospective matched cohort study used a 5% national sample of Medicare beneficiaries between 1996 and 2015. Data were integrated from Master Beneficiary Summary Files (MBSF) and claim files, including Medicare Provider and Analysis Review (MedPAR) Files, Carrier Claims, and Outpatient Standard Analytic Files (OutSAFs) for all years. The MBSF base segment contains demographic and enrollment information about beneficiaries enrolled in Medicare during a calendar year, such as date of birth, sex, race/ethnicity, date of death, state, Medicaid Dual Eligibility, original entitlement for Medicare, and monthly coverage. The claim files were used to extract diagnoses codes (International Classification of Diseases, Clinical Modification, 9th and 10th Revision [ICD-9-CM/ICD-10-CM]) and diagnosis dates for the chronic conditions. This study was approved by the University of Texas Medical Branch at Galveston Institutional Review Board (IRB # 20–0275). Additionally, a Data Use Agreement was established with the Centers for Medicare and Medicaid Services (CMS) prior to all data analysis.
Study Cohort
In this study, Medicare beneficiaries who were originally enrolled because of disability (disability insurance benefits and/or end-stage renal disease) were included alongside those who were enrolled upon reaching 65 years. These two groups are denoted as the disabled and the older age group, respectively. We identified a case cohort with newly diagnosed HIV infection and constructed a control cohort without an HIV diagnosis individually matched with cases based on demographic characteristics. We followed the Chronic Conditions Data Warehouse algorithm [18] to identify the individuals with HIV diagnosis (ICD-9-CM codes: DX 042, 042.0, 042.1, 042.2, 042.9, 043, 043.1, 043.2, 043.3, 043.9, 044, 044.0, 044.9, 079.53, 795.71, V08; ICD-10-CM codes: B20, B97.35, R75, Z21). For inclusion in the HIV group, beneficiaries had to have at least one diagnosis of HIV on an inpatient claim (MedPAR) or two outpatient or Carrier claims at least one day apart at any time. For those who qualified with an inpatient diagnosis, the date of admission was assigned as the index date. For those qualifying from outpatient/Carrier diagnoses, the first of the two diagnoses was assigned as the index date. Beneficiaries were required to have at least one year of continuous enrollment in Medicare parts A and B with no Health Maintenance Organization (HMO) enrollment prior to the index date and be at least 18 years old at the date of HIV diagnosis. Matching was accomplished in the following manner. A PWH was selected, and all potential controls were assigned with that case’s diagnosis date as their index date. Based on this index date, potential controls had to have at least one year of continuous enrollment in Medicare parts A and B with no HMO enrollment prior to the index date. The remaining potential controls were matched to the case based on age (at index date), sex, race, geographic region, original entitlement for Medicare enrollment, and dual Medicare-Medicaid eligibility. If the number of controls matching on these factors was greater than three, a random selection of three was taken. The selected controls were then removed from the control pool and the process was repeated for each remaining case.
Measures
The outcome of interest was all-cause mortality and time to death/censoring. Beneficiaries were censored at the end of the study period (12/31/2015) or final loss of coverage. The primary exposure variable of interest was HIV status. Other variables of interest were age at diagnosis (18–39, 40–49, 50–59, 60–69, 70–79, 80+ years), sex (Male, Female), race (White, Black, Hispanic, Other/Unknown), original entitlement for Medicare enrollment (disabled, older age), Medicare-Medicaid dual eligibility status (Yes, No), year of diagnosis/index date (1996–2000, 2001–2005, 2006–2010, 2011–2015), U.S. Census division (New England, Middle Atlantic, South Atlantic, East North Central, East South Central, West North Central, West South Central, Mountain, Pacific), and chronic conditions within one year prior to the index date. We identified 30 chronic conditions (Yes, No) included in Elixhauser comorbidity index. Due to the high prevalence of hypertension and diabetes, these conditions likely drove the comorbidity count. Additionally, because the individual effect of hypertension and diabetes has been under-studied, we examined them as two individual covariates. The count of remaining Elixhauser comorbidities was categorized into (0, 1, 2–3, 4–5, 6–8, 9+) based on the distribution of the variable and interested comparisons.
Statistical Analysis
All analyses were performed by stratifying disability status and sex (total of 4 strata: disabled male, disabled female, older age male, older age female). Descriptive statistics were calculated for all variables by HIV status. The unadjusted survival curves by HIV status were obtained using the Kaplan-Meier method. The Cox proportional hazards model was used to assess the effect of HIV status on survival from HIV diagnosis/index date. The main effect model was adjusted for all aforementioned covariates. To further assess the effect of comorbidities on the relationship between HIV status and survival, the interactions between comorbidity variables and HIV status were added individually to the main effect model. Due to significant interactions, stratified results by comorbidity levels were presented. Proportionality of the hazards for HIV infection status was assessed first by Schoenfeld residuals and then by adding an interaction term with logarithm of exposure time to the model. The results satisfied the assumption. For all results, hazard ratios (HR) and 95% confidence intervals (CI) were reported. All tests were two-sided with significance level of 0.05. All analyses were performed using SAS version 9.4 (SAS Inc., Cary, NC).
