OUP user menu

Survival among hospital in-patients with troponin T elevation below levels defining myocardial infarction

G. Cook, D. Taylor, M. France, G. Burrows, E. Manning, G. Lyratzopoulos, P. McElduff, P. Lewis, M. Martin, R.F. Heller
DOI: http://dx.doi.org/10.1093/qjmed/hci045 275-282 First published online: 10 March 2005


Background: Cardiac troponin T (cTnT) has an accepted place in the management of patients presenting with suspected acute coronary syndrome (ACS). Uncertainty remains about the significance and interpretation of elevated cTnT below the cut-off levels defining myocardial infarction (0.1 μg/l).

Aim: To compare the mortality risks for elevation of cTnT in the ranges 0.01–0.029 μg/l, 0.03–0.099 μg/l and <0.01 μg/l.

Design: Retrospective record study in three hospitals.

Methods: All cTnT measurements with values in the range >0.01–0.099 μg/l analysed during January 2002 were extracted from clinical biochemistry laboratory databases. Following agreed exclusion criteria, 179 patients with cTnT in the range 0.01–0.099 μg/l and 60 patients <0.01 μg/l were selected at random from across the three sites. Six-month follow-up was completed by review of case notes and contact with the patients’ GP.

Results: There was a graded increase in mortality with increasing cTnT, although only achieving statistical significance for patients in the 0.03–0.099 μg/l range. The graded increase in relative risk with cTnT was weaker after adjustment for potential confounding factors

Discussion: We found a trend for worse survival with increasing cTnT within the range 0.01–0.099 μg/l in unselected patient populations presenting with possible acute coronary syndrome. This suggests that the combined effects of assay imprecision and co-morbidity should be taken into account when interpreting borderline elevation of cTnT. The use of a cut-off based on current standards of assay precision should be used to define the sensitivity of cTnT as biochemical evidence of ischaemic cardiac damage and as an indicator of mortality risk. This level is likely to be between 0.03 and 0.1 μg/l.


Troponin T (cTnT) is accepted as a sensitive and specific marker of myocyte necrosis. The Joint European Society of Cardiology and the American College of Cardiology (JESC/ACC)1 guidelines recommended that a cTnT level above the 99th centile of the normal range (0.01 μg/l) be evidence of myocyte necrosis. They stipulated a desirable analytical coefficient of variation (CV) at this level of 10%. However, the CV of the latest available version of the cTnT assay can only achieve this precision at 0.03 μg/l.2 This conventional limit to assay performance is called the functional sensitivity. The accepted level of cTnT that equates with classic diagnostic standards for myocardial infarction is currently under debate, but the manufacturer continues to recommend 0.1 μg/l

The clinical specificity of troponin measurements is improved by sequential measurement, as recommended in the guidelines. However, this is not common practice in the UK, largely due to the need to contain costs. Clinical studies on the use of cTnT measurement in the management of acute coronary syndromes are confounded by lower analytical sensitivity of the assay used. In one study using the third-generation assay, 0.01 μg/l had higher diagnostic accuracy than higher cut-offs.3 However, this and older studies on the value of elevated cTnT between 0.01 and 0.1 μg/l were sub-studies within interventional trials,4–6 which may exhibit selection bias by excluding elderly patients with co-morbidity. Meta-analysis has shown different mortality between clinical trials and cohort studies.7

cTnT is used as a risk stratification tool as part of a global risk assessment in order to choose the most important care pathway for patients. In accordance with published standards,8 we evaluated the use of cTnT as an indicator of mortality risk in an unselected population of hospital patients with a differential diagnosis including acute coronary syndrome (ACS). We undertook a retrospective cohort study in three acute hospitals to determine whether mortality risk was graded in proportion to cTnT elevation within the range 0.01–0.099 μg/l or whether a threshold could usefully be set at the functional sensitivity of 0.03 μg/l.


