Protocols in Evidence-Based Healthcare

INFO-I530 (Foundation of Health Informatics) Protocol Based Systems Lecture #5

Lecture in a Nutshell • Protocols and Evidence Based Healthcare • Introduction - Protocols (Guidelines) • Protocols in Healthcare Delivery • Structure of Protocols • Care Pathways • Protocol Life Cycle • Departure of Protocols • Application of Protocols • Computer Based Protocol Systems • Passive Protocol Systems / Active Protocol Systems • Protocol Representations and Languages • Disseminating Protocols • Introduction • Clinical Impact • Improving the Uptake of Evidence Based • Designing Protocols • Protocol Construction and Maintenance • The Design of Protocols / Design Principles of Protocols

Protocols and Evidence Based Healthcare

Introduction • We had been taught to call the clinical judgment as an art • The efficacy of much clinical practice is still not validated • Delays in transferring research findings into routine • clinical practice  Streptokinase is useful in the treatment of myocardial infarction: first trial 1958, additional tests 1970s, meta-analysis 1980, routine treatment 1992 • Bottleneck between the publication of clinical trial data, and its conversion into clinical practice. • The exponential growth in medical knowledge is one of the causes of the many gaps we find in the knowledge of all practicing clinicians. • Only the data from patients enrolled in formal trials are analyzed scientifically. • Evidence-based medicine (EBM) or evidence-based healthcare (EBH): The hope is that if best-practice guidance can be distilled rapidly from the literature, then it should be convertible into a set of protocols that can be made readily and widely available to practicing clinicians. • Genuinely evidence-based practice will have the effect of more patients receiving up-to-date treatment, and allow the outcome data from all patients on the same protocol to be pooled for statistical analysis.

Protocols (Guidelines) • A protocol is a set of instructions (guidelines): might describe the procedure to be followed when investigating a particular set of findings in a patient, or the method to be followed in the management of a given disease. • A protocol is usually understood to be advice on the 'best' way to carry out some task. • Multiple terms are used: • Algorithm: In computer science, a set of instructions to carry out some task programmatically is called an algorithm. In clinical practice algorithms usually, but not always, involve some form of numerical calculation. • Protocol: Typically a protocol describes all the steps in the management of a clinical condition, and may cover the steps taken to both secure a diagnosis, or to treat the illness. • Guideline: Often used synonymously with the term protocol, the term emphasizes that the role of the patient management instructions is to offer guidance rather than dictate a specific course of action. • Care pathway: Pathways are most commonly used in nursing, and describe not just the steps to be taken in managing a patient, but the expected course of the patient's management. • Practice parameters: A North American term, used to describe evidence-based clinical guidelines for diagnosis and management of specific clinical conditions.

Protocols in Healthcare Delivery • Research: maximizes the likelihood that the same thing is being done to each patient, and that the data collected are representative of the effects of that particular treatment. • Delegation of responsibility: There is a strong case to be made for highly trained healthcare staff to delegate the care of minor or routine problems such as trained nursing staff for diabetic care. • Demarcation of responsibility: A protocol can make clear which tasks are to be carried out by different members of a healthcare team. • Education: A clinical protocol ensures that, even if there is a variation in ability or training, a certain minimum standard is adhered to in the delivery of care. • Safety-critical or complex situations: Even highly trained individuals will use protocols in situations in which errors can have significant consequences. • Uncommon conditions: Protocols are important in rare conditions.

Protocols in Healthcare Delivery cont. • Increased compliance with a guideline should result in decreased variation in clinical practice and the assumption is that this will then translate into improved health outcomes. • The hypothesis that guidelines will improve health outcomes has not been universally proven and there are significant variations in the impact they have on process and outcomes improvements, • The inefficiency of some protocols are not directly due to the evidence contained within a guideline, but due to the manner in which the guideline is represented, and the socio-technical environment in which it is delivered. • Protocols intended to be used in situations of time pressure or emergency will be designed very differently to ones used in routine practice.

