Abstract
Introduction
Timely detection, notification and response are critical for mitigating outbreak impact, yet evidence linking timeliness to public health outcomes is limited.
Methods
We hypothesised that timely detection, notification and response are associated with a decrease in the magnitude (ie, cumulative cases and deaths), severity (ie, case fatality ratio) and outbreak duration (ie, days from outbreak start to end). Using data from 84 outbreaks reported in Uganda between 2017 and 2022, we calculated time from outbreak start to detection, detection to notification and notification to initial response. We used regression analysis to measure associations between timeliness indicators and outbreak outcomes, controlling for disease categories.
Results
Longer time to detection was associated with an increase in cases (effect size (ES) = 1.092; p=0.010) and deaths (ES=1.106; p<0.001) and with a longer duration (ES=1.032, p=0.019). A longer time to notification was associated with a lower number of cases (ES=0.935; p=0.002) and deaths (ES=0.961; p=0.024). Time to initial response showed no significant association with any outcome.
Conclusion
Timely detection is associated with fewer cases, deaths and shorter outbreaks. Outbreaks with high impact potential (ie, more cases and deaths) are notified faster, possibly reflecting a heightened sense of urgency among health workers. The findings highlight the importance of frameworks like 7-1-7 in enhancing outbreak response timeliness and improving public health outcomes.
Keywords: Decision Making, Global Health, Health policies and all other topics, Health policy, Health services research
WHAT IS ALREADY KNOWN ON THIS TOPIC
In a previous publication, Bochner et al identified 409 articles and grey literature published in English, using the terms ‘timeliness’ and either ‘outbreak’ or ‘epidemic’. We conducted an additional rapid literature review using a search term (timeliness OR ‘7-1-7’) AND (outbreak* OR epidemic*) on PubMed for articles published in English between 1 January 2022 and 1 November 2024 (n=173), supplemented by a grey literature review for material published during the same period.
Only one report (Hampton et al) included analytical findings that suggest a potential association between timeliness of outbreak management and outbreak outcomes; Ebola virus disease outbreaks exceeding 33 days from symptom onset to outbreak confirmation were the only ones with over 150 reported cases, and these outbreaks incurred significantly higher costs.
WHAT THIS STUDY ADDS
Analysing data from 84 outbreaks in Uganda, we found that faster detection of disease outbreaks is associated with improved outcomes, including fewer cases, fewer deaths and shorter outbreaks.
To our knowledge, this is the first study to demonstrate a statistically significant association between the timeliness of outbreak management milestones and key outcomes.
These add to the evidence base on the public health impact of operationalising timeliness metrics.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Since the introduction of the 7-1-7 framework to improve timely detection, notification and response, many countries have adopted timeliness metrics in their public health emergency management systems to enhance preparedness.
However, evidence demonstrating the impact of these improved metrics on public health outcomes has been limited.
Synthesising evidence on this topic supports advocacy for countries to implement and operationalise timeliness frameworks like 7-1-7.
Such frameworks enable real-time data collection, facilitate timely identification and resolution of bottlenecks and support integration into various planning activities, ultimately strengthening preparedness and reducing the impact of future outbreaks.
Quantifying the impact of timeliness metrics on health outcomes also allows for the calculation of averted disability-adjusted life years; this, in turn, enables countries to articulate their return on investment through economic evaluations, which can serve as a powerful tool for bridging the gap between operational outcomes and policy decisions.
Introduction
Over the past decade, there has been a rise in public health events globally, with Africa experiencing a significant share of these crises.1 2 Frequent infectious disease outbreaks such as Ebola, cholera, measles, Marburg and Lassa fever have underscored the region’s vulnerability to health threats. These outbreaks are often exacerbated by factors like climate change, rapid urbanisation and zoonotic spillover, which results from increasing human-animal interaction in ecologically diverse areas influenced by deforestation and food insecurity.2 3 More recently, the global experience with the COVID-19 pandemic and mpox has further underscored the limitations of existing preparedness assessment tools under the International Health Regulations, such as Joint External Evaluations and States Parties Self-Assessment Annual Reporting, in facilitating real-time monitoring of surveillance and response actions for timely decision-making.3,5
In this context, timeliness indicators emerge as essential tools for enhancing outbreak response by promoting early detection, notification and response.6,10 By tracking timeliness indicators and analysing their performance against targets during outbreaks, countries can prioritise their interventions to prevent localised outbreaks from escalating into larger-scale epidemics or pandemics.11 12 There has been growing advocacy for and progress towards the integration of timeliness metrics into routine outbreak data collection and response management operations, including rapid field investigation, early action reviews and event management systems, to enhance the overall effectiveness of public health responses.11,13
Uganda, a country located in the Congo Basin, is particularly vulnerable to infectious disease outbreaks due to a complex interplay of ecological, social and economic factors.14 15 Uganda has experienced repeated outbreaks of viral haemorrhagic fevers, measles, malaria and cholera.16 17 As part of their continuous effort to improve their response to these challenges, Uganda launched the 7-1-7 framework in October 2021, which enables countries to monitor outbreaks in real time by establishing specific targets: outbreaks must be detected within 7 days of emergence, notified to public health authorities within 1 day of detection and seven early response actions to be completed within 7 days of notification.6
Despite the growing body of research on operationalising timeliness indicators, there remains a dearth of evidence on the association between timeliness measures and impacts on public health outcomes of the outbreaks, such as severity and magnitude. Previously, Liberia’s national strategy for coordinating rapid responses to remote Ebola outbreaks was reported to significantly improve outbreak control, cutting notification time by nearly half, reducing outbreak duration from 53 days to 25 days, doubling patient isolation rates and improving survival rates from 13% to 50%.18 A more recent commentary indicated that Ebola virus disease (EVD) outbreaks exceeding 33 days from symptom onset to outbreak confirmation were the only ones with over 150 reported cases, and these outbreaks incurred significantly higher costs.19 While operational outcomes are important for monitoring progress and processes, quantifying health impacts is crucial for evaluating the effectiveness of the timeliness indicators, ensuring that they lead to meaningful improvements in population health. In light of this gap, this study uses timeliness data collected from 84 One Health outbreaks reported in Uganda between 20172022 and investigates the association between the timeliness of detection, notification and response and key outcomes of outbreaks.
