Abstract
Clinical laboratories have an important role in improving patient care. The past decades have seen enormous changes with unpredictable improvements in analytical performance, range of tests and capacity to manage large volumes of work. At the same time, there has been a dramatic fall in the rate of laboratory errors. However, there is now a growing awareness that the testing process includes the time before samples reach the laboratory and after reports have been printed and that these areas need to be included in the quality assessment of the total testing process. Laboratory quality should include a focus on patient safety and clinical effectiveness. Services should be patient-centred, timely, efficient and equitable, and finally, should be moulded to ensure optimal outcomes. There is a need to define quality indicators that will ensure there is appropriate choice and selection of tests, use of the appropriate assay standardization and the correct interpretation of the assay results at the appropriate time. These are the areas in which a quality laboratory can, and should, now involve itself.
Introduction
Quality in laboratory medicine is often described as doing the right test at the right time for the right person. Despite the clear implication of this aphorism, current quality indicators in laboratory medicine are focused clearly on the performance and efficiency of the operational processes. The lack of attention to extra-laboratory factors contrasts with the wealth of evidence pointing to the predominance of errors in the pre- and postanalytical clinical arenas. 1 Over the past few years, health-care systems have been trying to reduce their costs; it is now timely for laboratory medicine to expand its contribution to patient care and safety by taking a more active role in this arena. This will be consistent with the changes being introduced by health-care systems which are now focusing on patient outcomes rather than counting clinical activity. To date, it has proved difficult for laboratory medicine to evaluate its contribution to patient outcomes, but it will be necessary to develop clinical metrics that will capture the benefits that its practitioners believe them to make. 2 This paper will explore surrogate markers than can be used both as drivers for improvement of patient care and as being indicators of quality.
What are quality indicators and why measure them?
Health-related indicators have evolved into two main types. The first indicator might be classified as having a primary utility for driving improvement and might include operational benchmarking and external quality assurance for comparison with peers. These should be used to monitor progression and improvement as well as having a function of providing a prompt for investigation to explain deterioration or outlier status. The second form of indicator has a more corrective function for performance management and accountability such as turnaround times, complaints and test costs. It should be noted that not all indicators are designed to reflect quality outcomes, and measurement may use incorrect metrics or miss areas where evidence and/or data are missing. The objective of collecting and analysing indicators is to have a positive impact on performance, but it is important to note that they will only be productive with useful outcomes if there is clarity about the aims and objectives at the onset. Otherwise they lose credibility even if the data have been correctly collected.
Quality is an all-embracing term that covers the entire spectrum of the total testing process. Since laboratories are subjected to external accreditation which should provide confidence to users that the correct answers are given, why do we need clinical quality indicators? Accreditation ensures that a laboratory performs its processes correctly, i.e. that it does the test right. It does not, however, ensure that the right test is performed and analysed. As an example, measurement of HbA1c within agreed turn-around times and with agreed precision would indicate good performance, but a high-quality service might be indicated by analysis based on International Federation of Clinical Chemistry (IFCC) standardization with IFCC units of reporting and reporting the time interval since the previous test.
Quality indicators can be considered to be an evolution from audit. The lessons learnt from the early audit processes were that they needed to be performed by individuals who could see benefits for themselves and with management support to ensure that any errors and inefficiencies were corrected and became incorporated into routine practice. Quality indicators should therefore be part of a coherent and integrated quality improvement strategy.
A broader aspect is the understanding that the design of metrics must fulfil the requirements for the needs for improvement and performance management while simultaneously being measureable. Additionally, they need to be measured easily so that they can be perceived to be relevant. Since the purpose of such indicators is to change practice, they should be studied with the same rigour as diagnostic tests with utility assessments such as sensitivity and specificity.
Current quality indicators
There are a number of programmes aiming to improve quality in laboratory medicine taking place at national society level around the world, but many are small and they are uncoordinated. 3 A more comprehensive programme is being driven by the Division of Laboratory Systems in the Centre for Disease Control in the USA. They have established a group to examine clinical quality issues with an aim of producing best practice guidelines. 4 To date they have pilot-tested new methods to conduct laboratory medicine quality improvement reviews and are seeking collaborative help to identify good practices.
An example of their work is the recent systematic review of quality indicators in laboratory medicine by Shahangian and Snyder 5 who recognized that there was a ‘…considerable challenge in identifying, defining, and, ultimately, implementing indicators that cover the … laboratory testing process in general and specific to different diseases and conditions’. This may explain why they found an evidence base for only 14 criteria across the whole range of pathology. These were test order appropriateness, wristband identification errors, patient satisfaction with phlebotomy, specimen quality (adequacy, contamination, information error), proficiency performance, cervical cytology–histopathology mismatch, availability of inpatient results, corrected laboratory reports, critical value reporting, turnaround time, clinician satisfaction and follow-up of abnormal cervical cytology. These metrics are largely operational and are clearly focused on prevention of errors.
