Turning Guidelines into Practice: Making It Happen With Standards - Part 1

• Abstract
• Introduction
• Part I: Issues and Barriers to the Success of Guideline-based Decision Support
• Barriers to the Success of Guideline-based Decision Support: Guideline-related Obstacles
• Barriers to the Success of Guideline-based Decision Support: Electronic Decision Support Issues
• Integrated Decision Support Systems Are Highly Complex
• Decision Support is Heterogeneous
• The Goals of Different Players Can Be Poorly Aligned
• Technical Requirements Are Complex
• Content Requirements Are Complex

Click here for Part 2 of this paper

Abstract
Successful delivery of the knowledge incorporated into guidelines requires a systemic approach which integrates knowledge with workflow using existing clinical information systems. Electronic clinical decision support (ECDS) systems are the means through which the knowledge embedded in guidelines can be managed and delivered effectively.

Barriers to the success of guideline-based ECDS are multiple and include:

• guideline-related obstacles, both extrinsic to the guideline (organisational and provider specific obstacles) and intrinsic to the guideline (such as failure to meet adequate standards in guideline development and format, identification and summary of evidence, and formulation of recommendations)
• electronic decision support issues which include such factors as:
• the extreme complexity of integrated decision support systems
• poor alignment of the goals of different players
• complex technical requirements
• complex content requirements.

Initiatives with the goal of "improving the guidelines" include the AGREE^{[ b ]} instrument for guideline quality appraisal, the GuideLine Implementability Appraisal (GLIA) instrument which predicts barriers to implementation and the COGS (Conference on Guideline Standardization) checklist of necessary components of practice guidelines.

Numerous projects in "guideline translation" have taken place worldwide, each using different approaches in guideline representation architecture and implementation. These attempts have advanced the process of transforming clinical knowledge expressed in a guideline to a computable format but the absence of commonly agreed standards has created major difficulties for guideline implementers and decision-support systems designers.

Standards are now seen as essential to improvements in the process of "guideline transformation" for ECDS and underpin a further critical requirement – the need for ECDS, which has typically been viewed as a software application, to be considered from a wider systems perspective.

The importance of a systems perspective and standards in ECDS development and implementation was highlighted in a New Zealand based development and implementation of a clinical decision support system for cardiovascular disease risk assessment and management.

The ECDS system developed, which is fully integrated within the electronic patient management system, is:

• integral to clinical workflow
• generalisable and scalable
• applicable to multiple diseases 
• applicable to multiple clinical settings (primary, secondary, community)
• discreet, defined and documented (ie, it truly is a platform for generalisable ECDS).

ECDS development in New Zealand has succeeded because of a shared, articulated strategic vision, an agreed definition of clinical decision support and its strategic value, wide scale sector commitment and the development and implementation of a comprehensive strategic plan.

Introduction
Clinical guidelines embody a rich source of knowledge designed to inform clinical decision-making and care planning and, over recent years, have gained support as vehicles for promoting best practice in clinical medicine.

The potential for guidelines to form part of the solution to widespread, unwarranted variations in clinical practice, poor implementation of recommended interventions (only 54.9% of adults in the US receive recommended care; refer table 1), inappropriate health care and escalating costs of health care has been the impetus behind an explosion of guideline publications.

Table 1: Percentage of adults in the US receiving recommended interventions (based on 439 indicators of quality of care for 30 acute and chronic conditions as well as preventive care)^{[ 1 ]}.

In a perfect world, guidelines embody a clear statement of the most appropriate practice based on available, high quality scientific evidence. Implementation is straightforward and intended users gratefully integrate guideline recommendations into their daily practice. As a result individual and population health improves!

In reality, the effective dissemination and use of guidelines in clinical care is far from perfect and presents major challenges.

Various approaches influence clinicians’ behaviour towards guideline adherence, some of which have proven to be more effective than others (refer table 2).

Table 2: Influences on clinicians’ behaviour toward guideline adherence grouped by effectiveness^{[ 2 ]}

Building on this theme, Zielstorff has described the steps from the development of a guideline to its integration into practice and the subsequent evaluation of its impact on practice and outcomes, as shown in table 3.^{[ 3 ]}

Evidence demonstrates that the most effective method of stimulating awareness of and compliance with best practices is computer-generated reminders provided at the point of care.^{[ 3 ]}

Table 3: Steps in the development of online practice guidelines

Successful delivery of the knowledge incorporated into guidelines requires a systemic approach which integrates knowledge with workflow using existing clinical information systems.

