Abstract
Background: Current evidence on the incidence and outcomes of patients with thyroid storm in the United States is limited to single-center case series. This study determined the national incidence of thyrotoxicosis with and without thyroid storm and clinical outcomes among hospitalized patients during a 10-year period in the United States.
Methods: Retrospective longitudinal analysis was conducted of clinical characteristics, mortality, hospital length of stay, and costs from 2004 to 2013. Adults (≥18 years of age) with a primary diagnosis of thyrotoxicosis with and without thyroid storm were included. To determine the incidence, outcomes, and cost of thyrotoxicosis with and without thyroid storm, the study used data from the National Inpatient Sample database, the largest public inpatient database, with a representative sample of all non-federal hospitals in the United States.
Results: Among 121,384 discharges with thyrotoxicosis during the study period (Mage ± standard error = 48.7 ± 0.11 years; 51.9% Caucasian; 77.3% female), 19,723 (16.2%) were diagnosed with thyroid storm. During the past decade, the incidence of thyroid storm ranged between 0.57 and 0.76 cases/100,000 U.S. persons per year, and 4.8 and 5.6/100,000 hospitalized patients per year. Thyroid storm was associated with significantly higher hospital mortality (1.2–3.6% vs. 0.1–0.4%, p < 0.01) and longer length of stay (4.8–5.6 vs. 2.7–3.4 mean days, p < 0.001) compared to patients with thyrotoxicosis without storm. Inflation-adjusted hospitalization costs progressively increased in patients with thyroid storm from $9942 to $12,660 between 2004 and 2013 (p < 0.01).
Conclusions: One of every six discharges for thyrotoxicosis was diagnosed with thyroid storm. Thyroid storm is associated with a 12-fold higher mortality rate compared to thyrotoxicosis without storm. The incidence and mortality of thyroid storm has not substantially changed in the past decade. However, hospitalization costs have significantly increased.
Keywords: thyrotoxicosis, thyroid storm, thyroid crisis, hospital mortality and costs
Introduction
Thyroid storm is the extreme manifestation of thyrotoxicosis, presenting as a clinical syndrome of multi-organ dysfunction, with or without a precipitating cause (1–5). The reported rates for thyroid storm are estimated between 2% and 16% of patients admitted with thyrotoxicosis, with an overall mortality rate between 8% and 30% (1–8). However, current knowledge on the incidence, mortality, and clinical outcomes of thyroid storm in the United States is based on single-center case series from several years ago (6,9,10). Population-based studies on incidence, mortality, and clinical outcomes of thyroid storm at a national level in the United States are lacking.
Only two large studies, both from Japan, have been published with a national representative population of patients with thyroid storm. In a nationwide physician survey with a response rate of 53%, Akamizu et al. reported an estimated incidence of thyroid storm of 0.20 cases/100,000 in the Japanese population per year, and 5.4% among hospitalized patients with thyrotoxicosis, with a hospital mortality rate ranging between 9.5% and 11% (8). In a second nationwide report in Japan, Ono et al. used a national discharge claim database during a two-year period, and estimated an incidence of 6.3/100, 000 hospitalized patients (11), with a similar mortality rate of 10.1%. Using data from a single center in the United States, Angell et al. (6) reported an estimated incidence of thyroid storm of 16.7% among patients hospitalized with confirmed thyrotoxicosis (25/150), with an in-hospital mortality rate of 8%.
Evidence on predisposing factors and associated conditions in patients with thyroid storm have been difficult to study and remain unknown at a national level in the United States. In large database studies in Japan, older age, central nervous system dysfunction, non-use of antithyroidal drugs and beta-blockers, need for mechanical ventilation, hemodialysis, shock, and disseminated intravascular coagulopathy were associated with higher mortality (11). This study used data from the National Inpatient Sample (NIS) database in order to obtain national estimates on incidence, predisposing factors, clinical characteristics, and hospital outcomes among hospitalized patients with thyrotoxicosis with and without thyroid storm.
Methods
Study design and data source
This was a retrospective study of discharges from U.S. hospitals (excluding long-term acute care or rehabilitation centers) with a principal diagnosis of thyrotoxicosis with and without thyroid storm. Data on predisposing factors, associated conditions, and procedures during hospitalization were analyzed in patients discharged between January 2004 to December 2013.
