Changing malaria fever test positivity among paediatric admissions to Tororo district hospital, Uganda 2012–2019

Indoor Residual Spraying (IRS) has regained attention as a means to prevent malaria infection in high transmission areas of Africa when used in combination with insecticide-treated nets, largely as a result of funding made available by the US President’s Malaria initiative (PMI) [1]. Carbamates, organochlorines, organophosphates, pyrethroids, and more recently neonicotinoids, have been used in 23 countries in SSA since 1997, resulting in approximately 21 million people being protected by one form of IRS in 2017 [2].

In Uganda, between 1959–1964, IRS using dichloro-diphenyl-trichloro-ethane (DDT) with three rounds of mass drug administration (MDA) with single doses of chloroquine-pyrimethamine was piloted in Kigezi and parts of Masaka districts [3]. During the first year of the operations parasite prevalence declined from 16.6 to 0.3% [4]. Between 1963 and 1964, a large-scale field trial of Malathion was carried out in Masaka district reducing vector densities to 0.0011 from 66 per house per day within a year [5]. Following on from these successes, the Uganda malaria pre-eradication programme was established in 1964 as a joint Government of Uganda-WHO project [6]. However, the period between 1970 and 1990 was characterized by a general collapse of the country’s health system as a result of civil war and political unrest. During this period, IRS alongside other forms of malaria prevention were largely abandoned.

Lambda-cyhalothrin IRS was introduced in 2004 to tackle localized epidemics in Kabale, Kisoro and Kanungu districts. In 2005, Uganda became one of the first countries in sub-Saharan Africa to benefit from PMI funding, supporting a sustained IRS campaign in different parts of the country. From 2006, IRS with lambda-cyhalothrin was introduced to Kabale and Kanungu districts [7]. PMI provided their implementing partner over US$ 150 million between 2007–2012 to continue, and expand, IRS activities with lamda-cyhalothrin across districts in the north (Lamwo, Kitgum, Pader, Agogo, Kole, Oyam, Apac, Gulu, Amuru and Nwoya) and Kanungu [8, 9]. For a brief period in 2008, the Ministry of Health used DDT as part of IRS programs in Apac and Oyam districts. In Katakwi and Kumi, in the Teso region, between 2008–2009 lambda-cyhalothrin was used in combination with mass screening and treatment as a combined intervention aimed at district-wide elimination [10].

In 2009, both DDT and pyrethroid resistance were detected at IRS districts [11] and the IRS programs switched to spraying with carbamate (Bendiocarb) [12]. Epidemiological impacts were few, however, at Apac district a 45% reduction in community parasite prevalence in children under five years was attributed to IRS using DDT (2007–08) and Bendiocarb (2009–10) [12], and a 62% reduction in health service use for malaria morbidity was associated with Bendiocarb IRS [13].

Between April and October 2014, IRS was scaled down in the northern districts (Acholi sub-region) and switched to seven mid-northern districts (Lango sub-region: Alebtong, Amolator, Dokolo, Kabaremaido, Lira, Otuke) and eight eastern districts (Serere, Pallisa, Kibuku, Budaka, Namutumba, Butaleja, Bugiri, Tororo) initially using Bendiocarb and later switching to Actellic 300 CS (organophosphate pirimiphos-methyl). Cessation of IRS in the Northern Districts resulted in large-scale epidemics from April 2015 through to 2016 affecting over 1 million people [14]. At Apac, 18 months following cessation of IRS, malaria morbidity had returned to pre-IRS baseline levels [15].

Unlike the introduction of IRS during the late 1950s, the rapid expansion of IRS in Uganda from 2007 was not accompanied by a comprehensive, longitudinal epidemiological impact analysis plan across all districts where IRS was implemented [6]. There has been one notable exception involving detailed parasitological, clinical and entomological studies at Nagongera sub-County in Tororo district since 2007 [16‐22]. These studies provided conclusive evidence that with > 80% LLIN coverage and four rounds of IRS (coverage ranging between 85–100%), the incidence of malaria among children was reduced by 92% and the prevalence of infection by 93% [20, 22]. No studies have been undertaken to assess the impact of IRS on child survival or hospitalization in Tororo District. Here, temporal monthly data have been assembled from a hospital surveillance system at Tororo district hospital to examine the impact of vector control approaches on paediatric malaria infections admissions and their age patterns since 2012 to 2019.

The analysis of fever malaria test-positivity rates at Tororo District Hospital from January 2012 to December 2019 highlights significant declines associated with the introduction of IRS in the district (Table 1; Fig. 2). Prior to the introduction of Bendiocarb IRS in December 2014, during a period when the only form of vector control was routine or mass delivery of LLIN, the average TPR among febrile hospital admissions was 63% (Table 1). However, TPR declined during the periods of IRS use to 43% (Bendiocarb) and 31% (Actellic) (Table 1). Interrupted time-series analysis, controlling for age, rainfall patterns and type of test, shows that compared to the pre-IRS period, IRS (Bendiocarb and Actellic) with the second LLIN campaign period resulted in significant overall monthly reductions in TPR (Fig. 2). Importantly, in this setting of historically high transmission the results demonstrate the combined added value of IRS and LLIN, compared to LLIN alone, findings supported across different settings in Africa [18, 41, 42].

The findings presented here correspond with longitudinal studies of community infection prevalence 2007–2012 (Pre-IRS) and 2014–2017 (Post-IRS, without chemoprophylaxis) among a cohort of children aged 6–30 months in Nagongera sub-county, Tororo [20] and the dramatic reduction in the incidence of clinical malaria in the same cohort [21, 32]. A comparison with the patterns of out-patient TPR at Nagongera Health Centre [18, 43], and the hospital TPR (Table 1) showed identical patterns of change.

