Thyroid storm (TS) is a rare manifestation of thyrotoxicosis. Clinical Features and Hospital Outcomes in Thyroid Storm: A Retrospective Cohort Study

Transcription

1 ORIGINAL ARTICLE Clinical Features and Hospital Outcomes in Thyroid Storm: A Retrospective Cohort Study Trevor E. Angell, Melissa G. Lechner, Caroline T. Nguyen, Victoria L. Salvato, John T. Nicoloff, and Jonathan S. LoPresti Division of Endocrinology and Diabetes (T.E.A., M.G.L., C.T.N., V.L.S., J.T.N., J.S.L.), USC Keck Medical Center, Los Angeles, California 90033; and Division of Endocrinology, Diabetes, and Hypertension (T.E.A., M.G.L.), Brigham and Women s Hospital, Boston, Massachusetts Context: Thyroid storm (TS) is a rare but life-threatening manifestation of thyrotoxicosis. Predictive features and outcomes remain incompletely understood, in part because studies comparing TS with hospitalized thyrotoxic patients have rarely been performed. Objectives: Our objectives were to compare the diagnosis and outcomes in TS versus hospitalized compensated thyrotoxic (CT) patients and to assess differences in diagnostic classification using the Burch-Wartofsky scores (BWSs) or Akamizu (Ak) criteria for identifying TS. Design, Setting, and Patients: This was a retrospective cohort study of hospitalized patients during a 6-year period at an academic tertiary hospital, with age 18 years, TSH 0.01 miu/l, and clinically diagnosed TS or CT. Outcome Measures: In-patient mortality, hospital and intensive care unit length of stay, intubation, and ventilator duration were assessed. Results: Twenty-five TS and 125 CT patients were identified and analyzed. All but 1 TS patient received thionamides, -blockade, glucocorticoids, and iodides within 24 hours of diagnosis. CT patients received thionamides and -blockade alone. In the acute hospital setting, rates of fever ( F), heart rate ( 100 beats/min), altered mentation, and a precipitating event were all higher for TS than for CT patients. Altered mentation was the only clinical feature significantly different between TS and the subset of CT patients defined as TS by BWS or Ak criteria (P.001). TS patients had greater in-patient mortality, hospital and intensive care unit length of stay, and ventilation requirements than CT patients. Conclusions: In acutely hospitalized thyrotoxic patients, the presence of central nervous system dysfunction distinguished clinically diagnosed TS from patients with BWS- or Ak-defined TS. Because TS patients had significantly worse outcomes in this study, thyrotoxic patients with possible TS and central nervous system dysfunction may derive the greatest benefit from aggressive supportive and TS-specific treatments. (J Clin Endocrinol Metab 100: , 2015) ISSN Print X ISSN Online Printed in U.S.A. Copyright 2015 by the Endocrine Society Received July 6, Accepted October 15, First Published Online October 24, 2014 Thyroid storm (TS) is a rare manifestation of thyrotoxicosis associated with substantial morbidity and mortality and requiring prompt recognition and treatment (1 5). A complete understanding of this condition remains difficult because of its rarity, nonspecific symptomatology, and variability in the diagnostic and treatment strategies applied. Traditionally, TS has been recognized as a clinical syndrome involving thyrotoxicosis, hyperthermia, alerted mentation, and a precipitating event, along with a wide array of other signs and symptoms (6 8). However, because of significant overlap between these features and other acute medical conditions (4, 6), more objective Abbreviations: AF, atrial fibrillation; Ak, Akamizu; AST, aspartate aminotransferase; BWS, Burch-Wartofsky score; CHF, congestive heart failure; CNS, central nervous system; CT, compensated thyrotoxicosis; GCS, Glasgow Coma Scale; GI, gastrointestinal; ICU, intensive care unit; INR, international normalized ratio; IQR, interquartile range; LOS, length of stay; RR, relative risk; TS, thyroid storm. doi: /jc J Clin Endocrinol Metab, February 2015, 100(2): jcem.endojournals.org 451

2 452 Angell et al Diagnosis and Outcomes in Thyroid Storm J Clin Endocrinol Metab, February 2015, 100(2): methods have been sought for the prompt and accurate diagnosing of TS. Burch and Wartofsky in 1993 (5) provided a quantitative diagnostic aid that is now considered to be precise criteria for TS (9). This method assigns points for dysfunction of the thermoregulatory, central nervous, gastrointestinal (GI)-hepatic, and cardiovascular systems, with increasing points given for greater severity of dysfunction. The authors proposed that a score 45 is highly suspicious, and very sensitive, for TS (5), and this cutoff has been widely adopted in the literature. More recently, Akamizu et al (10) analyzed TS patients reported by physician survey and