Repertoire review

The repertoire for the typical clinical laboratory should, theoretically, be reasonably standardised. However, local demographic factors, provision of specialist services (by the laboratory as a reference laboratory, and/or to the laboratory in terms of local clinical specialties) and workload/expertise for more specialist tests make this unrealistic. Consequently, each laboratory will need to constantly review their repertoire to respond to the local needs, national changes and new developments.31 For instance, the relatively recent implementation of the bowel cancer screening programme, along with accumulating evidence questioning the benefit of faecal occult blood testing in symptomatic patients32 has resulted in many laboratories withdrawing this test. Similarly, older cardiac enzyme assays have been progressively replaced by troponin tests, with newer candidates already on the horizon.33

Repertoire review should also cover repatriation of tests referred to other laboratories (or vice versa) in response to local patient population changes, capacity/cost review, network formation and so on. While this will not necessarily change the volume of testing, it may improve turnaround time and/or value for money for the local population, thereby aligning itself with the drive of demand management to reduce inconvenience for patients.34

Recommendation 5 : Review repertoire on a systematic basis, perhaps once every 6 months.

Vetting and within-laboratory alteration of requests

More centrally allied to demand management is the option to vet high cost/low volume tests, often those referred to specialist laboratories, on an individual basis. Using vetting, we have successfully reduced the number and cost of urine toxicology screens from 30 to <5 requests accepted per month (figure 1). Furthermore, through dialogue with the requestor, we discovered that most requests that were accepted required a ‘drugs of abuse’ profile (at a tenth of the cost) rather than a full toxicology screen, resulting in a saving of around £30 000 a year (and requestors receiving the correct test profile). While there is relatively little published work on the cost or clinical effectiveness of this approach (though several laboratories in the UK have evaluated this locally, with some success), Crispin et al 35 have advocated this from a quality improvement perspective in transfusion services.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Change in number of tests sent for urine toxicology screening (including drugs of abuse screens) per month following implementation of request vetting (two phases: October 2005 and June 2006).

Recommendation 6 : Systematically review the top 20 most expensive tests (cost per test) to determine whether they would benefit from vetting.

Many laboratories alter the requests internally, either by using ‘reflex testing’ protocols, or following clinical assessment of the results (‘reflective testing’).36 This is not a new concept. In 1988, Finn et al 37 showed that alteration of thyroid function test requests after they had been made, by either clerical staff implementing computer-based ordering menus or by laboratory staff utilising knowledge-based rules, significantly improved the appropriateness of test requesting. More recently, Srivastava et al 36 has used a combination of these approaches to improve the diagnosis of hypovitaminosis D, hypomagnesaemia, hypothyroidism and hyperthyroidism and haemochromatosis, a method used elsewhere to improve clinical utility of laboratory services.38 ,39

Recommendation 7 : Explore options to utilise IT to apply reflex testing. Examine the possible added clinical value of reflexive testing. Evaluate the effectiveness of such systems to ensure improved test appropriateness.

Duplicate requests and retest intervals

Of increasing utility in clinical laboratories is the identification of ‘minimum retest intervals’ to identify duplicate requests. This will be discussed further below in the section on electronic requesting, but is not limited to this approach. Many laboratory computer systems now allow the identification of a repeat request for specific tests within a user-defined time window. For instance, UK NICE guidance for Type I and Type II diabetes mellitus (DM) recommends HbA1c testing at 2–6-monthly intervals in patients with unstable diabetes, with a measurement made at an interval of less than 3 months being used as an indicator of direction of change rather than as a new steady state.20 ,21 In those with stable diabetic control on unchanging therapy, intervals of 6–12 months are recommended. Such data can be used to identify patients on whom repeat testing is requested within 2 months. We have recently examined the prevalence of such requests and showed that 21.3% of HbA1c repeat requests were received less than 2 months after the previous request.25