Doing this job effectively requires specialised clinical information applications. Electronic clinical decision support (ECDS) systems are the means through which the knowledge embedded in guidelines can be managed and delivered effectively (refer figure 1).

Figure 1: Delivering guidelines in clinical decision support systems

Developing and implementing systems and processes which make this approach scalable ^{[ a ]} and sustainable present a challenge.

This paper reports on the “Clinical Guideline and Decision Support Workshop” held at the 2nd Guidelines International Network Conference “Evidence in Action” held in Wellington, New Zealand, 1–3 November 2004. The workshop explored a number of issues related to the use of decision support for “turning guidelines into practice”, with a particular focus on the use of standards.

The paper is in two sections. "Part 1: Issues and Barriers to the Success of Guideline-based Decision Support" begins by looking at barriers to the success of guideline-based decision support. These are considered under two heads: guideline-related obstacles and electronic decision support issues.

In "Part 2: Solutions for Guideline-based Decision Support", steps toward solutions to the obstacles facing effective development of guideline-based ECDS systems are considered in two areas: improvement to the guidelines and improvements to the process of translating guidelines into ECDS systems. This section also includes a summary of the New Zealand experience in developing guideline-based decision support, setting out what has been learned from that experience.

Part I Issues and Barriers to the Success of Guideline-based Decision Support

Barriers to the Success of Guideline-based Decision Support: Guideline-related Obstacles
Guideline-related obstacles can be extrinsic to the guideline, including organisational and provider specific obstacles, or intrinsic to the guideline itself.

In a 1999 review of barriers to physician adherence to clinical practice guidelines, Cabana constructed a framework (figure 2) that organised barriers to adherence according to their effect on physician knowledge, attitudes or behaviour.^{[ 4 ]}

Figure 2: Barriers to physician adherence^{[ 4 ]}

Reproduced with permission from JAMA 1999

It is notable that, in this review, most of the barriers related to behavioural issues while only one perceived barrier related to the guideline content itself ("guideline factors"). This highlights that, when seeking successful adherence, addressing the external barriers is critical. It does not, however, diminish the important of factors intrinsic to the guideline itself, which remain critical.

A structured review in 1999^{[ 5 ]} showed that practice guidelines had limited adherence to established methodological standards (evaluated using a 25-item instrument covering methodological standards on guideline development and format, identification and summary of evidence, and formulation of recommendations; refer figure 3). While all areas of guideline development were shown to need improvement, greatest improvement was needed in the identification, evaluation and synthesis of scientific evidence.

Figure 3: Guideline quality evaluation: mean number of standards met based on a 25-item instrument

Reproduced with permission from JAMA 1999

In summary, while guidelines, as de facto repositories of the most current, high-quality knowledge about best practices, can be a superb knowledge source, a number of limitations for their use in ECDS have been identified.

Barriers to the Success of Guideline-based Decision Support: Electronic Decision Support Issues
An ECDS system is an "active knowledge system" designed to assist clinicians in decision-making by matching individual patient characteristics to computerised knowledge bases in order to generate patient-specific assessments or recommendations.

Inputs may be derived from many parts of the patient record or manually entered or amended.

Outputs can include:

• Clinical advice
• Patient specific care plan or order set 
• Information to patients
• Disease specific datasets
• Performance measures for audit and evaluation.

The benefits from ECDS can be considered within three broad categories:^{[ 6 ]}

• Improved patient safety, eg, by reducing medication errors, incidence of adverse events and improving choices of medication and laboratory tests
• Improved quality of care, eg, by increasing clinicians’ available time for direct patient care, promoting increased application of clinical pathways and guidelines, facilitating the use of up-to-date clinical evidence and improving clinical documentation and patient satisfaction
• Improved efficiency in health care delivery, eg, by reducing costs through faster order processing, reducing test duplication, reducing the incidence of adverse events, and changing patterns of drug prescribing to favour the use of less expensive but equally effective generic brands.

To date, ECDS has typically been regarded as a software application. However, to be successful ECDS must be considered from a wider systems perspective.

