Design: VALHUDES protocol diagnostic accuracy study in a colposcopy population
Population: 400 colposcopy attenders with self-collected vaginal and practitioner cervical samples
Primary outcome: Relative sensitivity and specificity for CIN2+
Effect: Relative sensitivity 0.982 and 1.037 (non-inferior); relative specificity lower

An Australian study used the international VALHUDES protocol to evaluate self-collection. Four hundred people attending colposcopy took their own vaginal swab, and a practitioner then took a cervical sample. Both were tested on two PCR-based HPV assays.

For biopsy-confirmed CIN2 or worse (58 cases), the self-collected swabs were non-inferior to practitioner samples, with relative sensitivities of 0.98 and 1.04 on the two assays. Relative specificity was lower with self-collection, meaning more HPV-positive results without disease.

A colposcopy population is enriched for disease, so this is the standard first step for validating a test–device combination rather than a screening-population study. It supports self-collection with these validated PCR assays, with the laboratory planning for a somewhat higher positive rate. For India, where HPV-based screening is expanding and access to a speculum examination is a real barrier, validated self-sampling could widen coverage. It was published in June 2026.

Use only PCR HPV assays clinically validated for self-collected samples.
Expect a somewhat higher HPV-positive rate with self-samples because of lower specificity.
Report sample type on the request and result so triage can take it into account.
Validate each test–swab combination locally before offering self-collection.

Why it matters

It removes the clinician examination from the first step of cervical screening without losing sensitivity.

The statistics, in plain English

A relative sensitivity of 0.98 means self-samples detected about 98% as many CIN2+ cases as clinician samples; 1.04 means slightly more. Non-inferiority means the difference fell within a pre-set margin. With 58 cases of disease, estimates have some uncertainty.