A major challenge facing South Africa is the concomitant HIV and tuberculosis epidemics. The National Health Laboratory Service provides testing for staging HIV-positive patients, monitoring patients on antiretroviral therapy (ART) and diagnosing tuberculosis. Not all health districts have equivalent ART-related coverage in particular for CD4 and HIV viral load testing.
The Integrated Tiered Service Delivery Model coverage precinct approach was used to address ART-related testing service coverage gaps in a manner that balances cost, quality and equity.
An algorithm was developed to identify and address ART-related diagnostic coverage gaps. Data was extracted from the corporate data warehouse and Oracle systems for the period of April 2015 to March 2016. Daily test volumes were based on 21.73 working days per month. Data were analysed using MS Excel and mapped using ArcCatalog and ArcMap. Capacity analysis was informed by the available testing-platforms.
Health district daily HIV viral load volumes ranged from 2 to 1308 samples. Nineteen candidate laboratories were identified to address the coverage gaps. Following the proximity analysis, testing was consolidated at four candidate laboratories, resulting in 13 revised candidate laboratories. The revised candidate laboratory daily HIV viral load referrals ranged between 5 and 205 samples, with CD4 volumes between 6 and 85 samples. Remaining coverage gaps were identified in seven municipalities.
The study demonstrated that the service coverage precinct approach could be used to identify coverage gaps for a defined ART-related testing repertoire.
A major challenge facing South Africa is the concomitant HIV and tuberculosis epidemics.
In 2006, the South African government approved the ambitious National Strategic Plan for HIV and AIDS and sexually-transmitted infections (2007–2011) and committed the government to providing ART to 80% of those eligible.
The required ART scale up in South Africa necessary to meet the HIV counselling and testing campaign targets led to the announcement that accreditation would be abandoned and that all public healthcare facilities would be geared up to provide ART.
Currently, the Joint United Nations Programme on HIV/AIDS (UNAIDS) estimated that seven million people are living with HIV in South Africa with a prevalence rate of 19.2% for adults (>15 years).
In 2015, six countries accounted for 60% of the global tuberculosis burden, with the highest burden in India, followed by Indonesia, China, Nigeria, Pakistan and South Africa.
The laboratory service plays a critical role in diagnosing tuberculosis and monitoring treatment.
For the HIV programme, the NHLS provides CD4, HIV viral load, Xpert MTB/RIF and other routine testing to support ART initiation, as defined in the national guidelines.
The Integrated Tiered Service Delivery Model (ITSDM) described for CD4 testing across South Africa to achieve coverage, proposed 104 CD4 testing sites consisting of 60 laboratories (Tier-3 to Tier-5), 22 point-of-care (POC) hubs (Tier-2), and 22 decentralised POC testing sites (Tier-1) that would only service one health facility.
A national burden of disease study conducted between 1997 and 2010 by the South African Medical Research Council reported that the four major broad causes of death in South Africa were HIV and tuberculosis, non-communicable diseases, injuries, and other Type One conditions such as nutritional deficiencies.
The lessons learnt from implementation of the CD4 ITSDM
Ethics clearance for this work was obtained from the University of the Witwatersrand (study number: M1706108). This study involved the secondary analysis of laboratory test volumes data that does not contain any patient identifiers. No patient recruitment was necessary as routine laboratory data was used for the study.
This study is based on the lessons learnt from the ITSDM and extends these concepts to develop an algorithm that addresses service coverage for a defined test repertoire. The algorithm developed was used to identify existing NHLS laboratories (candidates) that are not currently offering HIV viral load and CD4 testing that could be used to offer these diagnostic services and address ART-related testing coverage gaps (
Flow diagram steps used to identify candidate laboratories to address antiretroviral therapy-related testing coverage gaps, South Africa.
