The prevalence of preanalytical errors in a Croatian ISO accredited laboratory

Transcription

1 Clin Chem Lab Med 2010;48(7): by Walter de Gruyter Berlin New York. DOI /CCLM The prevalence of preanalytical errors in a Croatian ISO accredited laboratory Ana-Maria Simundic*, Nora Nikolac, Ines Vukasovic and Nada Vrkic University Department of Chemistry, University Hospital Sestre Milosrdnice Zagreb, Croatia Abstract Background: The preanalytical phase is the most common source of laboratory errors. The goal of this descriptive study was to analyze the prevalence and type of preanalytical errors in relation to the site of sample collection (inpatient vs. outpatient) and the type of laboratory unit (hematology and coagulation vs. biochemistry). For the biochemistry unit, the data were also analyzed relative to the type of the analysis (stat vs. routine). Methods: We retrospectively analyzed the sample and test request form error rate for a 1-year period, from January to December Results: The frequency of the sample errors differed significantly between the emergency and routine biochemistry unit (0.69% vs. 2.14%; p ), and between inpatients and outpatients (1.12% vs. 1.36%; ps0.0006). Hemolysis was the most frequent sample error, accounting for 65% of all unsuitable specimens in the emergency biochemistry unit. The total sample error rate did not differ between hematology and coagulation vs. the biochemistry unit. The frequency of test request form errors differed significantly with respect to the sample collection site (p ), laboratory unit (p ) and type of the analysis (p ). Errors in the test request form were least frequent in the outpatient unit (2.98%) and most frequent in the routine biochemistry unit (65.94%). Conclusions: Sample and test request form errors in our laboratory are occurring with a frequency comparable to that reported by others. Continuous educational action is needed for all stakeholders involved in laboratory testing to improve the quality of the preanalytical phase of the total testing process. Clin Chem Lab Med 2010;48: Keywords: hemolysis; preanalytical errors; quality; request form; sample. *Corresponding author: Assist. Prof. Ana-Maria Simundic, PhD, University Department of Chemistry, Department of Molecular Diagnostics, University Hospital Sestre Milosrdnice, Zagreb, Croatia am.simundic@gmail.com Received September 21, 2009; accepted February 19, 2010; previously published online May 4, 2010 Introduction The preanalytical phase is the most common source of laboratory errors (1). Continuous monitoring and management of preanalytical errors is therefore crucial to the quality of laboratory performance, and also required for all clinical laboratories accredited in accordance with ISO quality standard (2, 3). The reported prevalence of preanalytical errors depends on the definition of error, facility, detection system and specimen type, and ranges from -1% up to 10% (4 6). The goal of this descriptive study at the University Department of Chemistry of the University Hospital Sestre Milosrdnice (Zagreb, Croatia) was to analyze the prevalence and type of preanalytical errors in relation to the site of sample collection (inpatient vs. outpatient) and to the type of laboratory unit (hematology and coagulation vs. biochemistry). For the biochemistry unit, the data were considered with respect to the type of the analysis (stat vs. routine). Materials and methods This retrospective study was performed during a 1-year period, from January 2008 until December The University Department of Chemistry of the 900 bed University Hospital Sestre Milosrdnice (Zagreb, Croatia) is a large ISO accredited central clinical laboratory, providing general and specialized clinical chemistry, immunology, toxicology, molecular diagnostics, hematology and coagulation testing services. The laboratory performs up to 3 million tests/year. For inpatients, phlebotomy is performed by the nursing staff in the clinical wards, outside of the laboratory. The laboratory also has an outpatient laboratory unit where phlebotomy is performed by laboratory technicians. Samples and request forms are delivered to the laboratory by the hospital delivery personnel and processed by the laboratory staff. Upon admission, clinical data, test requests and respective non-conformities (errors) are registered