generated new diagnostic criteria for TS (designated TS1 and TS2) that largely paralleled the Burch-Wartofsky scores (BWSs). In contrast to the BWS, the Akamizu (Ak) criteria are not quantitative but instead categorize patients based on the aggregate presence of defined clinical features. The new Ak diagnostic criteria have not been widely applied to separate study populations, and neither diagnostic system has been systematically applied to a cohort of hospitalized thyrotoxic patients. Given the rarity of TS, most studies have collected cases over many years (1, 7, 11), likely yielding heterogeneous populations diagnosed in different ways and treated across eras with different standards of care. Additionally, comparisons made to thyrotoxic control groups from outpatient clinics may not be relevant to distinguishing TS from compensated thyrotoxicosis (CT) in the acute setting. The Los Angeles County-University of Southern California Medical Center (LAC-USC) is a public safety-net and tertiary-care hospital, serving a population lacking access to routine medical care, often due to homelessness, immigration status, and/or mental illness (12 15). Because of this, LAC-USC continues to see advanced stages of many diseases, including TS. In this single-center retrospective cohort study covering a 6-year period at LAC-USC, TS cases were compared with concurrently admitted thyrotoxic patients rather than an outpatient or historic population. Furthermore, patients were evaluated by the same physicians and treated in a similar manner, minimizing diagnostic and treatment heterogeneity. The purpose was to evaluate currently available diagnostic criteria and to identify the features of TS most important to hospital-based outcomes, many of which have not been addressed previously in the literature. Materials and Methods Study population After Institutional Review Board approval, consecutive records of adult patients admitted to LAC-USC from January 1, 2008, to December 31, 2013, with any ICD-9 diagnostic code for hyperthyroidism or thyrotoxicosis and an undetectable TSH level ( 0.01 miu/l) were retrospectively reviewed. The beginning of the study period represents when the medical records became available in electronic format. Repeated admissions by a single patient were recorded separately. An undetectable TSH level was chosen as the initial screening criterion to maximize the specificity for clinically severe thyrotoxicosis and exclude other etiologies, including euthyroid sick syndrome or mild overtreatment with thyroid hormone. After initial inclusion, medical records were reviewed and patient encounters excluded if known or incident pan-hypopituitarism, pregnancy, levothyroxine treatment without thyrotoxic symptoms, or medications known to reduce TSH levels were present. Previously thyrotoxic patients treated with radioactive iodine or antithyroid medication without continued evidence of thyrotoxicosis, but in whom TSH was still suppressed consistent with a delayed pituitary response (16), were similarly excluded. Patient admissions with incomplete or inadequate medical records were excluded. Diagnosis of TS The diagnosis of TS was determined by the consulting endocrinologist at the time of initial assessment (ie, ) without the use of a quantitative diagnostic schema. The identification of TS was based on severe thyrotoxicosis with manifestations suggesting decompensation, such as fever, altered mental status, an inciting illness, and clinical deterioration refractory to appropriate treatments. Patients were not restratified based on subsequent hospital course or retrospective assessment by the investigators. Hospitalized patients with thyrotoxicosis not diagnosed with TS, regardless of end-organ dysfunction or eventual outcome, were defined as CT. Data collection Demographic, clinical, laboratory, and outcomes data, originally recorded as part of standard medical care, were collected from the LAC-USC Electronic Medical Records system. Patients were separated by their clinical diagnosis of TS or CT. Clinical features of particular interest were those previously specified in diagnostic criteria, including hyperthermia, central nervous system (CNS) abnormalities, tachycardia, congestive heart failure (CHF), atrial fibrillation (AF), GI-hepatic dysfunction, and an identified precipitating event. All study patients were retrospectively categorized by BWS ( 25, 25 44, or 45) and Ak categories, as previously described (5, 10) (reproduced in Supplemental Methods). The Ak categories TS1 and TS2 were combined (AkTS1/2) because their similar mortality rates suggest both define TS (10, 17). Patients not meeting criteria for AkTS1/2 were defined as AkTS0. If ambiguity with respect to a diagnostic parameter could not be resolved by a blinded second reviewer, greater severity favoring TS was assumed. Outcomes measures included hospital length of stay (LOS), intensive care unit (ICU) admission and LOS, intubation, ventilator days, and in-patient mortality. Data analysis and statistics For descriptive statistics, means SDs or medians with interquartile range (IQR) are shown as indicated for demographic, clinical, laboratory, and outcome data. Statistical comparisons were made between TS and CT patients, and between TS patients and only those CT patients with BWS 45 or AkTS1/2 (BWS 45/AkTS1/2), but not other subgroups. For the comparison of

