Root cause analysis of laboratory turnaround times for patients in the emergency department

ED Administration

CJEM 2004;6(2):116-122

Abstract

Résumé

Introduction

In this time of emergency department (ED) overcrowding, efficient process and patient management during the ED visit may lessen the impact of overcrowding on strained resources.^1,2 Shortened waiting times for test results could potentially allow patients to be discharged in a more timely fashion, making ED stretchers available for patients waiting in hallways and on ambulance gurneys. Laboratory turnaround time (TAT), the time from physician ordering of an investigation to the return of test results, is a key measure of an important ED process. Many believe that prolonged laboratory TATs cause treatment delays and protracted ED stays.³

The College of American Pathologists Q-Probes Study summarized data from 700 institutions and 40 000 specimens per analyte.⁴ It identified TAT from phlebotomy to reporting of results as the most important performance measure for stat laboratory testing and documented a median inter-institutional TAT for hemoglobin of 25 minutes and a median TAT for potassium (which is considered a surrogate for all tests analyzed on serum or plasma) of 36 minutes. The Q-Probes report focused on laboratory performance and did not consider the interval from physician ordering to phlebotomy. This additional interval may be important when trying to determine whether causes of delay relate to the ED, the laboratory, or both.^5,6

The objectives of this study were to identify time intervals from physician ordering to result reporting for hemoglobin and potassium assays, to compare these times with established benchmarks from the Q-Probes Study and to identify the root causes of delays.

Methods

Design

Root cause analysis is an investigative method that tracks each of the relevant steps in a process and traces them back from the end failure to determine the initial or root cause.⁷ The 3 steps in RCA are to identify potential causes of delay using a cause-and-effect diagram, to collect data related to these potential causes, and to identify root causes based on these data that should clarify the impact of the causes on the particular process. A key component is the use of Pareto analysis, which specifies that 80% of the problem will often be the result of 20% of the causes and encourages the analyst to focus on the "vital few" rather than the "useful many."⁸ Wilson and colleagues provide an excellent overview of root cause analysis for interested readers.⁷

Setting

This study was conducted at Hamilton General Hospital Emergency Department of Hamilton Health Sciences, an acute care tertiary teaching hospital affiliated with McMaster University, Hamilton, Ont. The hospital serves as a regional cardiovascular, neurosciences and Level I adult trauma centre. The ED, staffed by full-time emergency physicians, has an annual volume of 39 500 patient visits, 50.8% of which are triaged as emergent or urgent (i.e., CTAS Levels I and III),⁹ and a 17.4% admission rate.

Sampling

This cross-sectional study examined TATs for all stat hemoglobin and potassium measurements during an 8-day period. Potassium analyses were performed using a Roche Integra 700 (Roche Diagnostics, Laval, Que.), and hemoglobin analyses were performed using a Coulter Gen-S (Beckman Coulter, Inc, Brea, Calif.). Blood samples from critically ill patients were excluded because of concerns about the accuracy of collection-time documentation in such cases. Blood tests submitted as "routine" were also excluded because our ED samples are ordered as "stat" or "urgent," and because benchmarks established in the Q-Probes Study were derived from stat, not routine, testing. Patient identifiers were removed from each sample and replaced by sequential numeric identifiers to ensure confidentiality.

The hemolysis rate was recorded because hemolysis may induce falsely elevated potassium levels, at times necessitating repeat collection of blood specimens. The proportion of potassium samples hemolyzed is reported as percent of total potassium samples.

Fig. 1. Key steps in the laboratory process

enlarge

Data collection

A flow chart was created to identify key steps in the laboratory process (Fig. 1). The key steps and times for this study were: physician orders test on paper (T1); bedside nurse processes physician orders and enters sample information in the hospital information system (T2); bedside nurse collects blood sample (T3); sample placed in tube system and transported to laboratory (T4); sample logged into the laboratory computer (accession time = T5) and queued for placement in the instrument; samples placed in instrument (T6); results reported on the hospital information system (T7). This allowed for the measurement of 7 event times (i.e., 6 time intervals [Table 1]) and overall TAT.