In addition to utilising published guidance to define recommended retest intervals, local agreements, analytical data (such as analyte half-life or reference change value) and consensus statements can be utilised.40 These data then allow duplicate requests to be automatically highlighted for laboratory review (either automatically rejected, or subject to individual vetting). Such intervals are either applied universally, or to selected requesters/locations. Table 3 shows some commonly used retest intervals from a recent survey by the UK West Midlands Association for Clinical Biochemistry Demand Management Forum. While this approach may appear straightforward, differences arise between primary and secondary care as routine tests performed in primary care can be simplistically divided into those used for monitoring or diagnosis, whereas, in the acute phase, the minimum retest interval is very dependent on the clinical state of the patient in addition to previous results. Nevertheless, this is a promising area and one which could certainly reduce the tendency to repeat tests at close intervals which may be inconsistent with the kinetics of the test in the disease in question.

While half of UK laboratories admit to using their laboratory computer system to identify such requests,41 and studies advocate this as an approach to demand management2 ,8 and estimation of its prevalence,25 ,42 ,43 there is a lack of published data on recommended minimum retest intervals or their effectiveness. An exception is the Guidelines & Audit Implementation Network (GAIN) Guidelines on the use of the Laboratory, published in 2008,44 a collaborative venture based in Northern Ireland. These guidelines provide recommended minimum retest intervals for a range of clinical laboratory tests spanning a number of laboratory medicine disciplines, though the underpinning evidence base specifically for the retest interval is unclear. The Clinical Practice Section of the Association for Clinical Biochemistry is currently collating data from UK laboratories to obtain consensus data on retest intervals. At the time of writing, these data are currently being circulated for comment and are due to be published in late 2012. This consensus, while useful in clinical practise, highlight that further research on the evidence for such intervals, and their impact on clinical and other patient outcomes, is required.

Recommendation 8 : Implement laboratory computer-based IT solutions to maximise the identification of duplicate requests within nationally agreed minimum retest intervals. Where possible, application of acceptance/rejection criteria and comments should be automated.

Education

There is a wealth of literature on the effect of a wide range of educational interventions on laboratory test utilisation, encompassing both verbal and written initiatives. These may include the general clinical liaison that is critical for the day-to-day operation of the clinical laboratory, simple memoranda and laboratory newsletters/bulletins, to formal educational material and/or training sessions, curriculum development and best practise guidance.2 ,6 ,45 ,46 However, the literature suggests that the effectiveness of such measures is variable.3 ,8 Table 4 lists some examples of educational initiatives, though this is by no means exhaustive. While these suggest that most (but not all) are successful, publication bias might suggest that the effectiveness of educational approaches is more variable than the literature indicates.

However, there is evidence of effectiveness in specific circumstances. For instance, we have previously shown that implementation of a locally agreed procedure, arising out of regular discussions with clinical teams, resulted in an 85% reduction in C-reactive protein requests from emergency admissions wards,58 though interestingly, this has been difficult to maintain following changes to the structure of the emergency portals in recent months. Wang et al 48 showed that development of guidelines on test-requesting by a multidisciplinary team in a coronary care unit resulted in reduction in test utilisation for a range of parameters that was evident a year after the intervention. However, while national guidance exists for many patient pathways, including advice on laboratory testing, their aim is not specifically to limit inappropriate testing.3 This, therefore, necessitates the development of locally agreed (and, therefore, potentially not standardised) guidelines.59 The effectiveness of such initiatives may therefore be limited to specific local circumstances due to variation in local expertise, structures and even personalities. Their reproducibility across many laboratories is therefore difficult to ascertain from the literature, and formal meta analyses are required.

It is also important not to neglect the ‘right process’ component of demand management education. Ensuring samples are collected in the correct manner is an essential facet of the laboratory's duty. The laboratory handbook is often used to address this issue by providing users with information on correct sample containers and volume, special conditions (eg, fasting, time of day, transport arrangements, etc) and correct phlebotomy procedure (to prevent haemolysis, ensure correct order of draw, etc). However, the number of incorrectly collected samples continues to be a major problem in many laboratories, suggesting that the laboratory handbook is more useful for laboratory staff and satisfying regulatory authorities, than influencing practise. Jacobsz et al 60 showed that 1.5% of samples were rejected, only half of which were repeated correctly, and 5% of these had results with critical values. Hence, it is important that further work be carried out on exploring more effective means of reducing this form of waste.

