The impact of computerised physician order entry on prescribing practices in a cardiothoracic intensive care unit*

Transcription

1 doi: /j x ORIGINAL ARTICLE The impact of computerised physician order entry on prescribing practices in a cardiothoracic intensive care unit* J. Ali, 1 L. Barrow 2 and A. Vuylsteke 3 1 Foundation Doctor, Addenbrookes Hospital, Cambridge, UK 2 ICU Clinical Pharmacist, Papworth Hospital, Cambridge, UK 3 Consultant in Anaesthesia and Intensive Care, Papworth Hospital, Cambridge UK Summary This prospective, time series, cross-sectional study was designed to compare the quality of handwritten vs computerised prescriptions in a tertiary 25-bedded cardiothoracic intensive care unit. A total of prescriptions for 613 patients were analysed over three periods of investigation: 7 months before; and 5 and 12 months after implementation of a clinical information system with computerised physician order entry capability. Errors in prescribing were common. Only (53%) of handwritten charts analysed had all immediate administration drugs prescribed correctly. Errors included omission of route 81 (8.0%), date of prescription 78 (7.7%), and time to be given 255 (25.2%), and 119 (11.7%) had no dose or an incorrect dose prescribed. All errors of completeness were abolished following implementation. The computerised system led to a significant improvement in prescribing safety, in a clinical area previously highlighted as having a high rate of adverse drug errors. Legibility, completeness and traceability are no longer possible sources of medication errors.... Correspondence to: Dr Jason Ali ja297@cam.ac.uk *Presented in part at the Association of Cardiothoracic Anaesthetists Annual meeting, Cardiff; November Accepted: 15 September 2009 Prescribing errors are an important cause of medical error, recognised in the reports: To Err is Human (US Institute of Medicine [1]) and An Organisation with a Memory (UK Department of Health [2]). Twenty percent of adverse events in hospitalised patients are attributable to medication errors [3]. A recent multinational prospective study in the intensive care unit (ICU) environment identified 74.5 errors per 100 patient days in the administration of parenteral drugs [4]. Several strategies to reduce the rate of medication errors concentrate on prescription process and a significant reduction in medication errors is reported when a clinical pharmacist is involved in reviewing prescriptions charts [5]. Information technology has the potential for improving the medication ordering process [6] and the Leapfrog group in the US [7] identified computerised physician order entry as one of its safety leaps to reduce preventable medical errors. There are however, very few direct comparisons of paper and electronic prescriptions in the literature to support this claim, especially in the ICU environment. This study was designed to assess the quality of handwritten medication orders vs computerised orders before and after the introduction of a clinical information system. Methods The study was undertaken in a 25-bed adult cardiothoracic teaching hospital ICU. Funding was obtained to buy a commercially available clinical information system (Metavision, imdsoft, Needham, MA, USA) to replace paper records. The system was introduced at the patients bedside after a team of doctors and nurses had customised Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 119

2 J. Ali et al. Æ The impact of computerised physician order entry Anaesthesia, 2010, 65, pages the software to meet the specific needs of the unit. As part of the implementation plan, working practices were observed and recorded to ensure a smooth migration from paper to electronic patient record. Observations were fed back to the implementation steering group on a regular basis to ensure that issues and problems were tackled early on. Our work was conducted with the agreement of the Trust authorities and Caldicott Guardian, and the Local Research Ethics Committee agreed that this was part of service evaluation audit and did not require ethical review. The unit used paper records for all prescriptions before November Prescriptions were hand written by medical staff or trained senior nursing staff on specifically designed documents. In addition to a standard drug chart the unit used several paper forms for the prescription of ICU specific treatments such as inotropes and haemofiltration. The hospital had a policy defining the minimum standard for completion of each prescription chart. Metavision software was developed specifically for use in ICU, allowing full integration of data gathered at the bedside into a highly customisable interface. It includes parameters from ventilators, monitoring devices, laboratory results, and medical and nursing records. This