3 doi: /jc jcem.endojournals.org 453 clinical parameters between TS and CT patients, positivity for each feature was defined as follows: 1) fever (temperature F), 2) CNS dysfunction (Glasgow Coma Scale [GCS] 15, agitation, disorientation, delirium, psychosis, seizure, lethargy/somnolence, or coma), 3) tachycardia (heart rate 100 beats/min), 4) CHF (lower-extremity pitting edema, pulmonary edema, jugulovenous distension, or cardiogenic shock), 5) AF (irregular heart rhythm confirmed by electrocardiogram), 6) GIhepatic dysfunction (nausea, vomiting, diarrhea, or unexplained jaundice), and 7) precipitating illness (a concurrent illness identified as complicating the patient s presentation). For complete details of the statistical analysis, see Supplemental Methods. Statistical analyses were performed using GraphPad Prism version 6.0 and SPSS Statistics version 21.0 (IBM), and graphs were created using GraphPad and Abode Photoshop software. Results Study population During the 6-year study period, there were acute hospital admissions to LAC-USC. Of 906 patients with a relevant ICD-9 diagnosis, 234 had a TSH 0.01 miu/l. After exclusion of nonthyrotoxic patients (n 44) and those with inadequate data (n 40), 150 thyrotoxic patients were included, of which 25 were diagnosed with TS. Table 1 summarizes the demographic characteristics of the 150 thyrotoxic patients. TS and CT patients were similar in age ( and years, respectively). Females comprised 97 of 150 patients (64.7%). The underlying etiology of thyrotoxicosis was definitively ascertained in 80 cases, with 75 (93.75%) having Graves disease, 2 with type 2 amiodarone-induced thyroiditis, and 1 case each of toxic adenoma, subacute thyroiditis, and toxic multinodular goiter, with similar proportions of new and previous diagnoses between TS and CT groups. Study patients were retrospectively categorized according to BWS and Ak criteria to evaluate the stratification of patients using these systems. As shown in Figure 1A, 44 of 150 patients (29.3%) had BWS 45, and an additional 55 (36.7%) had BWS of 25 to 44. Applying Ak criteria to the 150 patients, 30 (27.3%) and 11 (7.3%) were Ak TS1 and TS2, respectively. There were 47 patients who met either the BWS 45 or AkTS1/2 definition of TS, of whom 38 (81%) met both criteria. When comparing the clinical diagnosis of TS to the BWS and Ak designations, 20 of 25 TS patients had BWS 45 and AkTS1/2 criteria, whereas the remaining 5 TS patients exhibited BWS 25 to 44 and were AkTS0. Of the 125 patients clinically diagnosed as CT and not treated for TS, 27 were BWS 45/AkTS1/2. Clinical features The presence of clinical features were compared between TS and CT patients, as shown in Figure 1C. Consistent with the criteria used to make the diagnosis, those with TS were significantly more likely to have fever (relative risk [RR] 5.98 [95% confidence interval ], P.001), altered mentation (RR 8.87 [ ], P.001), and an identified precipitating event (RR 4.24 [ ], P.05). In 18 of 25 patients Table 1. Demographic Characteristics of Study Patients All Patients n 150 Thyroid Storm (TS) n 25 Thyrotoxic (CT) n 125 Age, mean years SD Sex M 53 (35.3) 9 (36.0) 44 (35.2) 9 (33.3) F 97 (64.7) 16 (64.0) 81 (64.8) 18 (66.6) Diagnosis of thyroid disorder New presentation 69 (46.0) 10 (40.0) 59 (47.2) 12 (44.4) Known diagnosis 71 (47.3) 14 (56.0) 57 (45.6) 13 (48.2) Unclear 10 (6.7) 1 (4.0) 9 (7.2) 2 (7.4) Season Winter 31 (20.7) 4 (16.0) 27 (21.6) 4 (14.8) Spring 31 (20.7) 7 (28.0) 24 (19.2) 6 (22.2) Summer 47 (31.3) 7 (28.0) 40 (32.0) 10 (37.0) Fall 41 (27.3) 7 (28.0) 34 (27.2) 7 (26.0) Admitting diagnosis Thyroid disorder 34 (22.8) 6 (24.0) 28 (22.6) 4 (14.8) Cardiovascular 37 (24.8) 2 (8.0) 35 (28.2) 10 (37.0) Respiratory 8 (5.4) 2 (8.0) 6 (4.8) 0 (0.0) Gastrointestinal 18 (12.1) 2 (8.0) 16 (12.9) 3 (11.1) Trauma/Musculoskeletal 10 (6.7) 4 (16.0) 6 (4.8) 1 (3.7) Infection/Fever 9 (6.0) 2 (8.0) 7 (5.6) 4 (14.8) Neurologic 14 (9.4) 3 (12.5) 11 (8.9) 1 (3.7) Malignancy/Other 12 (12.8) 4 (16.0) 15 (12.1) 4 (14.8) CT BWS >45 or AkTS1/2 n 27 Number of patients (% of group total) are shown for each category, except age, as indicated. Burch Wartofsky Score (BWS), Akamizu (Ak).

