Letters Variation in mortality and use of healthcare

Understanding variation in utilisation: start with health needs

BMJ 2013; 346 doi: http://dx.doi.org/10.1136/bmj.f1800 (Published 28 March 2013) Cite this as: BMJ 2013;346:f1800
  1. Michael A Soljak, clinical research fellow1,
  2. Azeem Majeed, professor of primary care1
  1. 1Department of Primary Care and Public Health, Imperial College London, London W6 8RP, UK
  1. m.soljak{at}imperial.ac.uk

In their paper on the effect of illness adjustment on regional mortality and spending rates using standard and visit corrected illness methods for adjustment, Wennberg and colleagues compare adjustment using diagnoses listed in administrative databases with additional adjustment for the frequency of doctors’ visits.1 They acknowledge that data on the use of healthcare cannot be used directly as a proxy for need or risk because these data also reflect differences in access to and supply of healthcare services. However, the methods of risk adjustment developed by these and other authors are based exactly on these data. Steventon’s editorial does not challenge this approach.2

We suggest that this problem should be approached using the fundamental public health principle of disease prevalence in a population (“health needs assessment”). Good measures of the incidence and prevalence of disease known to primary healthcare services and the prevalence of undiagnosed disease in the population (which can lead to emergency hospital admissions in particular) are needed with this approach. Of course access to, and supply of healthcare services, affect costs and utilisation, but these must be seen in the context of healthcare needs as manifested by disease incidence and prevalence and its severity, along with other measures such as frailty.

The UK is fortunate in having excellent and comparable primary care disease registers of diagnosed disease, fostered by the Quality and Outcomes Framework pay for performance programme. Previous research has developed several disease prevalence models for general practice populations and investigated the associations at practice level between diagnosed and undiagnosed disease prevalence, primary care quality and resourcing, and hospital utilisation.3 4 5 This work has also shown that, despite almost universally accessible primary care services, prevalence of undiagnosed disease in the UK is often high, especially for diseases with insidious onset, such as chronic obstructive pulmonary disease and dementia. This prevalence also varies geographically and may be associated with emergency hospital admission rates.6

We suggest that, in a population, completeness of diagnosis needs to be given similar recognition to quality of care for people with chronic diseases, and that estimated disease incidence and prevalence is the best starting point for research on healthcare utilisation.


Cite this as: BMJ 2013;346:f1800


  • Competing interests: None declared.