This evaluation describes how SAPG has coordinated a hospital wide PPS and built on this to implement regular measurement and feedback of nationally agreed prescribing indicators that have driven improvements in antimicrobial prescribing in Scottish acute hospitals.
SAPG coordinated the participation of 56% of acute hospitals in Scotland in the ESAC PPS 2009 (baseline PPS). This level of participation was due to a combination of national leadership [from SAPG] and local engagement from AMTs and clinicians. In the follow up PPS data were collected in all acute hospitals in Scotland following a Scottish Government instruction to all NHS boards to participate.
Previous data from the 2006 ESAC PPS[17], and a small Scottish survey in 2007[18], had indicated compliance with prescribing policy and antimicrobial use for surgical prophylaxis were areas of variable practice in Scotland. The baseline PPS data proved valuable in identifying key measures intended to drive improvement in the quality of antimicrobial use in hospitals; poor documentation of indication for treatment and compliance with local prescribing policy remained features of poor practice in empirical prescribing. Furthermore prolonged duration of surgical prophylaxis was also confirmed as an area where improvement was required.
Documenting the reason why an antimicrobial has been prescribed in medical notes is recommended in Scotland as essential for good clinical practice[4]. It ensures communication of diagnosis between clinical teams and supports review of treatment. It was therefore disappointing that no indication was documented in 24.1% cases, and although similar to Europe, SAPG regarded this level as unacceptable and an important target for improvement. All AMTs have produced guidelines on empirical treatment of commonly encountered infections based on advice from SAPG issued in 2008[19]. The baseline PPS revealed compliance with local policy of 81.0% (excluding cases where compliance was not assessable or where no information was available) confirming the need for improvement. Although the proportion of surgical prophylaxis with a duration of <24hr was greater than the European average, in 31.4% of cases it exceeded 24hr in the baseline PPS. The Scottish Intercollegiate Guideline Network (SIGN) guideline on antibiotic prophylaxis in surgery recommended a single dose of an appropriate antimicrobial is required for the majority of surgical procedures[13]. Unnecessarily prolonged surgical prophylaxis contributes to selection pressure for antimicrobial resistance[20] and associated risks such as Clostridium difficile infection (CDI)[21].
Disadvantages of hospital wide PPS include the labour required to collect and input data, their infrequent nature and the requirement for centralised data input resulting in a delay in reports being available for participating hospitals. When available the results of the baseline PPS were fed back to clinical teams by AMTs to encourage better local prescribing although their impact was probably reduced as the data were not recent. To overcome this and provide a further stimulus SAPG introduced regular systematic measurement of quality using national prescribing indicators in key clinical areas to promote improvement. The indicators chosen underpinned a national target for reduction in CDI[10]. By example compliance with local antimicrobial policy promoted agents less likely to lead to CDI and reducing excessively prolonged surgical prophylaxis would also reduce CDI[20][21]. We believe linking the national prescribing indicators to a key clinical area of concern combined with the support of Government to include the indicators as an integral component of national initiatives for patient safety and quality improvement was pivotal to their successful introduction and adoption by clinicians[22].
Establishing a culture of measurement and clinician feedback is an effective stewardship strategy[23], and using this approach the hospital empiric prescribing indicator has been successful in providing early evidence of improvement in medical and surgical admission units although this yet not consistent and reliable across all NHS boards. Our intervention was audit and feedback and three recent meta-analyses have used behavioural theories to synthesise evidence from audit and feedback studies in order to identify intervention components that may enhance effectiveness. One review used Feedback Intervention Theory[24] and two used Control Theory[25]. All three reviews suggested that the effectiveness of audit and feedback is enhanced by setting a target or behavioural goal, which was a component of our intervention. In addition effectiveness was improved in the Feedback Intervention Theory by providing specific, frequent and written suggestions for improvement[24]. One Control Theory review found that insufficient studies reported on use of action plans to allow reliable statistical analysis[25] However, a larger and more recent review indicated that feedback may be more effective when baseline performance is low, the source is a supervisor or colleague, it is provided more than once, it is delivered in both verbal and written formats, and when it includes both explicit targets and an action plan[26], results that are consistent with Control Theory[25]. We asked all AMTs to collect up to five examples of non-compliance per month and used this information both locally and nationally to identify common themes that could help shape national educational solutions. However, we did not attempt to standardize the way that this information was fed back to clinical teams (e.g. in writing versus verbally) or to what extent the information was used for action planning. We will consider the feasibility of more explicit application of these theoretical frameworks to future interventions and to understanding variation in the success of our current interventions.