4 454 Angell et al Diagnosis and Outcomes in Thyroid Storm J Clin Endocrinol Metab, February 2015, 100(2): Figure 1. Presenting clinical features and laboratory data for TS and CT patients. A, Study patients stratified by BWSs and Ak criteria. B, Polar chart displaying the frequency of composite presenting clinical features in TS and CT patients. The scale moves from 0% to 100% from center to outer line, with divisions every 20%. All features are reported dichotomously: fever (temperature F), tachycardia (heart rate of 100 beats/min), altered mentation (including presence of GCS 15, agitation, delirium, psychosis, lethargy, seizures, or coma), GI-hepatic signs (diarrhea, nausea, emesis, unexplained jaundice, or abdominal pain), signs of CHF (lower-extremity pitting edema, pulmonary edema, jugulovenous distension, or cardiogenic shock) AF, and a identified precipitating stressor. C and D, Frequency of traditional signs and symptoms of TS in TS and CT patients overall (C) or the subgroup of CT patients with BWS 45 or AkTS1/2 (D). Significantly different frequencies between TS and CT patients are indicated with asterisks: *, P.05; **, P.01; *** P.001. classified as having altered mental status, 6 (33%) had mild agitation, 2 (11.1%) had psychosis, 4 (22.2%) had extreme lethargy, 3 (16.7%) had delirium, and 3 (16.7%) others were unresponsive. Although not a specific consideration, the presence of tachycardia was also significantly associated with a diagnosis of TS (RR [ ], P.01). The prevalence of GI-hepatic manifestations, CHF, and AF did not vary significantly between TS and CT patients. When comparing TS patients (n 25) with CT patients with BWS 45/AkTS1/2 (n 27), a higher rate of altered mentation in TS patients was the only factor significantly different between the 2 groups (RR 3.08 [ ], P.0001) (Figure 1D). To better understand patient stratification, the clinical features of patients with incongruous diagnostic findings were further compared. As shown in Table 2, of the 27 CT patients with BWS 45/AkTS1/2, 7 of 27 (25.9%) were febrile ( F) and 6 of 27 (22.2%) had evidence of CNS dysfunction, although no patient had both. In the 6 patients with altered mentation, 2 had psychosis likely secondary to premorbid and untreated schizophrenia. In the remaining 4 cases, in addition to absent fever, tachycardia was mild, with a median heart rate of (range ) beats/min. Clinical features of tachycardia, GIhepatic dysfunction, AF, and CHF were present in 25, 17, 14, and 9 of these 27 patients, respectively. Conversely, in the 5 TS patients without BWS 45 or AkTS1/2 criteria, 2 (40%) had fever and 4 (80%) had altered mentation (agitation in 2 patients and extreme lethargy and coma in 1 case each). However, in these patients, none had AF or overt GI-hepatic manifestations, and only 1 had evidence of CHF, likely accounting for the lack of qualifying criteria under the BWS and Ak methods. Table 2. Clinical Features of Patients With Discordant Thyroid Storm Diagnosis Thyroid Storm (TS) BWS <45, AkTS0 (n,%) Thyrotoxic (CT) BWS >45 or AkTS1/2(n,%) Number of patients 5 27 Sex (F:M) 3:2 18:9 Age, median years (range) 44 (36 58) 40 (21 84) Fever, Temp F (%) 2 (40) 8 (29.6) CNS dysfunction (%) 4 (80) 6 (22.2) Tachycardia, HR 100 bpm (%) 5 (100) 25 (92.6) GI-hepatic dysfunction (%) 0 (0) 17 (63) Atrial Fibrillation (%) 0 (0) 14 (51.9) Congestive Heart Failure (%) 0 (0) 9 (33.3) Precipitating event (%) 2 (40) 17 (63) Burch Wartofsky Score (BWS), Akamizu (Ak), Central nervous system (CNS), Heart rate (HR), Gastrointestinal (GI).