Data were extracted from the NIS database, part of the Healthcare Cost and Utilization Project, sponsored by the Agency for Healthcare Research and Quality (https://www.hcup-us.ahrq.gov/nisoverview.jsp). The NIS is the largest publicly available inpatient database in the United States and contains de-identified information from patients covered by Medicare, Medicaid, or private insurance, and those who are uninsured. The NIS is designed as a 20% stratified random sample of all non-federal hospitals in the United States. Hospitals are stratified based on ownership, geographic area, urban or rural location, bed size, and teaching status. The NIS collects protected clinical and resource utilization information, including patient demographics, hospital charges, length of stay (LOS), severity and comorbid measures, and vital status from each hospital discharge. The NIS collects up to 25 diagnoses—one primary and the remaining as secondary—and up to 15 procedure codes from each discharge by using the International Classification of Diseases 9th revision coding system (ICD-9 code). The database includes hospitals from 44 states, providing a sample of hospitalized patients representative of >95% of the U.S. population.
Study population
The study included adult patients aged >18 years with a principal discharge diagnosis of thyrotoxicosis with or without thyroid storm after hospitalizations for elective and non-elective admissions. Patients discharged with thyrotoxicosis as a secondary diagnosis were excluded. Discharges from the NIS included the data of patients who died during hospitalization.
Patients meeting the inclusion criteria were identified by a combination of baseline characteristic provided by the NIS and ICD-9 codes generated during the admission. The ICD-9 code system has a unified diagnostic code for thyrotoxicosis of several causes. This code is then subdivided for patients without thyroid storm by the fifth-digit system as 242.x0, and for patients with thyroid storm as 242.x1. Patients with the following ICD-9 codes were included in the “thyrotoxicosis without storm” cohort: 242.00, 242.10, 242.20, 242.30, 242.40, 242.80, and 242.90. Patients with the following ICD-9 codes were included in the “thyrotoxicosis with thyroid storm” cohort: 242.01, 242.11, 242.21, 242.31, 242.41, 242.81, and 242.91. Neonatal thyrotoxicosis (ICD-9 code 775.3) was excluded from the above cohorts. The Institutional Review Board of the Icahn School of Medicine at Mount Sinai considered this an exempt study.
Main outcomes measures
The outcomes of interest included incidence, clinical characteristics, LOS, hospital costs, predisposing factors, and hospital mortality among patients hospitalized with thyrotoxicosis with and without thyroid storm. The incidence of thyroid storm was calculated by dividing the total number of patients discharged with a primary diagnosis of thyrotoxicosis with thyroid storm by the U.S. population estimate on July 1st of each corresponding year, as provided by the U.S. Census Bureau Web site (www.census.gov/data/datasets/2016/demo/popest/nation-total.html).
Study variables and definitions
Patient characteristics are provided by the NIS and include age, sex, race/ethnicity, median yearly income in the patient's zip code, insurance type, patient's comorbidities (Charlson Comorbidity Index [CCI] for administrative data) (12), hospital location (rural or urban), geographic region (Northeast, Midwest, West, or South), hospital teaching status, hospital bed size, and season of admission. Disease severity was defined based on the CCI as low if ≤1, medium if ≥2, and high if >3. The rate of admission per seasons was also determined. Spring was defined from March 1st to May 31st, summer from June 1st to August 31st, fall from September 1st to November 30th, and winter from December 1st to February 28th.
Predisposing conditions were identified by pre-specified ICD-9 codes (as shown in Supplementary Table S1; Supplementary Data are available online at www.lieberpub.com/thy), including Graves' disease, history of noncompliance, surgical procedures, thyroid and parathyroid surgery, and pregnancy. The presence of associated conditions and procedures was established by the corresponding ICD-9 code (as shown in Supplementary Table S1), and such conditions and procedures were stratified by organ system: cardiovascular (supra- and ventricular arrhythmias, cardiac arrest, acute heart failure, acute coronary syndrome, cardiogenic shock, cardiac assist devices, cardiac catheterization, cardioversion), pulmonary (acute respiratory failure, pulmonary emboli, mechanical and non-invasive ventilation), hepatic (acute liver failure, transaminasemia), renal (acute renal failure, hemodialysis), neurological (acute ischemic stroke, encephalopathy, intracranial hemorrhage), and endocrine (diabetes ketoacidosis, adrenal insufficiency).
