Injury, Int. J. Care Injured 45 (2014) 1942–1945
Contents lists available at ScienceDirect
Injury journal homepage: www.elsevier.com/locate/injury
Is body temperature an independent predictor of mortality in hip fracture patients? Murtuza Faizi a, Adam J. Farrier b, Murali Venkatesan c, Christopher Thomas d, Chika Edward Uzoigwe e,*, Siva Balasubramanian b, Robert P. Smith a a
Leicester Royal Inﬁrmary, Leicester, UK University Hospitals of North Tees, Stockton-On-Tees, UK Northampton General Hospital, Northampton, UK d University Hospitals of Coventry and Warwickshire, UK e Harcourt Building, 8 Harcourt Crescent, Shefﬁeld, UK b c
A R T I C L E I N F O
A B S T R A C T
Article history: Accepted 27 September 2014
Introduction: Admission body temperature is a critical parameter in all trauma patients. Low admission temperature is strongly associated with adverse outcomes. We have previously shown, in a prospective study that low admission body temperature is common and associated with high mortality in hip fracture patients (Uzoigwe et al., 2014). However, no previous studies have evaluated whether admission temperature is an independent predictor of mortality in hip fracture patients after adjustment for the 7 recognised independent prognostic indicators (Maxwell et al., 2008). Methods: We retrospectively collated data on all patients presenting to our institution between June 2011 and February 2013 with a hip fracture. This included patients involved in the original prospective study (Uzoigwe et al., 2014). Admission tympanic temperature, measured on initial presentation at triage, was recorded. The prognosticators of age, gender, source of admission, abbreviated mental test score, haemoglobin, co-morbid disease and the presence or absence of malignancy were also recorded. Using multiple logistic regression, adjustment was made for these potentially confounding prognostic indicators of 30-day mortality, to determine if admission low body temperature were independently linked to mortality. Results: 1066 patients were included. 781 patients, involved in the original prospective study (Uzoigwe et al., 2014), presented in the relevant time frame and were included in the retrospective study. The mean age was 81. There were 273 (26%) men and 793 (74%) women. 407 (38%) had low body temperature (<36.5 8C). Adjustment was made for age, gender, source of admission, abbreviated mental test score, haemoglobin, co-morbid disease and the presence or absence of malignancy. Those with low body temperature had an adjusted odds ratio of 30-day mortality that was 2.1 times that of the euthermic (36.5–37.5 8C). Conclusions: Low body temperature is strongly and independently associated with 30-day mortality in hip fracture patients. ß 2014 Elsevier Ltd. All rights reserved.
Keywords: Hip fracture Admission body temperature Hypothermia Pyrexia Mortality
Introduction The risk stratiﬁcation and prognostication of hip fracture patients is germane to safe clinical care. It is important for patient and family counselling, service provision, operative planning and treatment decisions [1,2]. Hu et al. performed a meta-analysis of studies examining prognosticators for hip fracture patients ,
* Corresponding author. Tel.: +44 07527907618; fax: +44 07527907618. E-mail address: [email protected]
(C.E. Uzoigwe). http://dx.doi.org/10.1016/j.injury.2014.09.024 0020–1383/ß 2014 Elsevier Ltd. All rights reserved.
