Point-of-Care Testing vs. Laboratory Testing during High Patient Volume Situations ()
1. Introduction
Our nation’s Emergency Departments (EDs) are in a state of crisis [1]. Crowding and long wait times have become commonplace. This has led to a demand for improved process efficiency and decreased turnaround times (TAT). Point-of-Care-Testing (POCT) has the potential to help accomplish this goal, since the accuracy of these devices has been shown to be equivalent to in-laboratory testing [2]. Decreased ED TAT in selected patient populations may help alleviate ED crowding [3] [4].
Several studies support the use of POCT for cardiac markers in patients with chest pain [5] - [9]. However, POCT has not been specifically compared to in-laboratory testing in high volume busy or mass casualty situations [10]. For example, even if the benefit can be shown in individual patients, what happens when there is a large volume of patients that require testing at the same time? Will the tenth POCT sample have the same logistic benefit as the first or will the laboratory’s capacity to run several samples simultaneously result in a quicker result? How many POCT machines would be required to match the laboratory’s ability to accommodate a high volume situation?
Given the need to develop strategies to adapt to high volumes, the results of this study could potentially identify a way to improve efficiency under these circumstances. Several studies have demonstrated that POCT decreases length of stay in the ED, providing evidence that POCT can alleviate ED crowding [11] [12] [13]. However, we acknowledge that having results faster does not always translate to short ED stays.
2. Objective
To evaluate the effectiveness of the POCT during high volume situations, POCT was compared to in-laboratory testing during busy periods with large numbers of patients.
3. Methods
3.1. Study Design, Setting, and Selection of Participants
This study was completed at an urban level 1 trauma center with an academic emergency medicine department with a residency and annual patient volume of 70,000. Blood samples requiring laboratory testing during the afternoon peak volume between 11 a.m. and 7 p.m. were included over a five-week period beginning January, 2012. These hours were chosen since review of Emergency Department Information manager (EDIM) demonstrated the highest volume of patients within these hours.
3.2. Methods, Measurements, and Outcomes Measured
Blood samples were drawn in the usual manner. One tube of blood was tested in the ED using POCT with the GEM Premier 4000 (Bedford, MA) and the others were sent to the laboratory for testing. The GEM Premier 4000 has a limited laboratory panel, including pH, pCO2, pO2, Na+, K+, Ca++, Cl−, glucose, lactate, hematocrit, total hemoglobin, oxyhemoglobin, carboxyhemoglobin, methemoglobin, deoxyhemoglobin, and total bilirubin. TAT was measured as time from when the tube was received to when the result was available, and each test was captured and compared. POCT was completed in the ED by a laboratory specialist stationed in the ED to complete the POCT and ensure all the appropriate quality control maintenance was done. All patients were treated in the usual manner of care. The study received a non-human subject research exemption from the Institutional Review Board.
A supplemental time-motion study was completed to detail the steps involved in lab processing, characterize the manual labor involved in laboratory tests, and estimate the number of POCT machines that would be necessary to handle the complete hospital volume in an hour.
3.3. Analysis
We compared POCT times from receipt to results for complete blood count (CBC) and chemistries with in-laboratory CBC and chemistry times. If the in-laboratory result time preceded the documented time the sample was received, the sample time was “negative”. It was assumed that the time of receipt was input incorrectly, and the observation was excluded. This was to account for samples dropped off to the POCT station prior to nursing completing collection status in the computer and sending samples to the labs. This would overestimate POCT times. If either the POCT or in-laboratory sample was not received, the observation was excluded. Missing values were not imputed. Descriptive statistics were used to summarize the sample distribution. For both POCT and in-laboratory testing, the numbers of TAT greater than 60, 90 and 120 minutes were calculated. Rank sum tests were used to assess significance of differences in processing time between in-laboratory and POCT.
We also performed a sensitivity analysis to determine whether times for in-laboratory CBC and chemistry labs were longer (relative to POCT) during high volume ED activity. We used scatter plots to examine the association between differences between in-laboratory CBC and chemistry and POCT processing times as a function of number of labs processed in an hour. Weighted linear regression analyses formally examined these associations, where weights were set equal to the number of labs samples received. Observations were grouped into hours of the day the labs were received and mean processing times calculated for each hour compared with the number of samples received. The analyses were repeated using day rather than hour as the unit of accumulation.
4. Results
Over a five-week period, 569 blood test samples were collected during times of peak ED volume. Of 569 observations, 30 (5.3%) had a negative number or a missing sample pair and were excluded. All analyses were conducted on the remaining 539 observations.
The POCT group was significantly faster than in-laboratory testing, with a mean POCT turnaround time (TAT) of 3.5 minutes, compared with in-laboratory CBC test time of 30.9 minutes and blood chemistry test time of 55 minutes (Table 1). Of note, while mean times were less than 60 minutes, in-laboratory testing frequently took longer than 60 minutes as described in (Table 2). Our sensitivity analysis found that as the number of samples increased and reached a peak, there was a slight decrease in POCT TAT, but the difference was insignificant. There was no time when in-laboratory testing was faster than POCT.
Since the peak volume of samples per hour did not sufficiently stress either method, we performed a time-motion study to determine the point at which POCT becomes less efficient than in-laboratory testing. The maximum number of samples per hour on a single POCT machine was 17 samples at 3.5 minutes per sample. There were several variables that determined the maximum number of in-laboratory samples per hour. The time-motion study assessed all the steps and time involved in getting specimens from blood draw in ED to result in the laboratory. We found that specimens the laboratory received from the ED were not run separately from those received from the rest of the hospital. Therefore, although we only assessed the TAT for ED specimens, the laboratory was processing many more specimens during the same hour. So, while our highest volume was 8 samples from the ED in an hour, the laboratory received an average of 54 specimens per hour (Figure 1).