Structure of Protocols • Entry criteria • Irrespective of its form, every protocol begins with an entry criterion that defines the context within which the protocol is used. • It is sometimes argued that protocols are of little value because the decision to use the protocol requires medical expertise. The simple counter-argument is that insufficient thought has been put into the design of the protocol's entry criteria. • Protocol Form and Function • Flowcharts (decision tree format) are probably the simplest way to represent a protocol, because they are graphical in nature, and make decision points and the flow of logic explicit. • Protocols can also be expressed more compactly as a simple set of logical rules, structured as if-then statements. • Chunked Protocols • One way to manage protocol’s complexity is to break the protocol down into a sequence of smaller chunks, each corresponding to a specific task. • A simple way to think about such chunks is as a finite state machine. • Chunking minimizes the effects of design change. Thus, in a well-designed protocol, it should thus be relatively easy to change some detail in a particular protocol segment, usually without affecting other segments.

Structure of Protocols cont. Flowchart In this sample protocol, a guideline is presented for an insulin-dependent diabetic's self-management of their intermediate-acting insulin regime.

Structure of Protocols cont. Logical Rules In this sample protocol, a guideline is presented for an insulin-dependent diabetic's self-management of their intermediate-acting insulin regime.

Structure of Protocols cont. Complex protocols can be broken down into chunks corresponding to different states in a process, or in the progression of a patient's condition. Each state within the protocol is entered or exited when a set of conditions is met. Movement back and forth between states can be permitted if necessary.

Care Pathways • The process of breaking down treatment into a set of stages or states, each with their own entry and exit criteria, is the basis of care pathways. • These protocols are at present especially of interest in nursing, where they are often used to support patient care. • In a pathway, a patient's care may be broken down into a sequence of days, corresponding to the ideal length of stay in hospital for that condition. • In this way, there is some allowance for individualization of care depending on a patient's rate of recovery, as well as a check to prevent inappropriate treatment being given. • As part of the chunking strategy, the pathway can be decomposed into multiple individual problems that are associated with a condition, and then be used to guide the management of each problem over the period of the patient's stay.

Protocol Life Cycle • The management of every patient could be considered, at an abstract level, as constituting a scientific experiment. Consequently, the results of that experiment could contribute to the advancement of knowledge. • If it is possible to treat most patients using protocols, then it should also be possible to close the loop, and use the knowledge gained from each patient's treatment. In effect, every patient treated according to a protocol can be part of a clinical trial. • Model-measure-manage cycle: the results of protocol application feed into subsequent revisions of the protocol. • The information loops are formed around the selection and application of protocols, and the analysis of the protocol's effectiveness.

Protocol Life Cycle cont. The three-loop model describes the way in which protocols can be used to manage individual patients in a uniform way, and use the results of treatment to advance medical knowledge.

Departure from Protocols • A common criticism often voiced of protocols is that patient states are too variable to be amenable to such programmatic care. • Variances are due to: • Patient condition: A patient may have an intercurrent illness that makes it difficult to carry out the care designed for a typical patient. Equally a patient may not respond as expected to a treatment because of their background fitness, nutritional status or genetic predisposition to respond to the treatment. • Treatment variation: Treatments may be given that are at variance with the planned protocol, for a variety of appropriate or inappropriate reasons. • Resource constraints: External factors such as the hospital system itself may contribute to variances. • Rather than being problematic, identification of variances can become a central part of the way in which protocol-assisted care is given: • Variances are signals to the care team to reassess whether it is appropriate for a patient to be maintained on a protocol. • When variances from a number of patients are pooled, they act as checks on the way in which care is delivered. • Variances offer an opportunity to assess the appropriateness of the protocol.

The Application of Protocols • When should NOT a protocol be used? • when clear goals cannot be isolated, and it is dangerous to make simplifying assumptions in order to isolate goals that would allow a protocol to be applied • when end states or outcomes cannot be clearly defined • when the utilities of different decisions are not independent of one another • when probabilities of outcomes or the utility of individual decisions are not independent of the context in which they are made • (when recognizing is harder than acting) • How should a protocol be used • In the passive approach (permissive), protocols act as a source of information only, and are not formally incorporated into the care process. • In contrast, the active use of protocols (prescriptive) shapes the delivery of care around a protocol. The steps in a treatment and data capture are explicitly guided by protocol, requiring the protocol to be consulted at most steps during the care process. Unsurprisingly, the introduction of a new prescriptive process is organizationally difficult, as it inevitably changes the way people work. However, no matter how desirable the changes may be from an organizational point of view - the new system should fit the users and not the reverse.

