In collaboration with the Ministry of Health (MOH) in Jordan, Wondfo Biotech (Guangzhou, China) conducted two validation studies on Finecare™ HbA1c Rapid Quantitative Test; one was performed in a private referral laboratory (n = 100 samples) and the other was performed in a public health laboratory that belongs to the MOH (n = 47 samples), and the other was performed in a private referral laboratory (n = 100 samples). HbA1c was measured from collected fingerstick and matched venous blood samples for a total of 147 participants from both laboratories. Testing results were provided to our lab for analysis, and that data was unaccompanied by any patient identifications or private information other than the primary demographic data, including age and gender. Accordingly, an Ethical approval exemption (QU-IRB 1766-E/22) was granted by Qatar University.

Finecare™ HbA1c Rapid Quantitative Test is based on fluorescence immunoassay technology and measures the level of HbA1c in human blood using a sandwich immunodetection approach. According to the manufacturer’s test leaflets and flyer, the Fincare™ HbA1c POC test, according to the manufacturer’s test leaflets and flyer, is traceable to the International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) reference method for measuring HbA1c and is certified by the National Glycohemoglobin Standardization Program (NGSP) as having documented traceability to the Diabetes Control and Complications Trial (DCCT) reference method [18, 19]. The NGSP awards certification to manufacturers for successfully meeting specific performance criteria [20]. The test was carried out according to the manufacturer instructions. The LFA reaction time is 5 min, and the measuring range is 4.0–14.5%

The Tina-quant Hemoglobin A1cDx assay is intended to diagnose diabetic patients. It is in vitro diagnostics assay to quantify hemoglobin A1c (mmol/mol) and % hemoglobin A1c in whole venous blood on the cobas pro c503 clinical chemistry analyzers. This approach is based on the turbidimetric inhibition immunoassay of blood samples that have been hemolyzed. The anti-HbA1c antibody forms a soluble complex with a single binding site on HbA1c. Polyhaptens react with excess anti-HbA1c antibodies to generate an insoluble compound, which is evaluated by turbidimetry. The measuring range is 4.0–14.5%.

Finecare™ and the reference technique, Cobas Pro c503, were compared using correlation and linear regression analysis. Because our data was not normally distributed, we estimated the spearman correlation coefficient (r), with r values of 0–0.39 indicating a weak correlation, 0.40–0.59 indicating a moderate connection, 0.6–0.77 indicating a high correlation, and 0.8–1 indicating a very strong correlation [21]. In addition, we assessed the area under the curve (AUC) of the Receiver-Operating Characteristic (ROC) curve, which measures the accuracy of a quantitative diagnostic test [22]. An AUC of 0.9–1.0 is denoted as excellent, 0.8–0.9 is denoted as very good, 0.7–0.8 is denoted as good, 0.6–0.7 is denoted as sufficient, 0.5–0.6 is denoted as bad, and 0.5 is denoted as not useful. Moreover, we generated Bland-Altman plot, which is based on the quantification of the agreement between two quantitative measurements by studying the mean difference and constructing limits of agreement (LOA). Further, concordance analysis between Finecare™ and the reference methods Cobas Pro c503 was conducted, which includes the overall percent agreement (OPA), positive percent agreement (PPA), and negative percent agreement (NPA), accuracy, and Cohen’s Kappa test. A Cohen’s Kappa coefficient 0.40 suggests a poor agreement, 0.40–0.59 suggests a fair agreement, 0.60–0.74 suggests a good agreement and $\ge$0.75 suggests an excellent agreement [23]. All statistical tests were conducted using GraphPad Prism (GraphPad Software, Inc. Version 9, San Diego, CA, USA).

We assessed the correlation between Finecare™ with the reference method, Roche Cobas Pro c503, using fingerstick and venous blood samples, as shown in Fig. 1. The correlation between Finecare™ employing fingerstick whole blood sample and Cobas Pro c503 was very high (r = 0.93, p 0.0001). Likewise, Cobas Pro c503 and Finecare™ utilizing venous whole blood samples showed a very significant correlation (r = 0.97, p 0.0001). The results of the linear regression study demonstrated that the built model could accurately predict the dependent variable between Finecare™ utilizing venous or fingerstick samples (R${}^2$ $>$ 0.95, p 0.0001).

Fig. 1.

Pairwise correlation and linear regression analysis for each assay. (A) Correlation plot of Finecare${}^{\text{TM}}$ with Cobas Pro c503 using fingerstick blood sample; (B) Correlation plot of Finecare${}^{\text{TM}}$ with Cobas Pro c503 using venous whole blood sample; and (C) Correlation plot of Fingerstick blood sample with venous whole blood sample using Finecare${}^{\text{TM}}$ machine. Spearman correlation coefficient (r) and p-value were indicated. The coefficient of determination (R${}^2$) was calculated to be 0.95, 0.97, 0.94 for (A–C), respectively. n = 100.

ROC curve analyses showed excellent performance for Finecare™ with an AUC of 0.994 and 0.998 using fingerstick and venous blood samples, respectively (Fig. 2).

Fig. 2.

ROC curve for Finecare™. Finecare™ showed excellent performance (AUC $>$0.9) compared to the reference method using fingerstick (Fig. 2A) and whole blood (Fig. 2B) samples. AUC of 0.9–1.0 is considered excellent. AUC, the area under the curve; ROC, Receiver Operating Characteristic.

