In recent years, there has been increasing interest in quality improvement and patient safety activities in healthcare. The clinical laboratory has a leader in the field of healthcare quality management with a focus on analytical quality born of its scientific background and was one of the first areas to use quantitative statistical control methods. However laboratories are now being asked to widen their focus to consider activities outside their immediate control. Accreditation agencies are increasingly requiring laboratories to go beyond analytical quality and take responsibility for the pre- and post-analytical (or extra-analytical) phases where most errors arise. These new challenges are a change from the traditional laboratory-based activities with which many laboratory staff is comfortable and this new role can cause some unease and discomfort. This article outlines the different phases of the total testing process, discusses laboratory accreditation requirements for the extra-analytical phase and describes some of the resources available for laboratories in managing this unfamiliar area.

The total testing process (or total testing cycle) is based on the original brain-to-brain loop concept described by Lundberg [1, 2]. He outlined a series of activities, starting with the clinical question in the clinician's mind, leading to test selection, sample collection, transport to the laboratory, analysis, reporting back to the clinician, and final interpretation and decision making by the clinician. These activities have traditionally been separated into three phases (pre-analytical, analytical and post-analytical). Some authors have introduced the "pre-pre-" and "post-post-" analytical phases to identify activities associated with the initial selection of tests and with the interpretation by clinicians respectively, to differentiate them for the pure collection/transport activities (pre-analytical phase) and reporting (post-analytical phase) [3, 4]. There is some evidence that these steps are more error-prone than other pre- and post-analytical activities [3-8]. However, the definition and use of such terms is not universal. Indeed the definition of even basic terms such as pre-analytical, analytical and post-analytical can vary between authorities.

Healthcare is a relatively high risk area and the overall defect rate in healthcare in the United States is estimated to be 31-69% [9]. Error rates are often described using the sigma concept, which refers to the number of standard deviations that lie between the process mean and the specification limit. As the process standard deviation becomes smaller, more standard deviations will fit between the mean and the specification limit, increasing the sigma number and decreasing the likelihood of items exceeding the specification limit. Using this measure, healthcare performs at a 1-2 sigma level, which compares poorly with non-healthcare industries such as airline baggage handling (approximately 4 sigma) [9]. Performance varies in different areas of healthcare, with values of 1 sigma (e.g., use of beta-blockers post myocardial infarction, detection and management of depression) to 3 sigma (e.g., adverse drug events, hospital-acquired infections). Higher error rates can be expected in institutions under pressure to increase revenue, lower costs and operate close to or over full capacity [10].

The analytical phase of laboratory medicine is arguably the best performing sector in healthcare with close to 5 sigma performance (0.002%) [9, 11]. This is more than 3,000 times lower than the rates of infection and medication errors and reflects the standardised quantitative nature of much of laboratory medicine testing, which is well suited to statistical quality control measures [12]. However, the accomplishments of laboratory medicine drop when errors in all phases of the total testing process are considered [13, 14]. The proportion of errors associated with the two extra-analytical phases is 4-5 times that seen in the analytical phase, with the pre-analytical phase consistently representing over half of all errors in published studies [12, 15-19]. In a representative study, an Italian stat laboratory used the same methodology to assess error rates in 1996 and 2006 and found that, despite a 34% reduction in error rate, the pattern of 62% pre-analytical, 15% analytical and 23% post-analytical phase errors remained basically unchanged [20]. Given the high volumes of laboratory tests performed globally, even a low prevalence of errors translates into significant absolute numbers of occurrences and opportunities for adverse patient outcome. Although some laboratories have developed mechanisms to detect errors and improve pre- and post-analytical quality, there remains significant room for improvement in the quality of the extra-analytical testing phase [21-23].

The commonest causes of errors in the total testing process as compiled by Plebani are shown below [22].

These lists illustrate the use of the pre-pre- and post-post-analytical categories - note, for example, that Plebani includes choice of container, collection, handling and transportation as pre-pre-analytical activities, resulting in most errors being categorised as pre-pre-analytical rather than pre-analytical. The lack of standardisation in such taxonomy accounts for some of the variation seen in reported error rates and can complicate discussions [24]. However, the concepts may have value in shaping the laboratory's approach to error management by acting as an explicit reminder of the error-prone nature of test selection and interpretation activities [25].

