Assessing the impact of regional laboratory networks in East and West Africa on national health security capacities

National laboratories are a fundamental capacity for public health, contributing to disease surveillance and outbreak response. The establishment of regional laboratory networks has been posited as a means of improving health security across multiple countries. Our study objective was to assess whether membership in regional laboratory networks in Africa has an effect on national health security capacities and outbreak response. We conducted a literature review to select regional laboratory networks in the Eastern and Western African regions. We examined data from the World Health Organization Joint External Evaluation (JEE) mission reports, the 2018 WHO States Parties Annual Report (SPAR), and the 2019 Global Health Security Index (GHS). We compared the average scores of countries that are members of a regional laboratory network to those that are not. We also assessed country-level diagnostic and testing indicators during the COVID-19 pandemic. We found no significant differences in any of the selected health security metrics for member versus non-member countries of the either the East Africa Public Health Laboratory Networking Project (EAPHLNP) in the Eastern Africa region, nor for the West African Network of Clinical Laboratories (RESAOLAB) in the Western Africa region. No statistically significant differences were observed in COVID-19 testing rates in either region. Small sample sizes and the inherent heterogeneities in governance, health, and other factors between countries within and between regions limited all analyses. These results suggest potential benefit in setting baseline capacity for network inclusion and developing regional metrics for measuring network impact, but also beyond national health security capacities, other effects that may be required to justify continued support for regional laboratory networks.

Introduction countries that are all Member States of the African Union; (2) have three or more countries in the network; (3) be located entirely within a specific region of Africa or contiguous to it; (4) focus on general diagnostics, surveillance, and other laboratory work rather than on specific diseases; and (5) be operational with publicly available information on specific projects as of October 2019.
We identified two networks that met these criteria: the West African Network of Clinical Laboratories (RESAOLAB), and the East Africa Public Health Laboratory Networking Project (EAPHLNP). RESAOLAB is described as the first regional network in West Africa to focus on creating a system of biological laboratories offering quality services, and to improve clinical diagnosis for public health aims [28]. EAPHLNP's objective is to "to establish a network of efficient, high quality, accessible public health laboratories for the diagnosis and surveillance of Tuberculosis and other communicable diseases" [29]. To this end, both networks are selfdescribed as having goals that are closely aligned with global health security outcomes. Fig 1 shows the breakdown of member and non-member countries for the West African region (RESAOLAB members and non-members) and the East African region (EAPHLNP members and non-members). Non-member countries for inclusion were defined per the African Union regions; [30] however, island nations were excluded given the lack of geographical continuity with other countries in the respective regions. For the purposes of additional analyses, scores were also collated within the RESAOLAB members based on the year of membership in the network (2009 or 2013) (Fig 1).

Health security metrics data collection and analysis
Three tools for measuring national health security capacity were used: the JEE [4], SPAR [31], and the Global Health Security Index (GHS Index). The GHS Index is a composite metric that combines elements of the JEE with indicators related to health systems, compliance with international norms, and risk environment [32].
The scores of member countries within each network, as well as non-member countries within the same region of Africa, were collated for: 1) the 48 JEE indicators of the first edition JEE tool; 2) the 13 SPAR indicators and the three C5 Laboratory sub-indicators for 2018; and 3) the 34 GHS Index indicators and the three 2.1 Laboratory Systems sub-indicators for 2019.
The JEE indicators were additionally sorted into four categories: 1) the four National Laboratory System indicators (D.1.1-D.1.4); 2) laboratory-related indicators, which were identified in the literature review as potentially being influenced by laboratory networks or where laboratories were cited in the JEE indicator description; 3) non-laboratory indicators, which made no mention of laboratories in either the JEE indicator description or guiding questions; and 4) other indicators, which included minimal mention of laboratories or laboratory services in the JEE guiding questions or only referenced non-biological laboratories, and which were discarded from further analysis to reduce the number of multiple comparisons. The inclusion of the non-laboratory indicators was used to establish whether there were differences in overall health security capacities between groups.
Differences in the National Laboratory Systems, laboratory-related, and non-laboratory JEE indicator scores of network member countries versus non-member countries were calculated using the Goodman-Kruskal (G-K) Gamma statistic. See Supplemental Material for further description of this statistic and corresponding derivation of p-values.
Differences in the median 2018 SPAR and 2019 GHS Index scores of network member countries versus non-member countries were calculated using Kruskal-Wallis tests, with adjusted p-values for multiple comparisons. The standard deviation of each average score was also calculated, and the data used to calculate each average score were checked for outliers. Data points were considered outliers if they were more than two standard deviations away from the mean. Sample sizes were insufficient for comparisons between countries that joined RESAOLAB in 2009 versus 2013 and non-members, across all metrics.