Recognition of errors in laboratory medicine has followed an evolution from purely analytical errors, through to all laboratory errors and has now progressed to include errors in the whole testing process. Plebani 1 has studied this area for many years and has documented and reviewed the fall in error rates over the past decades. He has developed a concept of laboratory medicine errors which is best illustrated as an hour glass. In this image, the preanalytical, analytical and postanalytical laboratory errors are closely gathered about the waist of the hour glass whereas the extra-laboratory errors fill the broad bulbs of the glass. This is based on data which suggest that the overall error rate in diagnostic testing is about 20% but that only a quarter of these errors are due to problems within laboratories. 6 A full understanding that so many errors in testing occur outside the laboratory is important in order to ensure that resources are appropriately utilized to minimize error rates. The problem is how to recognize and address them.
Recent evaluation of laboratory performance has started to examine clinical outcome indicators rather than operational indicators. These indicators are both within and without the laboratory and include topics such as quality of patient identification, number of repeat venesections, wasted samples, testing and reporting errors. 7 Newer reports are examining the failure of clinicians to act on abnormal results. 8 A further development into the misinterpretation of laboratory results is still in an anecdotal phase. 9 It must be restated that understanding the source of the errors can help explore solutions to poor patient outcomes.
Over the past years, hospital functions have become more specialized and this has led to health-care staff becoming isolated in their departments. This leads to unrecognized areas of discontent. The Institute of Quality in Laboratory Medicine has explored the effect of the remoteness between laboratory staff and nurses. They have highlighted a number of areas of discontent, many of which might be solved by breaking down the barriers that prevent communication, and their correction would thereby improve testing quality. 10 These discontents included speed of sample collection, skill of phlebotomy staff, lack of notification of change in procedures and inability to coordinate multiple tests for the same patient. A particular irritant for nurses was the turnaround time for acute samples – these were recorded differently by the laboratory and the ward-based nurses due to the inclusion or exclusion of the transport time!
Where are we right now?
Laboratory investigations are largely requested in an unrestricted manner by clinicians on the basis of clinical evaluation, and are analysed by laboratories that send the results of the analyses back to the requesting clinician. However, is this the best model? Clinicians request tests for all sorts of reasons and take samples at all times which may not necessarily give the best answer for a beneficial patient outcome. The many reasons why doctors do laboratory tests have been explored (Table 1). 11 Over past years, there have been significant changes in medical school training and huge changes in the range and quality of laboratory analyses which means that junior doctors are now uncertain over the best use of laboratory tests 12 and are almost certainly using them incorrectly on many occasions (Table 2). 13 The knowledge base for laboratory test utility has shifted towards the laboratory, and patient outcomes may be enhanced by including laboratory physicians and scientists in diagnostic pathways. Our preliminary data would suggest that we are not there yet. 14
Why do doctors request diagnostic tests?
The table is constructed on the basis of an in-depth review designed to explore all the factors that affect a doctor's decision to perform a diagnostic test 11
Classification of extra-laboratory test errors 13
Patient safety
The 1999 Institute of Medicine report, ‘To Err is Human: Building a Safer Health Care System’ estimated that up to 98,000 preventable deaths occurred per year due to medical errors in the USA. A further report in 2006 ‘Preventing Medication Errors’ suggested that 1.5 million Americans have an adverse drug effect every year. There has been no such systematic study of the adverse effects of laboratory tests, such as, the failure to perform an appropriate test or the effect of misinterpreting tests. The IFCC has an active working group on patient safety but which is focusing on measurable factors, engendering a more laboratory centric approach than the present author would suggest is ideal. 15 There is an international guide to the relative quality of laboratory testing which shows that the UK scores well in this field (Table 3). 16
International comparison of laboratory testing indicators
Data related to laboratory tests extracted from an international comparison of health care by the Commonwealth Fund. The data were obtained from surveys of both patients and physicians over the years 2007–2009 16
We do know that testing errors affect patient care. Errors in analysis have been reported to be related to adverse clinical interventional activity. 17 Errors in testing, in particular overtesting, have introduced unnecessary delays in patient care due to the requirement for follow-up of slightly abnormal tests, 18 and an increase in testing leads to an increase in clinical interventions. 19 Of more concern is that inappropriate tests can be frankly misleading since misinterpretation can lead to further unnecessary investigations and clinical interventions. 20
Laboratories contribute to patient safety by identifying results that require rapid action. These can be communicated to the appropriate clinician by telephone or computer notification. 21 However, although communicating a critical results can be lifesaving, it should not just consist of telephoning a clerk or providing an electronic alert, since neither guarantees a clinical response. 22,23 Telephoning critical results is probably routine practice in most laboratories, but despite this, there is no consensus regarding which tests or which values are critical. 24,25 There is a need to build an evidence base for this practice since outcome data are presently available only for calcium and haemoglobin. 26,27
Laboratory medicine has the potential to improve patient safety, since it crosses many pathways and organizational boundaries. Proactive interventions can be implemented to highlight high-risk situations such as in the preassessment and monitoring of drug therapy in the elderly where toxicity due to polypharmacy and underlying disease is very high. 28
Clinical effectiveness
Clinical effectiveness might be defined as the provision of services based on scientific knowledge to all who could benefit and withholding them from those unlikely to benefit. This should be interpreted in the context of laboratory testing as avoiding underuse and misuse of tests.