Errors in healthcare are of concern when they lead to actual or potential adverse outcomes for patients. Given the complex nature of healthcare and the difficulty in assessing the effect of a specific laboratory error on patient management, the prevalence of proven patient harm is difficult to assess. Obvious extreme errors in qualitative results with clear links to therapy or management decisions (e.g., histopathology, blood transfusion, microbiology, virology, genetic testing) are easiest to measure but assessing the effect of quantitative errors in clinical biochemistry and haematology results is much more difficult. Such difficulties mean that present measurements probably significantly underestimate the size of the problem in light of the high volume of quantitative testing performed in clinical laboratories. A review of the available literature on laboratory errors found great heterogeneity in the studies where the data collection method appeared to be the strongest influence on error prevalence and type [19]. Published data suggest that 24-30% of laboratory errors have an effect on patient care while actual or potential patient harm occurs in 3-12% [20, 22, 26]. Some areas, such as molecular genetics testing, can have actual harm rates of up to 100% [19, 27]. A recent study illustrating the dichotomy between the large potential for harm but the much smaller rate of actual harm describes a five-point scoring system for actual and potential adverse impact score elements [28, 29]. Errors were classified as pre-analytical (88.9%), analytical (9.6%) and post-analytical (1.5%). Classification and grading of quality failures in the clinical biochemistry laboratory showed that 72.7% of errors had an actual adverse impact score of 1 (least severe grade) while 65.9% of errors had a potential adverse impact score of 5 (most severe grade) [28].

Although the importance of the pre- and post-analytical phase has been acknowledged for many years, laboratories have often overlooked this area in their quality management programmes, focussing instead on analytical quality and associated activities within their direct control. The main reason for this neglect has been governance issues due to the variety of the different physical locations and staff groups (laboratory staff, clinicians, phlebotomists, porters) involved in the total testing process. Ignorance by non-laboratory staff of the importance of the extra-analytical phase, difficulties in capturing appropriate monitoring data, taxonomical issues in defining and classifying errors and narrow interpretations of the laboratory's role have all contributed to this inaction. The variety of different terms used to define errors, including mistakes, blunders, defects, outliers, unacceptable results, quality failures, have not helped discussion [22]. The term "laboratory error" is defined in International Organization for Standardization (ISO) 22367 as "failure of planned action to be completed as intended, or use a wrong plan to achieve an aim, occurring at any part of the laboratory cycle, from ordering examinations to reporting results and appropriately interpreting and reacting to them" and is the preferred term [22, 30]. A more recent and perhaps more useful description of laboratory error is "any defect from ordering tests to reporting results and appropriately interpreting and reacting on these" [31]. Recent changes to accreditation requirements are forcing laboratories to pay attention to this area.

The present interest in patient safety initiatives can be traced to studies in the 1990s showing that up to 4% of patients in the United States suffered iatrogenic injuries, of which two-thirds were mistakes [32, 33]. Even higher rates were noted in Australia (13%) and the UK (10%) [34, 35]. A series of publications in the US and UK between 1999 and 2004 subsequently led to greater requirements for active management of the extra-analytical phase of the total testing process [36-39]. The Institute of Medicine reports "To Err is Human: Building a Safer Health System" (1999) and "Crossing the Quality Chasm: a New Health System for the 21st Century" (2001) described the high rates of medical error in hospitals in the United States and outlined strategies to reduce their incidence. While the first report highlighted the many American patients who die each year from medical errors, the second described six aims for patient care, specifically safeness, effectiveness, efficiency, equitability, patient-centeredness, timeliness, and rules for care delivery redesign. Medical errors were defined as the failure of a planned action to be completed as intended or the use of a wrong plan to achieve an aim. The majority of medical errors was not the result of individual recklessness or the actions of a particular group but was caused by faulty systems, processes, and conditions that led people to make mistakes or fail to prevent them. Amongst the strategies proposed were the raising of performance standards and expectations for improvements in safety through the actions of oversight organizations and professional groups and the implementing of safety systems in healthcare organizations to ensure safe practices at the delivery level.

