Point-of-care tests for urinary tract infections to improve antimicrobial prescribing: early value assessment

3.1 Patient experts explained that having a UTI, particularly if it is recurrent or chronic, has a negative impact on people's social life and relationships, and can mean they need to take time off work. They explained that children with UTIs often have to miss school, which can affect their education and friendships. Patient and clinical experts said that current tests may not identify certain infections, even if a person has typical UTI symptoms, which can lead to a misdiagnosis and delayed treatment. This may increase pain and discomfort and can negatively affect people's mental health. Clinical experts explained that acute lower UTIs usually have mild, but uncomfortable, symptoms. They noted that most people do not go on to develop complications, such as pyelonephritis or bacteraemia. Delays to diagnosis of chronic UTI were highlighted as a particular issue. Clinical and patient experts noted that people are often willing to pay for private tests so that they can get a diagnosis and return to work sooner. Clinical experts explained that people with chronic UTI symptoms but negative tests are often referred to secondary care, where misdiagnosis is common. They noted that referral to tertiary care for further testing can improve diagnosis and treatment, but access to this in the NHS is extremely limited. This may increase the time to diagnosis while the person has to live with ongoing symptoms. The committee agreed that UTIs, particularly recurrent or chronic UTIs, can have a large mental and financial burden.

3.2 Patient experts explained that new point-of-care tests that can more accurately diagnose UTIs will improve patient outcomes and quality of life, and reduce side effects from taking unnecessary antibiotics. Faster access to effective treatment would also relieve symptoms more quickly and effectively, although some groups may receive antibiotics before test results are received in current care. New point-of-care tests may reduce the need to provide repeat urine samples, which may benefit groups who find this difficult, such as people who are pregnant, older people, people who are incontinent or people with dementia. Reductions in the need to travel to appointments (and associated costs), drop off samples, and pick up prescriptions may also benefit people with a lower income and disabled people. Patient experts noted that people with neurogenic bladder, diabetes, polycystic kidney disease, and people who are immunocompromised have a higher risk of complicated UTIs and need to receive treatment as soon as possible. They said that more accurate tests that could lead to quicker treatment may particularly help these groups. Clinical and patient experts also noted that there may be greater benefits for people who cannot have dipstick testing. Any improvements in antibiotic prescribing could reduce the risk of antimicrobial resistance, benefiting wider society. The committee accepted the potential benefits of newer point-of-care tests but acknowledged that less accurate results or delays to prescribing compared with current testing could lead to increased patient distress. A comment received at consultation highlighted the importance of educating patients and GPs, for example on collecting urine samples if new tests were introduced, to ensure any potential benefits are realised.

3.3 UTIs are currently diagnosed using a combination of clinical symptoms, dipstick tests (if appropriate) and laboratory-based culture testing (see section 2). Two randomised controlled trials, Butler et al. (2018) and Holm et al. (2017), evaluated how the culture-based Flexicult Human test affected antibiotic use. Both studies reported that there was no statistically significant difference in overall antibiotic use. Clinical experts said that, because the test takes up to 24 hours (and at least overnight) for results, clinicians may have continued to prescribe antibiotics if participants did not want to wait for test results before getting treatment. Clinical experts said that faster point-of-care tests could have a different effect on antibiotic use. The EAG said that the clinical and patient experts who they had consulted said that tests that gave results on the same day were likely to have the largest effect on antibiotic prescribing and patient outcomes.

3.4 The point-of-care tests evaluated for this guidance varied in how long they took to give a result. The EAG grouped them into rapid tests (results in less than 40 minutes) and culture-based tests (results in around 16 to 24 hours) in its assessment. It explained that this was a pragmatic choice based on the tests included in the scope for this assessment. It did not find any data on how the rapid tests affected antibiotic use. The committee concluded that the impact of tests on antibiotic use may be proportional to how quickly results can be returned, with quicker tests likely to have greater impact. It also concluded that, based on available data and expert opinion, the tests described as culture based by the EAG (taking around 16 to 24 hours for results) and tests that are unlikely to give same-day results are unlikely to be useful in primary or community care.

