OncoDoc: a successful experiment of computer-supported guideline development and implementation in the treatment of breast cancer

Abstract

Originally published as textual documents, clinical practice guidelines have poorly penetrated medical practice because their editorial properties do not allow the reader to easily solve, at the point of care, a given medical problem. However, despite the proliferation of implemented clinical practice guidelines as decision support systems providing an easy access to patient-centered information, there is still little evidence of high physician compliance to guidelines recommendations. Apart from physicians’ psychological reluctance, the incompleteness of guideline knowledge and the impreciseness of the terms used, another reason may be that, although suited to average patients, clinical practice guideline recommendations are not a substitute for the physician-controlled clinical judgement that should be applied to each actual individual patient. Therefore, computer-based approaches based on the automation of context-free operationalization of guideline knowledge, although providing uniform optimal strategies to problem-focused care delivery, may generate inappropriate inferences for a specific patient that the physician does not follow in practice. Rather than providing automated decision support, OncoDoc allows the clinician to control the operationalization of guideline knowledge through his hypertextual reading of a knowledge base encoded as a decision tree. In this way, he has the opportunity to interpret the information provided in the context of his patient, therefore, controlling his categorization to the closest matching formal patient. Experimented in life-size OncoDoc demonstrated good appropriation of the system by physicians with significantly high scores of compliance. We successfully tested the implemented strategy and the knowledge base in a second medical institution, giving then a noticeable example of reuse and sharing of encoded guideline knowledge across institutions.

Introduction

Beyond cost-effectiveness considerations, the need to reduce variations in patterns of clinical practice and to improve care for the same medical problem across providers, institutions and regions has founded the recent development of clinical practice guidelines (CPGs). CPGs were first elaborated as textual documents supposed to encapsulate the current understanding of best clinical practice to provide uniform optimal approaches to problem-focused care delivery, although locally adapted [1]. Numerous attempts have been made in order to promote their utilization in practice such as enhancing the dissemination of electronic versions of original paper-based textual material over networks [2], or designing computer-based decision support systems (DSSs) able to implement CPGs and to integrate guideline recommendations into clinical workflow.

However, despite the proliferation of such systems, there is still little evidence of any change in physician prescribing habits [3], [4]. Additional developments such as computer-generated reminders or individualized feedback to clinicians [5] have shown to enhance guidelines utilization. But yet, median compliance remains very low, even when users positively evaluate DSSs [6].

Among the barriers to the use of CPGs, psychological resistance comes from physicians’ concerns about clinical freedom, doctor autonomy [7] and importance of ownership in guideline implementation [8]. Another reason may come from the technical difficulty of automating the assignment of individual situations to formal categorizations of models. This is due to the very nature of medical knowledge, known to be incomplete, imprecise, and often lacking specificity [9], [10]. As a consequence, recommendations, suited to average patients, are not rules for all the patients [11]; although CPGs may point out the best research evidence to guide the care of “formal” average patients, they are not a substitute for clinical judgement, which has to be applied to each actual individual patient [12]. CPGs information should be interpreted in accordance with the intended meaning by a given physician for a given patient [4]. However, if textual versions of CPG, expressed in natural language, allow for a physician-controlled contextual interpretation, the automatization of a context-free formalization of guideline knowledge may generate unacceptable inferences. This could explain low compliance rates observed with implemented CPG systems [13].

Keeping the advantages of DSSs to optimize the access to patient-centered information and the advantages of text reading to allow for contextual interpretation, OncoDoc is a CPG system elaborated with a knowledge-based approach in a document-based paradigm. Borrowing knowledge representation principles from artificial intelligence, guideline knowledge has been organized in a rigidly structured representation, i.e. a decision tree. However, instantiation of decision parameters is not automatically performed by the system from patient data. Displayed with their modalities as textual documents expressed in natural language, parameters are dynamically instanciated by the user on the basis of his patient-based contextual interpretation of the information elements provided.

Rather than automatically processing the categorization of a given actual patient to his best corresponding formal equivalent in the knowledge-base (KB), OncoDoc actively involves the physician in the medical reasoning process. In contrast with, usual fully computerized approaches [13], [14], the clinician explores the explicit decision tree encoded as a flowchart while navigating through the pages displaying decision parameters. At each step, he has the opportunity of controlling by his free interpretation the best patient-centered contextual instantiation, thus building a sequence of instanciated parameters that corresponds to the best “formal” equivalent of his actual patient. He can, therefore, participate through his clinical expertise to the categorization of his patient in the most appropriate clinical theoretical situation and be provided with the best therapeutic decision advice through the CPG system expertise. Developed in collaboration with the “Service d’Oncologie Médicale Pitié-Salpêtrière” (SOMPS), the medical oncology department of the Pitié-Salpêtrière Hospital (known to be the largest French hospital), OncoDoc has first been applied to the treatment of breast cancer.

In the following section, we give an overview of guideline dissemination strategies, either as textual documents or as computerized systems, and analyze some of their inherent properties that limit physicians’ compliance in our point of view. The approach we used for OncoDoc’s development and implementation is described in Section 3 and is illustrated by a clinical case. Experimentations of OncoDoc, handled in two different medical institutions, allowed us to evaluate the system with respect to physicians’ compliance, knowledge base sharing and reuse, and clinical trial accrual. Section 4 reports some results from these routine uses. In Section 5, we address the issue of guideline knowledge flexibility and we explain how OncoDoc’s specific paradigm handled this difficulty. The paper ends with a conclusion and the presentation of our future plans.