Laboratory clinicians make a significant contribution in guiding appropriate clinical decision-making to ensure effective use of the laboratory. This may be proactive in the development of testing strategies or reactive by helping to interpret laboratory data. Data interpretation by a duty laboratory clinician is an important contribution to patient care in the UK, 29 Australia, the Netherlands and other countries. This interpretation may be reflective, i.e. interpretational or reflexive where additional analyses are added on within the laboratory, e.g. magnesium to help explain hypokalaemia or hypocalcaemia or by providing advice on antimicrobial sensitivity. 30 This extra-activity speeds the rate of diagnosis. 31,32
Laboratory clinicians can also contribute to clinical effectiveness by identifying factitious effects that alter laboratory analyses. Some are easy to identify such as haemolysis, but others may be more difficult to recognize because the effect occurs before the sample arrives in the laboratory, such as the ambient temperature effects on potassium, 33 or because the effect is an analytical issue not recognized by clinicians, such as the effects of very high leukocyte or platelet numbers on potassium, or the effect of high proteins on measurement of phosphate concentrations. 34
Many national and international guidelines include recommendations for laboratory testing. Laboratories can contribute to patient care by ensuring that their clinical testing policies conform to these guidelines. Laboratories should use methods that are consistent with international standardization so that their tests can be compared with published data and also that the laboratory results can be used as surrogate markers for patient outcomes.
Guidelines which incorporate laboratory test values in clinical decision-making processes are often developed from studies which use a single laboratory to perform the analyses. In practice, different laboratories use a myriad of different methods and analytical platforms. The combination of these factors increases the imprecision and bias at the decision-making values which will have a detrimental effect on the quality of clinical decisions. The financial cost of this effect is significant. The National Institute of Standards and Technology have calculated that for serum calcium, the cost is $34–89 per adult person for every shift in bias by 0.125 mmol/L. 35 The effects of analytical bias in diabetes, heart or renal disease may be measured in altered disease severity and life-expectancy. The minimum analytical performance standards (MAPS) project in the UK is currently trying to define how much imprecision and bias is clinically acceptable and it is to be hoped, will be able to publish its preliminary conclusions in the near future. 36
Patient-centred
A patient-centred service provides care that is respectful of and responsive to individual patient preferences, needs and values, and ensures that patient values guide all clinical decisions. It is almost 20 years ago since Stewart proposed ‘a patient's charter for laboratories’. He suggested the following: (i) blood test results should be returned in a timely manner; (ii) venepuncture should be available when convenient for the patient; and that (iii) the appropriate clinician would be telephoned if the sample was or became unsuitable for use. Since the publication of these standards, it is undoubtedly true that blood tests have been processed and returned faster but as investigations play a more central part of patient management, this turnaround will never be fast enough. Secondly, if care and thought is used to ensure that the optimal sample is taken, there will be no need for the inconvenience (to the patient) of a repeat venepuncture. Finally, I fear that workload has risen to such an extent that not all physicians are individually informed by telephone of unsuitable samples.
Since Stewart's proposals, patients have been empowered and given every support to help them care for their own diseases with the aim of enhancing their personal experience regardless of clinical outcome. This change will require the re-engineering of patient pathways. Some institutions have created outpatients with open access rather than conventional appointments and laboratories will need to follow this path and provide direct access. 37,38 Laboratories have an essential role in the provision and support of point-of-care testing by patients. For those patients undergoing laboratory tests, patient experience can be improved by ensuring the quality of phlebotomy is adequate to obtain samples on the first occasion, as this will minimize the pain, bruising and blood loss caused by repeated venepunctures. 39
There is a growing move to giving patients access to their own test results, which it could be argued that they already own! There has been some resistance but the experience of those physicians who do give access to their patients has been positive
40
and the move is supported by the Royal College of General Practitioners.