Approximately 1500 patients would have had to be recruited to have 80% power with 5% significance level to detect the optimal ‘discriminator’ value defining ‘high’ and ‘low’ risk patients among patients with intermediate troponin values. This calculation assumed that approximately half the patients would be in the ‘high’ and half the patients in the ‘low’ risk group; and that the hazard ratio between ‘high’ and ‘low’ risk patients would be between 2 and 5. For pragmatic and practical reasons, we decided to conduct a pilot study of about 200 patients first, to best inform the methodology of a future definitive larger study.

The study sample was drawn from patients who were either emergency medical admissions or elective medical or surgical in-patients, and for whom a cTnT measurement was undertaken for any reason, including symptoms of chest pain. Three Trusts in the North West of England were selected: one District General Hospital and two Teaching Hospitals, each assaying cTnT at least 12 h after the onset of chest pain. Serial cTnT measurements were not usual practice. We chose to sample across several hospital sites in order to provide a cross-section over a short period, since clinical protocols and assay systems change over time. Although this study was not powered to discriminate between hospitals, we felt it important to take a sample from different types of hospital, as clinical protocols might differ in relation to the on site resources. Each hospital would contribute an equal number of cases. We aimed for a target sample of 60 ‘positive’ cases from each trust and a total of 60 negative across the trusts to give a total sample of 240. There was uncertainty about the size of the pool of patients with ‘positive’ results from which a sample would be drawn. In the event of a pool in excess of 60, it was agreed to sample randomly to minimize the likelihood of any systematic bias. In respect of the negative cases, we were aware that there would be several hundred potential cases and elected to sample randomly to minimize selection bias. The cTnT results and dates along with patient identifiers were extracted from clinical biochemistry laboratory databases on the three sites for the month of January 2002 and stored in an Excel spreadsheet.

Data for all cTnT results <0.1 μg/l were extracted for the study. Patients with duplicate tests in the same admission were identified, and the result of the highest value was taken. Duplicate patients were removed, such that only patients with a first admission within that hospital during the period and whose highest value fell within the range of interest were included. A random number was assigned to the lists of patients using Excel and then sorted in order. Remaining cases in the random list were used where an earlier case was excluded or the case notes could not be found. Patients were excluded if they had clearly recorded statements in the medical record of chronic renal disease, fulfilled criteria for acute ST segment elevation myocardial infarction, had been thrombolysed, or in the previous 4 weeks had had a myocardial infarction, or percutaneous cardiac procedure, or heart surgery. Patients were also excluded if they fell outside the catchment area of the hospital; thus cases referred in for tertiary care were excluded. A total of 179 patients with cTnT in the range 0.01–0.099 μg/l entered the study. In addition, a random sample of 60 patients (20 for each site) with a cTnT value <0.01 μg/l were selected. Vital status of all study patients was ascertained up to 180 days after the cTnT measurement, by review of case notes and contact with patients GP.

Clinical data

Clinical data, abstracted from the clinical record by the researcher, included predominant symptom on admission, past medical history and final diagnosis as recorded on the discharge summary. The latter data item was used to group patients into four diagnostic categories: acute coronary syndrome; other cardiac; non-cardiac; and not stated (Table 3). Serum creatinine level on admission was also recorded. All of these data items were logged if there was a positive statement in the notes

Statistical methods

Patients were allocated to three groups determined by cTnT values <0.01 (negative), 0.01–0.029 (intermediate) and 0.030–0.099 (high) μg/l. Descriptive statistics are expressed as means or proportions with their associated 95%CIs. Cox proportional hazards models were used to estimate the relative risks of mortality at 28 days and 180 days for each of the groups defined by cTnT, with the <0.01 μg/l group as the reference. Relative risks are presented as crude estimates and also after adjusting for the effects of gender, age <75 or >75 years, hospital site, predominant symptom on admission, main diagnostic category, cardiovascular related drugs on admission and serum creatinine level. The crude relative risk presents the risk for a sub-group within a group relative to a subgroup chosen as baseline, which is denoted with a relative risk of 1, without taking into account other factors potentially affecting the comparison. The adjusted relative risk presents the same comparisons as the crude relative risk but takes into account other factors, which may affect the comparison.