COVID-19 testing data collection and analysis
We also sought to assess whether membership in a regional network might affect laboratory capacities in practice, as opposed to "on paper". To do so, we collated data provided to the Africa Centres for Disease Control and Prevention (Africa CDC) from Member States on: 1) the date at which testing capability was established in the country; 2) the number of reported severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) tests performed in the country per day; and 3) the incidence per 100,000 population of laboratory-confirmed COVID-19 cases per country. These data were reported publicly via the Africa CDC COVID-19 Dashboard [33]. To assess the effect of laboratory membership on testing capacity, we fit negative-binomial mixed effects models with number of daily tests as the outcome and laboratory membership as the primary covariate of interest. Further details on these models are available as a supplemental document (S1 Text).

Analysis of health security metrics
The analysis of JEE scores between EAPHLNP member countries and those Eastern Africa region countries outside the network showed no significant differences across any of the analyzed indicators (Table 1). Likewise, the comparison of RESAOLAB members to non-members revealed no statistically significant differences ( Table 2). Apart from network members showing higher, albeit not significantly, average score values for National Laboratory System indicators in both regions, there were no discernible patterns with respect to scores across the other analyzed indicators.
All three SPAR and all three GHS Index laboratory indicators for both EAPHLNP and RESAOLAB showed no statistically significant difference against their respective non-members (Figs 2 and 3). There was no consistent pattern, and no significant differences, between EAPHLNP or RESAOLAB members and non-members for the non-laboratory JEE, SPAR and GHS Index indicators (Tables 1 and 2, S1 and S2 Tables). Ethiopia, not an EAPHLNP member, was a high outlier across three 2019 GHS Index indicators (Biosafety [1.4], IHR Reporting Compliance and Disaster Risk Reduction [5.1], and Cross-Border Agreements on Public and Animal Health Emergency Response [5.2]), scoring higher than other non-member countries in East Africa. Nigeria, a non-RESAOLAB member, was a high outlier for two 2018 SPAR indicators (Legislation and Financing [C1] and Food Safety [C4]) and one GHS Index indicator (Biosecurity [1.3]), scoring higher than other non-network members in the Western Africa region. Table 1. JEE scores and score comparisons for the Eastern Africa region countries under consideration. The score difference was calculated as the member country average indicator score minus the non-member country average indicator score. Comparisons of the ordinal scores were made using the Goodman-Kruskal (G-K) gamma statistic; significance was tested using permutation-based, Benjamini-Hochberg (B-H) adjusted p-values. Results with significant p-values (at a 95% confidence level) are shown in bold.   Our model showed no impact of laboratory network membership on testing rate while controlling for daily incidence of reported COVID-19 cases and time since the first reported test in both networks (S3 Table). The EAPHLNP member countries reported more tests conducted than non-member countries, but the difference did not reach statistical significance. There was an insufficient number of countries with testing capability established prior to the COVID-19 pandemic to include this as a variable in the model.

