Rapid shifts in the age-specific burden of malaria following successful control interventions in four regions of Uganda

doi:10.21203/rs.2.17709/v2

Background: Malaria control using long-lasting insecticidal nets (LLINs) and indoor residual spraying of insecticide (IRS) has been associated with reduced transmission throughout Africa. However, the impact of transmission reduction on the age distribution of malaria cases remains unclear.

Methods: Over a 10-year period (January, 2009 to July, 2018), outpatient surveillance data from four health facilities in Uganda were used to estimate the impact of control interventions on temporal changes in the age distribution of malaria cases using multinomial regression. Interventions included mass distribution of LLINs at all sites and IRS at two sites.

Results: Overall, 896,550 patient visits were included in the study; 211,632 aged <5 years, 171,166 aged 5-15 years, and 513,752 >15 years. Over time, the age distribution of patients not suspected of malaria and those malaria negative either declined or remained the same across all sites. In contrast, the age distribution of suspected and confirmed malaria cases increased across all four sites. In the two LLINs-only sites, the proportion of malaria cases in <5 years decreased from 31% to 16% and 35% to 25%, respectively. In the two sites receiving LLINs plus IRS, these proportions decreased from 58% to 30% and 64% to 47%, respectively. Similarly, in the LLINs-only sites, the proportion of malaria cases >15 years increased from 40% to 61% and 29% to 39%, respectively. In the sites receiving LLINs plus IRS, these proportions increased from 19% to 44% and 18% to 31%, respectively.

Conclusions: These findings demonstrate a shift in the burden of malaria from younger to older individuals following implementation of successful control interventions, which has important implications for malaria prevention, surveillance, case management and control strategies.

Infectious Diseases

Malaria

routine surveillance

age distribution

reduced transmission

burden shift

In Africa, the burden of malaria has decreased significantly, primarily through the scale-up of vector control interventions, including long-lasting insecticidal nets (LLINs), and indoor residual spraying of insecticide (IRS) [1-3]. These interventions were coupled with improved case management using artemisinin-based combination therapies (ACTs), and intermittent preventive therapy with sulfadoxine-pyrimethamine for pregnant women [4, 5]. Whereas the impact of malaria control interventions is generally measured in terms of changes in P. falciparum infection prevalence and case numbers [1], more evidence is needed on how interventions influence the age distribution of malaria cases - a vital marker of progress in malaria control [6]. In high transmission settings, younger children bear the brunt of the malaria burden [7, 8], particularly for severe malaria and malaria deaths [9, 10]. However, it is unclear how quickly and to what extent the age distribution of uncomplicated malaria cases may shift with changes in transmission, following the successful implementation of control interventions.

Malaria surveillance efforts have generally focused on children under 5 years of age - a group that contributes the majority of reported cases [11]. School-aged children (5-15years) have received less attention due to the lower morbidity and occurrence of severe outcomes in this group [12]. However, older children often experience low density asymptomatic infections and have been identified as important contributors to the infection reservoir for onward transmission [13]. Even less attention is given to adults (over 15 years), except for pregnant women [14], despite documented high prevalence of asymptomatic infections with high parasitemia in adults [15, 16].

Commitments to malaria control made at the Abuja summit in 2000 [17] led to World Health Organization (WHO) recommendations of universal coverage with LLINs, beginning in 2007 [18]. In Uganda, access to LLINs increased gradually, first among pregnant women and children under 5 years of age [19, 20]. This later culminated in universal LLIN campaigns aimed at one LLIN for every two household residents [21], with nationwide distributions in 2013-14 [22] and 2017-18. Furthermore, IRS was implemented in ten districts effective 2010, then shifted to 14 districts from 2014 onwards [23, 24]. In northern Uganda, integrated community case management (iCCM) of malaria began from 2016-17 and the ten districts plus one also resumed IRS [25].

Following the scale-up of malaria control efforts in Uganda, malaria cases reported from health facilities declined by an average of 15.5% per year between 2013 and 2017 [26]. Moreover, malaria indicator surveys (MIS) showed that parasite prevalence in children under 5 years of age declined from 42% in 2009 to 19% in 2014-15 [27, 28], whilst malaria mortality reportedly decreased from 59 to 23 deaths per 100,000 between 2010 and 2017 [25].

This study aimed to investigate the impact of control interventions on the age distribution of malaria cases, using high-quality malaria surveillance data from four sites in Uganda, where mass LLIN distribution was conducted at all four sites and IRS implemented at two, one of which also received iCCM.

Malaria surveillance

The National Malaria Control Division (NMCD) of the Ministry of Health has conducted surveillance through the Health management information system (HMIS) since 2007. However, more detailed malaria surveillance including collection of individual-level data has been conducted in selected malaria reference centres (MRCs) since 2006. MRCs are level III or IV health facilities located across the country, representing varied transmission settings as previously described [29, 30]. In summary, patients visiting these centres are assessed at the outpatient department (OPD) and basic information recorded using an OPD registry including history of fever, age, gender, malaria diagnostic test results, and treatments prescribed. Individual visit-level OPD records are then entered into an MS Access database (Microsoft Corporation, Redmond, WA) and sent to the Uganda Malaria Surveillance Program (UMSP) data centre and cleaned before transfer to STATA (Stata Corporation, College Station, TX) for analysis. Data from these sites is used to generate monthly reports that are shared and reviewed by the National Malaria Control Division and other stakeholders.