Computer Based Protocol Systems

Passive Protocol Systems • A passive protocol delivery system acts as a source of information only, and is not intrinsically incorporated into the care process. • They make it less likely that steps will be inadvertently forgotten or altered. • The protocols are not integrated with other modules of the system such as order-entry or results reporting. • Improving access to clinical protocols through information technology does have a positive impact on adherence to guidelines. • If individuals are not reminded to check a protocol for each case, will the advantages of the passive system remain, or will healthcare workers' performance tend towards that without protocols? • Increasingly, many see the Internet as the vehicle of choice for the distribution of clinical practice guidelines. (US Guideline Clearinghouse, and the Cochrane Collaboration's Web site)

Active Protocol Systems • In an active system the protocol becomes central to the way care is delivered and care processes are designed around the protocol system. • There is a growing literature supporting the benefit that active protocol systems may have in supporting clinical practice. • Using computer representation of a protocol as a template to action, a variety of clinical activities can be supported or automated in some way. • These activities range from assisting with recording events into an electronic patient record, to driving medication ordering or test scheduling. • Protocol-driven record keeping can guide the entry of routine details into the electronic patient record (e.g. checkbox instead of descriptive report). • Protocol-driven record keeping is thus a good example of a system that delivers direct benefit to its users, by reducing their workload, as well as improving quality of data capture.

Active Protocol Systems cont. Protocol-driven information systems can integrate into many different clinical systems, with cumulative benefit

Active Protocol Systems cont. • If designed appropriately, such systems should reduce, but cannot eliminate, the amount of free-form data entry that occurs in the patient record. • As with passive systems, a primary role of active protocol systems is to provide healthcare workers with task recommendations and reminders. • Unlike a passive system, which requires a worker to make a conscious effort to check that their actions are appropriate, protocols in an active system should be a natural part of the workflow: • Firstly, an alert can be triggered by a computer-detected event such as a clinician ordering a medication, or the arrival of a laboratory result. • The second form of reminding is subtler, but just as important, and relies on the protocols being a natural part of the work situation. • There is a teaching effect that improves the knowledge levels of individuals participating in such studies. • The ability of individuals to assess a situation may be improved by indicating which and how data should be collected.

Active Protocol Systems cont. • Protocols affect workflow management. • The automated management of task scheduling in an organization is more generally carried out by workflow management systems. • The goal of workflow systems is to ensure that work processes are carried out in the mosttime- and cost-efficient method possible. • The degree to which protocols can drive workflow depends on the sophistication of the protocol care process, and the existence of order-entry and scheduling components in the organization's information system. • Data display can be modified by protocol. • Monitor alarms can be set by protocols: If patient-monitoring equipment like arrhythmia monitors or oxygen saturation probes are linked to a protocol system, then they too can be driven in a partially automated manner. The computer can detect that a new stage in the protocol has been entered by checking the events noted in the patient record.

Active Protocol Systems cont. • Protocols can be used either to advise the settings for biomedical equipment (open-loop control), or to control them directly (closed-loop control). • It should be possible to include features in the record-keeping part of an information system to record the variance and store it for later assessment. • Socio-technical issues: The rate of uptake for electronic guidelines in any specific location is influenced by many variables, some of which are local and some more generic. • These factors can limit the application: • Staff have very limited training in the functioning and use of the system • The primary care physicians were not the only decision-makers in this interaction space, and the system ignored the role of the patient in decision-making. • Significant problems were associated with the interaction design of the software. • The guidelines dealt with the ongoing management of established cases, rather than initial diagnosis or treatment. Clinicians may have a bias to sticking with the status quo, or seek evidence only at the beginning of clinical episodes. • Busy practitioners manage patients with complex, multiple conditions.

Protocol Representation and Languages • There is no clear 'best' way of capturing a protocol, and the choice of representational form is dependent on the protocol's intended use. • A computer system does not come with such background knowledge. As a consequence, computer protocols need to be specified in considerable detail. • Much of the advanced research into protocol languages for computers is aimed at creating ways that ensure that the knowledge is captured in as reliable a way as possible  formal protocol ontologies that would then be used to support the writing of specific protocols. • Protocol ontology would capture all the important knowledge about the things being described in the protocol (lab, radiology, pathology, …).

