CD4 T-lymphocyte count is an important qualifying test for antiretroviral treatment (ART) in HIV-positive individuals and is also used to monitor treatment efficacy.1-7 The scale up of public ART programs globally8 has led to an increased demand for CD4 count tests, especially to assess treatment eligibility. Despite expansion of laboratory infrastructure and services, access to CD4 testing remains a bottleneck to ART scale-up. In Zimbabwe, an estimated 380,000 adults are in need of ART9 and, by the end of 2009, an estimated 215,000 were on ART within the public sector.10 There is clearly a need to increase access to ART services and improving CD4 access may help.
In Zimbabwe, the “New Start” voluntary testing and counseling (VCT) centers (also known as client-initiated testing and counselling centers) are established by the Ministry of Health and Child Welfare in partnership with Population Services International (PSI) and provide free rapid HIV testing services to more than 360,000 clients nationwide on an annual basis. Clients testing positive at VCT centers are then referred to Opportunistic Infection (OI) clinics for HIV care and ART if eligible. After enrollment at the OI clinics, patients are scheduled for a CD4 count test. Due to high demand, delays in CD4 testing can occur for 2-3 weeks on average. There is substantial loss-to-follow-up of patients between HIV diagnosis and registration at the OI clinics and delays in CD4 testing can result in further loss of patients who do not return or who die before initiating treatment. The situation is exacerbated in rural areas where more limited CD4 access creates a significant bottleneck to the scale up of ART.
The recently developed PIMA point-of-care (POC) CD4 test system (Alere, Waltham, MA) conducts CD4 tests within 20 minutes of sample collection from a finger-prick drop of blood. The test can be conducted by nonlaboratory staff, and testing does not require laboratory infrastructure. We evaluated the use of the PIMA system in a VCT setting by comparing its performance against conventional laboratory-based CD4 testing. The ability of both nurses and laboratory technicians to run POC CD4 tests was also evaluated.
Newly diagnosed HIV-positive male and female participants were recruited for the study at the PSI “New Start” VCT center at New Africa House in Harare between December 16, 2009, and January 20, 2010. The center provided HIV counselling and testing services to approximately 45,000 walk-in and out-reach adult clients in 2009, of whom 52% were female and 48% were male. Approximately 9000 clients (20%) tested HIV positive in 2009.
This study was approved by the Medical Research Council of Zimbabwe. All participants provided written informed consent. Inclusion criteria included documented HIV infection and national identification. Exclusion criteria included serious medical conditions which might disrupt the accuracy of normal laboratory analysis and its interpretation; however, no potential participant met this criterion. There was no exclusion on the basis of gender, socioeconomic, racial, or ethnic grounds.
This was a cross-sectional study. Consecutive clients attending the PSI New Start VCT center were recruited for enrollment into the study during post-test counseling (PTC). Figure 1 illustrates how POC CD4 testing was integrated into the routine VCT clinic workflow. A target sample size of 150-200 study participants was set to ensure the accuracy of the bias analysis. Clients agreeing to participate provided signed consent (in Shona or English) and demographic information including gender and birth date. They then underwent immediate CD4 testing with the PIMA device followed by a blood draw by venipuncture for laboratory CD4 testing on a Becton Dickinson (BD) FACSCalibur (Becton-Dickinson, Franklin Lakes, NJ). Half of the study participants were tested on the PIMA device by nurses and the other half by laboratory technicians. Both the nurses and laboratory technicians were VCT staff who routinely carry out rapid HIV testing on clients attending the center.
Patients were not provided the PIMA CD4 result but instead received the reference laboratory FACSCalibur CD4 result when this was available, which was generally within 2-3 days. During the study, each participant was assigned a unique study identification number to link the results from the 2 tests performed on each patient and to ensure patient anonymity. To ensure blinding of the results, the technicians performing testing on the PIMA devices were not the same technicians performing the laboratory CD4 tests, and test records for each platform were kept separate.
PIMA CD4 Testing
For PIMA testing, each participant provided 1-2 drops of blood by lancet finger stick that were collected directly from the finger tip into the PIMA CD4 cartridge. A puncture depth of 1.8 mm with a blade-type lancet (Sarstedt) was used to achieve sufficient capillary blood flow. The PIMA cartridge collected the blood in a 25 μL receptacle. Of this initial volume, 5 μL of blood was drawn into the PIMA cartridge and further used for cytometric analysis. The cartridge was capped and inserted immediately into the PIMA analyzer to run the test. During the analysis process, the blood was automatically mixed with freeze-dried fluorescently labeled antibodies (anti-CD3 and anti-CD4) contained in the cartridge and transferred to a detection chamber where images were taken of the labeled cells to calculate the number of CD4 cells per μL of blood.
