Differences in therapeutic consequences of exercise testing between a rural and an urban Danish county: population based study -- Data Supplement

London School of Hygiene & Tropical Medicine
(University of London)
Keppel Street, London WC1E 7HT

Christopher Martyn
Associate Editor
British Medical Journal
BMA House
Tavistock Square
London
WC1H 9JR

Dear Chris

Re: Paper:Differences in therapeutic consequences of exercise testing in a rural and an urban Danish county.

I think this paper could be made fit for publication but the authors first need to familiarise themselves more with the published literature and clarify their theoretical perspective (i.e.) present a clear hypothesis. A bit more analysis is also needed.

With best wishes.

Yours sincerely
Nick Black
Professor of Health Services Research
and Head of Department

This paper reports on a study which compared the clinical management of patients with stable angina in a rural and an urban county in Denmark.

1.	The authors do not make clear what their hypothesis is in the Introduction. There is a clue on page 13 when they state "we therefore expected to find the same difference in bicycle exercise test rates" as they found for coronary angiography rates. If this is their theoretical starting point, they need to consult the literature. There is a considerable amount of published evidence that distance from a service is associated with its use. Specifically in the field of CHD management I know of two – one from the US (Every NR et al. NEJM 1993;329:546-51) and one I published (Black N et al. JECH 1995;49:408-412). This study is therefore a confirmatory rather than an innovative contribution. It is of interest that similar findings are obtained in Denmark as have been reported from the US and UK.
2.	Methods
	I was unclear (page 7 line 8) what was meant by clinicians deciding upon pre-examination waiting times. Does this mean they used an urgency rating scheme?
3.	Results
	Page 10 para 1: the variations in exercise test and angiography rates between hospital catchment areas quoted do not appear to take random variation into account. If this is true, then the authors need to use a method such as the Systematic Component of Variation to assess just how much the observed differences are simply random.
4.	Discussion
	The authors demonstrate that:
	(a)	levels of need (morbidity) are similar in urban and rural counties (RR 1.03)
	(b)	the likelihood of undergoing an exercise test is similar (RR 0.96)
	(c)	the likelihood of an abnormal test result is similar (RR 0.88)
	(d)	but the likelihood of undergoing angiography differs (RR 1.40) – men 1.27, women 2.06 – and was associated with distance from the facility (b = -0.78)

They offer the explanation that this was due to differences in clinical judgement (i.e.) a supply factor. For this to be true, there would have to be a systematic relationship between distance and judgement (i.e.) the further a clinician is from an angiography centre, the less likely he is to refer patients. This may be true but for economic reasons rather than cultural reasons (as suggested on page 14). In addition though, they need to consider a demand factor – that patients are less likely to want to be referred the further they have to travel. The latter is discussed on page 12 "There were no economic restrictions upon referral of patients" – maybe not to the health service but there certainly is to patients in terms of direct (travel) and indirect (loss of working time) costs.

The issue of gender differences in use of exercise tests have been investigated before. There is a new systematic review of that literature by Raine available in the October issue of J Health Services Research and Policy. The authors might want to look at it.