Trams, trains, planes and automobiles: logistics of conducting a statewide audit of medical records

Margaret Flood*, Wendy Pollock, Susan McDonald, Mary Ann Davey

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

Abstract

Objective: This paper reports on the logistics of conducting a validation study of a routinely collected dataset against medical records at hospitals to inform planning of similar studies. Method: A stratified random sample of 15 hospitals and two homebirth practitioners was included. Site visits were arranged following consent. In addition to the validation of perinatal data, information was collected regarding logistics. Results: Records at 14 metropolitan and rural hospitals up to 500 km from the research centre, and two homebirth practitioners, were audited. Obtaining consent to participate took between 5 days and 10 months. Auditors visited sites on 101 days, auditing 737 medical record pairs at 16 sites. Median audit time per record was 51.3 minutes; electronic records each took 36 minutes longer than paper. Travel time accounted for nearly one-quarter of audit time. Conclusions: Delays obtaining consents, long travel times and electronic records prolonged audit duration and expense. Employment of experts maximised use of available audit time. Conducting a validation study is a time-consuming and expensive exercise; however, confidence in the accuracy of public health data is vital. Implications: Validation studies are unquestionably important. Three alternative strategies have been proposed to make future studies viable.

Original languageEnglish
Pages (from-to)448-450
Number of pages3
JournalAustralian and New Zealand Journal of Public Health
Volume40
Issue number5
Early online date15 May 2016
DOIs
Publication statusPublished - 1 Oct 2016
Externally publishedYes

Fingerprint

Dive into the research topics of 'Trams, trains, planes and automobiles: logistics of conducting a statewide audit of medical records'. Together they form a unique fingerprint.

Cite this