The seven steps used in the algorithm are described below:
A list of HIV viral load laboratories was compiled, including latitude and longitude coordinates. These data were collected using a spreadsheet and converted to a shapefile using ArcCatalog (Redlands, California, United States) and loaded as a new layer on ArcMap (Redlands, California, United States)
HIV viral load samples tested within the NHLS are recorded in the laboratory information system through a data interface to the analyser. Specimen-level laboratory information system data are then replicated to the corporate data warehouse for national programmatic reporting. Health district HIV viral load test volumes for the period April 2015 to March 2016 (fiscal year [FY]2015/2016) were extracted from the corporate data warehouse. The data extract included the health district name and total HIV viral load test volumes, that is, the number of samples tested in the FY2015/2016. Daily health district HIV viral load test volumes were calculated assuming an average of 21.73 working days per month and 12 months per year. In South Africa, the Municipal Demarcation Board is responsible for determining municipal boundaries.
Using the symbology functionality in ArcMap,
A Microsoft Excel list of NHLS laboratories that do not currently provide HIV viral load testing, with their respective latitude and longitude coordinates, were loaded as a new layer on ArcMap. The maps were printed on an A3 paper in colour and given to three individuals to identify candidate laboratories in the current HIV viral load coverage gaps. The three lists were reviewed to generate the approved list of candidate laboratories. This list was converted to a shapefile using ArcCatalog.
For each test performed within the NHLS network, the laboratory information system generates single or multiple tariff code or codes to generate expenditure data (e.g. tariff code 3020 is allocated to the glucose test). These data are used to generate accounts on the Oracle enterprise resource planning system in the accounts receivable module,
In a geographic area within a 150 km circumferential service coverage precinct, one laboratory would be able to provide sufficient coverage, removing the need to over-capacitate multiple laboratories. The aim of this step was to prevent over-capacitation and provide optimal testing at a candidate laboratory with the required test repertoire. An independent visual inspection was again conducted by three single individuals. They were requested to identify laboratory clusters on the printed map for candidate laboratories in close proximity. The responses were consolidated to identify clusters. For each cluster identified, Google Maps was used to determine inter-laboratory distances (kilometres) and drive times (minutes). The Google Maps drive time is not able to factor aspects such as stop or go sections due to road construction, adverse weather, traffic congestion and terrain and road condition. Where laboratories in a cluster were >150 km from each other, they were added to the list of revised candidate laboratories, as the laboratories were too far apart for consolidated testing. Where the inter-laboratory distance was ≤ 150 km for laboratories in a cluster, the laboratory currently offering CD4 testing and reporting the highest test volumes was the preferred choice for consolidation in the cluster. Where consolidation was indicated, the combined HIV viral load and CD4 volumes were used to determine the required testing capacity.
Where a candidate laboratory did not currently offer HIV viral load or CD4 testing based on volumes and revenue year-to-date data, referred test volumes were used. All daily volume calculations assumed 21.73 working days per month and 12 months per year. This assumption is based on small district (rural) laboratories that offer an 8-hour service, but may not be applicable to larger laboratories that offer a 24-hour service impacting on testing capacity.
The capacity analysis was done in consultation with three individuals from the National Priority Programme unit, with expertise in either CD4 or HIV viral load testing. One HIV viral load and CD4 platform was then allocated to each candidate laboratory based on the daily volumes or referrals (consolidated volumes where applicable). For the purpose of this paper, daily CD4 volumes of less than ten samples were allocated to the FacsPresto (Becton Dickinson, San Diego, California, United States) platform.
In the identified clusters where consolidation was proposed, excluded laboratories were removed to generate a list of revised candidate laboratories. The list of revised candidate laboratories, with latitude and longitude values, was converted to a shapefile using ArcCatalog and added as a new layer (reported as red circles) on ArcMap.
For municipal areas without adequate coverage, town names were added to ArcMap. For these areas, either POC testing or additional laboratory sites are proposed.