3.5 A clinical expert explained that dipstick tests give quick results during a GP appointment. They noted that even if a point-of-care test returned results in 40 minutes, that would be after the appointment was finished so that the patient may need to come back for another appointment or have a phone call to discuss it. This would mean extra staff time and patients waiting for results. Other clinical experts disagreed that a second appointment would be needed, pointing out that electronic prescribing and better ways to communicate with patients, for example, text messaging, are now in place. They noted the importance of having good systems and pathways in place to achieve this. A clinical expert also noted that tests may only be able to run 1 sample at a time which, depending on how many samples need to be run a day, could affect the time to get results. The committee concluded that how testing was implemented in practice, and local demand for testing, could influence how quickly results are available and how they are acted on. This, alongside how quickly a test can be run, is likely to affect how much the tests can affect antibiotic prescribing.

3.6 Only 3 studies with test accuracy data identified were considered at low risk of bias by the EAG. Of these, only 1 (Boon et al. 2022) reported accuracy data for UTI detection for rapid tests (Uriscreen and UTRiPLEX). This study assessed the test in children. Six studies compared point-of-care tests to dipstick testing done in the same population (including Boon et al. 2022). Three studies evaluating Uriscreen reported higher sensitivity but lower specificity than dipstick testing. One study evaluating UTRiPLEX reported it to be less sensitive but more specific than dipstick testing. The committee noted that there is also very limited data on how well rapid tests identify bacteria or do antibiotic susceptibility testing. The EAG did not find any accuracy data for the Astrego PA‑100 system and TriVerity rapid tests. The committee concluded that there is limited good quality data on test accuracy. Particularly, to identify bacteria, assess susceptibility to antibiotics (for tests that can do this), and directly compare tests with dipstick testing. The committee noted that the EAG had raised issues with reference standards that were used to produce accuracy estimates for the tests (this is described further in section 7.2 of the diagnostics assessment report). Issues with detecting slower growing bacteria by laboratory-based tests were also raised during consultation on the draft guidance. At consultation, a commentator also highlighted that a UTI may be caused by more than a single causative pathogen and it is important to consider causative slow growth pathogens. The committee agreed that there was considerable uncertainty about how well the tests will perform if used in the NHS compared with current testing.

3.7 Clinical experts emphasised the limitations of current testing. They said that dipsticks are unreliable in some populations (see section 2.2), which is important to consider when assessing new technologies. Populations in the identified studies were women with uncomplicated UTIs (3 studies), pregnant women (4 studies), people with a catheter (1 study) and children and babies (1 study in under 18 years, 1 study in under 16 years and 1 study in under 24 months). Five studies described having mixed populations but did not report any further detail. It was not clear if any studies assessed people with recurrent or chronic UTI. Clinical experts noted that urine samples from some populations, such as people who have a catheter or are aged over 65, are often polymicrobial. These groups are likely to have asymptomatic bacteriuria, which could lead to overdiagnosis. They noted that, as with dipstick testing, the accuracy of the newer point-of-care tests is likely to differ in these groups. Clinical experts also highlighted that differences in acute, recurrent and chronic infections mean tests may also perform differently in these groups. The committee concluded that accuracy data in the identified studies is not generalisable across the whole population of people with suspected UTIs (for example, people with suspected chronic UTI, people with catheters, men, and people aged over 65).

3.8 Some clinical experts said that just knowing the type of bacteria in a urine sample would not affect prescribing decisions. For example, UTIs can be caused by Escherichia coli (E. coli), which has a broad antibiotic susceptibility profile. But other clinical experts said that it might affect prescribing if someone has a recurrent or chronic UTI. This is because it could identify the bacteria as one against which an antibiotic was previously effective. The EAG said that clinical experts had also considered that there would be some benefit from knowing the identity of bacteria present for prescribing decisions. A clinical expert said that bacterial identification may suggest what antibiotics are unlikely to work, but not what will work. Another said that tests that can more accurately rule out a UTI than dipstick testing may be useful as an initial tool to triage people for further testing, or to consider alternative diagnoses, and allow safe reduction of antibiotic use. Particularly if results are given as quickly and if it has better performance in groups that dipsticks are not recommended for.

