TOPLINE

A brief consultation with a rheumatologist within a three-step triage system correctly identified about two thirds of patients with inflammatory rheumatic and musculoskeletal disease (iRMD) and missed about one third. Patients with iRMD reached final assessment faster than those without the condition.

METHODOLOGY

• Researchers conducted a prospective study at a German rheumatology center to evaluate a three-step triage system designed to prioritize patients with unclear musculoskeletal symptoms suspicious for iRMD.
• They included referrals and used a standardized telephone interview by a trained health professional (step 1), a 10-minute consultation with a rheumatologist within 4 weeks (step 2), and a scheduled comprehensive inpatient assessment (step 3) immediately, within 4 weeks, or later.
• A total of 1180 referred adults (mean age, 51 years; 69% women) were included; 1128 were scheduled for step 3, and 980 completed the full diagnostic workup.
• Researchers compared triage judgments with the final diagnosis, recorded wait times between steps, documented patients who did not complete the pathway, and classified final diagnostic categories.
• Machine learning models were retrospectively developed to predict iRMD at steps 1 and 2, and model performance was assessed using area under the receiver operating characteristic curve (AUC-ROC).

TAKEAWAY

• Among patients who completed comprehensive assessment, iRMD was diagnosed in 32%, while 68% had noninflammatory conditions.
• The brief rheumatologist assessment (step 2) correctly identified 67.2% of patients with iRMD and correctly classified 75.4% of those without iRMD.
• The mean time from telephone interview to final assessment was 33 days for patients with iRMD compared with 55 days for those without iRMD.
• Machine learning models trained on data from step 1 and step 2 achieved AUC-ROC values of 0.73 and 0.78, respectively.

IN PRACTICE

"The triage approach evaluated in this study showed a fair performance, demonstrating that assessment of patients with iRMD can be prioritised within rheumatology referrals. Integration of ML [machine learning] may further improve performance," the authors of the study wrote.

SOURCE

The study was led by Imke Redeker, Marien Hospital Herne, Ruhr-Universität Bochum in Bochum, Germany. It was published online on September 4, 2026, in Annals of the Rheumatic Diseases.

LIMITATIONS

Researchers did not assess patient satisfaction with the triage approach. They did not calculate the costs of the triage approach.

DISCLOSURES

The study did not receive any specific funding. The authors declared having no competing interests.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.