Methods
Study hypotheses
We hypothesised that timely detection, notification and response to outbreaks are associated with a reduced magnitude, as indicated by the number of cumulative cases and cumulative deaths. Second, we hypothesised that timely detection, notification and response to outbreaks are associated with a reduced severity, as indicated by the case fatality ratio (CFR). Finally, we hypothesised that timely detection, notification and response to outbreaks are associated with a shortened outbreak duration.
Study data
This is a retrospective analysis of the Uganda One Health Timeliness Metrics for Outbreak Reporting, collected by investigators from the study team. The dataset includes timeliness metrics from the Salzburg Statement on Metrics for One Health Surveillance and cases, deaths and outbreak duration for 85 outbreaks that took place between 2017 and 2022 in Uganda.7 Detailed methods of the original data collection are described elsewhere.7 11 The data included only outbreaks that triggered a multisectoral response under the One Health framework and excluded outbreaks involving only human cases (eg, measles) as well as natural disasters.
The original data included substantial missing information across milestone dates. To address these gaps, we conducted an additional retrospective review of all National Public Health Emergency Operations Center (PHEOC) records related to a subset of 46 outbreaks with one or more missing milestone dates. Through this review, we obtained the PHEOC activation and closure dates for 40 of these outbreaks. We then used these dates as proxies to fill in missing milestone dates: for 13 outbreaks, the PHEOC activation date was used to approximate the date of initial response and for six outbreaks, the PHEOC closure date was used to approximate the outbreak end date. For the remaining six outbreaks, we were unable to find sufficient information on PHEOC records to approximate any of the missing milestone dates. However, these events were retained in the analysis despite the missing data. The only outbreak that was excluded from the analysis was the COVID-19 outbreak, which was considered a major outlier. As a result, the subsequent analysis considered 84 outbreak events.
Outcome variables
The magnitude of a given outbreak is defined by the total number of cumulative cases and deaths. The severity of the outbreak is measured by the CFR, calculated by dividing the reported number of deaths by the reported number of cumulative cases. Lastly, the duration of the outbreak is defined as the number of days between the reported start date and the end date of the outbreak. The distribution of outcome variables is shown in Supplementary Material 1.
Timeliness indicators
We calculated the number of days between key milestone events, which served as timeliness indicators for this study.6 Specifically, we assessed the following intervals: the number of days between the outbreak start and its detection, the days between outbreak detection and notification and the days between outbreak notification and the initial response. The initial response date was defined as the date on which any intervention was first enacted.7
Disease categories
Specific types of diseases may influence outbreak magnitude, severity and duration. For example, vector-borne diseases often lead to larger cumulative cases and fatalities due to their transmission from vectors to humans, while viral haemorrhagic fevers may present higher CFRs and prolonged outbreak durations due to their complex pathophysiology and response challenges.20 21 We also hypothesised that disease categories cause variations in outbreak management efficiency, leading us to examine the differential distribution of timeliness indicators stratified by these categories. Consequently, we included disease categories as a covariate in our analysis to account for their potential influence. To do so, diseases were classified into six mutually exclusive categories, following the schema established in the previous 7-1-7 publication by Bochner et al.12 The categories included: animal/zoonotic diseases (n=15), viral haemorrhagic fevers (VHF; n=26), vaccine-preventable diseases (VPD; n=7), foodborne or waterborne diseases (n=27), vector-borne diseases (VBD; n=4) and unknown illnesses (n=5). A breakdown of the diseases classified under each group is provided in the (online supplemental material table S1-1).
Analysis
We first conducted a descriptive analysis to observe the number and characteristics of events by disease category. Next, we examined the distribution of each outcome variable, stratified by the timeliness indicator values, to identify any distinct patterns and investigate our hypotheses. We also performed descriptive analyses on the patterns of missingness across the variables used in the analysis.