The results were printed and recorded after each test. Two PIMA instruments were used for the study, and each device was used equally by nurses and laboratory technicians. All test operators were formally trained on the PIMA device and sample collection methodology for half a day, and only trained operators were permitted to run tests. On each day of testing, manufacturer-provided internal quality control cartridges (normal and low CD4) were run on both instruments and were required to pass before testing proceeded.
Laboratory CD4 Testing
A second 2-3 mL blood specimen collected from each consenting participant via venipuncture into an evacuated K3 EDTA tube was sent for testing at the National Microbiology Reference Laboratory at Harare Central Hospital. The samples were tested on a BD FACSCalibur flow cytometer on the same day of collection using TriTest CD3/4/45 reagents and TruCount tubes. The CD4 results were sent back to the VCT center to be given to the study participants. Daily calibration and internal quality controls were also run on the FACSCalibur and were required to pass for CD4 data to be used in the study. The FACSCalibur was also enrolled in CD4 external quality assessment schemes via the United Kingdom National External Quality Assessment Service (NEQAS) and the South African National Health Laboratory Services CD4 External Quality Assessment Programme, and had passed assessments over the preceding 12 months.
The accuracy (bias and limits of agreement) of the PIMA device in comparison to the BD FACSCalibur was evaluated using the Bland-Altman method. Data were analyzed using SAS software (SAS Inc, Cary, NC).
Misclassification of ART eligibility was calculated using 2 × 2 tables to determine the percentage of patients who would be misclassified by the PIMA result as being below or above ART eligibility CD4 thresholds of 200 and 350 cells per microliter, compared with results from the FACSCalibur.
Table 1 contains summary characteristics of the study population tested with the PIMA device. The sample population was likely to have been representative of the actual client population at the VCT center in terms of age and gender distribution as participants were enrolled consecutively from all HIV-positive walk-in clients over the course of 5 weeks without preselection by the investigators.
Bland-Altman plots of the differences between paired PIMA and FACSCalibur results are displayed in Figure 2. The mean bias of the PIMA device was +7.6 cells per microliter relative to the FACSCalibur (95% CI: −6.6 to +21.8; P = 0.72), with 95% limits of agreement from −173.8 cells per microliter (95% CI: −198.4 to −149.1) to +189.0 cells per microliter (95% CI: +164.4 to + 213.7) (Fig. 2A). The bias was small at both low (<400 cells/uL) (Fig. 2B) and high (>400 cells/uL) CD4 counts: +8.6 cells per microliter (95% CI: −2.1 to +19.2; P = 0.54) and +2.4 cells per microliter (95% CI: −51.8 to +56.6; P = 0.96), respectively. Limits of agreement for the bias in these subsets ranged from −115.8 to +132.9 cells per microliter and −254.9 to +259.7 cells per microliter, respectively.
The performance of the PIMA device when operated by nurses or laboratory technicians was similar (P = 0.15). When testing was conducted by a laboratory technician, the bias of the PIMA device relative to the FACSCalibur was −3.1 cells per microliter (95% CI: −26.0 to +19.8; P = 0.93) (Fig. 2C). When testing was carried out by a nurse, the bias was +18.0 cells per microliter (95% CI: +0.6 to +35.3; P = 0.49) (Fig. 2D). The limits of agreement for the 2 cadres were: −205.9 to +199.7 cells per microliter (laboratory technicians) and −138.7 to +174.6 cells per microliter (nurses).
There was no significant difference between the bias of the 2 PIMA devices used in this study (P = 0.623), hence, inaccuracy inherent to individual instruments is unlikely to have significantly influenced the results. For PIMA device 1, the bias relative to the FACSCalibur was +11.1 cells per microliter (95% CI: −6.4 to +28.5; P = 0.66) and for device 2, the bias was +3.9 cells per microliter (95% CI: −19.3 to +27.1; P = 0.91). The limits of agreement were: −148.8 to +170.9 cells per microliter (device 1) and −199.2 to +206.9 cells per microliter (device 2). Both devices passed internal normal and low controls on each of the testing days and demonstrated high accuracy and repeatability with control samples (Table 2).
Using an ART initiation threshold of 200 cells per microliter, 2.4% of patients (4 of 165) were misclassified above this threshold by PIMA compared with the FACSCalibur, whereas 4.2% of patients (7 of 165) were misclassified below the threshold (total misclassification of 6.7%) (Table 3). Restricting the analysis to FACSCalibur results close to the threshold (100-300 cells/μL) increased the total misclassification to 13% (4% above and 9% below the threshold; Table 4). The misclassified results are listed in Table 5.
