Seite - 164 - in Applied Interdisciplinary Theory in Health Informatics - Knowledge Base for Practitioners

healthcare professionals on a regular basis, thus enabling multiple cycles through the control loop. For instance, in the Netherlands all 32 teaching intensive care units (ICUs) and 51 non-teaching ICUs participate in the National Intensive Care Evaluation (NICE) [9], and receive biannual feedback reports on standardised mortality ratios, readmission rates, length of stay, and other quality indicators. A&F is one of the most widely-used interventions in quality improvement and implementation research. It is generally used when the patient is not present, thereby making it distinctly different from clinical decision support. We can map the components of the feedback loop (Figure 1) to elements of A&F interventions as follows. The input function (perception) consists of the feedback on clinical performance that sits at the heart of each A&F intervention, and would typically materialise through recurring, paper-based or electronic reports issued by a national audit or governing body. Feedback reports summarise the performance of individual clinicians or clinical units over a set period of time (e.g. the last 3 months) using pre-defined indicators of clinical quality, typically using a combination of graphical and numerical information (scores). Reference values may be either explicitly provided or left implicit, and different for each quality indicator. Feedback on clinical processes is often determined by reference values provided by national guidelines, while feedback on outcomes would often be determined through benchmarking between care providers. For instance, in the NICE feedback report, outcome statistics are compared to the national average and the average of a group similar sized ICUs. In A&F interventions it is commonly left to the recipient of the report (i.e. the clinician or clinical unit) to interpret the information and translate it into behaviour. For instance, a unit might decide to start a quality improvement initiative based on poor performance scores in feedback reports. If that initiative is effective, improved performance should transpire in subsequent reports and can inform the decision about whether or not to continue the programme. However, there may also be disturbances (e.g. organisational barriers) that impede actual improvements to care quality, despite the efforts of the quality improvement initiative. Over the last decade, A&F interventions have increasingly moved from using static, paper-based (or PDF) documents to interactive electronic tools. The interactive computer interface of an e-A&F intervention may allow users to filter, drill down and further explore their performance summaries. For example, NICE participants can also view these data, updated after each monthly data upload, on a website called NICE Online and perform subgroup analyses [10]. In general, if an A&F system is linked to an electronic health record database, performance summaries may be generated on demand at each point in time, thus creating more flexibility for users. A recent review of e-A&F evaluation studies [11] classified them using the theoretical domains framework [12], an integrated theoretical framework synthesised from 128 theoretical constructs from 33 theories judged most relevant to implementation questions. The review found that the domains of knowledge; motivation and goals; and ‘social influences’ were most commonly targeted by these interventions. In contrast, professional identity and emotion were never targeted. Despite the clear roots of A&F in CT, it is uncommon that CT is explicitly mentioned in studies concerning the design or evaluation of A&F interventions – but it does happen that related theories are mentioned. Colquhoun and colleagues [13] conducted a systematic review of the use of theory in randomised controlled trials of A&F interventions. They found that only 20 out of 140 studies (14%) reported use of theory in any aspect of the study design, measurement, implementation or interpretation. Only 13 studies (9%) reported the use of a theory to inform development of the W.T.GudeandN.Peek /ControlTheory toDesignandEvaluateAuditandFeedback Interventions164