OUP user menu

Renal and cardiac function for long-term (10 year) risk stratification after myocardial infarction

Suetonia C. Palmer, Timothy G. Yandle, Christopher M. Frampton, Richard W. Troughton, M. Gary Nicholls, A. Mark Richards
DOI: http://dx.doi.org/10.1093/eurheartj/ehp132 1486-1494 First published online: 23 April 2009


Aims To determine whether combined renal and cardiac function after acute myocardial infarction (MI) predicts 10 year mortality and heart failure (HF).

Methods and results Estimated glomerular filtration rate (eGFR), plasma amino terminal pro-brain natriuretic peptide (NT-proBNP), and radionuclide ventriculography were obtained in 1063 patients with MI between 24–96 h of symptom onset. Mortality and HF were documented over follow-up of 9.3 years. Estimated GFR, NT-proBNP, and left ventricular ejection fraction (LVEF) each independently predicted 10 year mortality. Reduced eGFR (below 60 mL/min/1.73 m2) combined with increased NT-proBNP (above 1000 pg/mL) was associated with higher mortality rate compared with preserved eGFR together with lower NT-proBNP (60 vs. 14%, P < 0.001). Similar results for mortality were identified for eGFR combined with LVEF (dichotomized about 50%) (58 vs. 17%, P < 0.001). Corresponding analysis combining eGFR and NT-proBNP to predict HF yielded rates of 34 and 7% for high- and low-risk groups, respectively (P < 0.001). Similar risk stratification for HF was observed when combining eGFR with LVEF (35 vs. 7%, P < 0.001).

Conclusion Ten year rates of mortality and HF are 5–10 times higher when lower eGFR is present together with increased NT-proBNP or depressed LVEF.

  • Myocardial infarction
  • Kidney function
  • Mortality
  • Heart failure
  • Hospitalisation


Kidney function independently predicts mortality and morbidity after myocardial infarction (MI)1,2 and in heart failure (HF).3 Even small decrements of kidney function are associated with increased mortality.2,4 Left ventricular dimensions5 and contractile function6 are also prognostic following MI, as is plasma N-terminal pro-brain natriuretic peptide (NT-proBNP).7,8 No prospective data for MI have addressed the long-term prognostic utility of kidney function adjusting for left ventricular ejection fraction (LVEF) and NT-proBNP levels. We have assessed the utility of combining indices of cardiac and renal function to stratify risk of mortality and HF hospitalization to 10 years after MI.



This cohort study examined a community referral population of 1063 adults admitted to a coronary care unit with MI (the Christchurch Post Myocardial Infarction study).9,10 Subjects had ischaemic symptoms, ischaemic changes in two or more EKG leads, creatine kinase (CK) >400 U/L (reference range 25–175 U/L), and troponin T ≥0.01 µg/L (reference range <0.03 µg/L). Age over 80 years, immediate HF, and death within 24 h of symptom-onset were exclusion criteria. The study was approved by the local Ethics Committee in accordance with The Declaration of Helsinki; all patients gave informed consent.

Study design

Venous samples were drawn between 24 and 96 h of symptom onset. Plasma was stored at −80°C until extraction and radio-immunoassay for NT-proBNP (reference range 17–424 pg/mL).11 Radionuclide ventriculography was conducted within 24 h of blood sampling with a General Electric 400 AC gamma camera (Philadelphia, PA) interfaced with a 3000I computer system after in vivo technetium-99M red blood cell labelling. Glomerular filtration rate (eGFR) was estimated12 using the earliest recorded plasma creatinine level.

Study outcomes

The primary endpoint was all-cause mortality. Secondary outcomes were pre-discharge death or HF, or HF requiring re-admission to hospital. Deaths and HF events were documented during planned clinic assessments in the first year, annual questionnaires for 5 years and thereafter through the New Zealand National Health Information Service and Christchurch Hospital Patient Management System databases.

Statistical analysis

Variables are expressed as mean and standard deviation (SD); median with interquartile ranges (IQR) are provided when data were not normally distributed. Means were compared using independent t or Mann–Whitney tests as appropriate. χ2 tests were used to compare categorical variables. For univariable analyses, eGFR, NT-proBNP, and LVEF were dichotomized about clinical values (60 mL/min/1.73 m2, 1000 pg/mL, and 50%, respectively).