Study sites

Four MRCs were selected for this study including: Walukuba in Jinja District, Kasambya in Mubende District, Aduku in Apac District, and Nagongera in Tororo Distict (Figure 1). These sites were purposively selected because of their temporal representation of the malaria indicators of interest, with data covering the period of interest (January 2009 and July 2018), and have had malaria control intervention activities implemented in each of them. By transmission settings, Walukuba and Kasambya had an estimated annual entomological inoculation rate (aEIR) of less than 10 infective bites per person per year in the early 2000’s, while in Aduku and Nagongera aEIR was in excess of 1,500 and 550, respectively [31].

Variable description

We classified records in the database either as “suspected malaria” or “no suspected malaria”.

Suspected malaria was defined as patients who: a) had a laboratory test done for malaria (microscopy or rapid diagnostic test (RDT)) or b) were given a clinical diagnosis of malaria in the absence of laboratory testing. All records that did not meet this definition were classified as “no suspected malaria”.

Among suspected malaria cases, “malaria cases” were defined as all those patients with positive malaria diagnostic test results (microscopy or RDT). Moreover, “malaria negative cases” were those who were tested for malaria (microscopy or RDT) but had negative test results.

Description of malaria control interventions at the study sites

Using available information from the NMCD on when specific interventions were implemented and/or interrupted at each study site, we divided calendar time into three to four different intervention periods per site (Figure 2). The first period for each site, before large-scale interventions were implemented, is referred to as baseline. During baseline periods, control activities were largely limited to targeted distribution of insecticide treated nets (ITNs) to vulnerable populations such as children under 5 years of age and pregnant mothers [32, 33]. We then used the subsequent intervention periods to quantify the impact of interventions on the age-distribution of test-confirmed malaria cases.

The main control interventions implemented in Walukuba and Kasambya were two mass LLIN distribution campaigns conducted in 2013 and 2017 at both sites. In addition to LLINs in the same time periods, however, Aduku and Nagongera also received IRS. The baseline period in Aduku included one round of IRS with the pyrethroid alphacypermethrin insecticide, which was not considered to be effective due to insecticide resistance [30]. This period in Aduku, was followed by nine rounds of IRS with the carbamate bendiocarb insecticide approximately every six months, as well as the first mass LLIN distribution campaign. IRS was stopped for three years and then a single round of IRS with the organophosphate Actellic insecticide was conducted, immediately followed by the second mass LLIN distribution campaign. In Nagongera, the baseline period was followed by the first mass LLIN distribution campaign and then a sustained period of IRS including three rounds with the bendiocarb insecticide, approximately every 6 months, followed by three rounds with Actellic approximately every 12 months. A second mass LLIN distribution campaign was also conducted in Nagongera during the sustained period of IRS [34].

Statistical analysis

Data from patients with age missing (0.3%), age over 70 years (1.2%), and follow-up visits for any previously recorded illness episode (0.7%) were excluded from the analyses. We first explored the data by calculating, for each site and intervention period, the total number of patients seen and the proportion of those that were suspected, tested and classified as confirmed malaria cases. To characterize changes in testing practices, we also calculated the proportion of cases that were tested by microscopy versus RDT. We defined test positivity rate (TPR) as the proportion of patients tested for malaria that tested positive, a metric often considered a viable proxy for transmission intensity [35].

We also explored trends in the sex distribution of patients with or without suspected malaria and malaria cases across intervention periods using cross-tabulation with Chi-squared tests.

To characterize the impact of control interventions on the age distribution of malaria cases, we first used violin density plots to visualize changes over the intervention periods and compared the distributions of patients not suspected of malaria to those with confirmed malaria. We also used scatter plots of age (as a continuous variable) and test positivity, stratified by intervention periods, per site. We then fit multinomial logistic regression models, by site. The outcome in these models was ‘the age category of malaria cases’ (under 5 years, 5-15 years, and over 15 years) conveniently defined, while the main predictor was the intervention period. Models were adjusted for diagnostic test used (microscopy vs. RDT), as well as patient gender. To validate the potential impact of increasing use of RDTs across sites over time on these findings, we fit the multinomial regression models with an interaction between diagnostic test and time. Using the full models, we predicted the adjusted proportions of malaria cases in each of the age categories per intervention period. These marginal predictions were made at the mean values of the variables included in the model and analyses performed in R [36] and STATA 15 (Stata Corporation 2017, College Station, TX).

Patient composition and attendance by site

Between January, 2009 and July, 2018, the four health facilities of Walukuba, Kasambya, Aduku, and Nagongera recorded 896,550 patient visits in their outpatient department clinic registers, with over half of them (53%) suspected to be malaria cases. Walukuba recorded the highest number of both patient attendance (323,856) and suspected malaria cases (130,296), while Kasambya had the lowest attendance at 153,811 (Table 1).