clinical information system also has computerised physician order entry capability and displays the current prescribed medication, as well as allowing for prescription ordering. A multidisciplinary team at the hospital designed the initial interface to be used in the unit before implementation. The software allows the clinical design team to make changes on the fly and to react to staff feedback, so that the interface is constantly evolving. All orders are prescribed though a similar interface and most orders are pre-entered in the system. When customising, it is possible Figure 1 Screenshot demonstrating the computerised physician order entry software. The figure shows the box that is produced upon the selection of a drug from the available formulary. The route, dose, and frequency are predefined to the most common, but each can be modified by choosing alternative parameters from a drop down menu. 120 Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

3 J. Ali et al. Æ The impact of computerised physician order entry Table 1 List of criteria analysed on prescription charts. Patient identification present Allergy status completed Complete prescriptions Drug dose Route Start date Administration times Approved drug names used Traceability Legibility of signatures Prescription amendments Dose omissions to add a button leading directly to the prescription as chosen. All prescriptions available in the unit were entered by the lead intensivist on predefined templates (Fig. 1), and checked by a nurse and a clinical pharmacist. This included, for most drugs, standard dilutions, doses and administration times. Prescribing staff included all intensivists working in the unit (12) and doctors in training numbering two at any one time out of a pool of surgical and anaesthetic trainees working in the hospital for periods ranging from 3 months to several years. Due to the nature of the unit, most prescriptions on admission are written by the surgical trainee handing the patient over, and are subsequently reviewed by the ICU staff. The unit employs a clinical pharmacist who reviews each chart every day. The items to check were defined before the study (Table 1). All prescriptions on handwritten drug charts (phase one) and on computer records (phases two and three) were examined independently at the patient bedside by a medical student (JA) and a clinical pharmacist (LB). Each phase consisted of two continuous weeks of data collection, with prescription records of the last 24 h analysed each day. Phase one occurred 7 months before the computer record was introduced, phase two 5 months after introduction and phase three 1 year later. Descriptive statistics were obtained using Excel 2003 (Microsoft Corporation Redmond, WA, USA). As the differences between samples were obvious, statistical tools were not required. Results A total of prescriptions written in 613 charts were reviewed over the three phases. The breakdown of prescriptions by type during the three periods is shown in Table 2. Criterion 1: patient identification Although most handwritten charts (187 (98.4%)) had an addressograph, displaying the patient s name, date of birth Table 2 Number of prescriptions by type, before (Pre), and at 5 months (Post-1) and 12 months (Post-2) after implementation of a computerised information system. Values are number (proportion) of total prescriptions. Patients Total prescriptions STAT Regular Variable PRN Pre (22.7%) 2707 (60.7%) 11 (0.2%) 732 (16.4%) Post (15.1%) 4061 (79.9%) (5.0%) Post (15.4%) 4112 (79.7%) (5.2%) STAT, drugs to be administered as one off, at indicated time; Regular, drugs to be given at regular intervals; Variable, drugs such as warfarin that have a variable dose depending on measured parameters; PRN, drugs to be administered when required. and hospital number, only 103 (54%) had a named physician responsible for the patient, and 93 (49%) the location where the patient was hospitalised, written on the chart. The clinical information system requires complete patients information before allowing further clinical activities to be conducted so 100% of details were entered. Criterion 2: allergy status completion Allergy status notification was completed on 178/190 (93.7%) of handwritten charts. This initially fell to 174/ 210 (82.9%) at 5 months post implementation, but by 12 months this trend had reversed and just two charts (1%) had this information missing. Criterion 3: prescription completeness Data for regular and immediate ( STAT ) prescriptions are presented in Table 3 as these were the types of prescription that were most frequent, and with the greatest error. Of the 164 handwritten charts on which STAT drugs were prescribed, only five (3%) had no errors. The commonest errors were the prescriber omitting a time for the drug to be given or the nurse failing to note the actual time administered. Drugs were commonly prescribed with no dose or an incorrect dose. The commonest was Table 3 Errors in handwritten prescriptions identified when analysing regular drugs and STAT (one-off) prescriptions. Values are number (proportion). Regular drugs STAT drugs Number prescribed No route 11 (0.41%) 81 (8%) No date of prescription 66 (2.44%) 78 (7.7%) No time to be given 25 (0.9%) 255 (25.2%) No incorrect dose 110 (4.1%) 119 (11.7%) Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 121