Kaplan–Meier event curves, stratified according to eGFR, NT-proBNP, and LVEF dichotomized about clinical cut-off values (alone and in paired combinations) to predict long-term mortality and HF, were compared using log-rank tests. Estimated GFR was also analysed according to the National Kidney Foundation Kidney Disease Outcomes Quality Initiative (KDOQI™) categories of chronic kidney disease (above 90, 60–90, 30–59, 15–29, and below 15 mL/min/1.73 m2).13 Logistic regression was used to determine unadjusted risks of in-patient and long-term mortality or HF.

Multivariable regression analyses were conducted to test the independent predictive power of eGFR, NT-proBNP, and LVEF for predefined outcomes. These models incorporated accepted clinical predictors of each outcome as covariates, thereby allowing us to test the additional prognostic power of the three putative predictors. Age, prior MI, and diabetes were incorporated in the model used to predict in-patient HF; age, gender, antecedent co-morbidity, medication, and in-patient percutaneous transluminal coronary angioplasty (PTCA) were incorporated into the multivariable model used to predict long-term mortality and HF. To evaluate the relative independent predictive power of baseline NT-proBNP compared with LVEF for long-term mortality and HF, each variable (expressed as a standardized z-score) was separately incorporated into the multivariable regression model. P < 0.05 was taken to indicate clinical significance for all analyses. Statistical analysis was conducted using SPSS version 13 (SPSS Inc., Chicago, IL).


At baseline, all 1063 patients had a GFR estimated (Table 1), 98% had NT-proBNP measured, and 92% underwent cardiac scanning. Mean eGFR was 82.6 ± 23.6 mL/min/1.73 m2. Patients with ‘lower’ eGFR (below 60 mL/min/1.73 m2) had higher NT-proBNP levels, increased cardiovascular comorbidity, and lower post-infarct LVEF (Table 1). Peak troponin I, infarct territory, and proportions of patients with ST-segment elevation were similar for ‘higher’ (above 60 mL/min/1.73 m2) and lower eGFR. Rates of PTCA were also similar between levels of eGFR. Univariable correlations between eGFR and the cardiac variables (NT-proBNP and LVEF) were –0.26 and 0.10, respectively (P < 0.01).

View this table:
Table 1

Characteristics according to baseline estimated glomerular filtration rate

VariableOverall cohort (n = 1063)eGFR ≥60 mL/min/1.73 m2 (n = 904)eGFR <60 mL/min/1.73 m2 (n = 158)P-valuea
Age, years62.4 ± 10.761.2 ± 10.469.2 ± 9.7<0.001
Male gender834 (79)731 (81)103 (65)<0.001
Medical conditions, %
 Hypertension401 (38)317 (35)84 (53)<0.001
 Prior MI194 (18)137 (15)57 (36)<0.001
 Heart failure151 (14)111 (12)40 (25)<0.001
 Diabetes mellitus135 (13)103 (11)32 (20)<0.01
 Prior stroke102 (10)71 (8)31 (20)<0.001
 Peripheral vascular disease73 (7)49 (5)24 (15)<0.001
 Lipid disorder374 (35)318 (35)56 (35)0.99
Current or former smoker672 (63)577 (64)95 (60)0.32
Blood pressure, mmHg
 Systolic116.5 ± 15.7116.0 ± 15.4119.1 ± 16.8<0.05
 Diastolic66.7 ± 9.766.8 ± 9.666.4 ± 10.30.67
Medication at discharge
 ARB or ACE-inhibitor513 (48)423 (47)90 (57)<0.05
 Beta-blocker878 (83)767 (85)111 (70)<0.001
 Aspirin1008 (95)861 (95)147 (95)0.12
 Lipid-lowering therapy472 (44)407 (45)65 (41)0.33
Laboratory values
 Plasma NT-proBNP, pg/mL958 (389–1610)888 (554–1501)1370 (918–2593)<0.001
 Haemoglobin, g/L144.1 ± 14.0145.0 ± 13.5138.8 ± 15.7<0.001
 Peak CK, U/L1566 (902–2670)1575 (912–2730)1522 (864–2390)0.39
 Peak troponin T, µg/L0.04 (0.01–0.21)0.04 (0.01–0.20)0.05 (0.01–0.39)0.19
 LDL cholesterol, mg/dL155 ± 41158 ± 41147 ± 440.001
LVEF, %47.5 ± 12.548.0 ± 12.144.1 ± 13.8<0.001
LVESV, mL78 (59–104)76 (59–101)88 (63–134)<0.001
LVEDV, mL151 (127–184)150 (126–182)161 (131–202)<0.05
ST-segment elevation869 (82)746 (83)123 (77)0.11
Thrombolytic therapy611 (58)531 (63)80 (54)0.05
In-patient PTCA211 (20)185 (21)26 (16)0.23
MI territory0.85
 Anterior387 (39)333 (39)54 (38)
 Inferior530 (53)454 (53)76 (53)
  • Data are expressed as mean (SD), median (interquartile range), or number (%). Bold values indicate a statistically significant result for group comparisons i.e. P-value < 0.05