Results
Demographics
There were 6,865 PWH who met all inclusion criteria and 20,358 individuals who were selected for the matched (1:3 ratio) comparison group (Table SI). We were able to match 6,690 (97.5%) HIV patients with three controls, with the proportion matching three controls slightly higher among the older age group. Table 1 lists the descriptive statistics for demographic variables by four strata. Because the controls were individually matched to the cases on these demographics, all of these characteristics are nearly identical between the groups. The minor differences are due to slight variations for number of matching controls per HIV-infected person. Females and males were of similar age among the disabled group (48.3 vs. 47.4 years), but females were three years older than males on average among the older age group (76.6 vs. 73.3 years). Very few people reached age 80 at HIV diagnosis in the disabled group (0.5% and 1.1% for males and females, respectively), which comprised 17% of males and 32% of females among the older age group. There were more Black beneficiaries in the disabled group, especially Black females (52% vs. 45%), but fewer Black females compared to Black males in the older age group (22% vs. 27%). The South Atlantic and Middle Atlantic regions had more PWH. More females had Medicaid Dual Eligibility in both groups. There were more recent (2006 and beyond) HIV diagnoses in females than in males among the disabled group, but similar numbers in the older age group.
Table 1.
Medicare beneficiary characteristics and baseline comorbidities for HIV infected and matched uninfected individuals by original entitlement and sex
| Medicare original entitlement | Disabled | Older Age | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Sex | Male | Female | Male | Female | |||||
| HIV status | Level | HIV N (%) | Non-HIV N (%) | HIV N (%) | Non-HIV N (%) | HIV N (%) | Non-HIV N (%) | HIV N (%) | Non-HIV N (%) |
| Sample size | 3,313 | 9,789 | 1,558 | 4,610 | 1,064 | 3,181 | 930 | 2,778 | |
| Age at Index Date | Mean (SD) | 47.4 (11.7) | 47.4 (11.7) | 48.3 (12.9) | 48.4 (12.8) | 73.3 (6.3) | 73.3 (6.2) | 76.7 (7.4) | 76.6 (7.4) |
| 18–39 | 888 (26.8) | 2,595 (26.5) | 417 (26.8) | 1,226 (26.6) | |||||
| 40–49 | 1,097 (33.1) | 3,262 (33.3) | 439 (28.2) | 1,301 (28.2) | |||||
| 50–59 | 773 (23.3) | 2,293 (23.4) | 400 (25.7) | 1,192 (25.9) | |||||
| 60–69 | 443 (13.4) | 1,314 (13.4) | 221 (14.2) | 655 (14.2) | 370 (34.8) | 1,107 (34.8) | 194 (20.9) | 582 (21.0) | |
| 70–79 | 97 (2.9) | 285 (2.9) | 64 (4.1) | 186 (4.0) | 516 (48.5) | 1,543 (48.5) | 437 (47.0) | 1,305 (47.0) | |
| 80+ | 15 (0.5) | 40 (0.4) | 17 (1.1) | 50 (1.1) | 178 (16.7) | 531 (16.7) | 299 (32.2) | 891 (32.1) | |
| Race | Black | 1,487 (44.9) | 4,424 (45.2) | 811 (52.1) | 2,420 (52.5) | 289 (27.2) | 865 (27.2) | 204 (21.9) | 609 (21.9) |
| Hispanic | 212 (6.4) | 588 (6.0) | 89 (5.7) | 249 (5.4) | 57 (5.4) | 170 (5.3) | 82 (8.8) | 245 (8.8) | |
| Other/Unknown | 150 (4.5) | 396 (4.0) | 64 (4.1) | 163 (3.5) | 39 (3.7) | 112 (3.5) | 41 (4.4) | 121 (4.4) | |
