DES blog - clinical feedback (#284)

Please publish DES blog containing clinical feedback of our ideas

Author: chloe-eaborn

app/diabetic-eye-screening/2025/10/des-clinical-feedback/index.md

@@ -0,0 +1,49 @@
+---
+title: Clinical Voices – what we learned from feedback on DES cohorting ideas  
+description: The clinical feedback gathered on our DES cohorting ideas
+date: 2025-10-24
+author: Chloe Eaborn
+tags:
+  - discovery
+  - cohort manager diabetic eye screening
+---
+
+To ensure our proposed ideas were clinically sound and grounded in real-world practice, we engaged clinicians across several feedback sessions. These conversations helped validate our thinking, challenge assumptions, and surface new considerations for future scope discussions. 
+
+We focused on three key areas: 
+1. Validating cohorts using a second data source 
+2. Reducing the burden of SNOMED code selection 
+3. Defining the scope of a DES or diabetes wide repository 
+
+## Key Themes from Clinical Feedback:
+### Cohort Validation 
+- Type 1 diabetes patients were seen as easier to validate – clinicians noted that insulin prescriptions are a reliable indicator, making pharmacy data a strong secondary source. 
+- Type 2 diabetes and HbA1c patients posed more challenges, with cohort accuracy issues often stemming from inconsistent coding. 
+The 9 care processes were viewed as a good source of validation, though not without limitations. 
+- Emerging data sources like Digital Health Checks and the Health Information Exchange (HIE) were highlighted as promising future assets. 
+Clinicians questioned whether a “true” secondary source exists, noting that most rely on GP coding, which can be inconsistent or incomplete. 
+
+### SNOMED Code Complexity 
+-	GPs currently use thousands of SNOMED codes, often without clear guidance. Many patients have 50+ codes, while those with only one or two are harder to assess. 
+- A definitive DES SNOMED code was seen as valuable, especially for marking attendance, outcomes, and missed appointments (DNAs). 
+- However, clinicians noted that unless GPs are financially incentivised, uptake of a new code would be limited, especially since they are paid to complete the 8 care processes, not DES-specific coding. 
+- The historic context of codes (e.g. when they were applied) was seen as important for assigning confidence levels. 
+- Concerns were raised about non-clinical staff coding and the lack of standardisation across practices, which contributes to inconsistencies. 
+
+### Repository Scope and Integration 
+-	There was strong support for a centralised DES repository, with potential to expand to broader diabetes care. 
+- Clinicians emphasised the need for better integration between DES and wider diabetes management, noting that screening currently operates in a silo. 
+- The transition away from QOF reporting reduced GP engagement, particularly around managing inactive patients. 
+- A national decision was recommended for handling non-active patients who haven’t attended their GP in over five years. 
+- The date of diagnosis was flagged as a critical data point for understanding patient history and interpreting recent results. 
+
+## Additional Insights 
+- HbA1c levels were seen as useful for identifying high-risk patients and understanding engagement. 
+- There was a call for more specific patient information, including pregnancy status, language needs, and accessibility requirements. 
+- Clinicians suggested sending GPs a confirmation link to verify diabetes status for borderline cases. 
+- The London Care Record (based on HIE) was mentioned as a model for enabling read-only access to patient data across health organisations. 
+
+## How It Informed Our Next Steps 
+- We are refining our MVP scope based on clinical insight, programme objectives, and commercial and contractual dependencies. 
+- We’re assessing the feasibility of a feedback loop that enables bi-directional data flow between DESPs and GPs, supporting better cohort accuracy and care coordination. 
+- We’re considering how clearer guidance and standardisation could support improved coding practices at the GP level, reducing variation and enhancing data quality.