Physicians' Attitudes Towards the Advice of a Guideline-Based Decision Support System: A Case Study With OncoDoc2 in the Management of Breast Cancer Patients

From Clinfowiki
Revision as of 16:24, 16 November 2015 by Purvhi joshibhattarai (Talk | contribs)

Jump to: navigation, search

Review: Bouaud, J., Spano, J. P., Lefranc, J. P., Cojean-Zelek, I., Blaszka-Jaulerry, B., Zelek, L., . . . Séroussi, B. (2015). Physicians' Attitudes Towards the Advice of a Guideline-Based Decision Support System: A Case Study With OncoDoc2 in the Management of Breast Cancer Patients. MEDINFO 2015: EHealth-enabled Health: Proceedings of the 15th World Congress on Health and Biomedical Informatics 216, 264-269. doi: 10.3233/978-1-61499-564-7-264[1]

Introduction

Clinical Practice Guidelines ("CPGs") have been developed from principles of evidence-based medicine with the intent of addressing unwarranted variations, those variations in care due to "differences in population need, health status or patient preferences"[1]. Adherence to CPGs means uniformity of practice and therefore reduced variations. By itself, the availability of CPGs has not worked to remove unwarranted variations from medical practice, according to the authors. Some studies have shown improved adherence to CPGs by use of Clinical Decision Support Systems (CDSSs), but this is not universal, because of variations in CDSSs and practice environments.

The article notes that oncology practice is particularly prone to variation. The authors state that multidisciplinary meetings (MDMs) are conducted to improve adherence to CPGs in oncology, with CDSSs used to drive that adherence.

The study documented in this article was conducted to evaluate the use of OncoDoc2, a CPG-based CDSS used in a controlled clinical trial, by analyzing the decisions made during that study in comparison with the CPGs. The same authors had previously studied the accuracy of data entry in the same clinical trial, and their work in that study is heavily cited in this article, as they show the accuracy of data entry to be a factor in CPG compliance rates.

Materials and Methods

The guideline-based CDSS OncoDoc2

The authors describe OncoDoc2 as a CDSS based on local CPGs that can run automatically on data in the Electronic Medical Record (EMR), or interactively by users following a decision path via clicking answers to questions.

Data Collection

The authors automatically collected data from the CDSS from the abovementioned clinical trial. The data reflected the MDM participants' interaction with the system, the recommendations of the CDSS, and the MDM intervention decisions. After cases were considered by MDMs, research assistants ran the CDSS against the same patient cases to answer the decision support questions based solely on the case data.

The two types of compliance

For the purposes of the study, the authors used the CDSS interactions of the research assistants to provide a baseline of comparison for the real-world cases on the same patients. Based on these results, the study regarded those results that conformed to CPGs (based on alignment with the research assistant results) as "CGPs+" compliant, while results that aligned with the CDSS output were "CDSS+" compliant.

Quality of the CDSS Advice

If the CDSS is misused, the results may not be compliant with CPGs. Data input errors in answering the decision path questions lead to differences between the MDM interaction with the CDSS and the research assistant interaction. In these cases the CDSS recommendation does not generally comply with CPGs. The authors note that sometimes the results do align, however. Thus the authors defined three sets of result types: "Advice+", meaning the data input was accurate and the CDSS provided the CPG-aligned advice; "Advice+/-", meaning the data input was incorrect but at least one correct CDSS result was reached; and "Advice-", meaning the data input was incorrect and a wrong CDSS result was reached.

Physicians' attitudes towards the CDSS advice

Again the authors set up categories of "attitudes" of physicians towards the CDSS recommendations, by which they mean what the doctors do in relation to the advice. The categories are "Comp+", meaning the recommendation is aligned with the CPGs and is followed; "React+", meaning the MDM does not take the recommended course of action, but the action taken instead is still aligned with the CPGs; "Comp-", meaning the CDSS advice is wrong but the doctors follow it; and "React-", meaning the doctors take action not aligned with the CPGs despite the CDSS recommendation.

Consolidation of the CDSS use and physicians' attitudes

This section of the article presents a diagram of the various "Advice", "CDSS", "CPGs", and "Comp"/"React" categories represented as a decision tree, showing the hierarchical relationships between the categories.

Results

The decision tree model described above allowed the authors to assign proportional values to the various tree nodes. They determined that the CDSS was used incorrectly in more than half the cases, but that it often yielded the correct result regardless: 71% of cases had the CDSS recommend CPG-compliant action despite the high rate of incorrect data entry. The balance of correct decisions (16% of total cases) were made by the MDM panel contrary to the CDSS' incorrect recommendation. 4.6% of cases had decisions compliant with CDSS recommendation, but not compliant with CPG — this is regarded by the authors as "automation bias", the idea that the decision was placed with more confidence though wrong because of the guidance of the computer. 8.6% had decisions noncompliant with CDSS and also incorrect per CPG.

Discussion

The authors conclude that "the correct reminder of CPGs by a CDSS may foster their adoption"[1], which is to say that CDSS can be useful for increasing the adoption of CPGs. They note the possibility of unintended consequences of CDSSs, particulary in cases where incorrect data is collected or entered, and that usability of the CDSS might also come into play.

In the cases where physicians "reacted" to the CDSS (made a different choice than recommended), positive reactance was regarded as a sign the doctors were familiar with the CPGs and recognized the CDSS advice as faulty, while cases of negative reactance were more ambivalently regarded in that "CPGs are not known, or are not trusted, or are not appropriate to the patient"[1]. In the latter instance, the authors state, it could be that in certain cases CPGs are not applicable and physicians' judgement is more important.

Conclusion

The authors conclude that CDSS can be useful for encouraging the adoption of CPGs, but that some areas such as data quality, user interface, and physician training could use improvement.

Related articles

References

  1. 1.0 1.1 1.2 1.3 Bouaud, J., Spano, J. P., Lefranc, J. P., Cojean-Zelek, I., Blaszka-Jaulerry, B., Zelek, L., . . . Séroussi, B. (2015). Physicians' Attitudes Towards the Advice of a Guideline-Based Decision Support System: A Case Study With OncoDoc2 in the Management of Breast Cancer Patients. MEDINFO 2015: EHealth-enabled Health: Proceedings of the 15th World Congress on Health and Biomedical Informatics 216, 264-269. doi: 10.3233/978-1-61499-564-7-264