Burden of illness adjusted measures abound in health care, with the purpose being to avoid misleading analyses about health care spending, utilization and provider performance. These adjustments are supposed to help account for differences in measured outcomes that might be the result of treatment of patients with different health statuses and health needs. But a study in the British Medical Journal suggests that the three most common illness burden adjusters are flawed because they do not account for the effect of different amounts of health care use by patients. (BMJ Article) The underlying premise is that when using administrative data, like claims, to determine diagnoses, it appears to be the case that the more physician visits and tests a patient has, the more diagnoses will be given to them and the sicker they will appear on the various burden of illness adjusters. Since there may be a geographic factor in such utilization variations, there may appear to be more geographic variation in outcomes like mortality or spending then there really is. Even if the variation is at the individual physician or patient level, it can affect certain measures.
If you adjust the adjusters for physician visit rates, a substantial fraction of the geographic variation in mortality and spending disappears, although a significant amount remains and is unexplained. Since illness-adjusted measures are used for many purposes, including quality reporting, health plan payments under MA and analysis of health care issues for policymakers, if these adjustments introduce errors, then they are likely creating errors in the uses to which the measures are put. On the other hand, adjusting burden of illness methods for visit rates also seems to be somewhat circular. It may be that patients have more visits because they really are sicker. It may be that some doctors are more comprehensive in identifying every possible problem a patient has. The underlying problem may be that claims data is not the best source to identify actual patient illnesses, but abstracting medical records is painstaking at this point and there is increasing concern that even medical records may be “upcoded” for reimbursement or other purposes. How sick a particular patient is appears to be harder to discover than it looks at first glance.