`doi:10.1093/eurheartj/ehs337
`
`CLINICAL RESEARCH
`Chronic heart failure
`
`Predicting survival in heart failure: a risk score
`based on 39 372 patients from 30 studies
`
`Stuart J. Pocock1*, Cono A. Ariti1, John J.V. McMurray2, Aldo Maggioni3, Lars Køber4,
`Iain B. Squire5, Karl Swedberg6, Joanna Dobson1, Katrina K. Poppe7,
`Gillian A. Whalley7, and Rob N. Doughty7, on behalf of the Meta-Analysis Global Group
`in Chronic Heart Failure (MAGGIC)
`
`1Department of Medical Statistics, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK; 2Institute of Cardiovascular and Medical Sciences,
`University of Glasgow, Glasgow, UK; 3ANMCO Research Centre, Florence, Italy; 4Rigshospitalet—Copenhagen University Hospital, Copenhagen, Denmark; 5Department of
`Cardiovascular Sciences, The University of Leicester, Leicester, UK; 6Sahlgrenska University, Hospital/O¨ stra, Go¨ teborg, Sweden; and 7Department of Medicine, University of
`Auckland, Auckland, New Zealand
`
`Received 22 May 2012; revised 3 August 2012; accepted 13 September 2012; online publish-ahead-of-print 24 October 2012
`
`See page 1391 for the editorial comment on this article (doi:10.1093/eurheartj/ehs363)
`
`Aims
`
`Using a large international database from multiple cohort studies, the aim is to create a generalizable easily used risk
`score for mortality in patients with heart failure (HF).
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`Methods
`The MAGGIC meta-analysis includes individual data on 39 372 patients with HF, both reduced and preserved left-
`and results
`ventricular ejection fraction (EF), from 30 cohort studies, six of which were clinical trials. 40.2% of patients died
`during a median follow-up of 2.5 years. Using multivariable piecewise Poisson regression methods with stepwise vari-
`able selection, a final model included 13 highly significant independent predictors of mortality in the following order
`of predictive strength: age, lower EF, NYHA class, serum creatinine, diabetes, not prescribed beta-blocker, lower sys-
`tolic BP, lower body mass, time since diagnosis, current smoker, chronic obstructive pulmonary disease, male gender,
`and not prescribed ACE-inhibitor or angiotensin-receptor blockers. In preserved EF, age was more predictive and
`systolic BP was less predictive of mortality than in reduced EF. Conversion into an easy-to-use integer risk score iden-
`tified a very marked gradient in risk, with 3-year mortality rates of 10 and 70% in the bottom quintile and top decile of
`risk, respectively.
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`Conclusion
`In patients with HF of both reduced and preserved EF, the influences of readily available predictors of mortality can
`be quantified in an integer score accessible by an easy-to-use website www.heartfailurerisk.org. The score has the
`potential for widespread implementation in a clinical setting.
`- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
`Keywords
`Heart failure † Meta-analysis † Prognostic model † Mortality
`
`Introduction
`
`Heart failure (HF) is a major cause of death, but prognosis in indi-
`vidual patients is highly variable. Quantifying a patient’s survival
`prospects based on their overall risk profile will help identify
`those patients in need of more intensive monitoring and therapy,
`and also help target appropriate populations for trials of new
`therapies.
`There exist previous risk models for patients with HF.1 – 8
`Each uses a single cohort of patients and hence their generaliz-
`ability to other populations
`is questionable. Each model’s
`
`development is from a limited cohort size, compromising the
`ability to truly quantify the best risk prediction model. Also
`most models are restricted to patients with reduced left-
`ventricular ejection fraction (EF),
`thus excluding many HF
`patients with preserved EF.
`The Meta-analysis Global Group in Chronic Heart Failure
`(MAGGIC) provides a comprehensive opportunity to develop a
`prognostic model in HF patients, both with reduced and preserved
`EF. We use readily available risk factors based on 39 372 patients
`from 30 studies to provide a user-friendly score that readily quan-
`tifies individual patient mortality risk.
