Diagnostic test studies in nephrology: quantity, quality, and scope

Abstract

Background: Diagnostic errors represent an important cause of preventable harm in health care that may be reduced through evidence-based choice, use, and interpretation of diagnostic tests. We hypothesized that diagnostic errors are reduced through evidence-based choice, use, and interpretation of diagnostic tests.

Study design: Retrospective cohort study.

Setting & population: Diagnostic test studies.

Selection criteria for studies: Publications from 1966-2008 retrieved from MEDLINE.

Intervention: The Quality of Diagnostic Accuracy Studies (QUADAS) tool.

Outcomes: Number and coverage of diagnostic studies in nephrology and methodological quality of the test accuracy subset.

Results: Fewer diagnostic studies were published in nephrology than other areas of internal medicine, although the proportion of total citations that were diagnostic studies (4.9% ± 2.8% [SD]) was not statistically different from other specialties (P = 0.2). Within nephrology, some topic areas (eg, urinary tract infections) were over-represented, whereas others (eg, acute kidney injury) had relatively few diagnostic studies (range, 2.7%-12.5%). Examining the randomly selected subset of studies that were diagnostic test accuracy studies (120) showed variable quality. Ninety-seven percent (116 of 120) of studies adequately described index test procedure, but only 27% (32 of 120) adequately blinded investigators to results of index tests, and 36% (43 of 120), to results of reference tests. The quality of nephrology diagnostic test accuracy studies has not improved substantially during the past 30 years.

Limitations: Comparing nephrology with other specialties, some potential inequalities of scale could not be addressed, which may influence research output results across specialties.

Conclusions: Diagnostic research in nephrology is published less frequently than most other medical specialties. The quality of diagnostic test accuracy studies that are published is variable and leaves room for improvement.