Weighted national estimates are presented. Information on vital status (alive or death) at discharge, LOS, and hospitalization total charges are directly provided in the NIS for each hospitalization. Total hospital charges were converted to hospitalization costs based on a charge-to-cost ratio provided by the NIS. All costs were adjusted for inflation based on the Department of Labor's consumer price index, and presented in 2013 U.S. dollars (https://www.bls.gov/cpi/cpicalc.htm). The hospital costs represent the amount needed to produce the service, not the amount paid by payer, and do not include physician fees.
Statistical analysis
Survey analysis was conducted for the NIS data that were captured by stratified random sampling. Continuous variables were summarized by weighted mean and standard errors (obtained with Taylor Series Linearization Method), and categorical values by weighted frequency and percentage. Continuous variables, such as age and LOS, were compared by design-adjusted F-tests, and categorical variables, such as predisposing factors, associated conditions, and procedures were compared by the design-adjusted Rao–Scott chi-square test. Multivariate analyses were conducted for the binary mortality outcome based on design-adjusted logistic regression and for LOS and hospital cost based on design-adjusted linear regression.
Results
Clinical and demographic characteristics
Based on weighted data from the NIS, a total of 373,915,020 discharges occurred from U.S. hospitals between 2004 and 2013. A total of 121,384 (0.032%) discharges had a primary diagnosis of thyrotoxicosis during the 10-year study period. The number of admissions for thyrotoxicosis with and without storm per year is shown in Figure 1. Of these, 19,723 (16.2%) had a diagnosis of thyroid storm. Disease severity as defined by the CCI was similar in both groups, as well as the distribution of admission per seasons.
FIG. 1.
Admissions for thyrotoxicosis with (solid bars) and without (grid pattern bars) thyroid storm in the United States, 2004–2013.
The clinical and demographic characteristics are shown in Table 1. Thyrotoxicosis without storm was most common in middle-aged adults (Mage ± standard error [SE] = 49.8 ± 0.12 years), Caucasians (53.6%), and females (77.6%). Most patients were covered by private insurance (42.8%), followed by Medicaid (27.3%) and Medicare (17.4%). The thyroid storm cohort was similarly represented by a middle-aged (42.9 ± 0.26 years), predominantly Caucasian (43.6%), female population (75.8%). Most patients were covered by private insurance (33.6%), followed by Medicare (26.5%) and Medicaid (16.6%), but up to 23.2% of patients with thyroid storm were uninsured.
Table 1.
Demographic and Clinical Characteristics of Patients Hospitalized for Thyrotoxicosis and Thyroid Storm, 2004–2013
| Characteristics | Thyrotoxicosis without storm | Thyrotoxicosis with storm | p |
|---|---|---|---|
| Patients, n (%) | 101,661 (83.8) | 19,723 (16.2) | N/A |
| Age, median (SE) | 42.9 (0.26) | 49.8 (0.12) | <0.001 |
| Age group, n (%) | <0.001 | ||
| 18–40 | 34,221 (33.6) | 9699 (49.1) | |
| >40–65 | 45,941 (45.2) | 7967 (40.4) | |
| >65 | 21,500 (21.1) | 2057 (10.4) | |
| Sex, n (%) | 0.01 | ||
| Female | 78,630 (77.6) | 14,920 (75.8) | |
| Race, n (%) | <0.001 | ||
| Caucasian | 45,104 (53.6) | 7155 (43.6) | |
| Black | 21,866 (26.0) | 5038 (30.7) | |
| Hispanic | 10,206 (12.1) | 2682 (16.3) | |
| Other | 7042 (8.4) | 1543 (9.4) | |
| Insurance, n (%) | <0.001 | ||
| Medicaid | 25,850 (27.3) | 3001 (16.7) | |
| Medicare | 16,536 (17.4) | 4782 (26.5) | |
| Private | 40,547 (42.8) | 6057 (33.6) | |
| Uninsured | 11,843 (12.5) | 4179 (23.2) | |
| Hospital type, n (%) | 0.06 | ||
| Rural | 8328 (8.2) | 1749 (8.9) | |
| Urban non-teaching | 36,996 (36.6) | 7410 (37.8) | |
| Urban teaching | 55,824 (55.2) | 10,441 (53.3) | |
| Charlson Comorbidity Index, n (%) | <0.001 | ||
| 0 | 63,321 (62.3) | 11,268 (57.1) | |
| 1 | 24,448 (24.0) | 5748 (29.1) | |
| 2 | 8416 (8.3) | 1693 (8.6) | |
| ≥3 | 5476 (5.4) | 1013 (5.1) | |
| Season of admission, n (%) | |||
| Spring | 24,312 (24) | 4592 (23.3) | 0.39 |
| Summer | 24,366 (24) | 4594 (23.3) | 0.36 |
| Fall | 23,033 (22.7) | 4589 (23.3) | 0.40 |
| Winter | 22,983 (22.6) | 4458 (22.6) | 0.99 |
SE, standard error; N/A, not available.