5 doi: /jc jcem.endojournals.org 455 Table 3. Selected Laboratory Measurements in Thyroid Storm and Compensated Thyrotoxicosis Patients Laboratory Study Reference Range Thyroid Storm (TS) (n 25) Thyrotoxic (CT) (n 125) Free Thyroxine ng/dl 5.21 ( ) 3.93 ( ) Potassium mmol/liter 4.10 ( ) 4.00 ( ) Glucose mg/dl 111 ( ) 108 ( ) Corrected calcium mg/dl 9.28 ( ) 9.51 ( ) BUN 8 22 mg/dl 15.0 ( ) 14.0 ( ) a Creatinine mg/dl 0.57 ( ) 0.51 ( ) Alkaline phosphatase U/liter 122 (82 170) 119 ( ) Total bilirubin mg/dl 0.90 ( ) 0.60 ( ) b INR ( ) 1.14 ( ) b WBC count K/uL 9.30 ( ) 7.80 ( ) a % PMN 42 78% 76.3 ( ) 64.2 ( ) b Post-hoc analysis Sodium mmol/liter 137 ( ) 139 ( ) Carbon dioxide mmol/liter 22.0 ( ) 24.0 ( ) Albumin g/dl 3.30 ( ) 3.70 ( ) c AST U/liter 42.0 ( ) 29.0 ( ) c ALT U/liter 33.0 ( ) 31.0 ( ) Direct bilirubin 0.30 ( ) 0.30 ( ) Hemoglobin M: g/dl F: g/dl 12.4 ( ) 12.6 ( ) MCV fl 85.2 ( ) 83.4 ( ) Platelet count K/uL 220 ( ) 221 ( ) Data displayed as Median values ( interquartile range). Blood urea nitrogen (BUN), alanine aminotransferase (ALT), Aspartate aminotransferase (AST), White blood cell count (WBC), Mean corpuscular volume (MCV), Polymorphonuclear cells (PMN), International normalized ratio (INR). Corrected Calcium {[4 serum albumin (g/dl)] 0.8} serum calcium (mg/dl). a P.1 (trend), b P.05, b, c P.01. Laboratory parameters To determine whether there were any biochemical markers differentiating TS from CT patients, laboratory data at the time of presentation were compared (Table 3 and Supplemental Table 1). Free T 4 was assessed in all cases and median (IQR) values in TS and CT patients of 5.2 ( ) and ( ) ng/dl, respectively, were not significantly different. Planned analyses of laboratory testing (Table 3) demonstrated a significantly greater percent neutrophil count (9.30% [7.40% 15.3%] vs 7.80% [6.33% 10.6%], P.1) and a trend toward a greater peripheral white blood cell count (76.3 [ ] vs 64.2 [ ]K/ L, P.05) in TS compared with CT patients. Evaluation of liver function revealed a significantly higher median international normalized ratio (INR) of 1.21 ( ) and 1.14 ( ) and a median serum bilirubin of 0.90 ( ) and 0.60 ( ) mg/dl in TS vs CT patients, respectively (P.05 for each), but no difference in the proportion of patients in each group with a total bilirubin value 3 mg/dl at presentation. The planned analyses found no differences among TS and CT patients with respect to serum potassium, alkaline phosphatase, corrected calcium, glucose, creatinine, or blood urea nitrogen. Exploratory post hoc comparisons of other laboratory studies found that median serum aspartate aminotransferase (AST) was significantly higher (42.0 [ ] vs 29.0 [ ] IU, P.01) and median serum albumin level significantly lower (3.30 [ ] vs 3.70 [ ] g/dl, P.01) in TS compared with CT patients. All laboratory data for TS and CT patients as well as CT patients divided into BWS and Ak groups is provided in Supplemental Figure 1. Morbidity and mortality in patients with TS and CT Differences in outcomes were evaluated for TS and CT patients, followed by comparisons between TS and only those CT patients with BWS 45/AkTS1/2 (Figure 2 and Supplemental Table 2). Treatment for TS included a thionamide, typically propylthiouracil, -blockade (primarily propranolol), stress-dose glucocorticoids (primarily dexamethasone), and supersaturated potassium iodine in all but 1 case. Treatment with thionamides and -blockade only was used in CT patients. Both groups received appropriate supportive care and aggressive evaluation and treatment of potential underlying illnesses. In-patient mortality occurred in 2 of 25 patients (8%) with TS, and no patients in any other group, yielding a mortality rate of 1.3% among all study subjects. Of mortal cases, 1 patient was a 50-year-old man with no history of

6 456 Angell et al Diagnosis and Outcomes in Thyroid Storm J Clin Endocrinol Metab, February 2015, 100(2): Figure 2. Hospital outcome measures for patients. A F, Frequency of inpatient hospital mortality (A), median (IQR) hospital LOS (B), frequency of ICU admission during hospitalization (C), median (IQR) ICU LOS (D), frequency of intubation requirement (E), and median (IQR) ventilation duration (F) in TS patients, CT patients, and the subgroup of CT patients with BWS 45 or AkTS1/2. Statistical comparisons are TS vs CT patients, and TS vs CT (BWS 45/AkTS1/2) patients. Asterisks indicate significant differences between a group and TS patients: *, P.05; **, P.01; ***, P.001; ****, P thyroid disease presenting with chest pain, palpitations, lower-extremity swelling, and diarrhea. His temperature was F and heart rate was 150 beats/min. AF and CHF with pulmonary edema were present. He did not exhibit altered mentation. The BWS was 55, and Ak TS2 criteria were met. The patient received propylthiouracil but not other TS-specific treatments because of differing physician opinions regarding treatment. During conservative treatment for non-st elevation myocardial infarction, the patient required intubation and mechanical ventilation and hemodialysis for acute kidney injury, ultimately suffering cardiovascular arrest and death. The second case was that of a 29-year-old woman with a history of