By Doris Nessim, RPh, MA
EVER SINCE THE INSTITUTE OF MEDICINE’S 1999 LANDMARK REPORT, “To Err is Human: Building a Safer Health System,” there has been a heightened awareness of the risk of harm associated with medical errors, and specifically, medication errors. The national spotlight is now on initiatives aimed at reducing medication errors and medication error prevention strategies. As a result, a great deal of attention has been focused on systems that will intercept potential errors, such as computerized prescriber order entry (CPOE) and point-of-care and automated dispensing systems as technologies. The risk of serious medication incidents associated with intravenously administered medications has caused significant interest in infusion devices with safeguards to protect the patient from potential harm associated with administration-related errors.

The primary method by which information technology accomplishes error prevention and detection is through the use of intelligent clinical decision support software (CDSS), which improves communication and makes information accessible to the clinician. The software provides reference information on the appropriate dose or dosage range of a drug; dose and dosage range and frequency calculations (e.g., calculating weight-based doses of medications); and clinical checks, and monitoring (e.g., alerts that warn of potential excessive doses according to physiological changes occurring in the patient). In this manner, the clinical application utilizing the CDSS can track the frequency of clinical alerts, or potential adverse events averted.

Various technologies are aimed at reducing the error rate at each stage of the medication management process. For example, CPOE is aimed at reducing the prescribing error rate (39%), and the electronic medication administration record, combined with bedside point-of-care technology, is aimed at reducing the transcribing error rate (11%). Automated dispensing technologies and bar code scanning for drug dispensing seek to reduce the dispensing error rate (12%) while also improving medication access at the point of care. Lastly, ‘smart’ infusion technologies aim to reduce the 38% error rate associated with medication administration. Even when CPOE and bar code scanning technologies are implemented, serious errors may still occur through medication administration, especially with ‘high alert’ intravenously administered medications such as dopamine, dobutamine, heparin, and insulin.

The State of Infusion Pump Technology and the Need for Safety Enhancements About 60% of the most serious and potentially life-threatening adverse drug events are related to intravenous medications. However, observationbased studies and voluntary reporting of medication errors underestimate the true extent of the problem. Under reporting is hard to quantify; we don’t have an accurate representation of the near misses, or potentially life-threatening situations.

Intravenous medication administration using conventional infusion pump technologies may be delivered accurately when the infusion pump is properly programmed. However, these conventional infusion devices have been linked with medication errors and adverse drug events. Human error can occur when programming the device. Human Factor studies have demonstrated that, with current resource constraints, double-checking the programmed dose is unlikely to occur, and even if it does, it is still possible that either the information regarding dosage limits is not readily known or the misprogramming is not caught. In addition, a nurse could program the infusion pump with the right medication and dose, but inadvertently program the wrong rate of administration—older pump technologies would not catch such an error. In particular, if a ‘high risk’ medication is being used, then the programming error could result in a serious adverse event for the patient. Additional identified user problems have included incorrect flow rate, dose errors, and air-sensor problems. In some cases, the intravenous tubing in the pumps may be misloaded, however, if the condition were undetected by the misloading sensors, overor underinfusion would occur.

There was clearly a need for infusion device technology that would use software to provide safeguards against dosing and infusion rate errors. The emergence of infusion pump technologies with such clinical decision support has been receiving increasing attention because of its role in risk reduction. This new technology provides a level of patient protection by generating an alert message when programmed doses are over or under the standard limits set by the health system.

Medication Safety Using Infusion Pumps with Intelligence
Decision-support infusion devices (or ‘intelligent pumps’) represent a significant step forward in medication safety. A ‘smart pump’, (as they are referred to by the Institute of Safety Medication Practices (ISMP)), is an infusion system that performs a ‘test of reasonableness’. These devices have intelligence in both their CDSS and hardware that checks that the pump programming is within pre-established, institutionally defined upper and lower limits of dosage concentration and infusion rates. Alarms indicate that an infusion parameter is outside a predesignated range, and infusion will not begin. Thus, the ‘smart’ pump (or ‘pump with intelligence’) detects, alerts, and may stop or prevent (based on how soft/hard stop parameters are determined and set by the implementation team) a potential medication error, in either underor over-delivery of a medication dose, a potentially dangerous order, or in the calculation of an infusion rate.

