Cardiac troponin T: an important predictor of late death and myocardial infarction following hip fracture: an eight-year prospective observational cohort study
Cardiac troponin is an accurate marker of minimal myocardial injury and has been shown to predict adverse outcomes in various non-acute coronary disease settings. Patients with hip fracture (HF), an osteoporotic, mostly octogenarian population, are prone to sustain myocardial injury through both peri-fracture stress and subsequent surgery (osteosynthesis or hemiarthroplasty).
To assess the frequency and magnitude of myocardial injury on admission and in-hospital (post-surgery) as measured by troponin T (TnT) and to evaluate the long-term prognostic value of TnT on cardiovascular morbidity and mortality.
TnT levels were measured on admission (adTnT) and during hospital stay in 148 patients with HF and followed them prospectively until December 27, 2009. Each patient’s highest TnT value was designated peak TnT. TnT >0.1 ng/ml levels were identified as elevated, using ROC analysis to predict death. Survival was assessed using Cox regression analysis.
The study population was 81.3 ± 8.2 years of age, 32 men. Elevated adTnT and peak TnT were observed in 34 (22.9%) and 51 (34.4%) of patients, respectively. On December 27, 2009, 112 (75.7%) patients had died. After controlling for age, serum on-admission creatinine and CHF, adTnT emerged as a significant independent predictor of death (HR = 1.07, 1.02–1.12 95% CI P = 0.01). Late myocardial infarction (MI) observed in 13 patients at 1,314 ± 659 days was predicted by peak TnT (HR = 1.3, CI 1.06–1.57, P = 0.027).
Myocardial injury as evidenced by elevated TnT levels was observed in 1/3 of our patients with HF. TnT elevation, mostly observed on admission, reflected peri-fracture stress rather than surgery-related factors. Death and MI during the 8-year follow-up period were positively associated with magnitude of TnT levels during the 8-year follow-up period; Study contribution: We have demonstrated a continuous and graded association between increasing troponin levels on admission and risk of death. Previous studies have examined this association using a threshold effect. Additionally, the present study features a long follow-up time (8 years), suggesting that troponin levels on admission increase the risk of death over a prolonged period of time.
KeywordsTroponin T Hip fracture Cardiovascular events Survival
The incidence of hip fracture (HF) following low-velocity trauma to the osteoporotic proximal femur is steadily rising with increasing longevity and is accompanied by an increased early mortality rate [1, 2, 3, 4, 5]. HF occurs predominantly in female octo-and nonagenarian populations, although according to a German survey , the recent increase in HF is relatively greater among men. Post-fracture outcome is poor with a high early mortality rate, as evidenced in an English report  encompassing a 30-year follow-up period including 32,590 subjects aged 65 years or older admitted to hospital with fractured neck of femur. In the first month after fracture, standardized mortality ratios in women were 16 times higher, and those in men 12 times higher, than mortality in the same age group in the general population. Also, in a recently published Swedish study  of 766 women with HF, the authors reported 1-, 5- and 10-year survival rates of 79, 48 and 33%, respectively. Cardiovascular disease emerged as the most frequent (26.7%) cause of death in a study of 6,860 patients with hip fracture (14.7% dead at 6 months follow up) . Troponin T and troponin I are accurate markers of myocardial injury, enabling detection of even minor degrees of myocardial necrosis .
Cardiac troponins are detected in serum or heparinized plasma by using monoclonal antibodies against several different epitopes of the troponin T or I molecule. These antibodies have negligible cross reactivity to skeletal muscle. Elevated troponin levels have been observed in various medical conditions and following non-cardiac surgery [8, 9, 10, 11, 12, 13, 14, 15].
Recently, elevated troponin levels have been reported among HF patients undergoing hip surgery [16, 17, 18, 19, 20, 21, 22] and were found to be associated with adverse events during hospital stay or follow-up periods of up to 6 months. Consulting cardiologists are frequently required to evaluate patients with HF and provide clearance for non-elective hip operation. Besides, so far, no study in the literature examined the impact of troponin levels, measured on admission and during hospital stay, on cardiovascular morbidity and mortality of this population beyond short-term periods.
The objectives of the current study were originally designed to prospectively determine the incidence of elevated troponin T (TnT) in a consecutive cohort of patients with HF and correlate it with in-hospital course. Later on, we sought to estimate the predictive value of elevated TnT on death and cardiovascular events [congestive heart failure (CHF), cerebro-vascular accident (CVA) and myocardial infarction (MI)] at 8-year follow-up
Study design, population and laboratory methods
A convenience sample of 148 patients with HF admitted to the Orthopedics Department at E. Wolfson Medical Center, Holon, Israel, between October 10, 2000, and May 4, 2001, formed the patient population for the present study. Patients’ peripheral venous blood was measured for TnT (Troponin T STAT, Roche™ ) on three time points: admission; 24-h post-admission; and prior to hospital discharge. Admission and peak (most increased) levels were designated as adTnT and peak TnT, respectively. The patient care staff was unaware of TnT results in the enrolled patients.