| White | 1,464 (44.2) | 4,381 (44.8) | 594 (38.1) | 1,778 (38.6) | 679 (63.8) | 2,034 (63.9) | 603 (64.8) | 1,803 (64.9) | |
| US Census Division | East North Central | 335 (10.1) | 992 (10.1) | 183 (11.7) | 545 (11.8) | 139 (13.1) | 416 (13.1) | 132 (14.2) | 396 (14.3) |
| East South Central | 190 (5.7) | 568 (5.8) | 71 (4.6) | 213 (4.6) | 43 (4.0) | 128 (4.0) | 38 (4.1) | 114 (4.1) | |
| Middle Atlantic | 636 (19.2) | 1,883 (19.2) | 300 (19.3) | 881 (19.1) | 216 (20.3) | 646 (20.3) | 174 (18.7) | 520 (18.7) | |
| Mountain | 91 (2.7) | 262 (2.7) | 31 (2.0) | 89 (1.9) | 34 (3.2) | 102 (3.2) | 26 (2.8) | 78 (2.8) | |
| New England | 173 (5.2) | 487 (5.0) | 54 (3.5) | 156 (3.4) | 31 (2.9) | 93 (2.9) | 20 (2.2) | 59 (2.1) | |
| Pacific | 480 (14.5) | 1,424 (14.5) | 175 (11.2) | 519 (11.3) | 157 (14.8) | 470 (14.8) | 95 (10.2) | 283 (10.2) | |
| South Atlantic | 969 (29.2) | 2,867 (29.3) | 538 (34.5) | 1,602 (34.8) | 314 (29.5) | 939 (29.5) | 325 (34.9) | 973 (35.0) | |
| West North Central | 88 (2.7) | 260 (2.7) | 37 (2.4) | 102 (2.2) | 32 (3.0) | 95 (3.0) | 30 (3.2) | 89 (3.2) | |
| West South Central | 351 (10.6) | 1,046 (10.7) | 169 (10.8) | 503 (10.9) | 98 (9.2) | 292 (9.2) | 90 (9.7) | 266 (9.6) | |
| Medicaid Dual Eligibility | 2,020 (61.0) | 5,970 (61.0) | 1,141 (73.2) | 3,376 (73.2) | 268 (25.2) | 796 (25.0) | 341 (36.7) | 1,015 (36.5) | |
| Index Year | 1995–2000 | 1,056 (31.9) | 3,104 (31.7) | 367 (23.6) | 1,084 (23.5) | 231 (21.7) | 689 (21.7) | 206 (22.2) | 615 (22.1) |
| 2001–2005 | 800 (24.1) | 2,370 (24.2) | 341 (21.9) | 1,006 (21.8) | 215 (20.2) | 642 (20.2) | 179 (19.2) | 536 (19.3) | |
| 2006–2010 | 837 (25.3) | 2,480 (25.3) | 463 (29.7) | 1,370 (29.7) | 375 (35.2) | 1,124 (35.3) | 354 (38.1) | 1,057 (38.0) | |
| 2011–2015 | 620 (18.7) | 1,835 (18.7) | 387 (24.8) | 1,150 (24.9) | 243 (22.8) | 726 (22.8) | 191 (20.5) | 570 (20.5) | |
| Hypertension | 1,365 (41.2) | 3,913 (40.0) | 895 (57.4) | 2,291 (49.7) | 766 (72.0) | 1,940 (61.0) | 773 (83.1) | 1,989 (71.6) | |
| Diabetes | 767 (23.2) | 2,100 (21.5) | 558 (35.8) | 1,264 (27.4) | 413 (38.8) | 866 (27.2) | 412 (44.3) | 810 (29.2) | |
| Elixhauser Comorbidity Count | Mean (SD) | 3.0 (3.4) | 1.7 (2.1) | 3.9 (3.3) | 2.3 (2.3) | 3.9 (3.4) | 1.7 (2.2) | 4.3 (3.2) | 2.0 (2.2) |
| 0 | 1,014 (30.6) | 3,504 (35.8) | 266 (17.1) | 1,066 (23.1) | 175 (16.4) | 1,229 (38.6) | 91 (9.8) | 798 (28.7) | |
| 1 | 526 (15.9) | 2,408 (24.6) | 206 (13.2) | 1,105 (24.0) | 153 (14.4) | 665 (20.9) | 113 (12.2) | 628 (22.6) | |
| 2–3 | 673 (20.3) | 2,310 (23.6) | 358 (23.0) | 1,401 (30.4) | 227 (21.3) | 762 (24.0) | 231 (24.8) | 764 (27.5) | |
| 4–5 | 417 (12.6) | 935 (9.6) | 279 (17.9) | 601 (13.0) | 179 (16.8) | 292 (9.2) | 186 (20.0) | 358 (12.9) | |
| 6–8 | 392 (11.8) | 495 (5.1) | 297 (19.1) | 334 (7.2) | 213 (20.0) | 176 (5.5) | 199 (21.4) | 185 (6.7) | |
| 9+ | 291 (8.8) | 137 (1.4) | 152 (9.8) | 103 (2.2) | 117 (11.0) | 57 (1.8) | 110 (11.8) | 45 (1.6) | |
| Alcohol Abuse | 418 (12.6) | 563 (5.8) | 112 (7.2) | 121 (2.6) | 54 (5.1) | 66 (2.1) | 15 (1.6) | 16 (0.6) | |
| Blood Loss Anemia | 556 (16.8) | 1,027 (10.5) | 303 (19.4) | 541 (11.7) | 381 (35.8) | 657 (20.7) | 315 (33.9) | 536 (19.3) | |