into the laboratory information system (LIS). Samples are subsequently centrifuged, aliquoted and distributed to the different laboratory departments. Errors encountered later in the testing process are also entered into the LIS. The number and type of errors were retrieved retrospectively from the LIS. Two types of errors are recorded into the LIS for every sample processed by the laboratory: 1) errors in the test request form and 2) sample errors. We continuously record the following types of errors in the test request form: missing data on patient identification (ID), date of birth, hospital ward, requesting physician, diagnosis, sampling time, phlebotomist, sample type, missing tests, missing test request form. Errors not falling into any of the mentioned categories of the test request form were classified as other test request form errors. The following sample errors are recorded for every sample received in our laboratory: hemolysis, lipemia, inappropriate blood/additive ratio, inappropriate volume, clotted sample, inappro- 2010/512

2 1010 Simundic et al.: Preanalytical errors in Croatian ISO accredited laboratory priate delivery, inappropriate container and missing sample. Errors not falling into any of the mentioned categories of the sample were classified as other sample errors. Serum indices were assessed visually by laboratory technicians. For the purpose of this study, any visual hemolysis and lipemia was considered as sample error, irrespective to the degree of the serum interference and type of test requested. Statistical analysis Data are presented as counts and percentages. The error prevalence for each category was calculated relative to total number of errors (error number/total number of errors) and expressed as a percentage. The difference in total error rate between groups was tested using the z-test. The level of statistical significance was set at Statistical analysis was performed with MedCalc software (MedCalc , Frank Schoonjans, Mariakerke, Belgium). Results Over the 1-year period of this study, there were 332,319 patient samples (307,044 inpatient and 25,275 outpatient) analyzed in our laboratory, accounting for almost 2.5 million analyses. During the studied period, the majority of analyses were for the biochemistry (44.7%), hematology (31.2%) and coagulation unit (21.4%), while other analyses were much less frequent (1.2% immunology, 1% toxicology and 0.2% molecular diagnostics). Table 1 shows the frequency distribution for sample and test request form errors relative to the sample collection site (inpatients vs. outpatients). Sample errors were significantly less prevalent in inpatients than in outpatients (1.12% vs. 1.36%, respectively; ps0.0006). Sample hemolysis and lipemia were the most frequent sample errors in both the outpatient and inpatient units. There was no difference between the frequency of sample hemolysis seen between inpatients and outpatients (46.99% vs %; ps0.0021), whereas lipemic samples were more prevalent in outpatients than in inpatients (48.70% vs %; p ). The frequency of test request form errors differed significantly between inpatient and outpatient units (p ), being higher in inpatients. In inpatients, the most frequent test request form items that were missing were information about the phlebotomist, patient diagnosis, sample type and patient unique identification (ID) number. The majority of errors of the test request form recorded as other were missing signature of the requesting physician. The only error which was significantly more prevalent in outpatients (p ) was missing information about the requesting physician (unique identifying physician number). Table 2 shows the frequency of errors for sample and test request forms for the hematology and coagulation and biochemistry unit. We did not observe a statistically significant difference in the total number of sample errors between those two units in our laboratory (ps0.0210). Hemolysis and lipemia were significantly more prevalent (p ), and inappropriate volume and clotted samples significantly less frequent in the biochemistry unit (p ) when compared to the hematology and coagulation unit. The total number of errors in the test request form was significantly higher in the biochemistry unit than in the hematology and coagulation unit (p ). The most frequent errors in the biochemistry unit were missing signature of the requesting Table 1 Errors in the test request form and sample errors for inpatients and outpatients. Inpatients Outpatients p-value ns307,044 ns25,275 Sample errors Total number of errors, n (%) 3437 (1.12%) 345 (1.36%) Hemolysis, n (rel.