Medical history was extracted from electronic patient medical records and charts and the hospital’s electronic archive. Vital status at end of the follow-up was confirmed through Ministry of Interior records.
Outcomes analyzed included time death, MI, CVA, CHF or coronary angiography (cath). In all analyses, observations were censored at December 27, 2009.
Biochemical analysis (M.Z., F.A.)
Venous blood samples were collected into SST tubes, centrifuged at 1,500 g for 10 min at room temperature. Half ml of the separated serum was stored at −70°C. Troponin T determinations were performed in one run, using a Roche Diagnostic TnT electro-chemiluminescent immunoassay kit on an Elecsys 1010 analyzer (Troponin T STAT, Roche Diagnostics Manheim, Germany)  according to the manufacturer’s instructions. Intra-assay variability was 4.2% for TnT 0.1 μg/l (n = 10) and 2.9% for TnT 5 μg/l (n = 10). TnT levels were expressed in ng/ml.
Analysis of data was carried out using SPSS 10.0 statistical analysis software (SPSS Inc., Chicago, IL, USA). Distributions of continuous variables were tested for normality using the Kolmogorov–Smirnov test (cut-off P < 0.01). Normally distributed continuous variables were described as mean ± standard deviation. Variables with distributions differing significantly from normal, such as troponin (TnT) and triglycerides, were described as median (min–max). Categorical variables such as comorbidities and outcome were described using frequency distributions and are presented as n (%). Normally distributed continuous variables were compared by outcome using the t test for independent samples, while continuous variables with skewed distributions were compared by outcome or inclusion using the Mann–Whitney non-parametric U. Categorical variables were compared by outcome using the chi square test. Time to death was modeled using Cox regression analysis and included age, sex, peak TnT, serum admission levels of creatinine and hemoglobin. Hazards ratios with 95% confidence intervals were calculated for predictor variables. All tests were considered significant at P < 0.05.
The study cohort was comprised of 148 patients aged 81.3 ± 8.2 years, 32 men, diagnosed with unilateral acute per-trochanteric (n = 95) or sub-capital (n = 53) fractures. All but a single patient underwent hip surgery within 2.88 ± 2.65 days,1 during his/her index hospitalization.
Patients underwent closed or open reduction and internal fixation (n = 106) or hemi-arthroplasty (n = 41). TnT levels did not differ by type of surgery or anesthesia delivery method [general (n = 44), spinal (n = 95) and epidural (n = 8)].
On follow-up, 112 (76%) patients had died. Hospitalizations for endpoints were as follows: CHF (n = 30); CVA (n = 12); cath (n = 5) or MI (n = 13, five of which were fatal). Death or non-fatal late MI was observed in 120 of cohort patients. In 28 of the 36 survivors, follow-up was uneventful.
TnT levels by hospital course
Peak TnT levels were dichotomized to either normal (<0.1 ng/ml, n = 97) or elevated (≥0.1 ng/ml, n = 51). Elevated TnT levels were most frequently observed on admission; specifically, adTnT > 0.1 ng/ml was detected in 34 patients, comprising 67% of all elevated TnT cases.
Post operative TnT elevation was observed in 29 (20%) cohort patients, including: 18 patients with normal adTnT levels but exhibited an in-hospital increase; and 11 patients with elevated adTnT who developed an additional in-hospital TnT increase.
Peak TnT associations
Length of stay
Although elevated peak TnT levels were not associated with a higher frequency of ischemic (depressed ST segment) ECG pattern (observed in 14 and 6% of patients with elevated and normal peak TnT levels, respectively (P = 0.11), peak TnT elevation was more frequently observed among the 13 patients with ischemic ECG compared to the 95 counterparts with diagnostic yet non-ischemic ECG: 7 (54%) vs. 24 (25%), P = 0.03). ECG was non-diagnostic in 40 participants, 16 (40%) of whom exhibited elevated peak TnT (≥0.1 ng/ml).
Elevated peak TnT was more frequently observed among patients with known CAD [18 (47%) of 34 CAD versus 31 (27%) of 114 non-CAD counterparts, P = 0.01]. Similarly, elevated peak TnT was more frequently observed among the 51 CVD2 patients than in patients without CVD (39 vs. 28%, P = 0.003). TnT levels were higher in the 120 patients when compared to those in 28 patients without CVD endpoints (adTnT: 0.67 ± 3.30 versus 0.017 ± 0.077, P = 0.035, peak TnT: 0.91 ± 3.1 vs. 0.05 ± 0.09, P = 0.003).