| Cardiac Arrhythmia | 135 (4.1) | 116 (1.2) | 92 (5.9) | 80 (1.7) | 76 (7.1) | 47 (1.5) | 69 (7.4) | 54 (1.9) | |
| Chronic Pulmonary Disease | 872 (26.3) | 1,602 (16.4) | 557 (35.8) | 1,183 (25.7) | 355 (33.4) | 651 (20.5) | 337 (36.2) | 586 (21.1) | |
| Coagulopathy | 248 (7.5) | 274 (2.8) | 137 (8.8) | 145 (3.1) | 126 (11.8) | 135 (4.2) | 120 (12.9) | 82 (3.0) | |
| Congestive Heart Failure | 482 (14.5) | 860 (8.8) | 288 (18.5) | 467 (10.1) | 314 (29.5) | 415 (13.0) | 287 (30.9) | 397 (14.3) | |
| Deficiency Anemia | 557 (16.8) | 699 (7.1) | 379 (24.3) | 485 (10.5) | 245 (23.0) | 196 (6.2) | 232 (24.9) | 260 (9.4) | |
| Depression | 786 (23.7) | 1,498 (15.3) | 552 (35.4) | 1,236 (26.8) | 163 (15.3) | 218 (6.9) | 211 (22.7) | 347 (12.5) | |
| Drug Abuse | 462 (13.9) | 500 (5.1) | 179 (11.5) | 206 (4.5) | 27 (2.5) | 17 (0.5) | 11 (1.2) | NA | |
| Fluid and Electrolyte Disorders | 645 (19.5) | 980 (10.0) | 394 (25.3) | 623 (13.5) | 316 (29.7) | 302 (9.5) | 306 (32.9) | 396 (14.3) | |
| Hypothyroidism | 307 (9.3) | 639 (6.5) | 313 (20.1) | 731 (15.9) | 144 (13.5) | 291 (9.1) | 272 (29.2) | 653 (23.5) | |
| Liver Disease | 650 (19.6) | 578 (5.9) | 364 (23.4) | 275 (6.0) | 181 (17.0) | 126 (4.0) | 159 (17.1) | 107 (3.9) | |
| Lymphoma | 69 (2.1) | 49 (0.5) | 44 (2.8) | 26 (0.6) | 53 (5.0) | 28 (0.9) | 51 (5.5) | 30 (1.1) | |
| Metastatic Cancer | 41 (1.2) | 53 (0.5) | 30 (1.9) | 45 (1.0) | 55 (5.2) | 58 (1.8) | 22 (2.4) | 31 (1.1) | |
| Obesity | 166 (5.0) | 668 (6.8) | 234 (15.0) | 727 (15.8) | 51 (4.8) | 124 (3.9) | 88 (9.5) | 138 (5.0) | |
| Other Neurological Disorders | 428 (12.9) | 1,195 (12.2) | 252 (16.2) | 572 (12.4) | 151 (14.2) | 195 (6.1) | 138 (14.8) | 182 (6.6) | |
| Paralysis | 117 (3.5) | 422 (4.3) | 48 (3.1) | 153 (3.3) | 32 (3.0) | 44 (1.4) | 31 (3.3) | 35 (1.3) | |
| Pulmonary Circulation Disorders | 105 (3.2) | 154 (1.6) | 73 (4.7) | 128 (2.8) | 66 (6.2) | 95 (3.0) | 56 (6.0) | 76 (2.7) | |
| Psychoses | 561 (16.9) | 1,614 (16.5) | 288 (18.5) | 600 (13.0) | 64 (6.0) | 72 (2.3) | 82 (8.8) | 96 (3.5) | |
| Peptic Ulcer Disease excluding bleeding | 102 (3.1) | 151 (1.5) | 49 (3.1) | 86 (1.9) | 26 (2.4) | 56 (1.8) | 33 (3.5) | 62 (2.2) | |
| Peripheral Vascular Disorders | 438 (13.2) | 751 (7.7) | 259 (16.6) | 390 (8.5) | 323 (30.4) | 475 (14.9) | 320 (34.4) | 485 (17.5) | |
| Renal Failure | 571 (17.2) | 823 (8.4) | 334 (21.4) | 384 (8.3) | 243 (22.8) | 231 (7.3) | 214 (23.0) | 187 (6.7) | |
| Rheumatoid Arthritis | 207 (6.2) | 326 (3.3) | 214 (13.7) | 436 (9.5) | 102 (9.6) | 101 (3.2) | 147 (15.8) | 197 (7.1) | |
| Solid Tumor without Metastasis | 194 (5.9) | 354 (3.6) | 117 (7.5) | 220 (4.8) | 226 (21.2) | 456 (14.3) | 117 (12.6) | 240 (8.6) | |
| Valvular Disease | 327 (9.9) | 513 (5.2) | 207 (13.3) | 365 (7.9) | 234 (22.0) | 338 (10.6) | 232 (24.9) | 358 (12.9) | |
| Weight Loss | 404 (12.2) | 359 (3.7) | 209 (13.4) | 215 (4.7) | 191 (18.0) | 131 (4.1) | 172 (18.5) | 136 (4.9) | |
NA, count fewer than 11 was not included because of the CMS cell size suppression policy to protect the confidentiality of enrollees.