Such a systems perspective requires that ECDS:

• comprises problem-orientated clinical knowledge 
• is available within clinical information systems 
• is delivered at the point of care 
• is scalable across many diseases and clinical settings
• is supported by standards-based applications that are interoperable with related clinical information systems
• has built-in quality assurance.

Crucial elements of the ECDS system are humanistic rather than mechanistic and it should be recognised that multiple stakeholders are involved in their development and implementation.

The issues with and challenges to successful ECDS systems are numerous and wide-reaching. Key points follow.

Integrated Decision Support Systems Are Highly Complex
A fully integrated decision support system is highly complex, as illustrated in figure 4.

Figure 4: Complexity of fully integrated decision support systems (in the New Zealand setting).

Decision Support is Heterogeneous
The widespread use of the term "decision support" implies a homogeneous set of clinical applications which all serve a similar purpose and are possibly interchangeable. In fact, decision support exhibits significant heterogeneity with a "family" of decision support models providing different functionality. The members are not necessarily interchangeable as they have been designed to serve specific purposes and include:

• best practice decision support and systematised care planning
• alerts and reminders
• workflow enhancement (eg, referrals)
• advisory information (eg, support for laboratory results, support for prescribing)
• context relevant reference information (formularies, guidelines, references).

Success requires that the clinical relevance and user requirements for each model are better understood. Importantly, there is also a need to better understand how the knowledge incorporated into guidelines best fits or is fed into each model.

The Goals of Different Players Can Be Poorly Aligned
Guideline implementation requires that guideline developers offering clinical knowledge and decision support specialists offering informatics knowledge work effectively together. As both groups have differing perspectives, motivations and goals, this represents a significant challenge.

Guideline developers focus on crafting guideline recommendations to fit evidence whereas implementation requires a focus on effective implementation of recommendations so as to influence clinical behaviour.

Guideline narrative can be ambiguous, vague and lack the tight definition required for execution with ECDS. Terms such as "may benefit from", "some experts suggest" and "experts differ in opinion" are common and virtually impossible to translate into the specific recommendations demanded of ECDS.

The difficulty in translating guideline knowledge for decision support was illustrated in an evaluation of the translation of a clinical guideline from text into an encoded form (the GuideLine Interchange Format [GLIF]) which showed that different recommendations would be given for the same patient using GLIF-encoded representations of guidelines formulated by different individuals.

In summary, guidelines, as de facto repositories of the most current, high-quality knowledge about best practices, can be a superb knowledge source. The "different languages" spoken by guideline developers and implementers mean that guideline translation has been recognised as a key direction for moving forward.

Technical Requirements Are Complex
The complex technical demands placed on an effective decision support system include the need for the system to:

• have the ability to manage "localised" knowledge
• have the flexibility to handle multiple conditions and clinical settings
• be interoperable with many clinical information systems (CIS)
• be integratable with clinical workflow
• meet multiple standards for coding, messaging, interfacing, etc, to support system integration.

Content Requirements Are Complex
Numerous issues plague the management of ECDS content. For example:

• There is an increasing need to integrate guideline content from multiple sources, eg, national and local content, evidence-based guideline content and management protocols, etc. 
• Content needs to cover multiple diseases and co-morbidities.
• Guideline sources are numerous and diverse and, as a result, guidelines have variable formats and lack standard architecture (eg, the PREDICT^TM-CVD decision support tool in New Zealand [refer below] incorporates content from nine guidelines).
• Guidelines vary in coverage, depth and richness and can include omissions, gaps and/or ambiguities.
• Content needs vary and can be in conflict.
• There is a requirement for ongoing investment in management, maintenance, refinement and updating.

In addition, the process of translating guideline content to a functional form for ECDS needs to be conceptually sound, reliable, feasible, transparent and testable.

The items outlined above point to a series of factors critical to the successful implementation of ECDS:

• Clear definition of the clinical problem that ECDS will address.
• Clinical relevance and applicability of the intervention(s). 
• Clinical relevance of the decision support solution.
• Clinical workflow relevance – clarity as to when, where and how decision support is provided.
• Appropriate user interface to decision support.
• Clear definition of the scope of the expected behaviour change – new technology, new intervention, etc.
• Support for behaviour change – training, etc. 
• Clarity regarding requirements for and use of incentives – financial, etc.
• Suitable study design – both qualitative and quantitative approaches.
  
  

Click here for Part 2 of this paper

Footnote