a1 Institute of Health and Society, Newcastle University, Newcastle upon Tyne, UK
a2 Sunderland City Council, Sunderland, UK
a3 Leeds Institute of Health Sciences, University of Leeds, Leeds, UK
a4 Marsden Road Health Centre, South Shields, UK
Background Data on the uptake of clinical guidelines into practice are essential to guide and evaluate quality improvement interventions. Organizations responsible for service specification, monitoring and improvement need to consider the practicality of and trade-offs made in different data collection methods. We examined the feasibility of deriving and applying review criteria for clinical guidelines in English primary care.
Methods We selected two sets of guidance, on osteoporosis and depression, and used a consensus process to derive review criteria. We manually extracted data on adherence to review criteria from patient records in 20 general practices from three NHS primary care trusts in northern England. We compared the relative utility of extracted data with that of routinely available data, summarizing feasibility using what we termed a Resource Ratio.
Results Of 53 proposed review criteria we assessed, 41 were judged clinically important, valid, relevant and measurable. Thirty-one could be assessed in 10% or more of sampled patients, whereas 15 could be readily extracted (resource ratio of 15 or less). Only eight met all desirable attributes for use as review criteria. Resource ratios correlated poorly with local stakeholders’ prior views on feasibility of data collection. We observed wide variations in compliance with review criteria, with notably low levels among self-care standards.
Conclusions A minority of guideline recommendations were suitable for review criteria development, fewer still when using routinely available data. Local stakeholders tend to underestimate the actual resource requirements of data collection. Although improved design and use of clinical records may facilitate measurement of adherence to recommended practice, detailed assessments are still likely to rely upon some degree of manual data collection in the foreseeable future.
(Received July 09 2013)
(Revised December 18 2013)
(Accepted January 05 2014)
(Online publication February 14 2014)