41
The Association for Clinical Biochemistry, together with other professional associations, has been supporting this approach by providing education for the lay public with its labtestsonline website (
Timely
The most efficient patient process will also be the most timely. It will consist of the right test being ordered, analysed, reported promptly and viewed and acted upon. This will reduce patient delays which may be harmful for those who receive care.
Timeliness of laboratory tests are normally measured by laboratories as the ‘within laboratory turnaround time’. Kurec and Wyche 10 found that this was not the turn around that clinicians were interested in and that transport times need to be included. A further time frame that needs to be considered is the report delivery to clinical action time. Singh et al. 44 have explored results that were sent to clinicians with serious abnormal flags; 10% were not electronically acknowledged, i.e. unread; and almost 7% were not followed up in a timely manner. Perhaps more worryingly, 17% of the alerted results were for tests considered to be redundant. 45
How many tests are actually looked at? Callen et al. 8 have performed a systematic review of test results and noted 20–60% of inpatient results and up to 75% of emergency test results were not followed up. This study was reported in the national press in the UK by the Daily Telegraph on (7 February 2011) focusing on the figure of 75% of tests.
There are a number of reasons for lack of action taken on test results. Roy et al. have prospectively explored test results that arrived after patient discharge. Physicians were unaware of 61% of these tests and 13% of all the results required urgent action. 44 Significantly, 41% of all patients in that study had results that were returned after they had been discharged from hospital.
In the past, laboratories took considerable effort to design paper reports. These have now been replaced by electronic reports on computer screens over which the laboratory professional may have no influence. A survey of internal medicine physicians reported that a large number were dissatisfied with the way they managed their laboratory reports and often wished they had seen a result earlier. 46 This has resonances with a study from England in which 50% of emergency departments did not have a system to review their results, 47 which may explain why so many reports are not actively followed up as discussed above.
Efficient
Strategies to modify test-requesting patterns were last reviewed in this journal in 1987 by Fraser and Woodford (Table 4). 48 They reported on attempts to improve testing by feedback, education and even bribery, but noted all beneficial effects were lost after the intervention ceased. 49 Further studies have suggested that persistence in providing feedback may be more fruitful and lead to increased improvement in efficiency. 48
Interventions to improve testing 48
In order to understand how to improve the utilization of tests, it is necessary to understand why they are used. Whiting et al. 50 have reviewed this topic and Schattner 13 has described the permutations of extra-laboratory errors (Tables 1 and 2) and, distressingly, fewer than 5% of physicians formally evaluate the diagnostic utility of the tests they use. 51 If, therefore, laboratory tests are performed for the wrong reason, 52 and the results are frequently not looked at 8 and the majority do not contribute to patient management, 53 what can be done to improve the situation?
There is now little education in laboratory medicine in medical schools which leaves junior doctors feeling uncertain about the choice and interpretation of test results. 12 Postgraduate education is well received in primary care and is effective in improving laboratory testing. 54 This education can be purely didactic or by post-test feedback to clinicians. 55,56 We clearly need to redouble our efforts at improving knowledge about the optimal use of laboratory tests by our clinical colleagues.
Request card design has been used as a tool to improve test requesting. In the simplest example, a change from multiple tick boxes to a more limited list leads to a reduction in tests. 57 More sophisticated approaches have been described such as a checkerboard design with tumour markers on one side and location of primary cancer on the other using blocked areas to prevent inappropriate requests, 58 and another which used bar codes labelled by clinical symptoms to generate protocol led requests. 59 Both have helped to reduce the use of diagnostically useless tests. More recently, computer algorithms have been built into electronic-based ward ordering and this has been successful at implementing appropriate use of troponin testing. 60 It is fundamental that the use of decision support in primary care does not significantly reduce the number of requests, but it does improve the quality of testing as indicated by greater adherence to guidelines and protocols. 61 Computer testing algorithms are, however, not a panacea. They require computer input which replaces one set of problems with another set of unintended consequences such as increased work in the process of test ordering, recurrent training demands and reduced communication regarding patient care. 62
Computerized aid for reducing redundant tests has been studied in a randomized controlled trial. This has shown that redundant and duplicate testing can be reduced but that the savings appear to be quite modest, 63 which might be regarded as unconvincing by the silo mentality of current pathology management. However, the real savings are likely to be made by reducing patient time waiting for the results of these unnecessary tests and in the reduction of inappropriate actions based on their results. This latter thesis has yet to be systematically studied.