We then conducted regression analyses to confirm the statistical significance of the associations between the outcome and explanatory variables. We used the number of days between milestone events as explanatory variables. After examining the distribution of each outcome variable, we fitted a generalised linear model on a log-transformed form of cumulative cases and outbreak duration, which have positive values with a right-skewed distribution. We used a negative binomial model for the number of deaths and CFR, which have non-negative, over-dispersed, right-skewed distributions. For CFR, cumulative cases were used as an offset.
We tested each hypothesis through both crude and adjusted analysis, with disease categories introduced as covariates in the adjusted analysis. We anticipated that disease category might confound the relationship between timeliness indicators and outbreak outcomes, as variations in transmissibility, severity and public health urgency across different diseases could influence how outbreaks are detected, notified and responded to. Therefore, ignoring disease category may obscure or underestimate true associations. We reported whether the coefficient of each timeliness indicator was statistically significant, with its exponentiated effect size. Since all the regression analyses used a log link function, an exponentiated coefficient can be interpreted as a multiplicative effect on the outcome variable. This means that for each one-unit increase in the predictor variable, the outcome variable is multiplied by the exponentiated coefficient, assuming other variables remain constant. Additionally, we confirmed that the model including the covariate had a lower Akaike information criterion (AIC) value compared with the crude model, indicating a better overall model fit.
Role of the funding source
The funder of the study supported data collection but had no role in study design, data analysis, data interpretation, report writing or the decision to submit for publication.
Patient and public involvement
Patients or the public were not involved in this study.
Results
Descriptive analysis
Table 1 and online supplemental table S2-1 in the Supplementary Material provides the descriptive statistics of the 84 events included in the analysis, stratified by the disease categories. More than half (n=48 (57%)) of the events included in the analysis were detected by indicator-based surveillance (online supplemental table S2-1). The time taken between the start of the outbreak and its detection ranged from less than 1 day to 93 days, with the median of 3 days. The median time between detection and notification was 2 days (range 0–55 days), and the median time between notification and initial response was 1 day (range 13–71 days).
Table 1. Description of the data stratified by disease category; n=84.
| Animal/zoonotic (n=15) | Food/water-borne (n=27) | Undiagnosed illness (n=5) | Vector-borne (n=4) | Viral haemorrhagic fevers (n=26) | Vaccine-preventable diseases (n=7) | Overall (n=84) | |
|---|---|---|---|---|---|---|---|
| Cumulative cases | |||||||
| Mean (SD) | 30.0 (23.2) | 203 (423) | 39.0 (68.7) | 1660 (2680) | 15.7 (33.4) | 13.3 (8.62) | 144 (589) |
| Median (min, max) | 22.5 (2.00, 77.0) | 64.0 (6.00, 2120) | 5.00 (4.00, 142) | 222 (1.00, 4750) | 4.00 (1.00, 164) | 13.0 (1.00, 26.0) | 16.0 (1.00, 4750) |
| Missing | 1 (6.7%) | 2 (7.4%) | 1 (20.0%) | 1 (25.0%) | 1 (3.8%) | 1 (14.3%) | 7 (8.3%) |
| Confirmed cases | |||||||
| Mean (SD) | 2.54 (2.88) | 26.2 (62.8) | 0 (0) | 2380 (3360) | 7.28 (28.1) | 3.67 (2.16) | 78.4 (564) |
| Median (min, max) | 2.00 (0, 11.0) | 8.50 (1.00, 300) | 0(0, 0) | 2380 (1.00, 4750) | 1.00 (1.00, 142) | 3.50 (1.00, 7.00) | 2.00 (0, 4750) |
| Missing | 2 (13.3%) | 5 (18.5%) | 2 (40.0%) | 2 (50.0%) | 1 (3.8%) | 1 (14.3%) | 13 (15.5%) |
| Human deaths | |||||||
| Mean (SD) | 0.714 (0.914) | 3.08 (8.65) | 2.25 (2.06) | 6.33 (8.39) | 2.96 (10.9) | 2.67 (3.08) | 2.66 (8.07) |
| Median (min, max) | 0.500 (0, 3.00) | 1.00 (0, 44.0) | 2.50 (0, 4.00) | 2.00 (1.00, 16.0) | 1.00 (0, 55.0) | 1.50 (0, 7.00) | 1.00 (0, 55.0) |
