| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Address correspondence to Dr Sail Chun, Department of Laboratory Medicine, Asan Medical Center, 388-1 Pungnap-dong, Songpa-gu, Seoul 138-736 (South) Korea; tel 822 3010 4510; fax 822 478 0884; e-mail sailchun{at}amc.seoul.kr.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Keywords: critical value alert, short message service (SMS), personal data assistant (PDA) telephone
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Usually, in the case of an inpatient, the laboratory technologist telephones the nurses on the patient’s floor and asks them to notify the responsible physician of the patient’s critical value. However, this communication process can be delayed at times. The Joint Commission on Accreditation of Healthcare Organizations (JCAHO) requires health care organizations to track and improve the timeliness of the transmission and receipt of critical test results by the responsible licensed caregiver [3]. The Laboratory Authentication Program of the Korean Society for Laboratory Medicine requires that caregivers be promptly alerted to their patient’s critical values [4], and the parameters related to the transmission of critical value information are considered to be an important outcome measurement of clinical effectiveness and laboratory efficiency.
The Asan Medical Center (AMC) is a 2,200-bed tertiary care academic medical center in Seoul, Korea. Until early 2005, if the Laboratory Information Service (LIS) automatically flagged a test result that required a critical callback, the laboratory technologists (in the case of a clinical chemistry unit) monitoring the LIS would make a phone call to notify the appropriate individuals of the patient’s critical value. The phone call placed to the patient’s location was documented at the location receiving the phone call and in the laboratory. In 2005, the Department of Laboratory Medicine at AMC designed and implemented a Short Message Service (SMS) alerting system for notifying physicians of their patients’ critical values with a PDA phone. This is a real-time critical value alerting system that executes alerting algorithms on the LIS, and it has been added to the existing callback system.
At the AMC, serum or plasma calcium, potassium, sodium, chloride, magnesium, glucose, digoxin, urine ketone in children, and newborn screening tests have been listed as clinical chemistry tests that have critical value limits. We selected and studied critical hyperkalemia from among them because using the electronic medical record (EMR) database, it was simpler to recognize the clinical response to hyperkalemia than to the other tests. This paper discusses the clinical usefulness and effectiveness of the SMS alerting system for reporting critical values using the PDA phone in patients with critical hyperkalemia.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Alerting system used in 2005: SMS in addition to callback. The primary tool used for software development of the SMS alerting system for critical value notification was Microsoft Visual C++ 6.0 (Microsoft Co., Redmond, WA, USA). The database to store information regarding the development of the procedure was Oracle 8.0 (Oracle Co., Belmont, CA, USA). At the AMC, all physicians (including faculty and residents) carry a hospital-owned PDA phone that automatically notifies them of critical values using the SMS alerting system as soon as the laboratory result is entered into the LIS. Each PDA phone has a unique identification number. As in the previous system, a laboratory technician validates the laboratory test results, and the laboratory result and related information from the LIS are transmitted through the network and stored on the alerting system server. Then, the alerting system screens and detects all critical values that need to be reported. If a result is determined to be critical and worthy of notification by the alerting system’s algorithm, messages concerning the patient’s critical values are generated and transmitted to the SMS server through the Hospital Information Service (HIS) network. After that, an SMS text message containing the patient’s critical value is transmitted to the PDA phone of the appropriate physician. The SMS message contains the name of the faculty physician in charge, the patient’s name with identification, the laboratory test code, the name of the test, and the test result itself.
In an effort to make critical value alerting more effective and efficient, the system was not intended to be used to alert physicians of all critical laboratory results. Alerting algorithms continuously monitor new patient data to determine whether they satisfy any of the alerting algorithms criteria. Using the algorithm, physicians are only alerted in cases of either non-successive critical values, regardless of the time interval, or successive critical values over a time interval of more than 2 weeks. Consecutive critical values determined in a period of less than 2 weeks are not reported because the status of the patient is presumed to be already known to his or her doctor. For example, if a serum potassium level rises to a value >7.0 mmol/L, a second test detecting a similar value will not trigger a second alert if the test is performed within 2 weeks of the previous test. If a similar potassium level is detected after 2 weeks, the elevated potassium level is considered to be another event or the recurrence of a critical situation, and it will trigger a second alert. The phone call to the floor is not replaced by the SMS alerting system and the nurse still notifies the physician of the patient’s critical value. Using the SMS system, the physician can obtain the critical value of the patient at the same time as the nurse, or even a little earlier.
