Supporting surveillance capacity for antimicrobial resistance: Laboratory capacity strengthening for drug resistant infections in low and middle income countries

Desk based analysis and conceptual model

A desk-based analysis of capacity for AMR surveillance in Sub-Saharan Africa and South-east Asia was done, first by reviewing the WHO’s AMR report and then looking at those countries that had been able to contribute data, and identifying which components of the conceptual model they had put in place (or other) to do this. The challenges faced in these systems were also considered. Where there were few or no data in WHO sub-regions, examples of other surveillance systems that had been established, outside of the context of AMR surveillance were reviewed.

Case studies

Site visits were made to Malawi, Ethiopia and Vietnam, and capacity in these countries reviewed according to the components included in the conceptual model and the functions that would be needed to improve/develop AMR surveillance, as well as the strengths of the systems and models in place. This work involved extensive review of relevant guidelines and health system and policy literature, as well as discussions with in-country personnel, including laboratory staff, policy-makers, clinicians, microbiologists, veterinarians and epidemiologists.

Analytic framework

We identified the functions, sub-functions and activities needed for AMR surveillance at a national and site level, and defined four levels of AMR surveillance capacity involving increasing complexity. Capacity at Level 1 represents the most limited system, with incomplete clinical data, minimal clinical investigation and no laboratory testing. At level 2 there is some laboratory testing but no standardization of processes in place. Level 3 represents a core surveillance system with standardized processes, focusing on invasive disease in children under five years and the identification and antimicrobial susceptibility testing of WHO priority pathogens (Escherichia coli, Klebsiella pneumoniae, Staphylococcus aureus, Streptococcus pneumoniae, Acinetobacter baumannii, Salmonella spp., Shigella spp., Neisseria gonorrhoeae). Level 4 represents extended AMR surveillance capacity, with capacity to include AMR surveillance from all patients and all sample types. The four functions at the sentinel site level were 1) clinical admission assessment & investigation; 2) isolate identification and susceptibility testing; 3) isolate storage (local) & referral to reference laboratory; and 4) data system and review. At the national level these functions were 1) leadership; 2) training; 3) quality assurance; and 4) national reference laboratory. Each function was further broken down into sub-functions; for example “clinical admission assessment & investigation” has four sub-functions, namely i) clinical admission assessment; ii) clinical data; iii) clinical investigation; and iv) training and quality assurance, all describing the different activities relevant for clinical surveillance. Components are detailed in Table 1 and Table 2.

Costing

In each sub-function, the interventions, activities and resources needed to move from one level to the next were defined. The costs of establishing and implementing (start-up costs), as well as running (recurrent costs) of the described activities were included and recorded following an ingredients approach, where unit costs and quantities were required to calculate total costs for each input. This allowed the cost estimates to be adapted more easily to different country settings, by applying local unit costs. The cost results were presented by year, ranging from year 1 to year 5. Due to the different capacities and models used in the different sites that were visited, as well as logistical constraints, costs were assessed in a single highly functioning site in Kenya (KEMRI-Wellcome Trust Research Programme), where clinical, laboratory and demographic surveillance systems are integrated. We used site specific costs, for personnel, for subsequent specific country estimates. Data collection for the costing involved interviews with key personnel, review of key documents and observation of a functioning clinical, laboratory and demographic surveillance system. The approach to the costing was pragmatic, prioritising the most accurate recording of high cost items (with most influence on estimates of total cost). All information was collected and analysed in Excel 2016.

