Positive predictive value of a register-based method to identify polymyalgia rheumatica in primary and secondary care: a validation study in Denmark.

OBJECTIVE

Polymyalgia rheumatica (PMR) is diagnosed in both primary and secondary healthcare. However, general practitioners (GPs) do not report diagnostic codes to national health registers, limiting the representation of primary healthcare patients in register-based studies. We estimated the positive predictive value (PPV) of a register-based algorithm for identifying patients with PMR across healthcare sectors in Denmark.

METHOD: Potential patients were aged ≥50 years between 2003 and 2024. In primary healthcare, potential patients were identified by a first-time prescription of prednisolone/prednisone (PDN) issued by a GP and ≥375 mg PDN redeemed within 1 month. If other diagnoses, treated with PDN, were registered within 3 months of identification, patients were excluded. In secondary healthcare, potential patients were identified by a PMR diagnosis code in the Danish National Patient Registry (DNPR). Diagnoses were confirmed through chart review: in primary healthcare by GPs, and in secondary healthcare at two secondary hospitals and one tertiary hospital in Central Denmark Region (2013-2024). We estimated PPV with 95% confidence intervals (CIs) using the exact binomial method.

RESULTS

PMR diagnoses were confirmed in 62% (95% CI 54-69) of patients in primary care, and 89% (95% CI 84-93) in secondary care. Among primary care patients redeeming 1000-2000 mg PDN within 2 months following diagnosis, PPV was 84% (95% CI 76-90). Among secondary care patients redeeming PDN within 2 months of diagnosis, PPV was 91% (95% CI 86-94).

CONCLUSION

The algorithm provides a robust framework for register-based epidemiological studies of patients with PMR.