Further Thoughts on the Diagnosis and Diagnostic Criteria for Thyroid Storm

T hyroid storm is a rare condition with a high mortality rate. Even if diagnosed early, death can occur, and reported mortality rates have ranged from 2% to 75% in hospitalized patients (1 –3). Most immediate causes of death from fulminant thyrotoxicosis have been described as sudden cardiovascular collapse, with shock and arrhythmias including ventricular fibrillation. The cases have usually been described as resistant to all the usual resuscitative efforts (4). An accurate estimation of the incidence of thyroid storm is impossible to determine because of the considerable variability in the criteria for its diagnosis. Furthermore, prospective studies are impossible to conduct in a disease of such rarity on the one hand and acute occurrence on the other. With the available current therapeutic resources, death from thyrotoxicosis or from its treatment should be rare, and thyroid storm does indeed appear to be occurring less frequently today than in the past. This is most likely due to more widespread information among physicians and the general population on thyroid diseases in general and thyrotoxicosis in particular, as well as better diagnostic instruments to ensure earlier diagnosis and treatment of thyrotoxicosis. This improved management would thus preclude progression of thyrotoxicosis to the crisis stage. Nevertheless, as described very recently, it may occur in 1% to 2% of hospital admissions for thyrotoxicosis (5), and must therefore still be considered one of the more important and prevalent endocrine emergencies.

In the July 2012 issue of Thyroid, a landmark study by Akamizu et al. (6) described two stepwise national surveys of thyroid storm in Japan that looked at its incidence, diagnostic criteria, clinical features, mortality, and prevalence of survival with complications. The authors first developed diagnostic criteria based on 99 patients in the literature and seven of their own patients. Thyrotoxicosis was a prerequisite for both sets of diagnostic criteria. Additional criteria were a combination of classical features such as central nervous system manifestations, fever, tachycardia, congestive heart failure and gastrointestinal (GI)/hepatic disturbances. Similar parameters were also considered by Burch and Wartofsky (1).

Using these criteria, Akamizu et al. (6) subsequently conducted two nationwide surveys from 2004 to 2008, targeting all hospitals in Japan. The purpose of the first survey was to identify cases of thyroid storm. The second survey was intended to obtain detailed clinical and laboratory information regarding these cases. It is important to note that the follow-up survey did not stop with obtaining the information provided in the forms filled out by the respondents. As a further step, the information was vetted and respondents were contacted to fill in gaps in the data.

Ultimately, Akamizu et al. (6) identified 282 patients who were assigned the diagnosis of thyroid storm, grade 1 (TS1) and 74 patients who were assigned the diagnosis of thyroid storm, grade 2 (TS2). Readers should recognize that this terminology (i.e., TS1 and TS2) was developed after the authors completed the study. In developing the initial criteria for the diagnosis of thyroid storm, one that was based on a review of the literature, the authors formulated criteria for what they referred to as “definite thyroid storm” and “suspected thyroid storm.” This terminology was used in their surveys and data records. The results of the study were, however, at odds with “definite” and “suspected” as appropriate terminologies. Thus, based on the final diagnostic criteria for each of these categories, one can calculate that the mortality rate of suspected thyroid storm, 9.9%, was almost as high as the mortality rate of definite thyroid storm, 10.9%. Given these observations, and the fact that the mortality rate of thyrotoxic patients without thyroid storm was less than 1%, ^* it seemed better not to use “suspected” for a group that had such a high mortality. Hence the terminology TS1 and TS2 were used.

In addition to developing their own diagnostic criteria, Akamizu et al. (6) assigned to the patients the scores for thyroid storm proposed by Burch and Wartofsky (1) based on their seminal review of the literature. They noted that median scores (ranges) of patients with TS1 was 70 points (15–120), and those of patients with TS2 was 52.5 (25–90) points. Patients fulfilling TS1 criteria thus had a more evident disease burden and were notable for the presence of neurological disturbances. In this context it is worthwhile noting that TS1 was almost four times more prevalent in their series than TS2. This is in keeping with the classical concept of thyroid storm as being a syndrome with predominant central nervous system (CNS) manifestations.

Multiple organ failure was the most common cause of death in the patients with thyroid storm, followed by congestive heart failure, respiratory failure, arrhythmia, disseminated intravascular coagulation, gastrointestinal perforation, hypoxic brain syndrome, and sepsis. Glasgow Coma Scale results and serum urea nitrogen were associated with irreversible damages in 22 survivors. Interestingly, among the patients with thyroid storm, the occurrence of any one of the five major groups of clinical manifestations (CNS, fever, tachycardia, congestive heart failure, GI/hepatic) did not positively correlate with the occurrence of any other group of clinical manifestations. Seventy-six percent of the patients with thyroid storm had more than three major organ manifestations, consistent with multiple organ failure, which together with congestive heart failure alone was the direct cause of the 38 reported deaths.

Another interesting observation in the study was that there were similar values for serum free thyroxine and free triiodothyronine in the patients with thyroid storm, whatever their classification, as well as in the patients with thyrotoxicosis but without thyroid storm. This implies that measurement of thyroid function variables will not confidently predict the deterioration of thyrotoxicosis into thyroid storm. However, it does not necessarily mean that the thyrotoxicosis in patients without thyroid storm were similar in degree to that of patients with thyroid storm; the occurrence of multiorgan failure during development of thyroid storm will add an influence from nonthyroidal illness to the measured variables, thereby lowering first serum triiodothyronine followed by serum thyroxine (7).

As indicated by the authors, the Akamizu study (6) is the first large study to cast some clarification on the epidemiology and clinical presentation of patients with thyroid storm. It is by far the largest single series of cases of thyroid storm. The incidence of thyroid storm cases in Japan was estimated in the article to be 1283 (95% confidence interval: 1077–1489) per 5 years (0.2 persons/100,000 Japanese population/year), approximately 0.22% of all thyrotoxic patients and 5.4% of hospital admitted thyrotoxic patients. Triggers of the condition were mainly irregular or stopped use of antithyroid drug therapy and infections.

Another important contribution of the report of Akamizu et al. (6) is that it highlights the fact that an appreciable number of patients develop irreversible complications. In all, 22 such patients, or 9.9% of TS1 patients and 3.3% of TS2 patients, developed irreversible complications. This difference, though apparently large, was not considered statistically significant. However, irreversible neurological deficits were five times more prevalent in patients with TS1 than in those with TS2 (p<0.05). This is not surprising since a major distinction between TS1 and TS2 is the presence of CNS symptoms. The development of irreversible complications in patients with thyroid storm is an aspect that has not been emphasized in the literature.

Since Japan is an area with high iodine intake, it cannot be excluded that the criteria and predictions presented by Akamizu et al. (6) may not necessarily apply to other areas of the world. However, by using the literature criteria for thyroid storm as initial reference and later comparison, the authors correctly argued this to be unlikely. The study was retrospective in nature, which is always a potential source of bias, as well as the fact that only about 50%–60% of the hospitals responded to the invitation to participate in the surveys. The authors rightly mention that only prospective studies can fully clarify the true clinical features and outcome of patients with thyroid storm. This would, however, require carefully conducted multicenter participation, which is always particularly difficult in a rare condition.

The study by Akamizu et al. (6) is, to date, probably the best surrogate for a prospective study, which is a good start. It is our hope that clinical thyroidologists and epidemiologists in other countries around the world will perform similar studies in order to cover genetically different populations, different levels of iodine intake, as well as other factors. Only then can we have the confidence to deliver the best management for our patients with thyrotoxicosis.