Format

Send to

Choose Destination
Epidemiol Prev. 2013 Jul-Oct;37(4-5):289-96.

[Information on educational level from hospital discharge register: an analysis of validity].

[Article in Italian]

Author information

1
Dipartimento di epidemiologia, Servizio sanitario regionale, Regione Lazio m.ventura@deplazio.it.

Abstract

OBJECTIVE:

to analyze the validity of information on educational level from Hospital discharge register, by comparison with the 2001 Census; to develop a hierarchical algorithm which allows to maximize its validity.

METHODS:

Hospital Information System (HIS) of Lazio Region and 2001 Census were used to select all the hospital admissions between 2000 and 2009 of at least 35-year-old people living in Rome and registered at 2001 Census. For each hospitalisation, the information on education stated on Census was associated. Agreement with Census was measured using Cohen's kappa. A hierarchical algorithm based on the results of agreement analysis was developed. For each patient, it selects the most valid admission in order to find the most accurate information on education. The application of the algorithm was tested on four cohorts of 2008-2009 hospital admissions.

RESULTS:

a good agreement on education from HIS and Census (k between 0.5 and 0.6) was found. The information on education was better for planned hospitalisations placed in hospitals with a volume of care within the 12,000 admissions per year. The agreement between HIS and Census in hip fractures and acute myocardial infarction cohorts was considered sufficient (k=0.3), while it was found a good/excellent agreement in cholecystectomy and coronary artery bypass (k=0.6 for both conditions) cohorts. The application of the algorithm to the cohorts of acute care hospitalisations allowed moving to a level of good agreement (increase: 19%). The gain was much lower in the elective admission cohorts (4%).

CONCLUSIONS:

the overall agreement is good, but it depends on the characteristics of the hospital admission. However, these differences may be reduced by using hierarchical algorithm.

PMID:
24293494
[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for Inferenze scarl
Loading ...
Support Center