An alternative method for comparing Finecare™ and Roche Cobas Pro c503 using fingerstick and venous blood data and results agreement is by performing Bland-Altman plot analysis. Bland-Altman analysis aids in evaluating the agreement between two quantitative tests by graphically depicting the measurement variances by plotting the difference against the mean of the data [24, 25]. Bland-Altman method computes the average difference between two methods of measurement and standard deviation (SD) of the difference, and calculates the ‘95% limit of agreement’ (LOA) as the mean difference. The presentation of the ‘95% LOA’ on the Bland-Altman plot permits a visual decision of how well two measurement methods are in concordance with each other. A smaller range between the LOA is commented as better compliance.

As shown in Fig. 3A,B, Finecare™ fingerstick and venous blood data showed excellent agreement and compliance with Roche Cobas Pro c503 data as the mean bias are 0.05 (LOA: –0.58–0.68) and 0.003 (LOA: –0.49–0.5), respectively, which are very close to 0. Interestingly, a very small mean bias (0.047) was also observed between the fingerstick and the venepuncture data, indicating that the type of sample does not significantly affect the results, and that the assay has high reproducibility.

Fig. 3.

The bland-Altman plot. Dotted lines present mean bias and agreement limits. The bias is represented by the gap between the X axis and the mean bias dotted line. The mean bias was calculated to be 0.05, 0.003, 0.047 for (A–C), respectively. LOA, Limit of Agreement. n = 100.

We evaluated the performance of Finecare™ using fingerstick and venous blood samples compared to Cobas Pro c503. As depicted in Fig. 4, there was no significant difference between the results obtained from Finecare™ and CobasPro c503 analytical analyser using the fingerstick or venous blood samples.

Fig. 4.

Distribution of numerical results obtained from fingerstick whole blood and venous blood using Finecare™ machine and the reference method; Cobas Pro c503. The Nonparametric Kruskal-Wallis test was used to compare the differences between groups. ns, non-significant. n = 100.

According to the current recommendations, patients with an HbA1c level $\ge$6.5% would get intensive treatment. To assess the clinical applicability of Finecare™, the HbA1c results were compared with those of the reference method using a cut-off point of 6.5% for HbA1c using fingerstick and venous blood samples, as shown in Table 1A. Using fingerstick whole blood, 4 samples showed false positive results (4%), and two samples (2%) showed false negative results using Finecare™, as shown in Table 1A. Whereas using venous blood samples, only one sample showed a false positive result (1%), and none showed false negative results (0%) using Finecare™, as shown in Table 1A. Similarly, as shown in Table 1B, only 3 fingerstick whole blood samples showed false positive results (3%) when compared to venous blood. In contrast, only 2 fingerstick whole blood samples showed false negative results (2%) when compared to venous blood samples.

As shown in Table 2, Finecare™ demonstrated 100% (95% CI: 86.3–100) sensitivity and 98.7% (95% CI: 92.8–100) specificity compared to the reference method using venous blood samples. In addition, Finecare™ showed lower sensitivity of 92.0% (95% CI: 74.0–99.0) and a specificity of 94.7% (95% CI: 86.9–98.5) compared to the reference method using fingerstick whole blood. Moreover, we evaluated the sensitivity and specificity of Finecare™ using fingerstick compared to venous blood samples. As expected, using fingerstick whole blood showed a lower sensitivity of 92.3% (95% CI: 74.9–99.1) and a specificity of 95.9% (95% CI: 88.6–99.2) compared to venous blood.

The concordance assessment between the reference method Cobas Pro c503 and Finecare™ using venous and fingerstick whole blood samples is reported in Table 2. The OPA, PPV, and NPV between Cobas Pro c503 and Finecare™ using venous blood were 99.0% (95% CI: 94.6–100), 96.2% (95% CI: 78.1–99.4), and 100% (95% CI: 95.1–100), respectively. Whereas, using fingerstick whole blood, the OPA, PPV, and NPV were 94.0% (95% CI: 87.4–97.8), 85.2% (95% CI: 68.8–93.8), and 97.3% (95% CI: 90.4–99.3), respectively. Most importantly, Cohen’s Kappa coefficient denoted excellent agreement between Finecare™ and Cobas Pro c503 ($\kappa$ $>$ 0.84) using fingerstick and venous whole blood samples.

We classified the participants into three groups according to the American Diabetes Association (ADA): no diabetes (HbA1c 5.7%), pre-diabetes (HbA1c 5.7–6.5%), diabetes (HbA1c $\ge$6.5%). Similar to the reference method, Finecare™ showed a significant difference between the normal, pre-diabetic, and diabetic samples (p 0.001), as shown in Fig. 5.

Fig. 5.

Distribution of numerical results. (A) fingerstick whole blood using Finecare${}^{\text{TM}}$ machine; (B) venous blood using Finecare${}^{\text{TM}}$ machine; and (C) venous blood using the reference method; Cobas Pro c503. Results are represented as dot plots. Data are presented for 100 patients (normal 5.7%, prediabetic 5.7–6.4%, diabetic $\ge$6.5%) from each assay. Nonparametric Kruskal-Wallis test was used to compare the differences between groups. ** p 0.01, p 0.0001.

To ensure the reproducibility of the results, the performance of Finecare™ was evaluated in another laboratory. Finecare™ showed 80% sensitivity and 100% specificity compared to the reference method (Table 3). Nevertheless, similar to our results, a very strong correlation was observed (r = 0.97, p 0.0001) between Finecare™ and Cobas Pro c503 as shown in Fig. 6.

Fig. 6.

Pairwise correlation and linear regression analysis of Finecare™ in comparison to Cobas Pro c503. Spearman correlation coefficient (r) and p-value were indicated. The coefficient of determination (R${}^2$) was calculated to be 0.95. n = 47.