Desk based analysis

Capacity for AMR surveillance varies between sub-Saharan Africa and South-east Asia, and within countries in these regions. In the WHO African region (AFRO), a recent review reported only 2 of the 47 AFRO member states (Ethiopia and South Africa) had national AMR plans in place^5,10, and 7/47 members (Ethiopia, Ghana, Kenya, Lesotho, South Africa, Tanzania and Zimbabwe) had overarching national infection prevention and control (IPC) policies¹¹. Establishment of AMR surveillance is limited, but capacity is being developed, for example in 24 hospital laboratories in Ghana¹². In contrast, South-East Asian countries have, are developing, or are initiating AMR surveillance systems, for example, Thailand, Vietnam, and Cambodia¹³, respectively. This is in part supported by the more advanced development of networks and collaborations for AMR surveillance. Networks and examples of AMR surveillance and stewardship programmes are described in detail in Supplementary File 1, as well as surveillance of other infectious diseases. AMR surveillance strategies could also learn from and benefit from recent laboratory strengthening efforts as an important element of the work against the Ebola epidemic in West Africa since 2014 (Supplementary File 1: Box 3) and for specific diseases, for example rotavirus (Supplementary File 1: Box 4)¹⁴.

Emerging themes in strengthening laboratory capacity for AMR surveillance included:

1. Leadership

Leadership, and the national policy addressing AMR, as well as international collaboration, are necessary to strengthen AMR surveillance in LMIC countries, to ensure an enabling policy environment. However expertise in microbiology is limited in these settings, and, whilst in the longer term this capacity is important to develop, in the interim leaders may need to be drawn from clinical infectious diseases and public health. Leaders are important for advocacy/championing, and for local credibility of the surveillance effort, as well as providing the focal point for activities.

2. Training

Where health systems are weak and still developing, there is a need to develop human resources across cadres of staff: clinical, laboratory and data management. Well-trained staff are likely to move elsewhere for better salaries; it is important that there is sufficient budget to ensure staff can be recruited and retained.

3. Laboratory quality assurance

In terms of strengthening laboratory capacity, it is clear that inadequate laboratory infrastructure limits the quality and the ability to reliably detect pathogens and conduct antimicrobial susceptibility testing.

There are different options for models of AMR surveillance (outlined in Supplementary File 1 and summarised in Supplementary File 1: Table 1). Overall, a model with active, continuous, comprehensive integrated population and laboratory disease surveillance would provide the most robust, comprehensive data. However, within the constraints of resources, an appropriately designed sentinel site system would also be appropriate. Whilst it would be feasible to assess AMR outside of a health care context, for example through cross-sectional studies of population colonisation (as part of a population based survey), these would not include surveillance of the most serious drug resistant infections, and fall outside the remit of public health surveillance as observational research studies requiring appropriate ethical approvals and individual informed consent.

Laboratory based surveillance identifies “a resistant organism”, but without clinical or population data cannot be fully interpreted and inform policy. To achieve the aim of AMR surveillance, and inform interventions, data are needed on:

Within the different models of surveillance, the functions are broadly consistent, and can be split into those provided at a national level and those provided at one or more sites, depending on capacity and the model in place, and including the themes identified above. At a national level, there is a need for leadership to support training, and quality across the system. A national AMR laboratory can support development of site laboratories and conduct more extensive testing (thus focussing resources, Supplementary File 1: Table 3, Figure 1). At a site level an integrated model includes development of laboratory capacity (isolate identification and susceptibility testing), clinical surveillance (systematic investigation of patients according to diagnosis), health care utilisation surveys (to assess representativeness of the health facility data) and census or enumeration data to determine the population catchment (Supplementary File 1: Table 2, Figure 1). These functions were used to cost AMR surveillance (Supplementary File 1: Tables 2 and 3) and consider how they were being achieved in the different models and contexts of the country case studies.

Country case studies

Full site reports are included in Supplementary File 1, the findings for each country are summarised here.