Only 5 out of 52 districts reported daily HIV viral load test volumes between 601 and 1308, namely City of Cape Town Metro, City of Johannesburg Metro, Ehlanzeni, Ekurhuleni Metro and eThekwini Metro (
Current daily HIV viral load test volumes per day per health district with the laboratories offering testing (n=17), South Africa.
Nine health districts reported daily HIV viral load test volumes between 301 and 600; 22 health districts reported a daily HIV viral load test volume between 151 and 300 samples; only eight health districts had daily HIV viral load volume between 1515 and 300 while just 6 districts reported between 16 and 64 tests per day.
There were two districts that reported daily volumes ≤ 16 samples per day, namely Central Karoo and Namakwa health districts.
The ArcMap analysis reported adequate HIV viral load coverage in the Gauteng, KwaZulu-Natal and Mpumalanga provinces (
Current HIV viral load coverage provided using a service coverage precinct of 150 km circumference and the identification of candidate laboratories in the coverage gaps (
Current testing offered by the candidate laboratories to address the antiretroviral therapy-related testing coverage gaps, South Africa.
Number | Province | Laboratory alias | Testing offered on-site | |||||||
---|---|---|---|---|---|---|---|---|---|---|
HIV viral load | CD4 | CrAg |
Xpert MTB/RIF | eGFR | ALT | FBC | HbSAg | |||
1 | Eastern Cape | EC1 | No | No | No | Yes | Yes | Yes | Yes | No |
2 | EC2 | No | No | No | Yes | Yes | Yes | Yes | No | |
3 | EC3 | No | No | No | Yes | Yes | Yes | Yes | No | |
4 | Limpopo | LP1 | No | No | No | Yes | Yes | Yes | Yes | No |
5 | LP2 | No | No | No | Yes | Yes | Yes | Yes | No | |
6 | Free State | FS1 | No | No | No | Yes | Yes | Yes | Yes | No |
7 | FS2 | No | Yes | Yes | Yes | Yes | Yes | Yes | No | |
8 | North West | NW1 | No | Yes | Yes | Yes | Yes | Yes | Yes | No |
9 | NW2 | No | No | No | Yes | Yes | Yes | Yes | No | |
10 | Northern Cape | NC1 | No | Yes | Yes | Yes | Yes | Yes | Yes | No |
11 | NC2 | No | No | No | Yes | Yes | Yes | Yes | No | |
12 | NC3 | No | No | No | Yes | Yes | Yes | Yes | No | |
13 | NC4 | No | Yes | Yes | Yes | Yes | Yes | Yes | No | |
14 | Western Cape | WC1 | No | No | No | Yes | Yes | Yes | Yes | No |
15 | WC2 | No | Yes | Yes | Yes | Yes | Yes | Yes | No | |
16 | WC3 | No | No | No | Yes | Yes | Yes | Yes | No | |
17 | WC4 | No | No | No | Yes | Yes | Yes | Yes | No | |
18 | WC5 | No | No | No | Yes | Yes | Yes | Yes | No | |
19 | WC6 | No | No | No | Yes | Yes | Yes | Yes | No |
Reflexed CrAg screening offered routinely as from 01 July 2017.
CD4, Cluster differentiation 4; CrAg, Crytococcal antigen; Xpert MTB/RIF, GeneXpert ; eGFR, Estimated Glomerular Filtration Rate; ALT, Alanine Transminase; FBC, Full Blood Count; HbSAg, Hepatitis B Surface Antigen.
All candidate laboratories (
Four clusters were identified following visual inspection. Inter-laboratory distances ranged from 52.6 to 83.1 kilometres. Similarly, travel times ranged from 44 to 62 min (
Analysis of additional coverage provided by the 13 candidate laboratories that would provide the required diagnostic support for antiretroviral therapy services, South Africa.
Google Maps distances and travel times between candidate laboratories in close proximity across the four clusters identified, South Africa.