  • ARB, angiotensin-receptor blocker; ACE, angiotensin-converting enzyme; LVEDV, left ventricular end-diastolic volume; MI, myocardial infarction; PTCA, percutaneous transluminal coronary angioplasty; LVEF, left ventricular ejection fraction.

  • aP-values refer to the comparison of values for eGFR above 60 mL/min/1.73 m2 vs. below 60 mL/min/1.73 m2.

Twenty-three (2.2%) patients died and 192 (18.1%) developed HF during the index hospitalization. Patients with higher NT-proBNP and lower eGFR and LVEF values had higher unadjusted mortality and HF rates during the immediate post-infarct period (P < 0.05) (Table 2). In multivariable regression analysis, LVEF and NT-proBNP, but not eGFR, independently predicted in-patient HF (P < 0.05).

View this table:
Table 2

Unadjusted risks of mortality and heart failure during index hospitalization

VariableCut-off valueAbove cut-off valueBelow cut-off valueOR (95% CI)aP-value
In-hospital mortality (n = 23)
 eGFR60 mL/min/1.73 m216 (1.8)7 (4.4)2.6 (1.0–6.3)<0.05
 NT-proBNP1000 pg/mL20 (4.1)1 (0.2)23.7 (3.2–177.2)<0.01
 LVEF50%1 (0.2)11 (2.0)8.9 (1.1–69.4)<0.05
In-hospital heart failure (n = 192)
 eGFR60 mL/min/1.73 m2143 (15.8)49 (30.8)2.4 (1.6–3.7)<0.001
 NT-proBNP1000 pg/mL155 (32.0)27 (4.9)9.1 (5.9–14.1)<0.001
 LVEF50%33 (7.6)147 (27.1)4.5 (3.0–6.7)<0.001
  • Events are number (%).

  • aOdds ratios and 95% confidence intervals are shown for below vs. above cut-off values for eGFR and LVEF and above vs. below the cut-off value for NT-proBNP.

  • eGFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction.

Median follow-up was 9.3 (IQR, 5.0–13.4) years with 68% of patients completing 10 years observation. Thirty percent of patients died during follow-up (including 44% patients with completed 10 year follow-up). Lower eGFR and LVEF and higher NT-proBNP were associated with higher 10 year mortality (P < 0.001) (Figure 1). Long-term survival was also significantly lower in patients with an estimated GFR below 90 mL/min/1.73 m2 (36%) compared with an eGFR above 90 mL/min/1.73 m2 (19%) (P < 0.001).

Figure 1

Kaplan–Meier survival curves for estimated eGFR, plasma NT-proBNP, and LVEF dichotomized about clinical cut-off values of 60 mL/min/1.73 m2, 1000 pg/mL, and 50%, respectively.

Combining eGFR with NT-proBNP further refined risk stratification. Over half (60%) of patients with a combined lower eGFR and higher NT-proBNP level died compared with 14% of patients with lower NT-proBNP and higher eGFR values (P < 0.001) (Figure 2). The corresponding odds ratio (OR) for 10 year mortality was 9.1 [95% confidence interval (CI), 5.2–16.0] (Table 3).

Figure 2

Kaplan–Meier survival curves for all-cause mortality after MI stratified according to combined eGFR and NT–proBNP (upper panel) and eGFR and LVEF (lower panel).