4 J. Ali et al. Æ The impact of computerised physician order entry Anaesthesia, 2010, 65, pages prescribing a drug as x mg.kg )1 without weight being recorded on the chart. With regular drug prescriptions, errors were rarer. The commonest was the omission of a start date and no dose or an incorrect dose prescribed. In order to prescribe a drug on the computerised system, the approved drug name is selected from a list. A dose, dose interval, route of administration and time to be given (STAT drugs) must be selected. As a consequence, no drug can be prescribed without all of the important information that ensures safe administration. As such, all omissions were reduced to 0% in both of the post implementation phases of analysis. Criterion 4: approved drug names Of 4453 handwritten prescriptions in phase one, 463 (10.4%) were not prescribed with their approved drug name, making this the commonest error identified. In order to prescribe on the computerised system, a drug must be chosen from a list maintained by pharmacists. Therefore, no errors of this type were identified in the post-implementation phases of this study. Criterion 5: traceability Prescriptions should be signed in a legible manner. Of all 4453 handwritten prescriptions in phase one, only 109 (2.4%) were unsigned, ranging from 3 (0.24%) with regular, to 81 (8.0%) for STAT prescriptions. In total, 1368 (31.5%) of signatures were legible on analysis of all prescription types. Any prescription ordered on the computerised system is automatically linked to the person logged into the system, ensuring 100% traceability to prescriber in the post-implementation phases, the assumption being that the person logged in to the system is the person actually prescribing. In the event that a change is required, for example in dose, frequency or route, it is required that hand-written prescription are rewritten. Of the 2707 handwritten regular prescriptions 66 (2.44%) were found to have had the route and or dose altered, rather than rewritten. The clinical information system prevents the possibility of amending prescriptions, and requires re-prescription. As such there were none of these errors post-implementation. Discussion This study shows that the introduction of a clinical information system, including prescription functionality, remarkably improves the quality of prescribing in a large cardiothoracic ICU. The design of the system forces prescribers to fill in predefined fields with the required information. We did not investigate patient outcome, but it is logical to expect that better prescription practice will improve patient safety. This report also shows that introducing an electronic prescribing system might introduce unexpected problems. Although the system can alert a prescriber that an order is in conflict with the allergy list, this is only possible if the allergy list has first been completed. While it is a legal requirement that the allergy status of a patient is identified before prescription or administration of medications [8], only 178/190 (93.7%) of our paper charts had the information entered, and this fell to 174/210 (82.9%) after introduction of electronic prescribing. A simple modification that made obvious to users where to access the allergy electronic field, coupled with an automatic alert when not done, allowed us to correct this and achieve a near 100% completion. This simple illustration shows that monitoring the effect of the introduction of technology is required, and flexibility of customisation of the interface helps to fix unforeseen issues. With paper prescription charts, it was common for drug administration information to be omitted or incomplete, as described by others in similar settings [6]. In this study, we do not address the reasons behind this lack of responsible prescribing by clinicians, but these are alarming figures. Computerised order entry will significantly reduce medication errors [9, 10], mainly because parameters such as dose, route of administration, frequency and time to be given are mandatory and the user cannot complete the task if specified fields are left empty. Repetitive tasks are made easier, such as entering the date or instructions for preparation. We did not assess the validity of