Hispanics and African Americans were more likely to be diagnosed with thyrotoxicosis complicated by thyroid storm compared to Caucasians (20.8% and 18.7% vs. 13.7%, respectively; p < 0.0001). Hispanics and African Americans were also younger at presentation (Mage ±SE = 36.8 ± 0.26 and 41.7 ± 0.27 vs. 46.7 ± 0.27, respectively; p < 0.0001) compared to Caucasians. However, there were no differences in mortality rates among different ethnic groups.
Predisposing factors and associated conditions in patients with thyroid storm, 2004–2013
The number of episodes with a history of Graves' disease (51.1% vs. 41.4%, p < 0.001) and history of noncompliance to medications (15.4% vs. 5.9%, p < 0.001) were significantly higher in patients with thyroid storm, as shown in Table 2. The rate of surgical procedures and thyroid or parathyroid surgery were higher in the thyrotoxicosis without storm group (39.5% vs. 9.9% and 36.7% vs 8.6%, respectively; p < 0.001).
Table 2.
Predisposing Factors for the Development of Thyrotoxicosis and Thyroid Storm, 2004–2013
| Predisposing factors | Thyrotoxicosis without storm | Thyrotoxicosis with storm | p |
|---|---|---|---|
| Graves' disease | 42,067 (41.4) | 10,082 (51.1) | <0.001 |
| History of noncompliance | 5976 (5.9) | 3037 (15.4) | <0.001 |
| Surgical procedure | 23,947 (39.4) | 1117 (9.9) | <0.001 |
| Thyroid and parathyroid surgery | 37,323 (36.7) | 1696 (8.6) | <0.001 |
Patients with thyroid storm had higher rates of associated cardiovascular conditions, including: ventricular arrhythmias (2.5 vs. 1.2%, p < 0.001), cardiac arrest episodes (1.3% vs. 0.1%, p < 0.001), heart failure (19.4% vs. 10.3%, p < 0.001), and acute coronary syndromes (1.8% vs. 0.7%, p < 0.001), as shown in Table 3. Similarly, thyroid storm patients had higher rates of acute respiratory failure (5.9% vs. 0.8%, p < 0.001) during the hospital stay.
Table 3.