Using a threshold of 350 cells per microliter, 4.2% of patients (7 of 165) were misclassified above the threshold, and 2.4% of patients (4 of 165) were misclassified as below the threshold (total misclassification of 6.7%) (Table 6). Limiting the analysis to FACSCalibur results between 200-500 cells/μL increased the misclassification to 14% (8% above and 6% below the threshold; Table 7). Differences between PIMA and FACS Calibur results were higher in this subset of misclassified samples, with 5 of 11 misclassified samples differing by more than 100 cells (Table 8).
An important bottleneck to the scale up of HIV antiretroviral treatment services in Zimbabwe is access to diagnostics for identifying and staging patients. The widespread implementation of POC HIV rapid tests in VCT settings over the past decade has made it easier for individuals to access HIV screening services and subsequently care and treatment services. Access to CD4 testing has not been as widespread, especially in rural areas. Despite substantial investment in setting up CD4 laboratories, many rural patients need to travel long distances to reach testing centers and may have to make 2-3 visits before the test process is complete and results are available. At some OI clinics, there are waiting lists for CD4 testing. The coupling of POC HIV and CD4 test services in a VCT setting provides an opportunity to rapidly diagnose and assess ART eligibility in a single visit. Patients with severe immunodeficiency can also be prioritized for immediate follow-up.
This study has demonstrated that PIMA POC CD4 testing conducted in a VCT setting can produce results similar to those obtained with a laboratory-based flow cytometer. The study also demonstrated that capillary blood samples collected by finger prick are acceptable for testing with the PIMA CD4 device and yield results that are within acceptable limits of conventional CD4 testing. The bias was negligible, with a slight upward average bias of less than 10 cells per microliter at both high and low CD4 counts. The misclassification of patients at ART thresholds of either 200 or 350 cells per microliter was small with less than 10% upward or downward misclassification and less than 15% overall misclassification. Further analysis is required to establish standard thresholds for acceptable misclassification. The limits of agreement and discordance between individual PIMA and FACSCalibur results in this study were similar to that observed and accepted in other comparisons of conventional CD4 technologies.11-14 Discordance is likely to be attributable to error caused by variations in blood collection or pipeting.15
There was no difference between the performance of nurses and laboratory technicians on the PIMA device, demonstrating that laboratory technicians are not necessary to operate the instrument and that it can be used in nonlaboratory settings. Future studies should evaluate the performance of lay counselors and other lay staff in conducting POC CD4 testing.
In this study, we introduced POC CD4 testing within the workflow of routine HIV rapid testing services at a busy VCT center. Within an additional 30 minutes, clients were able to receive a CD4 test immediately after their HIV test. We demonstrated that by incorporating an offer of CD4 testing within post-test counseling, almost all eligible clients accepted, even within the context of a study and the need to provide informed consent. If CD4 testing is offered as a service in routine VCT settings, uptake rates are likely to be high. We also observed an unexpected 13% increase in client flow during the period of the study which may have been due to interest in accessing a CD4 testing service. If this service is combined with targeted referral into HIV care and treatment services, this may reduce loss-to-follow-up, allow immediate prioritization of patients with low CD4 counts, improve ART enrollment rates, and yield better outcomes for patients. “New Start” VCT centers in Zimbabwe currently diagnose approximately 70,000 clients as HIV positive annually and hence provide an important mechanism for identifying persons in need of ART. The strengthening of referral linkages between VCT and HIV treatment and care services is therefore essential to maximize the public health benefit of POC CD4 within VCT services. Assessing the magnitude of these benefits requires further study.
The PIMA device has features which suit its use in VCT settings or at district and rural clinics. It can be battery operated, requires no accessory instruments, computers, pipettes, or test consumables (other than lancets, swabs, and gloves for sample collection), and the test cartridges can be stored at room temperature. Because each test takes approximately 20 minutes to run, each device can run a maximum of 20-25 tests per 8-hour workday (1000 results are stored by the instrument). However, in settings where patients attend clinics in the morning only, the daily throughput per machine is closer to 10-15 tests per day. The PIMA device currently does not provide CD4%, hence, results of tests conducted on infants younger than 5 years should be used in conjunction with a corresponding lymphocyte percentage or with age/clinical stage tables16 to estimate CD4%.