prescriptions as neither the paper chart nor the electronic prescribing system make a difference to clinicians choice. While dose ranges can be defined, the wrong prescription at the wrong dose can still occur and the ICU pharmacist checking each chart daily has a major safety role to play. It is accepted that ambiguity and illegibility lead to medication errors that could result in serious harm and prescriptions should: always be clear, unambiguous and leave no doubt as to the prescribers intentions [2]. Illegibility affects traceability and responsibility as mistakes cannot be attributed. Implementation of a computerised system ensures that these prescription deficiencies are abolished, and that records are accurate [10]. Medication errors in the US occur at a rate of one per patient per day [11] with a cost of at least $1.5 billion per annum and an extra 1.5 million hospital days [12]. All of these errors are preventable and computerised order entry has the potential to improve the medication ordering process [3]. At the most basic level, we have shown that the system immediately offers the advantage of guaranteed legibility of prescriptions, completeness of orders and traceability. These errors 122 Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland

5 J. Ali et al. Æ The impact of computerised physician order entry might be considered as relatively minor [10] but the sheer number of small errors confers significance in itself. Computerised systems allow other clinical decision support functionalities, such as automatic dose modification in renal or hepatic failure, and detection of drug interactions, to be introduced. These have potential for reducing patient harm but we must remain vigilant that they might introduce unexpected problems. These results convincingly demonstrate the positive impact upon prescription legibility, completeness and traceability of the implementation of a computerised physician order entry-enabled clinical information system in a cardiothoracic ICU. Many errors avoided could be perceived as being minor, but all are errors nonetheless. There are many examples in the literature of where errors perceived as minor at first glance, have lead to fatal outcomes, and such a system affords an additional level of safety against these occurrences. Computerised order entry is a powerful tool, and we propose that there may be significant gain by implementing such systems throughout all ICUs, both in terms of patient safety, and with regard to financial gain associated with decreasing morbidity related to drug misadventures. References 1 Institute of Medicine, National Academy of Sciences (US). To Err is Human: Building a Safer Health System http: object.file Master ToErr- 8pager.pdf (accessed ). 2 Department of Health (UK). An Organisation with a Memory http: en Publicationsandstatistics publications publicationspolicyandguidance browsable dh_ (accessed ). 3 Leape L, Brennan T, Laird NE. The nature of adverse events in hospitalized patients: results from the Harvard Medical Practice Study. New England Journal of Medicine 1991; 342: Valentin A, Capuzzo M, Guidet B, et al. Errors in the administration of parenteral drugs in intensive care units: multinational prospective study. British Medical Journal 2009; 338: b Horn E, Jacobi J. The critical care clinical pharmacist: evolution of an essential team member. Critical Care Medicine 2006; 34: S Department of Health (UK). Building a Safer NHS for Patients: Improving Medication Safety http: www. dh.gov.uk en Publicationsandstatistics Publications PublicationsPolicyAndGuidance Browsable DH_ (accessed ). 7 The Leapfrog group (US). The Leapfrog Safety Practices http: for_hospitals leapfrog_ hospital_survey_copy leapfrog_safety_practices (accessed ). 8 British Medical Association and the Royal Pharmaceutical Society of Great Britain. British National Formulary, March edn. London: BMJ Publishing Group Ltd and RPS Publishing, Evans KD, Benham SW, Garrard CS. A comparison of handwritten and computer-assisted prescriptions in an intensive care unit. Critical Care 1998; 2: Shulman R, Singer M, Goldstone J, Bellingan G. Medication errors: a prospective cohort study of hand-written and computerised physician order entry in the intensive care unit. Critical Care 2005; 9: R Committee on Identifying and Preventing Medication Errors, Aspden P, Wolcott J, Bootman JL, Cronenwett LR, eds. Preventing Medication Errors: Quality Chasm Series. Washington, DC: The National Academies Press, Classen D, Pestotnik SL, Evans RS, Lloyd JF, Burke J. Adverse drug events in hospitalized patients. Excess length of stay, extra costs, and attributable mortality. Journal of the American Medical Association 1997; 277: Journal compilation Ó 2009 The Association of Anaesthetists of Great Britain and Ireland 123