Hello and welcome to this module entitled Incorporating Clinical Decision Support Systems into Antibiotic Stewardship. A good first step in understanding this topic is to define what a clinical decision support system is or does. A clinical decision support system by definition matches patient information to a clinical knowledge base and generates patient-specific recommendations at appropriate times during the course of clinical treatment. These systems combine individual patient data, population trends, and evidence-based medicine into their informatics infrastructure to produce meaningful, actionable alerts. In addition, they have the ability to collate data on interventions, measure drug utilization, improve the safe use of medications and quantify cost saving. The key word in the definition of a clinical decision support system is recommendation. It is not designed to replace the decision of a clinician but rather to help identify opportunities and to add efficiency to the intervention process. There are two general categories of clinical decision support systems. Those that are integrated into an electronic health record system or EHR and third party commercial systems. The main difference between the two is whether or not the user has to go into a separate system to view the alerts. In an EHR based system the data and technology needed to generate the alerts are imbedded within a single system. Not every EHR has the ability to generate actionable clinical alerts. The majority of EHR systems were not originally designed for this purpose. However, over time, this functionality is being embedded as newer versions come into the market. One of the key advantages of an EHR based clinical decision support system is that it has the ability to prospectively influence antibiotic use at the point of order entry in real time. And can be expanded easily into other areas of care such as the ambulatory clinic setting. Third party commercial systems and typically standalone running in parallel with the EHR. Therefore the user must go into a separate system to view the alert and take action. Because these products have been in the market longer, they generally tend to have more functionality and pre-built rules within the system. However, unlike EHR based tools, these systems do not have the capability to prospectively influence antibiotic use at the time of order entry. With the commercial system, the actual alert may be more retrospective in nature. This chart provides a comprehensive listing of the third party commercial systems that were available at the time of this recording. The majority of these products have both antibiotics stewardship capabilities as well as technology to support infection prevention activities. In addition many of them now have the capability to submit antibiotic utilization data electronically in to the antibiotic use and resistance modular AUR. The AUR module is part of the Center for Disease Control's National Healthcare Safety Network, which is also known as NHSN. If we go back to the mid to late 1990s, there were no commercial antibiotic clinical decision support systems that were available in the marketplace. Nor was this functionality embedded into EHRs. The process of connecting laboratory, pharmacy, patient demographics and clinical parameters was very fragmented. Although many hospitals had computers that contained some health information, paper charts were the primary repository for most information about the patient. Therefore the process of identifying patients for antibiotic clinical interventions was very manual and time consuming, which limited the number of patients that clinicians, or groups of clinicians could review at one time. In the modern age there was a significant increase in the amount of data available in a standardized electronic format. Resulting in the development of clinical decision support technology. The graphic on this slide provide a schematic of how the information flows into the clinical decision support tool. For a third party system, data streams feed into the software from different informatics sources such as pharmacy orders, medical records, admission, discharge and patient transfer information and billing data. For an EHR based product most of the information already is contained within the main system. Regardless once the data is integrated, clinical decision rules can be applied to generate alerts. Now large numbers of patients can be reviewed and the frequency in which these interventions can be applied goes up substantially. Now that we have reviewed the technical aspects of these products, let's discuss the benefits of using an electronic system to help identify antibiotic stewardship opportunities. One of the main advantages of using technology to identify antibiotic storage of opportunities is efficiency. An early study published in the New England Journal of Medicine evaluated the impact of a newly developed electronic clinical decision support tool on antibiotic related interventions in an ICU. The investigators did a time and motion study to look at the average duration of time that it took to complete an intervention. They found that when an infectious diseases specialist had to manually review charts before making the intervention, that it took an average of 14 minutes. The computerized clinical tool was able to complete the same action in only 3.5 seconds when the process was automated. Cost savings has been documented with the use of these systems. These costs may equate to dollars that go directly to the bottom line or what we call hard cost. An example of this would be a reduction in antibiotic spend or utilization or a reduction in the length of stay. In some cases, the cost savings measure may be soft costs, such as reduction in personnel time, or decrease in adverse events. These types of costs don't necessarily hit the budgetary bottom line, but they do have an impact on the quality and safety of care. Being able to measure the savings from these systems is important. As there is an associated purchase and implementation cost with these systems. So the measurement and reporting of costs may offset the financial investment that has to be made. This table provides a listing of common antibiotic stewardship alerts that are typically included in these different systems. A common alert that is generated is the Organism- Drug Mismatch, which is also referred to a Bug- Drug Mismatch. For example, an actionable alert may be generated