Email updates

Keep up to date with the latest news and content from BMC Family Practice and BioMed Central.

Open Access Highly Accessed Research article

Assessing Asthma control in UK primary care: Use of routinely collected prospective observational consultation data to determine appropriateness of a variety of control assessment models

Gaylor Hoskins1*, Brian Williams2, Cathy Jackson3, Paul D Norman4 and Peter T Donnan1

Author Affiliations

1 Population Health Sciences, School of Medicine, University of Dundee, Mackenzie Building, Kirsty Semple Way, Dundee, DD2 4BF, Scotland, UK

2 Nursing, Midwifery & Allied Health Professional Research Unit, Iris Murdoch Building, University of Stirling, Stirling, FK9 4LA, Scotland, UK

3 School of Medicine, University of St Andrews, St Andrews, KY16 9TF, Scotland, UK

4 School of Geography, University of Leeds, Leeds LS2 9JT, England, UK

For all author emails, please log on.

BMC Family Practice 2011, 12:105  doi:10.1186/1471-2296-12-105

Published: 29 September 2011

Abstract

Background

Assessing asthma control using standardised questionnaires is recommended as good clinical practice but there is little evidence validating their use within primary care. There is however, strong empirical evidence to indicate that age, weight, gender, smoking, symptom pattern, medication use, health service resource use, geographical location, deprivation, and organisational issues, are factors strongly associated with asthma control. A good control measure is therefore one whose variation is most explained by these factors.

Method

Eight binary (Yes = poor control, No = good control) models of asthma control were constructed from a large UK primary care dataset: the Royal College of Physicians 3-Questions (RCP-3Qs); the Jones Morbidity Index; three composite measures; three single component models. Accounting for practice clustering of patients, we investigated the effects of each model for assessing control. The binary models were assessed for goodness-of-fit statistics using Pseudo R-squared and Akaikes Information Criteria (AIC), and for performance using Area Under the Receiver Operator Characteristic (AUROC). In addition, an expanded RCP-3Q control scale (0-9) was derived and assessed with linear modelling. The analysis identified which model was best explained by the independent variables and thus could be considered a good model of control assessment.

Results

1,205 practices provided information on 64,929 patients aged 13+ years. The RCP-3Q model provided the best fit statistically, with a Pseudo R-squared of 18%, and an AUROC of 0.79. By contrast, the composite model based on the GINA definition of controlled asthma had a higher AIC, an AUROC of 0.72, and only 10% variability explained. In addition, although the Peak Expiratory Flow Rate (PEFR) model had the lowest AIC, it had an AUROC of 71% and only 6% of variability explained. However, compared with the RCP-3Qs binary model, the linear RCP-3Q Total Score Model (Scale 0-9), was found to be a more robust 'tool' for assessing asthma control with a lower AIC (28,6163) and an R-squared of 33%.

Conclusion

In the absence of a gold standard for assessing asthma control in primary care, the results indicate that the RCP-3Qs is an effective control assessment tool but, for maximum effect, the expanded scoring model should be used.