Associated Conditions and Procedures During Hospitalizations for Thyrotoxicosis and Thyroid Storm, 2003–2014
| Conditions and procedures | Thyrotoxicosis without storm | Thyrotoxicosis with storm | p |
|---|---|---|---|
| Supraventricular tachycardia/atrial fibrillation/atrial flutter | 20,245 (19.9) | 5403 (27.4) | <0.001 |
| Ventricular arrhythmia | 1232 (1.2) | 496 (2.5) | <0.001 |
| Cardiac arrest | 116 (0.1) | 255 (1.3) | <0.001 |
| Heart failure | 10,432 (10.3) | 3842 (19.4) | <0.001 |
| Acute coronary syndrome | 747 (0.7) | 360 (1.8) | <0.001 |
| Cardiogenic shock | 28 (0.03) | 206 (1.0) | <0.001 |
| Cardiac catheterization | 865 (0.9) | 229 (1.2) | 0.06 |
| Cardioversion | 554 (0.5) | 282 (1.4) | <0.001 |
| Acute respiratory failure | 834 (0.8) | 1162 (5.9) | <0.001 |
| Mechanical ventilation | 891 (0.9) | 1080 (5.4) | <0.001 |
| Noninvasive ventilation | 289 (0.3) | 175 (0.9) | <0.001 |
| Pulmonary emboli | 171 (0.2) | 55 (0.27) | 0.14 |
| Acute liver failure | 113 (0.1) | 365 (1.9) | <0.001 |
| Transaminitis/hepatitis | 744 (0.7) | 340 (1.7) | <0.001 |
| Acute renal failure | 2125 (2.1) | 958 (4.9) | <0.001 |
| Hemodialysis | 563 (0.6) | 242 (1.2) | <0.001 |
| Acute ischemic stroke | 381 (0.4) | 163 (0.8) | <0.001 |
| Encephalopathy/altered mental status | 1840 (1.8) | 828 (4.2) | <0.001 |
| Diabetic ketoacidosis | 227 (0.2) | 162 (0.8) | <0.001 |
| Adrenal insufficiency | 323 (0.3) | 128 (0.7) | 0.001 |
Incidence and mortality trends of patients with thyroid storm, 2004–2013
The incidence of thyroid storm in the U.S. population ranged between 0.57 and 0.76 cases/100,000 persons per year during the 10-year study period (Fig. 1). The incidence of thyroid storm among all hospitalized patients ranged from 4.8 to 5.6/100,000 hospitalizations per year. Among hospitalizations with a primary diagnosis of hyperthyroidism, the annual incidence ranged between 14.2% and 18.4%. Thyroid storm was associated with significantly higher hospital mortality, ranging from 1.2% to 3.6%, compared to 0.1–0.4% in patients with thyrotoxicosis as shown in Figure 2 (p < 0.007, after multiple comparison adjustments).
FIG. 2.
Trends in hospital mortality in hospitalized patients with thyrotoxicosis with (solid bars) and without (grid pattern bars) thyroid storm in the United States, 2004–2013.
Hospital costs and LOS
The trend in mean hospitalization costs per admission (adjusted for inflation) progressively increased from $9942 to $12,661 in patients with thyrotoxicosis with thyroid storm (p < 0.001) between 2004 and 2013. This increasing trend in costs started to accelerate since 2007–2008 (Fig. 3). During the study period, the mean LOS ranged from 4.8 to 5.6 mean days in patients with thyroid storm compared to 2.7–3.4 mean days in patients with thyrotoxicosis without storm (Fig. 4).
FIG. 3.
Trends in hospitalization costs (U.S. dollars) for patients admitted with thyrotoxicosis with (triangles) and without (circles) thyroid storm, 2004–2013.
FIG. 4.
Length of stay (mean days ± standard error) trend for patients admitted with thyrotoxicosis with (triangles) and without (circles) thyroid storm, 2004–2013.
Multivariate analysis of mortality, LOS, and hospital costs in patients with thyroid storm
In multivariate analysis adjusting for group, year, age, sex, race/ethnicity, CCI score, and hospital type, thyroid storm was associated with a higher risk of hospital mortality (odds ratio [OR] = 10.2 [confidence interval (CI) 7.1–14.6]) compared to patients with thyrotoxicosis without storm. Patients in the younger age group (18–40 years) and in the adult age group (40–65 years) were found to have a lower mortality risk (OR = 0.28 [CI 0.16–0.49] and OR = 0.48 [CI 0.31–0.73], respectively; p < 0.05) compared to the elderly group (>65 years). A CCI score <3 (0, 1, 2) points was also associated with a lower risk of mortality (p < 0.05) compared to a CCI score ≥3.
In multivariate analysis of LOS, it was found that among patients discharged with the diagnosis of thyroid storm, advanced age (>65 years), male sex, African American ethnicity, and a CCI score ≥3 were factors associated with a prolonged LOS (p < 0.001). In terms of type of hospital, being discharged from rural hospitals was associated with a shorter LOS compared to urban non-teaching hospitals, while urban teaching and non-teaching hospital were comparable (p < 0.001).
