Related Experiment Videos
[Minimum basic set of data of hospital discharges as a source of information for a study of congenital abnormalities]
N Alba Moratilla1, A M García García, F G Benavides
1Unidad de Salud Laboral, Centro de Salud Pública de Valencia. pablo.rodriguez@sanidad.m400.gva.es
Insights
Hospital discharge records accurately identify congenital defects (high specificity) but miss many cases (low sensitivity). This impacts their use in epidemiological studies.
Area of Science:
- Medical Informatics
- Public Health
- Pediatrics
Background:
- Congenital defects are a significant public health concern.
- Accurate data on congenital defects is crucial for epidemiological studies.
- Hospital discharge records are a potential data source.
Purpose of the Study:
- To validate computerized congenital defect diagnoses from hospital discharge records.
- To compare discharge diagnoses against detailed medical histories.
Main Methods:
- Random sampling of 100 children per hospital from 7 hospitals in Valencia.
- Analysis of discharges within the first year of life.
- Calculation of sensitivity, specificity, and predictive values.
Main Results:
- Records showed 64% sensitivity and 99.1% specificity for case detection.
- Diagnosis detection sensitivity was 46% with an 83% positive predictive value.
- Sensitivity varied significantly by specific diagnosis.
Conclusions:
- Hospital discharge records offer high specificity and predictive values for congenital defects.
- Low sensitivity limits their utility as a sole source for epidemiological case finding.
- Careful consideration is needed when using these records for congenital defect research.
Background:
The purpose of this study is that of assessing the validity of the computerized diagnoses of hospital discharges of congenital defects by comparing them with the information included in the medical history.
Methods:
Based on the discharge records generated over a one-year period at 7 hospitals in the Autonomous Region of Valencia, 100 children were selected at random from each hospital. As a standard, the diagnoses stated in the medical histories were indexed and coded. Solely those discharges having taken place during the first year of life were considered. A study was also made of the type, seriousness and individual or combinations of congenital defects. A calculation was made of the sensitivity, specificity, predictive values and the 95% confidence intervals thereof by the exact binomial method for the case studies (children) and the positive predictive value and sensitivity for the study of diagnoses.
Results:
126 children were detected as having congenital defects, and 201 diagnoses in medical records, and 83 children with congenital defects and 108 diagnoses on record. For the detection of cases, the records showed a 64% sensitivity, a 99.1% specificity and some positive and negative predictive values of over 90%. With regard to the detection of diagnoses, the sensitivity was 46% and the positive predictive value 83%. The sensitivity varied a great deal depending upon the diagnoses.
Conclusions:
The hospital discharge records revealed a high degree of specificity and high predictive values, but a low degree of sensitivity. These facts must be considered when these records are used as a source of cases for the epidemiological studies of congenital defects.