`
`* Corresponding author. Tel: +44 207 927 2413, Fax: +44 207 637 2853, Email: stuart.pocock@lshtm.ac.uk
`Published on behalf of the European Society of Cardiology. All rights reserved. & The Author 2012. For permissions please email: journals.permissions@oup.com
`
`1
`
`APPLE 1079
`Apple v. AliveCor
`IPR2021-00972
`
`
`
`Predicting survival in heart failure
`
`1405
`
`Methods
`The MAGGIC program’s details are documented previously.9 Briefly,
`we have individual patient data from 31 cohort studies (six randomized
`clinical trials and 24 observational registries). Here one registry is
`excluded since it had only median 3-month follow-up. The remainder
`comprised 39 372 patients with a median follow-up of 2.5 years (inter-
`quartile range 1.0 – 3.9 years), during which 15 851 patients (40.2%)
`died. Thirty-one baseline variables were considered as potential pre-
`dictors of mortality (Table 1).
`The Coordinating Centre at the University of Auckland assembled
`the database for 29 studies. The London School of Hygiene and Trop-
`ical Medicine team the added in the CHARM trial data. The online Ap-
`pendix lists the MAGGIC investigators (Supplementary material
`online).
`In 18 studies, a preference was for rounding the EF to the nearest
`5%. In these studies, such rounded values were re-allocated within
`2.5% either side using a uniform distribution.
`
`Table 1 Descriptive statistics for baseline variables
`
`Died
`Alive
`(n 5 15 851)
`(n 5 23 521)
`. . . . . . . . . . . . . . . . . .
`. . . . . . . . . . . . . . . . . .
`Mean
`SD Mean
`SD
`or %
`or %
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`
`Age (years)
`
`Male, %
`
`Non-Caucasian, %
`Body mass index (kg/m2)
`Current smoker, %
`
`Ejection fraction, %
`
`64.3
`
`69.0
`
`10.7
`
`27.5
`
`34.2
`
`36.6
`
`Systolic blood pressure (mmHg)
`
`131.0
`
`Diastolic blood pressure
`(mmHg)
`
`77.7
`
`11.8
`
`5.1
`
`14.0
`
`21.8
`
`12.1
`
`71.9
`
`65.1
`
`7.8
`
`26.0
`
`29.0
`
`33.6
`
`130.5
`
`75.5
`
`10.9
`
`5.0
`
`14.0
`
`25.6
`
`13.5
`
`Haemoglobin (g/L)
`
`133.7
`
`19.0
`
`119.0
`
`26.1
`
`Heart failure duration ≥18
`months, %
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`NYHA class, %
`
`48.8
`
`49.7
`
`I
`
`II
`
`III
`
`10.8
`
`53.8
`
`31.3
`
`6.7
`
`37.1
`
`42.8
`
`IV
`4.1
`13.4
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`Creatinine (mmol/L)
`109.4
`55.8
`126.9
`58.4
`
`138.9
`
`4.2
`
`Statistical methods
`Poisson regression models were used to simultaneously relate baseline
`variables to the time to death from any cause, with study fitted as a
`random effect. Since mortality risk is higher early on, the underlying
`Poisson rate was set in three time bands: up to 3 months, 3 – 6
`months, and over 6 months. Models were built using forward stepwise
`regression with inclusion criterion P , 0.01.
`For binary and categorical variables, dummy variables were used.
`Quantitative variables were fitted as continuous measurements,
`unless there was a clear evidence of non-linearity, e.g. body mass
`index, EF, and creatinine. Also two highly significant statistical interac-
`tions were included in the main model: the impact of age and systolic
`blood pressure both depend on EF.
`Each variable’s strength of contribution to predicting mortality was
`expressed as the z statistic. The larger the z the smaller the P-value,
`e.g.: z values 3.29, 3.89, 5.32, and 6.11 are associated with P-values
`0.001, 0.0001, 0.0000001, and 0.000000001, respectively.
