1. Experience using the Alere Pima ™ Point-of-Care CD4 analyzer in MSF field projects in nine countries Emmanuel Fajardo 1, Carol Metcalf 1, Pascale Chaillet 2, Erwan Piriou 3, Monique Gueguen 4, Céline Lastrucci 4, David Maman 5, Vivian Cox 6, Ruggero Giulliani 7, Syanness Tungal 8, Cara Kosack 9, Teri Roberts 10. MSF Southern Africa Medical Unit (SAMU) 1, MSF Operational Centre Brussels 2, MSF Operational Centre Amsterdam 3, MSF Operational Centre Paris 4, Epicentre 5, MSF Khayelitsha South Africa 6, MSF KZN, South Africa 7, MSF Roma, Lesotho 8, MSF Diagnostics Network, Amsterdam 9, MSF Access Campaign, Geneva 10

2. Implementation of lab-based CD4 has helped strengthen existing laboratory systems i.e. Infrastructure, transportation, etc. Trevor et al (2008) but challenges with lab-based CD4 still exist Source: Trevor Peter. Workshop on TB and HIV Diagnostics. 2011 Background: Access to lab-based CD4 testing Access to lab-based CD4 testing

3. Testing HIV RDT Testing HIV RDT Staging Eligibility CD4 test Eligibility CD4 test ART LTF Success Viral load Success Viral load LTF Enrollment and retention in care Background: HIV/AIDS Treatment Cascade

4. Testing HIV RDT Testing HIV RDT Staging Eligibility CD4 test Eligibility CD4 test ART LTF Interruptions 41% 2 32% 3 Success Viral load Success Viral load LTF 45-56% 2 23% 3 23% 4 (3 yrs) 1 Kenya AIDS survey; 2 Rosen & Fox, Plos Med (2011); 3 Kranzer & Ford, TMIH (2011); 4 Fox & Rosen, TMIH 2010; 5 Brinkoff et al. Plos One 2009 Enrollment and retention in care 46% RIP 5 56% 1 Not tested Background: HIV/AIDS Treatment Cascade

5.  In 2010 Alere Technologies GmbH, Jena, Germany officially launched in the market an innovative new Point-of-Care CD4 assay called Alere Pima™ CD4 test.  This test meets most of the ASSURED criteria for POCT (Affordable, Sensitive, Specific, User-friendly, Robust and Rapid, Equipment-free, Deliverable to those in need)  Fully automated  Self-contained disposable cartridge  Reagent stable at room temperature  CD4 absolute count (cells/ µL)  Highly portable analyzer  Results in 20 minutes  Capillary or venous blood  Rechargeable battery power  Connectivity Background: New POC CD4 technology

6. Alere Pima CD4 Test

7. Alere Pima™ CD4 On Board Controls The Pima CD4 analyzer performs a series of checks to ensure that all steps of the analysis process are completed successfully. If one of these checks discovers an error, the test will be aborted automatically resulting in an invalid result and the CD4 quantifiable result is not generated. These checks are: 1.Barcode and Expiry QC 2.Volume QC 3.Device QC 4.Reagent QC

8. Main Controls Barcode readable? – Barcode error Assay Software – Invalid barcode Expiry date – Error 203 1 2 3 Focus control – Error 820, 840, 870 Movement control – Error 880 Exposure control – Error 830, 850 Volume control – Error 201 Reagent control – Error 860 Channel filling – Error 810 4 5 6 7 8 9

