NL
| FR
| EN
Introduction
| Methodology
| Objectives
Results
| Contact
Methodology
Sampling of hospitals:
Hospitals eligible for the validation study have been randomly sampled
using two steps:
- 45 hospitals have been systematically sampled on the base of
a list of 268 participating hospital-trimesters using the period
01/01/1997 - 31/12/1999.
- For each of the 45 hospitals: random sampling of the trimester
to validate.
The initial sample size of hospitals was calculated with a relative
precision of 10%, which gave us an anticipated sensitivity of 65%
on a national basis.
Sampling of patients' charts:
A standardised methodology is used to select the charts to be validated.
The sampling method differs for the validation of nosocomial pneumonia
or nosocomial bacteraemia. All declared nosocomial PN and a random
sample of 20% of the patients' charts declared as negatives were
selected for validation (assessment of false-neg PN). Likewise,
all declared nosocomial BAC are selected. Besides, hospitals are
asked to provide a list of all records with a positive blood culture
reported on a laboratory list, together with a list of all admitted
patients on ICU during the surveillance period to be validated.
On the base of those two lists patients that comply with the case
definition criteria can be identified (assessment of false-neg Bac).
The surveyors are trained to use the definitions, described in
the protocol
in a uniform and standardised way. All trained data collectors are
members of the NSIH team of the Scientific Institute of Public Health.
They reviewed the selected charts and their evaluation-judgement
is used as 'golden standard'.
Investigators are using a blind design, they do not know to which
group a chart has been assigned in order to avoid observation bias.
Another surveyor reviews discrepant charts. After this process the
file is discussed with the clinician or the ICP to determine sources
of discrepancy (training opportunity).
(IM_031203)
|
