Item Performance and Internal Consistency of the KAP–SADARI Breast Self-Examination Questionnaire among Indonesian Women: A Pilot Validation Study

Abstract

Background: Breast cancer is the most common cancer in Indonesia, and most women present at an advanced stage, so programs that teach breast self-examination (BSE) need brief instruments with sound measurement properties to judge their effect. This study examined one such instrument.

Objective: To assess item performance and internal consistency of the 13-item knowledge, attitude and practice questionnaire on BSE (KAP–SADARI).

Methods: In this cross-sectional pilot study, 30 literate Indonesian women completed the knowledge, attitude and practice domains online in July–August 2026. Uncorrected item–total correlations and Cronbach's α were analyzed with Fisher-z and Feldt intervals. Corrected item–rest correlations and α-if-item-deleted were identified exactly for the attitude domain. For knowledge and practice, integer programming bounded them over every dataset that matches the observed statistics.

Results: All uncorrected correlations exceeded the critical r of 0.361 (0.422–0.919), yet two independent attitude items would meet this criterion with probability 1.000. Practice item P3 had a corrected correlation of 0.117–0.182 in every admissible dataset, and deleting it raised α in every possible dataset. Knowledge item K3 fell below 0.361 in 97% of randomly sampled admissible marginal sets. Cronbach's α was 0.733 (95% confidence interval [CI] 0.545–0.859) for knowledge, 0.725 (0.422–0.869) for attitude and 0.758 (0.594–0.871) for practice. None exceeded 0.70 significantly (p = 0.204–0.408).

Conclusion: KAP–SADARI shows acceptable but imprecise internal consistency. It is not yet ready for classifying women or evaluating programs. P3, and probably K3, need revision. The attitude domain needs more items before confirmatory validation in at least 319 women.