As with any educational intervention, guidance on requesting also needs to be repeatedly reinforced and kept up to date.

The overall view of these initiatives suggests a number of key messages:

1. Sustainability remains a challenge, though it can be achieved with regular reinforcement.52 It is, however, not an inconsiderable task given the range of scenarios where laboratory tests are required.3

2. Successful schemes more commonly use a combination of approaches.6 ,61 Solomon et al 45 performed a literature review of educational approaches and highlighted that interventions based on ‘multiple behavioural factors’ had a higher rate of success.

3. Joint initiatives from laboratories and service users (including patients) are more likely to lead to success, indicating that close laboratory–clinical liaison is critical (see also3).

Recommendation 9 : Cultivate links with all key specialties (including primary care) to allow establishment of suitable educational forums and ensure educational interventions have the backing of clinical teams.

Recommendation 10 : Develop a local educational strategy for requesting, drawing on national guidance and literature-based evidence where possible. Establish an ongoing plan to review its relevance.

Recommendation 11 : Utilise training programmes to address correct sample collection, timing, transport and so on.

Request form design

The design of the request form has long been a potential target for demand management strategies.3 ,8 ,62–64 However, these issues have largely been superseded by the issues arising from the implementation of electronic test requesting, which is addressed in more detail later.

Profiles and test combinations

Allied to request form redesign, is the area of the conventional ‘test profile’. Historically, these have been organ-based profiles such as liver, kidney and thyroid, which provide a set of tests that offer information about the state or functioning of the organ system. Not all of these are specific to the system concerned, and profiles vary considerably. The 2011 report from the National Pathology Benchmarking Service,41 which included approximately 50 UK laboratories, demonstrated that 12 different ‘liver function test’ profiles were used. A consensus report65 has recently been put together proposing to harmonise these different profiles as the additional tests that are included in some are not only a source of extra cost but also one of potential for confusion for doctors moving between different hospitals. This is, however, potentially only the start of a process which could move more towards diagnosis-based testing rather than organ-specific profiles. Many patients presenting with non-specific symptoms, for instance, are investigated for a range of possible diagnoses which may span different organ systems. In addition, monitoring tests could be considerably streamlined—there is, for instance, no clinical need to measure all the tests currently included in a ‘liver function test’ profile if the aim is to detect a change in a liver transaminase in a patient taking a drug which carries a risk of hepatocellular toxicity. This, potentially, significantly reduces the marginal cost of these tests by removing clinically unnecessary analytes. While these are measured in pence per test for most routine tests, the numbers involved make the potential savings from redundant tests worthy of consideration though, in some cases, the continued requirement for analyser maintenance, internal quality control, external quality assurance and so on, need to be taken into account when calculating savings. Winkleman,66 showed that a 10% reduction in utilisation of a single high-volume, low-cost test only released 1.32% of overall costs.67 Hence, the real savings associated with reduced test volumes are often significantly less than initially envisaged.66 ,68

Doing this successfully, however, requires more refined requesting systems than can be offered by traditional paper-based request forms. Far more clinical testing scenarios exist, for instance, in primary care, than could ever be listed on a paper request form, but it is not difficult to imagine how this could be approached by an electronic selection menu. This is a promising opportunity for the electronic requesting systems to offer a default test selection which can be supplemented if necessary, rather than a larger profile being selected by default. As yet, very little has been published on this subject.

A more immediate opportunity exists in hospitals with acute admissions where the list of common reasons for admission is shorter, and is compatible with paper request forms. Again, very little has been published on this subject, although the concept of the admission profile was described 30 years ago.69 This, again, offers the opportunity to select desired tests and ensure that first, important tests are not omitted, but second, by offering a default set of investigations, it potentially reduces the likelihood of inappropriate tests being added. Locally, this has been a problem with requesting of D-dimers in situations in which venous thromboembolic disease is not suspected, but in which the D-dimers are raised for other reasons, producing clinically false positive results.

