Guidelines Are Only as Good as Their Implementation: Closing the Diabetes Care Gap with GP Data

I’m fascinated by what data can tell us about performance, particularly when we compare care delivery data to assumed performance in the midst of busy clinics and stretched teams. For the health system, guidelines are written, what ‘good’ looks like is described, and then patients walk in through some 6,000 practices across England receiving care that varies far more than we’d like. As a clinical pharmacist who spent four years in one of the most deprived primary care networks in England before moving to clinical and population health management consultancy across almost 90 GP practices for two years, it’s fair to say that I’ve seen a lot first hand. Be it the guidelines clinicians are following, the different patient profiles that present or even just the reality of front line care in a consultation room.

We then often speak about health inequalities measured by the outcomes and the populations that are affected, though we often do not move upstream and question why equitable guideline implementation is failing in the first place.

That’s partly my motivation behind my re-entry to academia, especially as a proud Bradfordian who grew up seeing the impact in communities locally and further afield. I’m now studying a DPhil (PhD) at the University of Oxford, based at the Bennett Institute for Applied Data Science. My research looks at the equitable implementation of clinical guidelines in primary care, using diabetes as the clinical lens within applied data science methodology. In practice, it means that I will be contributing to improving diabetes care using 59 million electronic health records, with near complete coverage of England’s population.  

This blog explains what I’m working on, why it should matter and how the data many of you already contribute could help close the gap between current care and best practice.

Key Takeaways

• Diabetes care in England varies widely between practices and populations, and that variation is poorly understood at scale and in near real-time.
• OpenSAFELY now allows secure, transparent analysis of near whole-population NHS GP data for non-COVID research.
• My research aims to quantify the gap between current clinical performance and what good looks like, starting with guideline implementation and SGLT2 inhibitor prescribing.
• With this insight, it aims to generate evidence that can inform national strategy, reduce inequalities, and reduce avoidable downstream complications.

The Challenge is Unwarranted Variation

Diabetes prevalence in England is on the rise. Around 4.1 million people are now on the diabetes register equating to roughly 7.9% of the adult population, consuming around 10% of the NHS’ £200 billion budget. The pressure on the health system is only on the rise, especially with demand on primary care to prevent avoidable complications by delivering consistent, high-quality care. In order to meet the demand despite the pressure, it’s important we better understand variation of care, particularly, how much variation is clinically justified and how much of it is unwarranted.

Where guidelines meet the real world

Clinical guidelines are only as good as their adoption and implementation. And often, clinicians can dismiss them as just… ‘guidelines’, ‘guardrails’ and ‘suggestions’. NICE can recommend a therapy, the evidence can be excellent, and yet uptake on the ground can be slow, patchy and unequal.

A great example of this is medicines optimisation and the use of SGLT2 inhibitors. They are a class of medicines that have strong evidence for glycaemic control alongside cardio-renal protection, and the guidelines are reflective of the benefit.  However, are they reaching the right patients, at the right time? And similarly, are they de-prescribed when needed?

In answering these questions, data insight far beyond a single practice or primary care network is useful. Though before moving into population-scale health data analysis, my first focus is on the literature. I’m currently studying SGLT2is as a focus example within a scoping review on the implementation of clinical guidelines in primary care as part of my initial literature research. It will sit within a wider review of how guideline implementation actually works, what helps, what stops it and where the inequalities begin to develop. It will form a foundation for the quantitative applied data science work that will follow and shape the research outputs I am to produce as a result.

OpenSAFELY: The Go To Solution?

OpenSAFELY is a secure data analytics platform developed at the Bennett Institute for Applied Data Science, enabled by NHS England. It was originally limited to covid related research across England’s electronic health records, supporting more than 100 published outputs.

As of early 2026, it opened to non-COVID related research projects, facilitating my desired PhD research. Subject to project approval, I intend to study first and second line type 2 diabetes management to begin with. I’ll then run analyses that quantify, at population scale and in near real-time, the gap between current clinical practice and what best practice would look like in diabetes care.

For GP practices and clinicians, a few points are worth noting:

• The data never leaves the secure data environment – analysis runs inside the secure systems of the GP IT suppliers (TPP and EMIS/Optum).
• It’s transparent – every line of analysis code is published openly, and a public log records what was run and by whom.
• It’s whole-population -the platform covers >95% of people registered with a GP in England, the records of around 59 million people.
• It has the support of the BMA, RCGP and privacy campaigners.

Supporting Better Medicines Optimisation in Primary Care

At The Medicines Management Team (TMMT), we share a commitment to reducing unwarranted variation in care and improving outcomes through evidence-based medicines optimisation. Our clinical pharmacists, pharmacy technicians and advanced clinical practitioners work alongside GP practices and Primary Care Networks across England to support safe, effective and equitable prescribing, helping teams translate national guidance into meaningful patient outcomes.

As research continues to highlight the challenges of guideline implementation and healthcare inequalities, organisations such as TMMT play an important role in providing the clinical capacity, expertise and operational support needed to deliver high-quality medicines management services at scale.

To learn more about how TMMT supports GP practices and PCNs, visit https://medicinesteam.co.uk or contact the team at info@medicinesteam.co.uk.

Conclusion

Having worked in some of the most deprived communities in England, I’ve seen first hand how variation in care can affect people’s lives. Frontline primary care teams are continually adopting evidence based, guideline led care whilst balancing workforce challenges, increasing demand and system pressures.

My research is ultimately about understanding how clinical guidelines move from paper to practice. By combining implementation science with population-level health data, the goal is to reduce health inequalities by producing robust data, evidence and recommendations to inform policy and national strategy. This is so that the next guideline update, incentive or commissioning decision is shaped by what’s actually happening, narrowing the gaps we see today.

As the research progresses, I’ll be sharing my findings. I welcome thoughts and opinions on this topic and ultimately, the goal is collective benefit.