`Missing values are handled by multiple imputations using chained
`equations.10,11 This method has three steps. First, for each variable
`with missing values, a regression equation is created. This model
`includes the outcome and follow-up time,
`in this case the Nelson –
`Aalen estimator (as recommended by White and Royston10), an indi-
`cator variable for each study and other model covariates. For continu-
`ous variables, this is a multivariable linear regression,
`for binary
`variables, a logistic regression, and for ordered categorical variables,
`an ordinal logistic regression. Once all such regression equations are
`defined, missing values are replaced by randomly chosen observed
`values of each variable in the first iteration. For subsequent iterations,
`missing values are replaced by a random draw from the distribution
`defined by the regression equations. This was repeated for 10 itera-
`tions, the final value being the chosen imputed value. This is similar
`to Gibbs sampling.12
`This entire process was repeated 25 times, thus creating 25 imputed
`data sets. The next step was to estimate the model for each of these
`data sets. Finally, the model coefficients are averaged according to
`Rubin’s rule.13 This ensures that the estimated standard error of
`each averaged coefficient reflects both between and within imputation
`variances, giving valid inferences.
`We converted the Poisson model predictor to an integer score, which
`is then directly related to an individual’s probability of dying within 3
`years. A zero score represents a patient at lowest possible risk. Having
`grouped each variable into convenient intervals, the score increases by
`
`Sodium (mmol/L)
`
`Medical history, %
`
`Diabetes
`
`Angina
`
`MI
`
`Atrial fibrillation
`
`Stroke
`
`COPD
`
`Hypertension
`
`Rales
`
`Ischaemic heart disease
`
`CABG
`
`PCI
`
`Branch bundle block
`
`139.7
`
`3.6
`
`20.6
`
`40.3
`
`45.6
`
`17.8
`
`6.2
`
`5.7
`
`41.3
`
`22.3
`
`52.9
`
`15.4
`
`11.7
`
`22.1
`
`25.7
`
`38.6
`
`43.6
`
`23.5
`
`12.2
`
`17.0
`
`39.3
`
`41.7
`
`51.8
`
`13.9
`
`7.9
`
`24.5
`
`Oedema
`21.4
`31.9
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`Shortness of breath, %
`
`Resting
`
`15.9
`
`35.8
`
`Exercise
`80.8
`78.8
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`Medications, %
`
`Beta-blocker
`
`ACE-I
`
`ARB
`
`40.4
`
`68.0
`
`3.3
`
`24.4
`
`60.5
`
`4.3
`
`NYHA, New York Heart Association; COPD, chronic obstructive pulmonary
`disease; ACE-I, angiotensin-converting enzyme inhibitor; ARB,
`angiotensin-receptor blockers; PCI, percutaneous coronary intervention; CABG,
`coronary artery bypass grafting.
`
`an integer amount for each risk factor level above the lowest risk.
`Each integer is a rounding of the exact coefficient in the Poisson
`model, making log rate ratio 0.1 equivalent to 1 point.
`
`2
`
`
`
`1406
`
`S.J. Pocock et al.
`
`The data were analysed using Stata version 12.1 statistical package.
`
`Results
`
`This report is based on 39 372 patients from 30 studies: six were
`randomized controlled trials (24 041 patients) and 24 were regis-
`tries (15 331 patients). Supplementary material online Table S1
`describes each of the 30 studies. Overall, 15 851 (40.2%) patients
`died during a median follow-up of 2.5 years. The six largest studies
`(DIAMOND,14 DIG,15 CHARM,16 and ECHOS17
`trials and
`IN-CHF18 and HOLA19 registries) contributed 75.8% of patients
`and also 75.8% of deaths.
`There were 31 baseline variables available for inclusion in
`prognostic models. Table 1 provides their descriptive statistics
`for patients still alive and patients who died during follow-up.
`Using Poisson regression models for patient survival with
`forward stepwise variable selection, adjusting for study (random
`effect) and follow-up time (higher mortality rate in early follow-
`up), we identified 13 independent predictor variables (Table 2).
`All were highly significant P , 0.002, and most were overwhelm-
`ingly significant, i.e. P , 0.0001.
`Table 3 lists the extent of missing data for these 13 variables. A
`multiple imputation algorithm (see Methods) was used to
`
`overcome this problem. Consequently, all results are based on
`average estimates across 25 imputed data sets.