with over 64,000 patients included. There was strong evidence for 12 predictors of mortality: advanced age, male gender, nursing home or facility residence, poor preoperative walking capacity, poor activities of daily living, higher American Society of Anesthesiologists (ASA) grading, poor mental state, multiple comorbidities, dementia or cognitive impairment, diabetes, cancer and cardiac disease. Maxwell et al. developed a scoring system to predict 30-day mortality for hip fracture patients . They used multivariate regression analysis to distill the most potent predictors of mortality from the risk factors identiﬁed by previous researchers. They reported the seven most important predictors of
M. Faizi et al. / Injury, Int. J. Care Injured 45 (2014) 1942–1945
30-day mortality were increasing age, male gender, number of comorbidities (two or more), abbreviated mental test score (AMTS) on admission (six or less), haemoglobin (10 g/dl or less), the presence of malignancy and the source of admission (institutionalisation). We have previously shown that low admission body is common amongst hip fracture patients and is associated with a high mortality . This effect persisted even when taking into account patient age and American Society of Anesthesiologists (ASA) grade. However no previous studies have explored body temperature as a predictor of mortality, independent of the 7 recognised risk factors identiﬁed by Maxwell et al. . It is thus unclear if this parameter adds further prognostic information over and above that provided by these 7 predictors of mortality. This is surprising given the fact that body temperature is a fundamental physiological parameter. It is critical for the assessment of the trauma patient in every other context but appears to have been neglected in the hip fracture patient . Normothermia constitutes temperatures of 36.5–37.5 8C . Hypothermia is deﬁned as a core body temperature of less than 35 8C in the non-trauma setting. However, in the context of trauma, low body temperature is associated with such a poor prognosis that hypothermia is deﬁned as a core temperature of less than 36 8C [13,15]. In 1987 Jurkovich et al. reported a 100% mortality for trauma patients with a core temperature of less than 32 8C . Interrogation of the 1.1 million-patient strong US National Trauma Data Bank revealed a 41% mortality rate for patients involved in trauma, presenting with a core temperature less than 32 8C . An equally large study found the adverse effects of low body temperature on mortality persisted even after controlling for the degree of exsanguinations, age, sex, mechanism, injury severity score (ISS), head, chest, and abdominal injuries, Glasgow Coma Scale score, and base deﬁcit . Depending on environmental conditions signiﬁcant amounts of energy can be expending maintaining euthermia. Celi et al. reported that a drop in ambient temperature from 24 8C to 19 8C resulted in a 6% increase in energy expenditure . Marken et al. reported similar ﬁndings . Thermoregulatory control is impaired with senescence possibly to due its inherent energy demands . Trauma also has a deleterious effect on thermoregulation . There is evidence that in trauma the body adopts a more permissive attitude to hypothermia and energy is sequestered by the injury. Hence shivering, for example, is initiated at a lower core temperature in the context of trauma . Low body temperature adversely affects every cellular process on a fundamental enzymatic basis . There is incontrovertible evidence that low body temperature attracts a very poor prognosis in the context of trauma [8,9,14,15]. Hip fracture patients tend to be elderly with compromised homeostatic mechanisms. It is becoming increasingly appreciated that for the elderly patient, with limited physiological reserve, the hip fracture is the physiological equivalent to polytrauma in a young patient . The neck of femur fracture patient may therefore be vulnerable to low body temperatures that do not necessarily constitute hypothermia. We therefore used a value of 36.5 8C to deﬁne hypothermia in this population of patients. No previous studies have examined the effect of dysthermia as a prognostic indicator adjusting for the other principle predictors of hip fracture. We therefore sought to determine if low body temperature was an independent prognosticator of 30-day mortality for patients with hip fracture. Methods We collected data from all patients presenting to our institution with hip fracture between June 2011 and February 2013. This included patients involved in our prospective study
. Poly-trauma patients were excluded (ISS 16) In England, Wales and Northern Ireland, institutions are required to record a number of patient parameters. These must then be submitted to the National Hip Fracture Database. This is mandatory if they are to receive the ﬁnancial incentives in the form of the Best Practice Tariff, provided by the Department of Health for the care of hip fracture patients . From the National Hip Fracture data we determined all patients presenting to our institution during the relevant time period. From the database, patient age, gender, source of admission, AMTS and 30-day survivorship could be extricated. The presence and quantity of co-morbid disease and cancer was ascertained from the Coding Department. This unit of the institution transcribes patients’ admission details and comorbid status into an alphanumerical code. From this the tariff paid to health care institution is also evaluated. On presentation to emergency services all patients with hip fracture have their tympanic temperature recorded by a trained triage nurse in accordance with the Royal College of Nursing’s guidelines. A single model of infra-red thermometer is used (Braun 6021 Thermoscan) and the temperature taken by trained staff. Tympanic thermometers are regularly calibrated by our institution’s Medical Physics department. For the purpose of this investigation we noted the ﬁrst temperature recorded immediately on arrival at triage in the Emergency Department. No record of the ambient outside temperature was noted nor was the length of time from sustaining the hip fracture to presentation in the ED. Although this may have proven interesting, the purpose of this study was to reﬂect reality and to ascertain risk associated with patients’ presentation admission tympanic temperature. Serological investigations are performed, including haemoglobin. Hence we managed to determine the 7 most potent predictors of 30-day mortality as per the work by Maxwell , in addition to admission tympanic temperature. This enabled us to evaluate the latter as a predictor of mortality correcting for confounders. Statistical methods On univariate analysis continuous variables were compared with the ANOVA. Tukey’s post hoc test was used to determine any signiﬁcant differences. Proportions were compared with the Chi square test. Multivariate regression was performed including the seven predictors of age, gender, comorbidities (2), AMTS (6), haemoglobin (10 g/dl), source of admission (institutionalised), presence of malignancy. Admission temperature was added to this. This was treated as a trichotomous variable: the euthermic admission temperature of 36.5–37.5 8C, low body temperature with admission tympanic temperature less than 36.5 8C and pyrexial with an admission tympanic temperature of greater than 37.5 8C, which were later excluded. The euthermic range allows for diurnal variation observed. Correction for the 7 potential confounders allows for determination of any association between admission temperature and mortality.