for a patient that has a positive culture for an organism. But that organism is reported as resistant to the patient's active antibiotic therapy. These systems can be very helpful in automating key elements of a successful stewardship program. Including identifying patients for IV to oral conversion, identifying the need for an antibiotic time out, flagging duplicate therapy, identifying patients who need to have their dose adjusted due to organ dysfunction or flagging patients who are receiving an antibiotic but there is no indication for use documented in the medical record. Finally, customized rules can be built in these systems, so that alerts are generated in patients receiving restricted antibiotics or have a positive culture for healthcare-associated infection or a multidrug-resistant organism. In addition to infectious disease related alerts, these systems typically have the ability to generate other medication alerts as well. For example, alerts can be built in the clinical decision support system to identify patients with congestive heart failure who do not have an active order for an ace inhibitor or a patient who is receiving Warfarin but may have a high INR. These systems can also help supplement the organization's medication safety program. For example, a daily report or active listing can be developed that will contain the names of all patients that have received diphenhydramine or naloxone within the last 24 hours. A pharmacist reviews the report, and then conducts a follow up investigation to determine if the patient had an allergic reaction or opioid overdose, as evidenced by the diphenhydramine or naloxone respectively. Almost every system has the capability to generate an antibiogram on demand. This is extremely helpful, because it allows the users quite a bit of flexibility, and allows them to customize the information based on date, or the organisms that are included. Many of these systems allow the user to track their interventions, so that when the alert is received and an action is taken, it can be documented. This information can be aggregated and analysed and presented to key stakeholders along with the savings of G from these interventions. With the increasing emphasis on cost and cost effectiveness in health care, this savings quantification can be very valuable. However it is important to be able to delineate between hard and soft cost saving. Other functionality within a tool such as this includes tracking antibiotic utilization. Common ways that these systems track antibiotic utilization include define daily dose per 1,000 patient days, days of therapy per 1,000 patient days, or days of therapy per 1,000 patient days present. A small number of systems can also provide benchmarking data. Although this is typically facilitated through a separate manual request from the company as opposed to an on demand function. Finally newer technology embedded in these systems is allowing for predictive modeling and personalized antibiotic therapy. In predictive modeling, the system takes into account all clinically available information and tries to predict what infection may be present, along with the likely organisms and their antibiotic susceptibilities based on antibiotic resistance trends at the facility. For personalized antibiotic therapy, the software will take the prescriber through a list of questions to help confirm diagnosis. And then will generate a treatment recommendation that is specific to that patient, and is based not only on population resistance trends, but also on information from that individual patient's medical record, including previous hospital stays. There are several challenges to consider when evaluating the best system that will meet the institution and the clinician's needs. Both each are based in third party commercial systems require some type of IT personal involvement. What must be determined is how many people will be needed to be dedicated to the on boarding and maintenance of the product which can vary from system to system. Additional investment may be required as the technology is to be used with smartphones or other mobile technology. A big consideration with any of these systems is the ease of use. If a clinical decision support system is more intuitive and requires relatively few clicks of a mouse, then you may see easier or faster adaptation of technology, especially in today's environment where there is always demand on clinicians' time. If it requires multiple clicks of a mouse or several screens to scroll through in order to find the information, this makes the system less likely to be used. From a financial perspective there can be two cost categories. The first is what we call out of the box costs which are the costs associated to get the basic system up and running. However, there may be more fees charged on top of the initial investment in order to access additional functionality or capabilities that do not come as a standard feature. For example, some clinical decision support systems have the capability to do keyword searches in radiology or surgery reports. But this is not included in the basic model and additional investment is required. Another example might be additional fees associated with improving mobile user experience, or for supplemental training that is needed from the company when new staff members are on-boarded. Although these systems have many prebuilt clinical rules, additional time may be needed by physicians, pharmacists, infection preventionists and others to add customized rules. This can be challenging because frontline clinicians have to be removed from their patient care activities in order to build and refine actionable alerts. Finally, there needs to be an infrastructure in place whereby dedicated personnel are available to act upon alerts in a timely manner. If clinicians are not in place to quickly review the opportunities that have been identified, then any type of antibiotic clinical decision support system will be of little value. So in conclusion, antimicrobial clinical decision support systems may be either EHR based, or part of a third party commercial system. Many of these systems have both infection prevention as well as an antibiotic stewardship component. Hopefully it has been demonstrated in this presentation that there are pros and cons to each of the systems that are in the healthcare marketplace, but ultimately this medical artificial intelligence can save time and improve patient care. Thank you.