`For continuous variables, potential non-linearity in the predic-
`tion of survival was explored, as were potential statistical interac-
`tions between predictors. Hence the associations of EF, body mass
`index, and serum creatinine with mortality risk were, respectively,
`confined to EF ,40%, body mass index ,30 kg/m2, and serum
`creatinine ,350 mmol/L. The mortality association of increased
`age was more marked with higher EF, whereas the inverse associ-
`ation of systolic blood pressure with mortality became more
`marked with lower EF.
`Figure 1 displays the independent impact of each predictor on
`mortality risk. The impact of age (which varies with EF) is particu-
`larly strong, and hence is shown on a different scale to the other
`plots.
`From the risk coefficients given in Table 2, an integer score has
`been created (Figure 2). For each patient, the integer amounts con-
`tributed by the risk factor’s values are added up to obtain a total
`integer score for that patient. The bell-shaped distribution of this
`integer risk score for all 39 372 patients is shown in Figure 3.
`The median is 23 points and the range is 0 – 52 points, with 95%
`of patients in the range of 8 – 36 points. The curve in Figure 3
`relates a patient’s score to their probability of dying within 3
`
`Table 2 Multivariable model predicting mortality in all 39 372 patients
`
`Variable
`Rate ratio
`95% CI
`Log rate ratio
`P-value
`Z
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`
`Age (per 10 years)
`
`Males
`BMI (per 1 kg/m2 increase up to 30 kg/m2)a
`Current smoker
`
`SBP (per 10 mmHg increase)
`
`1.154
`
`1.115
`
`0.965
`
`1.159
`
`0.882
`
`(1.092, 1.220)
`
`(1.073, 1.159)
`
`(0.959, 0.972)
`
`(1.109, 1.210)
`
`(0.855, 0.910)
`
`0.143
`
`0.109
`
`20.035
`
`0.147
`
`20.126
`
`5.08
`
`5.58
`
`210.10
`
`6.65
`
`27.85
`
`,0.0001
`
`,0.0001
`
`,0.0001
`
`,0.0001
`
`,0.0001
`
`Diabetes
`1.422
`(1.365, 1.481)
`0.352
`16.85
`,0.0001
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`NYHA
`
`I
`
`II
`
`III
`
`0.788
`
`1.000
`
`1.410
`
`(0.732, 0.848)
`
`20.239
`
`26.35
`
`,0.0001
`
`(1.354, 1.467)
`
`0.343
`
`16.75
`
`,0.0001
`
`IV
`1.684
`(1.580, 1.796)
`0.521
`16.05
`,0.0001
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`Ejection fraction (per 5% increase up to 40%)a
`0.581
`(0.539, 0.627)
`20.542
`214.03
`,0.0001
`COPD
`
`6.36
`
`,0.0001
`
`1.228
`
`(1.152, 1.310)
`
`0.206
`
`HF duration .18 months
`
`Creatinine (per 10 mmol/L up to 350 mmol/L)
`
`Beta-blocker
`
`ACE-I/ARB
`Interaction of ejection fraction and ageb
`Interaction of ejection fraction and SBPc
`
`1.188
`
`1.039
`
`0.760
`
`0.908
`
`1.040
`
`1.012
`
`(1.139, 1.240)
`
`(1.035, 1.042)
`
`(0.726, 0.796)
`
`(0.856, 0.963)
`
`(1.031, 1.049)
`
`(1.008, 1.017)
`
`0.173
`
`0.038
`
`20.274
`
`20.096
`
`0.039
`
`0.012
`
`7.96
`
`19.82
`
`211.77
`
`23.26
`
`9.05
`
`5.13
`
`,0.0001
`
`,0.0001
`
`,0.0001
`
`0.002
`
`,0.0001
`
`,0.0001
`
`BMI, body mass index; SBP, systolic blood pressure; NYHA, New York Heart Association; COPD, chronic obstructive pulmonary disease; HF, heart failure; ACE-I,
`angiotensin-converting enzyme inhibitor; ARB, angiotensin-receptor blockers.
`aThe BMI variable has a linear trend up to 30 kg/m2, while above 30 kg/m2 the risk is constant. Similarly, for ejection fraction, the risk is constant above 40%, and for creatinine risk is
`constant above 350 mmol/L.
