Summary
The data recorded in the National Registry of Hospitalized Patients in Denmark (Landspatient-registeret
[LPR]) are based on reports from the somatic departments of hospitals throughout the
country. For the purpose of evaluating the quality of these data, information from
LPR concerning patient diagnoses was compared with corresponding information in the
records from 924 consecutive admissions to the Frederiksborg County Hospital, Hiller0d.
The diagnoses in these records were registered and given priority based on predetermined
criteria for selection of primary and secondary diagnoses, respectively. For each
hospitalized case a total of 4 primary and 4 secondary diagnoses might be registered.
In order to show the variation in choice of diagnoses between physicians, all records
from the two medical departments and 10% from the remaining departments were revised
by two different physicians. In the medical “block” (medical departments B and F,
and the departments of physical medicine, paediatrics, and neurology), there was agreement
between the two physicians on at least one of the maximally 4 primary diagnoses which
could be registered in 92% of the cases (2-digit code). For the surgical “block” (the
departments of parenchymal surgery, orthopaedic surgery, gynaecology, and otology),
the corresponding figure was 95%. If the two coders were in agreement on at least
one primary diagnosis, this was identical with LPR’s primary diagnosis in 76% of the
cases in the medical “block” and 85% of the cases in the surgical “block” (2-digit
code). However, in the case of 15% of the medical and 7% of the surgical cases, the
primary diagnosis, on which the two coders were in agreement, was not recorded at
all by LPR (3-digit code).
It is to be emphasized that a certain variation between different physicians in their
choice of diagnoses cannot be eliminated. In conclusion, the level of agreement between
LPR and the study material found in the present study is in general satisfactory if
LPR’s function as a basis of information for policy making in the hospital sector
is taken into consideration. On the other hand, agreement (especially on medical records)
is presumably not good enough in LPR to be used for research purposes.
Key-Words
Hospital Statistics - Quality Control - Evaluation of Diagnostics - Patient Registry