Data collection From the HIS database, we obtained and analyzed data on the clinical response rate and clinical response time to information regarding the critical values of patients with hyperkalemia in general wards and intensive care units (ICUs) during two 12-mo periods: the first was from 1 January 2001 to 31 December 2001, when only the callback system was being used, and the other was from 1 July 2005 to 30 June 2006, when SMS alerting was added to the callback system.
A clinical response occured when a physician entered an order for appropriate treatment for the critical value into the HIS. The clinical response rate was defined as the frequency of clinical responses divided by the number of critical value alerts during a given time period. In other words, the clinical response rate corresponds to the fraction of critical hyperkalemia alerts to which the physicians responded by placing a treatment order into the HIS. The clinical response time was defined as the length of time from the release of the SMS message containing the critical value information to the time when appropriate treatment was ordered in the HIS. The upper limit of the critical value for serum potassium (critical hyperkalemia) was defined as any value >7.0 mmol/L.
Orders for the initiation of various treatments, eg, regular insulin with glucose, calcium gluconate, polystyrene sulfonate calcium, furosemide, and sodium bicarbonate, or for the discontinuation of various therapeutic agents, eg, potassium chloride and spironolactone, were considered valid clinical responses to the hyperkalemia. To measure the impact of each system on the newly recognized critical laboratory value results alone, we excluded alerts for which treatment had been initiated before the critical value was reported.
Instances in which no treatment was administered within 24 hr after the critical laboratory value alert was sent were also excluded because they were deemed to be non-representative of the effect of the SMS system for critical value notification. For example, a physician might decide to order a treatment other than those we defined to be appropriate treatments.
Statistics Statistical analyses were performed using SPSS 12.0.1 for Windows (SPSS, Chicago, IL, USA) and Excel 2003 (Microsoft). The clinical response rate was analyzed using the Chi-square test, and the clinical response time was determined using the Mann-Whitney U-test. A p value <0.05 was considered statistically significant.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). This represented approximately 0.08% of all serum potassium test results reported. Of the 226 alerts of critical hyperkalemia, 124 (54.9%) were for patients in intensive care units (ICUs) and 102 (45.1%) were for patients in general wards. Sixty-one of the alerts were excluded from the clinical response analysis because the patients received treatment for the condition before the alert was sent, and 44 were excluded because no treatment was given for the condition within the following 24 hr.
|
). Fifty-seven of these alerts were excluded from the clinical response analysis because the patients had received prior treatment for the condition, and 25 were excluded because no treatment was given for the condition within the following 24 hr.
Therefore, we analyzed 121 critical hyper-kalemia alerts from 2001 and 96 from 2005 (Table 1 and Fig. 1).
|
, Fig. 2). The median and interquartile range of the clinical response time for patients in general wards were 249.0 and 548.0 min in 2001 and 63.0 and 193.0 min in 2005, respectively (p <0.001) (Fig. 3).
|
|
|
|
). The data for patients in the ICUs were divided into medical, surgical, and pediatric ICUs and then reanalyzed, but no significant differences were found between the data from each ICU in 2001 vs 2005 (p = 0.508, 0.791, and 0.698, respectively).
|
).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A survey in 2001 of hospitals in the United States revealed that the primary means of critical value notification in most institutions was the use of telephone calls to the physician and nursing personnel responsible for the patient’s care. The use of electronic devices that automatically dial the physician’s beeper has the advantage of reducing the amount of time laboratory technicians spend on the telephone with physicians and improves the likelihood that the physician or another person in charge of the patient’s care actually received the result [5]. Various types of automated wireless alert systems have been used to prevent medical errors and report critical laboratory values to physicians [6]. Even though these wireless technologies still cannot provide the rich and varied details given to physicians by nurses [7], they are becoming useful methods for delivering patient information quickly to the responsible doctor. Physicians are becoming accustomed to using portable technology in daily care. Mobile devices linked to clinical information systems can provide real-time event notification to health-care workers [8–10].