Similarly, admission with thyroid storm, advanced age (>65 years), male sex, and a CCI score ≥3 were factors associated with increased hospitalization costs (p < 0.001). In terms of type of hospital, being discharged from an urban teaching hospital was associated with higher costs compared to urban non-teaching hospitals, while discharge from a rural hospital had the lowest cost (p < 0.001).
Discussion
To the authors' knowledge, this is the first nationwide analysis on the clinical characteristics, hospital outcomes, and mortality among hospitalized patients with thyrotoxicosis with and without thyroid storm in the United States. Based on the results, thyroid storm is a relatively rare condition, with an annual incidence ranging from 4.8 to 5.9 cases/100,000 hospitalized patients per year during the 10-year study period. Thyroid storm was most common in the middle-aged, Caucasians, and females. A history of Graves' disease and noncompliance was most commonly noticed in this group, as well as higher rates of acute cardiopulmonary conditions requiring invasive procedures. The hospital mortality in hospitalized patients with thyroid storm ranged between 1.2% and 3.6% during the 10-year study period. An increasing trend in inflation-adjusted hospitalization costs was observed in patients with thyroid storm from $9942 in 2004 to $12,660 in 2013 (p < 0.001), while the trend for LOS remained stable from 4.8 to 5.6 days across the study period.
Thyroid storm is the extreme manifestation of thyrotoxicosis and is characterized by a clinical syndrome of multi-organ dysfunction (1–5). This disorder has long been recognized as a rare condition associated with high mortality (1–10). However, the incidence and mortality estimates were based on data from single-center series from more than three decades ago (9,10). Since the publication of the clinical score by Burch and Wartosky in 1993 (1), there have been only two nationwide studies published on the incidence and mortality of thyroid storm in Japanese populations. In a nationwide physician survey with a response rate of 53%, Akamizu et al. reported an estimated incidence of thyroid storm of 0.20 cases per 100,000 in the Japanese population per year, and 5.4% among hospitalized patients with thyrotoxicosis, with a hospital mortality ranging between 9.5% and 11% (8). The authors did not provide information on the numbers of cases with thyroid storm among the general hospitalized population. In the second study by Ono et al. (11), using a national discharge claim database with approximately 21 million admissions during a two-year study period, the authors found 1324 patients with thyroid storm and estimated an incidence is 6.3/100,000 hospitalized patients, with a similar mortality rate of 10.1%. Despite differences in the methodology and estimated incidence, both studies reported a similar mortality rate of approximately 10% (8,11).
In agreement with the report by Ono et al. (11) and also using a national discharge database, the present study found a similar rate in the United States ranging from 4.8 to 5.9 cases per 100,000 hospitalized patients per year. One could expect that the current results, along with those from Ono et al. (11), are more likely to be accurate compared to the extrapolation of data from a physician survey by Akamizu et al. (8). The rates of thyroid storm ranged between 15.3% and 18.3% among patients hospitalized with a principal diagnosis of thyrotoxicosis, similar to the 16.7% reported by Angell et al. in the largest case-series (n = 150) in the United States (6). As previously reported in the literature by Akamizu et al. (8) in a large national survey and by Angell et al. in small case series (6), higher rates of Graves' disease and noncompliance to medications were noticed as predisposing factors in the present study. As expected, these patients had an increased rate of acute severe cardiopulmonary complications requiring invasive procedures. A lower rate of surgical procedures and thyroid or parathyroid surgery was also noticed in patients with thyroid storm. One could speculate that patients with thyroid storm are less likely to undergo these procedures, since it has been known for years that they can be precipitating factors (1–5).
The trend of inpatient mortality from thyroid storm remained stable from 2004 to 2013 in U.S. hospitals, but it was lower than previously reported in small and single-center studies. The mortality rate in patients with thyroid storm ranged between 1.2% and 3.6% during the 10-year study period. These rates are lower than previously reported from a single-center study in the United States that found a similar estimated incidence of thyroid storm among patients with thyrotoxicosis but a hospital mortality of 8% (6). The present study represents data from academic and nonacademic, urban and rural hospitals throughout the United States, while the previous report was limited to a tertiary academic center where the acuity and mortality may be expected to be higher. In agreement with previous studies (6,8,11), it was found that advanced age (>65 years) and a CCI score >3 were associated with increased mortality among hospitalized patients with thyrotoxicosis and thyroid storm.