Ideally, these devices combine intelligence with connectivity to ‘close the loop’ between the infusion device and the technologies of the medication administration record, the automated medication dispensing cabinet, and the CPOE system. The ‘ideal’, however, must be aligned to the health system’s culture of medication safety, goals, its clinical and technical requirements and resources, and its current technical infrastructure. Selecting the ‘right’ infusion device and implementing planning stages and processes for a particular health system depend on matching the degree of intelligence and connectivity in the chosen device to the needs of the health system. Thus, the key to success lies in assembling a multidisciplinary team consisting of clinical, technical, and quality improvement/ patient safety-dedicated resources that will provide information about the system to selected vendors and work closely with them throughout the selection process.

All infusion devices classified as ‘smart pumps’ will contain a drug library of minimum and maximum dose rate guidelines for each drug and drug concentration. This is developed by a hospital interdisciplinary team and programmed by a clinician. Before implementation of a decision-support infusion pump, dose limits need to be defined and standardized for each patient care area (e.g., pediatrics requirements will differ from adult requirements). Implementation should include multidisciplinary team membership, with representation by nursing, pharmacy, medicine, biomedical engineering, and information technology support resources.
The system will produce an alert whenever the nurse enters program parameters that are out of drug limits for a patient care area. This ‘test of reasonableness’ has been shown to prevent programming errors, hence preventing adverse drug events. Infusion devices with intelligence will differ in the maximum number of drugs in the drug library that the particular manufacturer’s device will hold. However, in all cases, the device will provide a log and download of programming entries that could have resulted in a potential error and adverse drug event had it not been averted through the device’s alert/rule to the clinician, as well as situations in which a clinician elected to ‘over-ride’ a warning. Reviewing patterns of the types of errors, as well as when and where errors are occurring, is key to determining ‘best practices’ for enhancing patient medication safety through the use of infusion devices with intelligence.

Infusion devices differ in the degree of intelligence and connectivity that they provide. The next stage of intelligence uses bar code technology with the device’s hardware, with the bar code reader either built in or tethered to the infusion device. When a medication order is entered by the pharmacy into the pharmacy information system, the order may be translated into a bar code, which, when included on the label of an intravenous solution bag, may be programmed into the infusion device when the bar code is scanned by the nurse at the point of administration. Thus, the combination of the scanned bar code identifying the intravenous solution with the drug library programmed in the infusion device serves to further reduce errors by ensuring that the clinician has selected the right drug and right drug concentration from the drug library. Use of bar coding requires an investment by manufacturers and hospital pharmacies. Currently, only a limited number of manufacturers of intravenous solutions have bar coding on their products.

Increasing connectivity, some of these devices verify that the right patient has received the right drug and drug concentration by the implementation of Auto-ID, in which the patient, clinician, intravenous medication, and intelligent infusion device each contain a bar code identifier. This additional feature monitors potential errors and adverse drug events through an additional classification by patient population, and allows the audit of infusion device use and location.
In the near future, infusion devices with increasing levels of intelligence and connectivity will truly enable ‘closing the loop’ when vendors combine these increasingly intelligent pumps with an open system architecture and server-based network between the infusion pump server and the server for the electronic patient record (CPOE system) and point-ofcare system. Bi-directional, real-time communication between these servers would occur through the means of wireless radio frequency, such as an 802.11b Ethernet card. For those health systems that have CPOE, an electronic medication administration record, and bedside point-of-care devices in place, the process would be that a clinician would scan the label on the intravenous infusion solution and the patient and clinician ID, verifying the right medication for the right patient, and the right dose, time, and route. Bi-directional communication would involve sending information via the bar-coded label between the electronic patient record (hospital information system) or pharmacy information system, verifying accuracy of the patient and medication, and documenting administration in the patient’s electronic medication administration record.

Financial Considerations
In the struggle to prioritize among numerous competing health care technology selections and implementations, it is important to realize a solution that will be consistent with your hospital’s safety organizational agenda. The case for implementing infusion technologies with intelligence is supported by the immediate reduction in potential adverse drug events associated with high-risk intravenously administered therapies, and the associated reduction in hospital costs associated with length of stay, laboratory tests, and corrective follow-up treatments.

Since decision-support infusion devices are rapidly evolving, when negotiating a contract, cost considerations need to include contractual scalability for upgrades to the infusion device’s software and hardware. Additional cost considerations are implementation fees, fees associated with monitoring error logs, and on-going support fees and services.