Chen at al [11] reported that physicians prefer to use pagers and mobile phones as their alerting tool. Personal digital assistant (PDA) systems are advantageous over other mobile devices because they allow physicians to access clinical information wirelessly at the point-of-care [6]. Shabot and LoBue [10] were the first to send an alphanumeric text alert message directly to the responsible physicians from a clinical system via an early style PDA, bypassing nurses and laboratory technicians entirely [10]. The PDA phone, which is an evolved form of a PDA, has not only the function of a handheld computer but also that of a cellular phone. Even though permitting entry wherever it is available allows for greater efficiency and significant error reduction [6], patient information or an order are not allowed to be entered at the point-of-care through a PDA phone at the AMC due to security concerns and the narrow width of the screen, which is inconvenient for a full and detailed review of a patient’s EMR.
The use of wireless technology in physicians’ daily work promotes quicker event notification and faster delivery of information regarding critical events with reduced error in communication. Moreover, instead of waiting for nurses to notify them of abnormal test results, the fellows and residents must decide whether the alert requires immediate attention. Therefore, the use of such a thing makes the staff more proactive instead of reactive in patient care [7].
The SMS alerting system for critical values that we have designed is accompanied by an alerting algorithm so that the users (the laboratory technicians and physicians) can obtain more meaningful patient information. Howanitz et al [5] reported that the majority (71.4%) of the institutions studied did not have a policy for reporting repeat critical values for the same analyte on the same patient, although in some institutions (11.6%), the policy was that physicians were not alerted of repeat values. Wagar et al [12] examined the factors associated with improvement in critical value alerting in multiple institutions over time. One of the factors analyzed was the mandatory practice of requiring health care providers to be notified when handling inpatients known to have results repeatedly in the critical range. In clinical settings in which a critical test result is expected, alerting the appropriate individuals of this value may not contribute to improved patient care.
Critical value alerting on all occasions that a test result is beyond the critical limit may dilute the urgency of the alerts and may unnecessarily interrupt physicians. Especially, repeatedly alerting a physician of critical values already made known a few days prior will annoy him or her and may cause future alerts to be ignored. Therefore, the alerting algorithm used in our SMS alerting system for critical value notification was designed to take the current test result, the results of previous tests, and the length of time between the tests into consideration before sending an alert message. Moreover, the option for customized alerts and the use of a rule editor would enable us to change, add, and review the different values used to trigger an alert [13].
In the near future, alerting software should permit highly nuanced approaches to critical value alerting, a physician-specific critical values list, and context-sensitive critical value alerting, even though this would currently be constrained by regulations that require all critical results to be communicated [14]. It remains uncertain which policy appproaches on how to alert physicians of repeatedly occurring critical laboratory values is most beneficial to patient outcomes.
Iordache et al [13] made panic time rules that were similar in concept to the alerting algorithm designed in the present study. In their system, the repeated occurrence of a critical result during the panic time period would not trigger an alert, but would trigger an alert after this period because such a result may indicate the recurrence of a previously treated problem. They considered the panic time to be the length of time needed for a critical situation to be resolved and to return to within the critical limits. The panic time for hyperkalemia was 1 hr [13]. In contrast, we implemented a 2-week panic time period in our system.
In the present study, we found that SMS alerts sent to PDA phones had a remarkable impact on the care of patients with critical hyperkalemia. In 2001, when the AMC used only the callback system to alert physicians of critical laboratory values, nurses initially received the phone call alert of a critical value from the laboratory technician, and in turn notified the physician of the critical value. Not only was this process more time consuming than directly alerting the physician, but there was also the possibility for communication errors to take place. In 2001, the mean and median clinical response times for inpatients were 343.3 and 213.0 min, respectively. In 2005, after SMS was added to the callback system, the response times decreased by 40.8% and 65.0% to 203.2 and 74.5 min, respectively. Since the decreases in clinical response time were so remarkable, it was considered to be one of the most important factors in caring for patients in critical situations.
Kuperman et al [15] reported that the mean and median clinical response times to telephone-based alerts for critical values of sodium, potassium, glucose, and hematocrit for inpatients were 4.6 and 1.6 hr, respectively. Following the addition of an automated system for sending alerts directly to the physician’s pager, the mean and median physician response times were 4.1 and 1.0 hr, respectively. The decreases in mean and median response times were both statistically significant (p = 0.003) [15].