View this table:
Table 3

Unadjusted odds ratios (95% confidence intervals) for death or heart failure

1 year2 years5 years10 years
All-cause mortality
 Lower NT-proBNP, higher eGFR1.
 Lower NT-proBNP, lower eGFR1.3 (0.2–10.3)1.3 (0.2–10.2)1.4 (0.5–4.0)1.9 (0.9–4.0)
 Higher NT-proBNP, higher eGFR7.1 (3.3–15.4)6.0 (3.2–11.1)4.5 (2.9–7.0)2.9 (2.0–4.2)
 Higher NT-proBNP, lower eGFR15.6 (6.7–36.7)12.7 (6.2–25.9)10.6 (6.2–18.4)9.1 (5.2–16.0)
Heart failure hospitalization
 Lower NT-proBNP, higher eGFR1.
 Lower NT-proBNP, lower eGFR2.2 (0.5–10.9)4.4 (1.5–13.1)6.9 (3.1–15.7)3.6 (1.5–9.0)
 Higher NT-proBNP, higher eGFR4.6 (2.2–9.8)4.6 (2.3–8.9)4.6 (2.6–7.7)3.1 (1.9–5.2)
 Higher NT-proBNP, lower eGFR14.7 (6.5–33.4)16.8 (8.0–35.3)14.8 (7.7–28.4)11.7 (5.9–23.0)
All-cause mortality
 Higher LVEF higher eGFR1.
 Higher LVEF, lower eGFR4.5 (1.5–14.8)2.8 (1.0–8.1)1.4 (0.5–3.8)1.9 (0.9–3.9)
 Lower LVEF, higher eGFR2.8 (1.3–6.0)2.4 (1.3–4.4)2.4 (1.5–3.9)1.7 (1.2–2.4)
 Lower LVEF, lower eGFR8.3 (3.6–19.5)7.3 (3.6–14.8)7.7 (4.4–13.4)6.2 (3.5–10.7)
Heart failure hospitalization
 Higher LVEF higher eGFR1.
 Higher LVEF, lower eGFR4.7 (1.5–14.8)5.4 (1.9–15.7)6.4 (2.7–15.2)4.6 (2.1–10.3)
 Lower LVEF, higher eGFR2.9 (1.3–6.2)3.4 (1.7–7.0)3.5 (1.9–6.4)2.6 (1.6–4.3)
 Lower LVEF, lower eGFR10.3 (4.5–23.6)14.9 (6.8–32.9)15.1 (7.6–30.0)10.1 (5.1–20.0)
  • Higher and lower values represent values above and below cut-off values for each variable, respectively. Cut-off values for eGFR, NT-proBNP, and LVEF are 60 mL/min/1.73 m2, 1000 pg/mL, and 50%, respectively.

  • eGFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction.

Combining categories of eGFR and LVEF also refined risk stratification better than either marker alone. Higher LVEF and eGFR conferred lower mortality (17%) especially compared with combined reduced LVEF and lower eGFR (58%, P < 0.001) (Figure 2). In multivariable analysis, eGFR independently predicted mortality, as did NT-proBNP and LVEF (Table 4). Each 10 mL/min/1.73 m2 decrease in eGFR was associated with a significant increase in adjusted 10 year mortality risk of 7%. When incorporated into the multivariable regression analysis individually (as standardized z-scores), baseline NT-proBNP (adjusted HR 1.3, CI 1.2–1.5) and LVEF (adjusted HR, 1.4, CI 1.2–1.6) were associated with a similarly increased risk of long-term mortality.

View this table:
Table 4

Multivariable analysis for predictors of death or HF after MI

PredictorHazard ratio95% confidence intervalP-value
All-cause mortality (n = 319 events)
 Age, per year increase1.051.04–1.07<0.001
 Male gender0.960.72–1.290.78
 Prior MI1.611.20–2.15<0.001
 Prior stroke1.400.99–1.970.06
 Heart failure1.080.79–1.460.64
 Diabetes mellitus1.350.98–1.870.07
 Lipid-lowering therapy0.830.64–1.080.17
 In-patient PTCA0.610.42–0.900.01
 eGFR, per 10 mL/min/ 1.73 m2 decrease1.071.01–1.14<0.05
 NT-proBNP, per 100 pg/mL increase1.021.01–1.03<0.001
 LVESV, per 10 mL increasea1.041.02–1.06<0.001
 LVEF, per 10% decreasea1.251.11–1.41<0.001
Hospitalization for HF (n = 175 events)
 Age, per year increase1.051.03–1.07<0.001
 Male gender0.930.63–1.360.70
 Prior MI1.511.04–2.19<0.05
 Prior stroke1.430.91–2.240.12
 Heart failure1.130.75–1.690.57
 Diabetes mellitus1.781.18–2.68<0.01
 Lipid-lowering therapy1.150.82–1.620.42
 In-patient PTCA0.670.42–1.060.08
 eGFR, per 10 mL/min/ 1.73 m2 decrease1.081.01–1.08<0.05
 NT-proBNP, per 100 pg/ mL increase1.021.01–1.04<0.001
 LVESV, per 10 mL increasea1.051.02–1.080.001
 LVEF, per 10% decreasea1.251.07–1.46<0.01
  • Bold P-values indicate statistical significance.