A progressive inflation-adjusted increase in hospitalization costs is reported in both cohorts from 2004 to 2013, but the increase was almost double in patients with thyroid storm. The trend in costs started to accelerate progressively since 2007 for the thyroid storm cohort. From a policy perspective, these findings are remarkable, since the incidence, mortality, and LOS trends of thyroid storm were stable over the study period. Furthermore, almost half of thyroid storm patients comprised a young group (49.1% were between 18 and 40 years of age), with a very small group of elderly patients (10.4% were >65 years of age). A previous single-center study by Angell et al. showed that patients with thyroid storm required more intensive-care services (6), which is linked to increased hospitalization cost. It is suspected that the higher rate of acute severe cardiopulmonary conditions and the need for more invasive cardiac procedures and mechanical ventilation, which usually require admissions to coronary or intensive care units, may have impacted the increase in cost. In multivariate analysis, it was found that elderly males with thyroid storm and multiple comorbidities showed an independent association with higher hospitalization costs.
The high morbidity and mortality in hospitalized patients with severe hyperthyroidism and thyroid storm indicates the need for early diagnosis and treatment to prevent the progression to thyroid storm. Collaborative efforts are needed to design standardized strategies to diagnose and treat patients with thyroid storm. Clinicians should be adequately trained on the use of supportive care and thionamide drugs in patients with severe hyperthyroidism, and advised on consulting endocrinology specialists in a timely manner, since delaying treatment can lead to increased morbidity and mortality. This can be particularly challenging in regions where there is limited access to endocrinology specialists, such as rural and nonacademic hospitals.
A call-for-action was recognized to create a prospective patient registry, including clinical, laboratory, and treatment data, to inform future evidence-based management recommendations for this condition. The NIS was not developed as a research tool, but it provides valuable information on rare diseases at a national level. Nevertheless, the NIS database has been previously used for landmark studies in the thyroid field (13–16). Using ICD-9 codes always carries a risk of misclassifications. However, there is a unified diagnostic ICD-9 code for thyrotoxicosis of several causes, with or without thyroid storm. Thus, it is expected that the reliability of using these codes by clinicians is acceptable. In fact, the incidence and mortality results are in concordance with prior nationwide and single-center studies.
Several limitations to this study are acknowledged. Only patients with a principal discharge diagnosis of thyrotoxicosis were included in order to avoid the risk of including patients in whom thyroid disease was part of the past medical history and not an active condition at discharge. Thus, patients with thyroid storm as a secondary diagnosis were not included. The NIS does not contain information to link potential multiple admissions over time from a patient. The statistical analyses have assumed that all recorded admissions are independent incidences of thyroid storm or thyrotoxicosis without thyroid storm. The NIS does not contain clinical or laboratory data. Thus, it was not possible to create diagnostic criteria based on such data and make internal validations. Since the NIS does not include specific data on medications or type of treatment during the hospital admission, it was not possible to analyze the response to treatment and its impact on outcome and hospitalization costs. Moreover, there is potential for diagnosing patients with thyroid storm at discharge if they have had more interventions during the admission, or longer LOS, or suffer mortality. Nevertheless, the findings provide relevant data to understand the clinical burden and the impact on the healthcare system of this rare condition in the United States.
Conclusions
A nationwide analysis of the incidence and inpatient mortality of patients with hyperthyroidism and thyroid storm in the United States is presented. One in every six patients admitted with thyrotoxicosis was diagnosed with thyroid storm. Hospital mortality in thyroid storm was 12-fold higher than in patients with thyrotoxicosis. There have been no changes in hospital mortality over the last decade, but hospitalization costs have significantly increased.
