Tools, Technologies and Training for Healthcare Laboratories

Errors before the Laboratory

Several recent studies assert they have calculated the rate of error in laboratories, pre-analytical, analytical, and post-analytical. But what those numbers mean depends on what's being counted and who's doing the counting.

July 2007

Errors in the laboratory total testing process have been the subject of a couple of recent reports - one a published study in the journal of Clinical Chemistry [1] and another an abstract for a poster that was presented at the recent AACC/ASCLS meeting in San Diego [2]. Both point to the major source of errors being in the pre-analytic phase and particularly in specimen collection, labeling, and identification. Coincidentally, one of the Symposia at the San Diego meeting “Quality Management Approach to Improving Patient Safety” focused on “risk analysis” and included a presentation that demonstrated a process for reducing these errors in patient identification and specimen collection [3].

What are the estimates?

Plebani and his associates have provided the most definitive studies about the distribution of errors in the laboratory total testing process, beginning with an earlier assessment of errors in 1996 [4], adding a critical review of the data available in the literature [5], and now re-assessing the distribution of errors some ten years after their initial study. The most recent study found that 60% of errors are pre-analytic, 15% analytic, and 25% post-analytic. These figures can be compared with their earlier estimates of 68% pre-analytic, 13% analytic, and 19% post-analytic. The overall frequency of errors in the recent study was estimated as 3092 dpm (defects per million), compared to 4700 dpm in the 1996 study. Their earlier literature review [5] compiled estimates of errors that ranged from 31-75% for pre-analytic, 13-32% for analytic, and 9-31% for post-analytic errors.

The abstract/poster by Crim and Okorodudu documented the pre-analytic errors as 86%, analytic as 0%, and post-analytic errors at 15%.

In the presentation by Weber, the focus was strictly on specimen identification, collection, and labeling problems for which the initial error rate was observed to be 0.19%, and wase then reduced to 0.024% after a conducting a patient safety process improvement project.

Given the available estimates of frequency of errors, estimates of sigma-metrics can also be made. The Plebani 2006 error rate of 3092 dpm, or 0.31%, corresponds to 4.3 sigma and the 1996 error rate of 4700 dpm or 0.47% corresponds to 4.1 sigma. Weber’s 0.19% error rate corresponds to 4.4 sigma and the improvement to 0.024% corresponds to 5.0 sigma. If one considers that 60% of the overall errors in the most recent Plebani study are attributed to preanalytic errors, the error rate for pre-analytic errors would be 0.186%, which is remarkably close to the initial error rate of 0.19% observed by Weber.

How were the errors identified?

The Plebani study design depends on physicians and nurses to identify and notify the laboratory of questionable findings.

“The present study was conducted in 2006 according to the design we previously used in 1996 to monitor the error rates for laboratory testing in 4 departments (internal medicine, nephrology, surgery, and intensive care). For 3 months, physicians and nurses were asked to pay careful attention to all test results. Any suspected laboratory error was recorded with associated pertinent clinical information. Every day, a laboratory physician visited the 4 departments and a critical appraisal was made of any suspected results. Among a total of 51,746 analyses, clinicians notified us of 393 questionable findings, 160 of which were confirmed as laboratory errors…”

In the abstract from Crim and Okorodudu, the methodology is quite different:

“We reviewed the quality assessment and quality control records from our laboratory for a period of 6 months (1/06-6/06) for the total number of errors (n=196) and classified them according to the type of clinical error. The frequency of errors in the manual order entry was compared to that of the automated physician order entry.”

In the study reported by Weber, the errors are those identified by the laboratory in real-time monitoring during specimen processing.

How would a miss-matched test order and specimen tube be counted? If identified by the laboratory, it would likely be counted by Crim/Okorodudu and Weber, but maybe not by Plebani. If identified by the physicians or nurses, it would be counted by Plebani but maybe not by Crim/Okorodudu and Weber, depending on whether or not there was an incident report or a complaint to the laboratory.

How would an analytical error be counted? If a laboratory testing process were found to be out-of-control, would that count as an error? Not in the studies by Plebani because errors are only those identified by physicians and nurses on the ward. Not likely in the study by Crim/Okorodudu, unless you believe that a laboratory can operate for six months without any testing process having an out-of-control problem. If no out-of-control conditions were detected, then one must question whether the QC design was appropriate for detecting medically important errors!

How would a post-analytic error be counted? Plebani’s laboratory is highly computerized, thus all of the post-analytic errors noted in the 2006 were due to a single problem occurring with the laboratory information system. Does poor turn-around-time count as an error? Probably not, unless complaints have been made by the physicians or nurses.

