Reasons for a Laboratory's Inability to Report Results for Requested Analytical Tests

Department of Clinical Chemistry, Vejle County Central Hospital, DK-7100 Vejle, Denmark
^a author for correspondence: fax 45 75 82 18 14, e-mail kka{at}vs.vejleamt.dk

There is increasing understanding concerning the importance of the preanalytical circumstances on laboratory quality, including failures to report requested results. In the past, why the laboratory was unable to deliver a result was not investigated. Accordingly, when looking into the computer system for discrepancies between number of tests requested and number of tests reported, only the volume of this serious quality problem but not the reasons could be seen. Knowledge of the most frequently occurring reasons for report failure is essential in the process of continuous quality improvement.

Previous papers have focused on the total number of report failures, and numbers from 0.1% to 2.0% have been published (1)(2)(3). Ideally the number of unanswered test requests should be zero. In real life it is not possible to fulfill this quality demand, and a practical standard or quality goal concerning this number has not been agreed on, although some authors have suggested 0.2–0.5% (4)(5) as being possible and realistic.

We have studied the reasons for failure to report results in our 600-bed hospital, which serves an area with 120 000 persons and 90 general practitioners. Both general practitioners and hospital wards use request forms with a combination of bar codes and optical mark reading, ensuring that the data system receives key information, including the patient's unique national identification, the requesting doctor's identification, tests requested, and a request form number. The tubes are labeled at the time of sample collection with the bar code label attached to the request form. When a sample is not received with a request form, the reason is investigated and recorded in the computer. Similarly, at all phases of the analytical procedure, beginning at the patient contact, the reason is recorded for any inability to obtain blood, to mail a specimen, to process the specimen, or to calculate the results. These predefined reasons for failures are divided into five subgroups as follows:

Subgroup Examples A. Patient-related Patient not available for sam pling, patient not fasting, or patient discharged from the hospital B. Specimen-related Specimen too old, hemolyzed, lipemic, coagulated, or material incorrect. Air in the syringe, serum or CSF mixed with blood, volume too small, or tube not labeled C. Mailing-related Sample not received, sample mailed despite unsuitable for transportation, tube broken, or incorrect transport conditions D. Laboratory-related Sample not drawn (request overlooked), sample lost, or analysis failed E. Calculation/ evaluation Not evaluated according to laboratory's internal rules (e.g., U-albumin is not performed if U-protein >0.5 g/L, differential count not done if too few leukocytes, Urine M-component not analyzed if U-protein <0.05 g/L, or specific microbiological test not performed if screening test negative)

In every single case where the laboratory was unable to produce the test result, the reason was reported to the requesting doctor as a principal feedback quality assessment procedure.

During the last 5 years, the overall frequency of failures and the frequencies for the hospital wards changed only slightly (Table 1 ). The frequency of failures for the general practitioners fell from 0.93% (1 in 108) requests in 1992 to 0.74% (1 in 136 requests) in 1996 (P <0.001, test).

The main reasons for inability to answer test requests in 1992, 1993, and 1996 (Table 1 ) were specimen and transport related (Reasons B and C). These accounted for 73% in 1992 and 81% in 1993 and are reduced, but still high, accounting for 58% of all failures in 1996.

In 1992, the most frequent problem for the general practitioners was related to specimens (Table 1 ); 38% of unreported requests were related to the sample being too old. For the hospital wards, the main single problem was the subgroup reason D (Laboratory-related) "sample lost" and "sample coagulated" (~15% each).

In 1996 the pattern had changed. The most frequent single problem, 35% of all failures, for general practitioners was the mailing-related "sample not received" (which often means "sample not sent"); for hospital wards it was "patient not available for sampling because of other investigation/medical procedures", which accounted for 19% of all failures.

The problem of "samples not received", although of smaller size (~20%), can be seen for hospital wards as well and needs explanation. One likely explanation is that many patients are treated as outpatients at the hospital; however, they very often visit their own doctors just for blood drawing. The analyses are requested on the hospital's request forms; therefore, failures in such cases are registered as belonging to the hospital. A second explanation is that the hospital's intensive care unit, for practical reasons, often requests blood gas analyses in advance, before actual arterial puncture has taken place. Again, these situations are registered as "not received". Consequently, frequency of "sample not received" for hospital wards is probably <20% and for general practitioners is even >35%.