Graves disease who was noncompliant with methimazole. She was brought in by family with fatigue, heat intolerance, dizziness, and lower-extremity edema and rapidly developed confusion progressing to somnolence that necessitated intubation. Her temperature and heart rate were 98 F, and 155 beats/min, respectively. Again, AF and CHF with pulmonary edema were present. Her BWS was 80, and Ak TS1 criteria were met. Supportive and TS-specific treatments were initiated, but after initial improvement, the patient lost a secure airway causing respiratory compromise and subsequent cardiovascular arrest. Postresuscitation efforts, including aggressive administration of vasopressive agents, were unsuccessful, and the patient died with shock and multiorgan failure. Hospital outcomes including inpatient LOS, ICU admission and LOS, intubation, and ventilation duration were compared between TS and CT patients. As shown in Figure 2, median hospital LOS was significantly longer in TS patients compared with all CT patients (10 [ ] vs 4 [ ] days, P.0001), and CT patients with BWS 45/AkTS1/2 (5 [ ] days, P.05). The proportion of patients requiring ICU admission during hospitalization, and the median ICU LOS were also significantly greater in TS vs CT patients (100% vs 38.4%, P.0001; and 4.5 [ ] vs 0.0 [ ] days, P.0001). The higher ICU admission rate and median ICU LOS were still observed when TS patients were compared with only CT patients with BWS 45/AkTS1/2 (100% vs 63%, P.001; and 4.5 [ ] vs 2.0 [ ] days, P.005, respectively). Intubation was required during hospitalization in 11 of 25 TS patients (44%), with median ventilator duration of 10 ( ) days. No patients without TS required intubation. Predictors of outcomes Total BWS and its individual components were evaluated for prognostic significance in patients with clinically diagnosed TS. After log transformation, BWS was linearly correlated with hospital LOS (r 0.28, P.0005) and ICU LOS (r 0.26, P.05) in TS patients (Figure 3), but regression analysis of clinical features included in the BWS did not identify any independent predictors of hospital outcomes in TS patients. Discussion Decompensated thyrotoxicosis, or TS, is frequently characterized by multiorgan system dysfunction and exaggerated manifestations of thyroid hormone excess (1, 6, 18), although sometimes presenting with a paucity of symptoms (19, 20). The nonspecific findings of thyrotoxicosis often overlap with those of severe illness, making the identification of TS difficult (6). To facilitate the diagnosis, Burch and Wartofsky (5) proposed a quantitative diagnostic system for thyrotoxic patients, with a score of 45 indicating TS. More recently, Akamizu et al (10) assem-

7 doi: /jc jcem.endojournals.org 457 Figure 3. Correlation between BWS and patient outcomes. A and B, The correlation between BWS and log-transformed hospital LOS (A) or log-transformed ICU LOS (B) in days are plotted for individual TS patients. Spearman rank order correlation (r) and P values are shown with each graph. bled a large multicenter case series of TS and formulated new criteria for diagnosing TS. However, these diagnostic systems are difficult to evaluate because of the rarity of TS and the absence of a true gold standard and thus far have not been independently compared in a real-world setting. The current single-center cohort study directly compares patients who were diagnosed with TS with a contemporary thyrotoxic control group of similar severity and acuity. Of 906 patients admitted with a diagnostic code for hyperthyroidism/thyrotoxicosis, 25 (2.75%) were diagnosed with TS, which is similar to previously reported TS rates of 1% to 7% (8, 10, 21 23). When considering only the 150 patients who presented acutely to the Emergency Department with a TSH 0.01 miu/l and required hospitalization, the rate of TS was 16.7% (25 of 150). This high percentage may suggest a liberal clinical diagnosis of TS, although many more patients would have been classified as such using either BWS or Ak criteria. The TS rate more likely reflects severe illness within the LAC- USC patient population. Application of the BWS or Ak criteria to all 150 thyrotoxic subjects revealed notable differences in patient stratification. Of TS patients, 20 of 25 met the BWS 45 and AkTS1/2 criteria for TS, whereas the remaining 5 had BWSs of 25 to 44 (suspicious for TS) and were AkTS0. This finding suggests that Ak criteria may be less sensitive for the diagnosis of TS because there is no equivalent intermediate category. However, if all patients with a BWS of are considered TS, 66% of thyrotoxic patients in this study would potentially have received treatment for TS, suggesting that BWSs inform but do not supplant physician judgment (24). Although prudent to err on the side of overdiagnosing TS to assure appropriate treatment is provided, aggressive therapy is not without possible adverse consequences. High doses of thionamide drugs may increase adverse drug reactions, such as hepatotoxicity and agranulocytosis (25), and -blockers may cause further cardiovascular decompensation (26). Hence, unnecessary treatment for TS may entail potential harm that should be carefully weighed. A better understanding