While there are a considerable number of issues related to study design, let’s accept that pre-analytic errors are a big problem, particularly patient identification and specimen collection. I also believe analytical errors are still a problem and find it interesting that Plebani notes that of the 46 errors that led to inappropriate patient care, 24 of those (over half) were analytical errors. So, even though analytic errors are only 15% of the errors in the total testing process, they may still be the most important cause of inappropriate patient care.

Who mis-identifies patients and mis-collects their specimens?

The Plebani study represents the experience in a “stat” laboratory at a University Hospital in Italy. They clearly state that “blood drawing and sample collection are performed by physicians and nurses from the individual wards” [1].

The Crim/Okorodudu abstract comes also from a University laboratory (University of Texas Medical Branch, Galveston). The brevity of an abstract limits the description of the methodology, but they do indicate that “unfortunately, many extra-analytical factors are physician dependent…” [2], which suggests the specimen acquisition process is not under the control of the laboratory.

The study reported by Weber comes from a hospital laboratory at the University of Kansas and specifically focuses on the reduction of specimen collection problems by physicians, nurses, and ward staff. It seems that many (maybe most) of our pre-analytic problems are caused by non-laboratory personnel who are not properly trained or, more likely, do not find specimen collection to be their primary interest and/or duty.

How can this problem be solved?

Clearly, the work of Weber shows that the laboratory can contribute to resolution of this problem, particularly by providing the data for estimating and monitoring the errors in the specimen collection process. The laboratory can also provide the leadership for resolving this problem, and given the current interest in patient safety and the current JACHO national patient safety goals, the laboratory can more easily work across department barriers, as demonstrated by Weber. In fact, Weber was the leader for the patient safety improvement project, which followed a group problem solving process typical for TQM/CQI, but utilizing specific tools for risk analysis and error prevention. Hopefully, a complete description of this study will be published to make the methodology more widely available.

Given Weber’s blue-print for solving the problem, it will be of interest to see if this leads to resolution of the specimen identification/collection problem more widely throughout laboratories in this country. But, here’s the rub! It takes hard work by a laboratory to get this done and furthermore requires ongoing work to maintain any improvements. Do laboratories have the resources to get these problems solved when the faults exist outside of their area of control? Can improvements, once made, be maintained long term in the rapidly changing world of healthcare and the ever changing conditions in a hospital laboratory? Is it just coincidental that all three of these reports come from laboratories at University Hospitals? Do non-University laboratories have the resources to adopt the monitoring and improvement practices necessary to solve this problem?

Will the patient safety movement lead to sustainable improvements?

I recall working on this same problem almost 20 years ago [6]. At that time, there were no national patient safety goals, so it was a much harder “sell” to gain cooperation from outside the laboratory, particularly the nurses. When we got to the “monitoring” stage, it was very difficult to maintain their interest – they just had too many other more important things to do! To maintain improvements, we proposed a program that would have laboratory technologists regularly spend time on the nursing units to provide a friendly on-site interface to deal with any problems with the testing process. Unfortunately, we weren’t able to gain support to implement that program, in part because of concern in the laboratory for the cost of doing this and in part because nursing administrators didn’t view it as important for their operations.

I hope the patient safety movement has reached the top administrators and directors in hospitals and laboratories and, consequently, top management will now provide the leadership and commitment needed for long-term resolution of this problem.

References

  1. Carraro P, Plebani M. Errors in a stat laboratory: Changes in type and frequency since 1996. Clin Chem 2007;53:1338-1342.
  2. Crim J, Okorodudu A. Evaluation of changing patterns in clinical laboratory errors. Abstract. Clin Chem 2007;53(6S):A207.
  3. Weber S. Reducing preanalytic errors using Failure Mode and Effects Analysis (FMEA): A case study. Presented at the symposium Quality Management Approach to Improving Patient Safety, AACC National Meeting, San Diego, July 25, 2007.
  4. Carraro P, Plebani M. Mistakes in a stat laboratory: types and frequency. Clin Chem 1997;43:1348-1351.
  5. Bonini P, Plebani M, Ceriotti F, Rubboli R. Errors in laboratory medicine. Clin Chem 2002;48:691-698.
  6. Westgard JO. Total quality control: Evolution of quality management systems. Lab Med 1989;20:377-384.

James O. Westgard, PhD, is a professor emritus of pathology and laboratory medicine at the University of Wisconsin Medical School, Madison. He also is president of Westgard QC, Inc., (Madison, Wis.) which provides tools, technology, and training for laboratory quality management.