Results 1482 patients presented to our institution in the relevant time period. This included 863 patients were involved in our prospective study . 1066 had a comprehensive dataset of which 781 were also in our prospective study . There were 612 (57.4%) patients in the euthermic cohort (36.5–37.5 8C). 407 (38.2%) had low body temperature (<36.5 8C) and 47 (4.4%) were pyrexial (>37.5 8C). There were 273 (26%) men and 793 (74%) women. The mean age was 81. There was a statistically signiﬁcant difference between the ages of the cohorts (Table 1). Tukey’s post hoc analysis showed that there was no signiﬁcant difference in the ages
M. Faizi et al. / Injury, Int. J. Care Injured 45 (2014) 1942–1945
Table 1 Admission features of the three different admission groups.
% (number) Mean age yrs (range) % Female (number) 30-day mortality %
Low body temperature (<36.5 8C)
Euthermia (36.5–37.5 8C)
38.2 (407) 82 (21–105) 71% (291) 12.5%
57.4 (612) 80 (21–100) 76% (466) 5.2%
0.004 0.62 0.0002
between those with low body temperature (<36.5 8C) and those who were euthermic. The overall 30-day mortality was 8%. However there were notable differences between the groups (p = 0.0002). Post hoc analysis identiﬁed a signiﬁcant difference between those with euthermia and those with low body temperature (p < 0.0001). On multivariate analysis, adjusting for the seven most potent short-term prognosticators included in the comprehensive work by Maxwell et al. , as potential confounders, low body temperature remained very strongly linked to 30-day mortality in hip fracture patients. Those whose admission tympanic temperature was less than 36.5 8C had an odds of death at 30-day that was 2.1 times (p = 0.003) that of patients with a normal admission tympanic temperature (36.5–37.5 8C) (Table 2). We did not ﬁnd admission haemoglobin, the presence of malignant disease, nor institutionalisation to be predictors of 30-day mortality. Discussion Our ﬁndings suggest that low admission body temperature is strongly associated with 30-day mortality, notwithstanding adjustment for the seven recognised independent predictors of mortality . In our previous prospective work we noted that low body temperature was common and associated with high mortality in hip fracture patients . However, this clinical parameter has been neglected in the study of hip fracture patients. Hence it was not clear if low admission body temperate was an independent determinant of mortality or merely secondary to one of the seven recognised prognosticators. This study suggests that the former is the case. The relationship observed in this study does not imply causation. However, in other contexts of trauma, low body temperature has been thought to be the cause of the high mortality observed in this clinical combination [8,9,14,15]. The fact that we found low admission body temperature is strongly linked to mortality even after adjustment for the 7 most potent predictors of mortality is also suggestive. Low body temperature adversely impacts upon molecular, enzymatic, cellular and organ processes . Biochemical reactions occur at a slower rate as there is less kinetic energy [14,18]. Enzymes no longer function at their
optimum temperature level [14,18]. There is thus suppression of the immune system and coagulation cascades [15,18]. The function of critical homeostatic organs such as the liver and kidney are affected directly by the low temperature. There is also an indirect effect due to the depletion of energy and ATP in attempts at maintaining euthermia . It is intuitive that low body temperature may be a cause of increased mortality. However, what is antithetical is that no previous studies have explored the role of body temperature as an independent predictor of mortality notwithstanding its primacy in the management of all other trauma patients. We considered that geriatric patients sustaining a hip fracture, with their limited physiological reserve, may be vulnerable to low body temperatures which did not necessarily equate with hypothermia (36 8C). Hence we compared all those with low body temperature (<36.5 8C) to those, who