Country experiences with data linkage initiatives

This session highlights the experience from two countries in data linkage: Canada and the UK. Canada's ten provinces and three territories have primary responsibility for the organization and delivery of health care services and for the supervision of health care providers. As such, many data linkage initiatives are at the provincial level, and the recent Health Data Research Network aims to facilitate access to multi- province data for researchers. Also the Social Data Linkage Environment (SDLE) represents a national effort led by the federal government (Statistics Canada). The SDLE has a centrally maintained ‘key register' containing new identifiers based on a number of identifying characteristics which are used to link a wide range of data sets, including the Labour Force Survey, national health surveys, the hospital discharge database, census data, and a range of provincial data on crime, education, mental health, etc. Thus, the SDLE is not a consolidated database per se, but a tool to link other databases, and has potential for use in scientific research. Similar to Canada, the different nations of the UK operate separate data linkage initiatives. One internationally renowned national database, the Clinical Research Practice Datalink (CPRD), reflects the important role of GPs. The CPRD brings together de-identified microdata data from GP practices in the UK on the basis of the respective electronic practice information systems. The data collected includes demographic details, information on service providers, clinical presentation, referrals, vaccinations, laboratory findings and medication prescriptions, with the ability to link to other data sets to research the care pathway. As of 2019, the CPRD data covers 42 million patients and has been used for more than 2000 scientific publications that have researched issues such as drug safety, drug use, the effectiveness of health policy measures, the organization of service delivery and the development of risk factors.

[Development of ultrasonography in a radiological university department from 1994 to 2001]

BACKGROUND

In 1994, 5 % of a total of 25 718 examinations and 7 % of all 4630 B-mode sonograms performed in the Radiology Department, University of Cologne was classified as not indicated. In light of these results, the health care policy guidelines for sonographic indications have been amended.

PURPOSE

The aim of this study was to establish the current rate of non-indicated sonographic examinations performed in routine diagnostics by radiology departments at university hospitals, to determine the reasons for such over-diagnosis and identify which regulatory mechanisms can be implemented to prevent his.

METHOD

We counted the number of 1) B-mode and 2) color-flow Doppler ultrasound imaging procedures carried out in patients who had had no change in symptoms within the previous 4 weeks or who were scheduled without reference to an existing sonogram (double examinations). 3) The reasons for over-diagnosis were analyzed. 4) The 1994 survey was repeated in 2000 with an identical protocol and 5) additionally, a modified survey of the diagnostic questions was conducted.

RESULTS

1) Out of 4,119 patients presenting for the first time to receive a B-mode sonogram, 443 prior sonograms (11 %), 305 CT scans (7 %) and 57 MRI scans (1 %) were documented. 2) Double sonograms were carried out in 6 % of the 1,118 patients presenting for the first time for color-flow Doppler ultrasounds and in 16 % of the 651 patients assigned to receive catheter angiographies with arterial color-flow Doppler. 2) 41 out of 55 (75 %) prior sonograms from non-university settings stated by 94 surgery patients were listed in the medical records. 36 out of 43 (84 %) prior sonograms from the university hospital were repeated in the same patients despite the fact that the medical report with the findings was available. None of the 48 sonograms indicated to confirm a plausible finding yielded any information that was additionally relevant to therapy. 4) In the period April - June, 2000, 12 % of all 15,921 tests and interventions, 26 % of 3,569 B-mode sonograms and 58 % of 1,033 abdominal sonograms performed in adults were classified as having not been indicated. 5) Staging and follow-up were stated as the most common reasons that a sonography was carried out in 46 % of the 1,017 adults who were given B-mode sonograms conducted from Jan - Mar, 2000 and comprised 62 % of the 410 sonograms classified as not or probably not medically indicated.

CONCLUSION

The results showed that a multidisciplinary consensus was required to establish the diagnostic value of sonographic procedures. Therefore, this research group drafted a hospital-internal interdisciplinary guideline for "abdominal transcutaneous B-mode sonography in oncological questions".