Malawi. Malawi faces substantial challenges due to extremely limited country resources, which reduces health system capacity. However, whilst it does not have the capacity for a comprehensive AMR surveillance system, it has been developing institutional infrastructure and policy for AMR surveillance. It also has considerable capacity through international collaboration and an academic centre of excellence (Malawi-Liverpool Wellcome Trust (MLW)) undertaking microbiological surveillance as part of a research programme. In order to take advantage of available research infrastructure i.e. sentinel surveillance capacity beyond Queen Elizabeth Central Hospital (QECH), the following should be considered for new surveillance sites in Malawi:

Ethiopia. Ethiopia is developing a system for AMR surveillance that includes a national reference laboratory and sentinel regional laboratories across the country reporting data (continuous, passive surveillance). At present, whilst capacity is in development, there are some hospitals where the standard of clinical examination is adequate for surveillance, but there is limited microbiology laboratory capacity for processing specimens for culture, particularly blood cultures. Record keeping is not formalized or electronic, and there is potential for improved communication between the ward and the laboratory to link microbiology data to patients. In strengthening capacity for AMR surveillance, the following should be considered:

Vietnam. Vietnam has established national infrastructure for AMR and has worked with regional networks and international collaborators to develop a surveillance system, including a national reference laboratory and twenty laboratories attached to district hospitals. The hospitals are public and broadly representative, there are functioning microbiology laboratories and the physical, legal and ethical capacity to link these records for anonymised aggregation at individual case level. In strengthening capacity for AMR surveillance, the following should be considered:

Costing

We developed a flexible model, based on an assumption that each country would have a number of sentinel sites with clinical and laboratory surveillance taking place in each site. We made an assumption that a country would aim for one surveillance facility per five million population, with a minimum of five surveillance facilities per country. However, cost results are presented for varying number of sites per country (2, 5 and 10), rather than for the target number of sites, with a focus on costing inputs required to achieve surveillance at level 3 (a core level) and level 4 (an extended level).

For a sentinel site (Table 1), each sub-function was broken down into all input activities and resources required to achieve the different levels of surveillance sophistication (core and extended) and summarized under “interventions” below each sub-function in clinical surveillance, laboratory surveillance, isolate storage and transport, data system and review (Supplementary File 1: Tables 4A–D). A number of inputs were costed independently of the sub-function framework, for example personnel costs required for laboratory surveillance or the cost of setting-up a laboratory including the cost of building and equipment. They are presented directly as inputs. The additional demographic component to surveillance was costed as an initial and final population enumeration round in year 1 and 5. Inputs of activities and resources required at the national surveillance level are shown in Table 2. The inputs (Supplementary File 1: Tables 4A–D and 5) were used to cost the surveillance components by determining quantities and unit costs at different levels of surveillance sophistication.

Costs depend on existing capacity, number of sentinel sites proposed for surveillance, and, across countries, variability in costs of personnel. Illustrations are given for Kenya and Vietnam according to different numbers of sites (Supplementary File 1: Figures 7 and 8) across clinical, laboratory, national and demographic surveillance, as well as data system & review functions.

The largest contributor to the total cost of AMR surveillance is laboratory surveillance at the sentinel site (including isolate storage and transport to national reference laboratory), together with the national reference laboratory cost. These laboratory costs range, depending on number of sites and country, between 67% and 77% of the total cost at level 3 and between 78% and 85% at level 4. Within the laboratory cost the largest contributors to the total costs are the initial set up cost of the laboratory, and personnel. The increase in cost from core laboratory functions to extended laboratory functions is substantial (Supplementary File 1: Figures 6 and 7) and is driven by the increase in equipment and staff to support the increase in samples being processed (all sample types vs blood cultures), and isolates tested (all isolates vs WHO priority pathogens (excluding Neisseria gonorrhoeae, omitted as frequently tested for in outpatient settings with swabs)).

Within the overall envelope costed for AMR surveillance, the costs for clinical surveillance, data system and demographic surveillance are a much smaller proportion of total cost than laboratory surveillance. This is partly because for the clinical elements the costs of clinical staff providing care were not included, as in-post, except where additional clinical staff would be needed to undertake specific activities needed to support surveillance. In contrast, laboratory personnel costs were all included in the costing. However, this likely reflects the real-world situation, in terms of the need for recruitment of laboratory staff.