were euthermic, and not just those suffering from hypothermia. We speculate that a temperature of less than 36.5 8C may constitute ‘hypothermia’ for the elderly hip fracture patient . There is variation in body temperature dependent upon the site of monitoring. There is also diurnal variation. In the present study all recorded temperatures were tympanic temperature measures, with the same model of infra-red tympanic thermometer. This standardisation we envisage will minimise the effect of this. We did not record the modes or duration of transport of the patient to the hospital. Neither was transport environment temperature or environmental temperature noted. However, given that humans are homeotherms, euthermia is essential irrespective of the ambient temperature. Our aim was to determine any relationship between temperature on presentation and outcome, rather than to determine the causes of such a presentation. Tympanic temperature has been found to correlate well with other more invasive recordings core temperature such as via pulmonary artery catheterisation . We also selected the recognised normal range of temperatures (36.5–37.5 8C) to constitute euthermia rather than a single temperature of 37 8C  to cater for the possibility of daily variation. We included all patients presenting with hip fracture with exception of the multiply injured (ISS 16). It is possible that some of the patients with low body temperature may have been suffering from concomitant clinical or sub-clinical infection. It was our intention however to reﬂect the admission temperature and any link to mortality in this population as they present in reality rather than an artiﬁcial sub-cohort. In the present study some data were collected retrospectively. This may have impacted upon completeness. However, the data pertaining to age, gender, mortality, abbreviated mental score and source of admission are all collated contemporaneously in the National Hip Fracture Database. Clinical coding was required to determine the number of co-morbid diseases. The accuracy of the coding process has been called into question. However, this has largely been due to detail and inaccuracy of operative procedures
Table 2 Multivariate logistic regression of predictors of 30-day mortality. Parameter
Odds ratio lower bound (95%)
Odds ratio upper bound (95%)
Age at event Male vs. female Institution vs. non institution <36.5 vs. Euthermia Hb: <10 g/dl vs. 10 g/dl Comorbidities 2 vs. 2 Malignancy vs. none Ad AMTS vs. 6 vs. 6
0.064 0.985 0.504 0.737 0.582 0.983 0.313 0.930
0.0001 0.0001 0.061 0.003 0.080 0.0001 0.563 0.001
1.066 2.677 1.656 2.089 1.790 2.674 1.367 2.534
1.031 1.619 0.976 1.274 0.933 1.596 0.474 1.483
1.103 4.426 2.808 3.425 3.435 4.479 3.938 4.329
M. Faizi et al. / Injury, Int. J. Care Injured 45 (2014) 1942–1945
. Coding has been found to be more accurate particularly when the assessment of the number of comorbidities is required and in predicting mortality . Haemoglobin, the presence of malignant disease, or institutionalisation were not shown to be independent predictors of short term mortality in hip fracture patients. It is unclear why this may be the case. However, in the original paper by Maxwell et al. , both haemoglobin and institutionalization had the smallest coefﬁcient following regression analysis. It is likely that the addition of a further variable, in this case temperature, has ‘diluted’ their effect making them non signiﬁcant. There is a strong association with low body temperature and short-term mortality of hip fracture patients even after adjusting for potential confounding factors. We found this stronger than some of the other prognostic indicators. The relationship between low body temperature and mortality in trauma patients is thought to be causative [14,15]. It may also be so in this case. This study helps us better to determine which factors directly adversely affect outcome by applying current principles taken for granted in the young trauma patient, to the elderly hip fracture patients. Aggressive re-warming of cold hip fracture patients may go some way to reduce the dire mortality rates associated with this injury. This is likely to be the focus of a further study but it was necessary to ﬁrst prove that an independent relationship between admission temperature and 30-day mortality existed. Conﬂict of interests The authors conﬁrm that there is no conﬂict of interests. References  No authors listed. http://www.nice.org.uk/nicemedia/live/13489/54919/ 54919.pdf (accessed 30.08.13).  