3.9 Only 1 rapid test does antibiotic susceptibility testing (Astrego PA‑100 analyser with the PA AST panel U‑0501). No further tests were identified that have regulatory approval or expect to get it in the next 12 months, for inclusion in this assessment. Clinical experts said that rising resistance to trimethoprim has led to increased use of nitrofurantoin as a first-line empirical antibiotic choice for UTIs. This has led to a drop in trimethoprim resistance. Quicker identification of people whose infection is still susceptible to trimethoprim may allow it to be used more and allow for reduced use of nitrofurantoin (although potentially causing trimethoprim resistance to rise again). A clinical expert also highlighted that NICE's guideline on antimicrobial prescribing for lower UTIs recommends using a narrow-spectrum antibiotic if possible. But they noted that an antibiotic susceptibility test may be needed to prescribe them if there is concern about resistance in the local population. During consultation, a stakeholder said that antibiotic susceptibility information may be of limited use for uncomplicated lower UTIs because some antibiotics concentrate in the bladder and achieve levels well above that applied in laboratory resistance tests. They said it may be more useful for people with upper or complicated UTIs. Another clinical expert also highlighted concern about antibiotic susceptibility testing without bacterial identification, saying that you may need to know which bacteria it is to decide which dose of antibiotic to use. However, this was questioned by a stakeholder and another expert, who said that they did not think this would be done in primary care without clear guidance. The committee concluded that rapid tests that can test for antibiotic susceptibility (Astrego PA‑100 analyser with the PA AST panel U‑0501) are likely to have the greatest potential to improve antibiotic prescribing decisions. There is greater uncertainty about how antibiotic prescribing would be affected by tests that can identify bacteria but do not test for antibiotic susceptibility (Lodestar DX). But a clinical expert said that these tests could more accurately indicate the presence of bacteria in a reasonable timeframe for prescribing. Tests that can only identify if bacteria are present but do so very rapidly (UTRiPLEX and Uriscreen) may improve prescribing if they are more accurate and can be used more widely than dipstick tests.

3.10 The EAG identified ongoing studies for 2 of the included tests:

3.11 Butler et al. (2018) reported that the cost of Flexicult Human was £48 per person. Few companies provided their test costs. Costs for the Astrego and Lodestar DX tests were provided as commercial in confidence, because costs for the UK have not yet been finalised, so cannot be reported here. Clinical experts said that dipstick tests currently cost about 50p each and laboratory-based culture tests cost about £10 to £20 per person. So, using the example of Flexicult Human, the tests included in this assessment may cost more than current testing. The committee noted that, although these tests could reduce costs incurred after testing and may improve patient outcomes, the greater the increase in cost of testing the less likely the tests are to be cost effective. It added that they may only be cost effective in some groups of people who get a particular benefit. Clinical experts also noted that GP practices may be reluctant to buy new point-of-care tests with much higher costs, particularly if the tests could be used for all people with suspected UTIs. A clinical expert said that 80% of antibiotics are prescribed for women with suspected uncomplicated UTI based on clinical symptoms alone, and improved testing may reduce the associated cost burden in the NHS. The committee said that more information on the costs of new tests is needed to consider any recommendation for use in the NHS.

