Dissertations and Theses

Date of Degree


Document Type


Degree Name

Doctor of Public Health (DPH)


Epidemiology and Biostatistics


Denis Nash

Committee Members

Elizabeth Kelvin

Alexis Pozen

Kathy Anastos

Subject Categories

Epidemiology | Health Services Research | Public Health | Virus Diseases


aging, Medicare, chronic disease, insurance claims, polypharmacy, adherence


The number of people living with HIV (PLWH) ≥65 years old is increasing in the United States (US) as PLWH live longer. In 2015, there were nearly 1 million people living with diagnosed HIV in the US and under 10% were age ≥65. By 2035, the proportion of PLWH in this age group is projected to be 27%. Like the general population of elderly individuals, as they age, PLWH face age-related comorbidities, many of which require routine medical care and daily medications, in addition to daily antiretroviral therapy (ART) for treatment of HIV. Previous research has found that PLWH develop these conditions at higher rates and earlier ages than HIV-negative individuals. Therefore, elderly PLWH are particularly vulnerable to the challenges associated with multimorbidity and polypharmacy, including remaining adherent to their medications. However, in the context of PLWH, “older” has typically been defined as >50 years old, with few studies analyzing PLWH age ≥65, a relevant population given Medicare insurance eligibility in the US. Additionally, there is a dearth of literature focusing on older HIV-positive women and comorbidities. Gaining a better understanding of the non-HIV-related comorbid disease and non-ART comedication burden among PLWH age ≥65 is important because these comorbidities often require care coordination among multiple providers and because comorbidities may negatively impact adherence to ART.

With a focus on PLWH aged ≥65 living in the US, the objectives of this dissertation were: (1) to compare the non-HIV disease and non-ART medication burden among PLWH and HIV-negative individuals; (2) to describe patterns of non-HIV condition co-occurrence among PLWH and HIV-negative individuals; and (3) to examine the impact of comorbid disease and comedication burden on ART adherence among PLWH.

Using data from 2010 to 2015 in the IBM Watson Health MarketScan® Medicare Supplemental insurance database, I selected PLWH and HIV-negative individuals ≥65 years old based on diagnoses on medical claims and pharmacy claims for ART. Outcomes included common diagnoses and medication classes, prevalence and number of non-HIV conditions, daily non-ART medications, and ART medication adherence (using proportion of days covered (PDC) over a 1-year period, all based on medical and pharmacy claims. To address the first dissertation objective, I examined age-standardized prevalence rates for non-HIV conditions and prevalence ratios (PRs) and fit sex-stratified multivariable generalized linear models for the number of non-HIV conditions and number of daily non-ART medications. For the second dissertation objective, I used latent class analysis to identify classes of individuals based on the presence of non-HIV comorbid conditions. Separate latent class models were fit to cohorts of PLWH, HIV-negative individuals, and HIV-negative individuals matched to the PLWH cohort on demographic characteristics. For the third dissertation objective, I modeled the odds of being adherent to ART (defined as PDC ≥80%) using separate adjusted logistic regression models for PLWH treated with ART with the number of comorbid conditions, the number of comedications, and comorbidity classes as the exposures of interest.

Dissertation Objective 1: I assessed non-HIV conditions and daily non-ART medications among 2,359 elderly PLWH and 2,010,513 elderly HIV-negative individuals. PLWH were younger (mean age 71 vs. 76 years) and a larger proportion were men (81% vs. 45%). The most common diagnoses among both HIV-positive and HIV-negative cohorts were hypertension and dyslipidemia. Most non-HIV conditions were more prevalent among PLWH. The largest absolute difference was in anemia (29.6 cases per 100 people vs.11.7) and the largest relative difference was in hepatitis C (PR=22.0). The unadjusted mean number of non-HIV conditions and daily non-ART medications were higher for PLWH (4.61 conditions and 3.79 medications) than HIV-negative individuals (3.94 conditions and 3.41 medications), respectively. In generalized linear models with log link and negative binomial distribution where the outcome was the number of non-HIV conditions, PLWH had significantly more non-HIV conditions than HIV-negative individuals (ratios: men=1.272, [95% CI 1.233-1.312]; women=1.326 [1.245-1.413]). Among those with >0 daily non-ART medications, men with HIV had significantly more non-ART medications than HIV-negative men (ratio=1.178 [1.133-1.226]) in a generalized linear model with log link and gamma distribution where the outcome was number of non-ART medications. Dissertation Objective 2: When conditions with prevalence ≥15% among PLWH were included in latent class models, a 3-class solution was identified for cohorts of PLWH, all HIV-negative individuals, and matched HIV-negative individuals: a sickest class with high probabilities of multiple non-HIV conditions, a class characterized by hypertension and dyslipidemia, and a healthiest class with low probabilities of non-HIV conditions. Nearly 20% of the PLWH were assigned to the sickest class compared to 10.5% of the matched HIV-negative cohort, with PLWH having higher probabilities of specific non-HIV diagnoses, including kidney disease and anemia. Dissertation Objective 3: Lastly, when analyzing the odds of being adherent to ART among 1,644 elderly PLWH with logistic regression models, I found that odds of non-adherence were significantly higher among PLWH with 5-6 comorbidities compared to PLWH with 0-2 comorbidities (adjusted odds ratio [AOR]=1.420 [95% CI 1.035-1.947]). After controlling for the number of comedications, PLWH with 5-6 comorbid conditions (AOR=1.589 [95% CI 1.131-2.232]) and PLWH with ≥7 conditions (AOR=1.528 [95% CI 1.049-2.225]) were 50%-60% more likely to be non-adherent than PLWH with 0-2 conditions. Additionally, PLWH belonging to the hypertension/dyslipidemia/diabetes comorbidity class were more likely to be non-adherent than PLWH in the healthiest class (AOR=1.319 [95% CI 1.047-1.661]).

In conclusion, I found that the disease burden associated with aging is substantially higher among PLWH age ≥65 than similarly aged HIV-negative individuals, and that there was a trend of decreasing ART adherence with increasing comorbid disease burden among PLWH. Due to the greater comorbid disease burden, elderly PLWH in the US require additional services and care coordination to effectively manage both HIV and comorbid conditions, particularly those who have a large number of comorbid conditions. Because multimorbidity may have a negative impact of ART adherence, there is a need for interventions focused on primary and secondary prevention of comorbidities and ART adherence among elderly PLWH. Future analyses may include additional analyses with larger samples of older HIV-positive women, assessments of comorbidities and ART medication adherence using other data sources where HIV-specific variables like duration of infection, CD4 and viral load are available, an evaluation of PLWH’s adherence to comedications, such as antidiabetes medications and antihypertensives, and an estimation of the cost impact of comorbid conditions among older PLWH in the US.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.