Outcomes as of December 27, 2009
Cohort patients were followed up for 3,024 days (36 survivors and 112 deceased patients; 3,243 ± 58 vs. 1,134 ± 912 days, respectively, P < 0.00001). Cumulative mortality was 7 (5%) at 30; 29 (20%) at 365 days; 79 (53%) at 1,825 days and 112 (76%) on December 27, 2009.
Patients’ characteristics and predictors of survival by December 27, 2009 t test analysis
All (n = 148)
Survivors (n = 36)
Deceased (n = 112)
81.3 ± 8.2
75.5 ± 7.3
83.1 ± 7.6
0.94 ± 0.32
0.79 ± 0.18
0.99 ± 0.33
Enlarged heart silhouette by X-ray (%)
11.89 ± 1.72
12.48 ± 1.64
11.70 ± 1.71
0.54 ± 3.0
0.028 ± 0.088
0.71 ± 3.42
adTnT ≥0.7 (ng/ml)
12.47 ± 6.7
10.60 ± 4.72
13.06 ± 7.14
Peak TnT (ng/ml)
0.74 ± 2.8
0.06 ± 0.01
0.97 ± 3.19
Peak TnT ≥0.4 (ng/ml)
Cox regression survival model
Peak TnT (ng/ml)a
Late MI was observed in 13 patients at 1,314 ± 659 days post-HF and was predicted by peak TnT (HR = 1.3, 95% CI 1.06–1.57, P = 0.013) and prolonged AOI (likelihood ratio 14.7, P = 0.027).3
Predictors of late CHF (n = 30) by Cox regression
Other endpoints: neither late CVA nor late cath was associated with TnT levels.
The present prospective observational study demonstrated consistently with previous reports [16, 17, 18, 19, 20, 21, 22] a high incidence of elevated troponin (i.e. minimal myocardial injury) in this frail mostly octogenarian population, predicting death, independent of age, sex and other known cardiovascular and coronary risk factors including diabetes mellitus, ischemic ECG and admission hemoglobin.
An important innovation in the present study stems from the fact that we have demonstrated a continuous and graded association between increasing TnT levels on admission and risk of death. Previous studies have examined this association using a threshold effect. Additionally, the present study features a long follow-up time (8 years), suggesting that troponin levels on admission increase the risk of death over a prolonged period of time.
Patients with the highest TnT levels presented with already increased admission levels; these elevated figures exceeded the TnT levels observed in patients who were admitted with normal TnT but exhibited in-hospital elevation. This may suggest that myocardial injury in the HF population may be attributed to the impact associated with hip fracture (a cascade initiating sympathetic surge or at least theoretically small MI that triggered the fall and HF) and not related to subsequent hip surgery. The greater proportion of HF patients with elevated TnT on admission is consistent with Oscarsson et al. , who reported that 58% of their patients with TnI ≥ 0.06 ng/ml post-operatively had elevated levels preoperatively. On the other hand, these findings are not consistent with those of Dawson et al. , who followed 108 patients with HF for 4 months and reported that 71% exhibited TnT increases post-operatively only. Fisher and colleagues  reported that 29% of their 238 HF patient cohort had troponin I >0.06 ng/ml within the first 72 h following hospital admission and observed an equal proportion of TnI elevation pre- and post-operatively.
Ausset  and Chong  have independently reported that elective as opposed to emergency hip surgery confers lower risk accompanied by a lower incidence of troponin elevation; however, increased TnI impacted adversely on prognosis in both cohorts. Dawson et al.  reported increased 4-month follow-up mortality rates among patients with elevated (≥0.03 ng/ml) versus normal TnT levels: 21 versus 8%, P < 0.05. As previously stated, we observed decreased survival only among the ad TnT levels >0.7 ng/ml subgroup; Death at 1 year was predicted by peak TnT ≥ 1.0 ng/ml. Fisher et al.  observed that increased TnI was associated with LOS ≥20 days but detected a trend only for in-hospital death.
The 32% elevated TnT incidence rate observed in the present HF cohort is consistent with the estimated 45% CAD prevalence rate in people older than 80 years of age or the 20–30% symptomatic presentation rate among the elderly [26, 27]. A recent epidemiologic survey of patients with HF on Medicare reported an approximately 10% incidence of myocardial infarction.4 Considering the 27% non-diagnostic ECG pattern rate in the present study, it is not possible to ascertain the temporal association between events; specifically, it is not clear whether elevated adTnT resulted from a primary coronary thrombotic event complicated by a subsequent fall leading to HF or whether a catecholamine surge released due to fracture pain-induced stress causing myocardial injury.
