Psychometric testing for validating a cancer symptom recognition measure in to UK socioeconomically deprived population: learning from ABACus trial

Problem

Challenges have been identified in delivering interventions to deprived populations due to knowledge, language and literacy barriers. The phase 3 ABACus trial (Awareness and Beliefs About Cancer) aims to improve cancer symptom awareness and help-seeking behaviours among adults living in socioeconomically deprived areas in South West Yorkshire and South Wales. This trial used four adapted scales from an internationally validated scale (awareness and beliefs about cancer (ABC) measure) to assess cancer knowledge: Cancer symptom recognition (primary outcome), Anticipated symptom presentation, Barriers to presentation and Beliefs about cancer. Language within the cancer symptom recognition scale were simplified to meet the need in the targeted population. This presentation describes the findings of the psychometric tests on the adapted ABC measure.

Approach

Language within the cancer symptom recognition scale were simplified using feedback from focus groups during phase 1 development and tested during the phase 2 feasibility study. Changes were made to accommodate understanding of those with low literacy and health literacy skills through simplification of phrasing and terminology used throughout the scale. We conducted psychometric testing to validate the modified ABC scales using baseline data collected from our phase 3 randomised controlled trial. We explored scale properties such as item functioning, responsiveness, validity and reliability. Factor analysis, Cronbach’s alpha, Pearson’s correlations and agreement were used to determine item correspondence to the underlying constructs of the outcome measures and internal consistency. Multiple linear regressions were used to explore the intervention effect on each scale and the new aggregated scores.

Findings

The modified 12 item Cancer symptom recognition scale had strong internal consistency (Cronbach’s Alpha = 0.79) and strong reliability (0.80). Results were comparable to the original ABC measures. However, a high celling effect was found among five well-known cancer symptom items. Principal component analysis resulted in a 2-factor solution. Well-known cancer symptoms formed the first factor and less well-known cancer symptoms the second. There was no difference between the intervention and control groups for the 12 item Cancer symptom recognition score but when the new 2-factor solution was explored the intervention had a positive effect on the second factor; demonstrating that the ABACus intervention improved symptom knowledge on less well-known symptoms.

Consequences

The ABC is an internationally validated scale. However, the 12 items Cancer symptom recognition scale does not function well to capture the level of awareness in deprived populations in the UK due to a high proportion of adults already recognising those ‘red flag symptoms’. Item selection should be carefully considered due to cultural attitudes and presence of campaigns about the red flag symptoms.