Recommendation 12 : Examine the possibility of standardising common profiles in line with national benchmarking data.

Recommendation 13 : Consider the potential of implementing symptom-specific profiles.

Test report-based education

While many studies have advocated various approaches to this in an attempt to prevent inappropriate requests, the simplest means within the control of the laboratory is to provide education on appropriate testing either at request (see section on electronic requesting below), or via the final report. For instance, provision of interpretative comments on test reports is not only welcomed by users but has been suggested to influence requesting behaviour and, indeed, patient outcome.70 ,71 Vasikaran72 stated ‘Interpretative commenting should go hand in hand with regular contact with clinicians to develop a dialogue about appropriate testing…’. We have used such reports to identify and reject duplicate requests for HbA1c,25 for instance, and also provide data on when the previous test was performed, its level, and on the laboratory minimum retest interval. Such high-volume tests are amenable to automated comments, while more specialist tests warrant individualised information on appropriateness.71 While, in this latter case, it is difficult to quantify its impact, it is hard to argue that it has no beneficial effect on future requesting behaviour. However, some studies suggest that appending educational messages to reports, while effective, affects requesting behaviour to only a moderate extent,16 unless they interrupt the requestor's routine.3 Furthermore, the variability in content and appropriateness of interpretative comments themselves has led to the development of a peer-assessed external quality assessment scheme in an attempt to improve the comments themselves.73

Recommendation 14 : Develop a strategy to review all automated interpretative comments to include information on test appropriateness.

Feedback on test usage and costs

Various studies have been published showing the effects of feeding back either cost or volume information to users of laboratory services (table 4).6 ,51 ,74–76 The impact of feedback of test costs is likely to depend on the requesting physician's direct responsibility for laboratory expenditure. While it is possible that providing cost information would encourage physicians to concentrate on the quality of their investigations, it is unlikely that this information alone will be the key driver. As stated by Hoey et al,76 ‘A large proportion of tests are ordered because of clinically absolute reasons, which may be insensitive to price’. Hence, while generating comparative test volume or cost data (benchmarking) can be useful to highlight differences between requesting physicians to identify areas of potential high or low use, is not, in itself, sufficient to effect change.77 Furthermore, the complexity and, hence, cost of implementing regular feedback systems may make the resultant benefits insufficiently large to justify them.78 A combination of educational initiatives with facilitating mechanisms has been shown to be one of the more effective ways of influencing demand.14 ,45 Feeding back volume information is relatively straightforward in a primary care setting in which overall case mix between general practises is relatively homogeneous and differences in testing activity are too great to reflect practise demographics.23 This is, however, less straightforward on a hospital basis given the specialties and subspecialties of different consultants, and the multitude of clinical staff that may rotate through departments. In primary care, a group of approximately 20 tests (or profiles) make up the great majority of tests which are performed, most of which fall into the low-cost, high-volume category. In hospitals, however, the test mix is considerably larger, and test costs may differ by a factor of several hundred, so a more targeted approach is needed.

Recommendation 15 : Consider providing feedback on cost and volume data (and where possible, inappropriate request data), particularly to primary care using comparative data (benchmarking) to utilise peer pressure. Consider primary and secondary care feedback on a longitudinal basis to evaluate trends.

Financial incentives and penalties

With the advent of almost universal savings targets for healthcare agencies, the incentive to save on laboratory costs is an inevitable consequence. However, this is not a new phenomenon. For many years, the US model that paid a fixed tariff to physicians based on ‘episodes of care’, or other finance-based contracts between service users and laboratories, gave rise to increased financial awareness, though the savings resulting from this were often seen in ‘big ticket’ cost elements, such as referral rates, rather than laboratory test requesting.6 ,61 A wider review of financial control initiatives is provided elsewhere.6