The highest annual mean number of patients seen was in Walukuba (33,053) followed by Nagongera (22,701), then Aduku (20,079), and least in Kasambya (15,897). Mean monthly patient attendance per year remained fairly constant at all sites except Walukuba where this value peaked in 2011 at 3,400 and steadily declined to 1,952 by 2018 (Figure 1, additional file1). Mean monthly attendance of patients not suspected of malaria per year increased slowly over time at all sites with cyclic variations (Figure 2, additional file1). From 2009 to 2018, these increases were significant in Kasambya and Aduku though not in Walukuba or Nagongera by Wilcoxon rank-sum test (Table 1, additional file1). Conversely, mean monthly attendance of suspected malaria cases per year followed a general decline over time at all sites with cyclic variations (Figure 3, additional file1). From 2009 to 2018, these declines were significant in Walukuba, Aduku, and Nagongera, but not in Kasambya (Table 1, additional file1).

The majority of non-suspected and suspected malaria cases were female (67% and 63%, respectively). The difference between gender among confirmed malaria cases was smallest among children under 5 years of age (percent female: Aduku 49%, Nagongera 49%, Walukuba 54% and Kasambya 55%) and largest among patients over 15 years of age (percent female: Aduku 79%, Nagongera 75%, Walukuba 63% and Kasambya 66%). No observable trend in gender overall was found across intervention periods.

Trends in diagnostic testing over time

Throughout our study period, the majority of laboratory testing for malaria was by microscopy (89%) and the highest overall testing rate (percentage of suspected malaria cases that received a diagnostic test for malaria) was observed in Walukuba (97%). Across intervention periods however, the proportion tested by RDT increased at all four sites mostly in the last four years of study duration. By the last period, 77% of cases were tested by RDT in Kasambya, 72% in Aduku, 40% in Nagongera, and 25% in Walukuba (Figure 4, additional file1).

Test positivity rates

Although Aduku and Nagongera were historically the highest transmission settings, the highest TPR was observed in Aduku followed by Kasambya, Nagongera and Walukuba (Table 1). When considering only children under 5 years of age, however, baseline TPR levels were reflective of the historical transmission intensities. Baseline TPR in this group was highest at Nagongera (64%) and Aduku (63%), and lower in Kasambya (37%) and Walukuba (31%).

In all four sites, control interventions were associated with moderate reduction in overall TPR with acyclic secular trends in between. Larger reductions were observed in the two sites where both LLINs and IRS were implemented (Figure 2). In these sites, the declining trend in TPR is consistent except in Aduku, during the three years of withdrawn IRS, characterized by a sharp increase. Between baseline and the last intervention periods, TPR declined in Aduku from 56% to 27% and in Nagongera from 35% to 16%. In the two sites that received LLINs only, a similarly (with acyclic secular but less notably) reducing trend was observed. Between baseline and the last intervention periods, overall TPR decreased in Walukuba from 33% to 29% and in Kasambya from 39% to 30% (Table 1).

Over time and in all four sites, test positivity was seen to decline among the younger children while increasing among older participants. In all sites, we observed a shift in the peak age of test positivity from the youngest to the older ages, between baseline and last intervention period (Figures 5 and 6, additional file1). Interestingly, this pattern was reversed in Aduku during the three years when IRS was withdrawn, further confirming the effect of control interventions on test positivity with age (Figures 6, additional file1).

Differences in age distribution of malaria cases between sites at baseline

Although the duration of baseline periods varied between the sites due to the different timing of intervention activities, the age distribution of patients not suspected of malaria was very similar between all four sites at baseline, with median ages ranging from 23 to 25 years (Figure 3). In contrast, the age-distributions of malaria cases varied significantly between sites. These distributions were similar between the highest transmission sites of Nagongera and Aduku, with median ages of 2 and 3 years respectively. The distributions were also similar, but higher, between the lower transmission sites, with median ages of 8 and 11 years for Kasambya and Walukuba respectively.

Consistent with unadjusted analyses, results from the final adjusted multinomial regression models (adjusting for diagnostic test and gender of patient) showed that the proportion of malaria cases per age-group were significantly different between sites at baseline. The majority of malaria cases were among children under 5 years of age in the highest transmission sites (Aduku 58% and Nagongera 64%), while the highest proportion of malaria cases was among patients 5-15 years of age in Kasambya (35%), and over 15 years of age in Walukuba (31%) (Figure 4).

Changes in age distribution of non-suspected, test-negative, and laboratory confirmed malaria cases over time

The age of patients not suspected of malaria decreased slightly over the study duration at all four sites. Moreover, for malaria negative patients, the age distribution slightly shifted downwards and then upwards at all sites except Nagongera, where it shifted downwards across the intervention periods. In contrast, the age distribution of patients with laboratory confirmed malaria shifted upwards over time at all four sites. Comparing the last observation period to baseline, the median age of patients with malaria increased from 8 (IQR: 2.5 – 19) to 11 (IQR: 5 – 21) in Kasambya; 11 (IQR: 3.5 – 24) to 22 (IQR: 8 – 32) in Walukuba; 2 (IQR: 1.1 – 10) to 6 (IQR: 2 – 18) in Nagongera; and 3 (IQR: 1.2 – 13) to 14 (IQR: 5 – 22) in Aduku (Figure 3).

