Introduction 1 2 3 4 5 6 7 In view of the potential advantages and increasing use of handheld computers in medicine, we evaluated the benefits and drawbacks associated with introducing this technology in an academic ICU. Materials and methods Hardware 1 Software 1 8 Patient data were entered into the Memopad using a customized template generated with MemoPlus (Hands High Software Inc, Palo Alto, CA, USA). The information entered included demographic data, medical history, current diagnoses, therapy, procedures performed, and management plan. Data was transferred between medical personnel using the PDA's infrared beaming ability. As hospital policy requires a paper record, daily notes were generated by Palmprint software (Stevens Creek Software, Cupertino, CA, USA) using infrared transmission to an HP Laserjet 6P printer (Hewlett Packard, Palo Alto, CA, USA). Various software packages for patient data management (shareware or commercially available software) were evaluated during the study period. Critical Care Handbook of the Massachusetts General Hospital 9 Study subjects PDAs were given to the ICU attending physicians, the rotating resident trainees, and other medical staff. Four to six residents (postgraduate years 2 to 4) worked in the ICU at any one time. On the first day of their ICU rotation, residents were taught how to use the PDA in a 1-hour seminar. The principle investigator and research team were available for further help and troubleshooting throughout the study The research team was responsible for installing and updating software and schedules. Patient data was entered by residents, either during morning rounds or when patients were admitted to the ICU. The updated database was beamed to the on-call resident in the evening and transmitted back to the team in the morning, with new admissions added. Methodology 10 11 12 Critical Care Handbook of the Massachusetts General Hospital 9 P Results 11 Physical attributes Users found the PDA to be a convenient pocket size, allowing it to be available at all times. The screen was clear and easy to read, although not ideal for long text documents or large tables. Many users became proficient in text entry using Graffiti, while others preferred to use the GoType keyboards. Of the 19 PDA units used during the 6-month study period, only one had a technical malfunction requiring replacement. Two were damaged after being dropped and needed to have their screens replaced. No other problems were encountered. Medical reference databases Reference databases used regularly by medical residents included the critical-care drug dosing reference, ventilator weaning protocol, and electrolyte correction application. The calculation programs (creatinine clearance, ideal body weight) were found to be useful by the pharmacist and some residents. The ventilator weaning protocol was used by medical staff, as well as respiratory therapists, allowing regular assessment of whether patients met the criteria for extubation. Many databases were, however, not fully used. This appeared to relate more to inadequate training than to faults in the databases. In many cases, the PDA users were unaware that certain information was in their PDAs. This was because data were located on separate software programs (J-file, AvantGo, Cbas, Memopad) and may have been difficult to find. The PDA had a global 'Find' function to search for keywords, but this does not incorporate some of the added software programs, such as AvantGo. A unified database program with a search capability was suggested as a useful addition. Patient-management software Patient information was managed using the text-based MemoPlus software and a customized template. This required text entry on the PDA. Several modifications to the template were made during the study period. Residents responsible for patient data entry described difficulty entering data for new patients and keeping patient information updated during busy weekends. Attending staff found the patient data useful, particularly when they were taking over care of patients at the beginning of their on-call duties. Transferring the care of critically ill patients to a new physician is time-consuming and potentially stressful. The PDA patient database improved the staff's knowledge of patients, especially of previous medical problems in patients with complex conditions who had had a long stay in hospital. It also gave staff access to patient information when they were out of the ICU, aiding decision-making. During ICU rounds, the summarized chronological information was useful to find out how long intravenous lines had been in place and to review antibiotic therapy. Less benefit was noted in short-term patients. During night call, the patient summaries were of value when residents were called to see patients with whom they were not very familiar. In our ICU, a daily physician note is written in the patient record. The print function to create a daily note reduced duplication of work, but the process for entering patient data was found to be time-consuming initially. While residents did not feel that the patient-management application (MemoPlus) improved efficiency, it did increase their knowledge of the patients. During the study period, other commercially available patient-management software systems were evaluated. These had the advantage of easy data input using single keystrokes for date entry and 'pop-up' lists of drugs and diagnoses. While this simplified data inputting, no system was found to be ideal for the ICU. Many of these systems did not support the infrared data transfer or printing functions. Other uses of the software Study participants used a variety of other applications on a regular basis. Having the call and teaching schedules easily accessible was considered a benefit. The telephone list of hospital numbers was found to be valuable and the To Do list was used by most users to keep track of their work. Teaching rounds and morbidity and mortality rounds were facilitated by using archived patient data. Many participants used the Memopad to take notes in teaching seminars. Suggestions for change The focus-group discussions generated a number of suggestions for improvement. The hardware unit was considered suitable, but a more robust one may be needed in view of the two damaged screens. Because most of the users had had no previous experience with the PDA, additional teaching sessions and follow-up training were suggested to make optimal use of the technology. This would have helped users to become more aware of the many databases available on their PDA. In this regard, the medical information on the PDA would clearly benefit from integration into a single, searchable program. The patient-management software would be more user-friendly if the data could be entered with minimal effort, using customized pull-down lists of drugs, diagnoses, and procedures. The demographic data could be entered and updated daily by a ward clerk. Alarms were suggested - for example, to warn of prolonged intravenous line duration or the end of a course of antibiotic therapy. While transmission of data between staff by infrared was found to be useful, synchronization with the hospital electronic patient record was considered the optimal situation. Objective evaluation Two groups of four trainees took part in each crossover study. Half of the residents had prior experience with PDAs. No difference was noted in their subjective preference for the PDA or printed copy of the handbook, and the individual's preference did not correlate with previous PDA experience. Comparison of the test scores revealed no difference between the scores in the PDA-assisted test and the paper-assisted test, analyzed after correction for difficulty using the control mean and standard deviation. Discussion This study prospectively evaluated the benefits and drawbacks associated with the introduction of handheld computers in an academic-critical care environment. Who benefitted most? The introduction of handheld computers was well received by all users, despite differences in their familiarity with these devices. The most favourable response was from the more senior staff, namely, the attending physicians and fellows. This may be because of the longer time they were involved in the study, allowing more familiarity with the PDA platform. They were also more likely to benefit from having patient data available while on call outside the ICU. Furthermore, they were usually not responsible for entering patient data. Clearly, two conditions that might enhance the acceptance of these technological changes are adequate education and ease of data entry. Although an initial education session was held, it was when the junior medical staff in the study were beginning their rotation in an unfamiliar environment. Making the devices more user-friendly The patient data applications assessed were not ideal but did enable us to identify several criteria for a user-friendly system. These include ease of data entry using shortcuts and lists, limiting the range of data stored to that essential for patient management, and the ability to transmit data easily between staff. It is important that this computerized patient database should decrease workload and not cause duplication in work. In our study, enabling residents to print a daily note from their handheld computer offset the additional work of data entry. Ideally, the handheld system should be integrated with the hospital electronic patient record, allowing direct entry of demographic data as well as access to laboratory data. A wireless capability may also have significant benefits with respect to medical information databases. This would allow access to Medline searches and evidence-based guidelines. While internet access is available from desktop computers in the ICU, the ability to perform these searches on rounds or while consulting outside the ICU may be beneficial. Databases on paper or on screen? Critical Care Handbook of the Massachusetts General Hospital What is needed 13 Abbreviations HTML = hypertext markup language; ICU = intensive-care unit; IrDA = infrared data association; Mb = megabytes; PDA = personal digital assistant.