Agreement between general practitioners and central reviewers on spirometry interpretation in primary care
Italian Journal of Medicine · 2026-10-23 · 20 (4)
DOI: 10.4081/itjm.2026.2657 ↗Abstract
We assessed agreement between general practitioners (GPs) and central reviewers within a supported primary-care spirometry program in Piedmont, Italy. GP and central-review labels were recorded separately; reviewer blinding was not documented. Among 658 uploads dated 18 May 2022-18 January 2023, 481 from 25 GPs had paired obstruction labels. Confidence intervals (CIs) used 50,000 GP-cluster bootstrap resamples. The obstruction agreement was 92.9% (95% CI 90.3-96.0), and Cohen’s κ was 0.855 (0.797-0.918); discordance comprised 28 GP-positive/reviewer-negative and 6 GP-negative/reviewer-positive records. Relative sensitivity was 96.8% and specificity 90.4%. Agreement was 95.0% for possible restriction (κ 0.818), 97.5% for recorded reversibility (κ 0.737), and 86.2% for severity (weighted κ 0.789). Four Global Initiative for Chronic Obstructive Lung Disease categories were available, but no very severe classifications were recorded. One reviewer account contributed 397/481 assessments. Incomplete pairs, the concentration of reviews, possible repeat examinations, and the limited number of GP clusters constrain interpretation. Technical-quality data and the bronchodilator-tested denominator were unavailable. High agreement describes completed pairs within this supported pathway and does not establish diagnostic accuracy or clinical benefit. Hospital-community applications require prospective evaluation.
EUVIMED