Tests to help assess risk of acute kidney injury for people being considered for critical care admission (ARCHITECT and Alinity i Urine NGAL assays, BioPorto NGAL test and NephroCheck test)

4.3 Most of the available data for the tests were sensitivity and specificity estimates. These measured the tests' ability to identify people who will be diagnosed with acute kidney injury using current clinical criteria (serum creatinine or urine output). The time of acute kidney injury diagnosis varied from within 12 hours to within 7 days. The external assessment group (EAG) commented that there was considerable clinical and statistical heterogeneity seen across the studies, which included very different populations, and therefore the results should be interpreted with caution. The committee also noted that even the best estimates of sensitivity and specificity showed that using the tests could result in large proportions of falsely positive or negative results. The stage of acute kidney injury detected by the tests also varied in the studies; from any stage of the condition to higher stages only. Clinical experts commented that the staging of the condition in classification systems (such as KDIGO) was developed by clinical consensus and there was uncertainty about the clinical significance of subclinical or early stage (stage 0 or stage 1) acute kidney injury and its correlation with clinical outcomes. The committee concluded that there was uncertainty about how well the tests could detect emerging acute kidney injury, and the clinical significance of what they detect in studies of test accuracy.

4.4 Clinical experts commented that there was considerable uncertainty about how much benefit the care bundles used in the NHS would provide to prevent, or reduce the severity of, acute kidney injury if used earlier (when NephroCheck or NGAL tests indicate risk of acute kidney injury; see section 4.2). In its model, the EAG used data from Meersch et al. (2017) for the effect of test-guided preventive care (a KDIGO care bundle) on reducing the chance of developing acute kidney injury or reducing the severity of the condition if it developed (see section 3.23). The committee noted that people in the control arm did not have the KDIGO care bundle. This was unlikely to reflect NHS practice because although using the tests could allow earlier use of the care bundle, everyone at risk would eventually have an acute kidney injury care bundle at a later time, once serum creatine or urine levels showed acute kidney injury. The absence of the KDIGO care bundle in the control group could therefore have overestimated the treatment effect from Meersch et al. compared with NHS practice. Using the treatment effect size from Gocze et al. rather than Meersch et al. improved the cost effectiveness of the tests (see section 3.35). Clinical experts commented that standard care in Germany (the control arms of the 3 identified studies on the effectiveness of the KDIGO care bundle) may differ from standard care in the UK. Therefore, the generalisability of the results of these studies to the NHS was potentially limited. The committee concluded that there was substantial uncertainty about how much effect a KDIGO care bundle had on developing, or reducing the severity of, acute kidney injury. It also concluded that it was uncertain whether a treatment effect size determined in studies done in Germany would be seen in the NHS, and therefore if the modelled effect of the KDIGO care bundle on acute kidney injury would be seen in the NHS.

4.5 No studies were identified that showed the effect of NGAL-guided use of the KDIGO care bundle. So in base case 1, the EAG assumed that the effect of NephroCheck and NGAL-guided preventive care on acute kidney injury incidence was the same. It used data from Meersch et al. (2017), a study done in people who had a positive NephroCheck test, to estimate the effect of test-guided preventive care on acute kidney injury incidence. Clinical experts commented that the biomarkers used in the NephroCheck test (tissue inhibitor of metalloproteinase 2 [TIMP‑2] and insulin-like growth factor binding protein 7 [IGFBP‑7]) may perform very differently to NGAL as indicators of acute kidney injury because they are released during different physiological processes. The committee concluded that it was not appropriate to assume that the results of the NephroCheck and NGAL tests were equivalent. It also concluded that data from Meersch et al. should not be used to inform estimates of how well NGAL-guided use of the KDIGO bundle affects acute kidney injury incidence in the economic model.

4.6 In its model, the EAG used observational data to link incidence and severity of acute kidney injury to the probability of clinical outcomes, such as length of stay in hospital, 90‑day mortality and need for RRT. However, the committee noted that in Meersch et al. use of the KDIGO bundle reduced acute kidney injury incidence, but not length of stay in hospital or intensive care, need for RRT in hospital or 90‑day all-cause mortality. In Gocze et al. length of hospital and intensive care stay was significantly shorter in the KDIGO bundle study arm, but there was no significant difference in need for RRT or mortality in hospital. Clinical experts explained that how each stage of acute kidney injury affected shorter- and longer-term clinical outcomes was not clearly understood (see section 4.3). The EAG investigated how much varying the effect of having acute kidney injury, and severity, had on clinical outcomes in scenario analyses. This led to large variation in cost effectiveness (see section 4.9). The committee concluded that it was uncertain how much the incidence and severity of acute kidney injury affected clinical outcomes.

4.7 The EAG did not include the cost of analysers needed to run the NGAL assays in its estimates of cost per NGAL test. This was because it assumed that the NGAL tests are run on platforms already available in hospital laboratories, so the cost of these analysers was negligible. Clinical experts commented that the analysers needed to run the different NGAL assays would not be in every hospital. The committee concluded that it would have been reasonable to include the cost of analysers needed to run the NGAL assays in the cost per test used in the model, as had been done for the NephroCheck test.

4.8 The committee discussed the lack of data available for children. It noted that the EAG did a scenario analysis that used accuracy estimates from studies that enrolled children only (scenario Q; see section 3.33). Because of a lack of data for other parameters, the EAG had to use values derived from adult populations. The EAG cautioned that this analysis should be considered as exploratory. Also, clinical experts commented that the potential use for children could be very different to that for adults in the NHS. The committee concluded that, because of a lack of data to inform model parameters and uncertainty about the intended use of the tests, the cost-effectiveness estimates of the tests for children were highly uncertain. The committee considered that future studies should consider the utility of the tests for children (see section 4.11).

4.9 The EAG did multiple scenario analyses to reflect the uncertainty about which clinical outcomes would be affected by both the incidence and severity of acute kidney injury. It cautioned that the results of the cost-effectiveness modelling were largely speculative and should be interpreted with caution. Also, it considered it impossible to determine the best incremental cost-effectiveness ratio (ICER) given the available evidence. Incremental quality-adjusted life years (QALYs) were very low across the scenarios, with tests often having ICERs over £50,000 per QALY gained, or being dominated (that is, they had higher costs and lower QALYs) compared with standard monitoring. Varying the parameter values used in scenario analyses substantially affected the cost-effectiveness estimates for the tests. Changes to some parameters improved the cost effectiveness of the tests, to the extent that they dominated standard care (in base case 1) when compared in a pairwise manner (see section 3.30 and section 3.33). The committee further recalled that it did not consider it appropriate to use data from NephroCheck-guided use of the KDIGO care bundle to estimate the effect of NGAL-guided use of the KDIGO care bundle (see section 4.5). The committee concluded that there was substantial uncertainty about the best cost-effectiveness estimates for the tests in the defined clinical population. However, the estimates could potentially be much higher than what NICE normally considers cost effective.

4.10 The committee agreed that there was substantial uncertainty about how the tests could be used in the NHS (see section 4.2) and their likely cost effectiveness. This was mainly because there was uncertainty about the effect that test-guided care could have on the incidence and progression of acute kidney injury (see section 4.4) and on other clinical outcomes (see section 4.6) in the defined NHS clinical population. Also, how clinicians would react to the test results in the NHS was unclear (that is, the changes to care they would make in response to a positive or negative result). The cost-effectiveness estimates for the tests were very uncertain and, in most scenarios, much higher than what NICE normally considers a cost-effective use of NHS resources (see section 4.9). The committee concluded that there was too much uncertainty about the cost effectiveness of the tests to recommend their adoption in the NHS. Further research could provide clarity on how the tests would affect care and outcomes in the NHS and allow their cost effectiveness to be estimated.

4.11 The committee recalled that there was uncertainty about which patient populations in the NHS could benefit from test-guided use of preventive care for acute kidney injury (see section 4.2). If care bundles were already being used, in full or in part, in a patient population this would limit the effect that the test results can have on guiding care. Clinical experts commented that the potential use for children and young people can also be very different to adults, so specific consideration is needed for this group. The costs of the NephroCheck (about £90) and of providing the KDIGO care bundle earlier (about £105) were similar. Therefore, the committee questioned whether providing the care bundle earlier to everyone (that is, without testing) could be the most cost-effective strategy for some patient populations in the NHS. The committee concluded that, before further studies are done, it was important that companies define the patient populations in the NHS who could benefit from test-guided preventive care. It noted that people who are critically unwell in the NHS would likely already be having all available care to prevent acute kidney injury.