Supplementary Material
Acknowledgments
The abstract of this manuscript was submitted and received an Honorable Mention at the Annual Meeting of the American Association of Clinical Endocrinologists, May 2017, Austin, Texas. The study used data from the Nationwide Inpatient Sample (NIS), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality. This study was partially supported by the Jacobs Family Funds and the Atlanta CTSA. G.E.U. was supported in part by the National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number UL1TR000454, and 1P30DK111024-01 from the National Institutes of Health and National Center for Research Resources.
Author Disclosure Statement
None of the authors declare potential conflicts of interest relevant to this article.
References
- 1.Burch HB, Wartofsky L. 1993. Life-threatening thyrotoxicosis. Thyroid storm. Endocrinol Metab Clin North Am 22:263–277 [PubMed] [Google Scholar]
- 2.Tietgens ST, Leinung MC. 1995. Thyroid storm. Med Clin North Am 79:169–184 [DOI] [PubMed] [Google Scholar]
- 3.Pimentel L, Hansen KN. 2005. Thyroid disease in the emergency department: a clinical and laboratory review. J Emerg Med 28:201–209 [DOI] [PubMed] [Google Scholar]
- 4.Nayak B, Burman K. 2006. Thyrotoxicosis and thyroid storm. Endocrinol Metab Clin North Am 35:663–686, vii. [DOI] [PubMed] [Google Scholar]
- 5.Klubo-Gwiezdzinska J, Wartofsky L. 2012. Thyroid emergencies. Med Clin North Am 96:385–403 [DOI] [PubMed] [Google Scholar]
- 6.Angell TE, Lechner MG, Nguyen CT, Salvato VL, Nicoloff JT, LoPresti JS. 2015. Clinical features and hospital outcomes in thyroid storm: a retrospective cohort study. J Clin Endocrinol Metab 100:451–459 [DOI] [PubMed] [Google Scholar]
- 7.Swee du S, Chng CL, Lim A. 2015. Clinical characteristics and outcome of thyroid storm: a case series and review of neuropsychiatric derangements in thyrotoxicosis. Endocr Pract 21:182–189 [DOI] [PubMed] [Google Scholar]
- 8.Akamizu T, Satoh T, Isozaki O, Suzuki A, Wakino S, Iburi T, Tsuboi K, Monden T, Kouki T, Otani H, Teramukai S, Uehara R, Nakamura Y, Nagai M, Mori M, Japan Thyroid A. 2012. Diagnostic criteria, clinical features, and incidence of thyroid storm based on nationwide surveys. Thyroid 22:661–679 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Waldstein SS SS, Kagantec GI, Bronsky D. 1960. A clinical study of thyroid storm. Ann Intern Med 52:626–642 [Google Scholar]
- 10.Mazzaferri EL, Skillman TG. 1969. Thyroid storm. A review of 22 episodes with special emphasis on the use of guanethidine. Arch Intern Med 124:684–690 [DOI] [PubMed] [Google Scholar]
- 11.Ono Y, Ono S, Yasunaga H, Matsui H, Fushimi K, Tanaka Y. 2016. Factors associated with mortality of thyroid storm: analysis using a national inpatient database in Japan. Medicine (Baltimore) 95:e2848. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Deyo RA, Cherkin DC, Ciol MA. 1992. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol 45:613–619 [DOI] [PubMed] [Google Scholar]
- 13.Biron VL, Bang H, Farwell DG, Bewley AF. 2015. National trends and factors associated with hospital costs following thyroid surgery. Thyroid 25:823–829 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Al-Qurayshi Z, Robins R, Hauch A, Randolph GW, Kandil E. 2016. Association of surgeon volume with outcomes and cost savings following thyroidectomy: a national forecast. JAMA Otolaryngol Head Neck Surg 142:32–39 [DOI] [PubMed] [Google Scholar]
- 15.Rubio GA, Koru-Sengul T, Vaghaiwalla TM, Parikh PP, Farra JC, Lew JI. 2017. Postoperative outcomes in Graves' disease patients: results from the Nationwide Inpatient Sample database. Thyroid 27:825–831 [DOI] [PubMed] [Google Scholar]
- 16.Dhital R, Basnet S, Poudel DR. 2017. Impact of hypothyroidism on occurrence and outcome of acute coronary syndrome from the national inpatient sample. Am J Cardiol 120:2160–2163 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.