  • aLVESV was rotated through the model in place of LVEF.

  • ACE, angiotensin-converting enzyme; MI, myocardial infarction; PTCA, percutaneous transluminal coronary angioplasty; eGFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction.

Overall, 175 patients (16%) required hospitalization for HF. Higher NT-proBNP and lower eGFR and LVEF values were associated with increased HF hospitalization (P < 0.001). Combining NT-proBNP and eGFR improved risk stratification for HF over 10 years (Figure 3). Patients with combined higher NT-proBNP and lower eGFR incurred an HF rate of 34% compared with 7% of patients with low NT-proBNP and high eGFR values (OR 6.2, CI 3.5–10.7, P < 0.001). Combining LVEF with eGFR also permitted additional risk stratification for HF (Figure 3); a 5–10% difference in risk of HF was found within each category of LVEF when comparing the risk in sub-groups with higher and lower eGFR. This risk stratification for HF re-admission persisted through 10 years of follow-up (Table 3).

Figure 3

Kaplan–Meier survival curves for HF hospitalization after MI stratified according to combined eGFR and NT-proBNP (upper panel) and eGFR and LVEF (lower panel).

In multivariable analysis, eGFR, NT-proBNP, and LVEF remained independent predictors for new-onset HF (Table 4). When incorporated into the multivariable regression analysis individually (as standardized z-scores), baseline NT-proBNP (adjusted HR 1.4, CI 1.3–1.6) and LVEF (adjusted HR, 1.4, CI 1.2–1.6) were associated with a similarly increased risk of long-term HF.


In over 1000 patients with MI followed for 10 years, combining kidney function (eGFR) with cardiac performance (NT-proBNP or LVEF) improved risk estimation for both mortality and HF, compared with information provided by any one of these variables in isolation. Presence of both reduced eGFR and elevated NT-proBNP identified patients with a high 10 year risk of death or HF after MI. Similar refinement of risk stratification was provided through combining eGFR with LVEF, using the reproducible technique of radionuclide ventriculography.14

Whether the augmented risk of morbid and mortal cardiovascular events after MI associated with reduced renal function simply reflects the degree of cardiac injury or whether there is a specific pathogenetic role for renal dysfunction underlying later adverse outcomes remains unclear. Even minor kidney dysfunction is associated with altered homeostasis, including increased inflammatory cytokines and endothelial dysfunction, which predicts cardiovascular mortality.15,16 Experimental renal ablation leads to increased myocardial fibrosis17 and collagen content18 resulting in diastolic dysfunction and subsequent hypertrophic progression with systolic ventricular impairment. These findings suggest a potential causal role for kidney dysfunction in poorer outcomes after MI. The strong relationship between kidney dysfunction and elevated natriuretic peptide levels also might reflect a higher prevalence of pre-existing cardiac structural abnormalities in patients with kidney dysfunction.19,20

Lower kidney function (eGFR < 60 mL/min/1.73 m2) in our analysis was not associated with markers of more severe MI at baseline such as peak CK, infarct territory, or ST segment elevation but was associated with a greater burden of pre-existing cardiovascular disease and impaired ventricular function at baseline (Table 1). Kidney function may be a discriminating marker of sub-clinical cardiac dysfunction of prognostic importance in the early post-infarction period. Alternatively, kidney function may be a confounding variable acting as an index of severity for poorer cardiac function via a shared aetiology, such as atherosclerotic disease burden. Left-ventricular mass21 and diastolic dysfunction,22 independent predictors of mortality, are likely to be influenced by factors which concurrently influence renal function (including blood pressure, age, and diabetes). Their evaluation by echocardiography may provide further insight into the relationships between kidney dysfunction and long-term mortality after MI.