| Missing | 1 (6.7%) | 2 (7.4%) | 1 (20.0%) | 1 (25.0%) | 1 (3.8%) | 1 (14.3%) | 7 (8.3%) |
| Case fatality ratio (%) | |||||||
| Mean (SD) | 8.01 (16.5) | 2.37 (4.06) | 54.2 (53.4) | 35.7 (55.8) | 31.8 (39.0) | 19.2 (24.8) | 18.3 (31.8) |
| Median (min, max) | 0.649 (0, 50.0) | 0.758 (0, 16.7) | 58.3 (0, 100) | 7.21 (0.0421, 100) | 16.7 (0, 100) | 8.17 (0, 60.0) | 1.89 (0, 100) |
| Missing | 1 (6.7%) | 2 (7.4%) | 1 (20.0%) | 1 (25.0%) | 1 (3.8%) | 1 (14.3%) | 7 (8.3%) |
| Outbreak duration (days) | |||||||
| Median (min, max) | 88.0 (42.0, 272) | 64.0 (17.0, 130) | 126 (126, 126) | 106 (91.0, 120) | 46.0 (29.0, 159) | 144 (110, 179) | 64.5 (17.0, 272) |
| Missing | 5 (33.3%) | 12 (44.4%) | 4 (80.0%) | 2 (50.0%) | 6 (23.1%) | 3 (42.9%) | 32 (38.1%) |
| Days from outbreak start to detection | |||||||
| Median (min, max) | 3.00 (0, 9.00) | 2.00 (0, 5.00) | 93.0 (93.0, 93.0) | 5.50 (5.00, 6.00) | 4.00 (0, 43.0) | 4.50 (2.00, 16.0) | 3.00 (0, 93.0) |
| Missing | 5 (33.3%) | 10 (37.0%) | 4 (80.0%) | 2 (50.0%) | 5 (19.2%) | 3 (42.9%) | 29 (34.5%) |
| Days from detection to notification | |||||||
| Median (min, max) | 1.00 (0, 26.0) | 1.00 (0, 9.00) | 7.00 (1.00, 15.0) | 19.5 (0, 39.0) | 2.00 (0, 19.0) | 36.0 (2.00, 55.0) | 2.00 (0, 55.0) |
| Missing | 3 (20.0%) | 14 (51.9%) | 1 (20.0%) | 2 (50.0%) | 9 (34.6%) | 0 (0%) | 29 (34.5%) |
| Days from notification to initial response | |||||||
| Median (min, max) | 1.00 (−5.00, 13.0) | 1.00 (−5.00, 71.0) | 3.00 (−1.00, 7.00) | 7.50 (−1.00, 16.0) | 0 (−13.0, 10.0) | 6.50 (0, 46.0) | 1.00 (−13.0, 71.0) |
| Missing | 7 (46.7%) | 16 (59.3%) | 3 (60.0%) | 2 (50.0%) | 8 (30.8%) | 1 (14.3%) | 37 (44.0%) |
| Days from outbreak start to initial response | |||||||
| Median (min, max) | 7.00 (1.00, 12.0) | 4.50 (1.00, 73.0) | 107 (107, 107) | 32.5 (22.0, 43.0) | 8.00 (2.00, 47.0) | 56.5 (56.0, 66.0) | 8.00 (1.00, 107) |
| Missing | 8 (53.3%) | 13 (48.1%) | 4 (80.0%) | 2 (50.0%) | 6 (23.1%) | 3 (42.9%) | 36 (42.9%) |
| Days taken between detection to initial response | |||||||
| Median (min, max) | 3.00 (−1.00, 20.0) | 3.00 (0, 71.0) | 14.0 (14.0, 14.0) | 27.0 (16.0, 38.0) | 4.00 (−2.00, 34.0) | 47.0 (2.00, 62.0) | 4.00 (−2.00, 71.0) |
| Missing | 6 (40.0%) | 5 (18.5%) | 3 (60.0%) | 2 (50.0%) | 3 (11.5%) | 1 (14.3%) | 20 (23.8%) |
The time intervals between each milestone date varied by disease category (table 1 and figure 1). For example, while most outbreaks were detected within 7 days of the start date, VHF (median 4 days, range 0–43 days) and VPD (median 4.5 days, range 2–16 days) outbreaks included a few outliers that took much longer to be detected. VPD, vector-borne disease and undiagnosed illness outbreaks had a longer median time to notification (36, 19.5 and 7 days). Although response activities were initiated within 7 days of notification for most outbreaks, the median time between notification and response was longer for vector-borne disease (7.5 days) and VPD outbreaks (6.5 days).
Figure 1. Distribution of days taken between each milestone event by disease category. VHF, viral Haemorrhagichaemorrhagic Fev fever; VPD, vaccine-preventable diVaccine-Preventable Disease.
Missingness analysis
The variables that contained the greatest number of missing observations were the outbreak start date (32% missing), date of notification (31%) and the response date (21%). As a result, 38% of the recorded events had no information on the outbreak duration (online supplemental figure S3-1 in the Supplementary Material). The association between the key variables and missing observations, stratified by disease categories, is presented in the Supplementary Material 3 (online supplemental figure S3-2 and 3). Across disease categories, information was mostly missing at random. However, for undiagnosed illness outbreaks, the missingness pattern was observed to be not at random. For example, undiagnosed illness outbreaks with missing information on the outbreak start date reported a higher number of deaths and CFRs, while those with missing information on outbreak duration reported fewer deaths and CFRs.