In the UK, the introduction of the primary care Quality and Outcomes Framework (QOF) has provided general practitioners with financial incentives for performing certain tests in specific groups of patients. While this may potentially lead to increased testing, the argument is that it improves quality and equity of care. However, in terms of test appropriateness, there is sometimes a disconnect between other national guidance and QOF indicators. For instance, QOF guidance on testing frequency for HbA1c in patients with diabetes focuses on ensuring that patients have been tested within the previous 15 months,79 while NICE guidance suggests more frequent testing, especially within those with changing treatment or poor control.20 ,21 Hence, as expected, when we examined the impact of QOF guidance on testing outside NICE guidance for HbA1c, no impact was observed.25 This has not been the case for other examples, such as urine albumin:creatinine ratio testing following the introduction of the quality indicators from the Northern Ireland General Practise QOF for patients with diabetic nephropathy in 2004, where testing doubled following its introduction.80

Recommendation 16 : Consider aligning demand management initiatives to local/national financial incentives and/or penalties for requestors.

Specialty/staff grade limitations

A further way to prevent inappropriate requesting, particularly of the specialist tests, is to limit requesting to specific specialties and/or more senior staff. This approach is allied to the vetting approach described above, and while potentially effective, it is often difficult to police. Smith et al 81 implemented a scheme whereby out-of-hours requests were processed only following a discussion between the requesting consultant and consultant-level laboratory staff. This reduced out-of-hours requests by 40%. A similar approach by Morgan et al 82 reduced the number of out-of-hours calls to the laboratory by 80%, though evidencing the cost-effectiveness of these approaches, particularly given the increase in test volumes in the decades since this was trialled, may be a challenge. In our study on C-reactive protein requesting in the emergency portals, our approach included processing requests for tests only if a consultant's signature was supplied on the request form.58 Since the advent of electronic requesting, specialty- and staff grade-specific requesting is potentially more easily applied, and is therefore addressed in more detail below. However, it should be stated that while this approach can prevent tests being inappropriately requested, if it is not set up and policed carefully, it can result in the opposite effect.

Recommendation 17 : Review specialist tests to determine whether consultant-only and/or specialty-specific requesting would be appropriate.

Guideline and protocol development

National guidance (eg, NICE) provides some information on appropriate testing, though in some cases, this is somewhat hidden among other management and treatment instructions in large documents, spread among a range of sources and/or aimed at laboratory specialists rather than requestors.22 Hence, it may have little impact on appropriateness of requesting.4 For instance, we have shown that guidance on testing frequency for HbA1c in diabetes, though present in NICE documents for both Type I and Type II diabetes, has had little impact on HbA1c undertesting or overtesting.25 Hence, local guidance, or targeted protocols, may be more effective, though this then potentially introduces variability in practise and possible conflicting messages. It is essential, therefore, that such local guidance is based on national data and involves all stakeholders in its development and implementation.48 Ideally, as illustrated by Wang et al 48 and van Wikj et al,64 guideline development will comprise a component of a wider educational strategy. Despite its potential, influencing testing via development of clinical protocols,3 requires a long-term commitment and requires careful liaison between key stakeholders. This contrasts with laboratory-based protocols (eg, test repertoire, reflexive testing procedures, rejection parameters, etc), which are much easier to control, but do not prevent unnecessary phlebotomy episodes. However, there are some suggestions that the effectiveness of clinical guidelines in influencing clinical practise depends on the way in which the guidelines are implemented.83 Furthermore, protocol-driven requesting can, if not implemented and managed carefully, result in an increase rather than a decrease in inappropriate requesting.61 ,84

Whatever approach to education is used, a critical factor in ensuring the success of such approaches is the need to ensure a long-term sustained effect so as to prevent regression to original patterns,14 including taking into account healthcare staff turnover.85 Without careful planning, monitoring and reinforcement, the general impression is that long-term effectiveness is somewhat limited.46 Many studies suggest short-term effects,45 though this may reflect some degree of publication bias. However, this opens the door for opportunities to explore the role of electronic requesting as a means to reinforce educational messages and sustain their effects.

Recommendation 18 : Integrate a systematic review of effectiveness into the demand management strategy. Focus on educational approaches that can be monitored and, if necessary, re-emphasised on a regular basis to inform new staff.

Recommendation 19 : Develop local best-practise guidance, allied to national data, on appropriate testing. Implement an ongoing review to ensure that requestors are aware of the guidance.