Laboratory data

There were 2139 cTnT tests conducted in January 2002, and 436 (20.4%) of these where positive (Table 1); 486 (22.7%) were in the borderline range of interest.

View this table:
Table 1

Numbers of cTnT requests by result range and hospital

HospitalTotal tests in January 2002Negative (<0.01 μg/l)Borderline (0.01–0.099 μg/l)Positive (>0.1 μg/l) (not eligible)
Stockport660369 (56.0%)162 (24.5%)129 (19.5%)
MRI508252 (49.6%)119 (23.4%)137 (26.9%)
RLUH971596 (61.3%)205 (21.1%)170 (17.5%)
Total21391217 (56.9%)486 (22.7%)436 (20.4%)
  • MRI, Manchester Royal Infirmary; RLUH, Royal Liverpool University Hospital.

A total of 60 patients were excluded from the review (Table 2). Over 75% of the patients excluded were because the case notes of the relevant episode were unobtainable or missing. The remainder were excluded because the patients were from outside the district.

View this table:
Table 2

cTnT requests, patients and samples in the range 0.01–0. 099 μg/l by hospital

HospitalTestsPatients on sample listPatients excluded from sample listPatients remaining on list to be sampledFinal sample total
  • *Patients with more than one test counted only once. **Final sample list from which random sample drawn. MRI, Manchester Royal Infirmary; RLUH, Royal Liverpool University Hospital.

View this table:
Table 3

Demographic and clinical details of patients by cTnT range

VariableCTnT range (μg/l)p*
Age (mean in years%)60.875.275.5<0.001
Female32 (53.3%)38 (52.8%)54 (50.5%)0.923
Predominant symptom on admission
Chest pain35 (58.3%)24 (35.8%)33 (29.5%)0.003
Shortness of breath6 (10.0%)17 (25.4%)37 (33.0%)
Collapse/fall11 (18.3%)10 (14.9%)15 (13.4%)
Other8 (13.3%)16 (23.9%)27 (24.1%)
Past medical history
MI/angina24 (40.0%)23 (34.3%)48 (42.9%)0.503
Stroke3 (5.0%)4 (6.0%)10 (8.3%)
Diabetes6 (10.0%)6 (9.0%)11 (9.8%)
Other relevant12 (20.0%)10 (14.9%)22 (19.6%)
None relevant15 (25.0%)24 (35.8%)21 (18.8%)
Main diagnostic category
Acute coronary syndrome9 (15.0%)8 (11.9%)22 (19.6%)<0.001
Other cardiac8 (13.3%)18 (26.9%)42 (37.5%)
Non-cardiac/not stated43 (71.7%)41 (51.2%)48 (42.9%)
Cardiovascular-related drugs on admission
Number of people41 (68.3%)47 (70.1%)99 (88.4%)0.001
Mean number of drugs prescribed per patient2.
Serum creatinine levels on admission
Mean (μmol/l)100127145<0.001
% >130 μmol/l5 (8.8%)20 (29.9%)52 (46.9%)<0.001
  • *Kruskal-Wallis test for continuous data. χ2 test for categorical data.