Regression analysis
Overall, most of the associations between outbreak timeliness and magnitude, severity and duration were not statistically significant in the crude analysis (table 2). When adjusted for disease categories, a unit increase in the days between outbreak start and detection was associated with a higher number of cases (effect size 1.092, 95% CI 1.025 to 1.165; p value=0.010) and deaths (effect size 1.106, 95% CI 1.041 to 1.164, p value <0.001), as well as a longer outbreak duration (effect size 1.032, 95% CI 1.006 to 1.05], p value=0.019). Conversely, a unit increase in days between event detection and notification was significantly associated with fewer cases (effect size 0.935, 95% CI 0.898 to 0.973, p value=0.002) and deaths (effect size 0.961, 95% CI 0.929 to 0.995, p value=0.024). However, the significant association between timeliness indicators and the number of deaths was not observed in the CFR. The time taken between outbreak notification and initial response was not significantly associated with any of the outcome variables.
Table 2. Crude and adjusted effect sizes of regression analyses of outbreak timeliness metrics and outbreak magnitude, severity and duration.
| Crude | ||||
|---|---|---|---|---|
| Cumulative cases | Case fatality ratio | Number of deaths | Outbreak duration | |
| Time from outbreak start to detection (days) | 0.991 (95% CI 0.954 to 1.029; p value =0.634) | 1.031* (95% CI 0.999 to 1.065; p value =0.060) | 1.051**(95% CI 1.007 to 1.097; p value =0.022) | 1.012* (95% CI 1.000 to 1.024; p value =0.065) |
| Time from detection to notification (days) | 0.952** (95% CI 0.919 to 0.987; p value =0.011) | 1.025 (95% CI: 0.975 to 1.077; p value =0.334) | 0.961** (95% CI: 0.929 to 0.995; p value =0.024) | 1.014* (95% CI 1.000 to 1.028; p value =0.055) |
| Time from notification to initial response (days) | 1.007 (95% CI 0.987 to 1.028; p value =0.504) | 0.997 (95% CI 0.982 to 1.012; p value =0.713) | 1.006 (95% CI 0.992 to 1.021; p value =0.405) | 1.002 (95% CI 0.991 to 1.013; p value =0.724) |
| Adjusted by disease category | ||||
|---|---|---|---|---|
| Cumulative cases | Case fatality ratio | Number of deaths | Outbreak duration | |
| Time from outbreak start to detection (days) | 1.092** (95% CI 1.025 to 1.165; p value =0.010) | 1.007 (95% CI 0.966 to 1.050; p value =0.747) | 1.106*** (95% CI 1.051 to 1.164; p value <0.001) | 1.032** (95% CI 1.006 to 1.058; p value =0.019) |
| Time from detection to notification (days) |
0.935*** (95% CI 0.898 to 0.973; p value =0.002) |
1.013 (95% CI 0.971 to 1.056; p value =0.551) | 0.961** (95% CI 0.929 to 0.995; p value =0.024) | 1.000 (95% CI 0.970 to 1.031; p value =0.989) |
| Time from notification to initial response (days) | 1.007 (95% CI 0.990 to 1.024; p value =0.416) | 0.990 (95% CI 0.978 to 1.003; p value =0.124) | 1.006 (95% CI 0.992 to 1.021; p value =0.405) | 1.002 (95% CI 0.990 to 1.014; p value =0.756) |
Legend for bolded values, according to level of statistical significance: p<0.1, **p<0.05, ***p<0.01.
Discussion
Using timeliness data collected from 84 One Health outbreaks in Uganda reported between 2017 and 2022, we found that timely detection was associated with a reduction in the number of cases, deaths and the overall duration of outbreaks. To the best of our knowledge, our study is among the first to demonstrate the public health impact of rapid outbreak management across a large number of events. These findings support country efforts to strengthen surveillance systems and their integration with response operations to mitigate the impacts of large-scale outbreaks to reduce the burden and duration of such events. Investments in quality improvement approaches, including the 7-1-7 target can help to both identify critical systems bottlenecks for remediation and foster increased visibility and performance management to accelerate detection times to reduce morbidity and mortality from outbreaks.
We found that, contrary to our hypothesis, outbreaks resulting in higher case numbers and fatalities were reported to health authorities more swiftly, although the direction of causality of this association cannot be ascertained from the regression analysis. Holding time to detection constant, larger outbreaks or outbreaks with higher case fatality rates may trigger quicker notification by health workers to public health authorities due to the immediate visibility of the threat and the potential for overwhelming the health system. This complexity merits further research to better understand the interplay with the perceived risk of outbreak and reporting timeliness.
Additionally, disease category emerged as a significant factor in our adjusted analysis, highlighting how different types of diseases can influence detection, notification and response mechanisms. In line with our study hypotheses, we did not interpret disease category as a primary exposure variable, but rather treated it as a confounder to account for systematic differences across diseases and to more accurately isolate the relationship between timeliness and outbreak outcomes. The inclusion of disease category as a covariate in the adjusted analysis revealed associations not observed in the crude analysis. This confirms our hypothesis that varying transmissibility, severity and public health urgency across diseases may influence both timeliness of outbreak operations and outbreak outcomes. For example, diseases like Ebola or cholera often trigger more rapid detection and notification due to heightened awareness and established protocols. Conversely, slower responses may occur for less familiar or lower-priority diseases, leading to prolonged outbreaks and worse outcomes. This is consistent with observations from a previous study, which highlighted variability in response timeliness and outcomes across disease types.12 By accounting for disease category, we controlled for these variations, isolating the effect of timeliness on outbreak outcomes and revealing associations that were obscured in the crude analysis. Although not included in this study, we tested various disease categorization approaches from previous research9 12 22 to ensure the robustness of our analytical findings and minimise sensitivity to the choice of categorization methods as a confounding factor.
