Using data mining technology to predict medication-taking behaviour in women with breast cancer: A retrospective study.
Chen-Chen Kuo, Hsiu-Hung Wang, Li-Ping Tseng
PMID 34156764WHAT IT FOUND
Adherence to oral hormonal therapy declined from 85.5% in year 1 to 63.9% in year 5.
Younger and older women had lower persistence. Refill gaps, age, BMI and radiotherapy were top predictors.
Key findings
01Mean adherence fell from 85.5% in year 1 to 63.9% in year 5.
02Adherence and persistence were lower among women younger than 50 years and older than 70 years than among women aged 50 to 69 years.
03The top five predictors were duration of therapy discontinuation, duration of therapy use, age at diagnosis, BMI and receipt of radiotherapy.
STILL TO COME
How it was doneWhat they foundWhat it means for RNs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
Single regional hospital in Taiwan, so results may not apply to other populations or healthcare systems. Retrospective secondary data from medical records and cancer registry; many adverse-effect details were missing or incomplete. Adherence data were available for 272 of 385 patients and persistence for 292, with complete information for both in only 227. Prescription refill records were used as a proxy for taking medicine, so the study cannot confirm that patients swallowed the tablets. Psychosocial factors such as motivation, knowledge and patient-provider communication were not measured. The database was small for the artificial neural network, and the models were not externally validated. Observational associations cannot show that any factor caused poor adherence or persistence.
Declared interests
The authors declared no conflicts of interest. The supplied text does not state a funding source, although the publication types list non-U.S. government research support.
The easy way to misread this
Do not treat the 96.37% accuracy as proof that this prediction model is ready for clinical use. It was built from one hospital's incomplete records, used refill data that cannot confirm medication intake, and was not externally validated.