Candidate consolidated laboratory to improve coverage | Candidate laboratories in close proximity | Distance | Travel time |
---|---|---|---|
NW1 | NW2 | 74.2 km | 51 min |
WC2 | WC3 | 62.1 km | 56 min |
WC4 | 52.6 km | 44 min | |
WC5 | 59.4 km | 48 min | |
FS2 | FS1 | 80.0 km | 62 min |
EC2 | EC3 | 83.1 km | 53 min |
NW, North West ; WC, Western Cape; FS, Free State; EC, Eastern Cape; km, kilometres; min; minutes.
Daily HIV viral load referrals for the revised candidate laboratories (
Daily HIV viral load and CD4 test volumes, capacity required and testing tier for candidate laboratories, South Africa.
Number | Laboratory | HIV viral load | CD4 | ||||
---|---|---|---|---|---|---|---|
Daily volumes | Proposed platform | Quantity | Daily volumes | Proposed platform | Quantity | ||
1 | EC2 | 36 | GX16 | 1 | 42 | BC Aquios | 1 |
2 | LP1 | 26 | GX16 | 1 | 25 | BC Aquios | 1 |
3 | LP2 | 23 | GX16 | 1 | 18 | BC Aquios | 1 |
4 | FS2 | 205 | GX16 | 4 | 83 | BC Aquios | 1 |
5 | NW1 | 138 | GX16 | 3 | 54 | BC Aquios | 1 |
6 | NC1 | 18 | GX16 | 1 | 18 | BC Aquios | 1 |
7 | NC2 | 5 | GX4 | 1 | 6 | BD FacsPresto | 1 |
8 | NC3 | 46 | GX16 | 1 | 56 | BC Aquios | 1 |
9 | NC4 | 25 | GX16 | 1 | 19 | BC Aquios | 1 |
10 | WC1 | 7 | GX4 | 1 | 8 | BD FacsPresto | 1 |
11 | EC1 | 10 | GX4 | 1 | 8 | BD FacsPresto | 1 |
12 | WC6 | 6 | GX4 | 1 | 9 | BD FacsPresto | 1 |
13 | WC2 | 80 | GX16 | 2 | 85 | BC Aquios | 1 |
LP, Limpopo; NW, North West; WC, Western Cape; FS, Free State; EC, Eastern Cape; GX, GeneXpert; BD, Beckton Dickinson; BC, Beckman Coulter
For CD4 testing, four laboratories with lower daily volumes (6–9 samples) would cope with the low throughput of a BD FacsPresto platform. The remaining laboratories (
The majority of the ART-related diagnostic coverage gaps would be addressed with the exception of some rural municipal areas that include Brandvlei, Bredasdorp, Pofadder, Pomfret, Prieska, Swellendam and Williston, with daily HIV viral load test volumes of ≤ 6 samples per day (
This article describes an approach to establishing where service deficiencies lie across a national programme, using a step-by-step approach. Each step provides detail of the analysis undertaken to make the final decision to identify and address coverage gaps. The work is by no means fully comprehensive but should be viewed as a guideline of how to approach the mammoth task of where to start rolling out laboratory services across a national programme, irrespective of whether the programme is a vast network of laboratories that may be required for extensive ART support services or a smaller network that involves other clinical laboratory testing.
The adequacy of ART-related diagnostic coverage is best addressed across a defined test repertoire that provides access to all testing required for ART initiation and monitoring of HIV-positive patients enrolled on ART. Addressing the coverage gaps for a single test would not have the desired impact on ART initiation, as patients would be required to wait for those tests results that are not included in the local test repertoire.
The 250 km service coverage precinct was based on the hub-and-spoke logistics model used within the NHLS.
An advantage of providing ART-related coverage is the ability to support integrated patient care envisaged by the Ideal Clinic Initiative.
Integrated care could be supported by providing decentralised ART-related diagnostics. Improving the capacity of local haematology and chemical pathology testing would have a broader impact for the diagnosis of chronic diseases such as diabetes and cardiovascular diseases.