9. Validation Studies on Alere Pima CD4 AuthorYearCountrySettingUserPopulationFinger prickEDTAStandard Zinyowera 2010ZimbabweVCTNurse; Lab techHIV+ adultsn=165FACSCalibur Sukapirom 2011ThailandLabLab techHIV+ adultsn=203FACScount; FACScan Jani 2011MozambiqueClinicNurse; Lab techHIV+ adultsn=135n=140FACSCalibur Diaw 2011SenegalClinicPhysician, nurse, lab techHIV+; HIV-n=95n=100FACSCount N van Shalk 2011South AfricaMobile ClinicNurseHIV+ adultsn=48n=160PLG Beckman Coulter Glencross 2012South AfricaLab, Hospital, clinicLab tech, nurseHIV+ adultsn=91; n=96n=100PLG Beckman Coulter Mnyani 2012South AfricaClinicNurseHIV+ pregnant womenn=296PLG Beckman Coulter Barnabas 2012South AfricaHome-based testingNursesHIV+ adultsn=177FACSCalibur Thakar 2012IndiaART centreLab techHIV+ adultsn=175n=1790FACSCalibur, FACScount, Partec Herbert 2012UKClinicNursesHIV+ adultsn=254Becton Coulter FC500 Manabe 2012UgandaHospital clinicNurseHIV+ adultsn=176 FACSCalibur Baker 2012UKLaboratoryLab techHIV+ adultsn=100Beckman Coulter CDC 2012USALaboratoryLab techHIV+ adultsn=105FACSCalibur Yothipitak 2012ThailandLaboratoryLab techHIV+ adultsn=300FACSCalibur Mwau 1 2013KenyaLaboratoryLab techHIV+ adults and childrenn=521FACSCount Mwau 2 2013KenyaLaboratoryLab techHIV+ adults and childrenn=162Partec Mwau 3 2013KenyaLaboratoryLab techHIV+ adults and childrenn=176Guava Mwau 4 2013KenyaLaboratoryLab techHIV+ adults and childrenn=396FACSCalibur Mwau 5 2013KenyaLaboratoryLab techHIV+ adults and childrenn=822FACSCount Mwau 6 2013KenyaLaboratoryLab techHIV+ adults and childrenn=407Partec Mwau 7 2013KenyaLaboratoryLab techHIV+ adults and childrenn=191Guava Myer 2013South AfricaClinicNurse, counsellorHIV+ pregnant womenn=521PLG Beckman Coulter Galiwango 2014UgandaClinicLab techHIV+ adultsn=903FACSCalibur Rathunde 2014BrazilLaboratoryLab techHIV+ adultsn=107FACSCalibur

10. Validation Studies on Alere Pima CD4 Capillary blood EDTA venous blood Bland-Altman Analysis

11. Impact Studies on Point-of-Care CD4 Testing

12. Larson et al (2012) in South Africa found an average cost per CD4 test in a mobile HTC program of $23.76 compared to the fee of laboratory-based testing of $7-8 per test (provider’s perspective). Estimation of costs included material, labour, equipment, insurance and QC. Supervision, training and EQC were not included in the costing analysis. Ilesh Jani in Mozambique reported a cost per CD4 test of $11.78 compared to $10.50 lab- based CD4. As testing volumes rise, POC CD4 becomes more cost-competitive. Costing Studies on POC CD4 Testing Stander et al in South Africa built a mathematical model to estimate the effect of POC CD4 staging in patient’s outcomes which are measured in economic, financial and public health metrics. They found that the CD4 POC technology is highly cost-effective compared to lab-based CD4 testing (economic perspective). The model predicts that, within the first six months, R48.6 million can be saved through the use of POC CD4 testing (financial perspective). The model also predicts that at five years 35,549 infections are averted (public health perspective)

13. - Commercially launched in 2010 - Prequalified by WHO in December 2011 - Procured in more than 50 countries by 2014 Adoption of Alere Pima CD4 Worldwide Source: http://alerehiv.com/studies-implementation-world-maphttp://alerehiv.com/studies-implementation-world-map

14. Use of Alere Pima CD4 in MSF  In 2011, Médecins Sans Frontières introduced Pima CD4 testing in multiple HIV projects in order to decentralize CD4 testing and reduce loss to follow- up. A high proportion of invalid results have been reported in several projects.  A high proportion of invalid results has important operational implications:  A second cartridge needs to be used.  Increases the average price of testing per patient.  Doubles the turnaround time.  Reduces CD4 testing throughput.  If finger-prick is used, the patient needs to be re-pricked, causing discomfort.  Instrument may have to be sent for technical maintenance and programmes incur in extra costs related to shipment.  Loss of confidence of the test and, ultimately, in frustration by the end-users.

15. Study Description Study objectives 1. Describe the frequency of invalid Pima CD4 results in participating MSF projects, and the settings, conditions of use of Pima CD4 testing in these projects. 2. Identify factors associated with the occurrence of invalid results and the sources of errors. 3. Describe the advantages and challenges experienced with the use of the Pima CD4 analyzer in MSF projects under programmatic conditions. Timeframe: from January 2011 to June 2013. Study countries: 9 countries (India, Guinea, DRC, CAR, Kenya, Malawi, Mozambique, South Africa, Lesotho). Study sites: laboratories, clinics, mobile teams (including home-based testing). Study Design: retrospective analysis of routinely-collected programmatic data.