We suggest the reported difference between the baseline PPS and the subsequent prescribing indicators may be due in part to the PPS evaluating prescribing across the whole hospital rather than only in medical and surgical admission units where the prescribing indicators were applied. Consequently there may have been a greater emphasis on improvement of prescribing in admission units compared with other parts of hospitals. Indeed, improving the quality of prescribing and antibiotic review in continuing care inpatient wards has been identified as an area of priority for SAPG in the future. In the follow up PPS there were higher levels of indication documented and policy compliance than observed in the baseline survey. The increase is welcome but remains below the targets used in the national prescribing indicators and confirms the need to introduce improvement initiatives in other inpatient departments.
For sustainable clinical engagement we believe it is important that national prescribing indicators are open to review and seen as drivers for improving clinical outcomes as opposed to being viewed as either punitive or restrictive measures. Our review in March 2011 showed that although compliance with the hospital empiric prescribing indicator measures had improved AMTs indicated that monitoring and reporting acute medical and surgical admissions units separately would provide greater clarity. Data following the review illustrated a lower proportion of indication documented and policy compliance in surgical units and led to AMTs targeting improvement activity more closely with clinicians in surgical units.
The initial prescribing indicator for surgical prophylaxis was less successful in providing a consistent and homogenous national dataset as AMTs could select which surgical procedures to include. The review of the surgical prophylaxis indicator to focus on colorectal surgery from April 2011 has achieved a consistency of approach not possible when there was local discretion over which procedures to include. Nationally aggregated data for 12 months following the modification illustrate the proportion receiving single dose prophylaxis has exceeded the target of 95% and the policy complaint proportion has increased but remains below the target indicating that further improvement work is required. Although data for the initial prescribing indicator were collected in a number of surgical procedures the proportion with duration <24hr was much higher at the start of data collection than observed in the baseline PPS. It is possible that surgeons improved their practice after the baseline PPS but it is more plausible that the difference is due to the way doses for surgical prophylaxis were recorded and captured. In the baseline PPS only standard prescription charts were used to identify doses for prophylaxis. However, in routine clinical practice in Scotland prophylactic single doses for prophylaxis may be prescribed in the once only section of standard prescription charts, on theatre record sheets or on the fluid prescription chart. If only standard prescription charts are reviewed, then single doses prescribed on other records will not be captured with a resultant overestimation of the proportion of prolonged prophylaxis. This may require to be considered in reviewing the methodology used in the PPS. The follow-up PPS used all case notes, nursing notes and theatre records. In the follow up PPS there was a higher proportion of surgical prophylaxis with a duration of <24hr than in the baseline survey but it remains lower than observed in the national prescribing indicator dataset. These data indicate that improvement may still be required in procedures other than colorectal surgery.
We believe involvement of the prescribing and clinical community in collecting and feeding back these data was important. The improvement in the national prescribing indicators has only been possible through engagement with AMTs and the clinical community in Scotland. This has been challenging with surgeons, where discussions have centered on the choice of prophylaxis regimen. The move away from cephalosporins to narrower spectrum beta-lactams with or without aminoglycoside has met with some resistance, partly because the evidence based to support these regimens against traditional agents such as cefuroxime was not available. As part of ongoing dialogue with the surgical community, and other clinicians, SAPG has committed to measure unintended consequences of changes in prescribing policies including aminoglycoside and flucloxacillin related renal toxicity, increased surgical site infection and mortality. SAPG believe such balancing measures are critical in reassuring clinicians of the safety and effectiveness of our interventions.
There are some limitations to the methods used. In the PPS data were collected across hospitals with different specialties and case mix. This may influence the prevalence of antimicrobial use but quality measures such as recording in notes, compliance with policy and duration of surgical prophylaxis should be less affected. As numerous individuals collected baseline and prescribing indicator data the extent of inter-rater variation or observer bias is unknown. Although in the follow-up PPS inter-rater reliability was tested and found to indicate an excellent level of agreement between data collectors[27]. The improvements in the prescribing indicator data have not been statistically assessed. The sample size for prescribing indicators as national targets may have influenced the results i.e. with a recommended sample of 20 patients per month, to achieve the target of ≥95% compliance a score of ≥19/20 is required. When sample size was <20 patients adherence to policy and documenting indication had to be perfect to achieve the target.