MR-proADM Improves Disposition Strategies in Acute Dyspnea
MR-proADM Improves Disposition Strategies in Acute Dyspnea
Data used for this analysis derive from the BACH trial, which was a prospective, 15-centre international study of 1641 patients presenting to the emergency department (ED) with dyspnoea. The primary prognostic endpoint of the original trial was 90-day survival and the investigators tested the superiority of MR-proADM versus BNP for prognosis in patients admitted with heart failure (n=568, 34.6%). Details on the BACH trial have been described in detail elsewhere.
The BACH trial was approved by the institutional review boards of all 15 participating centres. The study complies with the Declaration of Helsinki and is registered in clinical Trials.gov (registration number NCT00537628).
The current study analyses the data of all 1557 patients of the eight US sites (n=831) and of the six European sites (Germany, Switzerland, Italy, Greece, UK and Poland, n=726). Data from the site in New Zealand (n=84) were omitted from the analysis due to their low number of patients.
Characteristics of blood collection and sampling have been described. For MR-proADM, the limit of quantification was 0.23 nmol/l; the within run imprecision (coefficient of variance, CV) was 1.9% and the between run CV was 9.8%. The 97.5th percentile in a normal range population was 0.49 nmol/l with a mean (SD) of 0.33 (0.07) nmol/l.
Patient disposition levels differed between USA and European Union (EU) sites. Patients in Europe were discharged, admitted to the general ward or admitted to a cardiac care unit (CCU)/intensive care unit (ICU). CCU/ICU in Europe were treated as one level as in many European hospitals CCUs were part of the ICUs rather than being separate units.
In the USA, CCUs or monitoring units as a care level between general ward and ICU were also available for disposition. As both, the levels as well as the distribution of patients to each level, were not comparable between EU and USA sites, the analysis was performed for both global regions separately.
All outcome prediction results reported in the current analysis are for all-cause mortality within 90 days after presentation to the ED.
Values are expressed as means and SDs, medians and interquartile ranges (IQR), or counts and percentages as appropriate. As MR-proADM concentrations were non parametric, data were log-transformed before plotting. A two-sided p value of 0.05 was used for significance.
All statistical analyses were performed using R V.2.5.1 (http://www.r-project.org), SAS V.9.1 (SAS Institute), or Statistical Package for the Social Sciences (SPSS) V.16.0 (SPSS Inc).
To simulate patient disposition with MR-proADM guidance, we applied the following adapted NRI- model. We used the original disposition as a starting point, assuming that standard risk stratification was the main influence on disposition. Based on that, we used the MR-proADM concentration to change the disposition. The original disposition was changed by one level up (upgrading) or down (downgrading) only. We compared the original disposition with the hypothetical disposition by adapting the method of NRI introduced by Pencina et al.
The goal was to identify an algorithm both effective and simple for both global regions. Therefore, cut-off values were identified as follows: A MR-proADM cut-off was considered useful only if
The identified cut-offs for MR-proADM for upgrading care were ≥5.0 nmol/l in both the EU and USA. For downgrading, a MR-proADM value <0.5 nmol/l was used in USA. In the EU, the same cut-off was used for downgrading from general ward to discharge, but a higher cut-off (1.0 nmol/l) was applied for downgrading from CCU/ICU to general ward. Following the disposition strategy as suggested by the MR-proADM cut-off could result in an unfavourable disposition for individual patients.
To make sure that both survivors and non-survivors have improved reclassification, reclassification for survivors and non-survivors was applied separately, as suggested by Pencina et al.
Methods
Study Setting and Population of the Original Trial
Data used for this analysis derive from the BACH trial, which was a prospective, 15-centre international study of 1641 patients presenting to the emergency department (ED) with dyspnoea. The primary prognostic endpoint of the original trial was 90-day survival and the investigators tested the superiority of MR-proADM versus BNP for prognosis in patients admitted with heart failure (n=568, 34.6%). Details on the BACH trial have been described in detail elsewhere.
The BACH trial was approved by the institutional review boards of all 15 participating centres. The study complies with the Declaration of Helsinki and is registered in clinical Trials.gov (registration number NCT00537628).
Population of the Current Study
The current study analyses the data of all 1557 patients of the eight US sites (n=831) and of the six European sites (Germany, Switzerland, Italy, Greece, UK and Poland, n=726). Data from the site in New Zealand (n=84) were omitted from the analysis due to their low number of patients.
Measurements and Outcome Measures
Characteristics of blood collection and sampling have been described. For MR-proADM, the limit of quantification was 0.23 nmol/l; the within run imprecision (coefficient of variance, CV) was 1.9% and the between run CV was 9.8%. The 97.5th percentile in a normal range population was 0.49 nmol/l with a mean (SD) of 0.33 (0.07) nmol/l.
Patient disposition levels differed between USA and European Union (EU) sites. Patients in Europe were discharged, admitted to the general ward or admitted to a cardiac care unit (CCU)/intensive care unit (ICU). CCU/ICU in Europe were treated as one level as in many European hospitals CCUs were part of the ICUs rather than being separate units.
In the USA, CCUs or monitoring units as a care level between general ward and ICU were also available for disposition. As both, the levels as well as the distribution of patients to each level, were not comparable between EU and USA sites, the analysis was performed for both global regions separately.
All outcome prediction results reported in the current analysis are for all-cause mortality within 90 days after presentation to the ED.
Data Analysis
Values are expressed as means and SDs, medians and interquartile ranges (IQR), or counts and percentages as appropriate. As MR-proADM concentrations were non parametric, data were log-transformed before plotting. A two-sided p value of 0.05 was used for significance.
All statistical analyses were performed using R V.2.5.1 (http://www.r-project.org), SAS V.9.1 (SAS Institute), or Statistical Package for the Social Sciences (SPSS) V.16.0 (SPSS Inc).
NRI Analysis
To simulate patient disposition with MR-proADM guidance, we applied the following adapted NRI- model. We used the original disposition as a starting point, assuming that standard risk stratification was the main influence on disposition. Based on that, we used the MR-proADM concentration to change the disposition. The original disposition was changed by one level up (upgrading) or down (downgrading) only. We compared the original disposition with the hypothetical disposition by adapting the method of NRI introduced by Pencina et al.
The goal was to identify an algorithm both effective and simple for both global regions. Therefore, cut-off values were identified as follows: A MR-proADM cut-off was considered useful only if
it provided a positive NRI (ie, that outcome classification was improved by upgrading non-survivors and down-grading survivors)
it provided a positive effect on resource utilisation (ie, an increased number of patients moving onto a lower level of care)
it was a multiple of 0.5 nmol/l
and, as far as possible, identical cut-off values for both regions could be chosen.
The identified cut-offs for MR-proADM for upgrading care were ≥5.0 nmol/l in both the EU and USA. For downgrading, a MR-proADM value <0.5 nmol/l was used in USA. In the EU, the same cut-off was used for downgrading from general ward to discharge, but a higher cut-off (1.0 nmol/l) was applied for downgrading from CCU/ICU to general ward. Following the disposition strategy as suggested by the MR-proADM cut-off could result in an unfavourable disposition for individual patients.
To make sure that both survivors and non-survivors have improved reclassification, reclassification for survivors and non-survivors was applied separately, as suggested by Pencina et al.
Source...