If we were to assume all 54 samples arrived at once, similar to a theoretical surge or mass casualty situation, it would take 3 hours to process the 54 specimens using only POCT even if single sample duration remained only 3.5 minutes. Alternatively, 3 POCT machines would be necessary to process all samples within an hour. POCT has the potential to be associated with delay during surge or batch testing situations.
![]()
Table 1. Mean and standard deviation of turnaround times (TAT) and distribution of processing turnaround times within 25th, 50th, or 75th percentiles for point-of-care testing (POCT) compared to in-laboratory complete blood count (CBC) and chemistry (Chem) testing. N = 539 observations. POCT processing times were significantly less than either in-lab CBC or Chemistry processing times, both p < 0.0001 using Sign Rank test).
![]()
Table 2. Percent of times when testing modality took longer than 60, 90, or 120 minutes to complete; machine run and manual labor times from detailed time motion study machines that would be necessary to handle the complete hospital volume in an hour.
a. Shorter CBC run times compared to the prior observations reflect changes in laboratory procedures; b. turnaround time (TAT); c. standard deviation (SD); d. eligible N (eN) refers to 60, 90, 120 minute distributions.
![]()
Figure 1. Number of samples processed per hour. The laboratory receives 75 samples per hour during peak times of 5 a.m. to 9 a.m. when all inpatient samples are run and turnaround times are extended. During routine times, the laboratory processes about 54 specimens an hour. If point of care testing (POCT) were to handle the same volume at its maximum rate of 17 samples per hour, POCT could handle the total hospital morning surge volume with 3 systems, with turnaround times measured in minutes rather than tens of minutes.
5. Discussion
We found that during routine ED high volume situations, POCT provided results significantly faster than in-laboratory testing. We also found that if POCT were used for total hospital laboratory testing volumes approaching 54 samples an hour, 3 POCT machines would be necessary to maintain rapid turnaround times for most specimens. While POCT holds promise as a solution for fast specimen processing during high volume situations, some caution is advised because most institutions will not typically have the number of POCT systems necessary to maintain fast processing for all specimens within a short period. Further, if samples arrived in a significant batch such as from a mass casualty incident rather than a more evenly distributed daily census, scalability of POCT may be further stressed. However, these concerns remain theoretical given the moderately limited volumes of specimens available in our study.
The time-motion study found that in most instances, the observed manual labor was not very different between POCT or in-laboratory testing. However, in-laboratory TAT was hampered by the many other hospital samples being run simultaneously, with placement of the just arrived sample in a relatively long automated que tray on its way to the laboratory testing machine. This step allowed the laboratory technician to complete other tasks, increasing their efficiency but potentially accounting for the wide variability in laboratory median run times.
Planning for mass casualty incidents perhaps needs to include discussions with the laboratory to limit testing to specimens from the ED and ICUs to temporarily improve system efficiency. If POCT will be the solution to meet high volume demands, the number of systems that may be required in a mass casualty bolus situation should be considered. Moreover, unlike during this trial, in many institutions with POCT, nursing staff perform the test, which may impact their ability to perform other tasks during mass-casualty situations. Nevertheless, in our study, POCT significantly outperformed in-laboratory testing during routine high patient volume situations, affirming its diagnostic efficiency.
We acknowledge a number of limitations. During our busiest high volume times, neither laboratory testing nor POCT system was sufficiently stressed to simulate a mass casualty incident. We attempted to account for this with an additional time motion study and sensitivity analysis. Due to lab concerns for maintenance of quality control, laboratory personnel performed the POCT within the ED, which may not be generalizable to all hospitals.
Additionally, the nursing workflow mechanism for documenting time of collection in our electronic medical record may have affected the actual collection time of some specimens as they may have documented collection well after specimens were already in the laboratory, creating a negative number and causing their data to be excluded from analysis. Rarely, some specimens were sent to the laboratory without a matching POCT specimen and their time data was excluded from analysis.
We anticipated a learning curve in the beginning of the study, with the expectation of increasing samples collected progressively over a five-week period. However, we found the number of samples collected each week decreased, possibly revealing a selection bias. For example, if a patient appeared pale the nurse may favor CBC POCT to obtain faster results to expedite care. In addition, this study excluded weekends and samples from the pediatric emergency department. The blood samples were collected based on our busiest hours, calculated from our ED electronic medical record system. We did not adjust for the day of the week, month, or season, which may have provided a higher patient volume.
6. Conclusion
Point-Of-Care Testing was significantly faster in providing results compared to in-laboratory testing during routine high patient volume situations. However, during a true mass casualty incident requiring simultaneous blood specimen testing, multiple POCT machines may be necessary. Future research on POCT should take into account the effects of workload on staff who will actually perform the test and how that may affect their availability of other duties.
Acknowledgements
1) This project was sponsored by the University Center for Disaster Preparedness and Emergency Response. This study received funding from the US Department of Defense Grant No. W9132T-10-1-0001. The views, opinions, positions, conclusions, or strategies in this work are those of the authors and do not necessarily reflect those of the Department of Defense or any agency of the U.S. Government and no official endorsement should be inferred.
2) Jonathan McCoy received funding from Boehring-Ingelheim for REVERSE AD site-related costs which do not relate to this work.