The MRI sample fell short by one, because of the relatively smaller sampling frame and the greater number of excluded records. In summary, the three-hospital random sample of 179 (45%) for the cTnT 0.01–0.099 μg/l group was drawn from a potential pool of 399 patients. For the negative patient group, a total of nine patients were excluded for similar reasons. The final sample of 60 patients in the cTnT <0.01 μg/l group represents a 5% random sample from the population of 1217 negative tests in this group. Three (5%) of the negative group and 22 (12.3%) of the intermediate group were surgical patients. The remaining patients were medical cases. The overall mean age of the sampled patients was 71.7 ± 15.4 years. Fifty-two percent of patients were female. The chief presenting complaint was chest pain, which occurred in 92 (38.5%) patients, followed by shortness of breath in 60 (25.1%). Thirty-six patients who collapsed or fell made up a further 15.1% and the rest had a multitude of other reasons for admission. Ninety-five (39.7%) patients had a documented past medical history of myocardial infarction and or angina, although 187 (78.2%) patients were being prescribed 1–9 (mean 2.4) cardiovascular-related drugs on admission (in 12 patients, drug history was unavailable). Thirty-nine (16.3%) had an acute coronary syndrome related primary diagnosis at discharge. A further 68 (28.5%) had other cardiac related diagnoses such as heart failure. The majority (132; 55.2%) had a non-cardiac-related/not stated primary diagnosis recorded in the medical record in relation to the hospitalization under scrutiny. Mean serum creatinine was 129 ± 89 μmol/l. Sixty-four (26.8%) patients were discharged on anti-anginal treatment. Twelve (5%) patients had had angiography following admission within the follow-up period. Fewer patients were diagnosed as an acute coronary syndrome in the lower range, but overall there were more cardiac related cases in this group.

The negative cTnT group were generally younger than the positive groups. (Table 3). The age and sex distribution was similar to that of the two subgroups of the intermediate range. Chest pain as the presenting complaint predominated in the negative group, whereas both chest pain and shortness of breath were the main presenting problems in the intermediate range. A majority of all ranges had a relevant past medical history, with a stated history of ischaemic heart disease in 30–40%. A diagnosis of acute coronary syndrome was made in a minority of cases for each range, but more so with increasing troponin level. Less than 30% of the negative cases had a cardiac-related diagnosis, compared with 60–80% of the intermediate ranges. In contrast, 68% of negatively tested patients received on average two cardiovascular-related drugs. This figure increased to almost 90% of patients in the higher of the two intermediate ranges receiving three such drugs. Mean serum creatinine showed a stepwise increase over the three ranges.

With the exception of gender and past medical history, the remaining demographic and clinical features showed a highly significant difference in the proportions of the different categories between the troponin ranges. In the analysis of survival at 28 days and 180 days (Tables 4 and 5, respectively), adjustments (see statistical methods for explanation) were made for all the categories, with the exception of past medical history.

View this table:
Table 4

Number of participants dead or alive at 28 days, by gender, age group, hospital site, tertile of creatinine level and cTnT range, with crude and adjusted relative risk (RR) of death

Status at 28 daysRR (95%CI)
Alive n (%)Dead n (%)CrudeAdjusted*
M100 (87)15 (13)1.001.00
F110 (89)14 (11)0.79 (0.38, 1.67)0.96 (0.40, 2.27)
Age group
≤75 years108 (93)8 (7)1.001.00
>75 years102 (83)21 (17)2.47 (1.09, 5.98)1.48 (0.57, 3.85)
Stockport71 (89)9 (11)1.001.00
MRI74 (94)5 (6)0.62 (0.20, 1.90)0.94 (0.28, 3.11)
RLUH65 (81)15 (19)1.97 (0.83, 4.64)2.33 (0.85, 6.40)
Predominant symptom on admission
Chest pain88 (96)4 (4)1.001.00
Shortness of breath46 (77)14 (23)5.82 (1.91, 17.7)5.27 (1.42, 19.5)
Collapse/fall29 (81)7 (19)4.81 (1.41, 16.5)2.95 (0.65, 13.4)
Other47 (92)4 (8)1.38 (0.31, 6.18)1.16 (0.22, 6.24)
Main diagnostic category
ACS35 (90)4 (10)1.001.00
Other cardiac60 (88)8 (12)1.15 (0.35, 3.83)0.48 (0.11, 2.01)
Non-cardiac/not stated115 (87)17 (13)1.20 (0.40, 3.59)0.79 (0.20, 3.06)
Cardiovascular-related drugs on admission
No34 (85)6 (15)1.001.00
Yes167 (89)20 (11)0.68 (0.27, 1.70)0.39 (0.13, 1.12)
Creatinine (μmol/l)
≤9370 (88)10 (12)1.001.00
93–13071 (91)7 (9)0.71 (0.27, 1.87)0.67 (0.22, 2.10)
>13065 (84)12 (16)1.15 (0.49, 2.71)0.84 (0.30, 2.35)
Troponin (μg/l)
<0.0157 (95)3 (5)1.001.00
0.01–0.0362 (91)6 (9)1.84 (0.46, 7.37)2.52 (0.48, 13.3)
0.03–0.09991 (82)20 (18)3.81 (1.13, 12.89)4.92 (0.97, 24.8)
  • *Relative risks adjusted for all other variables in the table. MRI, Manchester Royal Infirmary; RLUH, Royal Liverpool University Hospital.

