What question did this study set out to answer?

To compare the accuracy of risk-adjusted hospital mortality predictions using administrative data alone, administrative plus laboratory data, and administrative plus full clinical data.

March 1, 1997Annals of Internal Medicine

Predictions of Hospital Mortality Rates: A Comparison of Data Sources

View Full Paper

Ask AI

Bookmark

Why the study?

Does adding laboratory data to administrative data improve the accuracy of predicting inpatient death in hospitalized patients compared to administrative data alone?

Population

46,769 patients hospitalized with acute myocardial infarction, cerebrovascular accident, congestive heart…

Comparison

Mortality prediction models using restricted… vs Mortality prediction models using administrative…

Design

Cohort

Follow-up

Inpatient stay

Authors

MPMichael PineBoston University

Discussion

Member takes

Implication

Laboratory-augmented administrative models may enable accurate mortality prediction without chart review; leaves open their validity for unbiased hospital comparisons.

Key Points

To compare the accuracy of risk-adjusted hospital mortality predictions using administrative data alone, administrative plus laboratory data, and administrative plus full clinical data.
Retrospective cohort study analyzing 46,769 patients hospitalized across 30 acute care hospitals for acute myocardial infarction, cerebrovascular accident, congestive heart failure, or pneumonia.
Predicted inpatient mortality probability across four models: unrestricted administrative, restricted administrative (excluding potential hospital-acquired complications), laboratory (restricted administrative plus lab data), and full clinical models.
Average areas under the receiver-operating characteristic (ROC) curves were 0.87 for unrestricted administrative models, 0.75 for restricted administrative models, 0.86 for laboratory models, and 0.87 for clinical models.
Laboratory models ranked 93% of patients and 95% of hospitals similarly to comprehensive clinical models, outperforming unrestricted administrative models (75% and 69%) and restricted administrative models (72% and 77%).

Structured PICO

Does adding laboratory data to administrative data improve the accuracy of predicting inpatient death in hospitalized patients compared to administrative data alone?

Population

46,769 patients hospitalized with acute myocardial infarction, cerebrovascular accident, congestive heart failure, or pneumonia across 30 acute care hospitals

Intervention

Mortality prediction models using restricted administrative data plus laboratory data (laboratory models) or restricted administrative data plus laboratory and abstracted clinical data (clinical models)

Comparator

Mortality prediction models using administrative data only (unrestricted administrative models) or administrative data restricted to secondary diagnoses unlikely to be hospital-acquired complications (restricted administrative models)

Outcome

Prediction of inpatient death (hospital mortality rates)hard clinical

Adding electronically available laboratory data to restricted administrative data provides accurate predictions of inpatient mortality, avoiding the cost of manual chart abstraction and the biases of using unrestricted administrative data.

Cite This Study

Michael Pine (1997) studied this question.

synapsesocial.com/papers/6a7fe8eaeaf37ab6fb502d81 https://doi.org/10.7326/0003-4819-126-5-199703010-00002

View Full Paper

Ask AI

Bookmark

Implication

Laboratory-augmented administrative models may enable accurate mortality prediction without chart review; leaves open their validity for unbiased hospital comparisons.

Predictions of Hospital Mortality Rates: A Comparison of Data Sources

Discussion

Member takes

Implication

Key Points

Structured PICO

Cite This Study

Predictions of Hospital Mortality Rates: A Comparison of Data Sources

Discussion

Member takes

Implication

Key Points

Structured PICO

Cite This Study

Also Consider

Also Consider