of the clinical features associated with worse outcomes in TS may help identify patients with the greatest potential to benefit from aggressive therapy. The clinical diagnosis of TS in this study relied heavily on the presence of fever and altered mental status in thyrotoxic patients presenting with a precipitating illness. As would be expected, these features, as well as the presence of tachycardia, were significantly more common in patients with TS compared with CT. When TS patients were compared with only those CT patients with BWS 45/ AkTS1/2, altered mentation was the only feature more frequently found in TS. The vital importance of CNS dysfunction to the diagnosis of TS was recently recognized in Akamizu et al (10) and was incorporated as a key element to their TS1 criteria, in part explaining the closer similarity between the TS and AkTS1/2 groups in this study. The rates of GI-hepatic manifestations CHF, and AF did not differ between TS and CT patients, suggesting that in the acute hospital setting, these findings may be less informative, although small sample size may have limited the ability to detect differences in these parameters. In particular, GI-hepatic manifestations were frequent in all hospitalized thyrotoxic patients but were not associated with the clinical diagnosis of TS. This study also compares for the first time several hospital outcomes relevant to patient morbidity and healthcare use between TS and CT patients. Compared with CT controls, TS patients had a greater rate of ICU admission as well as greater median hospital and ICU LOS. These differences were smaller but preserved when comparing TS patients with the CT patients with BWS 45/AkTS1/2, and patient outcomes remained worse for 5 TS patients with BWS 45/AkTS0 compared with CT patients with BWS 45/AkTS1/2 (Supplemental Table 2). The frequency of altered mental status was a key clinical feature distinguishing TS patients, suggesting that CNS dysfunction identifies those at risk for worse outcomes. In support of this, Akamizu et al (10) found that a lower GCS, indicating poorer neurologic function, was associated with irreversible neurologic deficits after hospitalization and greater severity of illness (using APACHEII or SOFA scores). Additional factors associated with disease severity in TS patients in that study were advanced age, comorbid CHF, and higher total serum bilirubin. Although a serum bilirubin value 3 mg/dl was not associated with a clinical diagnosis of TS in the current study, both patients suffering in-patient mortality had a serum bilirubin 3

8 458 Angell et al Diagnosis and Outcomes in Thyroid Storm J Clin Endocrinol Metab, February 2015, 100(2): mg/dl, supporting the association with mortality previously reported. In this study, 2 of 25 TS patients (8%) died during hospitalization. This is consistent with the mortality rate of 7% reported by Mazzaferri and Skillman (1) and far lower than the universal mortality in early reports of TS (27, 28). Greater physician awareness, improvements in TS-specific treatments, and myriad advances in critical care medicine likely all contribute to improved TS mortality rates over time. The pathogenesis of TS remains uncertain, but the observed collective association of tachycardia, fever, and altered mentation with TS in this study suggests to the authors that hemodynamic decompensation may be of importance. Thyroid hormone-induced peripheral vasodilation to dissipate excess heat, and impaired ability to augment cardiac function with exertion (29 31), may produce susceptibility to cardiovascular decompensation when challenged, such as with volume depletion, sepsis, or other stress-induced adrenergic stimulation. The greater rate of tachycardia may indicate poorer perfusion of the peripheral circulation, causing heat retention and hyperthermia, and the CNS, instigating altered mentation. Furthermore, insufficient perfusion to meet the metabolic demands of the thyrotoxic liver (32) may account for the higher serum AST, total serum bilirubin, and INR found in TS cases. That cardiovascular decompensation may underlie critical aspects of TS argues that therapeutic interventions aimed at optimizing intravascular volume and cardiovascular function may be the most important components of treatment for TS. We recognize several limitations of the current study. Although the diagnosis and treatment of TS at a single center likely provided homogeneity in the classification and treatment of patients, the methodology for diagnosing TS may be difficult to replicate in future investigations or clinical practice. Retrospective data may contain ascertainment bias and is limited to correlative analyses. Any negative findings in this study must recognize a possible type II error given the sample size; however, the cohort design of this investigation is more robust than the casecontrol design most often employed in early studies of TS. This study was unable to assess many historical aspects, such as the interval from symptoms onset to TS, as has been done in other studies (1, 8, 10), because this information was not reliably documented, but given the frequency of altered mentation in TS patients, such history was unlikely to