Based on questionnaires, case record forms, and focus group discussions, the overall evaluation of the GPs participating in the evaluation study regarding the recommendations made in the chest pain guideline, was positive. While the long version was partially criticized for a perceived lack of clarity, the short version and the heart score were especially appreciated. Reported change of behaviour as consequence of the guideline was inconsistent. Several suggestions to improve guideline implementation were made.
A remarkable aspect of the guideline evaluation concerns the heterogeneous feedback of the long version. As the focus group discussions revealed, some GPs criticized the layout and perceived lack of clarity of the long version, whereas they did not generally refuse the recommendations. Nevertheless, individual GPs were reluctant to accept some recommendations, despite the fact that its content (e.g., regarding the troponin test) is based on solid evidence derived from several studies. A reason may be the complex presentation since the diagnostic effectiveness of each item from the history must be discussed in relation to several outcomes. Another reason may be that GPs are used to randomized controlled trials (RCT) informing therapeutic decisions, but not a patient's history and physical signs; these are still regarded as areas for intuition.
Due to low levels of feedback from nurses and patients regarding the flowchart and patient leaflets, respectively, the benefit of these modules must be questioned. A reason for the low acceptance of the flowchart may be found in the low collaboration of nursing staff and GPs in Germany. Most German practices are small, with one to three GPs; therefore, nursing staff typically have little scope for decisions so that a patient presenting with chest pain is immediately referred to a GP. Another reason could be that an informal rule how to treat patients with chest pain is already implemented within the practices so that the flowchart seems to have no additional benefit for the nursing staff.
GPs' reasons for their low use of the patient leaflet needs to be investigated in further research and, if necessary, the leaflet's content and/or layout should be modified. Since patient interviews were performed six weeks after the index-consultation, the few patients who received a leaflet may have forgotten its use and content. Whether the patient leaflet reduced patients' anxiety in our research, like a study of Arnold and colleagues [24
] showed in the setting of a hospital emergency department, can not be definitely answered. It is possible that chest pain patients presenting to their GP are less anxious than patients presenting to the emergency department. As Jones and Mountain recommend, further research regarding the benefits of patients leaflet should be undertaken [13
According to the findings of other authors [8
], a high quality of guidelines and the agreement of GPs are no guarantors for a successful implementation of guidelines. The heterogeneous feedback concerning the perceived additional diagnostic value by the guideline and the physicians' behaviour change in consequence of the guideline knowledge reveals that agreement alone is not a sufficient precondition for a lasting implementation of the chest pain guideline [10
The attitude of the physician towards guidelines plays an important role in the decision to implement guideline recommendations. This assumption is in line with the theory of planned behaviour, where attitude, in addition to subjective norm and perceived behavioural control, is an important predictor for behavioural intention [26
]. A variety of studies have shown the theory's relevance for the medical sector [27
Some of the participating GPs reported that they don't recognize noticeable differences between the guideline recommendations and their own previous behaviour. Thus, to increase implementation success of guidelines, significant diagnostic or treatment innovations should be indicated by a well-arranged design of the guideline (e.g., desk version of the heart score), so that differences to previous behaviour become obvious. A reason for the different feedback concerning GPs behaviour change on a concrete (see case record forms (CRFs)) and more general level (see questionnaires and focus groups) might be that the CRFs were filled in directly after the consultation, so the specific behaviour was more present to the GPs, while the general feedback, in retrospect, was more prone to recall bias.
Being aware that the GPs' perception of conformity doesn't necessarily correspond to real facts, further research must be undertaken to investigate a supposed perception-reality gap. Additional recommendations on how to improve the design and evaluation of medical innovations are proposed by Murray and colleagues in their normalisation process theory (NPT) [30
]. NPT shall serve as a sensitizing tool, enabling researchers to think through issues of implementation while designing and evaluating complex interventions. By integrating interventions into routine work implementation potential is enhanced.
However, as opposed to further laboratory investigations, behaviour regarding first clinical assessment of a patient is difficult to define. Questions and expert reasoning help specify the probabilities for relevant conditions. Deviations from the standard proposed by the guideline cannot be observed within a study design of this kind, since most reasoning occurs inside GPs' thought process.
Our study was not based on a conceptual framework. Nevertheless, due to the research question and the evaluative and pragmatic character of the study, we presume the chosen method to be appropriate.
As a result of the small and non-randomized sample, the representativeness of the data may be limited. The low participation rate of GPs in the focus group discussions (10 out of 17 GPs) was most likely due to the additional effort required to visit our department after a long work day.
Social desirability may have biased focus group discussions and answer patterns from the questionnaires causing the reported evaluation of the guideline to perhaps be rated more positively than it actually was. Another limitation concerns the lacking psychometric evaluation of the used questionnaires, although we carefully considered standard format. For confidentiality reasons we could not match focus group contributions with the questionnaire answer pattern of the participating GPs.