Talsnes O, Vinje T, Gjertsen JE, Dahl OE, Engesæter LB, Baste V, et al. Perioperative mortality in hip fracture patients treated with cemented and uncemented hemiprosthesis: a register study of 11,210 patients. Int Orthop 2013;37(6):1135–40.  Hu F, Jiang C, Shen J, Tang P, Wang Y. Preoperative predictors for mortality following hip fracture surgery: a systematic review and meta-analysis. Injury 2012;43(June (6)):676–85. http://dx.doi.org/10.1016/j.injury.2011.05.017.  Maxwell MJ, Moran CG, Moppett IK. Development and validation of a preoperative scoring system to predict 30 day mortality in patients undergoing hip
      
fracture surgery. Br J Anaesth 2008;101:511–7. http://dx.doi.org/10.1093/bja/ aen236. Uzoigwe CE, Khan A, Smith RP, Venkatesan M, Balasubramanian S, Isaac S, et al. Hypothermia and low body temperature are common and associated with high mortality in hip fracture patients. Hip Int 2014. http://dx.doi.org/ 10.5301/hipint.5000124. Moran CG, Forward DP. The early management of patients with multiple injuries: an evidence-based, practical guide for the orthopaedic surgeon. J Bone Joint Surg Br 2012;94:446–53. Hutchison JS, Ward RE, Lacroix J, He´bert PC, Barnes MA, Bohn DJ, et al. Hypothermia therapy after traumatic brain injury in children. Hypothermia Pediatric Head Injury Trial Investigators and the Canadian Critical Care Trials Group. N Engl J Med 2008;358(June):2447–56. http://dx.doi.org/10.1056/ NEJMoa0706930. Jurkovich GJ, Greiser WB, Luterman A, Curreri PW. Hypothermia in trauma victims: an ominous predictor of survival. J Trauma 1987;27:1019–24. Martin RS, Kilgo PD, Miller PR, Hoth JJ, Meredith JW, Chang MC. Injuryassociated hypothermia: an analysis of the 2004 National Trauma Data Bank. Shock 2005;24:114–8. Shaﬁ S, Elliott AC, Gentilello L. Is hypothermia simply a marker of shock and injury severity or an independent risk factor for mortality in trauma patients? Analysis of a large national trauma registry. J Trauma 2005;59(November (5)):1081–5. Celi FS, Brychta RJ, Linderman JD, Butler PW, Alberobello AT, Smith S, et al. Minimal changes in environmental temperature result in a signiﬁcant increase in energy expenditure and changes in the hormonal homeostasis in healthy adults. Eur J Endocrinol 2010;163:863–72. van Marken Lichtenbelt WD, Schrauwen P, van De Kerckhove S, WesterterpPlantenga MS. Individual variation in body temperature and energy expenditure in response to mild cold. Am J Physiol Endocrinol Metab 2002;282: E1077–83. Brown DJA, Brugger H, Boyd J, Paal P. Accidental hypothermia. N Engl J Med 2012;367:1930–8. Tsuei BJ, Kearney PA. Hypothermia in the trauma patient. Injury 2004;35: 7–15. Moffatt SE. Hypothermia in trauma. Emerg Med J 2013;30:989–96. Greenspan L, McLellan BA, Greig H. Abbreviated injury scale and injury severity score: a scoring chart. J Trauma 1985;25:60–4. No authors listed. http://www.hscic.gov.uk/hipfracture (accessed 31.08.13). Mallet ML. Pathophysiology of accidental hypothermia. QJM 2002;95:775–85. Seekamp A, van Griensven M, Hildebrandt F, Wahlers T, Tscherne H. Adenosine-triphosphate in trauma-related and elective hypothermia. J Trauma 1999;47(October (4)):673–83. Rotello LC, Crawford L, Terndrup TE. Comparison of infrared ear thermometer derived and equilibrated rectal temperatures in estimating pulmonary artery temperatures. Crit Care Med 1996;24:1501–6. Dalal S, Roy B. Reliability of clinical coding of hip facture surgery: implications for payment by results? Injury 2009;40:738–41. http://dx.doi.org/10.1016/ j.injury.2008.11.018. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, et al. Coding algorithms for deﬁning comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 2005;43:1130–9.