UK self-tests overstate accuracy and pose usability risks

By Dr. Priyom Bose, Ph.D.Reviewed by Lauren HardakerJul 28 2025

New research uncovers how popular high street self-tests for conditions like HIV, cancer, and COVID-19 may overstate their accuracy, hide critical evidence, and leave consumers vulnerable to unsafe results.

Study: Direct-to-consumer self-tests sold in the UK in 2023: cross sectional review of regulation and evidence of performance. Image credit: New Africa/Shutterstock.com

Considering the easy availability of numerous self-tests, assessing their safety and effectiveness is imperative to protect the public from the effects of poorly performing diagnostic self-tests. Researchers reviewed the available literature of self-tests commercially available on the UK high street. This review has been published in the BMJ journal.

Self-tests in the UK

In 1971, the first self-test was marketed in the UK to examine pregnancy. Since then, numerous self-tests have been developed and sold in the UK for various conditions, including severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the causal agent of the coronavirus disease 2019 (COVID-19) pandemic.

Although the National Health Service (NHS) and the National Institute for Health and Care Excellence (NICE) endorse certain self-tests (e.g., blood glucose levels and pregnancy tests), they do not recommend many due to the lack of published guidance. The accuracy claims of many self-tests are subject to debate. For instance, assessments of instructions for use (IFU) documents indicated that average sensitivity was overestimated by 12% (with some tests showing up to 45% overestimation) in some SARS-CoV-2 rapid tests.

All self-tests have false-negative (1−sensitivity) and false-positive (1−specificity) results. Test performance claims may vary due to several factors, including the study group's representativeness and design. Furthermore, the degree of sensitivity and specificity obtained in laboratory conditions may differ because experts conduct these tests, while self-tests are performed by consumers, mostly non-experts.

The point-of-care tests used in hospitals or clinical practice are typically selected and interpreted by trained healthcare personnel. In contrast, self-tests are chosen and interpreted by users who lack medical training. Hence, it is essential that the information, sampling, and test equipment are easy to handle, user-friendly, and have limited false negatives. The BMJ study also notes that predictive values, which users understand as the chance of having a condition after testing, were absent from all instructions, and one test misinterpreted sensitivity and specificity as predictive values.

About the review

The authors conducted a cross-sectional review to understand whether self-test devices available in the UK high street shops fit their specified purposes, reduce health service demands, and are safe and reliable. Only businesses with 10 or more UK outlets were selected to ensure a generalizable snapshot of nationwide availability. The review was limited to stores within a 10-mile radius of Birmingham, meaning it did not capture independent pharmacies or newer tests entering the market.

The data collected included details of the test, such as the manufacturer, biomarker characteristics, regulatory status, and costs, among others. Data were also compiled on claims regarding performance, as mentioned in the IFU, supporting evidence from the manufacturer’s technical file, and aspects of sampling, instructions, equipment, patient readability, and interpretation. The manufacturer’s technical file contained the IFU, clinical study reports (CSRs), and lay person study reports (LSRs). The authors noted that many manufacturers either refused to provide these reports, citing commercial confidentiality, or did not respond, which limited the ability to verify performance claims.

The paper also notes that many tests had been approved mainly based on analytical studies with unrepresentative or poorly described populations, a regulatory pathway that contributes directly to the gaps in reliability and transparency identified in the study.

Study findings

There were 35 self-tests identified, of which 30 were obtained from 14 manufacturers that tested for 20 different biomarkers. According to regulations, 23% of the tests could be deemed moderate or high risk. The IFU documents for 24 out of 30 tests contained accuracy claims: 19 included an “accuracy” or correlation value, 16 gave a specificity value, and 17 showed a sensitivity value.

A more useful measure for users could be the predictive values, which were not mentioned by any test. For 12 out of 30 tests, the statistical precision of estimates of sensitivity and specificity could be deduced. Of the tests investigated, 29% based their accuracy assessments on comparisons with other rapid tests rather than gold-standard clinical reference tests, which the authors note may reduce the reliability of these performance claims.

Concerning safety, 18 out of 30 tests (60%) had at least one high-risk usability or safety concern, such as unclear instructions, challenging sampling methods, or problematic equipment, particularly for tests relating to serious conditions like HIV, cancer, and blood glucose. The BMJ study also notes that only the COVID-19 and influenza tests showed no such concerns.

Ten manufacturers and distributors shared nine sets of CSRs and LSRs, covering 12 tests. Three manufacturers refused to provide the performance documents on the grounds of commercial confidentiality. Nine CSRs and six LSRs were analyzed using the QUADAS-2 tool, which groups the risk of bias and applicability concerns under four domains: flow and timing, patient selection, reference standard, and index test.

High applicability concerns because of inappropriate study designs were raised in 58% of tests. Due to poor reporting, 73% were rated as having an unclear risk of bias. Only two (22%) of the CSRs reported participant characteristics, and half described the origin and country. About 78% of LSRs provided few participant characteristics, with 22% not describing participants. Only one menopause test that described the most demographic characteristics did not represent the intended users, raising applicability concerns. For example, 59% of women in one menopause test study were aged 20 to 40, well below the typical menopause age range of 45 to 55.

Nearly 83% of LSRs were rated as having low applicability concerns for the index test domain, and 44% of CSRs stated that a similar rapid test was used as the reference standard. For 93% of CSRs and LSRs, the flow and timing risk of bias was unclear.

Conclusions

This study underlines the need for clearer standards and enhanced regulatory oversight to ensure the reliability and safety of self-tests available in the UK market. The authors highlight that regulatory frameworks (CE marking under EU IVDR/IVDD and upcoming UKCA rules) currently allow approval based largely on laboratory studies, often with unrepresentative populations and limited transparency, a model the authors argue undermines consumer confidence and the evidentiary basis for these products.

A key limitation of the study stems from poor documentation provided by manufacturers, which made it difficult to comment on many tests. The study did not include self-tests sold only in independent pharmacies or some of the newer tests available now. Furthermore, there could be some subjectivity in some parts of the analysis, particularly relating to concerns or errors.

Download your PDF copy now!