It is well known that errors in the preanalytical and postanalytical phases of laboratory medicine are the main source of erroneous results (1)(2)(6)(7). Examples of errors include use of wrong sample or container, missing identification of the sample or of the requesting clinician, specimens that are too old, hemolysis, inadequate volume, lost sample, negligence of fasting rules, and so on. Therefore, the laboratory should incorporate systems to detect these situations and to avoid performing analyses on unacceptable specimens. In some clinical situations, a less-accurate result may be better than no result; the error of the result may have no medical importance, but the absence of a result would (1)(2)(6). Because these considerations can be difficult to make in the actual situation in the laboratory, our opinion is that all laboratory results should be reliable and principally of the same quality.

Our purpose is to improve quality of laboratory testing and not to judge or control our users. The described system is a good tool for identifying the type, location (ownership), and severity of the single problem and thus to focus on the most problematic areas; the source of a problem at one type of location (e.g., the ward or general practitioner) does not necessarily exist at another.

The laboratory has focused on solving the major problems and has reduced the specimen- and transport-related problems (Table 1 ). However, there are still problems. One of them is microcoagulation in tubes. It was identified as restricted to arterial punctures for blood gas status investigations, and procedures have been introduced to reduce the problem. Another problem is that general practitioners do not always follow the instructions and rules on mailing and stability of the components. Low quality of slides for differential counts made in general practice is another example. Furthermore, the four laboratories in the county exchange specimens for rare tests; during this procedure, specimens or results can be lost.

The intensity of diagnostic work and treatment of the patients in the hospital per time unit has grossly increased. The duration of stay has decreased from 5.5 days to 4.5 days; hence a new problem is that patients frequently are not available at the time for specimen sampling because of other medical procedures. In 1996, this problem accounted for 13% of all request failures.

The described procedure seems to be a useful tool for quantifying not only the severity of the problem with unreported results but also for diagnosing the reasons, and it thus enables corrective procedures. Yet the method requires a large database with the facility for continuous registration over long periods of time on the single user level so that the periodical fluctuations caused by occasional problems can be minimized.

The reasons for failure can be listed according to the frequency, and hence action can be taken to reduce the most typical causes. Furthermore, a specific pattern can be seen for each individual user; the laboratory, hospital wards, and general practitioners can focus attention on the most important causative problems for each area. We regularly send a data sheet to our users listing, for every single doctor or ward, the number and specific reasons for inability to report results, so that their own pattern of problems can be seen and compared with their colleagues.

Our goal is to detect the failures and prevent them to reduce report failures to ~0.5% and to maintain this level. The suggested frequency of 0.5% report failures should not be treated as a goal for laboratory performance. What is important is to recognize the problem and to consequently introduce procedures minimizing it. Cooperation with the clinicians and the personnel outside the laboratory is of crucial importance for achieving this. Whether 0.5% is an acceptable rate of report failure is a matter of discussion. The question is to balance out the percentage of failures with the risk of effects of biased results.

To enable interlaboratory comparison at the national and international level, a standardized classification and definition of causes for report failures could be useful, and we suggest use of the proposed nomenclature.

Acknowledgments

We thank chief technologist Carsten Thomsen for his kind help with data collection.

References

1. Plebani M, Carraro P. Mistakes in a stat laboratory: types and frequency. Clin Chem 1997;43:1348-1351. [Abstract/Free Full Text]
2. Stuart J, Banfi G, Deom A, Narayanan S, Rabinovitch A. Major themes of the symposium. Proceedings of the second symposium: the impact of the pre-analytical phase of laboratory result 1996:5-9 Becton Dickinson Oxford, UK. .
3. Hinckley CM. Defining the best quality-control systems by design and inspection. Clin Chem 1997;43:873-879. [Abstract/Free Full Text]
4. Stahl M, Lund ED, Brandslund I. Frequency and reasons for failure to report laboratory results [Abstract]. Scand J Clin Lab Investig 1994;54(Suppl 218):87.
5. Pascual C, Garcia-Beltran L, Ricos C, Pelegri MD. The usefulness of a quality assurance program in the preanalytical phase. Proceedings of the second symposium: the impact of the pre-analytical phase of laboratory result 1996:18 Becton Dickinson Oxford, UK. .
6. Witte DL, Van Ness SA, Angstadt DS, Pennell BJ. Errors, mistakes, blunders, outliers, or unacceptable results: how many?. Clin Chem 1997;43:1352-1356. [Abstract/Free Full Text]
7. Guder WG, Narayanan S, Wisser H, Zawta B. Samples: from the patient to the laboratory. The impact of preanalytical variables on the quality of laboratory results. Darmstadt, Germany: Git Verlag, 1996:101 pp..