`bThe interaction between ejection fraction and age indicates an extra 4% increase in mortality for each simultaneous 10-year increase in age and 5% increase in ejection fraction on
`top of the risks of ejection fraction and age considered independently, i.e. the protective effect of increased ejection fraction function diminishes as a patient ages (Figure 1).
`cThe interaction between ejection fraction and SBP indicates an extra 1.2% increase in mortality for each simultaneous 10 mmHg increase in SBP and 5% increase in ejection
`fraction on top of the risks of ejection fraction and SBP considered independently, i.e. the protective effect of increased ejection fraction function diminishes as a patient’s SBP
`increases (Figure 1).
`
`3
`
`
`
`Predicting survival in heart failure
`
`1407
`
`Table 3 Extent of missing data
`
`Model variable
`
`Studies with no data
`Studies with some data
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`Studies
`Missing patients
`Studies
`Missing patients
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`
`Total patients missing data
`
`Age
`
`Gender
`
`BMI
`
`Current smoker
`
`SBP
`
`Diabetes
`
`NYHA class
`
`Ejection fraction
`
`COPD
`
`HF duration
`
`Creatinine
`
`Beta-blocker
`
`ACE-I/ARB
`
`0
`
`0
`
`17
`
`6
`
`9
`
`1
`
`5
`
`6
`
`10
`
`20
`
`5
`
`3
`
`1
`
`0
`
`0
`
`14 515
`
`9166
`
`12 016
`
`348
`
`2503
`
`3279
`
`16 788
`
`11 679
`
`2800
`
`7890
`
`97
`
`0
`
`0
`
`13
`
`24
`
`21
`
`29
`
`25
`
`24
`
`20
`
`10
`
`25
`
`27
`
`29
`
`0
`
`0
`
`2686
`
`448
`
`276
`
`341
`
`1128
`
`3558
`
`253
`
`1066
`
`17 245
`
`709
`
`649
`
`0
`
`0
`
`17 201
`
`9614
`
`12 292
`
`689
`
`3631
`
`6837
`
`17 041
`
`12 745
`
`20 045
`
`8599
`
`746
`
`BMI, body mass index; SBP, systolic blood pressure; NYHA, New York Heart Association; COPD, chronic obstructive pulmonary disease; HF, heart failure; ACE-I,
`angiotensin-converting enzyme inhibitor; ARB, angiotensin-receptor blockers.
`
`years. For instance, scores of 10, 20, 30, and 40 have 3-year prob-
`abilities 0.101, 0.256, 0.525, and 0.842, respectively. Table 4 details
`the link between any integer score and the probabilities of dying
`within 1 year and 3 years.
`Figure 4 shows mortality over 3 years for patients classified into
`six risk groups. Groups 1 – 4 comprise patients with scores 0 – 16,
`17 – 20, 21 – 24, and 25 – 28, respectively, approximately the first
`four quintiles of risk. To give more detail at higher risk, groups 5
`and 6 comprise patients with scores 29 – 32 and 33 or more, ap-
`proximately the top two deciles of risk. The marked continuous
`separation of the six Kaplan – Meier curves is striking: the 3-year
`% dead in the bottom quintile and top decile is 10 and 70%,
`respectively.
`Regarding model goodness-of-fit, Figure 5 compares observed
`and model-predicted 3-year mortality risk across the six risk
`groups. In the bottom two groups, the observed mortality is slight-
`ly lower than that predicted by the model, but overall the marked
`gradient in risk is well captured by the integer score.
`Tables 5 and 6 show two separate models for patients with
`reduced and preserved left-ventricular function (EF ,40 and
`≥40%, respectively). For most predictors, the strength of mortality
`association is similar in both subgroups. However, the impact of
`age is more marked and the impact of lower SBP is less marked
`in patients with preserved left-ventricular function, consistent
`with the interactions in the overall model.
`In this meta-analysis of 30 cohort studies, we explored between-
`study heterogeneity in mortality prediction. From fitting separate
`models for each study, we observe a good consistency across
`studies re the relative importance of the predictors (data not
`shown). We have also repeated the model in Table 2, now fitting
`study as a fixed effect (rather than a random effect). This reveals
`substantial between-study differences
`in mortality risk not
`explained by predictors in our model. However, a comparison of
`
`the seven randomized trials with the 23 patient registries reveals
`no significant difference in their mortality rates.
