Age-bias in assessments of medical decision-making capacity: a cross-sectional experimental vignette study

Abstract

Accurate assessments about medical decision-making capacity (DMC) are vital but could be susceptible to age-bias. We examined if client age impacts on DMC likelihood assessment. One hundred and forty-seven adults were trained to identify four cognitive abilities (i.e., understanding, appreciation, reasoning, expression). Six experimental vignettes described a person making a diabetes treatment decision with systematic variation of client age (unspecified, 50-, or 95-years-old) and abilities (demonstrated vs uncertain). The dependent variable was DMC likelihood. Standardised tests of explicit (Kogan's Attitudes Toward Old People scale) and implicit age-bias (IAT) were administered. There was a statistically significant effect of abilities on DMC likelihood. DMC was judged more likely when cognitive abilities were demonstrated. There were no other significant effects. Age-bias, measured implicitly and explicitly, did not influence DMC likelihood. Contrary to the hypothesis, DMC likelihood was unaffected by client age. This suggests that training can mitigate age-bias in simulated clinical DMC evaluation.

Keywords: ageism; aging; clinical judgement; consent; dementia; healthcare; implicit bias; medical decision-making capacity; unconscious bias.

Figure 1.

Age implicit association test (Age-IAT): example of stimuli. Note: Taken from the Age-IAT (Millisecond Software, 2016).

Figure 2.

Participant flowchart and data selection process. Note: DMC = decision-making capacity; KAOP = Kogan’s Attitudes Toward Old People Scale; Age-IAT = Implicit Association Test. Stage 1: The online survey was administered in Qualtrics [included vignette exposure, capacity judgement (Likert-type rating of medical DMC), and explicit bias task (KAOP)]. Stage 2: the computer task (Age-IAT) was administered using Inquisit. All participants were instructed to download the Inquisit application to undertake the task and create a unique code for data matching back to the questionnaire. A total of 31 participants did not undertake this task, or their data could not be matched to their questionnaire. The comprehension test was a 4-item first letter prompted recall task for the name of the traditional medical DMC model elements (e.g. what does ‘A’ stand for?), with failure defined as three or more incorrect.

Figure 3.

DMC = decision-making capacity. DMC likelihood (with 95% confidence intervals) by vignette client age and abilities (N = 147).