We did not observe any significant association between the timely initiation of response actions and outbreak outcomes. The observed lack of association between timely response and outbreak outcomes likely stems from the limitations in how ‘response date’ was defined in the dataset. By marking only the date when any initial response activity was launched,11 the data did not capture the scope and timing of the full set of critical response activities, nor did it assess the quality of response activities once activated. This distinction is important because effective outbreak control generally requires a coordinated and timely execution of all essential response activities, rather than just the timing of the first action.6 23 24 The 7-1-7 framework addresses this crucial aspect by clearly defining the timeliness of response using the ‘date of early response action completed’, which is defined as the latest date on which all of the seven relevant predefined early response actions are completed.25 However, this retrospective data collection largely predated the institutionalisation of 7-1-7 in Uganda, which limited the ability for identification of the completion of all seven response actions.6 These findings highlight the importance of the completion of comprehensive response actions aimed at initial containment of outbreaks and indicate an area of further research to assess impact. In addition, outbreak duration, magnitude and severity are often influenced by the choice and effectiveness of disease-specific prevention and control strategies. While some diseases can be rapidly contained through targeted interventions like ring vaccination or vector control, others require sustained efforts such as case isolation or behaviour change.26 27 The current 7-1-7 metrics do not fully capture this variability, and there is growing recognition of the need to adapt these metrics to better reflect differences across disease types and hazard categories, which can, in turn, influence both detection mechanisms and response strategies.
We found that most outbreaks were detected through indicator-based surveillance (IBS), with the exception of undiagnosed illnesses and vector-borne diseases. Despite a growing emphasis on the implementation of event-based surveillance,28 29 IBS is valuable for outbreak detection and its ongoing role during outbreaks for epidemiologic monitoring and accurate case counting. A notable finding was that endemic diseases like vector-borne and VPDs experienced significantly longer detection times, which may indicate that IBS data are underused for generating timely alerts for these diseases, suggesting that a more proactive use of IBS including threshold alerts that are automated in surveillance systems might improve early detection and response for endemic disease outbreaks.
Effective surveillance systems allow for early detection of public health threats, enabling timely interventions and preventing localised outbreaks from escalating into larger epidemics.30 However, detection alone is not enough and subsequent rapid decision-making is essential to ensure appropriate follow-up actions.31 Furthermore, routinely collected surveillance data and records of response measures can improve decision-making frameworks. For example, these data can support modelling efforts to assess the return on investment of surveillance systems and workforce capacity, helping to quantify their impact.32 Such analyses provide evidence for sustained funding and inform policy prioritisation. Our findings emphasise the need for continued investment in integrated surveillance systems, efficient decision-making processes and data-driven approaches to strengthen public health security.
Interpreting the study findings requires caution due to several limitations and potential biases. First, this is an observational study, and while our analysis found significant associations between the outcome and exposure variables, these associations should not be interpreted as causal due to risks of confounding, selection bias and the potential for reverse causality.33 The dataset is relatively small and includes incomplete data, despite retrospective efforts to address missing information. Furthermore, the data were collected retrospectively from publicly available sources and reported cases, which may not fully reflect the true magnitude and severity of the outbreaks. Another potential bias is underestimating ‘time to detect’, since unreported index cases may occur before the first known case. We also observed wide variability in the time intervals between timeliness milestone dates, including several outliers that reported over 30 (n=10) or 50 (n=4) days between onset, detection, notification or response. Because the 7-1-7 benchmarks are intentionally short, these extreme delays may carry especially valuable lessons, likely reflecting differences in disease characteristics (eg, incubation periods), uneven surveillance and reporting capacities across regions and data quality issues such as incomplete or delayed records. The small number of disease types in some categories (online supplemental table S1-1) may also exaggerate this variability and limit generalisability. We argue that these outliers reflect real-world delays with meaningful consequences. Many were associated with poor outbreak outcomes, emphasising the critical importance of rapid detection, notification and response. For example, three yellow fever outbreaks with delays over 30 days in detection, notification or response had durations exceeding 100 days, with case fatality rates as high as 60%. Similarly, an Ebola outbreak with a 43-day delay in conducting initial response resulted in 164 cases and 55 deaths, a pattern consistent with findings from a previous study.19 Excluding them risks underestimating the true relationship between timeliness and outbreak outcomes. We therefore interpret the full dataset as offering a more complete and policy-relevant view of outbreak response variability.
