Establishing how best to target resources remains a challenge within COPD as this is a heterogeneous patient group with complex needs often poorly reflected by routinely collected clinical measurements such as FEV₁.

Jones et al. created The DOSE score (dyspnoea (MRC score), obstruction (FEV₁ percentage predicted), smoking status and exacerbation number in a year) (Table 1) a validated, clinically useful measure of risk stratification in COPD which utilises data already routinely collected in Primary Care for QOF review.

• Download figure
• Open in new tab
• Download powerpoint

Abstract P20 Table 1

DOSE INDEX SCORING SYSTEM (Jones et al. AJRCCM 2009;180(12):1189–95): The DOSE Index points associated with every category of all four variables are added to build the DOSE Index score

By using a collaborative approach with informatics, statistical and clinical input we developed a database approach to calculating a DOSE score using routinely collected and coded Primary and Secondary Care data. A local NHS database holding anonymised clinical records for over one million patients was used to identify a cohort of over 13,000 patients with codes diagnostic of COPD.

Microsoft Structured Query Language Server was used to identify, cleanse and standardise the required clinical information and calculate the DOSE score, creating a series of functions that can be replicated across other database management systems.

Date of FEV₁ percentage predicted was taken as the index date for DOSE score calculation. Where only FEV₁ was recorded, a percentage of predicted FEV₁ was calculated using available height and age data.

Read codes (the routine coding system used in primary care) and ICD-10 codes were used to compile lists identifying those symptoms, diagnoses and prescriptions indicative of COPD exacerbations. These lists were applied in the year prior to the chosen FEV₁ value and, functions were written to cluster those events felt to be reflective of a single exacerbation.

Read codes reflecting MRC score and smoking status closest in time to the index FEV₁ measurement were combined with the above measurements, generating a complete score in approximately 10,000 patients. Partial scores were created for a further 1500 patients with incomplete data for the individual score components.

This approach provides a simple way for clinicians to risk stratify their COPD population without increasing their clinical workload. This gives an opportunity to identify those at highest risk of hospital admission and death and proactively allocate resources accordingly.