Baseline comorbidities
Baseline individual comorbidities and number of comorbidities are also displayed in Table 1. In addition, Fig. 1 presents the histograms of prevalence by four strata. Hypertension was the most prevalent comorbidity in both the disabled (40–60%) and older age groups (61–83%), with PWH having a significantly higher prevalence than controls throughout all four strata. Diabetes was the second most prevalent comorbidity and the differences in prevalence between PWH and controls were more pronounced in females for both groups. Other comorbidities were also common. At least 61% of individuals had at least one comorbidity other than diabetes or hypertension, with the highest prevalence observed in the older females with HIV (90%). For other individual comorbidities, PWH almost universally had a higher prevalence than controls. Among these conditions, chronic pulmonary disease, congestive heart failure, depression, fluid and electrolyte disorders, anemia, and peripheral vascular disorders were most prevalent. Only a few exceptions were observed. Obesity was slightly higher in controls in the disabled group, but not in the older age group. This was especially true for females with HIV, who showed nearly double (9.5% vs. 5.0%) the prevalence of obesity compared to the control females in the older age group.
Fig. 1.
Comorbidities for HIV infected and matched uninfected individuals by Medicare original entitlement and sex
ESRD: end-stage renal disease.
Compared to the older age group, the disabled group had a higher prevalence of alcohol abuse and drug abuse. We also observed a few conditions that varied significantly according to sex. For example, alcohol abuse was much higher in males with HIV compared to females with HIV (12.6% vs. 7.2% in the disabled groups, 5.1% vs. 1.6% in the older age groups). Females with HIV had higher prevalence of depression compared to males (35% vs. 24% in disabled groups, 23% vs. 15% in older age groups). Females with HIV also had a higher prevalence than males with HIV of hypothyroidism, obesity, and rheumatoid arthritis.
Mortality
The median survival times were lower for the groups with HIV (Table 2) compared with their matched non-HIV controls, with a difference of 4–5 years in the older age groups. Since the disabled group did not reach a mortality of 50% by the end of the study, 50th and 75th percentile survival times were not estimated. The Kaplan-Meier curves showed a distinct difference by HIV status (log-rank test p<0.0001) across all strata (Fig. 2). The 10- and 15-year mortality rates were higher in PWH. The curves showed more separation by HIV status among older males compared to females (lower panel), which showed that the risk of death due to HIV infection was amplified in older males compared to females. Cox proportional hazards models showed a significant interaction between HIV status and sex for the older age group (p=0.004) but not for the disabled group (p=0.66), after adjustment for baseline covariates.
Table 2.
Survival statistics for HIV infected and matched uninfected individuals by Medicare original entitlement and sex
| Disabled | Older Age | |||||||
|---|---|---|---|---|---|---|---|---|
| Male | Female | Male | Female | |||||
| HIV | Non-HIV | HIV | Non-HIV | HIV | Non-HIV | HIV | Non-HIV | |
| Death Event (N, %) | 1,323 (39.9) | 2,204 (22.5) | 515 (33.1) | 817 (17.7) | 614 (57.7) | 1,212 (38.1) | 505 (54.3) | 1,120 (40.3) |
| Age at death, Median (25th, 75th) | 54.7 (47.2, 63.6) | 58.5 (51.1, 67.4) | 56.7 (47.2, 66.3) | 60.9 (52.0, 70.6) | 77.6 (72.4, 83.7) | 81.4 (76.6, 86.8) | 82.5 (77.1, 88.4) | 85.5 (80.3, 90.6) |
| Age at death/censoring, Median (25th, 75th) | 54.2 (47.0, 61.9) | 56.0 (49.0, 63. 8) | 55.1 (46.3, 63.3) | 56.4 (48.0, 64.6) | 77.3 (72.6, 82.8) | 79.5 (74.5, 84.7) | 81.7 (76.5, 87.5) | 83. 5 (78.0, 88.8) |
| Survival Time (Years), Median (25th, 75th) | 12.1 (4.0, −) | − (10.6, −) | 14.8 (4.6, −) | − (11.9, −) | 5.2 (1.4, 11.8) | 10.4 (5.2, 17.9) | 6.6 (2.2, 14.0) | 10.9 (5.1, 17.2) |
| 10 years mortality | 45% | 24% | 41% | 21% | 69% | 47% | 64% | 47% |
| 15 years mortality | 55% | 34% | 51% | 31% | 82% | 68% | 80% | 68% |
-, indicates that there are no estimates due to the reason that the disabled group did not reach a mortality of 50% by the end of the study, 50th and 75th percentile survival times were not estimated.
Fig. 2.
Kaplan Meier curves for HIV infected and matched uninfected individuals by Medicare original entitlement and sex
ESRD: end-stage renal disease.
Since there were significant interactions between HIV status and comorbidities in disabled groups, and trends in that direction in older age groups, we performed stratified analyses by examining the level specific HR for HIV status by comorbidity levels within each stratum (Table 3). We found those without hypertension [HR: disabled males 2.4 (2.2, 2.7), disable females 2.4 (2.0, 2.9), old age males 1.9 (1.6, 2.3) and old age females 1.4 (1.1, 1.8)] had a higher relative hazard of death (PWH vs. controls) compared to those with hypertension [1.4 (1.2, 1.5), 1.4 (1.2, 1.7), 1.6 (1.4, 1.8) and 1.3 (1.1, 1.4)] across four strata. Disabled groups showed a more pronounced impact of HIV infection by having a larger HR in general. These groups also exhibited more significant interactions between HIV infection and comorbidities, indicated by larger differences in HR between individuals with hypertension and those without hypertension. No sex differences in HIV-mortality association were observed in the disabled group; however, males in the older age group had higher HRs than females. There were similar findings for diabetes and other comorbidities. Those with one or no comorbidities had higher HR for HIV status, especially males and those in the disabled group. However, sex differences were more pronounced in the older age group.
