Guest Editorial

Collecting Data on Ambulatory Care

This issue provides two research papers each grounded in the context of the National Primary Medical Care Survey (NatMedCa). NatMedCa, carried out over 2001/02, was undertaken to describe consultations between primary health care providers and their patients. The survey featured a nationally representative sample of general practitioners (GPs) and their patients. To make the picture of ambulatory care in New Zealand more complete, a sample of four Emergency Departments (EDs) spread across New Zealand was also drawn. The NatMedCa reports are available from http://www.moh.govt.nz/natmedca.

In the first paper, Peter Crampton, Roy Lay-Yee, Daniel Patrick, Peter Davis and Antony Raymont report on a â€œComparative study of electronic pilot and paper data collection methods in a survey of general practice consultations”. They find a low number of problems-per-visit and very low number of reasons-for-visit in the electronic arm, involving data collection direct from the practice management software (PMS), as compared to paper methods. Conversely, they find recoding of prescriptions to be relatively good in the electronic arm. The comparison is based on a small number of practices, and conclusions must be qualified on those grounds. Nonetheless, the paper provides a clear lesson – the mode of data collection matters. This is rather obvious, but easily forgotten where information systems are involved. It is far too easy to think, “I got the information directly from the system through which they work – it must be complete”.

The second paper is “‘Data salvage’ in hospital emergency departments: extracting usable information from electronic data systems” by Martin von Randow, Peter Davis, Antony Raymont, Roy Lay-Yee and Daniel Patrick. This paper explores the anatomy of problems encountered in aiming to get a consistent data extract from the information systems of four distinct EDs. The data resisted combination at most every conceivable level, including formatting, the availability of fields, the coding of fields and (quite probably, but only implicitly) the method of encoding variables where coding otherwise appears compatible. The paper concludes by pointing out the importance of consistent ED data to the understanding of primary care and the health care system at large. The New Zealand Health Information Service’s National Non-admitted Patient Collection (NNPAC) appears well-targeted to address the state of affairs encountered by the authors.

All data collections are merely models of reality; and models, at best, tend to be complete only in so far as their practical use. “A model of a cruise ship” could be anything as diverse as a set of artist’s conceptions of the facilities (very useful in testing the market for its passenger services) or a series of fluid flow equations (very useful in assessing fuel consumption, maximum speed and related engineering concerns). Neither model is more right than the other – each is fit to a purpose. And in this light we must view any clinical or administrative data collections. Crampton et al find the PMS to be a rich source of prescriptions because there is a functional link – the PMS data must model prescriptions accurately so the patient gets the right medications. However, data with a less strong functional link will tend to be less well populated. It would be difficult and error-prone to estimate a cruise ship’s rate of fuel consumption from its market test brochure (and the artist would rightly be put out if you said she did a bad job because the brochure did not include a comprehensive set of drawings of the engine, the propeller, etc.). As there is no one universally right model of a cruise ship, there is no one universally right data collection for a healthcare phenomenon. We can (and should!) exercise our best foresight in choosing how we collect data; but researchers will always find themselves piecing together clues from inadequate data to answer new questions.

This month’s issue also includes a review by Alec Holt of Sue Whetton’s book "Health Informatics: A Socio-Technical Perspective" giving a thorough run-down on its contents and perspective. This book provides a particularly strong focus on Australasia (with a healthy mix of Australian and New Zealand authors dominating the contributions). And, as the title suggests, views health informatics as far more than merely a technology topic.