Author + information
- Brahmajee K. Nallamothu, MD, MPH, FACC,
- Saeed Payvar, MD,
- Yongfei Wang, MS,
- Mikhail Kosiborod, MD,
- Frederick A. Masoudi, MD, MSPH, FACC,
- Edward P. Havranek, MD, FACC,
- JoAnne M. Foody, MD,
- S. Ward Casscells, MD, FACC and
- Harlan M. Krumholz, MD, SM, FACC⁎ ()
- ↵⁎333 Cedar Street, P.O. Box 208088, New Haven, Connecticut 06520-8088
To the Editor:
Activation of the renin-angiotensin system in advanced heart failure may result in low body temperatures (1). Despite this theoretical association, body temperature is rarely reported in patients with heart failure, and no large population-based study has examined its association with mortality. We hypothesized that low body temperatures would be associated with worse survival when compared with patients with normal or elevated body temperatures.
To test this hypothesis, we used data from the National Heart Care (NHC) Project, a large national quality-of-care initiative in hospitalized patients with heart failure. Details regarding the NHC Project are described elsewhere (2). Briefly, fee-for-service Medicare beneficiaries hospitalized with a principal discharge diagnosis of heart failure were identified between April 1998 and March 1999 or July 2000 and June 2001. Discharged patients were sorted by age, gender, race, and treating hospital, and up to 800 patients per state were selected randomly. Clinical data were collected using trained abstractors, and data quality was ensured using random record re-abstraction (2). For this analysis, we excluded patients: 1) age ≤65 years (n = 6,558); 2) not admitted with a primary diagnosis of heart failure (n = 5,003); 3) with severe systemic illnesses such as human immunodeficiency virus or metastatic cancer (n = 6,216); 4) receiving hemodialysis (n = 549); 5) transferred from another hospital (n = 2,419); 6) with previous admissions (n = 3,732); or 7) with missing or extreme values for body temperature (<32.2°C or >43.3°C) (n = 464).
Patients were grouped a priori into three categories: 1) <36°C; 2) 36°C to 38°C; and 3) >38°C. The primary outcomes we evaluated were in-hospital and one-year mortality. We used multivariable survey logistic regression models to determine the independent association of low body temperature with outcomes after adjusting for covariates that were identified by unadjusted analyses as potentially associated with body temperature or mortality. These included age, gender, race, admission season, method of body temperature measurement, noncardiovascular comorbidities including acute infections, cardiovascular comorbidities including left ventricular systolic dysfunction, medications, vital signs, and imaging or laboratory studies. (A full list of specific covariates is available in the Appendix.) We also used multivariable fractional polynomial regression models to evaluate the sensitivity of our results to the choice of specific body temperature categories. All analyses used probability weights proportional to the inverse sampling fraction for each state and adjusted for hospital-level effects (3). Adjusted risk ratios (RRs) for body temperature were estimated from adjusted odds ratios. Statistical analyses were conducted using Stata version 8.0 (Stata Corp., College Station, Texas) and SAS version 9.0 (SAS Institute Inc., Cary, North Carolina).
Our final study population included 56,659 patients. The mean age was 79.7 (±0.05) years, 59.3% of the patients were women, 59.5% had coronary artery disease, and 19.2% had severe left ventricular systolic dysfunction. The mean body temperature was 36.5°C (±0.0045°C); 18.5% of patients had body temperatures <36°C and 1.9% had body temperatures >38°C. Patients with lower body temperatures were more likely to have thyroid diseases, severe left ventricular systolic dysfunction, and higher creatinine levels and less likely to be women or to be diagnosed with pneumonia.
Compared with patients with body temperatures between 36°C and 38°C, patients with body temperatures <36°C had a higher risk for in-hospital (adjusted RR 1.28, p < 0.001) and one-year mortality (adjusted RR 1.14, p < 0.001) (Table 1).Patients with body temperatures >38°C did not have a significantly higher risk for in-hospital mortality; however, an elevated body temperature was associated with a lower risk for one-year mortality (adjusted RR 0.80, p = 0.001). Fractional polynomial regression models showed a steep and approximately linear increase in the in-hospital and one-year mortality rates as temperatures fell below 37°C (Fig. 1).
Our findings suggest that lower body temperatures on admission are associated with an adverse prognosis in patients with advanced heart failure. Body temperature has the potential to be an important marker for assessing risk in these complex patients given that it is simple to measure and widely available. Understanding possible mechanisms behind this relationship may also offer new insights into the pathophysiology of heart failure.
Disordered thermoregulation is a recognized feature of advanced heart failure (3). An earlier report by Casscells et al. (4) documented an independent association between low body temperature and in-hospital death in 291 patients with heart failure. This report only included patients from a single institution, however, and did not examine long-term survival. How could body temperature be related to survival in heart failure? First, low body temperature may be a marker of advanced disease and greater neurohormonal activation. Another possibility is that low body temperatures directly contribute to worse survival. Although not fully related to whole-body thermoregulation, external cold temperature stimuli trigger vasoconstriction, and increase heart rates and blood pressure in heart failure (5), whereas environmental warming improves endothelial function (6). Low body temperatures may also explain, at least in part, the well-described link between seasonal changes and heart failure (7).
Of note, we also found that body temperatures >38°C were associated with a lower adjusted risk for one-year mortality. There are two potential reasons for this finding. Hospitalized patients with high body temperatures are more likely to have a precipitating event for heart failure (such as pneumonia), and those who generate a fever may represent a group with more intact cardiovascular systems.
Our study has important limitations. This was an observational study, and admission body temperatures were not collected using uniform methods. Although we adjusted for several covariates, the potential for residual confounding remains. We were also limited to a single body temperature measurement.
For a list of variables that were adjusted for in the multivariable models, please see the online version of this article.
Please note: Dr. Nallamothu is supported as a clinical scholar under a K-12 grant from the National Institutes of Health (RR017607-01). Dr. Kosiborod is currently affiliated with the Mid America Heart Institute of the University of Missouri, Kansas City, Missouri. The authors do not have potential conflicts of interest to disclose, with the exception of Drs. Casscells and Payvar, who may receive royalties from LifeSentry, Inc., which has an option to license the University of Texas Health Science Center at Houston’s patent on hypothermia. The terms of this arrangement have been reviewed and approved by the University of Texas Health Science Center at Houston in accordance with its conflict of interest policies.
- American College of Cardiology Foundation
- Masoudi F.A.,
- Rathore S.S.,
- Wang Y.,
- et al.
- Imamura M.,
- Biro S.,
- Kihara T.,
- et al.
- Stewart S.,
- McIntyre K.,
- Capewell S.,
- McMurray J.J.