To the Editors:
Tenofovir disoproxil fumarate (TDF) is a key component of World Health Organization recommended first-line antiretroviral therapy (ART) and preexposure prophylaxis (PrEP),1 with 20 million people predicted to use TDF by 2020.2 TDF is only recommended for patients with normal renal function, meaning serum creatinine and estimated glomerular filtration rate (eGFR) should be assessed before initiation, and during follow-up.1 However, limited access to laboratory testing in low- and middle-income countries can restrict compliance with these guidelines, delay receipt of results, and impede ART uptake. Point-of-care (POC) creatinine devices could overcome these barriers and enable “same-day” ART/PrEP initiation and subsequent “same-day” monitoring. This could be particularly useful in home and community-based programs that enhance test and treat efforts in high-prevalence settings.3,4
The Statsensor Xpress-i (Nova Biomedical, Waltham, MA) is a battery powered, hand-held POC device that measures creatinine concentration in 30 seconds from capillary, venous or arterial whole blood. Although evaluations of the analytical performance of this assay have shown mixed results,5–7 several ART trials in Southern Africa are currently using the assay to screen for renal impairment.3,8,9 However, no clinical evaluations have yet been published in this setting. Here, we evaluate the implementation of the Statsensor Xpress-i for renal function assessment among HIV-positive patients receiving first-line ART in a primary care setting in South Africa.
We conducted a prospective, clinic-based evaluation of the POC Statsensor Xpress-i compared with a laboratory-based reference method, as part of the STREAM (Simplifying HIV TREAtment and Monitoring) study. STREAM is a randomized controlled implementation trial of POC viral load testing and task shifting being conducted in Durban, South Africa. Consenting participants were randomized in a 1:1 ratio to receive either POC monitoring (viral load, CD4, and creatinine) versus standard laboratory monitoring only.9 Eligible participants were clinically stable, nonpregnant, HIV-positive adults who initiated first-line ART within the past 6 months. One patient with an eGFR of <30 mL/min in the 3 months before screening was deemed clinically unstable and excluded from the trial. The primary outcome is virological suppression and retention in care after 1 year.
In this substudy, we only include participants allocated to POC monitoring, as they had both finger-prick capillary blood sampling and venepuncture performed by a study nurse at enrollment. Finger-prick capillary whole blood creatinine was measured immediately using the factory calibrated Statsensor Xpress-i. Venous plasma samples were transported to the central laboratory, where creatinine was measured using the Dimension EXL 200 (Siemens Healthcare, Erlangen, Germany) isotope dilution mass spectrometry aligned assay. Before use, each new lot of Statsensor strips was tested using manufacturer supplied linearity controls, with all results within expected ranges. The vendor provided on-site training before study commencement, with internal quality assurance reviews performed thereafter.
POC Statsensor Xpress-i and laboratory serum creatinine were compared using Spearman correlation and a Bland–Altman plot. eGFR was calculated from age, sex, and creatinine values using the Modified Diet in Renal Disease equation (without ethnicity factor), which has been validated and is widely used in our setting.10 We calculated sensitivity, specificity, positive predictive value, and negative predictive value of the Statsensor Xpress-i to detect an eGFR below internationally recognized thresholds, which define <60 mL/min as moderate to severe impairment, and 60–89 mL/min as potential mild renal impairment (in the presence of other indicators of kidney disease).10 The Statsensor Xpress-i does not have functionality to provide eGFR results and calculating eGFR by hand can be complex for primary care providers. Therefore, we also used a previously validated creatinine cutoff of 106 µmol/L (1.2 mg/dL), which approximates an eGFR of 60 mL/min depending on age and sex.7
This study was approved by the University of KwaZulu-Natal Biomedical Research Ethics Committee and the University of Washington Institutional Review Board.
Between February and August 2017, a total of 390 participants were enrolled into STREAM. One hundred ninety-five were randomized to laboratory monitoring alone and were excluded from this substudy. One hundred ninety-five were randomized to receive POC monitoring, with laboratory reference samples drawn for comparison. Eight reference samples were excluded because they were either not processed during a laboratory strike, or processed more than 48 hours after sampling, leaving 187 participants with paired creatinine samples for inclusion in this analysis. Median age was 31 years [interquartile range 27–38], 61.5% (115/187) were female and median CD4 count was 459 cells/mm3 (interquartile range 289–658).