The use of biomarkers to monitor disease has been well established over past years and is keenly embraced by enthusiasts. It is only recently that studies have explored the relationship between testing rate and outcome. O'Kane et al. 64 have challenged the concept that the rate of testing is related to outcomes. In contrast, a different study, using glycated haemoglobin as an indicator, has shown that more tests are done in patients with complications, suggesting that the greater testing rate reflected greater clinical activity and was probably appropriate. 65
Equitable
Laboratory medicine is the area of health care for which it is probably easiest to ensure equitability since samples are largely anonymous and are processed independent of gender, ethnicity, geographic location and socioeconomic status. However, this may not be true outside the UK due to different laboratory funding arrangements.
It is probable that the most important source of inequitability arises when appropriate testing is not performed because of the absence of systems to ensure comprehensive delivery of care. This is a particular problem for long-term management of medication. There are, however, good examples of laboratory-based automated systems to ensure that no patients are unintentionally missed. These have been developed for thyroid disease 66 and for monitoring lithium therapy. 67 Due to the potentially serious side-effects that disease-modifying antirheumatic drugs (DMARDs) can occasionally cause, regular blood monitoring is necessary. A number of community-based programmes have been developed in England which improve monitoring. 68 The DAWN system for anticoagulant therapy delivers improved control. 69
Outcomes
It is difficult to isolate the exact contribution that laboratory medicine makes to patient outcomes as patient care is delivered by so many individuals and affected by so many external forces. Paradoxically, the reverse has been proposed with patient outcomes being used to evaluate laboratory error rates. 70
Patient outcomes are worse where analyses are performed in laboratories with low volumes of work, or if serial tests are done by multiple laboratories, whereas they are improved where there are strong laboratory regulatory programmes. In the case of prothrombin time measured in physicians' offices as an example, where test volume is below 40 per month, the rates for stroke (CVA) or acute myocardial infarction (AMI) are 1.96 and 3.43 times greater, respectively, than in laboratories with greater workload. The effect of switching between laboratories increases the risk of CVA or AMI to 1.57 and 1.32 times greater, respectively. 71
The effect of diagnostic errors on patient outcomes has been examined by two studies in USA-based primary care practices. Hickner et al. 72 reported on a survey of 58 physicians who analysed 590 errors and noted that 18% of patients had experienced some harm ranging from financial costs, time costs, delays in care and adverse clinical consequences. A further study in primary care suggested that an error in the testing process had had an effect on patient care in 27% of patients in whom a testing error had occurred. 73 Laboratories have a duty to educate their clinical colleagues about these findings.
It is clear that laboratories have a huge role to play in diagnostic and therapeutic decision-making and monitoring safety. There are many such examples highlighted in this paper but increasing the knowledge of the use and abuse of testing, and studying the outcomes should be part of the value-added function of laboratory services. This will improve the quality of care and affect the final outcomes.
How should quality indicators be selected?
Choosing quality indicators for clinical laboratories is a balance between satisfying patients' and clinicians' convenience and requirements, helping clinicians avoid diagnostic mistakes and providing a service that is optimal as measured by analytical performance. There are some prerequisites for quality indicators. They must be timely so that users receive the measures before they become outdated. They need to have thresholds for acceptable performance. Quality indicators might be used as a measure of laboratory improvement so they should not be chosen lightly.
How should the best indicators be chosen? A list of potential indicators is listed (Table 5). This list has been developed from a number of focus group meetings with laboratory physicians and scientists who have identified indicators of the functions they feel are most important. There are, however, far too many on that list and collecting so much data would divert so many resources away from the primary function of the laboratory as to reduce quality. The list should probably be narrowed down by users of the laboratories. 74
Potential quality indicators
This list of potential indicators has been developed from a number of focus group meetings with laboratory physicians and scientists (n ∼ 60) who have identified indicators of the functions they feel are most important
GP, general practitioner; EQA, external quality assessment; POC, point-of-care; DMARD, disease-modifying antirheumatic drug
Once chosen, indicators should be evaluated with the same rigour as diagnostic tests with investigation of their specificity and sensitivity and overall diagnostic utility. This will ensure that users will have confidence in their value.
Summary
Clinical laboratories have an important role in improving patient care. The past decades have seen enormous improvements in analytical performance, range of tests and capacity to manage large volumes of work. There is now a growing awareness of other aspects of laboratory quality. These include appropriate choice and selection of tests, use of the appropriate assay standardization and the correct interpretation of the assay results at the appropriate time. These are the areas in which a quality laboratory can, and should, now involve itself.
DECLARATIONS