© 1997 Faculty of Public Health Medicine of the Royal Colleges of Physicians of the United Kingdom
research-article |
Regional variation in intervention rates: what are the implications for patient selection?
Nick Black, Professor of Health Services Research
Joanne Griffiths, Research Fellow in Statistics
Mark.E. Glickman, Professor of Statistics
Health Services Research Unit, Department of Public Health and Policy, London School of Hygiene and Tropical Medicine Keppel Street, London WC1E 7HT
Department of Mathematics, Boston University 111 Cummington St, Boston, MA 02215, USA
Address correspondence to Professor Black
BACKGROUND: Whereas geographical variations in intervention rates are well recognized, little is known about their implications for patient selection. This study looks at how the relative probability of being treated in different regions within England vary with a person's need for treatment, and whether higher intervention rates are associated with a greater probability of treatment at all levels of need or confined to only certain levels.
METHODS: The method was modelling of retrospective data from population surveys, patient cohort studies and population intervention rates. Two southern regions (SW Thames and Wessex) and two northern regions (Northern and Mersey) were compared. Subjects were men aged 55 years and above in the population with urinary symptoms suggestive of benign prostatic hyperplasia and men undergoing surgical treatment. The ratio of probability of surgery in the southern regions by level of symptom severity was determined.
RESULTS: The rate of surgery in the southern regions was 26.5 per cent higher than in the north. A higher proportion of patients in the north had severe symptoms before surgery (58 per cent vs 52 per cent; p = 0.002). The probabilities of being operated on in a given year varied by symptom severity in both the north and the south. The probability was higher in the south at all levels of symptom severity: none/mild (ratio = 1.44; p>0.01), low-moderate (ratio=1.35; p = 0.003), high-moderate (ratio =1.53; p< 0.0001), and severe (ratio= 1.15; p>0.01). On testing the sensitivity of the key assumptions by assuming a more severe distribution of symptoms in the south, the differences at none/mild and low-moderate symptom levels were enhanced but differences at high-moderate and severe symptom levels were reversed.
CONCLUSIONS: As few men with mild symptoms qualify for surgery and most men with severe symptoms are operated on, any difference in patient selection between high and low rate regions is inevitably confined to the intermediate group of men with moderate symptoms. Surgeons appear to be rationing their resources in a sensible way, though perhaps not as stringently as could be achieved.
Keywords: Prostatectomy, rationing, regional variations, patient selection