Throughout the time-series the mean age of paediatric febrile admission malaria test positives increased from 24 months in 2012–2013 to 46 months in 2019 (Fig. 3; Table 1; SI Table 1). Among test-positives during the baseline period (January 2012–October 2013), 29.5% were aged 1–11 months and 8.6% were aged 5–14 years, conversely during the last period of Actellic IRS (June 2016–December 2019), 12.5% were aged 1–11 months and 22.7% were 5–14 years (Table 1). Changing age-patterns of TPR have also been shown among febrile, paediatric out-patient attendees associated with sustained vector control elsewhere in Uganda [43] and Mozambique [44]. The majority of febrile admissions associated with malaria infection continue to be aged < 5 years (87%), however the changing age-pattern among young children suggests a possible impact of sustained, high coverage vector control on the acquisition of immunity among childhood populations [45].

Despite the entomological inoculation rate in Nagongera sub-county declining from 129 infective bites per person per year in 2013 to an apparent zero in 2017 [19], findings from this study suggest that transmission has clearly not been interrupted in the district. During the last period of Actellic IRS (June 2016 to December 2019), 31% of children with fever and tested for malaria presented to the district hospital still harboured malaria infections (Table 1). Limiting the data to all admissions aged 1–11 months, reflecting more recent community transmission in 2019, 90/857 (10.5%) were either slide or RDT positive. Malaria therefore continues to be associated with a substantial proportion of hospitalizations in the district, despite high coverage of LLIN and IRS using a currently efficacious residual organophosphate. The experimental addition of monthly chemoprophylaxis in Nagongera appears to accelerate a reduction in transmission when used in combination with IRS [17, 20, 22]. Similar findings have been observed when IRS is combined with Mass Drug Administration (MDA) in Zambia [46] and Mozambique [47]. Combinations of vector and chemoprevention of parasite reservoirs are likely to accelerate reductions in transmission [48]. However, the complete interruption of transmission in Uganda is as challenging today as it was during early experimental studies of IRS and drug-based interventions during the 1960s [3‐5]. Furthermore, monthly TPR values varied significantly during periods of IRS (Fig. 2), suggesting variations with time since spraying but importantly that any cessation of IRS in Tororo is likely to rapidly result in resurgence of infections among paediatric admissions. Uganda has ample evidence of the dramatic, immediate impact of removing intensive vector control though IRS on out-patient and in-patient hospital burdens [14, 15]. Relying only on moderate-to-high LLIN coverage alone will not mitigate this potential change in malaria burdens.

The WHO recommends that IRS or universal LLIN coverage should be selected as district-wide vector control, not both [49]. The available controlled trial evidence suggests little additional gains of combined LLIN and IRS as compared to the effective implementation of either approach independently [49]. However, in practice national malaria programmes, such as Uganda, continue with LLIN distribution in areas selected for IRS with the benefit of mitigating insecticide resistance [49]. The combination of IRS and LLINs had greater impact on TPR than LLIN used in isolation (Fig. 2; Table 1). This may have arisen because of LLIN and IRS campaigns used different classes of insecticides during a phase of emerging pyrethroid resistance [50, 51]; LLIN distributions have never achieved 100% coverage in Tororo [31], high bed net attrition rates [52], and those who own LLINs might not always use them [53]. Importantly, as shown in Uganda, sustaining IRS is heavily dependent on donor support, and when donors elect to change support for IRS, or the location, it is essential that these communities continue to be protected by LLIN.

The use of fever TPR has avoided the attribution of admissions to a specific admission diagnosis of malaria, in part because admissions have multiple causes, but largely because fever surveillance for infection is a much simpler, less ambiguous surveillance outcome. Unlike routine health information systems in Uganda which provide only crude age groupings above and below 5 years of age, it has been possible in the hospital analysis presented here to explore in more detail the precise age of test positives with time. However, long-term hospital surveillance suffers from several intrinsic problems that require contextualization, for example changing diagnostic protocols, access to diagnostics and patient access. Furthermore, TPR is less sensitive to measuring the true magnitude of the impact of interventions as it ranges from 0–100% and cannot represent incidence expressed by the number of people at risk of hospitalization. Defining incidence through routine data, affected by access and use of services, is complex [54] and the relationships between facility based TPR and incidence [55] and community prevalence [56] are non-linear. Nevertheless, routine surveillance of the public health impact of vector control will provide early warnings of a changing disease epidemiology. There is a growing body of evidence that suggests that the impact of IRS can be monitored through health facility and hospital data from routine health information systems in Uganda [13‐15, 43, 56‐62]. The continued surveillance of the health impact of IRS in Tororo is critical to understand possible further changes in TPR and the age-patterns of malaria admission.

The findings from this study have important policy implications regarding malaria control in Uganda. The combination of IRS with LLIN distribution has a larger impact than LLIN alone. Sustained vector control increases the mean age of febrile test-positive admissions to hospital a likely result of delayed immune acquisition. However, even after many years of IRS the majority of febrile test positive are still among young children. LLINs and IRS used in combination will not be enough to achieve elimination in Uganda. Combinations of interventions that target vectors and the parasite might be required to interrupt transmission including MDA and vaccines. Routine enhanced health facility-based malaria surveillance data offers a cost-effective means of monitoring the impact of malaria control interventions on malaria infection rates over-time. Expanding the adoption of a standardized MRF for paediatric admissions across the country by the Ministry of Health, and regular interrogation of hospital data will provide insights into intervention impacts, failures and an epidemiological understanding of the changing clinical patterns of disease.