Discussion
The lack of statistically significant differences in any of the analyzed health security framework indicators suggests a number of possible interpretations. Despite member countries having numerically higher average scores across most laboratory-related indicators across all three metrics, the small sample sizes limited the power of our statistical analyses, in both regions. This suggests that other approaches that employ qualitative methods might be better suited to understanding potential differences with small samples, in order to further understand the impact of networks in these regions. The small sample sizes also may have overrepresented  outliers. Non-network members, Nigeria and Ethiopia, in the Western and Eastern Africa regions respectively, were both high outliers across different SPAR or GHS Index indicators. Neither are network members, but both were priority countries for United States Global Health Security Agenda (GHSA) assistance, of which laboratory capacity strengthening was a core activity. Other countries in the two regions did also benefit from GHSA investment, and may have received other types of bilateral or multilateral assistance. Future analyses of the effect of laboratory networks on national capacities should therefore attempt to consider largescale assistance and cooperation programs as potential competing or confounding variables.
Our analysis of COVID-19 testing data showed no statistically significant relationship between network membership and testing rates. The analysis was limited by focusing only on COVID-19 testing data. A more robust test of national laboratory network resilience would have also measured diagnostic activity for non-COVID suspected specimens (i.e. HIV viral load testing), to ensure that the scale-up of COVID-19 testing did not come at the expense of other laboratory activities. Similarly, regional entities, including the Africa CDC, have provided substantial targeted support for various aspects of COVID-19 diagnostic testing since the literature review for this paper took place, as have other bilateral and multilateral donors. Africa CDC support has included the provision of training, supplies, reagents, and web-based resources, any of which could have influenced testing rates independent of network membership [34,35]. Future studies could also consider analyzing notifications and alerts related to public health events prior to COVID-19, to identify a potential link between laboratory reactiveness and network membership; however, databases of outbreak alerts have known limitations, which would need to be addressed before a comprehensive analysis of this sort could be undertaken [36].
Our results provide no evidence of a beneficial effect of regional laboratory networks on national level laboratory-related health security capacities, at least based on the metrics analyzed here. It is worth highlighting that we did not attempt to analyze the existence of baseline capacity inclusion criteria or the success of the networks, based on their own stated goals and objectives; our analysis was limited to the potential influence of network membership on national health security indicators. We also did not seek to examine potential regional level benefits; to our knowledge, such regional level indicators do not yet exist within the context of global health security frameworks. The indicators used by RESAOLAB and EAPHLNP themselves were largely applied at the national, rather than regional, level, but could constitute a starting point for development of relevant regional indicators for laboratory systems, for example with respect to proportion of disease events that receive confirmatory testing, irrespective of whether the test is performed within the same country or another country in the network [37,38].
There are numerous examples of efforts to establish regional laboratory networks from regions outside of Africa [39][40][41]. Like RESAOLAB and EAPHLNP, these networks have received widespread interest and promote open communication and data sharing [42][43][44]. Despite this, to our knowledge, there have been no analyses to date to demonstrate or quantify the potential benefits of membership in these types of networks in terms of national compliance with health security frameworks, or regional capabilities. To this end, our research adds to this area of scholarship from a methodological perspective and suggests there may be opportunities, especially in world regions where multi-country political or economic coalitions are already in place and functional, to complement existing nationally-focused health security metrics with indicators that capture capacities at a regional level as well.
The ongoing COVID-19 pandemic has highlighted the importance of improving core capacities, like laboratory diagnostics, in health security systems [45]. The pandemic has emphasized the need for scalable, rapidly-activated diagnostic testing platforms, strengthened global reagent supply chains, and sufficient trained laboratory personnel in countries worldwide. While there is an ongoing debate about the value of metrics such as JEE, SPAR, and GHS Index scores in predicting response outcomes at national levels [46][47][48][49], there may yet be further potential in leveraging regional initiatives to support national health security capacities [50]. In Africa for example, additional laboratory networks are being established to strengthen health security like the Institute of Pathogen Genomics (IPG) laboratory network. IPG operates through Africa CDC's Regional Coordinating Centers and its Regional Integrated Surveillance and Laboratory Network (RISLNET) [51] and has already shown great success in establishing regional and continental genomics capacity for COVID-19 that can be leveraged for future pathogen discovery [52][53][54]. Additionally, Africa CDC and the African Union has called for African countries to pool resources to procure and distribute COVID-19 diagnostic tests, allowing for greater national-level access to testing [55]. This kind of sharing of resources can be further facilitated through regional health security systems, including regional laboratory networks such as RESAOLAB and EAPHLNP. Such benefits could be captured through new regional health security metrics, as described above, as a means of more accurately characterizing the benefits of regional networks.
Our study had numerous limitations. It cannot be ascertained whether the scores on laboratory-related indicators reflected membership in the regional networks, a self-selection bias, legacies of pre-membership laboratory capacity status, one or more with unmeasured factors, or a confounding factor. Individual country-level variation in size, political attention or priority given to health security, access to resources, and other bilateral and multilateral relationships and partnerships outside of the networks may influence laboratory capacity, as well as the ability to benefit from networks. Numerous commentators have questioned the value of health security metrics as measures of laboratory and other essential public health capacities. While we attempted to address this limitation by including testing rate data, numerous other factors could have contributed to testing capacity. These include domestic attention to testing as a core response strategy and bilateral or multilateral support from non-network sources.

Conclusion
Our study showed no statistically significant benefits of membership in regional laboratory networks on national-level laboratory capacities, in either studied region, nor on any of the other non-laboratory health security indicators analyzed. The lack of observed benefit across these indicators may in part reflect imprecision in the original indicators, the methodological approach, or shortcomings of indicators as a proxy for capacity. Taken at face value, it could be suggested that national health security capacities are, as a stand-alone outcome, insufficiently addressed by these regional networks. However, our analysis did not directly consider network inclusion criteria or other stated goals and outcomes of the selected networks; for example, improved quality of services, which could itself justify continued support for regional laboratory networks. Moreover, there may be opportunities to focus on development of regional capacity indicators as a promising and important area for future exploration.
Supporting information S1 Table. Table comparing member Table. Table comparing member