There are several differences between our study and that of Kuperman et al [15], including the tests considered, the critical value limits, the number of patients for whom a doctor was responsible, and the study design. The most significant difference was the way in which physicians were alerted. Under the system developed by Kuperman et al [15], a notification program automatically paged the patient’s physician. Then, the specific callback number was displayed on the digital pager, which only indicated that an automated alert had been generated for one of the physician’s patients. After noticing that a special code has arrived on his or her pager, the physician had to log on to an available computer workstation to review the alert [15]. In a study by Rind et al [16], e-mail alerts of rising creatinine levels were shown to reduce the relative risk of developing serious renal impairment in hospitalized patients receiving nephrotoxic medications, and the medications were adjusted or discontinued an average of 21.6 hr earlier than when no e-mail alerts were delivered.
In our study, the mean and median clinical response times in general wards in 2005 decreased by 54.3% and 74.7%, respectively, in comparison to the response times in 2001. This decrease was statistically significant. On the other hand, the mean and median clinical response times in ICUs in 2005 decreased by 11.8% and 51.8%, respectively, in comparison to those in 2001. The decrease in the clinical response times to critical value alerts was less dramatic for patients in the ICUs than for those in the general wards and was not statistically significant. There are several possible explanations for this discrepant result. First, under the callback only system used in 2001, critical test results may not have directly reached the responsible physician but may rather have been communicated to him or her via a nurse, especially in the general wards. Therefore, the length of time for a critical result to be communicatied from a laboratory technician to a nurse and then from the nurse to a responsible physician might be longer in non-ICU floors than in ICUs. Second, the physicians in charge were more available in the ICU setting. Third, the patients in the ICUs had complicated clinical situations requiring prudence in choosing the best treatment option and may have involved consultations with other departments. However, it was unlikely that the volume of information processed by the LIS/HIS and transported to the physician was so large as to hamper the reduction of clinical response time for patients in the ICU.
The clinical response rate to critical hyperkalemia was 73.3% in 2001 (callback only) and 79.3% in 2005 (SMS with callback), but this difference was not statistically significant. This finding suggests that the physicians were aware of the critical situation and responded to it in a similar manner, regardless of the mode of alert. Howanitz et al [5] reported that therapy was influenced by the critical values in 64.9% (by chart review) or 62.9% (by physician survey) of cases. Tate et al [17] found that only 50% of physicians responded to a critical value alert in an appropriate manner. However, it is difficult to compare the clinical response rate determined in our study with these authors’ findings due to the differences in study design.
It takes considerable time and labor for laboratory and nursing departments to notify physicians of critical values using the callback system. Howanitz et al [5] proposed that the critical value alerting system should be automated in order to conserve early response time. The SMS message to the PDA phone is one of the various automated tools used to alert physicians of critical values. At the AMC, we still use both the SMS and callback systems to alert physicians of critical values because SMS reception is not always consistent, and it is difficult to confirm that a message was in fact received by the physician. We hope that in the near future, by the advent of bidirectional communication through PDA phones, the SMS critical value alerting system will completely replace the traditional callback system.
To measure the effects of alerting systems like the one described in the present study, a controlled, randomized trial in which physicians receive critical value alerts with or without SMS message being sent to physicians’ PDA phones would need to be conducted. However, such trials may be difficult to justify ethically considering the critical nature of the events reported using these systems [18]. The clinical response time and rate to critical value alerts for analytes other than potassium would not be the same as in the current study of critical hyperkalemia. Evaluations of the effectiveness of critical value alerts with SMS messages for the other listed analytes remain to be done. A study by Shea et al [19] showed that a computer-generated informational message sent directly to physicians shorted the patients’ length of stay (LOS) [19] and reduced medical costs. Our SMS system for alerting physicians of critical values using an alerting algorithm is expected to have beneficial effects on patient care, LOS, and patient survival, but the parameters used to assess patient outcomes need to be measured in future studies.
In conclusion, under the SMS alerting system for critical value notification using the latest messaging device, the PDA phone, the length of time was decreased between the point at which critical laboratory results were directly communicated to the responsible physician and the point at which an appropriate treatment was ordered. The mean and median clinical response times with the SMS alerting system were decreased by 40.8% and 65.0%, respectively, in comparison to the results obtained using only the callback system.
|
Acknowledgments |
|---|
This work was supported by grants from the Korea Health Industry Development Institute.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. W. Armstrong, A. J. Watson, M. Makredes, J. E. Frangos, A. B. Kimball, and J. C. Kvedar Text-Message Reminders to Improve Sunscreen Use: A Randomized, Controlled Trial Using Electronic Monitoring Arch Dermatol, November 1, 2009; 145(11): 1230 - 1236. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