For additional ART monitoring tests, such as fasting cholesterol and triglycerides, it would be beneficial to perform these tests at larger referral NHLS laboratories, where both the required platforms and mono-specialist trained medical technologists are available. These test results are not required for initiation onto ART, but can be available for the subsequent clinic visit. If a decision is made to implement fasting cholesterol and triglyceride testing in a candidate laboratory, the Alere Cholestech LDX analyser (Alameda, California, United States) has been evaluated and found to provide clinically-equivalent results when compared to conventional laboratory based platforms.
Outcomes from the CD4 ITSDM suggest that using existing NHLS laboratories first to extend diagnostic coverage before investigating POC hubs or decentralised POC (true-POC) could reduce costs related to fit-for-implementation site development, training and competency costs and improve capacity.
Once approval has been obtained to implement ART-related testing at the proposed revised candidate laboratories, extensive planning is required, including site visits, gap analysis and identifying and addressing staffing requirements. In the implementation phase, activities include laboratory preparation, staff recruitment, analyser procurement followed with delivery and set-up, user training and finally, competency assessment and verification (fit-for-purpose). These activities need to be synchronised through a harmonising coordinating body, described as Tier-6 in the ITSDM, which will coordinate, streamline and standardise all processes relating to implementation, training and quality. In the South African context, this harmonising tier is provided for by the National Priority Programme Unit of the NHLS. Additional cadres of staff, such as the phlebotomist technicians or medical technicians,
Improved access to ART-related testing should improve the overall access to laboratory service delivery to support clinical services in these remote areas. Furthermore, building capacity for ART-related testing could be an important stepping stone to improving access for the diagnosis of chronic diseases such as cardiovascular diseases, diabetes, chronic respiratory conditions and cancer.
The algorithm is intended to be an iterative process. Once a solution has been implemented, the algorithm could be repeated until all coverage gaps are addressed. If the step-by-step methodology is automated on the corporate data warehouse, this approach could be applied to multiple tests in real time, enabling key coverage decisions within the NHLS.
This study was a desktop exercise and is based only on data extracted for FY2015/16. It does not reflect NHLS plans for expansion of services but is merely an exercise to demonstrate the processes followed in the development of the ITSDM and how the model can be applied to inform on other laboratory services.
The concepts and methodology described here can be developed in any organisation to provide real time analysis of coverage gaps in a national network of laboratories as well as extend the concept to other laboratory test services with minimal effort. The visual assessment of candidate laboratories in proximity could be further replaced by an automated proximity analysis, as described elsewhere.
Costing of the various tiers, including decentralised ITSDM, has been undertaken and published elsewhere
This article has demonstrated that a service coverage precinct approach could be used to identify coverage gaps for an ART-defined test or -relevant repertoire depending on the programmatic requirements that need to be addressed. The ITSDM approach could be used as a starting point to define best practices to introduce sustainable testing as part of a national service extending laboratory coverage to other chronic disease testing. This approach addresses coverage gaps in a cost effective manner.
The authors thank the staff of the NHLS corporate data warehouse and finance department, as well as Pedro da Silva, Natasha Gous, Somayya Sarang and Puleng Sheila Morokane of the NHLS National Priority Programme for their assistance. We would like to thank Dr Sergio Carmona for his expert advice on HIV viral load testing within the NHLS.
The authors declare that they have no financial or personal relationships which may have inappropriately influenced them in writing this article.
None.
D.K.G. supervised the study by providing leadership and oversight, and was also the project leader. N.C and L.M.C. designed the study, developed the methodology and conducted the research. N.C. developed the algorithm that was used to identify candidate NHLS laboratories. N.C and L.M.C conducted the data analysis and prepared the maps. D.K.G and W.S.S. provided editorial comments and technical input. All authors contributed to the manuscript development.