Metadvice

Context

Metadvice is an advanced AI company that combines the power of Neural Networks and Machine Learning with advancements in medical science with the goal of helping clinicians manage patients better and faster, enhance preventative care, and ensure improved medical outcomes.

Their AI-powered software platform works by identifying at-risk populations, then, bringing together the patient’s medical records alongside the latest clinical guidelines, in order to recommend the next step in the treatment pathway. This aims to ensure clinical workflows and decision-making are optimised at a population and individual patient level.

Unity Insights, commissioned by Health Innovation Kent Surrey Sussex, conducted a mixed-methods evaluation of Metadvice at Ball Tree Surgery in Lancing and Sompting PCN.

Method

The evaluation considered the acceptability and usage rates of Metadvice amongst clinicians, the effectiveness of Metadvice in supporting clinicians to make more impactful treatment decisions within the realm of cardiometabolic disease, thoroughness of implementation schematics, monetary benefits stemming from reduced appointment times, and impacts on health inequalities.

Qualitative insights

Interview and survey data suggests that considerable thought was placed into the long-term establishment of Metadvice at Ball Tree Surgery, with extensive documentation accompanying all necessary legal, bureaucratic, and human elements of implementation. Some additional considerations could have been made to provide staff with more bespoke training sessions in order to improve the quality of education delivered and subsequently maximise usage of the tool.

Quantitative insights

Metadvice was demonstrated to result in improved reductions in blood pressure and non-HDL over a usual care pathway:

• Blood pressure: a reduction of 2.55/0.014mmHg amongst patients recommended any medication change. When only considering antihypertensive medication change recommendations, this changes to a reduction of 2.72 in systolic blood pressure, and an increase of 0.44 in diastolic blood pressure.
• Non-HDL: a reduction of 0.046mmol/L. Non-HDL was reduced by 40% or more in 14.3% of patients eligible for analysis with 0 lifetime CVD events.

Due to the ability of Metadvice to be used as a screening tool through automatic calculation of QRISK and SMART2 scores enabling more effective risk stratification of patients, the tool allowed clinicians to prioritise outreach to patients of higher risk levels than in a usual care pathway. As such, the above benefits in patient outcomes are realised in patients who may have otherwise not received medical attention and treatment optimisation.

Health economic insights

The lead pharmacist at Ball Tree Surgery estimated that Metadvice is responsible for a 30% saving in clinical time. After applying optimism bias corrections to this measure, it was estimated that Metadvice could potentially lead to a saving of £4.71 per CVD-related GP appointment, totalling just under £60k, and a saving of £1.75 per pharmacist appointment, totalling just over £1.4k.

Limitations

• Quantitative data limitations affected the possibility of attributing greater effect sizes to Metadvice.
• Qualitative data collection was conducted by Metadvice representatives, presenting the possibility of bias in responses.
• Health economic benefits are theoretical and relied on qualitative judgements as opposed to extant data. No value judgements on Metadvice could be created due to the absence of cost modelling.
• Sample sizes of ethnic minority groups were too small to conduct intricate analyses on health inequalities

Recommendations

• Developers of Metadvice should continue to expand data access and summary features for clinicians.
• Clinician training on Metadvice use should be delivered as early and comprehensively as is feasible in order to maximise use of the tool.
• Data collection for further evaluations should be expanded to include recommendations for lifestyle changes, whether clinicians enacted recommendations provided by Metadvice, clinician usage patterns, and more complete data on cardiovascular event risk.
• When considering further implementation sites for Metadvice, consideration should be given to areas of greater ethnic minority representation to facilitate more detailed analysis on health inequalities.