Across all sites, we observed a progressive decline in the proportion of malaria cases from the youngest age-group and a progressive increase in the proportion of cases from the oldest age-group. Comparing the last intervention period to baseline, the adjusted proportion of malaria cases among children under 5 years of age decreased from 58% (95% CI: 57% – 59%) to 30% (95% CI: 28% – 31%) in Aduku; 31% (95% CI: 30% – 31%) to 16% (95% CI: 15% – 17%) in Walukuba; 64% (95% CI: 63% – 65%) to 47% (95% CI: 45% – 48%) in Nagongera; and 35% (95% CI: 34% – 36%) to 25% (95% CI: 23% – 27%) in Kasambya. Comparing the same periods, the proportion of malaria cases among patients over 15 years of age increased from 19% (95% CI: 19% – 20%) to 44% (95% CI: 42% – 46%) in Aduku; 40% (95% CI: 40% – 41%) to 61% (95% CI: 59% – 64%) in Walukuba; 18% (95% CI: 17% – 18%) to 31% (95% CI: 29% – 32%) in Nagongera; and, 29% (95% CI: 28% – 29%) to 39% (95% CI: 37% – 41%) in Kasambya.

The upward shift in the age distribution of malaria cases occurred gradually throughout the study periods in all sites except Aduku, where IRS was withdrawn in 2014 for three years (defining the 3^rd intervention period) before another round was implemented in 2017. In Aduku, during the intervals from the 2^nd to the 3^rd intervention periods, the proportion of malaria cases among children under 5 years of age increased from 38% (95% CI: 37% – 39%) to 44% (95% CI: 43% – 44%), followed by a decrease to 30% (95% CI: 28% – 31%) following the last round of IRS. At this site, the proportion of malaria cases among patients over 15 years decreased from 35% (95% CI: 34% – 36%) during the 2^nd to 30% (95% CI: 29% – 31%) during the 3^rd intervention period, before increasing to 44% (95% CI: 42% – 46%) in the last period (Figure 5).

The upward shift in age distribution of confirmed malaria occurred consistently in both males and females (Figure 7, additional file 1). Whilst the majority of all patients, non-suspected and suspected malaria cases were female across the study durations, small differences were observed in age distribution between males and females that were not suspected to be malaria cases (Figure 8, additional file 1), but the age-distribution of females was older than that of males among malaria negative patients across all sites (Figure 9, additional file 1). Moreover, models allowing an interaction between sex and intervention period suggest greater increase in proportion of males than females among confirmed malaria cases over time (Figure 10 and Table 2, additional file).

Overall, Aduku experienced the largest change in the age distribution of malaria cases throughout the study period. The odds of an upward shift in the age category of confirmed malaria cases in the last relative to the baseline intervention periods were 3.27 (95% CI: 2.97 – 3.61) in Aduku, 2.35 (95% CI: 2.14 – 2.58) in Walukuba, 2.03 (95% CI: 1.90 – 2.17) in Nagongera, and 1.59 (95% CI: 1.44 – 1.76) in Kasambya (Table 2). Whereas the interaction between intervention time and diagnostic test used was significant in Kasambya and Nagongera, the same did not notably impact the overall effect observed (Table 3, additional file1). Also, RDT use increased gradually at all sites, reaching 20% only in the last three to four year of the study except Aduku (Figure 4, additional file 1).

We investigated the impact of reductions in malaria transmission following the scale up of malaria control interventions on the age distribution of malaria cases, using routine surveillance data from four sentinel health facilities in Uganda. The study included data from sites with historically varied transmission intensity where large-scale programmatic control interventions of either LLINs alone or LLINs plus IRS were implemented over the approximately ten-year study period.

Our findings provide empirical evidence of rapid shifts in the malaria burden to older individuals, following implementation of effective malaria control interventions. Over time, the proportion of test confirmed malaria cases progressively decreased in children under 5 years of age while it progressively increased in those over 15 years of age, irrespective of transmission settings. The absence of similar changes in age patterns among patients not suspected of malaria suggests that the primary drivers of this shift were declines in malaria transmission intensity associated with effective control interventions and not changes in patient demographics. The reverse shift observed in Aduku during a period of malaria resurgence after three years of interrupted IRS [37], provides further support for this conclusion.

Many factors may have contributed to the shift in age distribution of malaria cases in this study. For many endemic infectious diseases, decreases in transmission are expected to result in increased age of infection and cases [38]. For pathogens like P. falciparum, where partial immunity develops gradually as a consequence of repeated exposure, waning immunity due to decreased infection rates may lead to reduced ability to control parasites [39]. This waning immunity may result in a relative increase in the disease burden among adolescents and adults [12, 40]. Concentration of the malaria burden among older age-groups following reduced transmission has been predicted in other studies [8], and seen in children for both severe and non-severe outcomes [10, 41].