`
`Discussion
`
`This study identifies 13 independent predictors of mortality in HF.
`Although all have been previously identified, the model and risk
`score reported here are the most comprehensive and generaliz-
`able available in the literature. They are based on 39 372 patients
`from 30 studies with a median follow-up of 2.5 years, the largest
`available database of HF patients. Also, we include patients with
`both reduced and preserved EF, the latter being absent from
`most previous models of HF prognosis.
`Given the wide variety of different studies included, with a global
`representation, the findings are inherently generalizable to a broad
`spectrum of current and future patients. Conversion of the risk
`model
`into a user-friendly integer score accessible by the
`website www.heartfailurerisk.org facilitates its use on a routine in-
`dividual patient basis by busy clinicians and nurses.
`All 13 predictors in the risk score should be routinely available,
`though provision will be made in the website for one or two vari-
`ables to be unknown for an individual. Note, the ‘top five’ predic-
`tors age, EF, serum creatinine, New York Heart Association
`(NYHA) class, and diabetes are important to know. The inverse as-
`sociation of EF with mortality is well established, and as previously
`reported,9 in above 40% there appears no further trend in progno-
`sis. We included serum creatinine rather than creatinine clearance
`or eGFR. The latter involve formulae that include age, which would
`artificially diminish the huge influence of age on prognosis.
`We confirm the association of body mass index with mortality,20
`but with a cut-off of 30 kg/m2, above which there appears no
`further trend. While others report heart rate as a significant pre-
`dictor of mortality,21 we find that once the strong influence of
`
`4
`
`
`
`1408
`
`S.J. Pocock et al.
`
`Figure 1 Mortality rate ratios (and 95% CIs) for each variable in the predictive model. All charts are on the same scale except that for the
`interaction between ejection fraction and age, where the impact on mortality is more marked.
`
`beta blocker use is included, heart rate was not a strong independ-
`ent predictor. A modest association of ACE-inhibitor and/or angio-
`tensin-receptor blockers (ARB) use with lower mortality was
`highly significant, though many of our cohorts were established
`before ARBs were routinely available.
`infarction,
`Cardiovascular disease history (e.g. myocardial
`angina, stroke, atrial fibrillation, LBBB) was considered in our
`model development. What mattered most was the time since
`
`first diagnosis of HF, best captured by whether this exceeds
`18 months. Besides the powerful influence of diabetes, the other
`disease indicator of a poorer prognosis was prevalence of COPD.
`Previous myocardial
`infarction, atrial fibrillation, and LBBB were
`not sufficiently strong independent predictors of risk to be included
`in our model.
`For patients with reduced and preserved EF, we developed sep-
`arate risk models (Tables 5 and 6). Nearly all predictors display a
`
`5
`
`
`
`Predicting survival in heart failure
`
`1409
`
`Figure 2 A chart to calculate the integer risk score for each patient.
`
`similar influence on mortality in both subgroups. Two exceptions
`are age (better prognosis of preserved EF compared with
`reduced EF HF is more pronounced at younger ages) and systolic
`blood pressure, which have a stronger inverse association with
`mortality in patients with reduced EF. These two interactions are
`incorporated into the integer risk score, as displayed in Figure 1.
`Our meta-analysis of 30 cohort studies enables exploration of
`between-study differences in mortality risk. Separately, for each
`of
`the 10 largest studies, we calculated Poisson regression
`models for the same 13 predictors.
`Informal
`inspection of
`models across
`studies
`shows a consistent pattern to be
`expected, given there are no surprises among the selected
`predictors.
`An additional model, with study included as a fixed effect (rather
`than a random effect), reveals some between-study variation in
`mortality risk not captured by the predictor variables. This may
`be due to geographic variations or unidentified patient-selection
`criteria varying across registries and clinical trials, though overall
`patients in registries and trials appear at similar risk. Also, calendar
`
`Figure 3 Distribution of the integer risk score for all 39 372
`patients, and its association with the risk of dying (and 95% CI)
`within 3 years.
`
`6
`
`
`
`1410
`
`S.J. Pocock et al.