Selection and Implementation of Infusion Devices with Intelligence
Any recommendation of how to proceed with selecting the ‘right’ infusion device for your health care system involves first a clear understanding of the goals your health system has for enhancing medication safety within your institution. The successful selection and implementation of this technology will include a plan involving clinical, technical, quality assurance/ risk management, and financial considerations. Implementation should include multidisciplinary team membership, with representation from nursing (including nursing administration and nurse educators), pharmacy, medicine, biomedical engineering, information technology, and importantly, the quality improvement/ patient safety officers.

Critical to success is working in partnership with your selected vendor to define your implementation needs so as to attain your team’s goals. Keep in mind that this technology is still relatively new and continuously evolving—new adopters may experience implementation challenges. Hence, the implementation team needs to continue to work closely together throughout each step in the implementation and postimplementation evaluative processes. In addition, the team should work with the vendor to identify areas of support as well as functionality features that will further enhance user acceptance, ease of use, and most importantly, medication safety.

Selection
Infusion devices differ in the degree of intelligence and connectivity that they provide. There are many different safety features to evaluate from the location of settings, which ensures ease of use, to how the settings are determined which ensures that they can meet the safety and flexibility requirements established by your implementation team. Some important features that should be considered in purchasing the right infusion device system for your health care system are:

• Maximum number of drugs allowed in the drug library
• Drug library flexibility
• Dose range checking features
• Start/stop/restart features
• Event history reports
• Alarm/trigger features
• Integration options (bar coding, RFID, PIS, HIS)

Manufacturers differ in the safety features of their devices, in the visual and alarm features, in the location of settings, and how those settings are determined.
Another important area to evaluate in the information-gathering stage of the selection process pertains to additional or newly developing features that are awaiting 510K approval by the FDA. Additionally, you should ask about services vendors can provide to you and your team to guide evaluation of the data gathered by the infusion device around errors averted, and therefore, harm prevented. This is critical to developing behavior and practice changes in patient care areas.

Implementation
Nothing will replace the decision-making process and role of the clinician at the bedside, and how these decision-support infusion devices are implemented and deployed can either complement or hinder the safety features of the pump. It is important, for example, to simplify the selection among several dosing units as well as the number of possible drug names.

Therefore, prior to implementing infusion devices with intelligence, the interdisciplinary team will need to consider ways to reduce the risk of errors associated with the selection among different drug names, concentrations, and units of measure through:

• standardizing nomenclature
• standardizing dosing units of measure
• standardizing drug concentrations
• assuring consistency between the information programmed in the library of the pumps with that used in the pharmacy information system, electronic patient record, and clinical documentation, including the medication administration record
• agreement as to the upper and lower dose range limits
• agreement as to the hard and/or soft dose limits
• deciding how alert overrides will be monitored so as to determine whether settings for the upper and lower dose limits need to be reevaluated or whether practice change and education may be required
• including human factors design and considerations as well as failure modes and effects analysis (FMEA) in the implementation process

Since the use of decision-support infusion devices may be rolled out throughout the hospital at different times or stages, the implementation process should also consider:

• a procedure for staff to follow when a prescribed medication is not in the library or it does not have a standard concentration
• a procedure to use when patients are transferred from one area utilizing a ‘smart pump’ to an area that uses a conventional pump
• using the same type of pump throughout the hospital so as to reduce the necessity for staff to remember where to find safety features on different smart pumps or how to program one pump versus the other

Summary
This article is intended to provide an introductory description to the developing state of infusion technology that uses clinical decision-support intelligence and connectivity features to enhance patient medication safety with intravenously administered medications. Keep in mind that, in addition to the important features that are presented, there are many more features of intelligence, infusion device hardware, and connectivity that are important considerations in the selection, implementation, and evaluation of intelligent infusion devices. Infusion devices with intelligence are still relatively new and are continually evolving and rapidly changing. Vendors are working to enhance both the degrees of software intelligence as well as the connectivity features that will allow bi-directional communication with the electronic patient record (hospital information system) and point-of-care system.

Doris Nessim, R.Ph., M.A., is an independent healthcare consultant providing clients with objective assessments, process improvement, and project management with health enabling technologies. In addition, she is the author of an upcoming Neuenschwander Report entitled i.v. IQ: A review of increasingly intelligent and communicative infusion pumps, with commentary. Doris may be contacted at dnessim@rogers.com.