%) 1615 (46.99%) 172 (49.86%) Lipemia, n (rel.%) 477 (13.88%) 168 (48.70%) Inappropriate volume, n (rel.%) 213 (6.20%) 3 (0.87%) Clotted sample, n (rel.%) 456 (13.27%) 1 (0.29%) Inappropriate delivery, n (rel.%) 60 (7.75%) N/A Inappropriate container, n (rel.%) 8 (0.23%) 0 (0.00%) Sample missing, n (rel.%) 256 (7.45%) 2 (0.58%) Other, n (rel.%) 251 (7.30%) 8 (2.32%) Errors of the test request forms Total number of errors, n (%) 110,368 (35.95%) 753 (2.98%) Patient ID not known, n (rel.%) 25,983 (23.54%) N/A Date of birth not known, n (rel.%) 551 (0.50%) 2 (0.27%) Hospital ward not known, n (rel.%) 253 (0.23%) N/A Requesting physician not known, n (rel.%) 3558 (3.22%) 558 (74.10%) Diagnosis not known, n (rel.%) 37,127 (33.64%) 180 (23.90%) Sampling time not known, n (rel.%) 19,763 (17.91%) 19 (2.52%) Phlebotomist not known, n (rel.%) 51,356 (46.53%) 26 (3.45%) Sample type not known, n (rel.%) 24,576 (22.27%) N/A Tests missing, n (rel.%) 108 (02.10%) N/A Test request form missing, n (rel.%) 284 (0.26%) 5 (0.66%) Other, n (rel.%) 36,275 (32.87%) 16 (2.12%)

3 Simundic et al.: Preanalytical errors in Croatian ISO accredited laboratory 1011 Table 2 Errors in the test request forms and sample errors for the hematology and coagulation unit and biochemistry unit. Hematology and Biochemistry unit p-value coagulation unit ns181,705 ns125,339 Sample errors Total number of errors, n (%) 1337 (1.07%) 2100 (1.16%) Hemolysis, n (rel.%) 317 (23.71%) 1296 (61.71%) Lipemia, n (rel.%) 61 (4.56%) 416 (19.81%) Inappropriate volume, n (rel.%) 190 (14.21%) 23 (1.10%) Clotted sample, n (rel.%) 421 (31.49%) 35 (1.67%) Inappropriate delivery, n (rel.%) 51 (3.81%) 9 (0.43%) Inappropriate container, n (rel.%) 5 (0.37%) 3 (0.14%) Sample missing, n (rel.%) 104 (7.78%) 152 (7.24%) Other, n (rel.%) 147 (10.99%) 104 (4.95%) Errors of the test request forms Total number of errors, n (rel.%) 33,945 (27.08%) 76,423 (42.06%) Patient ID not known, n (rel.%) 8195 (6.54%) 17,788 (9.79) Date of birth not known, n (rel.%) 267 (0.21%) 284 (0.16%) Hospital ward not known, n (rel.%) 99 (0.08%) 154 (0.08%) Requesting physician not known, n (rel.%) 1556 (1.24%) 2002 (1.10%) Diagnosis not known, n (rel.%) 12,438 (9.92%) 24,689 (13.59%) Sampling time not known, n (rel.%) 6308 (5.03%) 13,455 (7.40%) Phlebotomist not known, n (rel.%) 19,081 (15.22%) 32,275 (17.76%) Sample type not known, n (rel.%) 1420 (1.13%) 23,156 (12.74%) Tests missing, n (rel.%) 72 (0.06%) 36 (0.02%) Test request form missing, n (rel.%) 119 (0.09%) 165 (0.09%) Other, n (rel.%) 1383 (1.10%) 34,892 (19.20%) physician (recorded under other errors) (p ), information on the phlebotomist (p ) and patient diagnosis (p ). The total number of sample errors and errors in the test request form were higher in routine than in the emergency biochemistry unit (p ; Table 3). We observed more Table 3 Errors in the test request forms and sample errors in the biochemistry unit for stat and routine analysis. Emergency Routine p-value biochemistry unit biochemistry unit ns123,464 ns58,241 Sample errors Total number of errors, n (%) 853 (0.69%) 1247 (2.14%) Hemolysis, n (rel.%) 556 (65.18%) 742 (59.50%) Lipemia, n (rel.%) 50 (5.86%) 366 (29.35%) Inappropriate volume, n (rel.%) 13 (1.52%) 10 (0.80%) Clotted sample, n (rel.%) 30 (3.52%) 5 (0.40%) Inappropriate delivery, n (rel.%) 8 (0.94%) 1 (0.08%) Inappropriate container, n (rel.%) 3 (0.35%) 0 (0.00%) Sample missing, n (rel.%) 135 (15.83%) 17 (1.36%) Other, n (rel.%) 68 (7.97%) 36 (2.89%) Errors of the test request forms Total number of errors, n (rel.%) 38,017 (30.79%) 38,406 (65.94%) Patient ID not known, n (rel.%) 8575 (22.56%) 9213 (23.99%) Date of birth not known, n (%) 159 (0.42%) 125 (0.33%) Hospital ward not known, n (rel.%) 76 (0.20%) 78 (0.20%) Requesting physician not known, n (rel.%) 1033 (2.72%) 969 (2.52%) Diagnosis not known, n (rel.%) 10,780 (2.84%) 13,909 (36.22%) Sampling time not known, n (rel.%) 5796 (15.25%) 7659 (19.94%) Phlebotomist not known, n (rel.%) 17,562 (46.20%) 14,713 (38.31%) Sample type not known, n (rel.%) 9783 (25.73%) 13,373 (34.82%) Tests missing, n (rel.%) 27 (0.07%) 9 (0.02%) Test request form missing, n (rel.%) 127 (0.33%) 38 (0.10%) Other, n (rel.%) 13,782 (36.25%) 21,110 (5.50%)

