What if clinicians like the idea of pharmacogenetic testing, but the way results are delivered determines whether they would actually use it? This study tests that question directly with GPs and pharmacists. In 2022 to 2023, 1.18 billion prescription items were dispensed in the community in England, costing an estimated 10.4 billion pounds.

That total had increased by 14 percent over the previous decade. This growth occurred while the volume and complexity of contacts in primary care also rose considerably, increasing the pressure around everyday prescribing. The NHS Long Term Plan recognised this increasing demand and highlighted medicines optimisation as a way to use the prescribing budget more effectively in primary care.

The plan also committed to informed prescribing, reducing ineffective treatment, avoiding overprescribing, and preventing problematic polypharmacy. For pharmacogenetic guided prescribing to be used routinely, the service around it must meet the needs of healthcare professionals.

A discrete choice experiment, or DCE, is a survey-based method used to elicit choices and quantify stated preferences. Earlier DCEs examined pharmacogenetic service attributes in specific clinical scenarios, but not for primary care in the NHS. This study therefore aimed to identify the preferences of healthcare professionals likely to deliver a pharmacogenetic testing service in primary care.

The choice question asked respondents: if they had to choose one of two pharmacogenetic tests to help guide a patient’s treatment, which would they choose? The DCE used an unlabelled design, so the tests were simply Test A and Test B, and it included an opt-out representing no pharmacogenetic test.

That opt-out reflected current prescribing and allowed the study to investigate potential uptake of pharmacogenetic testing in primary care. Table one defines the attributes and levels used in the discrete choice experiment.

It includes categorical options for the scope of pharmacogenetic reporting—focused, narrow, or broad—and for returning results by post, email, a standalone web portal, or directly into the electronic health record. Continuous attributes specify effectiveness from fifty to eighty percent, adverse drug reaction risk from twenty to five percent, and turnaround time from five to twenty days, establishing the hypothetical choices participants evaluated.

The relevant population was healthcare professionals who might be responsible for pharmacogenetic guided prescribing if such a service became available. The sample frame identified general practitioners and pharmacists working in primary care or community pharmacy settings.

Pharmacists were included because their clinical practice is changing in the United Kingdom. Around one quarter of registered UK pharmacists were currently licensed to prescribe, and from 2026 all newly qualified pharmacists would be qualified independent prescribers on registration.

Community pharmacies in England were also increasingly digitally enabled, giving pharmacists access to view and update patient records within the GP system. The analysis dataset included 235 completed surveys: 121 from general practitioners and 114 from pharmacists.

The majority, 90.2 percent, had no experience of pharmacogenetic guided prescribing. The median response time in the analysis dataset was 15.2 minutes. Table three reports an uncorrelated random parameter logit model for general practitioners and pharmacists, with effects-coded coefficients, confidence intervals, and p values.

Both groups show positive coefficients for narrow data reporting and via electronic health record return, while via post has negative coefficients; the constant is also positive in both groups, consistent with choosing pharmacogenetic testing over standard care. The table quantifies how reporting format, expected effectiveness, adverse-drug-reaction risk, and turnaround time relate to predicted uptake.

Respondents consistently chose pharmacogenetic testing over standard of care, shown by the positive coefficient for the constant term in the estimated model. Both professional groups preferred pharmacogenetic data reported as a narrow panel compared with other reporting strategies.

Focused reporting was associated with a negative coefficient in both professional groups. Broad reporting increased uptake for pharmacists, but not for GPs; for GPs, broad reporting was associated with reduced uptake. Table four lays out the service profiles compared in the study.

The base-case service reports focused data post, with a fifty percent chance of effectiveness, a twenty percent chance of an ADR, and a twenty-day turnaround; the optimised service reports narrow data through the electronic health record, with a ten-day turnaround and continuous values varied between fifty-three and sixty-five percent for effectiveness and five and twenty percent for ADR.

These optimised categorical levels had the highest coefficient in the final selected model. Figure two compares predicted pharmacogenetic-testing uptake under the base-case and optimised service designs. The charts vary either the chance of medicine effectiveness or the chance of an adverse drug reaction, separately for pharmacists and GPs; the shaded region represents potential improvement from adapting the service.

This matters because the authors report that service adaptations meaningfully affect predicted uptake across these ranges, with GPs described as most sensitive to the changes. At a medicine-effectiveness chance of 53 percent, an absolute improvement of 3 percent over prescribing without pharmacogenetics, the base-case service had estimated uptake of 55.2 percent among pharmacists and 33.3 percent among GPs.

For the optimised service at the same improvement in effectiveness, predicted uptake was 96.3 percent for pharmacists and 93.3 percent for GPs. The paper describes this as a potential improvement of up to 41.1 percent for pharmacists and up to 59.9 percent for GPs from changing the service design.

DCEs measure choices and infer stated preferences, and estimates suggest they produce reliable predictions, but they are susceptible to hypothetical bias. Even if an optimised service were developed, pharmacogenetic testing might not be routinely requested among the daily demands experienced by GPs and pharmacists in primary care.

The survey may also have sampling bias, because respondents may not fully represent primary care and may have been people already interested in pharmacogenetics. The findings may not generalise to other healthcare professionals or healthcare settings, so independent replication would be required.

The central message is that interest in pharmacogenetic testing is high, but uptake depends heavily on service design—especially for GPs. Designing the information, delivery route, speed, and performance carefully could make a major difference.