T1 through T6 were registered on the data collection form via synchronized time stamps, and T7 was collected from the hospital information system. Overall TAT incorporated all of these, as per the Q-Probes Study, with the addition of the physician-order time (T1). For the data collection period, every effort was made to blind the ED and laboratory staff to the objectives of the study.

Analysis

Sample size determination for this study was based on a previous 1-week study of stat hemoglobin and potassium TATs.¹⁰ Central tendency and dispersion were described using median values and interquartile ranges (IQRs). Where appropriate, 95% confidence intervals were calculated to illustrate data precision. A log transformation (ln [minutes + 1]) was performed to normalize TAT data. Transformed data were compared with data from the Q-Probes Study using 1-sample t tests minus T1. Our ethics review board considered this a quality assessment study involving no human subjects or experiments and ruled it exempt from formal approval.

Results

Of 837 patients registered in the ED during the study period, 207 had their hemoglobin or potassium measurements checked. Overall, 43% of the 207 samples were drawn on patients who had been admitted but were still in the ED waiting for a bed. Figure 2 summarizes eligible, excluded and missed patients (i.e., patients not identified by the laboratory as having had these measurements checked), and shows that 147 stat samples were analyzed. The median TAT was 18 (IQR 13-26) minutes for hemoglobin and 49 (IQR 42-58) minutes for potassium. Table 1 shows that hemoglobin TAT was significantly less than the Q-Probes benchmark (18 v. 25 min; p < 0.001) and the potassium TAT was significantly greater (49 v. 36 min; p < 0.001). Our data reveal that 75% (105/140) of hemoglobin samples and 10.7% (13/121) of potassium samples met the Q-Probes benchmark time. Hemolysis was noted in 7 potassium samples (5.8%), but only 2 of these required repeat sample collection.

Fig. 2. Flow chart of emergency department patients for whom potassium or hemoglobin measurements were tested

Root cause analysis demonstrated 3 root causes of delay -- order processing for both hemoglobin and potassium, volume of tests for admitted patients, and queue and instrument times for potassium -- although we could not determine the proportionate contribution of these 3 main causes to the overall delay. A barrier analysis (Table 2) provides insight into the contribution of the work environment and existing processes to TAT delays.

Discussion

These data show that the median TAT for hemoglobin met the Q-Probes benchmark in 75% of cases, but this was the case for potassium in only 10.7% of cases. The main contributors to delay were processing times, admitted-patient testing and laboratory queue and instrument times. Unlike previous studies, hemolysis was not a significant contributor to laboratory delays.¹⁰

Test processing times

Order processing is identified as a root cause of delay for both tests. The Leapfrog Group¹¹ recently identified computerized physician-order entry as an important mechanism to improve patient safety. As more hospitals move to computerized physician-order entry, processing times should decrease and patient safety should improve, based on the elimination of illegible physician orders and transcription errors.

Although Q-Probes did not assess time from order to processing and collection, we believed it could be an additional cause of delay. Our root cause and barrier analysis suggests that reliance on ED nurses having the time to perform phlebotomy may further prolong the processing. It may be that use of dedicated personnel for phlebotomy would reduce order-processing time. Excessive queue and instrument times suggest the need for process change in the laboratory for chemistry analysis. Finally, the workload of caring for admitted patients who are still in the ED (43% of stat ED lab tests in this study) slows processing of true "stat" tests on non-admitted patients by increasing total requests being processed.