These recommendations have been translated in new specific requirements to enhance patient safety by US-based accreditation bodies with similar provisions in other international standards. They can also be found in voluntary guidelines such as those of the National Quality Forum whose 2009 publication "Preferred Practices for Measuring and Reporting Patient Safety and Communication in Laboratory Medicine" focuses on the same areas of patient/sample identification, sample acceptability, test order accuracy, verbal communication and critical result reporting targeted by the accreditation bodies described below [40].

There are a variety of free on-line resources available to the laboratory seeking background information and suggested procedures for pre-analytical phase procedures. Manufacturers of laboratory sample containers are also a valuable source of reference and educational material which can be tailored to the specific collection and analytical equipment used by a particular laboratory. The following is an incomplete list of sites and publications (both commercial and non-commercial) that provides free information in this area. At present there is unfortunately little on-line information on the post-analytical phase.

Until recently, there has been little available information on quality indicators in laboratory medicine. A review of laboratory quality indicators could find an evidence base for only 14 indicators of which 10 were extra-analytical: test order appropriateness, wristband identification errors, patient satisfaction with phlebotomy, specimen quality, availability of inpatient results, corrected laboratory reports, critical value reporting, turnaround time, clinician satisfaction and follow-up of abnormal cervical cytology [56]. Many of the indicators in common use by laboratories suffer from inconsistency in definition, measurement methodologies and reporting practices and a general lack of basic supporting evidence. Even the denominator (per patient, per sample, per test) used in reporting rates of errors can vary between authors [19].

In recent years, several national and regional external quality assurance programmes to examine extra-analytical quality have been developed [75-78]. The approach with the greatest potential global utility is that of the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) Working Group on "Laboratory errors and patient safety" (WG-LEPS), which has published preliminary benchmarking data for all phases of the analytical phase [79-81]. The working group's mission is to stimulate studies on the topic of errors in laboratory medicine, to collect available data and to recommend strategies and procedures to improve patient safety. One of their projects has been to create a systematic common reporting system for clinical laboratories based on standardised data collection, and to define state-of-the-art and quality specifications for each quality indicator independent of the size of organization and type of activities, complexity of processes undertaken, and different degree of knowledge and ability of the staff. A website is available for data submission (www2.csinet.it/mqiweb/) and the working group recently published the results of data collected from 39 laboratories (25 from Europe, 3 from USA, 3 from Asia-Pacific, 8 from elsewhere) from February 2008 to December 2009.

The quality indicators for the pre- and post-analytical phase together with the proposed standards are shown in Table 2. Although zero defects are the ultimate goal, the quality standards suggested represent "state of the art" performance. In some cases, a single quality criterion is given due to the low value observed in the study while in other, no value is given due to the small number of laboratories responding obtained and the wide range of values reported. In terms of the JCI and CAP patient safety goals regarding sample identification and handling of critical results, the table classifies performances of <0.4% misidentified samples, <50 min average time for critical result communication, and >96% critical result communication as meeting optimum levels. It should be appreciated that these are preliminary goals reflecting the heterogeneous group of laboratories around the world contributing to the data collection exercise. For laboratories that already exceed these levels, an expectation of even higher performance may be more appropriate. For example, a CAP Q-Tracks study of 180 institutions showed the 25% best performing laboratories had reported critical result rates of >99% in 2001, a level probably driven by US regulatory requirements [82]. There are differences between US and European practices, particularly with respect to the notifier and the choice of critical values [83, 84]. The laboratory's focus on internal versus external activities also appears to vary between the US and the UK, with greater attention to implementation of clinical guidelines, test utilisation and result interpretation by physicians in the UK [85]. Nevertheless these criteria represent excellent starting points for laboratories in setting benchmarks for extra-analytical quality monitoring and are a significant step forward in developing consensus standards [81]. The group plans to eliminate and modify some existing indicators and develop new ones over the next year [81]. The project's success depends on the participation and collaboration of laboratories enrolled in the project to help define best practice and improve performance and laboratories interested in participating are encouraged to contact the group through the website mentioned above.