Protocol Representation and Languages cont. • Some of the more significant guideline representation languages include: • Arden syntax: Developed by American Society for Testing and Materials (ASTM). This language encodes the actions within a clinical protocol into a set of situation-action rules known as medical logic modules (MLMs). The Arden syntax resembles the Pascal computer programming language, and is procedural in its design. Arden has recognized deficiencies in the type of things that can be described using it and has little tolerance for errors and interdependencies between rules. • PROforma: is designed to emphasize safe and robust guideline creation. PROforma's ontology is structured around the notion of clinical tasks, which are subdivided into plans, decisions, actions and enquiries. Many PROforma-based systems are in routine clinical use such as RetroGram, CAPSULE, RAGs, ARNO and MACRO. A user display from a computerized protocol system for managing adult acute asthma built using PROforma.

Protocol Representation and Languages cont. • Prodigy: A UK system, Prodigy (Prescribing Rationally with Decision-Support in General Practice Study) has been developed to support chronic disease management in primary care. The main protocol structure is hierarchical, and each protocol is decomposed into scenarios, therapy groups and therapy details. The therapy group level then offers a choice of types of drug therapy, and the prescription details. • Protege: Also structured around an ontology of tasks, Protege has been an ongoing research activity at Stanford University. Protege is essentially a protocol design tool that allows a user to build a protocol, guided by an ontology. Once constructed, the protocol is translated into a machine-readable form. In both Protege and Proforma, the researchers have spent much effort in developing ways for people to specify a protocol in a simple way. • Guideline Interchange Format (GLIF): GLIF is a research system that has not yet effectively been employed in the real world, but was designed with the intention of acting as an interchange format that supported the sharing of guidelines between different institutions and software systems. Its expression language was originally based on the Arden syntax, and its default medical data model is based on the HL7 Reference Information Model (RIM).

Protocol Representation and Languages cont. • All of the reviewed protocol representation formalisms contain primitives that represent specific clinical tasks that will be recommended to clinicians. These primitives can be classified into two categories: • Actions: An action is a clinical or administrative task that the protocol system recommends should be performed, maintained or avoided, e.g. a recommendation to give a medication. • Decisions: A decision is made when one or more options are selected from a set of alternatives based on pre-defined criteria, e.g. selection of a laboratory test from a set of potential tests. • Computer representations also contain primitives that are used by the computer system to record intermediate states: • The clinical status of a patient: records the state that the computer system believes a patient to be in. • The execution state of the system: records the stage of completion of a task, such as an action or decision, during the process of computer guideline execution.

Protocol Representation and Languages cont. Computer protocols typically are constructed from a set of standard primitives, and assembled to produce complete protocols.

Protocol Representation and Languages cont. • These primitives are used to construct the specific steps in a protocol: • Scheduling constraints: specify the temporal order in which representation primitives can be executed during guideline application. The execution of steps may be in a linear sequence, but many systems permit parallel execution if more than one set of actions are required. • Plans or nesting of guidelines: Nesting of guidelines defines the hierarchical relationship among guidelines during guideline application. • For an active protocol to be applied in clinical practice, it will require access to data about a specific patient's state as well as clinical context data, such as medication orders for a patient. This is ideally provided by integrating the active protocol system into an EMR and an order-entry system. • Striking a balance between prescription and permission, protocol systems need to be only as formal as is necessary to ensure appropriate outcomes, without restricting the permission needed by clinical workers to vary their work patterns.

Disseminating Protocols

Introduction • In an ideal world, every clinician would have immediate and easy access to advice on best practice and the supporting evidence. • The Cochrane Collaboration is now perhaps one of the most influential organizations working in the area of evidence-based practice. • Further, evidence suggests that, even when clinical protocols are available, clinicians forget to follow them, or deviate from them without clear cause. • A further broad area of difficulty is socio-technical and lies within the culture of clinical practice (intrusion on their clinical freedom). • Evidence-based recommendations have to compete with other piece of information for clinicians (information marketplace). • The global store of clinical evidence is growing rapidly, perhaps exponentially. • The amount of information that can be accessed or 'consumed' is fundamentally limited by the constraints on human attention. • For producers of information, the uncomfortable consequence of an ever growing information supply and scarce human attention is an economic Malthus' law of information the fraction of information produced that is actually consumed will, with time, asymptotically approach zero. • Dominant Design theory: widespread adoption makes it hard to change.