Behavioral factors may have also contributed to rapid shifts in the age distribution of malaria. A reduced proportion of cases among children under 5 years of age may have been due to an increased use of LLINs among this age-group relative to older age-groups [22, 41, 42]. In adults, behavioral factors including travel, leisure and social activities, and occupational activities such as agriculture or night-time work may have increased the risk of exposure outside the household as compared to children. Travel has been reported as a risk factor for P. falciparum infection in East and Southern Africa [43, 44] and for cases of imported malaria being older than those not imported in Southern Africa [45]. Whereas we were unable to evaluate occupational hazards in our study, considerable occupational risk of malaria has been documented among mobile male workers in Asia [46, 47] and populations involved in agriculture in Africa [48, 49].

Our results suggest that implementation of LLINs plus IRS was associated with larger decreases in transmission and larger shifts in age distribution of cases than LLINs alone. However, among LLIN-only sites, Walukuba recorded a much larger shift than Kasambya and in Walukuba the magnitude of the shift was comparable to that of Nagongera, a site with both LLINs and IRS. This suggests that other non-intervention factors for example urbanization [50] may have contributed to the shifts, consistent with other reported findings [38, 51].

Findings from this study showed that a significantly higher proportion of females, as compared to males, seem to seek care for febrile illnesses among the school aged (5-15 years) and adults (over 15 years old) at all sites. This is consistent with the generally older age distribution of females among both the confirmed and the negative malaria cases, but similar age distribution of males and females among those not suspected of malaria. However, shifts in the age distribution of malaria cases observed after control interventions seem to have disproportionately affected males, suggesting a role of gender-based occupational or behavioural differences. Nevertheless, reported greater involvement of females in agriculture in the region [52] than males, as well as documented associations between occupation and education status of mothers and malaria infection risk of families [53] may explain the prevailing significantly high proportion of older females and hence the need for continued interventions that address this vulnerable group.

This study had limitations. First, without well-defined catchment populations for our study sites, we were unable to estimate changes in malaria incidence over time and thus were limited to describing shifts in the age distribution of laboratory confirmed cases of malaria. Additional research is necessary to determine impacts of reduced transmission on incidence of malaria. Second, the increased use of RDT’s over time, may have influenced our results through varied sensitivity [54]. Nevertheless, in our regression analyses, diagnostic test used was accounted for. Moreover, patterns seen among suspected and confirmed cases were absent in the cases not suspected of malaria, suggesting that the impact of diagnostic method would be minimal. Third, inability to account for levels of uptake of the interventions implemented at our study sites could have masked some important variations. However, we believe that the community-wide benefit of reduced vector populations [55], coupled with the multiple rounds of interventions, would have mitigated this effect and thus its impact on our results. Fourth, whereas fever or history of fever in the past 48 hours is the main indicator of suspected malaria, this specific measure was not consistently recorded in the health facilities, and so inconsistency in consideration of the diagnosis may have impacted our results. Nevertheless, other proxy indicators of suspected malaria were also considered to ensure a fairly complete capture of these cases. Lastly, having analyzed health facility surveillance data, this would only include participants that seek care in the public health facilities. As such community level interventions, especially iCCM that target young children could have had an impact on our findings. However, iCCM was limited to one of the four sites and for less than two years.

The findings from this study have important implications regarding targeted malaria control interventions that have historically focused on children under 5 years of age. Whereas these efforts need to continue, new strategies may be necessary to address the shift in burden to older age-groups following the implementation of successful malaria control interventions. For instance, extending iCCM for malaria to older age-groups after intensive IRS, with the post IRS duration often associated with malaria resurgence. These findings may also have implications for the optimal allocation of health care resources for the diagnosis, treatment and control of malaria amidst changing transmission. Further, they highlight the usefulness of considering age distribution in surveillance and monitoring change, and the role of surveillance in understanding the epidemiology of malaria.

Acknowledgements

We would like to especially thank God in this work. We also would like to thank the administration and staff at the Infectious Diseases Research Collaboration, and HMIS officers and in-charges at each of the clinics for all their hard work and great contributions to the success of this project. We also acknowledge the district health leadership in Jinja, Tororo, Mubende, and Apac districts for all their cooperation with and support toward this study.

Ethics approval

This study was approved by the Uganda National Council for Science and Technology (UNCST Ref SS 4455), the Makerere University School of Medicine Research and Ethics Committee (REC Ref. 2017-119), and the London School of Hygiene & Tropical Medicine Ethics Committee (LSHTM Ref. 13902)

Written informed consent of participants included in this study was not applicable.

Consent for publication

Not applicable.

Availability of data and material

The datasets used and/or analyzed for this study are available from the corresponding author on reasonable request.

Funding statement

Funding for parent study was provided by the President’s Malaria Initiative, U.S. Agency for International Development under the terms of Interagency Agreement (1U51CK000117) with the CDC. For the sub-study however, research reported in this publication was supported by the Fogarty International Center of the National Institutes of Health under Award Numbers (D43TW7375) and (D43TW010526). JIN is supported by the Fogarty International Center (Emerging Global Leader Award) grant number (K43TW010365). The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the President’s Malaria Initiative, U.S. Agency for International Development or the National Institutes of Health.

