Point-of-care NT-proBNP monitoring for heart failure: observational feasibility study in primary care

Summary

This study found that it was feasible to do routine POC NT-proBNP testing in primary care, with testing successfully carried out at every planned study visit. GPs used the results of the POC tests to inform changes to more than a third of treatment plans, and patients found POC testing acceptable. Between-person variability in POC NT-proBNP was around twice as large as within-person variability over a 12-month period.

Comparison with existing literature

The CV for within-patient POC readings over the course of the year was relatively low at 37.6%, and lowest at reduced concentrations where a diagnosis of HF is generally least obvious.¹⁵ Since each patient’s readings were taken 6 months apart, this CV reflects instrument variation as well as biological variation over the year-long study.

Comparison to laboratory NT-proBNP results showed good agreement, with the POC device showing slight positive bias, as has been previously reported by the Scandinavian Evaluation of Laboratory Equipment for Primary Health Care, and Hex et al.^16,17 Although simple bias calculations revealed a significant positive bias, non-significant results from the Passing–Bablok regression and linear regression on the Bland–Altman plot suggest that this bias is only borderline significant and is acceptably small for a primary care setting. Moreover, the magnitude of difference between the POC and laboratory devices is similar to those seen between the various laboratory assays for BNP and NT-proBNP produced by different manufacturers,^18,19 as immunoassays vary with respect to antibody specificity²⁰ and the population characteristics of those assessed.²¹ Perhaps more important are the few recorded cases where there has been significant variation from visit to visit, and lack of concordance between POC and laboratory readings (Figures 2 3), suggesting operator error rather than measurement bias. If POC testing is to be incorporated into routine primary care, it is imperative that operators are trained in its usage, and checks are developed to prevent and identify irregular results.

¹⁵

Strengths and limitations

This study had some limitations. Firstly, the sample size was limited due to a low response rate from invited patients. Despite this, the study was able to draw statistically significant conclusions and clearly determine both that a larger cohort study of POC NT-proBNP testing is feasible, and that it is possible to routinely carry out planned POC NT-proBNP testing in primary care. Another limitation was the accidental loss of two-thirds of planned laboratory NT-proBNP tests due to a laboratory error. This has limited the robustness of the data comparing POC and laboratory results. Nonetheless, a comparative analysis was still possible using the remaining data, and results were comparable to previous research.¹⁷ Finally, POC testing only occurred at planned appointments during the study, and not at any unplanned appointment. This likely reflects current pressures in UK primary care. If unscheduled NT-proBNP testing is to be successfully used, other factors such as sufficient time, training, and staff available to carry out testing will need to be considered. Nonetheless, this study's data indicate that future research into the importance of operator skill, lack of trust in POC devices, and short-term variability of measurements, is all feasible. As this was a feasibility study, the authors did not follow patients up to identify hospital admission following POC testing, but future research could investigate the ability of primary care POC testing in HF to avoid hospital admissions or readmission.

The unique strength of this work was its pragmatic design. Only one or two nurses per practice were trained on the new device, the population was typical of those who would benefit from close monitoring in primary care, and the device used is available to GPs in the UK today.

Implications for practice

More than a third of patient treatment plans were altered because of the POC NT-proBNP readings, suggesting that clinicians found the added information useful. Moreover, in all cases of altered treatment, doctors noted that their decision would have been the same regardless of their access to confirmatory laboratory findings. Adlbrecht et al found that the Roche Cobas h 232 POC NT-proBNP measurement device had a 100% negative predictive value for HF-related hospitalisation, and better positive predictive value than any clinical sign or symptom when NT-proBNP was in the low range of 100–500pg/mL.¹⁵ In precisely these low concentrations, this study has shown that POC testing has the greatest precision. Therefore, for patients on the borderline of current diagnostic cut-offs who may never be referred to a hospital for BNP or NT-proBNP testing may stand to benefit greatly from POC testing in primary care. Moreover, POC testing is cheaper and more accessible in terms of instrument costs, time, and personnel, making implementation in primary care cheaper than laboratory testing. On the other hand, in 12% of the decisions not to change treatment, doctors said their decision may have been different if they had access to laboratory data instead of the POC device, suggesting a potential lack of trust in the POC device.

In this setting, NT-proBNP monitoring was found to be acceptable to patients, feasible for implementation in UK general practice, and sufficiently accurate, especially for the patients on the border of clinical cut-offs who may stand the most to gain. Most importantly, GPs were willing to use this new technology to improve the care of more than one-third of patients with HF in primary care.