These limitations emphasise the crucial need for real-time, prospective data collection on timeliness indicators by outbreak response authorities to both improve early warning and response capabilities but also to allow for stronger data generation and analysis.6 Since the study data were collected, Uganda has made significant strides by adopting the 7-1-7 timeliness metrics and integrating the 7-1-7 framework and timeliness data collection tool into the rapid response teams (RRTs). These teams are tasked with collecting 7-1-7 timeliness milestones, as well as information on bottlenecks and enablers of timely response. RRTs submit the collected data to the PHEOC after each deployment, enabling Uganda to evaluate its public health response times against 7-1-7 targets and to address identified bottlenecks in real time. Additionally, briefings on the 7-1-7 framework were incorporated into various stakeholder meetings and operational planning to raise awareness and improve preparedness. As we accumulate more data on outbreaks with complete 7-1-7 milestone dates, we plan to conduct a similar study using the 7-1-7 timeliness metrics, rather than the crude number of days between each milestone. This will allow us to test whether meeting the 7-1-7 targets is associated with better outbreak outcomes.
Conclusion
Our findings reveal a significant association between timely detection and reduced cases, deaths and outbreak duration, underscoring that timely detection is essential for effective public health responses. The limitations of data we highlighted reinforce the necessity for real-time collection of timeliness indicators and their integration into operational reviews. The implementation of the 7-1-7 framework marked a pivotal advancement in Uganda’s public health emergency management strategy. By prioritising the collection of timeliness indicators as outbreaks unfold, Uganda aims to enhance its ability to identify bottlenecks, respond effectively and ultimately improve public health outcomes in future outbreaks. These insights carry important implications for other countries as well. By adopting similar real-time data collection practices and frameworks like 7-1-7, nations can improve their outbreak detection and response capabilities, ultimately leading to better health outcomes.
Supplementary material
Footnotes
Funding: This study was funded by United States Agency for International Development One Health Workforce (Next Generation Project Cooperative Agreement 7200AA19CA00018).
Provenance and peer review: Not commissioned; externally peer reviewed.
Handling editor: Naomi Clare Lee
Patient consent for publication: Not applicable.
Ethics approval: Not applicable.
Data availability free text: The dataset generated and analysed in the study is available from the corresponding author on request.
Patient and public involvement: Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Data availability statement
Data are available upon reasonable request.
References
- 1.Ngom R, Gueye AS, Lassieur S, et al. Five decades of infectious diseases outbreaks in the African region (1970–2018) a geographic snapshot. Soc Sci Humanit Open. 2023;8:100625. [Google Scholar]
- 2.Koua EL, Njingang JRN, Kimenyi JP, et al. Trends in public health emergencies in the WHO African Region: an analysis of the past two decades public health events from 2001 to 2022. BMJ Glob Health. 2023;8:e012015. doi: 10.1136/bmjgh-2023-012015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Ellwanger JH, Fearnside PM, Ziliotto M, et al. Synthesizing the connections between environmental disturbances and zoonotic spillover. An Acad Bras Cienc. 2022;94 doi: 10.1590/0001-3765202220211530. [DOI] [PubMed] [Google Scholar]
- 4.World Health Organization Independent panel for pandemic preparedness and response. 2024. https://theindependentpanel.org/wp-content/uploads/2021/05/COVID-19-Make-it-the-Last-Pandemic_final.pdf Available.
- 5.World Health Organization WHO’s work in health emergencies strengthening preparedness for health emergencies: implementation of the international health regulations. 2005. https://cdn.who.int/media/docs/default-source/documents/emergencies/a74_9add1-en.pdf Available.
- 6.Frieden TR, Lee CT, Bochner AF, et al. 7-1-7: an organising principle, target, and accountability metric to make the world safer from pandemics. The Lancet. 2021;398:638–40. doi: 10.1016/S0140-6736(21)01250-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fieldhouse JK, Randhawa N, Fair E, et al. One Health timeliness metrics to track and evaluate outbreak response reporting: A scoping review. EClinicalMedicine. 2022;53:101620. doi: 10.1016/j.eclinm.2022.101620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Crawley AW, Divi N, Smolinski MS. Using Timeliness Metrics to Track Progress and Identify Gaps in Disease Surveillance. Health Secur. 2021;19:309–17. doi: 10.1089/hs.2020.0139. [DOI] [PubMed] [Google Scholar]