View this table:
Table 5

Number of participants dead or alive at 180 days, by gender, age group, hospital site, tertile of creatinine level and troponin range, with crude and adjusted relative risk (RR) of death

Status at 180 daysRR (95%CI)
Alive n (%)Dead n (%)CrudeAdjusted*
M84 (73)31 (27)1.001.00
F100 (81)24 (19)0.68 (0.40, 1.17)0.83 (0.45, 1.53)
Age group
≤7598 (84)18 (16)1.001.00
>7586 (70)37 (30)2.13 (1.21, 3.76)1.40 (0.73, 2.68)
Stockport63 (79)17 (21)1.001.00
MRI63 (80)16 (20)1.01 (0.50, 2.01)1.20 (0.57, 2.54)
RLUH58 (73)22 (28)1.44 (0.75, 2.74)1.45 (0.69, 3.02)
Predominant symptom on admission
Chest pain84 (91)8 (9)1.001.00
Shortness of breath39 (65)21 (35)4.77 (2.11, 10.8)4.50 (1.66, 12.2)
Collapse/fall25 (69)11 (31)4.19 (1.68, 10.4)3.05 (1.00, 9.30)
Other36 (71)15 (29)3.5 (1.45, 8.24)3.41 (1.17, 9.98)
Main diagnostic category
ACS32 (82)7 (18)1.001.00
Other cardiac51 (75)17 (25)1.35 (0.56, 3.26)0.54 (0.19, 1.58)
Non-cardiac/not stated101 (77)31 (23)1.29 (0.57, 2.93)0.65 (0.24, 1.81)
Cardiovascular-related drugs on admission
No31 (78)9 (23)1.001.00
Yes148 (79)39 (21)0.89 (0.43, 1.75)0.59 (0.27, 1.33)
Creatinine (μmol/l)
≤9367 (84)13 (16)1.001.00
93–13064 (82)14 (18)1.09 (0.51, 2.32)0.90 (0.37, 2.16)
>13049 (64)28 (36)2.35 (1.21, 4.55)1.55 (0.69, 3.48)
Troponin (μg/l)
<0.0152 (87)8 (13)1.001.00
0.01–0.0353 (79)14 (21)1.67 (0.70, 3.99)0.97 (0.35, 2.68)
0.03–0.09979 (71)33 (29)2.61 (1.20, 5.66)1.55 (0.59, 4.08)
  • *Relative risks are adjusted for all other variables in the table. MRI, Manchester Royal Infirmary; RLUH, Royal Liverpool University Hospital.

Table 4 summarizes the crude and adjusted survivals at 28 days for the different demographic and clinical categories. The overall death rate was 12% at 28 days. There was no difference between the sexes, but the crude rate was significantly higher in those aged >75 years. This significance disappeared when adjusted. There was a significant difference in crude 28-day death rates between the hospitals, which was also lost on adjustment. However, in respect of predominant symptom on admission, there was a high crude death rate for those presenting with shortness of breath, which remained significant when adjusted. The crude death rate was also significantly higher with elevated creatinine, but the relationship disappeared after adjustment. The crude relative risk associated with cTnT in the range 0.03 to 0.099 μg/l was significant and only marginally failed to achieve significance after adjustment {RR 4.92 (0.97–24.8)}. The confidence for this estimate is very wide, due to the size of the study

In Table 5, the same analysis is presented for survival at 180 days. Similar observations are made to those at 28 days. Once again, the crude death rates were significantly higher with greater age, with the three non-ACS symptom presentation groups, with elevated creatinine and in the highest troponin range under scrutiny. The significance disappeared on adjustment, with all but predominant symptoms of shortness of breath and other non-cardiac, although the confidence intervals are again wide.