16. METHODOLOGY Field staff at each site completed a questionnaire to provide information related to the use of the Pima CD4 analyzer. Export files for Pima test and Control Beads from each instrument were collated and analysed in Stata 11. Logistic regression was used to determine factors associated with errors, controlling for confounders. Country Number of sites Number of instruments Number of facilities Type of facility (clinic, mobile team, lab) Number of users Type of user (clinician, lay worker, lab staff) Type of sample (finger prick, EDTA whole blood) Sample procedure (type of transfer pipette, etc) Type of training (local, by MSF, refresher trainings) Maintenance: breakdowns, instrument swap, place for maintenance (local or international) Export files for test and controls Test ID, Device ID Assay ID, Sample ID CD4 value Error message Operator Test date Control Barcode Control Cartridge expiry date Control Sample volume Control Device Control Reagent Software version QuestionnaireExport files

17. Data Characteristics  Time period: 1 January 2011 - 30 June 2013  Records in original dataset: 26,901  Exclusions: 1,152 (<50 tests using device, user unknown, user did <20 tests)  Records analysed: 25,749  9 countries  33 sites  54 devices  155 users  Users per device: Median: 3 (IQR: 2-5)  Tests per device: Median: 470 (IQR: 252 – 672)  Tests per user: Median: 106 (IQR: 50 – 216)  Errors: 3,393 (13.2%)  Errors per device: 12.7% (IRQ: 10.3 – 19.9)  Errors per user (of users who did ≥50 tests): Median: 12.1% (IQR: 7.1 – 19.2)

18. TEST CHARACTERISTICS Most devices were used in fixed clinics, followed by home-based testing and mobile clinics. There was a similar proportion of users i.e. clinicians (registered nurse, nurse assistant, clinical officer, doctor), lay workers (counsellors, community health workers) and lab staff Most tests were run with EDTA whole blood (62%) while 38% of the tests were done with capillary blood.

19. Invalid Rate per country

20. Invalid Rate by Instrument Number of instruments = 54 Number of tests = 25,749 Number of valid tests = 22,293 Number of invalid tests = 3,393 Range invalid rate per instrument = 2.2% - 28.3% Median invalid rate per instrument = 13.0% 92.5% of instruments had an invalid rate >5% 8.5% (n = 4) of instruments had an invalid rate <5% 12.8% (n = 6) of instruments had an invalid rate of 5 – 10% 61.7% (n = 29) of instruments had an invalid rate of 10 – 20% 17.0% (n = 8) of instruments had an invalid rate of 20 – 30% Manufacturer’s recommended rate

21. Invalid Rate by Period Errors did not differ significantly by period or show a time trend.

22. Invalid Rate by User Users <50 tests done were excluded Number of users with >20 tests= 155 Number of tests = 25,749 Number of valid tests = 22,356 Number of invalid tests = 3,393 Range invalid rate per user = 1.3% to 49.2% Median invalid rate per user = 13.2% 87.8% of users had an invalid rate >5% 12.2% (n = 14) of users had an invalid rate <5% 25.2% (n = 29) of users had an invalid rate of 5 – 10% 38.3% (n = 44) of users had an invalid rate of 10% - 20% 20.9% (n = 24) of users had an invalid rate of 20% - 30% 3.5% (n = 4) of users had an invalid rate >30% Manufacturer’s recommended rate

23. Errors according to test characteristics Errors were more likely to happen in the laboratory (adjusted Relative Risk (aRR) = 1.17, p <0.001). There was no significant difference in the proportion of errors by user type; however, lay workers were less likely to generate errors compared to clinicians (aRR = 0.85, p = 0.027).

24. Errors were less likely to happen with capillary blood (adjusted Relative Risk = 0.93 p <0.001). There was no significant difference between invalid rate and number of users Errors according to test characteristics

25. Type of Errors No Error code No of errors Percentage 18501,23336.53% 288061318.16% 39103169.36% 48602958.74% 52002326.87% 62012286.76% 79401343.97% 8810912.70% 9203471.39% 10825330.98% 11No result260.77% 12210190.56% 13870180.53% 14930180.53% 15206170.50% 16820170.50% 17840150.44% 1892090.27% 1920770.21% 2983070.21% TOTAL3,393100.00% 73% Can be caused by the instrument or the user Operator related 850, 860, 880, 910 Bubbles and optics blocks 200: Test aborted This can be caused intentionally by manually aborting the analysis 201: Volume Too little sample volume Unusual sample characteristics Failure of the volume sensor of analyzer Cartridge is run for a 2 nd time 14% In order to know whether the problem is due to the instrument or the user, a full diagnostic assessment needs to be done by the Alere Technical Hub. This entails sending technical files that are huge in size and require alternative transfer methods and good internet connection