Clinical Impact • The clinical impact of a guideline is determined both by its efficacy as well as its adoption rate. Irrespective of whether the content of a guideline reflects best practice, if it is not used it will have no impact: Efficacy x Adoption Rate= Clinical Impact • The level of adoption of any guideline is a reflection of the ease with which the product can be accessed, and its perceived utility amongst clinicians once it is accessed. • Combining these two measures produces a measure of the actual clinical impact of the guideline. Thus, a guideline describing a treatment that does not have the best clinical outcome may nonetheless be the best when we consider its ease of adoption and consequent impact on the health of the population.

Clinical Impact cont. Treatment A gives a 90% success rate, and achieves an adoption rate of 1% within the population of patients. The remaining patients use the baseline treatment with 50% success. The overall improvement to population health produced by A is thus: Impact of A = 0.9 x 1/100 + 0.5 x 99/100 = 0.504 Treatment B gives a 80% success rate, and is used by 10% of the population. The remaining patients use the baseline treatment with 50% success. The overall improvement to population health produced by B is thus: Impact of B = 0.8 x 10/100 + 0.5 x 90/100 = 0.530 The improvement produced by introducing A is thus 0.004 (0.504-0.5), and by introducing B is 0.03 (0.530-0.5). B thus has an impact 7.5 times as great as that of A (assuming the baseline of 50% success produces a 0.5 impact rate) Sample calculations for a given impact factor

Improving the Uptake of Evidence Based • A substantial body of work now exists in the social and behavioral sciences that examines how personal and systemic changes occur  used to develop strategies that encourage clinicians for evidence-based practices. • There are only two ways to increase the uptake of a product: • The 'cost of ownership' of information can come down, making resource-strapped clinicians more able to access and apply evidence. • The value of information to the clinician could go up, increasing the benefit to clinical practice. Clinicians should then be willing to devote more resources to accessing evidence than to other activities. • Much of the benefit of EBH is couched in terms of benefit to the healthcare system or to patients, and there has been little emphasis on finding ways to make individual clinicians derive direct benefit (e.g. CME points). • Evidence-based practice imposes costs on clinicians in both making the initial change to their practice to become evidence-based, and the effort in maintaining this way of working.

Improving the Uptake of Evidence Based cont. A non-exhaustive catalogue of the costs and benefits of using guidelines in clinical practice for the individual clinician, the patient, and the healthcare system

Improving the Uptake of Evidence Based cont. • One study estimates that up to 50% of the variation in compliance rate by clinicians with guidelines can be ascribed to the clinical setting alone. • Providing access methods that are optimized to local needs can also enlarge the range of clinical contexts in which evidence is used (e.g. mobile devices for emergency settings). • Clinician perception is subject to a range of normal human biases that need to be accounted for when the value of EBH is explained to them (e.g. humans seem to give greater weight to losses than to gains). • Altering the perceived cost-benefit ratio of using guidelines for clinicians: Separate gains, Combine losses, Avoid valuation of sunk clinical costs, Separate small gains from large losses, and Combine a small loss together with a larger gain. • Organizational, social and professional (socio-technical) factors have been hypothesized to be at least as important as technical and practical factors in preventing more widespread uptake. • Nurses seemingly place greater value on policies and procedures while doctors have a stronger emphasis on the role of evidence from the biomedical literature in their decision-making culture.

Designing Protocols

Protocol Construction and Maintenance • A protocol should be feasible and accessible. Failure to meet either of these conditions of protocol designability or protocol usability will cause difficulties. • Two key features of the protocol design and maintenance process are: • Creation of a protocol does not occur at a single moment in time, but is part of an ongoing process that assesses a protocol's performance, and refines it accordingly. • Creation of a protocol cannot be an isolated event; its form and content must be designed to reflect the context within which it will be used. • Despite the rewards contemplated by massprotocolization, it is by no means yet certain whether it is technically feasible. Protocols are designed on the basis of a set of assumptions about the nature of the disease that is to be treated, as well as the context within which the protocol will be used. This context of the care process defines who will be delivering care, their available resources and the local expectations of the goal of care.

