Factors Partitioning Physical Frailty in People Aging With HIV

A Classification and Regression Tree Approach

Mehmet Inceer; Marie-J. Brouillette; Lesley K. Fellows; José A. Morais; Marianne Harris; Fiona Smaill; Graham Smith; Réjean Thomas; Nancy E. Mayo


HIV Medicine. 2022;23(7):738-749. 

In This Article

Abstract and Introduction


Objective: To estimate the extent to which comorbidity and lifestyle factors were associated with physical frailty in middle-aged and older Canadians living with HIV.

Design: Cross-sectional analysis of 856 participants from the Canadian Positive Brain Health Now cohort.

Methods: The frailty indicator phenotype was adapted from Fried's criteria using self-report items. Univariate logistic regression and classification and regression tree (CaRT) models were used to identify the most relevant independent contributors to frailty.

Results: In all, 100 men (14.0%) and 26 women (19.7%) were identified as frail (≥ 3/5 criteria) for an overall prevalence of 15.2%. Nine comorbidities showed an influential association with frailty. The most influential comorbidities were hypothyroidism [odds ratio (OR) = 2.55, 95% confidence interval (CI): 1.29–5.03] and arthritis (OR = 2.54, 95% CI: 1.58–4.09). Additionally, tobacco (OR = 1.79, 95% CI: 1.05–3.04) showed an association. Any level of alcohol consumption showed a protective effect for frailty.

The CaRT model showed nine pathways that led to frailty. Arthritis was the most discriminatory variable followed by alcohol, hypothyroidism, tobacco, cancer, cannabis, liver disease, kidney disease, osteoporosis, lung disease and peripheral vascular disease. The prevalence of physical frailty for people with arthritis was 27.4%; with additional cancer or tobacco and alcohol the prevalence rates were 47.1% and 46.1%, respectively. The protective effect of alcohol consumption evident in the univariate model appeared again in the CaRT model, but this effect varied.

Cognitive frailty (19.5% overall) and emotional frailty (37.9% overall) were higher than the prevalence of physical frailty.

Conclusions: Specific comorbidities and tobacco use were implicated in frailty, suggesting that it is comorbidities causing frailty. However, some frailty still appears to be HIV-related. The higher prevalence of cognitive and emotional frailty highlights the fact that physical frailty should not be the only focus in HIV.


Treatment success for HIV infection over the past two decades has resulted in a shift from a disease with a poor prognosis to a chronic yet manageable condition.[1,2] There is now a population ageing with HIV and they wish to age well.[3] One impediment to ageing well is the emergence of frailty.[4,5] As mortality in people living with HIV has shifted from infection-driven to comorbidity-driven, this shift may also apply to frailty.[6]

Frailty has been defined as 'a multidimensional syndrome characterized by decreased reserve and diminished resistance to stressors'; however, there is no agreed-upon way of operationalizing frailty.[7] According to the 2013 Consensus Statement on frailty, age is considered a necessary, but not sufficient, contributor to frailty, and no age restrictions are recognized. Other key dimensions of frailty are an accumulation of health deficits and clinical signs of weakened reserve,8–10 and existing measures of frailty focus on quantifying these dimensions.[11] Clinical signs of weakened reserve are those that produce the Fried's frailty phenotype,[9] the most commonly used and adapted approach to classify frailty.[8] Lifestyle factors such as smoking and alcohol use have been found to be associated with both the development and progression of frailty.[12,13]

There has been an interest in frailty in HIV since the early 2000s and this interest is growing; there are some 70 reviews of this topic, 40 in the past 5 years. In HIV, viral infection has been proposed to accelerate ageing through immunocompromise and chronic inflammation. In contrast to those without HIV of the same age, people living with HIV show a higher prevalence of frailty as well as a higher prevalence of cognitive impairment and psychological disorders,[14–17] which in themselves can be considered as different types of frailty.[18]

Biomarkers of physiological reserve in frailty pathogenesis in HIV have been considered to contribute to frailty.[19] Potential biomarkers were identified in the Veterans Aging Cohort Study (VACS), including markers of infection and organ damage.[20] Markers of chronic inflammation have also been implicated.21,22 These reserve indicators can also contribute directly to disability, comorbidity and frailty.[9]

Frailty is a multifactorial syndrome with causes originating from morbidities, genetics, lifestyle and the environment.23–25 Comorbidities that are not directly associated with HIV infection such as diabetes or heart disease are of interest in HIV as they show strong associations with frailty in people without HIV.[26] While most of the comorbidities are manageable, the link between comorbidities and frailty is important, as they are another contributor to depletion of reserves which could result in the emergence of disabilities, eventually limiting mobility and independence.

It is important to distinguish between frailty, comorbidity and disability[27] as the causes and consequences of each one differ. Additionally, age-related comorbidities (e.g. arthritis) are also important considerations in frailty as they may be reasons why people perform poorly on tests that are part of classifying people as frail. In HIV, there is a lack of clarity about the effects of age, sex, viral control, lifestyle and comorbidity on frailty. A challenge with conducting research in frailty is that the most common approach for frailty classification, the Fried's frailty phenotype, requires performance-based tests. These tests are not routinely carried out in research or in clinical practice. What is widely available are self-reported limitations on functional activities requiring a certain degree of capacity on the performance tests. These self-reports can be adopted as surrogate indicators of a person's status on the frailty criteria.[8]

The purpose of this study is to estimate for a cohort of middle-aged or older Canadians living with HIV the extent to which comorbidity and lifestyle factors are associated with a greater prevalence of frailty as defined using frailty indicators.