26. TROUBLESHOOTING ERRORS  A 62-page document on error troubleshooting is available from Alere (Pima Error Resolution Guide). It requires assistance from Alere because one error can have multiple causes making troubleshooting complex. Error source Operator: 200, 201, 203, 206, 207, 810, 880, 910 Operator/instrument: 850, 860 Instrument: 820, 825, 840, 870 Sample: 830, 920, 940 Error type Volume: 200, 201, 206, 207, 810, 820, 880, 910, 920, 940 Bubbles: 810, 880, 910 Volume/optic block: 850, 860 Autofocus: 820, 825, 840, 870 Expired cartridge: 203

27. TROUBLESHOOTING ERRORS 5 countries (South Africa, Mozambique, Kenya, India and Malawi) had a local Alere distributor with capacity for technical maintenance. 5 countries had to send instruments to another country (Lesotho, DRC, Guinea, CAR). We were unable to analyze errors according to cartridge lot number because this is not contained in the cartridge barcode; and sites did not document lot numbers systematically. Constant monitoring and troubleshooting are important aspects of POC activities in order to: -Reduce operator errors -Identify and investigate ambigous errors -Repair faulty instruments -Retrain users -Improve training -Improve quality of materials

28. AuthorCountryError rate Sukapirom, 2011Thailand5% Diaw, 2011Senegal10% N van Shalk, 2011South Africa14% Glencross, 2012South Africa11% Thakar, 2012India2% Manabe, 2012Uganda8% CDC, 2012USA12% Myer, 2013South Africa15% Gous, 2013South Africa16-19% Validation studies reporting errors

29. MAINTENANCE  The Alere™ CD4 Analyser is warranted for 12 months from the date of sale. The warranty cover all costs for repair but no shipment costs.  After the warranty is expired, Alere offers 3 options for extended care plan:  5 year contract (cheaper and instrument is changed for a new one at year 3)  1 year contract (more expensive)  Pay-as-you-go: no contract, instrument is sent only when it requires maintenance  Given the high occurrence of errors, having a maintenance contract is important, but it can be costly. 1 year maintenance contract would cost MSF $96,000 for 80 instruments. No shipment costs included.

30. Manufacturer’s recommendations when using finger prick whole blood Automatic lancets are preferred for better blood flowDon’t milk the tip of the finger... Instead hold base of the fingerAvoid bubbles!!Clean the cartridge window

31. Manufacturer’s recommendations when using EDTA venous blood The use of a glass capillary avoids formation of bubbles compared to pipettes Clean the cartridge window from dust, glove powder or dirt before running the test.

32. Data Management -Data from Pima device were exported to a USB and read in an excel-spread sheet in a computer; however, it required staff (e.g. VCT coordinator or lab staff ) to go around in the clinics with certain frequency to collect the data. -POC CD4 registers were not in place or were not properly filled and data was incomplete and prone to transcriptional errors. -The Alere connectivity allows data transfer on real-time via a modem and SIM card  FTP server. -This is being piloted in KZN, South Africa in 11 clinics Modem SIM card Web portal

33. CONNECTIVITY - POC manager can then view data on web portal (Alere Data Point and produce several reports from a dashboard. - Connectivity facilitates real-time monitoring of : 1. devices: maintenance 2. Users: targeted training 3. QC results: instrument performance 4. Stock levels: manage cartridge stock 4. Indicators: tests CD4 <350, <500. Disadvantages: 1. No patient information (eligibility vs. FU) 2. Requires good mobile coverage. 3. Additional cost at the moment ($500 for one connectivity pack). Alere is planning to include it in new sold instruments. Errors QC

34. CONCLUSIONS  POC CD4 is an important tool to increase access to CD4 testing and has a significant impact on patients and the health system.  Alere Pima POC CD4 is the only available POC CD4 technology in the market has been valuable to identify patients eligible for ART and increase access in resource-limited settings.  Our study shows that invalid rate is caused my multiple factors and this is a limitation of the system  There are other POC CD4 technologies in the pipeline (BD FACSPresto, Zyomix, Dakrari, Visitect, MBio), but they need to be validated. Once validated they can be considered for implementation.

35.  MSF Laboratory Working Group  Field projects  Alere technical support, especial thanks to Ralph Labugger (Alere, Germany) and Greg Khoury (Alere, South Africa) for their technical advice and constant support ACKNOWLEDGEMENT