Jump to comment:

• Authors reply to Kearney and Nolan: Predicting outcome in chronic heart failure
  Marcel L. Bouvy
  Published on: 9 January 2022
• Predicting outcome in chronic heart failure
  Mark T Kearney
  Published on: 9 January 2022

• Published on: 9 January 2022

  Authors reply to Kearney and Nolan: Predicting outcome in chronic heart failure

  • Marcel L. Bouvy
  • Other Contributors:
    • E. Rob Heerdink, Bert G.M. Leufkens, Arno W. Hoes

  Dear Editor

  We do not agree that our study population was highly selected. The only exclusion criteria in our study were severe psychiatric problems and/or dementia, planned admission to a nursing home, not taking care of their own medication (e.g. filled or administered by relatives or district nurses) and/or life expectancy of less than 3 months. We feel that we sufficiently described the characteristics of our...

  Show More

  Dear Editor

  We do not agree that our study population was highly selected. The only exclusion criteria in our study were severe psychiatric problems and/or dementia, planned admission to a nursing home, not taking care of their own medication (e.g. filled or administered by relatives or district nurses) and/or life expectancy of less than 3 months. We feel that we sufficiently described the characteristics of our population (co-morbidity, physical examination and medication that describe our population). We choose not to report cardio thoracic ratio and echocardiography data, because although they were available on the majority of patients, there were many differences in the way these data were reported between the participating hospitals. More objective assesments of cardiac function such as ejection fraction or fractional shortening were mostly missing. Once more we want to emphasise that the major aim of our study was to develop a comprehensive and easily applicable prognostic model predicting the risk of death in patients with heart failure, based on information that is readily available in daily care. In their study Kearney et al.[1] did not include co-morbidity and relatively easy to obtain characteristics from physical examination such as weight, ankle oedema and blood pressure. In addition patients with diabetes were excluded in their study.[2] In our opinion it is very important to first study whether these relatively easily obtainable characteristics are predictive, before assessing the additional value of more difficult to obtain data from cardiovascular imaging. We also want to emphasise that several similarities between Kearney’s and our study are noticeable. Both studies find renal function as an important predictor. We used a history of renal insufficiency as a binomial value, but inclusion of the continuous variables serum creatinine or creatinine clearance (calculated with Cockroft and Gault’s formula) led to similar findings. Kearney et al find a predictive value of low sodium levels. We also found low sodium levels to be predictive, although slightly less than other variables and therefore not included in the model. Mean frusemide dose (or bumetanide equivalent) in our study was 124 mg/day. We did not include dose in the model, but high frusemide dose and low ACE-inhibitor dose could also be important and readily available predictors of mortality and might be subject of future studies. Although not reported bootstrapping was performed and yielded similar results.

  References

  (1) Kearney MT, Fox KA, Lee AJ et al. Predicting death due to progressive heart failure in patients with mild-to-moderate chronic heart failure. J Am Coll Cardiol 2002;40:1801-8.

  (2) Bouvy ML, Heerdink ER, Leufkens HGM, Hoes AW. Predicting mortality in patients with heart failure: a pragmatic approach. Heart 2003;89:605-609.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 9 January 2022

  Predicting outcome in chronic heart failure

  • Mark T Kearney, British Heart Foundation Intermediate Fellow and Consultant Cardiologist
  • Other Contributors:
    • James Nolan

  Dear Editor

  We read with interest the article by Bouvy and colleagues [1] and the accompanying editorial by Professor Cowie.[2] We would like to congratulate the investigators in attempting to construct a prognostic rule for patients with chronic heart failure. This is a neglected area in heart failure research.

  We have recently completed 5-year follow up of a study with the specific aim of identifying independe...

  Show More

  Dear Editor

  We read with interest the article by Bouvy and colleagues [1] and the accompanying editorial by Professor Cowie.[2] We would like to congratulate the investigators in attempting to construct a prognostic rule for patients with chronic heart failure. This is a neglected area in heart failure research.

  We have recently completed 5-year follow up of a study with the specific aim of identifying independent predictors of mortality in patients with mild to moderate chronic heart failure.[3,4] Using over 2000 patient years of follow up and 201 deaths we have constructed our own prognostic index.[5] This index integrates abnormalities of cardiac autonomic and electrical function, metabolic disturbances and abnormalities of cardiac size.

  The report by Bouvy et al. has a number of problems, which one should bear in mind before the results are incorporated into clinical practice. As the authors acknowledge the study cohort is a highly selected group of patients taking part in a study not designed to evaluate variables in predicting outcome in chronic heart failure. The variables employed were not selected a priori and this should be recognised as a potential source of bias. Furthermore, this may account for the very impressive C statistic of 0.80 for the model described.

  Before one can apply a prognostic rule clinicians should know the exact characteristics of the population in question. Bouvy et al do not present data describing the whole population, markers of heart failure severity such as mean frusemide dose, cardiothoracic ratio and echocardiographic variables are an important part of patient characterisation. Furthermore, renal insufficiency is part of the index yet the authors do not clearly define what they mean by this.

  As the authors also admit the prognostic index requires validation in a separate cohort. To give the reader an idea of its internal validity and potential transportability Bouvy et al. should have performed ‘bootstrapping’ [5] or a similar technique.

  In conclusion we agree that development of prognostic indices is an important goal for patients with chronic heart failure, and Bouvy et al. have made a commendable attempt at this. However, further details of their study and prospective validation in an adequately powered trial, in a clearly defined population needs to be performed before clinicians can consider transfering the proposed model into clinical practice.

  References

  (1) Bouvy ML, Heerdink ER, Leufkens HGM, Hoes AW. Predicting mortality in patients with heart failure: a pragmatic approach. Heart 2003;89:605-609.

  (2) Cowie MR. Estimating prognosis in heart failure: time for a better approach. Heart 2003;89:587-588.

  (3) Kearney MT, Fox KA, Lee AJ et al. Predicting death due to progressive heart failure in patients with mild-to-moderate chronic heart failure. J Am Coll Cardiol 2002;40:1801-8.

  (4) Kearney MT, Zaman A, Eckberg DL et al. Cardiac size, autonomic function, and 5-year follow-up of chronic heart failure patients with severe prolongation of ventricular activation. J Card Fail 2003;9:93-9.

  (5) Kearney MT, Nolan J, Lee AJ et al. A prognostic index to predict long-term mortality in patients with mild to moderate chronic heart failure stabilised on angiotensin converting enzyme inhibitors. Eur J Heart Fail; in press

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top