4 1012 Simundic et al.: Preanalytical errors in Croatian ISO accredited laboratory hemolyzed and missing samples in the emergency than in routine biochemistry unit, and the difference was statistically significant (p ). Lipemic samples were more frequent in the routine biochemistry unit (p ). The most frequent errors in the test request form for the routine biochemistry unit were missing signature of the requesting physician (recorded under other errors) (p ), information on the phlebotomist (p ) and patient diagnosis (p ). Discussion Preanalytical errors are the most common types of errors occurring within the total testing process, and greatly influence the quality of laboratory performance. The goal of this study was to assess the prevalence and type of preanalytical errors relative to the different laboratory units, and the site of sample collection. In addition, for the biochemistry unit, the data were also analyzed in relation to the type of the laboratory analysis (stat vs. routine). We observed that sample errors occurred with a frequency of 0.69% 2.14%. The frequency of test request form errors were observed with the frequency of 65.94% in routine biochemistry, and were the least prevalent in outpatients (2.98%). The sample error rate was significantly different with respect to the site of sample collection (inpatient vs. outpatient), and type of analysis (stat or routine). The total sample error rate did not differ between hematology and coagulation vs. the biochemistry unit. Test request form errors differed significantly between different laboratory units (hematology and coagulation vs. biochemistry), respective to the site of sample collection (inpatient vs. outpatient), and the type of analysis (stat or routine). Sample and test request errors were most frequent in the routine biochemistry unit. Hemolyzed specimens are very frequent in clinical laboratories, accounting for the majority (40% 70%) of all unsuitable specimens (7). Hemolysis was also the most frequent sample error in this study. As expected, hemolysis was most frequent in our emergency biochemistry unit where it accounted for 65% of all unsuitable specimens. Even in our outpatient laboratory unit, hemolysis was the most prevalent laboratory error where it accounted for 49% of the total sample errors. Of interest, the lowest frequency of hemolysis (23%) was recorded in our hematology and coagulation unit, where clotted samples were the most frequent type of sample error (31%). Lipemic samples were the second most frequent types of sample errors in this study, accounting for 4.5% 48.6% of all unsuitable specimens. Of interest, lipemic samples were most frequently observed in the outpatient unit (48.6% of all sample errors) where they were the second most frequent sample error. This clearly indicates that the laboratory should be more involved in educating patients about the importance of fasting prior to laboratory testing. Phlebotomy is crucial to the quality of the sample, and is recognized as the most critical preanalytical step (8, 9). Bilic- Zulle et al. recently published a cross-sectional national survey study to investigate the self-reported quality of extra-analytical practices in Croatian laboratories (10). The questions in this survey were designed as statements describing the frequency of the laboratory procedures and were divided into three groups: 1) questions considering the criteria for sample acceptance; 2) questions considering procedures for phlebotomy; and 3) questions considering reporting of results. Of interest, the lowest quality, according to what was self reported by the laboratory personnel, was related to the procedures related to phlebotomy performed within the laboratory by laboratory personnel. These data further support the need for reinforced continuous educational activities for all stakeholders (patients, physicians, nurses and laboratory staffs) to improve the extra-analytical activities of the testing process. Special emphasis should be placed on the practices and policies related to the collection of biological samples. Compared to sample errors, errors in the test request