Updates to the SHFM 2013
has been updated to allow application of the model to higher
risk hospitalized patients by including IV diuretics, inotropes, IABP, ventilator,
ultrafiltration, incorporation of newer LVADs and reflect newer guidelines for
. These changes will more accurately reflect the SHFM
as it is being employed in the ROADMAP and REVIVE-IT LVAD protocols. The changes
to the SHFM are outlined below. The 2013 AHA/ACC Heart Failure Guidelines state
that "validated multivariable risk scores can be useful to estimate subsequent
risk of mortality in ambulatory or hospitalized patients with HF" (Class IIa)2
1) The updated model allows use of IV diuretics by checking the IV box next to the diuretic based on the oral bioavailability of the diuretics.
2) The updated model incorporates the hazard ratio associated with intra-aortic balloon pump/ventilator/ultrafiltration based on the COCPIT analysis3
. This is the same methodology we used for the REMATCH, ADVANCE, and European HeartWare analysis4-6
3) The updated model incorporates the use of inotropes from the Cochrane meta-analysis as we did for REMATCH, ADVANCE, and European HeartWare analysis4-6
. The model automatically assigns a patient on a ventilator, IABP, ultrafiltration, or inotrope to NYHA 4 (i.e. ambulatory inotropes are NYHA 4).
4) We changed the LVAD hazard ratio from 0.52 (REMATCH) to 0.3 (70% mortality
benefit) for the newer generation LVADs. This reflects an additional 46% mortality
reduction with HeartMate II vs. MeartMate XVE DT trials (hazard ratio 0.52*0.54=0.3)7,8
For a patient with a 25% one year survival (i.e. REMATCH), the addition of
the XVE LVAD would have increased the one year survival to 52% (prior model)
and to 66% with a HeartMate II (updated model). This is concordant with the 1-year
results in REMATCH (52%) and the HeartMate II (68%) destination therapy trials.
The online version remains programmed to only ungray the LVAD box (make the choice
available) if the patient is NYHA 4, EF ≤25% with a 2-year survival of 50% (<2
year median life expectancy) to be concordant with the entry criteria for REMATCH
(i.e. patients expected to live ≤2 years). If the user wishes, they change alter
values to meet the LVAD criteria, click the LVAD box and the LVAD box will remain
checked but now grayed out. This will allow estimates in NYHA 3B or annual mortality
of <29%. These are clinical areas being explored in the ROADMAP registry and
the REVIVE-IT randomized trials of LVADs.
5) The model incorporates a variable ICD benefit based on the estimated annual mortality without an ICD. There is ~50% ICD mortality benefit at ~5% annual mortality decreasing to 0% benefit at ~25% annual mortality. This is based on the prospective application of the SHFM-D to SCD-HeFT9
. The SHFM-D is a version that included similar variables, but only serum sodium and creatinine as the other lab variables were not available in SCD-HeFT. The major effect of this is to decrease the ICD benefit in very high risk patients where the primary mode of death is pump failure not sudden death and the ICD has no or minimal benefit in this group9-11
. Thus, at an annual mortality of >25%, there likely is no benefit from the addition of an ICD and the health care provider should choose alternative therapies such as palliative care, transplant, or LVAD. This is concordant with the very high risk patients in MADIT 2 having no ICD benefit11
. This is consistent with requiring "a reasonable expectation of meaningful survival for >1 year" (i.e. 1 year survival has to be >75%)2
. The ICD selection box is now programmed to allow adding the ICD for NYHA 2-3 ischemic/non-ischemic with EF ≤35% and NYHA 1 ischemic with EF ≤30% (MADIT 2) consistent with the AHA/ACC 2013 Heart Failure Guidelines (Class I)2
6) We changed the estimate of the 0 to 5-year mortality (baseline survival function) to reflect the observed survival in the derivation of the SHFM-D model in 10,038 patients 9-11
with >5 years of observation. In the original SHFM, the 5-year survival was estimated from the PRAISE trial from the 0 to 3 year survival fitted with an exponential decline to estimate values at 5 years. We now employ the observed 1 to 5-year survival which is slightly higher at times beyond 1-year. This has a very modest effect (i.e. for a 5% 1-year mortality, it increases the 5-year mortality estimate from 22% to 25%, with a corresponding decrease in the estimated life expectancy from 10.6 to 9.8 years).