Correlation between the Statsensor Xpress-i and laboratory serum creatinine was modest (Spearman ρ = 0.58, Fig. 1A). Mean POC creatinine was 79.4 µmol/L (range 50–128) compared to 69.0 µmol/L (range 28–116) with the laboratory test [bias +10.4 µmol/L, 95% limits of agreement ±27.9 µmol/L (−17.6 to +38.3 µmol/L) Fig. 1B]. Results were unchanged in a sensitivity analysis restricted to the 167 patients who had laboratory reference samples performed on the same day as the POC test. The Statsensor Xpress-i had 87.1% sensitivity [95% confidence interval (CI) 76.2 to 94.3] and 52.0% specificity (95% CI: 42.9 to 61.0) to detect an eGFR of <90 mL/min, and correctly identified the 1 patient with a laboratory eGFR of <60 mL/min. Using an abnormal creatinine cutoff of >106 µmol/L, the Statsensor Xpress-i had sensitivity of 100%, specificity of 95.1% (95% CI: 90.9 to 97.7), negative predictive value of 100%, and positive predictive value of 30.8% (95% CI: 9.1 to 61.4).
Nurses reported that the analyzer was easy to use, and required minimal training because of similarity with other widely used POC tests (eg, finger-prick glucose). However, a separate eGFR calculator was needed to facilitate eGFR assessment. Cost per POC test in our study, including analyzer cost, staff time, controls, and consumables, was estimated at 10 USD compared with approximately 4.5 USD per laboratory test.
The Statsensor Xpress-i demonstrated modest agreement with the laboratory reference assay and generally overestimated creatinine levels, with limits of agreement just outside the Clinical Laboratory Improvement Amendments (CLIA) acceptable test performance criteria of ±27 µmol/L.5 However, this POC assay detected the few patients with potential renal impairment, suggesting that it could be used to rapidly screen for renal impairment in HIV care programs.
Previous evaluations have indicated that the Statsensor Xpress-i is negatively biased at higher creatinine levels but positively biased at lower levels. Two studies from the Netherlands of 60 and 133 patients, with higher laboratory creatinine values than our study (up to 1000 µmol/L), demonstrated a negative bias of −10.7 and −12.38 µmol/L, and correlation coefficients of 0.97 and 0.95, respectively.5,6 However, when restricted to patients with normal creatinine levels, there was a small positive bias of 3 µmol/L, and a correlation coefficient of 0.69, which was more similar to our results.5 Overall, sensitivity and specificity to detect a creatinine >115 µmol/L was 100% and 91%, respectively.5 In an evaluation from Nicaragua of 100 patients with creatinine levels between 44 and 320 µmol/L, the Statsensor Xpress-i was positively biased with a median creatinine of 92 µmol/L compared with 64 µmol/L with the laboratory method. As in our analysis, the POC assay had a high sensitivity (100%) and acceptable specificity (84%) to detect a creatinine >106 µmol/L.7
Strengths of our analysis include the relatively large sample size and the novel implementation of the Statsensor Xpress-i among HIV-positive patients. Patients in our analysis had generally lower creatinine values than in the above studies, which corresponds with low rates of renal disease seen in other HIV-positive cohorts on TDF in Southern Africa.10 This limits the interpretation of our results. First, the positive bias seen in our study likely reflects an inherent positive bias in the assay at these lower creatinine levels. Second, the modest correlation seen in this analysis may not reflect performance of the assay over a higher range of creatinine levels, as demonstrated in previous studies.5,6 Third, our analysis is limited by having few participants with abnormally high creatinine, or eGFR <60 mL/min. However, the Statsensor Xpress-i did identify all these correctly.
