Objective:

To examine and adjust for misclassification in disease prevalence estimates by linking primary care records with text-mined outpatient letters, focusing on psoriatic arthritis (PsA) and its implications for accurate health data.

Key Findings:

• Observed prevalence of PsA in primary care was 0.13% (95% CI, 0.11%-0.15%), indicating a significant underestimation.
• Primary care codes identified 188 true PsA cases but missed 196 hospital-diagnosed cases, leading to over 2-fold underestimation, highlighting the need for improved coding.
• Adjusted prevalence accounting for misclassification was 0.25% (95% CI, 0.21%-0.28%), demonstrating the effectiveness of the integration approach.

Interpretation:

Linking primary care with hospital records effectively identified both false positives and negatives, enabling more accurate prevalence estimates for PsA.

Limitations:

• The study may not be generalizable beyond the specific region and population studied, which could limit the applicability of findings.
• Reliance on text-mining may introduce variability in the accuracy of extracted data, potentially affecting the reliability of results.

Conclusion:

Integrating primary and secondary care data enhances the accuracy of disease classification and prevalence assessments, underscoring the importance of addressing misclassification in health data for future research and policy.