Use of fresh-frozen plasma in 2012 at the Fondazione Ca' Granda Hospital of Milan: assessment of appropriateness using record linkage techniques applied to data routinely recorded in various hospital information systems

Abstract

Background: The Quality Unit of a research and teaching hospital in Milan assessed the increased clinical use of fresh-frozen plasma in patients treated during 2012 in order to evaluate the appropriateness of this use.

Materials and methods: For each patient in the study, a pathology profile was generated by means of record linkage techniques involving data collected through different information systems. Patients' information was combined using the patient identifier key generating pathology profiles exported to an Excel file. The profiles were reviewed by two haematologists who identified 101 potentially inappropriate treatments for which the medical records had to be reviewed manually.

Results: In 2012, 490 patients were transfused and for 473 cases the automatic record linkage provided a complete profile. The information relating to the remaining patients did not match, mainly because the patients underwent outpatient procedures for which clinical information is not automatically recorded. In the overall audit only 13 treatments were judged inappropriate.

Discussion: Our study supports the view that record linkage techniques applied to data routinely recorded in different hospital information systems could be potentially extended to support clinical audits, enabling the generation of automated patient profiles that can be easily evaluated, relegating manual checks on medical records to doubtful cases only. Moreover, the method applied in this study allows the analysis of a full set of cases instead of sample surveys, increasing the robustness of the audit results.

Figure 1

Record linkage algorithm for the generation of patients’ pathology profiles (PPP). FFP: fresh-frozen plasma; ID PAT: patient identification code; PATs: patients; UNIT: unit of fresh-frozen plasma; Data ADM: data of hospital admission; Data UNIT: data of transfusion of the fresh-frozen plasma unit; Data LAB: data of laboratory test request; Data DISCHARGE: data of discharge; Data Outpatient ID: data of outpatient treatments; H ADM: hospital admission.

Figure 2

Example of a patient’s pathology profile (PPP). ID PAZ: identification code of the patient in the study; DT LAB: date of laboratory request; ID RIC LAB: identification code of laboratory request; PT SEC: prothrombine time; PT RATIO: prothrombine time standardised ratio; APTT SEC: activated partial thromboplastin time; APTT RATIO: partial tromboplastin time standardised ratio; ID_HOSP: identification code of the hospitalisation; ADM DT: date of admission for hospitalisation; TYPE HOSP: type of hospitalisation; N BAGS: number of units transfused; DISC DT: date of discharge; DIAGN 1–6: ICD9-CM codes of diagnoses 1 to 6; DT INT1–6: dates of interventions from 1 to 6; INT 1–6: ICD9-CM codes of interventions from 1 to 6.