Clinical Criteria for the Diagnosis of Thyroid Storm

T hyroid storm (or “thyrotoxic crisis”) represents a clinical diagnosis based upon usually exaggerated signs and symptoms of hyperthyroidism accompanied by variable manifestations of multisystem decompensation. Because of the likelihood of earlier diagnosis of hyperthyroidism today, storm is thought to occur less commonly than in prior decades, but its true incidence is unknown (1). In view of earlier reports of mortality if thyroid storm was not treated early and aggressively, criteria designed to establish the diagnosis were proposed (2) and have been generally well accepted. In this issue of Thyroid, Akamizu et al. (3) report on a far-reaching attempt to both establish an estimate of the incidence of the syndrome as well as potentially enhance the criteria for its diagnosis. As was also the case for the Burch and Wartofsky publication (2), this was a retrospective analysis, but on a much broader scale in terms of the survey instruments applied.

The authors first developed tentative criteria for the diagnosis by an analysis of patients reported with thyroid storm in the literature. They based determination on either a definite or suspected diagnosis on five clinical variables, with central nervous system (CNS) manifestations, including Glasgow coma score, being a major variable, such that the diagnosis was considered valid in patients with CNS findings and one other clinical manifestation. Other criteria included atrial fibrillation, shock, multi-organ failure, hypotension, and elevated serum bilirubin. In the absence of CNS signs, the diagnosis could be met with multiple minor criteria in the setting of thyrotoxicosis.

They then conducted a nationwide survey of hospitals for the prevalence of thyroid storm over a 5-year period, identifying 671 cases, with a 53% survey response rate. Finally, a follow-up survey, with a survey response rate of 65%, was conducted to refine the diagnostic criteria as applied to 356 cases of either suspected or definite thyroid storm. Although survey responses were incomplete, the study provides useful information on the annual incidence of storm (0.2 patients/100,000 population) and the mortality rate (9.5%–11%). As a consequence of these surveys and their analysis, the authors imply that they have also refined the criteria for diagnosis of storm.

In citing possible deficiencies in the prior Burch–Wartofsky criteria for thyroid storm, the authors indicate without documentation that use of the latter scoring system is associated with possible false-positive diagnosis. Indeed, in view of the potentially high mortality that may be seen with delays in diagnosis and treatment, Burch and Wartofsky considered it prudent to err on the side of diagnosis with somewhat liberal criteria rather than delaying treatment (2). In general, the clinical features assessed as criteria for thyroid storm by Akamizu et al. (3) are not novel and do not significantly differ from those previously employed. Similar to the Burch–Wartofsky criteria, gastrointestinal manifestations including jaundice or hyperbilirubinemia were found to be of heavy weight. Akazimu et al. do not propose a “scoring system” to render the diagnosis somewhat more objective, and state that the Burch–Wartofsky scoring system was not validated when indeed it was, albeit retrospectively, against the patients' hospital records. They also indicate that the latter scoring system is complex and difficult to memorize. I am not sure why anyone would want to memorize it, and I for one have personally not done so. Finally, they state that in contrast to their criteria, hyperthyroidism was not a prerequisite for the diagnosis by the Burch–Wartofsky criteria. However, this is not the case; thyrotoxicosis was considered a given in all the patients analyzed by Burch and Wartofsky on the basis of their thyroid function test data. In this regard, it is of some concern that Akamizu et al. included some patients with normal thyroid hormone blood levels albeit low thyrotropin (TSH) levels, which could be considered problematic given that TSH may be suppressed by systemic illness alone (4).

They do acknowledge, however, that either their criteria or the Burch–Wartofsky scoring system “may be helpful in diagnosing thyroid storm.” In summary, the authors are to be commended for developing the largest and most thorough epidemiological and outcomes data on thyroid storm to date. The incidence data thereby derived for the Japanese population do not differ significantly from prior estimates, although their validity may still be questioned due to the relatively low response rates from the hospitals surveyed. Would hospitals with more cases of storm be likelier to have responded? One can only speculate how the estimates might have varied based on a greater survey response.

Clearly, the importance of using appropriate and accurate diagnostic criteria relates to the ability to make the diagnosis as soon as possible so as to start treatment and reduce mortality risk. Unfortunately, Akamizu et al. make no mention of the therapeutic interventions employed for the patients analyzed and whether successful outcomes might have correlated with the diagnostic criteria for those patients. Perhaps this information is to come in a future communication. Future assessments of diagnostic criteria could be more useful if they were linked to the therapy employed to inform us as to which therapies, either specific or supportive, were associated with good outcomes if certain clinical manifestations are present. Until such information is available, we can hope that the use of extant scoring systems for thyroid storm or those proposed by Akamizu et al. will lead to earlier diagnosis and be associated with effective implementation of treatment and successful outcomes.