This study used finger-prick blood samples for testing on the PIMA device. One of the drawbacks of finger-prick testing is that VCT clients may be subject to up to 3-4 separate lancet pricks to complete HIV screening and subsequent CD4 testing. It is not yet clear if this level of discomfort will discourage patients from testing, and further studies are needed to assess the acceptability of multiple finger sticks. It is also possible to test blood collected in tubes by pipetting a sample into the PIMA test cartridge. If HIV screening is routinely conducted on blood collected by venipuncture into tubes, PIMA CD4 testing can be conducted with blood from the same tube, and finger prick sampling is not required. However, finger-prick testing is faster and not dependent on the availability of trained phlebotomists. It is important to ensure that PIMA test operators are well trained on finger-prick sample collection. Preliminary observations in this study suggest that incorrect finger-prick sampling affects the reliability of POC CD4 results. The test operators in this study were highly experienced at finger-prick blood collection from HIV rapid testing, and this may have contributed to the overall high performance of the PIMA device. Future studies should evaluate the machine in the hands of less experienced operators.
This study has demonstrated that POC CD4 testing can be integrated into a VCT setting and that the PIMA POC CD4 system, whether operated by a nurse or a laboratory technician, performs acceptably compared with the BD FACSCalibur for absolute count CD4 testing. The implementation of POC testing in VCT clinics may help improve access to ART for patients in need, especially for rural populations and populations accessible mainly through mobile and outreach health services.
The authors thank the staff of both Harare Hospital OI Clinic and the PSI New Africa House “New Start” VCT center for assistance with conducting the study. The authors are also grateful for the support of Muchaneta Mugabe and Standford Mupandasekwa of the NMRL for the FACSCalibur analysis.
1. Bendavid E, Young SD, Katzenstein D, et al. Cost-effectiveness of HIV
monitoring strategies in resource-limited settings. Arch Intern Med
2. Phillips AN, Pillay D, Miners AH, et al. Outcomes from monitoring of patients on antiretroviral therapy in resource-limited settings with viral load, CD4
cell count, or clinical observation alone: a computer simulation model. Lancet
3. DART Trial Team. Routine versus clinically driven laboratory
monitoring of HIV
antiretroviral therapy in Africa (DART): a randomized noninferiority trial. Lancet
4. Lawn SD, Little F, Bekker L, et al. Changing mortality risk associated with CD4
cell response to antiretroviral therapy in South Africa. AIDS
5. O'Gorman MR, Zijenah LS. CD4
T cell measurements in the management of antiretroviral therapy-a review with an emphasis on pediatric HIV
-infected patients. Cytometry B Clin Cytom
6. Sterne JAC, May M, Costagliola D, et al. When To Start Consortium. Timing of initiation of antiretroviral therapy in AIDS-free HIV
-1-infected patients: a collaborative analysis of 18 HIV
cohort studies. Lancet
7. Peter T, Badrichani A, Wu E, et al. Challenges in implementing CD4
testing in resource-limited settings. Cytometry B Clin Cytom
9. Ministry of Health and Child Welfare. Zimbabwe National HIV/AIDS Estimates
. Harare Zimbabwe: Zimbabwe Ministry of Health and Child Welfare; 2009.
10. Ministry of Health and Child Welfare. Zimbabwe AIDS and TB Unit, Monitoring and Evaluation Report
. Harare Zimbabwe: Zimbabwe Ministry of Health and Child Welfare; 2009.
11. Sippy-Chatrani S, Marshall S, Branch S, et al. Performance of the panleucogating protocol for CD41 T cell enumeration in an HIV
facility in Barbados. Cytometry B Clin Cytom
12. Kovit P, Yuwadee P, Surada L, et al. Evaluation of a single-platform microcapillary flow cytometer for enumeration of absolute CD4
T-lymphocyte counts in HIV
-1 infected Thai patients. Cytometry B Clin Cytom
13. Kovit P, Surada L, Egarit N, et al. Evaluation of a new single-parameter volumetric flow cytometer (CyFlowgreen) for enumeration of absolute CD4_ T lymphocytes in Human Immunodeficiency Virus Type 1-infected Thai patients. Clin Diagn Lab Immunol
14. Kovit P, Hla S, Egarit N, et al. A multicenter evaluation of the panleucogating method and the use of generic monoclonal antibody reagents for CD4
enumeration in HIV
-infected patients in Thailand. Cytometry B Clin Cytom
15. Mandy F, Nicholson J, McDougal J, et al. Guidelines for performing single-platform absolute CD4
+ T-cell determinations with CD45 gating for persons infected with human immunodeficiency virus. Centers for Disease Control and Prevention. MMWR Recomm Rep
16. World Health Organization. Antiretroviral therapy of HIV
infection in infants and children in resource-limited settings: Towards universal access. Recommendations for a public health approach. 2006. Available at: http://www.who.int/hiv/pub/guidelines/WHOpaediatric.pdf
. Accessed April 16, 2010.