Author’s contributions

SPK AND RK conceived of the study with input from RP, KH, JIN, GD & IR-B. AM, AS, AY, & RK led the data collection activities with input from GD, SGS, DR, CS, JO & MK. RK led the data management with input from GD. SPK, IR-B, GD & RK led the data analysis, with support from AE, BG. All authors, with all the data in this study at their disposal, reviewed the manuscript and gave permission for publication. SPK, the corresponding author, takes full responsibility for submission towards publication.

Declaration of interests

The authors declare that they have no competing interests.

ACT – Artemisinin-based Combination Therapies

aEIR – annual Entomological Inoculation Rate

CI – Confidence Interval

HMIS – Health Management Information System

iCCM – integrated Community Case Management

IQR – Inter quartile range

IRS – Indoor Residual Spraying of insecticide

ITN – Insecticide Treated Net

LLIN – Long-lasting insecticidal net

MIS – Malaria Indicator Survey

MRC – Malaria Reference Centre

NMCD – National Malaria Control Division

OPD – Outpatients Department

RDT – Rapid Diagnostic Test

TPR – Test Positivity Rate

UMSP – Uganda Malaria Surveillance Project

WHO – World Health Organization

Bhatt, S., et al., The effect of malaria control on Plasmodium falciparum in Africa between 2000 and 2015. Nature, 2015. 526(7572): p. 207-11.
Murray, C.J., et al., Global, regional, and national incidence and mortality for HIV, tuberculosis, and malaria during 1990-2013: a systematic analysis for the Global Burden of Disease Study 2013. Lancet, 2014. 384(9947): p. 1005-70.
Noor, A.M., et al., The changing risk of Plasmodium falciparum malaria infection in Africa: 2000-10: a spatial and temporal analysis of transmission intensity. Lancet, 2014. 383(9930): p. 1739-47.
Nkumama, I.N., W.P. O'Meara, and F.H.A. Osier, Changes in Malaria Epidemiology in Africa and New Challenges for Elimination. Trends Parasitol, 2017. 33(2): p. 128-140.
WHO, World Malaria Report 2008. 2008, World Health Organization: Geneva, Switzerland.
WHO, Malaria surveillance, monitoring & evaluation: a reference manual. 2018, World Health Organization: Geneva. p. 8.
Molineaux L., G.G., The Garki Project: Research on the epidemiology and control of malaria in the Sudan Savanna of West Africa. 1980, World Health Organization: Geneva.
Carneiro, I., et al., Age-patterns of malaria vary with severity, transmission intensity and seasonality in sub-Saharan Africa: a systematic review and pooled analysis. PLoS One, 2010. 5(2): p. e8988.
Schellenberg, D., et al., The changing epidemiology of malaria in Ifakara Town, southern Tanzania. Trop Med Int Health, 2004. 9(1): p. 68-76.
Ceesay, S.J., et al., Changes in malaria indices between 1999 and 2007 in The Gambia: a retrospective analysis. Lancet, 2008. 372(9649): p. 1545-54.
WHO, World Malaria Report 2017. 2017, World Health Organization: Geneva, Switzerland.
Nguyen, H.V., et al., Marked age-dependent prevalence of symptomatic and patent infections and complexity of distribution of human Plasmodium species in central Vietnam. Am J Trop Med Hyg, 2012. 87(6): p. 989-95.
Nankabirwa, J., et al., Asymptomatic Plasmodium infection and cognition among primary schoolchildren in a high malaria transmission setting in Uganda. Am J Trop Med Hyg, 2013. 88(6): p. 1102-8.
WHO, Global technical strategy for malaria 2016-2030. 2015, World Health Organization: Geneva, Switzerland.
Fekadu, M., M.K. Yenit, and A.M. Lakew, The prevalence of asymptomatic malaria parasitemia and associated factors among adults in Dembia district, northwest Ethiopia, 2017. Arch Public Health, 2018. 76: p. 74.
Oladeinde, B.H., et al., Asymptomatic Malaria among Blood Donors in Benin City Nigeria. Iran J Parasitol, 2014. 9(3): p. 415-22.
WHO, World Malaria Report 2005. 2005, World Health Organization and UNICEF: Geneva, Switzerland.
WHO, World Malaria Report 2010. 2010, World Health Organization: Geneva, Switzerland.
MoH, The Uganda Malaria Reduction Strategic Plan 2014-2020. 2014, Ministry of Health: kampala, Uganda.
UBOS, Ugand Demographic and Health Survey 2006. 2007, Uganda Bureau of Statistics (UBOS) and Macro International Inc: Calverton, Maryland, USA.
MoH, Uganda National Malaria Control Strategyic Plan 2010/11 - 2014/15. 2011, Ministry of Health: Kampala, Uganda.
Wanzira, H., H. Katamba, and D. Rubahika, Use of long-lasting insecticide-treated bed nets in a population with universal coverage following a mass distribution campaign in Uganda. Malar J, 2016. 15: p. 311.
MoH, Annual Health Sector Performance Report - FY 2010/2011. 2011, Ministry of Health - Uganda: Kampala, Uganda.
International, U.B.o.S.U.a.I., Uganda Malaria Indicator Survey 2014-15. 2015: Kampala, Uganda, and Rockville, Maryland, USA.
MoH, Annual Health Sector Performance Report - FY 2016/2017. 2017, Ministry of Health - Uganda: Kampala, Uganda.