Protocol Construction and Maintenance cont. • The dynamics of the information market means that as the quantity of available evidence grows, good-quality evidence will become harder to find over time (e.g. Cochrane Collaboration). • We will soon need automated means for exploring, collating and disseminating best-practice knowledge. Progress has already been made by biomedical journals through the adoption of standard formats for article abstracts, ensuring that key aspects of a paper are always present. • Statistical meta-analysis is one of the tools used to decide what constitutes 'best practice' on the basis of published clinical studies. There has to be a consensus process in which individual studies are selected to be pooled for such analysis. • Equally, where the literature is equivocal about the best way to treat a condition, then decisions need to be based on other criteria. In both cases, it is important for experts to discuss and reach consensus.

The Design of Protocols • The cause of this protocol rigidity can often be attributed to poor design. • It is still a commonplace for those designing protocols to spend most of the time 'getting the evidence right' rather than focusing on interaction design. • The contextual factors include the following: • Patients are rarely so accommodating as to present a typical pattern of disease. (e.g. medication side effects varies between patients and the protocol should include different treatment plans) • The goal of treatment for the same problem varies with the clinical situation. Notions of best treatment are always constrained by local goals. (e.g. disaster situations and leaving behind the ones that will not survive). • Goals are also constrained by the local resources available for treatment. There is little point in specifying protocols that cannot be carried out. • The skill level of the individuals required to carry out the protocol also affects the form and content of protocols. • Local care processes evolve to reflect not just best-practice knowledge, but local resources and goals. A protocol has to be designed with an understanding of how it will be used within such existing processes.

Protocol Design Principles • Principle 1: Make any assumptions about the context of use explicit • What is the goal of the protocol? • What are the protocol entry and exit criteria, and how will these be determined at the time of use? • Who will decide protocol entry, and who will apply the protocol? • What terminology will be understood by those using the protocol? • How much time will be available to follow the protocol? • What treatment resources are available, including medication and biomedical devices? • Are multiple treatment options to be considered? • How much detail should be included? • Which representation is most appropriate (e.g. flowchart, decision tree, rules)? • Will users wish to, or be able to, access the evidence used in creating the protocol? • How is the protocol to be used in the care process? For example, how is it to be accessed? • How is the protocol's performance to be reviewed, and how are variances to be recorded? • What mechanisms will be available to update the protocol? • How long should a protocol be in use before it is considered to be out of date?

Protocol Design Principles cont. • Principle 2: A protocol should not be more specific than is necessary to achieve a specific goal The more specifically a method models a given situation, the more useful it will be in that situation, but the less useful it will be in others  identifying the appropriate level of detail is important. Protocol designers should thus be aware of the degree of generalizability they will require of the protocol, and that the wider the expected adoption. One technique that has been suggested for managing these difficulties is to initially specify a protocol only in general terms, and then instantiate it with local data  skeletal plan refinement. (e.g. Normal blood pressure through intravenous fluid replacement' might be an appropriate goal for a protocol, rather than specifying exactly how much fluid is to be given)

Protocol Design Principles cont. • Principle 3: Protocol design should reflect the skill level and circumstances of the users The level of description used in a protocol should also match the abilities of those using it. Very simple steps will probably be best for relatively inexperienced users. For example, protocols for first-aid resuscitation taught to the public are kept very simple. Protocols for trained paramedical or medical staff in exactly the same circumstances may be much richer and more complex, despite the similarities in overall goal. Flowchart for situations in which the ability to understand the protocol is limited, such as an emergency situation, or when the user has had limited instruction.Rule-based representation is more likely to be used as a reminder for individuals who are under less pressure of time.

Protocol Design Principles cont. • Principle 4: Protocols should be constantly reviewed Human knowledge tends to decay with time as circumstances change. Part of the difficulty many people have with protocols is that they represent a snapshot in time of what some people consider the best way to carry out a task.The rigidity that some see in the use of protocols lies not in the protocols themselves, but in the failure to update protocols as knowledge evolves over time.

Summary • Protocols and Evidence Based Healthcare • Introduction - Protocols (Guidelines) • Protocols in Healthcare Delivery • Structure of Protocols • Care Pathways • Protocol Life Cycle • Departure of Protocols • Application of Protocols • Computer Based Protocol Systems • Passive Protocol Systems / Active Protocol Systems • Protocol Representations and Languages • Disseminating Protocols • Introduction • Clinical Impact • Improving the Uptake of Evidence Based • Designing Protocols • Protocol Construction and Maintenance • The Design of Protocols / Design Principles of Protocols

Protocols in Evidence-Based Healthcare