`
`Table 4 Predicted probabilities of death for each integer risk score
`
`Integer risk score
`
`3-year probability
`1-year probability
`Integer
`3-year probability
`1-year probability
`of death
`of death
`risk score
`of death
`of death
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`
`0
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`8
`
`9
`
`10
`
`11
`
`0.015
`
`0.016
`
`0.018
`
`0.020
`
`0.022
`
`0.024
`
`0.027
`
`0.029
`
`0.032
`
`0.036
`
`0.039
`
`0.043
`
`0.039
`
`0.043
`
`0.048
`
`0.052
`
`0.058
`
`0.063
`
`0.070
`
`0.077
`
`0.084
`
`0.092
`
`0.102
`
`0.111
`
`26
`
`27
`
`28
`
`29
`
`30
`
`31
`
`32
`
`33
`
`34
`
`35
`
`36
`
`37
`
`38
`
`0.175
`
`0.191
`
`0.209
`
`0.227
`
`0.248
`
`0.269
`
`0.292
`
`0.316
`
`0.342
`
`0.369
`
`0.398
`
`0.427
`
`0.458
`
`0.397
`
`0.427
`
`0.458
`
`0.490
`
`0.523
`
`0.556
`
`0.590
`
`0.625
`
`0.658
`
`0.692
`
`0.725
`
`0.756
`
`0.787
`
`12
`
`13
`
`14
`
`15
`
`16
`
`17
`
`18
`
`19
`
`20
`
`21
`
`22
`
`23
`
`24
`
`25
`
`0.048
`
`0.052
`
`0.058
`
`0.063
`
`0.070
`
`0.077
`
`0.084
`
`0.093
`
`0.102
`
`0.111
`
`0.122
`
`0.134
`
`0.147
`
`0.160
`
`0.122
`
`0.134
`
`0.146
`
`0.160
`
`0.175
`
`0.191
`
`0.209
`
`0.227
`
`0.247
`
`0.269
`
`0.292
`
`0.316
`
`0.342
`
`0.369
`
`39
`
`40
`
`41
`
`42
`
`43
`
`44
`
`45
`
`46
`
`47
`
`48
`
`49
`
`50
`
`0.490
`
`0.523
`
`0.557
`
`0.591
`
`0.625
`
`0.659
`
`0.692
`
`0.725
`
`0.757
`
`0.787
`
`0.816
`
`0.842
`
`0.815
`
`0.842
`
`0.866
`
`0.889
`
`0.908
`
`0.926
`
`0.941
`
`0.953
`
`0.964
`
`0.973
`
`0.980
`
`0.985
`
`Figure 4 Cumulative mortality risk over 3 years for patients
`classified into six risk groups. Risk groups 1 – 4 represent the
`first four quintiles of risk (integer scores 0 – 16, 17 – 20, 21 – 24,
`and 25 – 28, respectively). Risk groups 5 and 6 represent the
`top two deciles of risk (integer scores 29 – 32 and 33 or more,
`respectively). 95% CIs are plotted at 1, 2, and 3 years follow-up.
`
`Figure 5 Observed vs. model-predicted 3-year mortality in six
`risk groups.