Ssempiira, J., et al., The effect of case management and vector-control interventions on space-time patterns of malaria incidence in Uganda. Malar J, 2018. 17(1): p. 162.
MoH, Annual Health Sector Performance Report - FY 2014/2015. 2015, Ministry of Health - Uganda: Kampala, Uganda.
Macro, U.B.o.S.U.a.I., Uganda Malaria Indicator Survey 2009. 2010, UBOS and ICF Macro: Calverton, Maryland, USA.
Sserwanga, A., et al., Improved malaria case management through the implementation of a health facility-based sentinel site surveillance system in Uganda. PLoS One, 2011. 6(1): p. e16316.
Kigozi, R., et al., Indoor residual spraying of insecticide and malaria morbidity in a high transmission intensity area of Uganda. PLoS One, 2012. 7(8): p. e42857.
Okello, P.E., et al., Variation in malaria transmission intensity in seven sites throughout Uganda. Am J Trop Med Hyg, 2006. 75(2): p. 219-25.
Sexton, A.R., Best practices for an insecticide-treated bed net distribution programme in sub-Saharan eastern Africa. Malar J, 2011. 10: p. 157.
Teklehaimanot, A., J.D. Sachs, and C. Curtis, Malaria control needs mass distribution of insecticidal bednets. Lancet, 2007. 369(9580): p. 2143-2146.
Katureebe, A., et al., Measures of Malaria Burden after Long-Lasting Insecticidal Net Distribution and Indoor Residual Spraying at Three Sites in Uganda: A Prospective Observational Study. PLoS Med, 2016. 13(11): p. e1002167.
Bi, Y., et al., Can slide positivity rates predict malaria transmission? Malaria Journal. 11: p. 117.
Team, R.C., R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria., 2018.
Raouf, S., et al., Resurgence of Malaria Following Discontinuation of Indoor Residual Spraying of Insecticide in an Area of Uganda With Previously High-Transmission Intensity. Clin Infect Dis, 2017. 65(3): p. 453-460.
O'Meara, W.P., et al., Changes in the burden of malaria in sub-Saharan Africa. Lancet Infect Dis, 2010. 10(8): p. 545-55.
Farnert, A., et al., Epidemiology of malaria in a village in the Rufiji River Delta, Tanzania: declining transmission over 25 years revealed by different parasitological metrics. Malar J, 2014. 13: p. 459.
Vliegenthart-Jongbloed, K., et al., Severity of imported malaria: protective effect of taking malaria chemoprophylaxis. Malar J, 2013. 12: p. 265.
O'Meara, W.P., et al., Effect of a fall in malaria transmission on morbidity and mortality in Kilifi, Kenya. Lancet, 2008. 372(9649): p. 1555-62.
Ferrari, G., et al., Identifying risk factors for Plasmodium infection and anaemia in Kinshasa, Democratic Republic of Congo. Malar J, 2016. 15: p. 362.
Arinaitwe, E., et al., Association between Recent Overnight Travel and Risk of Malaria: A Prospective Cohort Study at Three Sites in Uganda. Clin Infect Dis, 2018.
Mutegeki, E., M.J. Chimbari, and S. Mukaratirwa, Assessment of individual and household malaria risk factors among women in a South African village. Acta Trop, 2017. 175: p. 71-77.
Lowa, M., et al., Human mobility and factors associated with malaria importation in Lusaka district, Zambia: a descriptive cross sectional study. Malar J, 2018. 17(1): p. 404.
Erhart, A., et al., Epidemiology of forest malaria in central Vietnam: a large scale cross-sectional survey. Malar J, 2005. 4: p. 58.
Yang, Y., et al., Epidemiology of Malaria in Yulin, South China 1999-2016: Imported Malaria Threatens Zero Local Case Status. Vector Borne Zoonotic Dis, 2018. 18(10): p. 533-538.
Janko, M.M., et al., The links between agriculture, Anopheles mosquitoes, and malaria risk in children younger than 5 years in the Democratic Republic of the Congo: a population-based, cross-sectional, spatial study. Lancet Planet Health, 2018. 2(2): p. e74-e82.
Naidoo, S., et al., Occupational activities associated with a reported history of malaria among women working in small-scale agriculture in South Africa. Am J Trop Med Hyg, 2011. 85(5): p. 805-10.
Kigozi, S.P., et al., Associations between urbanicity and malaria at local scales in Uganda. Malar J, 2015. 14: p. 374.
Marsh, K., Africa in transition: the case of malaria. Int Health, 2016. 8(3): p. 155-6.
Woldu, D.O. and Z.T. Haile, Gender roles and perceptions of malaria risk in agricultural communities of Mwea Division in Central Kenya. Women Health, 2015. 55(2): p. 227-43.
Essendi, W.M., et al., Epidemiological risk factors for clinical malaria infection in the highlands of Western Kenya. Malar J, 2019. 18(1): p. 211.
Laban, N.M., et al., Comparison of a PfHRP2-based rapid diagnostic test and PCR for malaria in a low prevalence setting in rural southern Zambia: implications for elimination. Malar J, 2015. 14: p. 25.
Hawley, W.A., et al., Community-wide effects of permethrin-treated bed nets on child mortality and malaria morbidity in western Kenya. Am J Trop Med Hyg, 2003. 68(4 Suppl): p. 121-7.