Our findings have implications for HIV programs in other low- and middle-income countries, where the young age of HIV-positive cohorts means prevalence of renal impairment is likely to be similarly low.10 Overall, the positive bias, wide limits of agreement, and modest Spearman correlation mean that the Statsensor Xpress-i should not be used to determine exact creatinine levels, or monitor trends, within the normal range. Instead, as demonstrated in previous studies, the acceptable sensitivity and specificity to detect renal disease suggests that this assay could be used to rapidly rule out renal impairment, allowing patients with normal POC results to be safely initiated or continued on TDF during the same clinical visit. Those with abnormal POC results (7% in our study) will require a confirmatory laboratory creatinine to prevent misdiagnosis of renal impairment and unnecessary withholding of TDF. Nurses and lay health care workers could use this assay in clinics and community settings, although a separate eGFR calculator would be needed. In settings such as South Africa, where creatinine monitoring for TDF is mandatory, the relatively high cost per POC test may be offset by costs saved through earlier treatment initiation and reductions in clinical visits to review test results. Although there is ongoing debate about the value of universal creatinine monitoring for TDF,10 this POC assay could still be useful to screen for renal impairment in patients at higher risk of kidney disease.
In conclusion, among HIV-positive, primary care patients with generally normal renal function, the Statsensor Xpress-i displayed modest agreement with laboratory references, but accurately detected the few patients with high creatinine or renal insufficiency. This assay has potential to facilitate rapid screening for renal impairment within HIV programs, which is particularly important given growing interest in “same-day” ART initiation and community-based ART delivery.3,4 However, further evaluations are warranted, including cost-effectiveness studies, particularly among populations with higher levels of renal impairment and those considered for PrEP initiation.
The authors thank all participants in the study and acknowledge the work and support of staff at CAPRISA and Ethekwini Municipality, in particular Dr. Yukteshwar Sookrajh and Dr. Ayo Olowolagba.
1. World Health Organisation. Consolidated Guidelines on the Use of Antiretroviral Drugs for Treating and Preventing HIV Infection: Recommendations for a Public Health Approach. 2nd ed. Geneva, Switzerland: World Health Organization; 2016.
2. World Health Organization. Combined Global Demand Forecasts for Antiretroviral Medicines and HIV Diagnostics in Low- and Middle-Income Countries From 2015 to 2020. Geneva, Switzerland: World Health Organization; 2016.
3. Labhardt ND, Ringera I, Lejone TI, et al. Same day ART initiation versus clinic-based pre-ART assessment and counselling for individuals newly tested HIV-positive during community-based HIV testing in rural Lesotho—a randomized controlled trial (CASCADE trial). BMC Public Health. 2016;16:329.
4. Grimsrud A, Barnabas RV, Ehrenkranz P, et al. Evidence for scale up: the differentiated care research agenda. J Int AIDS Soc. 2017;20(suppl 4):22024.
5. Kosack CS, de Kieviet W, Bayrak K, et al. Evaluation of the Nova StatSensor® XpressTM creatinine point-of-care Handheld Analyzer. PLoS One. 2015;10:e0122433.
6. Van Lint CL, Van Der Boog PJ, Romijn FP, et al. Application of a point of care creatinine device for trend monitoring in kidney transplant patients: fit for purpose? Clin Chem Lab Med. 2015;53:1547–1556.
7. Minnings K, Kerns E, Fiore M, et al. Chronic kidney disease prevalence in Rivas, Nicaragua: use of a field device for creatinine measurement. Clin Biochem. 2015;48:456–458.
8. Barnabas R. Delivery optimization for antiretroviral therapy (the DO ART study). 2017. Available at: https://clinicaltrials.gov
/ct2/show/NCT02929992. Accessed October 31, 2017.
9. Dorward J, Garrett N, Quame-Amaglo J, et al. Protocol for a randomized controlled implementation trial of point-of-care viral load testing and task shifting: the simplifying HIV TREAtment and monitoring (STREAM) study. BMJ Open. 2017;7:e0175071–8.
10. de Waal R, Cohen K, Fox MP, et al. Changes in estimated glomerular filtration rate over time in South African HIV-1-infected patients receiving tenofovir: a retrospective cohort study. J Int AIDS Soc. 2017;20:21317.