Admitted-patient testing

As reported in a previous study, the volume of tests for admitted patients contributed to prolonged order-processing time.¹⁰ Admitted-patient testing volume has grown significantly along with the severity of ED overcrowding; in our ED, admitted patients routinely occupy 25%-100% of ED stretchers. This contributes to loss of ED productivity because the time that ED staff spend with stable admitted patients detracts from the evaluation and management of incoming acutely ill patients. It is common for the laboratory to view any test routed from the ED as a "stat" test (including those from stable admitted patients). This is likely to have a further impact on ED stat test TATs because an increasing number of admitted-patient tests, which are commonly processed as stat, displace the test specimens taken from newly arrived acute ED patients.

Sample processing

The sample-processing time for potassium is expected to be a minimum of 15 minutes because of the need for centrifuge. Use of whole-blood analyzers would preclude centrifuge, although these instruments offer a smaller menu of tests than large automated multi-analyte instruments.

Instrument time, depending on the machine, can be expected to be a minimum of 8.5-10 minutes, depending on the rate-limiting step. Multi-analyte laboratory instruments generally hold results on a sample until all results are ready to be reported. Thus, the analytical time is that of the longest test requested on each sample. Taken together, there is room for improvement in the sample processing and instrument times.

Limitations

A limitation of this study could have been a Hawthorne effect introduced by ED staff when recording the event times, although efforts were made to keep the staff unaware of the objectives. Synchronized electronic time stamps were employed to ensure integrity of the recorded times. Given that the study was conducted over 8 days with rotating ED staff, it is unlikely that a concerted effort to falsify the recorded times was made. External validity is also uncertain, and our findings may not apply to all EDs. This study measured hemoglobin and potassium as reflections of the system, without evaluating the need for these tests. Elimination of unnecessary tests may ultimately be the most efficient improvement in laboratory TATs.

There are a number of opportunities for further research, including seeing whether changes to these root causes improve the system, if implementation of an improved system will allow favourable function of computerized physician-order entry, if ED length of stay for acute patients will be affected, and if reduced TAT will result in quicker decision-making.

Conclusions

In this study, the TAT for hemoglobin (18 min) met the established benchmark, but that for potassium (49 min) did not. Root causes of delay were order-processing time; excessive queue and instrument times for potassium and volume of tests for admitted patients. Further study of these identified causes of delays is required to see whether laboratory TATs can be reduced.

References

1. Kaplan R, Norton D. The strategy-focused organization. Boston: Harvard Business School Publishing Corporation; 2001.
2. Joint Commission tackles emergency department overcrowding. ACEP News 2003;May:1-2.
3. Steindel SJ, Howanitz PJ. Physician satisfaction and emergency department laboratory test turnaround time. Arch Pathol Lab Med 2001;125:863-71.
4. Howanitz PJ, Steindel SJ, Cembrowski GS, Long TA. Emergency department stat test turnaround times. Arch Pathol Lab Med 1992;116:122-8.
5. Howanitz PJ, Cembrowski GS, Steindel SJ, Long TA. Physician goals and laboratory test turnaround times. A College of American Pathologists Q-Probes study of 2763 clinicians and 722 institutions. Arch Pathol Lab Med 1993;117:22-8.
6. Howanitz JH, Howanitz PJ. Laboratory test results. Timeliness as a quality attribute and strategy. Am J Clin Pathol 2001; 116:311-5.
7. Wilson PF, Dell LD, Anderson GF. Root cause analysis: a tool for total quality management. Milwaukee: American Society for Quality Control;1993. p. 59-72, 131-50.
8. Juran JM. Juran on planning for quality by design. New York: The Free Press; 1988.
9. Beveridge R, Clarke B, Janes L, Savage N, Thompson J, Dodd G, et al. Canadian Emergency Department Triage and Acuity Scale: implementation guidelines. Can J Emerg Med 1999;1(3 Suppl). Available: www.caep.ca/002.policies/002-02.ctas.htm
10. Fernandes CMB, Walker R, Price A, Marsden J, Haley L. Root cause analysis of laboratory delays to an emergency department. J Emerg Med 1997;15:735-9.
11. The Leapfrog Group Web site [Internet]. Available: www.leapfroggroup.org (accessed 2004 Feb 6).