3.12 The EAG's proposed modelling approach assessed how point-of-care tests affect empirical or targeted antibiotic use. It used test accuracy data and assumed that targeted antibiotics could always be used as the first antibiotic for rapid tests. It also assumed that tests that can identify the bacteria and tests that can assess antibiotic susceptibility affected prescribing in the same way. The committee recalled that how the test results affect prescribing decisions can be complex and may depend on local prescribing pathways (see sections 3.3 and 3.4). Also, the impact of tests that identify bacteria but do not identify antibiotic susceptibility was particularly uncertain (see section 3.8). The committee considered that more detail was needed on other ways that point-of-care tests could contribute to changes in antibiotic use, beyond empirical or targeted use. The committee concluded that there was too much uncertainty around the EAG's exploratory modelling to be able to accurately estimate how the test affects antibiotic prescribing. Direct data is needed. Clinical experts suggested that electronic prescribing meant that more detailed patient level data may be increasingly available, including age, sex and what antibiotics were prescribed. The committee also thought that future modelling work should give greater consideration to use of an individual patient simulation approach.

3.13 The EAG explored using a linked evidence approach to assess how test-guided treatment decisions affect health outcomes, such as the incidence of pyelonephritis, sepsis and kidney failure. Clinical experts questioned the inclusion of pyelonephritis and sepsis, noting that these complications overlap. They also said that kidney failure was not a common outcome for people with UTI. The committee noted the importance of considering how the tests affected health outcomes as well as antibiotic use. It concluded that direct data on how test use affected clinical outcomes would be useful but agreed that a linked evidence approach could also be used. The EAG identified several areas in which data may be needed for a linked evidence approach. These included the risk of complications for people with UTIs depending on the antibiotic they were treated with, or if they had no antibiotic treatment. But it highlighted that it had not done a systematic literature review, so it was possible that this data was available.

3.14 The committee questioned the approach proposed by the EAG for modelling suspected recurrent or chronic UTIs. It noted that the EAG's model did not distinguish between first and repeat UTIs, or between recurrent and chronic UTIs. Clinical experts highlighted issues with this approach and said that there are meaningful differences between a first UTI and later UTIs. For example, each time someone has a UTI, the risk of having another increases. Decisions about antibiotics are also different for recurrent or chronic UTIs, which can be influenced by which antibiotics have already been used. The committee recalled that chronic UTIs can have a substantial impact on people's health-related quality of life (see section 3.1) and said that this should be considered in future modelling work. The EAG did not specifically search for this data and so it was not clear if data on the health-related quality of life for people with chronic UTI exists. The committee concluded that future modelling approaches should distinguish between acute and recurrent or chronic UTI.

3.16 The committee recalled that there is ongoing research on some of the point-of-care tests (for example the TOUCAN study; see section 3.10) focusing mainly on test accuracy. The committee considered that data on how the tests affect antibiotic prescribing was needed to assess their potential impact in the NHS (see section 3.12). But better accuracy data may also help with assessing the benefits and harms of allowing early routine use in the NHS while further data is collected. The committee concluded that once the data from ongoing studies is available, it could be reviewed to decide if the NHS could be given early access alongside further data collection.

3.17 There was limited data for some groups of people with suspected UTI. The accuracy of tests and how they affect prescribing choices may vary in different populations and data is unlikely to be generalisable across groups (see section 3.7). The committee concluded that further research should be done in populations that fully represent people who the test could be used for in the NHS. The committee noted that ongoing studies (see section 3.10) are mostly in women aged over 18 with acute UTI. It encouraged further research in other populations. Clinical and patient experts also highlighted that it was important to assess the new tests in groups that have limited options in current standard care; for example, dipstick tests are not currently recommended for use in certain groups (such as people aged over 65 or who have a catheter). People with recurrent or chronic UTIs were highlighted as a group that may particularly benefit from improved testing. A comment received at consultation also highlighted that children may also benefit from improved testing. The committee concluded that it was important to evaluate the tests in groups for who current testing methods are known to be less accurate for. A clinical expert also highlighted that it may be useful to assess test performance in groups who currently have no testing, for example women aged over 18 with 2 or 3 symptoms of UTI.

3.18 The committee concluded that further research was needed before it can recommend early routine use of the tests in the NHS while further evidence is collected (see section 4). It also discussed what further data may be needed to fully assess the clinical and cost effectiveness of the technologies. It concluded that further data is needed on: