This article has Open Peer Review reports available.
A meta-analysis of the performance of the PimaTM CD4 for point of care testing
© Scott et al. 2015
Received: 10 March 2015
Accepted: 11 June 2015
Published: 25 July 2015
The Alere point-of-care (POC) Pima™ CD4 analyzer allows for decentralized testing and expansion to testing antiretroviral therapy (ART) eligibility. A consortium conducted a pooled multi-data technical performance analysis of the Pima CD4.
Primary data (11,803 paired observations) comprised 22 independent studies between 2009–2012 from the Caribbean, Asia, Sub-Saharan Africa, USA and Europe, using 6 laboratory-based reference technologies. Data were analyzed as categorical (including binary) and numerical (absolute) observations using a bivariate and/or univariate random effects model when appropriate.
At a median reference CD4 of 383 cells/μl the mean Pima CD4 bias is -23 cells/μl (average bias across all CD4 ranges is 10 % for venous and 15 % for capillary testing). Sensitivity of the Pima CD4 is 93 % (95 % confidence interval [CI] 91.4 % - 94.9 %) at 350 cells/μl and 96 % (CI 95.2 % - 96.9 %) at 500 cells/μl, with no significant difference between venous and capillary testing. Sensitivity reduced to 86 % (CI 82 % - 89 %) at 100 cells/μl (for Cryptococcal antigen (CrAg) screening), with a significant difference between venous (88 %, CI: 85 % - 91 %) and capillary (79 %, CI: 73 % - 84 %) testing. Total CD4 misclassification is 2.3 % cases at 100 cells/μl, 11.0 % at 350 cells/μl and 9.5 % at 500 cells/μl, due to higher false positive rates which resulted in more patients identified for treatment. This increased by 1.2 %, 2.8 % and 1.8 %, respectively, for capillary testing. There was no difference in Pima CD4 misclassification between the meta-analysis data and a population subset of HIV+ ART naïve individuals, nor in misclassification among operator cadres. The Pima CD4 was most similar to Beckman Coulter PanLeucogated CD4, Becton Dickinson FACSCalibur and FACSCount, and less similar to Partec CyFlow reference technologies.
The Pima CD4 may be recommended using venous-derived specimens for screening (100 cells/μl) for reflex CrAg screening and for HIV ART eligibility at 350 cells/μl and 500 cells/μl thresholds using both capillary and venous derived specimens. These meta-analysis findings add to the knowledge of acceptance criteria of the Pima CD4 and future POC tests, but implementation and impact will require full costing analysis.
Globally, 34 million individuals are infected with HIV, and currently nearly 14 million worldwide are receiving antiretroviral therapy (ART) . The number of additional HIV-positive patients eligible for ART has increased a further 12 million for a total of 25.9 million eligible patients. The treatment gap, however, remains large and better methodologies or healthcare system changes are required to improve the number of individuals initiating treatment . Many HIV-positive patients, however, do not have reliable access to required diagnostic laboratory tests, including CD4 enumeration since CD4 testing is often only available in regional laboratories. This longer turnaround time on results impacts on patient retention in care [3–5]. It should also be noted that the need for such testing and the thresholds of CD4 counts that clinicians deem relevant for treatment initiation are moving targets . In addition to ART initiation, CD4 counts are also being used as a screening tool for reflex testing to screen for and prevent Cryptococcal meningitis in patients with a CD4 count <100 cells/μl . There is therefore a critical need to expand access to HIV diagnostic testing services.
Generally, method comparison studies of new technologies compared to the reference technologies are performed to address these critical issues. The Pima CD4 (Alere, Jena, Germany) was one of the first commercially available point-of-care (POC) CD4 technologies. It entered the market in 2009 and provides a CD4 result in 20 min, is very easy to use, requires no refrigeration of reagents or controls, and can be operated with battery power . Over many years the ART initiation target in many low- and middle-income countries has been CD4 counts <200 cells/μl, expanded more recently to include thresholds of <350 cells/μl  and was further raised to <500 cells/μl in the WHO 2013 guidelines . The selection of accurate and affordable POC CD4 technologies that can increase access to testing remains necessary in many regions for attaining ambitious 2015 treatment initiation goals . Implementation of POC CD4 testing in primary health care facilities has been shown to reduce test turnaround time, reduce pre-ART loss to follow-up, and increase prompt ART initiation [10, 11], yet implementing an inaccurate and imprecise CD4 testing platform would be costly to patients and national programs.
Despite the more than 50 technical evaluation studies of the Pima CD4 being performed in dozens of countries, this has not been reported in a consolidated format nor has the venous versus capillary blood detection debate reached a conclusion. Each study adds to the breadth of knowledge, but there is little guidance on acceptable evaluation criteria specifically for CD4 testing technologies . We sought to conduct a pooled data meta-analysis to address these issues and generate guidance for national programs and future CD4 test developers. The objectives of this pooled multi-data analysis were to summarize the performance of the Alere Pima POC CD4 technology at three clinical thresholds [100 cells/μl (to identify patients in need of reflex testing for prevention of Cryptococcal meningitis); 350 cells/μl (to identify patients eligible for ART according to the 2010 WHO guidelines) and 500 cells/μl (to identify patients eligible for ART according to the 2013 WHO guidelines)] compared with several laboratory-based reference technologies and across global regions.
Study selection and data pooling
An initiative between researchers at the University of the Witwatersrand, the Clinton Health Access Initiative (CHAI), the World Health Organization (WHO) and the US Centers for Disease Control and Prevention (CDC) led to the formation of a Pima CD4 consortium comprising 34 individuals. Studies were either undergoing publication, already published evaluations on the Pima CD4, or were in-country regulatory evaluations of the technology and were willing to supply their study data. A PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) analysis was performed with a modified checklist since this “meta-analysis” involved re-analysis of observation pairs from groups willing to supply their data . The STARSD (Standards for Reporting Studies of Diagnostic Accuracy) analysis criteria were followed where applicable to method comparison of CD4 paired observations . Data sets from each group were received in MS Excel format and merged into one worksheet containing the following minimum set of variables: observation pair number, country, Pima CD4 count (cells/μl), reference CD4 count (cells/μl) and type, specimen type (capillary or venous derived) and year in which observations were collected. The “predicate”, “in-country”, “gold standard”, “standard” and “reference” CD4 technology terminology often applied to CD4 enumeration evaluation studies are collectively referred to in this article as reference CD4 technology. These included the Beckman Coulter PanLeucogated CD4 (Beckman Coulter, Miami, FL, USA), the FACSCount, FACSCalibur and FACScan (Becton Dickinson Biosciences, San Jose, CA, USA), the CyFlow (Partec, Munster, Germany) and the Guava EasyCD4 (Merck Millipore, Billerica, MA).
Description of data analysis
(a) Catagorical data analysis
The number (proportion) of CD4 observations in the following CD4 categories: <100 cells/μl; 100 – 350 cells/μl; 350–500 cells/μl and >500 cells/μl was determined for both Pima CD4 and reference methods. The data were further divided into the type of specimen (venous or capillary) tested on the Pima CD4
Significance (p ≤0.05) between categories was determined using the proportions test.
The Pima CD4 and reference CD4 observations were also converted to binary (0 = above the specified threshold and 1 = below the threshold). The observation pairs were also sorted by specimen type, comparator reference technology and year when observations were collected.
The false positive, false negative, sensitivity (ability to correctly identify patients requiring treatment) and specificity (ability to correctly identify patients not requiring treatment) were calculated for the three clinical thresholds of the entire dataset. The total misclassification rate (percentage) was calculated as the addition of false positive rate and false negative rate. The upward (percentage of patients requiring treatment incorrectly identified by the Pima CD4 as above the threshold) and downward (percentage of patients not requiring treatment incorrectly identified by the Pima CD4 as below the threshold) misclassification rates were calculated. The Q-statistic was calculated  to quantify and account for the presence of any study heterogeneity due to differences in sample size, study quality, study designs, and/or data collection methods. A bivariate and/or univariate random effects model was applied using METANDI commands in STATA 13.
(b) Numerical data analysis
Methods applied (where applicable, 95 % confidence intervals (CI) were reported)
The CD4 count paired observations were described by mean (using random effects models), median and standard deviation (SD).
The agreement between the Pima CD4 and reference technology was measured using the Bland-Altman (bias [or mean difference] and SD of the bias) ,
The Bland-Altman measures the difference between observation pairs (a-b), where method ‘a’ is the Pima CD4. The mean paired difference (the bias or accuracy) and SD of this bias (precision) were determined. A zero mean difference implies good accuracy between reference and Pima CD4 and a small SD of the bias implies good precision (low variability). The accuracy and precision are visually represented on a modified Bland-Altman difference plot with the paired difference on the vertical axis and the absolute CD4 count of the reference on the horizontal axis.
The agreement between the Pima CD4 and reference technology was also measured using the percentage similarity (mean, SD and coefficient of variation [CV]) ,
The percentage similarity is calculated as the average between the reference and Pima CD4 technology represented as a percentage of the reference technology: [([a + b]/2) /b] × 100, where ‘b’ is the reference method. Observation pairs with the same value will be 100 % similar (accurate) and observation pairs where the Pima CD4 is greater than the reference will be > 100 %, and conversely <100 % if Pima CD4 has a value smaller than the reference. The amount of variability (precision) is represented by the percentage similarity SD and overall agreement by the percentage similarity CV.
The agreement between the Pima CD4 and reference technology was also measured using the percent difference (bias, SD) 
The percentage difference is calculated as (a-b)/b (or the average between ‘a’ and ‘b’) × 100 % . Observation pairs with the same value will have no difference and therefore low percent difference, as the percentage difference method is more relative than absolute difference over the range of data.
The formula applied is pc (concordance correlation) = p (Pearson correlation [measure of precision]) x Cb (bias correction factor [measure of accuracy]) [17, 36]. The value of pc (strength of agreement) is suggested as: <0.9 (poor); 0.90-0.95 (moderate); 0.95 – 0.99 (substantial); >0.99 (almost perfect) [17, 36].
(c) Subset analysis
Description of subset
Sample size in method comparison: Few CD4 method comparison studies’ sample sizes are based on statistical criteria, but rather constrained by costs. This pooled meta-analysis data set afforded the ability to investigate potential impact of sample size on statistical outcomes. An analysis was therefore performed on a subset of data from the comparison between the Pima CD4 and FACSCount of venous derived specimens, as this was the largest subset of paired observations from a single reference and Pima CD4 comparison.
Once the data pairs were entered in MS Excel, random sample numbers (between 1 and 3,486) and irrespective of CD4 category were generated for each CD4 observation pair. This would ensure selection of sample sizes would be independent of the CD4 count and range of CD4 count. The misclassification and agreement analysis was then performed in STATA for sample sizes ranging from 50 to 4,000. The bias, SD of the bias, percentage similarity mean and SD, total misclassification, sensitivity and concordance correlation were all plotted against sample size to determine the impact of sample size on method comparison parameters.
Performance of the Pima CD4 compared to various reference technologies.
The data were sorted based on the reference CD4 method comparator performed in comparison to the Pima CD4, irrespective of study, region or year when the study was performed. The data selection, however, took into account the outcome of the analysis performed in (c) on sample size. Categorical and numerical statistical analyses were applied and results visualized in scatter plots and bar charts.
Performance of the Pima CD4 by different cadre of staff
A subset of 3,751 paired observations was evaluated for total misclassification rates based on different healthcare worker cadres of Pima CD4 operators. This subset was from 11 studies that provided such information with their data. Three cadres were defined: laboratory technician/technologist (includes scientists); laboratory assistant (a lower level of training than technicians) and clinical staff (includes nurses and lay counselors).
Categorical data analysis
The percentage contribution of observations in the four CD4 categories (<100 cells/μl; 100 – 350 cells/μl; 350–500 cells/μl and >500 cells/μl) as determined by the Pima CD4 and reference technologies found the Pima CD4 had more observations (48.2 %) with CD4 counts <350 cells/μl than reference technologies (44.0 %). In addition, more observations had CD4 counts <350 cells/μl from capillary derived (51 %) than venous derived (46 %) specimens. The proportion test indicated a significant difference (p < 0.001) between the Pima CD4 and reference technologies in the overall numbers of observations in all categories except the 350–500 cells/μl category (p = 0.243). This was similarly found among capillary derived specimens. Venous derived specimens showed no significant difference in the 0–100 cells/μl (p = 0.148) and 350–500 cells/μl (p = 1.06) category assignment by the Pima CD4 compared to reference technologies.
A subset of 584 paired observations from two studies [15, 16], that tested the performance of the Pima CD4 with specimens from HIV treatment-naïve patients, was analyzed to ensure that the results found in this pooled data meta-analysis (n = 11,803) can be applied to this critical population. This would also be useful to determine if changes in clinical thresholds for ART eligibility criteria (350 cells/μl clinical change  to 500 cells/μl ) using the Pima CD4 would differ from the above analysis. The percentage difference for naïve and meta-analysis observation pairs showed little difference: <100 cells/μl (1 % versus 0.6 %); 100–350 cells/μl (4 % versus 3.6 %); 350–500 cells/μl (−2 % versus 0 %) and >500 cells/μl (−3 % versus −4 %). The Pima CD4, therefore, performed comparably to the reference CD4 technologies overall and in each CD4 category in a subset of HIV-positive treatment-naive patients.
Categorical meta-analysis summary including random effects modeling
n = 11,803
n = 7,648
n = 4155
Mean (absolute range)
Mean (absolute range)
1.4 % (0.9 % - 2.0 %)
1.1 % (0.9 % - 1.5 %)
2.1 % (1.3 % - 3.3 %)
1.0 % (0.7 % - 1.4 %)
0.8 % (0.6 % - 1.0 %)
1.6 % (1.1 % - 2.4 %)
2.3 % (1.7 % - 3.1 %)
1.8 % (1.5 % - 2.2 %)
3.5 % (2.4 % - 5.0 %)
1.5 % (1.0 % - 2.2 %)
1.2 % (0.9 % - 1.6 %)
2.2 % (1.3 % - 3.6 %)
14.3 % (11.2 % - 18.1 %)
11.9 % (9.1 % - 15.3 %)
21.0 % (16.1 % - 27.0 %)
7.5 % (5.9 % - 9.4 %)
6.3 % (4.6 % - 8.6 %)
9.3 % (7.3 % - 11.7 %)
2.9 % (2.2 % - 3.8 %)
2.3 % (1.7 % - 3.2 %)
3.9 % (2.8 % - 5.3 %)
11.0 % (9.6 % - 12.5 %)
9.2 % (7.5 % - 11.1 %)
13.8 % (12.1 % - 15.8 %)
6.7 % (5.1 % - 8.6 %)
5.7 % (4.1 % - 7.9 %)
8.2 % (5.9 % - 11.2 %)
13.7 % (10.9 % - 17.2 %)
10.9 % (8.0 % - 14.6 %)
17.9 % (14.1 % - 22.5 %)
Cadre of staff analysis at 350 cells/ul
n = 1133, 12.0 % (4.7 % - 14.9 %)
n = 510, 11.5 % (7.2 % - 17.8 %)
n = 623, 12 % (9.3 % - 15.3 %)
n = 558, 12.1 % (9.1 % - 15.9 %)
n = 254, 6.6 % (2.2 % - 17.9 %)
n = 304, 15 % (6.3 % - 31.9 %)
n = 2060, 9.2 % (7.1 % - 11.9 %)
n = 1850, 8.3 % (6.5 % - 10.7 %)
n = 210, 13 % (7.3 % - 22.1 %)
6.7 % (5.6 % - 8.1 %)
6.3 % (5.0 % - 7.8 %)
7.5 % (5.8 % - 9.6 %)
2.6 % (2.1 % - 3.3 %)
2.0 % (1.5 % - 2.7 %)
3.6 % (2.8 % - 4.6 %)
9.5 % (8.3 % - 10.8 %)
8.3 % (7.0 % - 9.8 %)
11.3 % (9.6 % - 13.2 %)
3.9 % (3.1 % - 4.8 %)
3.1 % (2.3 % - 4.2 %)
5.0 % (3.9 % - 6.5 %)
21.8 % (18.0 % - 26.1 %)
18.7 % (14.8 % - 23.4 %)
26.3 % (20.7 % - 32.8 %)
85.7 % (81.9 % - 88.8 %)
88.1 % (84.7 - 90.9 %)
79.0 % (73.0 % - 83.9 %)
93.3 % (91.4 % - 94.9 %)
94.3 % (92.1 - 95.9 %)
91.8 % (88.8 % - 94.1 %)
96.1 % (95.2 % - 96.9 %)
96.9 % (95.8 - 97.7 %)
95.0 % (93.5 % - 96.1 %)
98.5 % (97.8 % - 99.0 %)
98.8 % (98.4 % - 99.1 %)
97.8 % (96.4 % - 98.7 %)
86.3 % (82.8 % - 89.1 %)
89.1 % (85.4 % - 92.0 %)
82.1 % (77.5 % - 85.9 %)
78.2 % (73.9 % -82.0 %)
81.3 % (76.6 % -85.2 %)
73.7 % (67.2 % -79.3 %)
The overall sensitivity of the Pima CD4 at 100 cells/μl compared to reference technologies was less (86 %) than its performance at the ART thresholds (sensitivity >93 %). The Pima CD4 may, therefore, have less ability in identifying all necessary patients requiring reflex CrAg testing. There was also a significant difference in this sensitivity between specimens tested from venous (88 %) compared to capillary (79 %) derived specimens, since the CIs do not overlap. Patients, however, not requiring CrAg reflex testing will be correctly identified by the Pima CD4 since the specificity of the Pima CD4 compared to reference technologies is high (98.5 %), and there was no significant difference between type of specimen (venous or capillary).
The impact of the sensitivity and specificity of the Pima CD4 used at the three clinical thresholds was further investigated through the extent of total numbers of patients who would be misclassified (false positive + false negative rates). The total misclassification rate of Pima CD4 was 2.3 %, 11.0 %, and 9.5 % at the 100 cells/μl, 350 cells/μl and 500 cells/μl thresholds, respectively (Table 2). In addition, the false positivity rates were higher across all clinical thresholds indicating that more patients are found eligible for treatment using the Pima CD4 than reference CD4 technology. This relationship was the same irrespective of specimen type; however, there was greater total misclassification with capillary derived blood specimen testing (3.5 % ≤100 cells/μl; 13.8 % ≤350 cells/μl and 11.3 % ≤500 cells/μl) compared to venous derived specimen testing (1.8 % ≤100 cells/μl; 9.2 % ≤350 cells/μl and 8.3 % ≤500 cells/μl).
This is similarly reflected in the downward misclassification rates, where 14 % of patients would be identified by the Pima CD4 as incorrectly requiring treatment at the ART eligibility threshold of 350 cells/μl and up to 22 % at the ART eligibility threshold 500 cells/μl compared to reference CD4 technology. The upward misclassification of Pima CD4 at the two ART initiation clinical thresholds was less: 7 % (at 350 cells/μl) and 4 % (at 500 cells/μl). Both upward and downward misclassification rates were higher among capillary derived specimens.
A subset of the data (n = 3,751 paired observations) was further analyzed to investigate any differences in the Pima CD4’s performance based on cadre of operator. Seventy percent (n = 558 laboratory assistant; n = 2,060 laboratory technician/scientist) of the tests were conducted by laboratory technicians and 30 % (n = 1,133) by clinical staff. Table 2 highlights that the total misclassification rate at 350 cells/μl was below 13 % for laboratory assistants, laboratory technicians and clinical staff. Laboratory assistants performing the Pima CD4 using venous-derived specimens had the lowest total misclassification rate (7 %), yet they also had the highest misclassification rate of 15 % performing the Pima CD4 on capillary derived specimens. Clinical staff had similar misclassification rates (12 %) using either venous or capillary derived specimens. All analyses however showed misclassification rates with overlapping CI’s indicating that technical performance of the Pima CD4 does not alter when used by different cadre of operators.
Numerical data analysis
Method comparison meta-analysis summary using numerical data
n = 11,803
n = 7,648
n = 4155
Mean (absolute range)
Mean (absolute range)
Accuracy and Precision (cells/ul)
Mean bias (Pima - Reference)
Mean bias (CI)
Percentage similarity mean %
Percentage similarity SD %
Percentage similarity CV %
Percent bias (SD) >100 cells/μl
n = 11037, −3.26 % (26.4)
n = 7190, −3.1 % (22.5)
n = 3487, −3.54 % (32.3)
Concordance correlation (Pc)
0.914 (0.911, 0.917)
0.934 (0.931, 0.937)
0.874 (0.867, 0.881)
Strength of agreement
Overall cell variance
Percentage bias across all rangesc
Performance of the Pima CD4 compared to various reference technologies
This pooled data meta-analysis not only comprises the largest single data set to date published on the performance of a single CD4 enumeration technology, but also comprises observation pairs that are representative of CD4 counts across different geographic regions, observations collected over a fairly short time period (three years) and predominantly (69 %) from high HIV prevalence settings with 55 % from resource-limited settings. There is good representation of six reference comparator technologies that is seldom possible in a single evaluation study. In addition sub-analyses were possible comparing the performance of the Pima CD4 on venous and capillary derived specimens, different cadres of staff and sub-population of HIV ART-naïve patients. The median CD4 (383 cells/μl) from the reference CD4 technology also shows that conclusions drawn from this study can be well applied to the important 350 cells/μl clinical threshold for ART initiation and categorizing this large sample size (11,803) allows for conclusions also to be extrapolated to the 100 cells/μl and 500 cells/μl clinical thresholds. This meta-analysis, therefore, provides a unique opportunity to evaluate the Pima CD4’s technical performance independent of influence from patient age, immunological status, gender, pregnancy, geographic location, HIV status, HIV subtype (by geographic location), instrument, reagent lot, assay version, operator training and sample size that may otherwise influence a smaller study’s analyses.
Overall, the Pima CD4 generates lower CD4 count values than reference technologies with the effect that more patients’ CD4 counts are categorized <350 cells/μl by the Pima CD4, and this is more marked among capillary than venous tested specimens. In absolute cell numbers this equates to an average bias between the Pima CD4 and reference technologies of −23 cells/μl with variability in the bias (SD) across the range in CD4 counts (1-2,800 cells/μl) increasing to SD = 93 cells/μl difference (23 % relative bias and 67 % similarity CV) for venous derived specimens and up to SD = 126 cells/μl difference (32 % relative bias and 113 % similarity CV) for capillary derived specimens. The overall bias across all CD4 ranges may be summarized as 10 % for venous derived and 15 % for capillary derived specimens.
Some variability was noted among the reference technologies (as has been noted by others ), with the outlier (higher variability, least similarity) being the Pima CD4 compared to the CyFlow. When the Pima CD4 was compared to the CyFlow reference technology only, this generated a positive bias and the most variability. This may be due to both technologies being based on volumetric testing and using testing volumes <50 μl. The Pima CD4 compared to the Beckman Coulter PanLeucogated reference CD4 technology yielded the least variability, but only among venous tested specimens. This may be due to the Beckman Coulter technology being based on counting total white cells to generate a CD4 count and therefore differences between fresh capillary tested Pima CD4 specimens versus >1 hour old anti-coagulated Pima CD4 and Beckman coulter tested specimens . The Pima CD4 also compared well to the FACSCalibur and FACSCount technologies, but for the FACSCount this was found only for venous derived specimens. The FACSCalibur testing requires a highly skilled operator’s input for manual gating, interpretation and complex software compared to the FACSCount and Pima CD4 which are closed with no operator input to refine the software selection of the CD4 positive cell cluster. The heterogeneity among the reference technologies illustrates the importance of selecting the most appropriate reference technology comparator for such technical evaluations of new technologies. Furthermore, it is critical that the reference technologies meet all quality requirements including participation in external proficiency testing before commencing evaluations.
Improvement in misclassification of the Pima CD4 over time was also noted, and may be due to changes in software, hardware, changes in the type and use of lancets as well as training of operators during implementation. Operator training is key to successful implementation of new technologies. This should be considered for future evaluations of early versions of new platforms that may not be fully optimized, to ensure that promising products are not unduly excluded from consideration for implementation. It is also important for national HIV treatment programs wishing to implement the Pima CD4 (or fast followers) to be aware of allowable differences in bias, resulting in misclassification rates and reduced sensitivity compared to the reference CD4 (or current “in-country”) technology, that may impact treatment costs and weigh the performance and costs with the increased patient access such a technology will allow.
Clinical relevance of meta-analysis findings, for venous and capillary derived specimen testing by the Pima CD4
Venous derived specimen testing
Capillary derived specimen testing
Is Pima suitable for screening for reflex testing of CryAg testing at the 100 cells/μl threshold?
Suitable: 88 % sensitive, Negative bias of 34 cells/μl, 1.8 % total misclassification, Good specificity >97 %
Not suitable: 79 % sensitivity, Negative bias of 73 cells/μl, 3.5 % total misclassification, Good specificity >97 %
Is Pima suitable for identifying patients eligible for ART initiation at 350 cell/μl (WHO 2010 guidelines)?
Suitable: >91 % sensitive, Negative bias 38-51cells/μl.
Suitable: >91 % sensitive, Negative bias 38-51cells/μl.
Expect 9.2 % (6.3 % false positive) total misclassification with specificity of 89 %
Expect 13.8 % (9.3 % false positive) total misclassification with specificity of 82 %,
Will increase treatment costs significantly more than venous testing.
Is Pima suitable for identifying patients eligible for ART initiation at 500 cells/μl (WHO 2013 guidelines)?
Suitable: >95 % sensitive, Negative bias 53-79 cells/μl
Suitable: >95 % sensitive, Negative bias 53-79 cells/μl
Expect 8.3 % (6.3 % false positive) total misclassification with 81 % specificity
Expect 11 % (7.5 % false positive)total misclassification with 74 % specificity Will increase treatment costs significantly more than venous testing.
Not only does this analysis highlight some difference in reference technologies, but also in some method comparison parameters. Testing a CD4 blood specimen on the same or on a different platform or test will yield a different CD4 count due to variability in accuracy and precision of both the platforms and tests. Where this variability in the CD4 count becomes important is whether or not the variability becomes clinically relevant and patient management is altered. It is this variability that is investigated in method comparisons and we are beyond the inappropriate use of correlation and linear regression for performing such analyses with CD4 counts [23, 24] but also realize a newer approach using concordance correlation has value in scaling the strength of agreement between two technologies. Appropriate methods reported in the literature for the analysis of continuous values of CD4 counts are the difference , the percentage difference , the percentage similarity  and the ratio . The latter three transform the observation pairs into values that can be compared between studies (even where different samples were tested). Specific parameters from these methods are also more informative than others for interpreting acceptable versus non-acceptable performance limits, for example, the mean bias interpreted with the confidence interval for accuracy and standard deviation of the bias for precision, and both accuracy and precision interpreted in the context of the median CD4 count of the observation set. This pooled data meta-analysis also highlighted the flaws associated with using stand-alone method comparison parameters. The Bland-Altman mean bias is not relative over the range of CD4 counts, especially >100 cells/μl and the percentage similarity and relative percentage mean bias is influenced by outliers (non-clinical) in the <200 cells/μl range. The combination of these method comparison parameters provides a more optimal evaluation across the range of CD4 counts. Analyses such as sensitivity, specificity and misclassification are not typical of CD4 technical evaluations, but in the context of CD4 being used for treatment initiation or screening for reflex testing have proved informative.
The sub-study analyses showed no difference in the Pima CD4 performance in a subset of HIV-positive ART-naïve individuals versus the meta-analysis findings. This was also true for the Pima CD4 testing performed by different cadres of operators. In addition to these findings, the subset analysis of the impact of sample size on method comparison parameters determined an average optimal sample size of 280 paired observations (n = 164 for sensitivity and n = 370 for bias calculation) for analyzing CD4 enumeration technologies. This therefore may be a guide to inform future evaluation studies for minimum sample size requirements for different methods of comparison.
While designing and conducting technical evaluations takes time and significant resources, it is critical to ensure that a technology performs comparably to reference standards. This pooled data meta-analysis implies that immunological population differences do not significantly affect the performance of CD4 diagnostic tests, especially in countries within the same geographic region. Performing a technical evaluation in every country considering a new product would, therefore, lead to significant delays in product approval, implementation in health care facilities, and improving the lives of patients. Thus, a harmonized approach could be attained with one large evaluation across sites and pooled data.
This meta-analysis focused on a method comparison using CD4 observation pairs, and no qualitative analysis of the Pima CD4 technology itself was investigated. Implementation of POC CD4 technologies will require strengthening of decentralized health care networks, including supply chain, quality assessment and program monitoring. POC CD4 technologies, however, will help achieve the bold goals set out by WHO, UNAIDS, and other global stakeholders of initiating significantly more patients on ART and improving patient access to quality care. In conclusion, this meta-analysis demonstrated that the Pima CD4 platform can generate accurate CD4 counts to be used for ART initiation in both laboratory and non-laboratory settings used by either skilled or non-skilled operators.
We thank Willy Urassa from the World Health Organisation, Geneva, Switzerland and Braimoh Bello (consultant statistician residing at Centre for Statistical Analysis and Research in South Africa), and Marta Prescott and Jessica Joseph (from the Clinton Health Access Initiative) for added discussions and statistical input.
The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the U.S. Centers for Disease Control and Prevention. This research has been supported by the President’s Emergency Plan for AIDS Relief (PEPFAR) through the Centers for Disease Control and Prevention. Use of trade names is for identification purposes only and does not constitute endorsement by the U.S. Centers for Disease Control. Grand Challenges Canada (grant 0007-02-01-01-01 [to W.S.]) provided support for WS and LES. All authors were involved in communications with the consortium members.
On behalf of the PIMA CD4 consortium: Sekesai Mtapuri-Zinyowera (National Microbiology Reference Laboratory, Harare, Zimbabwe) and Douglas Mangwanya (Ministry of Health and Child Welfare, Harare, Zimbabwe) ; James McIntyre (Anova Health Institute, Johannesburg, South Africa) ; Kovit Pattanapanyasat and Kasama Sukapirom (Office for Research and Development, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkok, Thailand; Center for Emerging and Neglected Infectious Diseases, Mahidol University, Nakhon Pathom, Thailand) ; Ilesh Jani (Instituto Nacional de Saude, Maputo, Mozambique) ; Yukari Manabe (Infectious Diseases Institute, Makerere College of Health Sciences, Kampala, Uganda; Division of Infectious Diseases, Department of Medicine John Hopkins University, Baltimore, Maryland, USA) ; Papa Alassane Diaw and Souleymane Mboup (Laboratoire de Bacteriologie-Virologie, CHU Aristide le Dantec, Universite Cheikh Anta Diop, Dakar, Senegal) ; Debbie Glencross and Lindi Coetzee (Department of Molecular Medicine and Haematology, School of Pathology, University of the Witwatersrand, National Health Laboratory of South Africa, Johannesburg, South Africa) ; Natasha Gous and Matilda Nduna (Department of Molecular Medicine and Haematology, School of Pathology, University of the Witwatersrand, National Health Laboratory of South Africa, Johannesburg, South Africa) ; Johan Potgieter and Lumka Ntabeni (Department of Haematology, Tshwane Academic Division and University of Pretoria, Pretoria, South Africa); Ntate Mothabeng (CHAI) on behalf of the Ministry of Health in Lesotho; Harry Hausler (TB/HIV Care Association; School of Pulic Health, University of the Western Cape); Jacques Boncy (Ministry of Health in Haiti) and Anna Osborne (CHAI); Linda-Gail Bekker and Nienke van Schaik (The Desmond Tutu HIV Foundation, South Africa; Department of Medicine, University of Cape Town, South Africa) ; Jozef Chisaka (Ministry of Health in Malawi); Matilu Mwau (Kenya Medical Research Institute) ; Madhuri Thakar (National AIDS Research Institute in India); Djibril Wade (Laboratory of Immunology, Department of Biomedical Sciences, Institute of Tropical Medicine, Antwerp, Belguim; Unit of Immunology, Laboratory of Bacteriology Virology, Le Dantec University Teaching Hospital, University Cheikh Anta Diop, Dakar, Senegal) and Department of Biomedical Sciences, University of Antwerp, Antwerp, Belgium . Jonathan Lehe (Clinton Health Access Initiative, Boston, MA, USA).
This meta-analysis was initiated and led by LS and WS as a need realized during their “Multiple Point of care for HIV ART initiation feasibility” project funded by Grand Challenges Canada (grant: 0007-02-01-01-01).
- UN AIDS. The Gap Report 2014. 2014. http://www.unaids.org/en/resources/campaigns/2014/2014gapreport/gapreport. Accessed November 2014.
- World Health Organisation. Consolidated guidelines on the use of antiretroviral drugs for treating and preventing HIV infection 2013. 2013. http://www.who.int/hiv/pub/guidelines/arv2013/en/.
- Larson BA, Brennan A, McNamara L, Long L, Rosen S, Sanne I, et al. Lost opportunities to complete CD4+ lymphocyte testing among patients who tested positive for HIV in South Africa. Bull World Health Organ. 2010;88:675–80.View ArticlePubMedPubMed CentralGoogle Scholar
- Rosen S, Fox MP. Retention in HIV care between testing and treatment in sub-Saharan Africa: a systematic review. PLoS Med. 2011;8:e1001056.View ArticlePubMedPubMed CentralGoogle Scholar
- Faal M, Naidoo N, Glencross DK, Venter WD, Osih R. Providing immediate CD4 count results at HIV testing improves ART initiation. J Acquir Immune Defic Syndr. 2011;58:e54–9.View ArticlePubMedGoogle Scholar
- Stevens WS, Ford N. Time to reduce CD4+ monitoring for the management of antiretroviral therapy in HIV-infected individuals. S Afr Med J. 2014;104:559–60.View ArticlePubMedGoogle Scholar
- UNITAID. HIV/AIDS diagnostics technology landscape. 4th ed. 2014. http://www.unitaid.eu/images/marketdynamics/publications/UNITAID-HIV_Diagnostic_Landscape-4th_edition.pdf.
- World Health Organisation. Antiretroviral therapy for HIV infection in adults and adolescents. Recommendations for a public health approach: 2010 revision. 2010. http://www.who.int/hiv/pub/arv/adult2010/en/.
- World Health Organisation. HIV Programme: Achieving our Goals 2014. http://apps.who.int/iris/bitstream/10665/112666/1/9789241507110_eng.pdf. UNAIDS. Treatment 2015, http://www.unaids.org/sites/default/files/media_asset/JC2484_treatment-2015_en_1.pdf
- Jani IV, Sitoe NE, Alfai ER, Chongo PL, Quevedo JI, Rocha BM, et al. Effect of point-of-care CD4 cell count tests on retention of patients and rates of antiretroviral therapy initiation in primary health clinics: an observational cohort study. Lancet. 2011;378:1572–9.View ArticlePubMedGoogle Scholar
- Jani IV, Peter TF. How point-of-care testing could drive innovation in global health. N Engl J Med. 2013;368:2319–24.View ArticlePubMedGoogle Scholar
- Stevens W, Gelman R, Glencross DK, Scott LE, Crowe SM, Spira T. Evaluating new CD4 enumeration technologies for resource-constrained countries. Nat Rev Microbiol. 2008;6:S29–38.View ArticlePubMedGoogle Scholar
- Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097.View ArticlePubMedPubMed CentralGoogle Scholar
- Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig LM, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ. 2003;326:41–4.View ArticlePubMedPubMed CentralGoogle Scholar
- Mtapuri-Zinyowera S, Chideme M, Mangwanya D, Mugurungi O, Gudukeya S, Hatzold K, et al. Evaluation of the PIMA point-of-care CD4 analyzer in VCT clinics in Zimbabwe. J Acquir Immune Defic Syndr. 2010;55:1–7.View ArticlePubMedGoogle Scholar
- Mnyani CN, McIntyre JA, Myer L. The reliability of point-of-care CD4 testing in identifying HIV-infected pregnant women eligible for antiretroviral therapy. J Acquir Immune Defic Syndr. 2012;60:260–4.View ArticlePubMedGoogle Scholar
- McBride GB. A proposal for strength of agreement criteria for Lin’s Concordance correlation coefficient. Hamilton, New Zealand: NIWA client report HAM2005-062 Report for Ministry of Health; 2005.Google Scholar
- Mwau M, Adungo F, Kadima S, Njagi E, Kirwaye C, Abubakr NS, et al. Evaluation of PIMA(R) point of care technology for CD4 T cell enumeration in Kenya. PLoS One. 2013;8:e67612.View ArticlePubMedPubMed CentralGoogle Scholar
- Glencross D, Scott LE, Jani IV, Barnett D, Janossy G. CD45-assisted PanLeucogating for accurate, cost-effective dual-platform CD4+ T-cell enumeration. Cytometry. 2002;50:69–77.View ArticlePubMedGoogle Scholar