`
`7
`
`
`
`Predicting survival in heart failure
`
`1411
`
`Table 5 Main effects model for EF <40 (21 442 patients of whom 8900 died)
`
`Variable
`Rate ratio
`95% CI
`P-value
`Z
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`
`Age (per 10 years)
`
`Male
`BMI (per 1 kg/m2 increase up to 30 kg/m2)
`Current smoker
`
`SBP (per 10 mmHg increase)
`
`1.407
`
`1.101
`
`0.970
`
`1.154
`
`0.936
`
`(1.375, 1.439)
`
`(1.044, 1.161)
`
`(0.961, 0.978)
`
`(1.091, 1.222)
`
`(0.924, 0.948)
`
`29.54
`
`3.57
`
`27.32
`
`4.99
`
`210.06
`
`,0.001
`
`,0.001
`
`,0.001
`
`,0.001
`
`,0.001
`
`Diabetes
`1.421
`(1.347, 1.499)
`13.00
`,0.001
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`NYHA classs
`
`I
`
`II
`
`III
`
`0.828
`
`1.000
`
`1.372
`
`(0.744, 0.922)
`
`(1.303, 1.445)
`
`23.44
`
`12.03
`
`0.001
`
`,0.001
`
`IV
`1.640
`(1.503, 1.790)
`11.21
`,0.001
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`Ejection fraction (per 5% increase)
`0.915
`(0.902, 0.928)
`212.34
`,0.001
`
`COPD
`
`HF duration .18 months
`
`Creatinine (per 10 mmol/L up to 350 mmol/L)
`
`Beta-blocker
`
`1.191
`
`1.191
`
`1.041
`
`0.736
`
`0.834
`
`(1.096, 1.295)
`
`(1.127, 1.259)
`
`(1.035, 1.046)
`
`(0.694, 0.781)
`
`(0.770, 0.905)
`
`4.17
`
`6.22
`
`15.65
`
`210.21
`
`24.47
`
`,0.001
`
`,0.001
`
`,0.001
`
`,0.001
`
`,0.001
`
`ACE-I/ARB
`
`BMI, body mass index; SBP, systolic blood pressure; NYHA, New York Heart Association; COPD, chronic obstructive pulmonary disease; HF, heart failure; ACE-I,
`angiotensin-converting enzyme inhibitor; ARB, angiotensin-receptor blockers.
`
`Table 6 Main effects model for EF ≥40 (17 930 patients of whom 6951 died)
`
`Variable
`Rate ratio
`95% CI
`P-value
`Z
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`
`Age (per 10 years)
`
`Male
`BMI (per 1 kg/m2 increase up to 30 kg/m2)
`Current smoker
`
`SBP (per 10 mmHg)
`
`1.589
`
`1.113
`
`0.960
`
`1.174
`
`0.982
`
`(1.536, 1.643)
`
`(1.053, 1.177)
`
`(0.951, 0.969)
`
`(1.095, 1.258)
`
`(0.968, 0.998)
`
`27.14
`
`3.77
`
`28.50
`
`4.54
`
`22.30
`
`,0.001
`
`,0.001
`
`,0.001
`
`,0.001
`
`0.024
`
`Diabetes
`1.401
`(1.311, 1.498)
`9.90
`,0.001
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`NYHA class
`
`I
`
`II
`
`III
`
`0.756
`
`1.000
`
`1.458
`
`(0.682, 0.838)
`
`(1.361, 1.561)
`
`25.32
`
`10.83
`
`,0.001
`
`,0.001
`
`IV
`1.756
`(1.599, 1.928)
`11.82
`,0.001
`. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
`COPD
`1.284
`(1.181, 1.396)
`5.91
`,0.001
`
`HF duration .18 months
`
`Creatinine (per 10 mmol/L up to 350 mmol/L)
`
`Beta-blocker
`
`ARB/ACE-I
`
`1.166
`
`1.035
`
`0.798
`
`0.938
`
`(1.088, 1.250)
`
`(1.029, 1.041)
`
`(0.746, 0.855)
`
`(0.842, 1.044)
`
`4.37
`
`11.39
`
`26.47
`
`21.21
`
`,0.001
`
`,0.001
`
`,0.001
`
`0.233
`
`BMI, body mass index; SBP, systolic blood pressure; NYHA, New York Hear Association; COPD, chronic obstructive pulmonary disease; HF, heart failure; ACE-I,
`angiotensin-converting enzyme inhibitor; ARB, angiotensin-receptor blockers.
`
`year may be relevant since improved treatment of HF may enhance
`prognosis in more recent times. We will explore these issues in a
`subsequent publication.
`The integer risk score gives a very powerful discrimination of
`patients’ mortality risk over 3 years, and also has excellent
`
`to the data across all 30 studies combined
`goodness-of-fit
`(Figures 3 and 4). Specifically, the score facilitates the identification
`of low-risk patients, e.g. score ,17 has an expected 90% 3-year
`survival, and very high-risk patients, e.g. score ≥33 has an expected
`30% 3-year survival.
`
`8
`
`
`
`1412
`
`S.J. Pocock et al.
`
`We recognize some limitations. In