Providing the foundation for effective decision support for skin problems in primary care


Providing the foundation for effective decision support for skin problems in primary care: exploring case mix, consultation patterns, scope for improvement, and decision support needs in SystemOne.  


Using ResearchOne anonymised electronic patient records we will develop a foundation for the development, implementation and evaluation of decision support that reflects the (potentially) differing requirements of GPs and Practice Nurses. 

We wish to explore ResearchOne records for differences in case mix between, and judgement/decision performance, and the source of any differences.


Secondary analysis of >70,000 electronic patient records 


24% of the population will seek medical help for a skin disease and skin is the commonest reason to consult a GP with a new problem; an average GP will undertake 630 skin based consultations per year. ~ 50% of the primary care workforce (~16,000) are practice nurses, with 58% of them seeing between 1 and 5 patients with skin problems per week, with 20% seeing 6 or more.  ~ 4,000 deaths annually in the UK are due to skin disease. Effective diagnosis, management and referral by general practice is key to the provision of stepped, high quality, systemic provision of NHS dermatological care. Referral rates for dermatology vary between vary between 10 and 22 per 1000 population and nurses clearly have a role to play in reducing these (unwarranted) variations and enhancing the quality of primary care. 

Improving practice is not straightforward, whilst nurses often achieve the same outcomes as GPs for many conditions (but also take longer and recall patients more often) medical dermatological care (e.g. appropriate referral, accuracy of diagnosis) is itself suboptimal. Research suggests that “traditional” approaches such as upskilling nurses or training may not deliver the step change in quality that is required to improve clinical outcomes. There is considerable potential for computerised decision support (CDSS) to enhance the judgements and decisions of practice nurses, but current systems (e.g. VisualDx) are aimed primarily at doctors confronted with visual signs and symptoms in patients. The epidemiology behind the aids to differential diagnosis is unclear; there is no probabilistic reasoning behind decision support recommendations; there is no clear (research) evidence-base for the recommendations. 

This “one-size-fits all” approach to decision support (such as simply presenting differential diagnoses) is unlikely to benefit both professional groups equally. Nurses and General practitioners vary in the amount of dermatological training they receive, and key variable such as tolerance of uncertainty and risk thresholds for decisions are unknown. More fundamentally, if nurses are seeing patients with different case mix, demography and biography, then the underlying prevelance (prior probabilities of disease) will also be different from a GP’s workload. Key questions necessary for the design of a theory and evidence based decision support intervention  include:

• Do nurses see different patients to GPs (i.e. is the case mix [severity, ICD diagnosis], biographic or demographic profile different)  
• Is the workflow (e.g. appt length for skin conditions, assessment information recorded) different for nurses and GPs? 
• Do nurses and GPs differ in their adherence to national guidelines and best practice (e.g. NICE 2006)
• Do nurses and GPs differ in their Dx accuracy and appropriateness of referrals?


Carl Thompson
Steven Ersser
Louise Schreuder

Funding body: AHSN Yorkshire and Humber, £10,000 

Contact : Professor Carl Thompson