Table 3.
Adjusted hazard ratios (HIV infected vs. HIV uninfected) by Medicare original entitlement and sex within comorbidity strata
| Disabled | Older Age | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Comorbidities | Level | Male | Female | Male | Female | ||||
| HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p | HR (95% CI) | p | ||
| Hypertension | No | 2.43 (2.21, 2.68) | <0.0001 | 2.44 (2.03, 2.94) | <0.0001 | 1.91 (1.59, 2.29) | 0.06 | 1.42 (1.11, 1.82) | 0.38 |
| Yes | 1.37 (1.23, 1.53) | 1.42 (1.22, 1.65) | 1.55 (1.36, 1.76) | 1.25 (1.11, 1.42) | |||||
| Diabetes | No | 2.26 (2.08, 2.46) | <0.0001 | 1.89 (1.62, 2.19) | 0.08 | 1.69 (1.49, 1.92) | 0.48 | 1.31 (1.13, 1.51) | 0.66 |
| Yes | 1.13 (0.99, 1.29) | 1.54 (1.29, 1.84) | 1.57 (1.32, 1.87) | 1.25 (1.05, 1.48) | |||||
| Elixhauser Comorbidity Count | 0 | 3.10 (2.70, 3.56) | <0.0001 | 2.91 (2.17, 3.90) | <0.0001 | 2.08 (1.63, 2.66) | 0.14 | 1.40 (0.98, 2.00) | 0.88 |
| 1 | 2.26 (1.89, 2.70) | 2.42 (1.71, 3.42) | 1.45 (1.07, 1.95) | 1.31 (0.95, 1.82) | |||||
| 2–3 | 1.87 (1.62, 2.16) | 1.75 (1.38, 2.21) | 1.79 (1.45, 2.21) | 1.19 (0.95, 1.47) | |||||
| 4–5 | 1.41 (1.17, 1.68) | 1.98 (1.52, 2.56) | 1.49 (1.17, 1.90) | 1.29 (1.01, 1.66) | |||||
| 6–8 | 1.14 (0.94, 1.38) | 0.95 (0.73, 1.23) | 1.35 (1.04, 1.74) | 1.41 (1.10, 1.81) | |||||
| 9+ | 0.90 (0.69, 1.17) | 1.45 (1.03, 2.05) | 1.87 (1.26, 2.77) | 1.11 (0.73, 1.68) | |||||
Discussion
This study observed significantly higher comorbidities at baseline and shorter survival among newly diagnosed beneficiaries with HIV as compared to matched controls for all groups. Moreover, we observed sex differences in baseline comorbidities and survival, especially in the older age group. We also found interactions between baseline comorbidities and HIV status in terms of the survival outcome within each stratum, especially in the disabled group.
The two most prevalent chronic conditions were hypertension (72% and 83% of older HIV men and women, respectively) and diabetes (39% and 44% for the same groups), which were higher than those reported in previous studies and likely due to the less healthy population studied [11,16]. Hypertension and diabetes were even more common among the older females with HIV. We found other chronic conditions were sex dependent as well. Similar to Turrini’s study [11], we found that older females with HIV had more depression than males with HIV (23% vs. 15%), although less discrepancy was found in our analysis.
The HRs of death for HIV infection that we detected in the older age groups were lower than those reported in previous studies [2,11,16]. For example, Turrini et al. analyzed all prevalent HIV cases among older Medicare beneficiaries between 2011 and 2016 and estimated an HR of 2.0 from a model adjusted by similar baseline covariates and 11 individual chronic conditions [11]. Although we targeted the newly diagnosed HIV population, it is likely that the impact of HIV infection is attenuated after individual exact matching, in additional to the effect of controlling for these baseline characteristics and all comorbidities in the modeling. Our data also demonstrated sex differences in the risk of death when compared to controls, where the risk was amplified in male older age groups. We found that baseline comorbidities significantly interacted with HIV status in the disabled group. A similar trend was observed in the older age group, but the association did not reach statistical significance. PWH with more chronic conditions had a less pronounced increase in risk of death than those with fewer conditions when compared to controls. Previous studies showed that an increase in hospitalization has been associated with chronic conditions [19]. Female Medicare beneficiaries were more likely than their male counterparts to have multiple chronic conditions, and those with more chronic conditions were more likely to use more Medicare services [20]. This increase in health care utilization may have improved the management of their HIV and comorbidities and thus decreased the impact of these conditions on other outcomes. Therefore, this higher level of use of health care may explain the discordance in the baseline comorbidities and mortality and the sex differences found among older age groups.