Interpretation of borderline cTnT elevation between 0.01 and 0.099 μg/l is problematic. Manufacturers continue to recommend 0.1 μg/l as a diagnostic threshold. In this study of routine patient care in hospital, we wished to compare the utility of borderline cTnT elevation within the ranges 0.01–0.29 μg/l (intermediate) and 0.03–0.099 μg/l (high) with negative values <0.01 μg/l. In this study, 22.7% of results fell in the borderline range compared with 20.4% >0.1 μg/l and 56.9% negative. Only one published prospective study based on a sample of patients from the GUSTO IV trial supports the use of any elevation of cTnT >0.01 μg/l as a graded indicator of mortality risk. Two recent clinical studies have used cut-offs of 0.03 μg/l9 and 0.04 μg/l,10 based on appropriate considerations of assay sensitivity. None compared outcome between those <0.01 μg/l and those between 0.01 μg/l and the cut-off used. One prospective study in an Emergency Medicine department showed that patients with marginally elevated troponin within the range 0.01–0.09 μg/l were at higher risk of adverse events than those with undetectable levels (<0.01 μg/l).11 A British Cardiac Society working group12 has recently reiterated the NACB13 recommendation for a two-stage approach for interpretation of troponin results: one to fit in with the classical definition of myocardial infarction and a lower level signifying myocardial injury. This emphasizes the idea that myocardial necrosis does not necessarily equate with infarction and indeed may not be ischaemic in origin. Both regard any detectable cTnT (>0.01 μg/l) as an indication of myocardial necrosis.

Our results show a graded crude relative risk of mortality at 28 and 180 days with increasing cTnT, but this only achieved significance in the higher group (0.03–0.099 μg/l). With increasing cTnT, patients were older, had an increased likelihood of a cardiovascular presentation, had a greater number of prescribed cardiovascular drugs, were more likely to have had assigned a cardiovascular diagnosis, and were likely to have a higher creatinine level. After adjustment for these factors, only shortness of breath (RR 5.27, 95%CI 1.42–19.5, compared with patients with chest pain only) and cTnT >0.03–0.99 μg/l (RR 4.93 95%CI 0.97–24.8, compared with negative cTnT) were strong predictors of relative mortality, the latter just failing to achieve statistical significance due to increased uncertainty in the estimate of relative risk. At 180 days, shortness of breath and collapse, suggesting haemodynamic instability, increased the risk of death compared with those with chest pain only, even after adjustment for cTnT.

The increase in crude 28-day mortality risk with increasing cTnT in the range up to 0.099 μg/l is consistent with previous studies. Our concern was that raised cTnT, due to its release from heart muscle by a non-ischaemic mechanism consequent on co-morbidity, would obscure this relationship. Troponin release occurs in a wide range of clinical situations.14 Such release of low levels of cTnT by apoptosis by non-ischaemic mechanisms is a confounding factor in planning management of ACS. There were significant differences in admission creatinine between our three groups, possibly reflecting non-ischaemic cardiac morbidity. The independent value of cTnT in predicting short-term mortality is likely to relate to its association with ischaemic necrosis, i.e. infarction as an acute event. Our results suggest that the level of cTnT that will be clinically useful to distinguish such an ischaemic event from that due to non-ischaemic release, as is likely to occur for instance in heart failure, lies somewhere between 0.03 and 0.099 μg/l. These results should be interpreted with caution, because of the small size of this pilot study and the consequent wide uncertainty of the estimates of relative risk. Nevertheless, our results suggest that there is little useful distinction to be made between patients with cTnT <0.01 μg/l and those within the range 0.01–0.03 μg/l, a group defined by measurements made below the accepted limit of assay performance. A larger study is required to ascertain the exact cut-off.


View Abstract