One Size Fits None. How can we do better? using patient reported experience measure findings to drive local quality improvement across wards in a large Australian metropolitan hospital.

Abstract

Patient reported experience measures (PREMs) are being collected across entire jurisdictions, resulting in large volumes of rich qualitative patient feedback. However, this collection of data is often not connecting with local quality improvement efforts. This study aims to answer the question: "Are there meaningful differences in the patient experience of care, as measured through qualitative survey feedback, among wards at a large metropolitan hospital?" to assess the need to analyse PREMs data at a ward level to identify actionable insights. We utilise 6-months of PREMs surveys from a jurisdictional level survey in a large metropolitan hospital in Australia, focusing on Gynaecology, Maternity, Surgical and Short Stay wards. Responses to two qualitative questions concerning (i) what was good about their care, and (ii) what could be improved about their care were analysed using a semi-automated machine learning based content analysis tool, Leximancer. We performed a quantitative comparison between the hospital wards of the concepts identified from the text and their frequencies, estimated with the Cramer's V, and a qualitative comparison between wards of the three most prevalent concepts and the details reported by patients. In the quantitative comparison, we found a moderate association of the concepts reported between the wards (Cramer's V: 0.36-0.67). The qualitative analysis showed that even when the high-level issue being reported was shared across wards, the nuances often differed, especially for feedback related to improvements in care. Our study found there were substantial differences between the issues and details reported by patients across different wards, highlighting the importance of analysing PREMs at a ward level to inform quality improvement. We demonstrated a standardised way to analyse this data at ward level by employing semi-automated content analysis. These findings provide a clear method that health services can use to analyse PREMs data to drive on-the-ground quality improvement for patients.