Table 1. Malaria associated demographics of study participants for each site, by intervention period

Site	Intervention period	Total observations*	Suspected malaria (%)	Tested for malaria (%) ^β	Microscopy (%)	Positive test result (%) ^β
Walukuba	Jan2009 – Oct2013	183,327	95,080 (51.9%)	92,784 (97.6%)	92,738 (99.9%)	30,785 (33.2%)
	Nov2013 – May2017	111,506	27,483 (24.7%)	26,051 (94.8%)	24,360 (93.5%)	9,239 (35.5%)
	Jun2017 – Jul2018	29,023	7,733 (26.6%)	7,024 (90.8%)	5,227 (74.4%)	2,037 (29.0%)
Kasambya	Jan2009 – Nov2013	85,200	65,768 (77.2%)	63,479 (96.5%)	60,070 (94.6%)	24,538 (38.7%)
	Dec2013 – Nov2017	60,488	41,823 (69.1%)	38,328 (91.6%)	30,199 (78.8%)	15,996 (41.7%)
	Dec2017 – Jul2018	8,123	5,272 (64.9%)	5,083 (96.4%)	1,156 (22.7%)	1,529 (30.1%)
Aduku	Jan2009 – Aug2010	34,596	19,024 (55.0%)	17,877 (94.0%)	17,877 (100%)	9,921 (55.5%)
	Sep2010 – Apr2014	71,329	40,428 (56.7%)	40,277 (99.6%)	39,056 (97.0%)	9,825 (24.4%)
	May2014 – May2017	66,092	34,467 (52.2%)	29,410 (85.3%)	17,466 (59.4%)	15,311 (52.1%)
	Jun2017 – Jul2018	25,389	9,013 (35.5%)	8,158 (90.5%)	2,280 (28.0%)	2,189 (26.8%)
Nagongera	Jan2009 – Nov2013	124,711	82,762 (66.4%)	76,915 (92.9%)	74,959 (97.5%)	27,007 (35.1%)
	Dec2013 – Jan2015	24,530	15,893 (64.8%)	15,723 (98.9%)	14,874 (94.6%)	3,977 (25.3%)
	Feb2015 – Jul2018	72,236	27,297 (37.8%)	27,064 (99.2%)	16,213 (59.9%)	4,309 (15.9%)

*Patients seen at the health facility excluding those with a missing record of age; β-Testing for malaria includes both microscopy and RDT

Table 2. Association between being malaria confirmed within ages (<5, 5-15, >15years) and control intervention period.

Factor	Categories	Odds	95% CI	P - Value
Walukuba
Sex	Male	1	Ref
Sex	Female	1.34	1.29 - 1.39	<0.001
Malaria test done	Microscopy	1	Ref
Malaria test done	RDT	2.11	0.61 - 7.26	0.237
Intervention period	Baseline	1	Ref
	First period	1.55	1.48 - 1.62	<0.001
	Last period	2.27	2.05 - 2.51	<0.001
Kasambya
Sex	Male	1	Ref
Sex	Female	1.39	1.34 - 1.45	<0.001
Malaria test done	Microscopy	1	Ref
Malaria test done	RDT	0.98	0.93 - 1.03	0.368
Intervention period	Baseline	1	Ref
	First period	1.39	1.34 - 1.44	<0.001
	Last period	1.59	1.44 - 1.76	<0.001
Aduku
Sex	Male	1	Ref
Sex	Female	2.64	2.54 - 2.75	<0.001
Malaria test done	Microscopy	1	Ref
Malaria test done	RDT	1.26	1.19 - 1.32	<0.001
Intervention period	Baseline	1	Ref
	First period	2.23	2.11 - 2.35	<0.001
	Second period	1.78	1.68 - 1.89	<0.001
	Last period	3.27	2.97 - 3.61	<0.001
Nagongera
Sex	Male	1
Sex	Female	2.18	2.09 - 2.28	<0.001
Malaria test done	Microscopy	1
Malaria test done	RDT	1.24	1.15 - 1.34	<0.001
Intervention period	Baseline	1
	First period	1.19	1.11 - 1.27	<0.001
	Last period	2.03	1.90 - 2.17	<0.001

KigoziRapidshiftsinagespecificburdenAdditionalfile1.doc

Rapid shifts in the age-specific burden of malaria following successful control interventions in four regions of Uganda

Status:

Journal Publication

Version 2

Abstract

Figures

Background

Methods

Results

Discussion

Conclusions

Declarations

Abbreviations

References

Tables

Supplementary Files

Status:

Journal Publication

Version 2