The close temporal proximity between hip trauma and TnT elevation added to (see Alcalay et al. ) advanced age, high frequencies of renal dysfunction, and lower than 1.0 ng/ml peak TnT elevation are consistent with a non-thrombotic (i.e. not acute coronary) etiology of myocardial injury in the majority of our HF cohort.
Osteoporosis and (coronary) atherosclerosis may be linked  . A recently published study  identified osteoprotegerin, a glycoprotein central to bone turnover, as an independent marker of atherosclerosis among post-menopausal osteoporotic women with cardiac risk factors but unknown CAD. Varma et al.  studied bone density in 198 women and reported significantly increased incident obstructive coronary artery disease among women inflicted with osteoporosis. However, these observations have not been uniformly confirmed; for example, Tekin et al.  measured bone mineral density in 227 consecutive women subjected to coronary angiography and reported lack of association with cardiovascular risk factors or presence of CAD.
The present study is based on prospectively collected observational data at a single institution. Additionally, none of the patients with elevated TnT underwent further cardiac evaluation during their index hospitalization, so it is not possible to know whether such an investigation could have been prognostically beneficial or the degree to which such measures would have been associated with in-hospital TnT. We could not collect death certificates of all deceased patients. Nevertheless, the observation that TnT elevation positively predicted late MI (an ischemic event) but not late admissions for CHF (presumably a non-ischemic manifestation) supports an ischemic etiology of late death.
Measurement of troponin levels identifies HF patients with clinically unrecognized myocardial injury and stratifies their future cardiac risk. The long-term follow-up in this study enabled us to determine that the main impact of troponin elevation is on late outcome and is not necessarily related to immediate surgical risk. The median magnitude of in-hospital TnT increase was small in most patients; therefore, it is possible that surgery conferred only a small incremental cardiac risk. To date, troponin measurement is not routinely recommended prior to non-elective non-cardiac surgery. Dawson and colleagues  proposed that troponin measurements should be used in conjunction with other screening tests to identify patients at greatest in-hospital risk. The present study adds that TnT elevation in patients with HF affected long-term prognosis. A randomized clinical trial comparing the clinical outcomes of patients with HF managed with vs. without clinicians’ awareness of troponin levels represents a step toward basing the diagnostic value of TnT on evidence. At present, it seems that in certain HF subsets presenting with markedly elevated troponin levels, subsequent cardiac evaluation may be useful.
Admission-Operation Interval (AOI): mean ± SD.
Cardiovascular disease = CAD, CVA or heart valve disease.
Peak TnT and AOI were closely associated; peak TnT levels were higher in patients operated >3 days hospital stay (peak TnT: 2.14 ± 7.16 versus 0.23 ± 0.78 ng/ml for AOI > 3 and AOI ≤ 3, respectively, P = 0.022).
Type 2 MI: Myocardial infarction secondary to ischemia due to either increased oxygen demand or decreased supply, e.g. coronary artery spasm, coronary embolism, anemia, arrhythmias, hypertension, or hypotension .
The authors express their sincere gratitude to Mr. Yosi Abuhav for his indispensable investment enabling manuscript’s creation.
Conflict of interest
- 8.James P, Ellis CJ, Whitlock RM, McNeil AR, Henley J, Anderson NE (2000) Relation between troponin T concentration and mortality in patients presenting with an acute stroke: observational study. BMJ 3;320 (7248):1502–1504Google Scholar
- 21.Oscarsson A, Fredrikson M, So¨rliden M, Anska¨r S, Eintrei C (2009) N- terminal fragment of pro-B-type natriuretic peptide is a predictor of cardiac events in high-risk patients undergoing acute hip fracture surgery. Br J Anaesth 103(2):206–212Google Scholar
- 23.Troponin T STAT, cardiac T, Roche ™ Diagnostics, Manheim, GermanyGoogle Scholar
- 25.Thygesen K, Alpert JS, White HD, Joint ESC/ACCF/AHA/WHF Task Force for the Redefinition of Myocardial Infarction (2007) Universal definition of myocardial infarction. J Am Coll Cardiol 50(22):2173–2195Google Scholar
- 26.E Centers for Disease control and Prevention (2003) MMWR series on public health and aging. MMWR 52:101–106Google Scholar
- 27.Schwartz JB, Zipes DP (2008) Cardiovascular disease in the elderly. From: chapter 75, Braunwald’s heart disease, a textbook of cardiovascular medicine, 8th edn. Saunders Elsevier, Philadelphia, PAGoogle Scholar