Also, as expected, after adjusting for social demographic variables and comorbidities as effect modifiers, disabled PWH still had a much higher risk of death as opposed to the older PWH when compared to their control groups. This result is likely because the disabled group had generally poorer health conditions, disadvantaged socioeconomic status, greater financial and access barriers to care, and unmeasured behavioral factors, even after accounting for the existing factors in this analysis [21].
This study has several limitations. We did not examine ART or clinical biomarkers, such as CD4 count and viral load, as covariates, which can have a major impact on mortality. The Medicare Prescription Drug Benefit program became available in 2006 and there were no laboratory data in Medicare datasets, so we could not incorporate the effect of treatment information or laboratory data on HIV infection effectively in this analysis. We studied newly diagnosed HIV patients; however, newly diagnosed people living with HIV, especially older people with HIV, tend to have more advanced disease due to delayed diagnosis, which results in negative health outcomes [5,22–24]. Although our findings provide important information on those newly diagnosed PWH in the Medicare population, the results may not be generalizable to the general population of PWH who have been infected with HIV and have been on ART for years; the associations between HIV status and death would be expected to be lower in the whole population of PWH. We evaluated overall mortality, but this result cannot be used to assess HIV-related death. We evaluated the impact of baseline comorbidities on survival outcomes. However, comorbidities change over time with the development of diseases and their treatment. Future studies should incorporate time-varying covariates for comorbid conditions. We focused on the chronic conditions included in the Elixhauser comorbidities index. Further analysis could include more conditions, including HIV-related conditions and additional sex-specific conditions. Although we applied a one-year continuous enrollment period to identify people ‘newly’ diagnosed with HIV, some individuals may have had HIV infection but delayed diagnosis, impacted their general health conditions [22,23]. We applied a two-year continuous enrollment period to assess the impact of our observing period and found a similar pattern (results are not shown).
Strengths of the study include its large sample size and long follow up extending up to 20 years, which made it possible to evaluate long-term survival, observe interactions, and conduct stratified analysis. We matched on seven baseline variables which largely controlled confounding effects due to these factors before modeling. Thus, crude estimates were more comparable between groups. We also provided sex stratified results in terms of comorbidities and survival, which lies beyond the scope of most previous studies. We further examined interactions between baseline comorbidities and HIV status and provided stratified estimates of the impact of HIV status.
These results have important implications for management and treatment of HIV disease and comorbidities and could also help inform policy makers in the design of screening and prevention strategies. Given the high prevalence of comorbidities among people with HIV, enhanced regular screening for comorbidities upon HIV diagnosis and comprehensive management of comorbidities are needed, especially among males who are less likely to be retained in care. The higher relative risk of mortality among those with fewer chronic conditions highlights the necessity of closely monitoring the treatment and supporting the engagement in care of those HIV patients, which may impact health outcomes. Sex differences in comorbidities and survival suggest the need for an awareness of the differential needs of women and men. Finally, the disabled HIV population requires additional clinical support, since this group has a higher risk of worse health outcomes.
Conclusion
Medicare enrollees with newly diagnosed HIV had more baseline comorbidities and were at higher risk of death following diagnosis in both disabled and older populations. HIV infection has a more pronounced effect on survival in subgroups, especially those with fewer comorbidities. Compared to males with HIV, older females with HIV had more baseline comorbidities, but their HIV infection had less of an impact on their survival.
Supplementary Material
Acknowledgements
Funding support: Drs. Yu and Berenson are supported by a research career development award (K12HD052023: Building Interdisciplinary Research Careers in Women’s Health Program-BIRCWH; Berenson, PI) from the National Institutes of Health/Office of the Director (OD)/National Institute of Allergy and Infectious Diseases (NIAID), and Eunice Kennedy Shriver National Institute of Child Health & Human Development (NICHD). Dr. Giordano is supported by the MD Anderson Foundation Chair at Baylor College of Medicine. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Department of Veterans Affairs.
Footnotes
Conflict of Interest:
There are no conflicts of interest.
