In the present study, we showed that guideline adherence to European and national recommendations concerning pre-anaesthesia evaluation of adults undergoing non-cardiac surgery leaves room for improvement. Adherence rates vary widely and are better in patients with a higher ASA risk class, age > 65 years, in the absence of obesity and with higher surgical risk. A considerably high number of examinations not covered by guidelines are ordered prior to surgery. Our results are consistent with the self-assessment of German anaesthesiologists where almost 40% of respondents admitted to not having sufficient guideline knowledge (Aust et al. 2013). However, lack of adherence to the guidelines may be caused not only by differences in individual knowledge of the anaesthesia staff but also by lack of enforcement, lack of training of the disciplines involved in the perioperative process, and obstacles due to the complexity of the organisational processes (Hoorn et al. 2019).

The impact of physicians’ knowledge, the level of training and adequate resources on guideline adherence can account for some findings of this study. The easier availability of the German guidelines, for instance, may explain why anaesthesiologists in our study are more familiar with their national recommendations than with those of the ESAIC. In addition, the national guidelines are shorter and less comprehensive than the ESAIC guidelines and seem to be more focused on clinical applicability, e.g. actions that can still be performed before surgery to improve patient condition. Furthermore, the native language seems to provide easier access to the information. On the other hand, the higher adherence rates in patients of higher ASA risk classes (Hackett et al. 2015), in elderly patients, and in patients scheduled for procedures with a high risk of perioperative cardiac complications (Glance et al. 2012) suggests that anaesthesiologists attached more importance to a thorough risk evaluation when it seems to matter.

The results of our study indicate that the focus of pre-operative evaluation is very much on cardiac assessment: recommended ECGs and non-invasive stress testing are performed in most cardiac-risk patients. In contrast, metabolic diseases may often not be perceived as a risk: HbA1c values in diabetic patients or serum protein levels in malnourished patients were not determined in a single case.

Therefore, it is obviously a management task to demand care not only for cardiac high-risk patients. A good example is the blanket disregard of the recommendation for pulse oximetry despite the presence of equipment in the anaesthesiology outpatient department. It was therefore only consequent that the documentation of a pulsoxymetrically measured oxygen saturation, similar to blood pressure and pulse rate, was defined as a mandatory field of the pre-operative anaesthesia log of the department. Integrated into an appropriate interprofessional and digitalised environment, such measures can potentially help to mitigate the increased staffing and time required for guideline-compliant procedures that physicians often complain about (Bohmer et al. 2012).

Another reason for the non-compliance with guidelines may also lie in the shared responsibility for the pre-operative process (Baron et al. 2017). Therefore, it is crucial that not only anaesthesiologists but also surgeons know the respective guidelines and implement them in a coordinated manner. This starts most simply with a common understanding of the necessary time between pre-operative evaluation and elective surgery, especially for the medically necessary and sometimes elaborate pre-operative measures, and not just for legal considerations (Geldner et al. 2017). The importance of time pressure is evident in this study regarding adherence rates for examinations that are particularly time-consuming and organisationally demanding and may probably not lead to patient improvement before surgery. For example, recommendations concerning transthoracic echocardiography are followed only moderately, or others, such as carotid Doppler, pulmonary function testing and cardiac catheterisation, are followed poorly.

The traditional approach to pre-operative diagnostics has been to use standardised lists of laboratory tests and technical examinations used by clinicians and general practitioners, who often assume that a standard preoperative laboratory and ECG are mandatory for every patient. We must point out that the software used in this study does not allow us to differentiate between laboratory values determined in the hospital and those brought in by the general practitioner. Findings brought in by the general practitioner have, for sure, contributed to a certain extent to the not indicated laboratory tests. Although the individual examination only accounts for a small amount, non-indicated laboratory tests and non-indicated ECGs cause high costs in total, which have been described in former studies (Flamm et al. 2011). An additional minor factor for over-analysis is the application of automated laboratory tests, which often do not allow analysis of a single parameter (e.g. sodium/potassium) but will provide a whole set of parameters that come with it.

Another aspect that needs to be discussed is that the guideline recommendations only deal with specific preoperative issues. For example, when taking medication that can cause kidney or liver damage or blood count changes, a laboratory test may be medically justified without finding a counterpart in the guidelines for preoperative evaluation. Due to the retrospective study design, this is always a case-by-case decision for the individual patient and cannot be verified for every case.

On the other hand, it must also be mentioned that many preoperative laboratory tests recommended by the guidelines were not performed. This mainly concerns tests that are not included in the standard profiles, such as HbA1c or protein levels. It is particularly striking that not a single diabetic in our patient group had an HbA1c value documented in the records or requested by the anaesthesiologist. However, it must be said that at the time of the study, it was only a weak recommendation, which has been upgraded to a strong recommendation in the current ESC guidelines.

However, poor adherence to guidelines often results from a combination of the above reasons, which can be vividly illustrated by the disregard for obesity-related recommendations. It probably started with the fact that not every BMI > 30 kg m−2 was recognised as long as the entry of height and weight into the anaesthesia log was not mandatory and the BMI was not displayed. Furthermore, blood glucose, HbA1c and pulmonary function tests, which are frequently requested by guideline recommendations for these patients, are not part of routine lists. Finally, it is often unlikely that clinical consequences will be drawn from these findings because most of the established measures would rarely be effective in the time available until surgery, especially as these were often weak recommendations according to the GRADE system (Guyatt et al. 2011).

Our study has limitations due to the retrospective design. Data were taken from the visit logs before anaesthesia, but no study-related patient interview was conducted. As a result, under-documentation, especially of examinations already performed in the outpatient setting, but also of non-pathological findings, is likely. Such missing information may also have influenced the retrospective assessment of guideline adherence. Furthermore, and despite the agreement with similar rates of guideline adherence of 50–60% (Bohmer et al. 2012), we report from only one centre from a German university hospital.

It is a strength of the study that in our pre-anaesthesia outpatient clinic, an electronic documentation system guides doctors through the pre-anaesthetic assessment in a structured way. However, electronic decision support is not used during the standard pre-anaesthesia visit. To give sufficient time to disseminate the guidelines published between October 2014, June 2017, and February 2018 we started the survey in June 2018.

Regarding the software tool used, the implementation of guideline recommendations was performed using simple control structures such as conditional statements and branches. If the corresponding conditions are met and correctly entered into the system, we assume a correct output of the applicable guideline recommendations. This was tested using case vignettes before conducting the study. However, a further requirement for the software is the creation of a data set providing the required information in structured form. The rigorous avoidance of free text in the data input form is a possible source of bias and might have led to information loss, especially as data entry was performed manually, and thus to an inconsistent implementation of the guidelines on the part of the software.

It has to be taken into account that guideline adherence is difficult to determine, and there is no unique definition of it. Often, only the adherence rate to grade 1A recommendations is taken as a benchmark (Kentenich et al. 2023). The strength of our study is that we used a software-guided approach which enabled us to analyse every guideline recommendation. As a drawback of this approach, weaker recommendations may have biased the results towards lower adherence.