Assessing the frailty index in trials of glucose-lowering therapy for type 2 diabetes

Conference: 
SAPC ASM 2024 - Bristol
Talk Code: 
6E.1
Presenter: 
Peter Hanlon
Twitter: 
@PHanlon17
Co-authors: 
Heather Wightman, Elaine Butterly, Lili Wei, Kenneth Rockwood, Andy Clegg, Miles Witham, Sofia Dias, Nicky Welton, David Phillippo, David McAllister, Peter Hanlon
Author institutions: 
University of Glasgow, Dalhousie University, University of Leeds, University of Newcastle, University of York, University of Bristol

Problem

Frailty is an age-related state of reduced physiological reserve associated with increased risk of adverse clinical outcomes. Frailty is common among people with type 2 diabetes. However, representation of frailty within diabetes trials is unclear as trials do not routinely measure or report frailty in trial participants. As a result, the applicability of trial findings to people living with frailty is uncertain. This study aims to use individual participant data (IPD) from randomised controlled trials (RCTs) of glucose-lowering therapies for type 2 diabetes to quantify frailty and assess the association between frailty and adverse events, hypoglycaemia and trial attrition.

Approach

We analysed IPD from 28 RCTs of SGLT2 inhibitors, GLP1 receptor agonists and DDP4 inhibitors. Frailty was quantified using a frailty index based on Rockwood’s cumulative deficit model of frailty: a count of health deficits identified from medical history, laboratory data, and patient-reported measures from baseline questionnaires. For each trial, we quantified the distribution of the frailty index. We then assessed the association between frailty and trial attrition using logistic regression, and between frailty and total adverse events, total serious adverse events, and number of hypoglycaemic episodes using negative binomial regression. All models were adjusted for age and sex.

Findings

Across the 28 RCTs, the total number of trial participants analysed was 24,744. Mean age ranged from 53.8 to 74.2. Mean frailty index ranged from 0.04 to 0.26. Using a conservative cut-point of 0.2 (mild frailty), median frailty prevalence was 5% (interquartile range 1.5% to 8.2%). Four trials had frailty prevalence ≥25%, and focused either on older people or those with renal impairment. In all but one trial, <​5% of participants had a frailty index >0.3. Across all trials, frailty was associated with increased odds of trial attrition (pooled odds ratio 1.41 per 0.1-point increase in the frailty index, 95% confidence interval 1.26 to 1.58), and with an increased incidence of adverse events (incident rate ratio [IRR] 1.44, 95% confidence interval 1.34 to 1.55), serious adverse events (IRR 2.08, 95% confidence interval 1.79 to 2.41) and of hypoglycaemic events (IRR 1.24, 95% confidence interval 1.03 to 1.50).

Consequences

Frailty was rare in most trials, but was more common in trials focused on higher-risk populations such as older people or those with renal impairment. However, even in trials assessing higher-risk populations, severe frailty was uncommon. Frailty was associated with clinically important adverse events and trial attrition. Clinicians should be cautious when applying trial findings to people living with severe frailty, who are largely excluded from many trials. However, it is also clear that trials focusing on higher-risk groups can and do successfully recruit people living with frailty. Future analysis will assess whether treatment efficacy varies by frailty.

Submitted by: 
Peter Hanlon
Funding acknowledgement: 
Tenovus Scotland, Medical Research Council