References
- 1.May MT, Gompels M, Delpech V, Porter K, Orkin C, Kegg S, et al. Impact on life expectancy of HIV-1 positive individuals of CD4+ cell count and viral load response to antiretroviral therapy. AIDS 2014; 28(8):1193–1202. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.de Coninck Z, Hussain-Alkhateeb L, Bratt G, Ekstrom AM, Gisslen M, Petzold M, et al. Non-AIDS Mortality Is Higher Among Successfully Treated People Living with HIV Compared with Matched HIV-Negative Control Persons: A 15-Year Follow-Up Cohort Study in Sweden. AIDS Patient Care STDS 2018; 32(8):297–305. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Eyawo O, Franco-Villalobos C, Hull MW, Nohpal A, Samji H, Sereda P, et al. Changes in mortality rates and causes of death in a population-based cohort of persons living with and without HIV from 1996 to 2012. BMC Infect Dis 2017; 17(1):174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Fontela C, Aguinaga A, Moreno-Iribas C, Reparaz J, Rivero M, Gracia M, et al. Trends and causes of mortality in a population-based cohort of HIV-infected adults in Spain: comparison with the general population. Sci Rep 2020; 10(1):8922. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Croxford S, Kitching A, Desai S, Kall M, Edelstein M, Skingsley A, et al. Mortality and causes of death in people diagnosed with HIV in the era of highly active antiretroviral therapy compared with the general population: an analysis of a national observational cohort. Lancet Public Health 2017; 2(1):e35–e46. [DOI] [PubMed] [Google Scholar]
- 6.Legarth RA, Ahlstrom MG, Kronborg G, Larsen CS, Pedersen C, Pedersen G, et al. Long-Term Mortality in HIV-Infected Individuals 50 Years or Older: A Nationwide, Population-Based Cohort Study. J Acquir Immune Defic Syndr 2016; 71(2):213–218. [DOI] [PubMed] [Google Scholar]
- 7.Trickey A, May MT, Vehreschild J, Obel N, Gill MJ, Crane H, et al. Cause-Specific Mortality in HIV-Positive Patients Who Survived Ten Years after Starting Antiretroviral Therapy. PLoS One 2016; 11(8):e0160460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hessol NA, Kalinowski A, Benning L, Mullen J, Young M, Palella F, et al. Mortality among participants in the Multicenter AIDS Cohort Study and the Women’s Interagency HIV Study. Clin Infect Dis 2007; 44(2):287–294. [DOI] [PubMed] [Google Scholar]
- 9.Trepka MJ, Fennie KP, Sheehan DM, Niyonsenga T, Lieb S, Maddox LM. Racial-ethnic differences in all-cause and HIV mortality, Florida, 2000–2011. Ann Epidemiol 2016. March; 26(3):176–82.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Friedman EE, Duffus WA. Chronic health conditions in Medicare beneficiaries 65 years and older with HIV infection. AIDS 2016; 30(16):2529–2536. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Turrini G, Chan SS, Klein PW, Cohen SM, Dempsey A, Hauck H, et al. Assessing the health status and mortality of older people over 65 with HIV. PLoS One 2020; 15(11):e0241833. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Heron JE, Bagnis CI, Gracey DM. Contemporary issues and new challenges in chronic kidney disease amongst people living with HIV. AIDS Res Ther 2020; 17(1):11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Loutfy MR, Sherr L, Sonnenberg-Schwan U, Walmsley SL, Johnson M, d’Arminio Monforte A, et al. Caring for women living with HIV: gaps in the evidence. J Int AIDS Soc 2013; 16:18509. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Collins LF, Sheth AN, Mehta CC, Naggie S, Golub ET, Anastos K, et al. The Prevalence and Burden of Non-AIDS Comorbidities among Women living with or at-risk for HIV Infection in the United States. Clin Infect Dis 2020. ciaa204. doi: 10.1093/cid/ciaa204. Online ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Palella FJ, Hart R, Armon C, Tedaldi E, Yangco B, Novak R, et al. Non-AIDS comorbidity burden differs by sex, race, and insurance type in aging adults in HIV care. AIDS 2019; 33(15):2327–2335. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kong AM, Pozen A, Anastos K, Kelvin EA, Nash D. Non-HIV Comorbid Conditions and Polypharmacy Among People Living with HIV Age 65 or Older Compared with HIV-Negative Individuals Age 65 or Older in the United States: A Retrospective Claims-Based Analysis. AIDS Patient Care STDS 2019; 33(3):93–103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.KFF. Medicare and HIV. https://www.kff.org/hivaids/fact-sheet/medicare-and-hiv/. Published 2016. Accessed November 1, 2020.
- 18.Chronic Condition Data Warehouse. Other chronic or potentially disabling conditions. https://www2.ccwdata.org/web/guest/condition-categories. Accessed October 20, 2020.
- 19.Cubanski J, Neuman P. Medicare doesn’t work as well for younger, disabled beneficiaries as it does for older enrollees. Health Aff (Millwood) 2010; 29(9):1725–1733. [DOI] [PubMed] [Google Scholar]
- 20.Buchacz K, Baker RK, Moorman AC, Richardson JT, Wood KC, Holmberg SD, 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(11):1345–1354. [DOI] [PubMed] [Google Scholar]
- 21.Centers for Medicare and Medicaid Services. Chronic Conditions among Medicare Beneficiaries, Chartbook, 2012 Edition. Baltimore, MD. 2012. [Google Scholar]
- 22.Brooks JT, Buchacz K, Gebo KA, Mermin J. HIV infection and older Americans: the public health perspective. Am J Public Health 2012; 102(8):1516–1526. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Grinsztejn B, Hosseinipour MC, Ribaudo HJ, Swindells S, Eron J, Chen YQ, et al. Effects of early versus delayed initiation of antiretroviral treatment on clinical outcomes of HIV-1 infection: results from the phase 3 HPTN 052 randomised controlled trial. Lancet Infect Dis 2014; 14(4):281–290. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Rein SM, Lampe FC, Johnson MA, Bhagani S, Miller RF, Chaloner C, et al. All-cause hospitalization according to demographic group in people living with HIV in the current antiretroviral therapy era. AIDS. 2021; 35(2):245–255. [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.