- 9.Impouma B, Roelens M, Williams GS, et al. Measuring Timeliness of Outbreak Response in the World Health Organization African Region, 2017-2019. Emerg Infect Dis . 2020;26:2555–64. doi: 10.3201/eid2611.191766. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chan EH, Brewer TF, Madoff LC, et al. Global capacity for emerging infectious disease detection. Proc Natl Acad Sci U S A . 2010;107:21701–6. doi: 10.1073/pnas.1006219107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Fieldhouse J, Nakiire L, Kayiwa J, et al. How feasible or useful are timeliness metrics as a tool to optimise One Health outbreak responses? BMJ Glob Health. 2024;9:e013615. doi: 10.1136/bmjgh-2023-013615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bochner AF, Makumbi I, Aderinola O, et al. Implementation of the 7-1-7 target for detection, notification, and response to public health threats in five countries: a retrospective, observational study. Lancet Glob Health. 2023;11:e871–9. doi: 10.1016/S2214-109X(23)00133-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Elston J, Eteng W-EO, Ihekweazu C, et al. Development and Implementation of a Public Health Event Management System, Nigeria, 2018-2024. Emerg Infect Dis . 2025;31:4–13.:e240379. doi: 10.3201/eid3101.240379. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Buregyeya E, Atusingwize E, Nsamba P, et al. Operationalizing the One Health Approach in Uganda: Challenges and Opportunities. J Epidemiol Glob Health. 2020;10:250–7. doi: 10.2991/jegh.k.200825.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Redding DW, Atkinson PM, Cunningham AA, et al. Impacts of environmental and socio-economic factors on emergence and epidemic potential of Ebola in Africa. Nat Commun. 2019;10:4531. doi: 10.1038/s41467-019-12499-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Kiggundu T, Ario AR, Kadobera D, et al. Notes from the Field: Outbreak of Ebola Virus Disease Caused by Sudan ebolavirus - Uganda, August-October 2022. MMWR Morb Mortal Wkly Rep. 2022;71:1457–9. doi: 10.15585/mmwr.mm7145a5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ario AR, Aliddeki DM, Kadobera D, et al. Uganda’s experience in establishing an electronic compendium for public health emergencies. PLOS Glob Public Health . 2023;3:e0001402. doi: 10.1371/journal.pgph.0001402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kateh F, Nagbe T, Kieta A, et al. Rapid response to Ebola outbreaks in remote areas - Liberia, July-November 2014. MMWR Morb Mortal Wkly Rep. 2015;64:188–92. [PMC free article] [PubMed] [Google Scholar]
- 19.Hampton LM, Luquero F, Costa A, et al. Ebola outbreak detection and response since 2013. Lancet Microbe. 2023;4:e661–2. doi: 10.1016/S2666-5247(23)00136-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Bwire G, Sartorius B, Guerin P, et al. Sudan Ebola virus (SUDV) outbreak in Uganda, 2022: lessons learnt and future priorities for sub-Saharan Africa. BMC Med. 2023;21:144. doi: 10.1186/s12916-023-02847-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Mbonye AK, Wamala JF, Nanyunja M, et al. Ebola viral hemorrhagic disease outbreak in West Africa- lessons from Uganda. Afr Health Sci. 2014;14:495–501. doi: 10.4314/ahs.v14i3.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.World Health Organization World health organization. annex 2 of the international health regulations (2005) 2017. https://www.who.int/publications/m/item/annex-2-of-the-international-health-regulations-(2005) Available.
- 23.Olu OO, Lamunu M, Chimbaru A, et al. Incident Management Systems Are Essential for Effective Coordination of Large Disease Outbreaks: Perspectives from the Coordination of the Ebola Outbreak Response in Sierra Leone. Front Public Health. 2016;4:254. doi: 10.3389/fpubh.2016.00254. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Perscheid C, Benzler J, Hermann C, et al. Ebola Outbreak Containment: Real-Time Task and Resource Coordination With SORMAS. Front ICT. 2018;5:7. [Google Scholar]
- 25.7-1-7 ALliance 7-1-7 milestone dates reference guide. 2024. https://717alliance.org/wp-content/uploads/2024/05/717-milestones-reference-guide.pdf Available.
- 26.Peak CM, Childs LM, Grad YH, et al. Comparing nonpharmaceutical interventions for containing emerging epidemics. Proc Natl Acad Sci U S A . 2017;114:4023–8. doi: 10.1073/pnas.1616438114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Bedford J, Farrar J, Ihekweazu C, et al. A new twenty-first century science for effective epidemic response. Nature New Biol. 2019;575:130–6. doi: 10.1038/s41586-019-1717-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Balajee SA, Salyer SJ, Greene-Cramer B, et al. The practice of event-based surveillance: concept and methods. Global Security: Health, Science and Policy . 2021;6:1–9. doi: 10.1080/23779497.2020.1848444. [DOI] [Google Scholar]
- 29.World Health Organization Early detection, assessment and response to acute public health events: implementation of early warning and response with a focus on event-based surveillance. 2014. https://iris.who.int/bitstream/handle/10665/112667/WHO_HSE_GCR?sequence=1 Available.
- 30.Resolve to save lives. Epidemics that didn’t happen. 2024. https://etdh.resolvetosavelives.org/2024/ Available.
- 31.World Health Organization . Epidemic analysis for response decision-making: systematic organization of multi-source information to inform response decisions. 2020. Epidemic analysis for response decision-making: systematic organization of multi-source information to inform response decisions. [Google Scholar]
- 32.de Vries L, Koopmans M, Morton A, et al. The economics of improving global infectious disease surveillance. BMJ Glob Health. 2021;6:e006597. doi: 10.1136/bmjgh-2021-006597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Ejima K, Li P, Smith DL, Jr, et al. Observational research rigour alone does not justify causal inference. Eur J Clin Invest. 2016;46:985–93. doi: 10.1111/eci.12681. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data are available upon reasonable request.