Another source of information on present practices in quality indicator monitoring is a recent article describing the results of a survey of members of the Association of Clinical Biochemists in the United Kingdom, which elicited responses from 335 individuals [85]. The author lists 11 questions as pre-analytical, 5 as analytical and 7 as post-analytical, although their classification would not necessarily match the CAP and ISO 15189:2007 phase definitions discussed earlier. Most laboratories had an electronic handbook (84%), provided help and advice in interpreting clinical laboratory data (80%) and discussed turnaround times with clinical staff (75%) but only 58% had a written critical limits (alert) list. This is a useful snapshot of quality practices in UK laboratories and highlights some of the areas where improvement is required.

Review of the accreditation criteria, patient safety concerns and discussions in the literature suggest some clear areas for action for laboratories looking to expand their quality focus outside the physical confines of the laboratory. The two areas of highest priority are patient/sample identification (pre-analytical quality) and the handling of critical results (post-analytical quality). For many laboratories, attention to these issues is required by regulation or patient safety goals and for the remainder, they are becoming part of the good laboratory practices expected of all laboratories worldwide. A variety of approaches, both procedural and information technology-based, are now available for laboratories seeking guidance [24, 83, 87].

The next step could be to expand these items to include more of the testing process. In the pre-analytical area, laboratories can develop clear sample acceptance and rejection criteria which are linked to monitoring of the collection and transport processes. Sample haemolysis and clotting are the commonest causes of unsuitable blood specimens and most laboratories have procedures to handle such specimens at sample receipt, but a more pro-active approach extending back to the point of collection is required [26, 50, 88]. The laboratory's procedures regarding unlabelled or mislabelled specimens should be clear and sample relabeling by laboratory personnel, clinical staff or third parties is strongly discouraged [67]. Collection procedure, container, transport temperature/time/safety and within-laboratory pre-analytical temperature/time/safety criteria should be stipulated and monitored. In the post-analytical area, attention to critical result reporting can be expanded to include all reports, ensuring that the right report goes to the right clinician within the right timeframe. This could include monitoring of the wider definition of turnaround time from test request to clinician review commonly used by clinicians, rather than the traditional sample receipt to result reporting approach favoured by many laboratories [19, 89-91].

Laboratories looking to incorporate the pre-pre- and post-post-analytical phases into their management plans may wish to monitor the appropriateness of test requesting and utility of interpretative reporting. Duplicate laboratory requests repeated within defined intervals can represent wasted and unnecessary testing [92-95]. Repetition of tests can result from poor access to previous results or lack of standardisation between different laboratories [96]. Given the suggestion that up to 50% of requests may be inappropriate, introducing strategies to manage duplicate testing can be a useful first step in initiating demand management without challenging the autonomy of clinical decision makers [97]. Information technology, such as electronic medical records, clinician order entry, expert systems, electronic handbooks and embedded hyperlinks in reports, is probably the easiest way to both provide solutions and monitor performance in these phases [98-100]. Clinical audits and clinician satisfaction surveys can also be useful measures of overall laboratory effectiveness [43, 101].

For many years, the clinical laboratory has been at the forefront of quality improvement activities in the healthcare sector. Its focus on analytical quality has resulted in an error rate of 4-5 sigma which surpasses most other areas in healthcare. However, greater appreciation of the prevalence of errors in the pre- and post-analytical phases and their potential for patient harm has led to increasing requirements for laboratories to take greater responsibility for activities outside their immediate control. Accreditation bodies such as JCI and CAP specifically require healthcare organisations to have clear and effective procedures for patient/sample identification and communication of critical results and to monitor their performance in these areas. There are a variety of free on-line resources available to aid in managing the extra-analytical phase and the recent publication of quality indicators and proposed performance levels by the IFCC WG-LEPS provides useful benchmarking data for laboratories embarking on extra-analytical quality improvement programmes. Managing the extra-laboratory phase of the total testing cycle is the next challenge for laboratory medicine. By building on its existing quality management expertise, quantitative scientific background and familiarity with information technology, the clinical laboratory is well suited to play a greater role in reducing errors and improving patient safety.