have been available at presentation to inform diagnostic judgment. In this study, the clinical factors associated with TS suggest decompensation of homeostatic mechanisms normally preserved in thyrotoxicosis, possibly related to cardiovascular insufficiency. However, only altered mentation distinguished TS from CT patients with BWSs 45 or Ak TS1 or TS2 categorization, suggesting that recognition of CNS dysfunction is of paramount importance in assessing for TS. Given the greater rate of mortality and adverse hospital outcomes observed in patients clinically diagnosed with TS at our institution, patients with suspected TS and altered mentation may derive the greatest benefit from aggressive TS-specific and supportive treatments. Acknowledgments Address all correspondence and requests for reprints to: Trevor E. Angell, MD, Instructor, Brigham and Women s Hospital, Endocrinology, Harvard Institutes of Medicine, 77 Avenue Louis Pasteur, HIM Suite 641, Boston, MA TANGELL@ partners.org. Disclosure Summary: All authors have no conflicts of interest to declare. References 1. Mazzaferri EL, Skillman TG. Thyroid storm. A review of 22 episodes with special emphasis on the use of guanethidine. Arch Intern Med. 1969;124: Pimentel L, Hansen KN. Thyroid disease in the emergency department: a clinical and laboratory review. J Emerg Med. 2005;28: Tietgens ST, Leinung MC. Thyroid storm. Med Clin North Am. 1995;79: Nicoloff JT. Thyroid storm and myxedema coma. Med Clin North Am. 1985;69: Burch HB, Wartofsky L. Life-threatening thyrotoxicosis. Thyroid storm. Endocrinol Metab Clin North Am. 1993;22: Ingbar SH. Management of emergencies. IX. Thyrotoxic storm. N Engl J Med. 1966;274: McArthur JW, Rawson RW, Means JH, Cope O. Thyrotoxic crisis; an analysis of the thirty-six cases at the Massachusetts General Hospital during the past twenty-five years. J Am Med Assoc. 1947;134: Waldstein SS, Slodki SJ, Kagantec GI, Bronsky D. A clinical study of thyroid storm. Ann Intern Med. 1960;52: Bahn Chair RS, Burch HB, et al. Hyperthyroidism and other causes of thyrotoxicosis: management guidelines of the American Thyroid Association and American Association of Clinical Endocrinologists. Thyroid. 2011;21: Akamizu T, Satoh T, Isozaki O, et al. Diagnostic criteria, clinical features, and incidence of thyroid storm based on nationwide surveys. Thyroid. 2012;22: Nelson NC, Becker WF. Thyroid crisis: diagnosis and treatment. Ann Surg. 1969;170: Lavarreda SA, Cabezas L, Jacobs K, Roby DH, Pourat N, Kominski GF. The state of health insurance in California: Findings from the California Health Interview Survey. Los Angeles: UCLA Center for Health Policy Research Available at: ucla.edu/publications/documents/pdf/shic2009-apr2012.pdf. Accessed November 12, The annual homelessness assessment report. Washington, DC: U.S. Department of Housing and Urban Development, Available at: pdf. Accessed November 12, Wallace SP, Torres JM, Mobari TZ, Pourat N. Undocumented and

9 doi: /jc jcem.endojournals.org 459 uninsured: barriers to affordable care for immigrant populations. The Commonwealth Fund and the UCLA Center for Health Policy Research, August Available at: org/ /media/files/publications/fund-report/2013/aug/1699_wallace_ undocumented_uninsured_barriers_immigrants_v2.pdf. Accessed November 12, Derose KP, Escarce JJ, Lurie N. Immigrants and health care: sources of vulnerability. Health Affairs. 2007;26: Uy HL, Reasner CA, Samuels MH. Pattern of recovery of the hypothalamic-pituitary-thyroid axis following radioactive iodine therapy in patients with Graves disease. Am J Med. 1995;99: Feldt-Rasmussen U, Emerson CH. Further thoughts on the diagnosis and diagnostic criteria for thyroid storm. Thyroid. 2012;22: Klubo-Gwiezdzinska J, Wartofsky L. Thyroid emergencies. Med Clin North Am. 2012;96: Lahey FH. Apathetic Thyroidism. Ann Surg. 1931;93: Feroze M, May H. Apathetic thyrotoxicosis. Int J Clin Pract. 1997; 51: Roizen M, Becker CE. Thyroid storm. A review of cases at University of California, San Francisco. Calif Med. 1971;115: Dillon PT, Babe J, Meloni CR, Canary JJ. Reserpine in thyrotoxic crisis. N Engl J Med. 1970;283: Klubo-Gwiezdzinska J, Wartofsky L. Thyrotoxic storm. In: Wass J, Steward P, eds. Oxford Textbook of Endocrinology and Diabetes. 2nd ed. Oxford, UK: Oxford University Press; 2011: Nayak B, Burman K. Thyrotoxicosis and thyroid storm. Endocrinol Metab Clin North Am. 2006;35: , vii. 25. Cooper DS. Antithyroid drugs. N Engl J Med. 2005;352: Fraser T, Green D. Weathering the storm: beta-blockade and the potential for disaster in severe hyperthyroidism. Emerg Med (Fremantle). 2001;13: Bayley RH. Thyroid crisis. Surg Gynecol Obstet. 1934;59: Maddock WG, Pedersen S, Coller FA. Studies of the blood chemistry in thyroid crisis. JAMA. 1937;109: Klein I, Danzi S. Thyroid disease and the heart. Circulation. 2007; 116: Forfar JC, Muir AL, Sawers SA, Toft AD. Abnormal left ventricular function in hyperthyroidism: evidence for a possible reversible cardiomyopathy. N Engl J Med. 1982;307: DeGroot LJ. Thyroid and the heart. Mayo Clin Proc. 1972;47: Myers JD, Brannon ES, Holland BC. A correlative study of the cardiac output and the hepatic circulation in hyperthyroidism. J Clin Invest. 1950;20: