Can community healthcare practitioners believe point-of-care tests for haemoglobin and renal function? Agreement study of point-of-care tests and subsequent central laboratory tests for acute frailty syndromes in older adults.
The problem
Rapid identification of underlying disease is important in the community management of acute frailty syndromes. The NHS England Five Year Plan, Future Hospital Commission (Royal College of Physicians) and the Silver Book (British Geriatrics Society) all call for more acute care out of hospital, supported by rapid diagnostics. Therefore the accuracy of near patient tests for common pathologies in the acutely ill older adult is crucial for development of contemporary acute care practice in the community. We set out to determine the agreement between near patient tests for renal function and haemoglobin compared with subsequent central laboratory values in older patients presenting with frailty syndromes to a community based unit providing out of hospital care.
The approach
We analysed samples from patients who were referred by GPs or paramedics for acute ambulatory care for a frailty syndrome to a community unit staffed by interface GPs and hospital clinicians. The Abbott iSTAT point of care platform was used for clinical chemistry and haemoglobin testing in the community and the same blood draw was then used to fill lithium/heparin and EDTA blood bottles for the same analyses in a central hospital laboratory using conventional transport and storage. Agreement was analysed with Bland Altman plots and correlation with Passing-Bablok linear regression using STATA and MedCalc.
Findings
In 123 older adults (median age 81 years, 64% female) with acute frailty syndromes the agreement between the iSTAT creatinine, urea and haemoglobin and the paired measurements on a central laboratory analyser demonstrated a mean difference of 16.4 µmol/l (95% limits of agreement (LOA) -2.7 to 35.5) for creatinine, 0.7 g/dl (95% LOA -0.5 to 1.8) for haemoglobin, and -0.5 g/dl (95% LOA -2.8 to 1.8) for urea. A slope of 0.99 (95%CI 0.96-1.02) for creatinine (y=0.99 x + 16.9), 1 (95%CI 0.96-1.05) for haemoglobin (y=1x + 0.7) and 0.99 (95%CI 0.96-1.00) for urea (y=0.99x - 0.38) were found. The overestimation by the iSTAT haemoglobin resulted in a sensitivity of 68% (95% CI 48.6-83.3%) for detecting lab haemoglobin values below 11g/dl however with a positive predictive value of 95.5% for iSTAT values below 11g/dl and overall area under the ROC curve of 0.97. Adjusting the iSTAT values by -1.8 to increase sensitivity to 100% (95% CI 88.8-100%) decreased the specificity to 75% (95% CI 64.9-83.4%).
Consequences
The community-based near patient renal function tests demonstrate high agreement and precise measurements in older patients with frailty syndromes. For haemoglobin, an overestimation by the point-of-care test was found in this sample but a high c statistic for detecting laboratory confirmed anaemia. Bias and precision studies of near patient tests can support acute care pathways and inform clinical management decisions for acute ambulatory care.
Credits
- Jan Verbakel, Institute of Liver Studies, King's College Hospital, London, UK
- Charlotte Richardson, Department of Clinical Biochemistry, Oxford University Hospitals NHS Trust, Oxford, UK
- Brian Shine, NIHR Oxford Diagnostics Evidence Cooperative, Oxford, UK
- Ann Van den Bruel, Institute of Liver Studies, King's College Hospital, London, UK
- Sarah Pendlebury, NIHR Oxford Biomedical Research Centre, Oxford, UK
- Daniel Lasserson, Institute of Liver Studies, King's College Hospital, London, UK