Abstract
Our knowledge of the incidence of scoliosis and scoliosis surgery is based on a few small scale studies. The National Health Service (NHS) in the United Kingdom has long collected data on hospital based activity. We have used a five year English database (1998–2002) of hospital admission statistics to study age-adjusted admission rates for scoliosis (code M41 in the International Classification of Diseases, 10th revision) and for two scoliosis surgery codes (V41 ‘instrumental correction of deformity of spine’ and V42 ‘other correction of deformity of spine’ (the latter includes ‘anterolateral release of spine for correction of deformity’).
Results: Three thousand, seven hundred and eighty three patients (2533 females and 1240 males) aged 5–29 years had diagnosis M41 recorded over the five year sample period. Most of the patients were teenagers. 971 (males and females) of these had operation V41 and 1212 had V42, it is likely that the vast majority of these cases had idiopathic scoliosis. We made regional maps based on age-adjusted admission rates/100000 population. Admission rates varied from 5.75/100000 (95% confidence intervals x to y) in London to 2.8/100000 (x to y) in the Yorkshire-Humberside region.
Interpretation: There was wide geographical variation in admission rates. We considered 5 hypotheses:
-
Social deprivation – we were able to study this, and admission rates appeared independent of social deprivation.
-
Availability of spine surgeons – this may be an explanation, but not very convincing. Scoliosis surgery is concentrated in 15 centres that do not obviously link with the variations we found.
-
Variation in decision making about referral and/or treatment (by general practitioners, patients or surgeons). This is possible, but cannot be studied using our data.
-
Regional genetic variation. Some of our maps were consistent with concepts of local biological variation, but are not very convincing.
-
Incomplete or inaccurate coding in routine hospital statistics. Cannot be studied using our database alone.
Conclusion: There is wide variation in recorded rates of diagnosis and surgical treatment without obvious explanation. It might be possible to study clinical case notes, identified from the statistical database, to check whether variation is simply attributable to unreliability of coding. To determine whether there may be a genetic explanation for the geographical variation found by us, the possibility could be explored of comparing the scoliosis maps with other maps of genetic profiles of the English population.
Correspondence should be addressed to Jeremy C T Fairbank at The Nuffield Orthopaedic Centre, Windmill Road, Headington, Oxford OX7 7LD, UK