- Wynberg E, Cooke G, Shroufi A, Reid SD, Ford N. Impact of point-of-care CD4 testing on linkage to HIV care: a systematic review. J Int AIDS Soc. 2014;17:18809.View ArticlePubMedPubMed CentralGoogle Scholar
- Hyle EP, Jani IV, Lehe J, Su AE, Wood R, Quevedo J, et al. The clinical and economic impact of point-of-care CD4 testing in Mozambique and other resource-limited settings: a cost-effectiveness analysis. PLoS Med. 2014;11:e1001725.View ArticlePubMedPubMed CentralGoogle Scholar
- Larson B, Schnippel K, Ndibongo B, Long L, Fox MP, Rosen S. How to estimate the cost of point-of-care CD4 testing in program settings: an example using the Alere Pima Analyzer in South Africa. PLoS One. 2012;7:e35444.View ArticlePubMedPubMed CentralGoogle Scholar
- Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–10.View ArticlePubMedGoogle Scholar
- Scott LE, Galpin JS, Glencross DK. Multiple method comparison: statistical model using percentage similarity. Cytometry B Clin Cytom. 2003;54:46–53.View ArticlePubMedGoogle Scholar
- Pollock MA, Jefferson SG, Kane JW, Lomax K, MacKinnon G, Winnard CB. Method comparison–a different approach. Ann Clin Biochem. 1992;29:556–60.View ArticlePubMedGoogle Scholar
- Denny TN, Gelman R, Bergeron M, Landay A, Lam L, Louzao R, et al. A North American multilaboratory study of CD4 counts using flow cytometric panLeukogating (PLG): a NIAID-DAIDS Immunology Quality Assessment Program Study. Cytometry B Clin Cytom. 2008;74:S52–64.View ArticlePubMedGoogle Scholar
- Sukapirom K, Onlamoon N, Thepthai C, Polsrila K, Tassaneetrithep B, Pattanapanyasat K. Performance evaluation of the Alere PIMA CD4 test for monitoring HIV-infected individuals in resource-constrained settings. J Acquir Immune Defic Syndr. 2011;58:141–7.View ArticlePubMedGoogle Scholar
- Jani IV, Sitoe NE, Chongo PL, Alfai ER, Quevedo JI, Tobaiwa O, et al. Accurate CD4 T-cell enumeration and antiretroviral drug toxicity monitoring in primary healthcare clinics using point-of-care testing. AIDS. 2011;25:807–12.View ArticlePubMedGoogle Scholar
- Manabe YC, Wang Y, Elbireer A, Auerbach B, Castelnuovo B. Evaluation of portable point-of-care CD4 counter with high sensitivity for detecting patients eligible for antiretroviral therapy. PLoS One. 2012;7:e34319.View ArticlePubMedPubMed CentralGoogle Scholar
- Diaw PA, Daneau G, Coly AA, Ndiaye BP, Wade D, Camara M, et al. Multisite evaluation of a point-of-care instrument for CD4(+) T-cell enumeration using venous and finger-prick blood: the PIMA CD4. J Acquir Immune Defic Syndr. 2011;58:e103–11.View ArticlePubMedGoogle Scholar
- Glencross DK, Coetzee LM, Faal M, Masango M, Stevens WS, Venter WF, et al. Performance evaluation of the Pima point-of-care CD4 analyser using capillary blood sampling in field tests in South Africa. J Int AIDS Soc. 2012;15:3.View ArticlePubMedPubMed CentralGoogle Scholar
- Gous N, Scott L, Potgieter J, Ntabeni L, Enslin S, Newman R, et al. Feasibility of performing multiple point of care testing for HIV anti-retroviral treatment initiation and monitoring from multiple or single fingersticks. PLoS One. 2013;8:e85265.View ArticlePubMedPubMed CentralGoogle Scholar
- Myer L, Daskilewicz K, McIntyre J, Bekker LG. Comparison of point-of-care versus laboratory-based CD4 cell enumeration in HIV-positive pregnant women. J Int AIDS Soc. 2013;16:18649.View ArticlePubMedPubMed CentralGoogle Scholar
- Wade D, Daneau G, Aboud S, Vercauteren GH, Urassa WS, Kestens L. WHO multicenter evaluation of FACSCount CD4 and Pima CD4 T-cell count systems: instrument performance and misclassification of HIV-infected patients. J Acquir Immune Defic Syndr. 2014;66:e98–107.View ArticlePubMedPubMed CentralGoogle Scholar
- Takkouche B, Cadarso-Suarez C, Spiegelman D. Evaluation of old and new tests of heterogeneity in epidemiologic meta-analysis. Am J Epidemiol. 1999;150:206–15.View ArticlePubMedGoogle Scholar
- Lin LK. A note on the concordance correlation coefficient. Biometrics. 2000;56:324–5.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.