form were far more prevalent. These were observed with the frequency of up to 65.94% for routine biochemistry, and were least prevalent for outpatients (2.98%). Among errors in the request form, the most frequent were missing physician signature (recorded under other errors) and information on the phlebotomist and diagnosis. Sample type and patient ID were also very frequent errors in the test request form. Education of the clinical staff about the importance of correct and proper information on the test request form is the responsibility of the laboratory. Due to the high prevalence of test request form errors in this study, we concluded that there is space for potential improvement. Our next corrective action will be intensive education of the hospital staff with the goal of reducing the total error rate. Without such data on the error rate obtained using this error tracking system, we would not be able to quantify and monitor major preanalytical problems, which is a prerequisite for targeted educational activities as well as for the redesign of processes within the laboratory. As reviewed by Bonini et al., data on the prevalence and type of preanalytical errors in the clinical laboratory are heterogeneous, differing in the period during which data is collected, number of tests/patients involved and reported error frequency (11). The real estimate of errors is hard to assess, due to fear, underestimation and underreporting (12). Such heterogeneity of the reported frequencies of preanalytical errors is to a certain extent due to the fact that error identification is highly subjective. One significant finding reported by Bilic-Zulle et al. was also that 21% of the participants in the survey never or rarely kept records of non-conformities (10). This means that the real number of errors is actually unknown and probably substantially higher. Based on our observations, we hypothesize that laboratory personnel are more sensitive to errors with the potential for having clinically significant consequences. Therefore, our study found clotted samples and samples with inappropriate volume and ratio of sample and additive were more often recorded in the hematology and coagulation unit, whereas hemolysis was predominantly recorded in biochemistry unit. The error detection rate may also vary depending on the detection sys-

5 Simundic et al.: Preanalytical errors in Croatian ISO accredited laboratory 1013 tem, visual inspection being of inferior performance (13), and automated systems being recommended (14). It should be noted that careful monitoring and proper management of unsuitable specimens with various serum indices is very important since results from such specimens may be inaccurate, causing some types of medical error and being a considerable hazard to patient health (15, 16). The question to be answered is: does data on the prevalence of preanalytical errors reflect the real error rate or the rate of errors considered relevant by the operating personnel? Some future studies should provide answers to this question. Therefore, we need an objective standardized methodology for definition of error (17), and systematic error detection and classification techniques to provide data on segments of the total testing process that are more susceptible to errors (18, 19). We conclude that sample errors and test request form errors in our laboratory are occurring with the frequency comparable to that already reported by laboratories in developed countries, with sample errors being less frequent than test request form errors. The frequency of errors varies with respect to the laboratory unit, with hemolysis being the most commonly recorded sample error and physician signature and data on phlebotomist, patient diagnosis, sample type and patient ID being the most commonly observed test request errors. Laboratory personnel should be educated in order to take actions to reduce the error rate in the preanalytical phase of the laboratory testing process within and outside the laboratory. Continuous educational actions should be targeted to patients and clinical staff in the hospital wards. Effective error reporting systems in laboratory medicine is essential if the laboratory is to learn from its own mistakes. Objective and standardized mechanisms should be adopted for systematic error tracking (20), with the ultimate goal of promoting successful risk management and safety awareness. Reporting errors, process analysis and investigation should always lead to corrective and preventive actions that address weak points in the system. This closed cycle is the essence of continuous quality improvement, a key prerequisite to total quality management. Patient safety requires a complex approach, encompassing full control of all laboratory activities, from test request to reporting of results. As pointed out by Lippi, we should start thinking outside our traditional laboratory box, and try to target those extra-analytical processes that are more vulnerable to errors (21). Acknowledgements This work is supported by the Ministry of Science, Education and Sports, Republic of Croatia. Project number: Conflict of interest statement Authors conflict of interest disclosure: The authors stated that there are no conflicts of interest regarding the publication of this article. Research funding: None declared. Employment or leadership: None declared. Honorarium: None declared. References 1. Plebani M. Laboratory errors: how to improve pre- and postanalytical phases? Biochemia Medica 2007;17: Kirchner MJ, Funes VA, Adzet CB, Clar MV, Escuer MI, Girona JM, et al. Quality indicators and specifications for key processes in clinical laboratories: a preliminary experience. Clin Chem Lab Med 2007;45: Simundic AM, Topic E. Quality indicators. Biochemia Medica 2008;18: Salvagno GL, Lippi G, Bassi A, Poli G, Guidi GC. Prevalence and type of pre-analytical problems for inpatients samples in coagulation laboratory. J Eval Clin Pract 2008;14: Lippi G, Salvagno GL, Favaloro EJ, Guidi GC. Survey on the prevalence of hemolytic specimens in an academic hospital according to collection facility: opportunities for quality improvement. Clin Chem Lab Med 2009;47: Carraro P, Plebani M. Errors in a stat laboratory: types and frequencies 10 years later. Clin Chem 2007;53: Lippi G, Blanckaert N, Bonini P, Green S, Kitchen S, Palicka V, et al. Haemolysis: an overview of the leading cause of unsuitable specimens in clinical laboratories. Clin Chem Lab Med 2008;46: Lippi G, Salvagno GL, Montagnana M, Franchini M, Guidi GC. Phlebotomy issues and quality improvement in results of laboratory testing. Clin Lab 2006;52: Lippi G, Salvagno GL, Montagnana M, Guidi GC. The skilled phlebotomist. Arch Pathol Lab Med 2006;130: Bilic-Zulle L, Simundic AM, Supak Smolcic V, Nikolac N, Honovic L. Self reported routines and procedures for the extraanalytical phase of laboratory work in Croatia cross-sectional survey study. Biochemia Medica 2010;20: Bonini P, Plebani M, Ceriotti F, Rubboli F. Errors in laboratory medicine. Clin Chem 2002;48: Plebani M, Lippi G. To err is human. To misdiagnose might be deadly. Clin Biochem 2009;43: Simundic AM, Nikolac N, Ivankovic V, Ferenec-Ruzic D, Magdic B, Kvaternik M, et al. Comparison of visual versus automated detection of lipemic, icteric and hemolysed specimens: can we rely on a human eye? Clin Chem Lab Med 2009;47: Lippi G, Luca Salvagno G, Blanckaert N, Giavarina D, Green S, Kitchen S, et al. Multicenter evaluation of the hemolysis index in automated clinical chemistry systems. Clin Chem Lab Med 2009;47: Ryder KW, Glick MR. Erroneous laboratory results from hemolyzed, icteric, and lipemic specimens. Clin Chem 1993;39: Lippi G. Governance of preanalytical variability: travelling the right path to the bright side of the moon? Clin Chim Acta 2009; 404: Sciacovelli L, Plebani M. The IFCC Working Group on laboratory errors and patient safety. Clin Chim Acta 2009;404: Lippi G, Bassi A, Brocco G, Montagnana M, Salvagno GL, Guidi GC. Preanalytic error tracking in a laboratory medicine

6 1014 Simundic et al.: Preanalytical errors in Croatian ISO accredited laboratory department: results of a 1-year experience. Clin Chem 2006;52: Lippi G, Banfi G, Buttarello M, Ceriotti F, Daves M, Dolci A, et al. Recommendations for detection and management of unsuitable samples in clinical laboratories. Clin Chem Lab Med 2007;45: Lippi G, Bonelli P, Rossi R, Bardi M, Aloe R, Caleffi A, Bonilauri E. Development of a preanalytical errors recording software. Biochemia Medica 2010;20: Lippi G, Simundic AM. Total quality in laboratory diagnostics. It s time to think outside the box. Biochemia Medica 2010; 20:5 8.