7) We update the programming to reflect ACC/AHA 2013 Heart Failure Guidelines (Class I or IIa) for BiV pacers and BIV ICDs in NYHA 2-ambulatory 4 patients with EF≤35% with LBBB >120 ms and NYHA 3-ambulatory 4 patients with QRS >150 ms with EF ≤35%2
. Clicking these boxes does not alter the risk of the patient in the current model, but merely ungrays the appropriate device box to allow selection by the user. Patients with non LBBB QRS 120-149 ms, who are now ACC/AHA Class IIb criteria, are not suggested to have a BIV or BIV ICD (i.e. box remains grayed out). Previously BIV and BiV ICD were available for NYHA 3 and ambulatory 4, EF≤35% with QRS ≥120ms. We have tried to incorporate changes from one device to another (i.e. ICD to CRT-D, CRT-D to ICD, CRT to CRT-D, etc). The user is cautioned to use clinical judgment in all device decisions.
8) We removed the option of adding a statin and deriving benefit given the negative results with the addition of statins in large randomized clinical trials.
9) The user is reminded that the SHFM provides a mean (average) life expectancy and not a median life expectancy. This is the area under a Kaplan Meier curve for the lifetime of the patient. This is not the same as a median survival (50% alive/50% dead) due to the nonlinear distribution of survival. For example, at a 50% 2-year median survival (median survival), the estimated mean survival for the will be 2.6 years.
10) If a user would like to perform a cohort validation study of the SHFM, the user should contact email@example.com to ensure the analysis is performed correctly. The appropriate value to enter into a Cox proportional hazards model is the Seattle Heart Failure Model Score (SHFMScore) as described in the original publication1
and not the % survival, % mortality, or mean (average) life expectancy. Similarly, to construct a predicted Kaplan Meier curve use the average of the individual predicted survivals for the cohort at 1-5 years. Failure to understand these differences can result in errors 12. The SHFMScore is not provided in the online model, but can be calculated with the following equation. Representative SHFM Scores and 1-year survivals are shown.
SHFMScore=Ln(Ln(SHFM estimated 1-year survival)/Ln(0.9604))
| SHFM Score|| SHFM Estimated 1-Year Survival|
| 0|| 96.0%|
| 1|| 89.5%|
| 2|| 74.0%|
| 3|| 44.0%|
| 4|| 10.8%|
1. Levy WC, Mozaffarian D, Linker DT, et al. The Seattle Heart Failure Model: prediction of survival in heart failure. Circulation. Mar 21 2006;113(11):1424-1433.
2. Yancy CW, Jessup M, Bozkurt B, et al. 2013 ACCF/AHA guideline for the management of heart failure: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Journal of the American College of Cardiology. Oct 15 2013;62(16):e147-239.
3. Smits JM, Deng MC, Hummel M, et al. A prognostic model for predicting waiting-list mortality for a total national cohort of adult heart-transplant candidates. Transplantation. Oct 27 2003;76(8):1185-1189.
4. Levy WC, Mozaffarian D, Linker DT, Farrar DJ, Miller LW, Investigators R. Can the Seattle heart failure model be used to risk-stratify heart failure patients for potential left ventricular assist device therapy? The Journal of heart and lung transplantation : the official publication of the International Society for Heart Transplantation. Mar 2009;28(3):231-236.
5. Strueber M, O'Driscoll G, Jansz P, et al. Multicenter evaluation of an intrapericardial left ventricular assist system. Journal of the American College of Cardiology. Mar 22 2011;57(12):1375-1382.
6. Aaronson KD, Slaughter MS, Miller LW, et al. Use of an intrapericardial, continuous-flow, centrifugal pump in patients awaiting heart transplantation. Circulation. Jun 26 2012;125(25):3191-3200.
7. Rose EA, Gelijns AC, Moskowitz AJ, et al. Long-term use of a left ventricular assist device for end-stage heart failure. The New England journal of medicine. Nov 15 2001;345(20):1435-1443.
8. Slaughter MS, Rogers JG, Milano CA, et al. Advanced heart failure treated with continuous-flow left ventricular assist device. The New England journal of medicine. Dec 3 2009;361(23):2241-2251.
9. Levy WC, Lee KL, Hellkamp AS, et al. Maximizing survival benefit with primary prevention implantable cardioverter-defibrillator therapy in a heart failure population. Circulation. Sep 8 2009;120(10):835-842.
10. Mozaffarian D, Anker SD, Anand I, et al. Prediction of mode of death in heart failure: the Seattle Heart Failure Model. Circulation. Jul 24 2007;116(4):392-398.
11. Barsheshet A, Moss AJ, Huang DT, McNitt S, Zareba W, Goldenberg I. Applicability of a risk score for prediction of the long-term (8-year) benefit of the implantable cardioverter-defibrillator. Journal of the American College of Cardiology. Jun 5 2012;59(23):2075-2079.
12. Levy WC. Seattle Heart Failure Model. The American journal of cardiology. Apr 15 2013;111(8):1235.