Patients with impaired kidney function may be less likely to receive medication proved to improve outcomes after MI. We observed that lower eGFR was associated with reduced prescription of beta-blockers and inhibitors of the renin–angiotensin system at discharge, therapies both known to reduce mortality23 and HF.24 The reasons for lower prescription rates cannot be determined by our study design although reduced tolerance of beta-blockers may reflect increased comorbidity (including an increased prevalence of peripheral vascular disease) (Table 1).25

Our data extend previous reports of the influence of kidney function on outcomes after acute coronary syndromes2,4,2629 and confirm that small decrements in kidney function within the normal range portend poorer clinical outcomes. The findings of our study corroborate a recent report demonstrating an association of kidney dysfunction with increased adjusted 10 year mortality after MI in those with an eGFR below 74 mL/min/1.73 m2.4 Combining kidney function with concurrent NT-proBNP or LVEF allowed us to further stratify estimates of 10 year mortality risk using cut-off values for kidney and cardiac function that are clinically intuitive. We identified a population with a 10 year risk of death or HF 5–10 times that for patients with preserved cardiorenal function.

Combining NT-proBNP or LVEF with kidney function has demonstrated refinement of risk stratification after MI over short-term follow-up (≤1 year) in the Global Utilisation of Strategies To Open occluded arteries IV (GUSTO-IV) trial.26 NT-proBNP combined with eGFR provided the strongest prediction of death at 1 year, with unadjusted mortality for those within both lowest quartile of creatinine clearance (≤51 mL/min) and highest quartile NT-proBNP (>1869 pg/mL) of 25.7%, compared with 0.3% for those with creatinine clearance >84 mL/min/1.73 m2 and NT-proBNP ≤237 pg/mL. Cardiac imaging was not available in the GUSTO-IV analysis. We confirm a similar graded stratification for the combination of eGFR and NT-proBNP with escalating mortality rates in patients with poorest cardiorenal function and demonstrate this risk stratification persists to 10 years. In another study of 1051 patients with stable coronary heart disease,30 no graded increase in combined fatal and non-fatal cardiovascular events was observed over 48.7 months of follow-up with decreasing eGFR when adjusted for NT-proBNP levels. However, mortality (2.9%) was far lower than in the current analysis, suggesting insufficient power to discern the relationships observed in the current report.

Combined measures of cardiac structure and kidney function have also been shown to refine risk stratification at 12 months after MI. In the VALsartan In Acute myocardial iNfarcTion study (VALIANT), enrolling patients with impaired ejection fraction after MI, an eGFR below 60 mL/min/1.73 m2 adjusted for LVEF and LVESV31 was associated with increased all-cause mortality and hospital stay. Similar findings were observed in 3074 patients with chest pain without ST elevation. Creatinine clearance below 30 mL/min together with LVEF below 35% was associated with a "3–4-fold increase in all-cause mortality at 1 year compared with normal cardiorenal function.32 We extend this knowledge, showing the increased risk of mortality and HF persists for at least 10 years after MI for patients with impairment of both renal and cardiac function.

Study limitations

The present cohort included unselected patients hospitalized to a single coronary care unit with acute MI; its strength lies in the duration of follow-up of a well-characterized cohort for cardiac and renal function. Our study however has limitations. The study excluded people over 80 years and analyses may not be representative for this older population who represent ∼20% of admissions for acute coronary syndrome. Further, the cohort was derived from patients admitted to a coronary care unit with a higher proportion of patients with ST-segment elevation at admission than is usually seen in a hospitalized MI population. While the current cohort was unselected for eGFR, very few (1.0%) patients had an eGFR <30 mL/min/1.73 m2. The study included predominantly Caucasian subjects and no conclusions can be drawn on interactions between ethnicity and cardiorenal function. Finally, as cause of death was not adjudicated, the effect of cardiorenal function on risk of cardiovascular mortality specifically cannot be determined in this study.


Combining eGFR and cardiac function following MI can refine risk stratification for death and HF and identify patients for whom 10 year risks vary over a 5–10-fold range. Our findings suggest future studies of interventions to improve outcomes after MI should stratify patients according to both cardiac and renal function.


This work was supported by the National Heart Foundation of New Zealand; and the Health Research Council of New Zealand. S.C.P. is supported by a Health Research Council of New Zealand Fellowship. A.M.R. holds the National Heart Foundation (New Zealand) Chair of Cardiovascular Studies.

Conflict of interest: none declared.


View Abstract