Multi-source Surveillance Research Articles

Multi-Task Intelligent Monitoring of Construction Safety Based on Computer Vision

Effective safety management is vital for ensuring construction safety. Traditional safety inspections in construction heavily rely on manual labor, which is both time-consuming and labor-intensive. Extensive research has been conducted integrating computer-vision technologies to facilitate intelligent surveillance and improve safety measures. However, existing research predominantly focuses on singular tasks, while construction environments necessitate comprehensive analysis. This study introduces a multi-task computer vision technology approach for the enhanced monitoring of construction safety. The process begins with the collection and processing of multi-source video surveillance data. Subsequently, YOLOv8, a deep learning-based computer vision model, is adapted to meet specific task requirements by modifying the head component of the framework. This adaptation enables efficient detection and segmentation of construction elements, as well as the estimation of person and machine poses. Moreover, a tracking algorithm integrates these capabilities to continuously monitor detected elements, thereby facilitating the proactive identification of unsafe practices on construction sites. This paper also presents a novel Integrated Excavator Pose (IEP) dataset designed to address the common challenges associated with different single datasets, thereby ensuring accurate detection and robust application in practical scenarios.

Multisource surveillance conducted by the Tokyo Metropolitan Government during the Tokyo 2020 Olympic and Paralympic Games.

The Tokyo 2020 Olympic and Paralympic Games (the Games) were held from 23 July to 5 September 2021 in Tokyo, Japan, after a 1-year delay due to the coronavirus disease (COVID-19) pandemic. The Tokyo Metropolitan Government was responsible for monitoring and responding to infectious disease outbreaks other than COVID-19 during the Games. A multisource surveillance system was used from 1 July to 12 September 2021 for the early detection and rapid response to infectious diseases. This included routine notifiable disease surveillance, sentinel surveillance, syndromic surveillance, cluster surveillance, ambulance transfer surveillance and the Tokyo Infectious Alert system. Daily reports were disseminated summarizing the data collected from the multisource surveillance system. No case of infectious disease under the Tokyo Metropolitan Government system required a response during the Games. The multisource surveillance was useful for providing intelligence during the Games and, if required, could contribute to the early detection and rapid response to outbreaks during other mass gatherings. The system could be improved to overcome the challenges implied by the findings of this multisource surveillance.

Refining epidemiological forecasts with simple scoring rules.

Estimates from infectious disease models have constituted a significant part of the scientific evidence used to inform the response to the COVID-19 pandemic in the UK. These estimates can vary strikingly in their bias and variability. Epidemiological forecasts should be consistent with the observations that eventually materialize. We use simple scoring rules to refine the forecasts of a novel statistical model for multisource COVID-19 surveillance data by tuning its smoothness hyperparameter.This article is part of the theme issue ‘Technical challenges of modelling real-life epidemics and examples of overcoming these’.

At the tip of an iceberg: citizen science and active surveillance collaborating to broaden the known distribution of Aedes japonicus in Spain

BackgroundActive surveillance aimed at the early detection of invasive mosquito species is usually focused on seaports and airports as points of entry, and along road networks as dispersion paths. In a number of cases, however, the first detections of colonizing populations are made by citizens, either because the species has already moved beyond the implemented active surveillance sites or because there is no surveillance in place. This was the case of the first detection in 2018 of the Asian bush mosquito, Aedes japonicus, in Asturias (northern Spain) by the citizen science platform Mosquito Alert.MethodsThe collaboration between Mosquito Alert, the Ministry of Health, local authorities and academic researchers resulted in a multi-source surveillance combining active field sampling with broader temporal and spatial citizen-sourced data, resulting in a more flexible and efficient surveillance strategy.ResultsBetween 2018 and 2020, the joint efforts of administrative bodies, academic teams and citizen-sourced data led to the discovery of this species in northern regions of Spain such as Cantabria and the Basque Country. This raised the estimated area of occurrence of Ae. japonicus from < 900 km2 in 2018 to > 7000 km2 in 2020.ConclusionsThis population cluster is geographically isolated from any other population in Europe, which raises questions about its origin, path of introduction and dispersal means, while also highlighting the need to enhance surveillance systems by closely combining crowd-sourced surveillance with public health and mosquito control agencies’ efforts, from local to continental scales. This multi-actor approach for surveillance (either passive and active) shows high potential efficiency in the surveillance of other invasive mosquito species, and specifically the major vector Aedes aegypti which is already present in some parts of Europe.Graphical abstract

Towards an integrated animal health surveillance system in Tanzania: making better use of existing and potential data sources for early warning surveillance

BackgroundEffective animal health surveillance systems require reliable, high-quality, and timely data for decision making. In Tanzania, the animal health surveillance system has been relying on a few data sources, which suffer from delays in reporting, underreporting, and high cost of data collection and transmission. The integration of data from multiple sources can enhance early detection and response to animal diseases and facilitate the early control of outbreaks. This study aimed to identify and assess existing and potential data sources for the animal health surveillance system in Tanzania and how they can be better used for early warning surveillance. The study used a mixed-method design to identify and assess data sources. Data were collected through document reviews, internet search, cross-sectional survey, key informant interviews, site visits, and non-participant observation. The assessment was done using pre-defined criteria.ResultsA total of 13 data sources were identified and assessed. Most surveillance data came from livestock farmers, slaughter facilities, and livestock markets; while animal dip sites were the least used sources. Commercial farms and veterinary shops, electronic surveillance tools like AfyaData and Event Mobile Application (EMA-i) and information systems such as the Tanzania National Livestock Identification and Traceability System (TANLITS) and Agricultural Routine Data System (ARDS) show potential to generate relevant data for the national animal health surveillance system. The common variables found across most sources were: the name of the place (12/13), animal type/species (12/13), syndromes (10/13) and number of affected animals (8/13). The majority of the sources had good surveillance data contents and were accessible with medium to maximum spatial coverage. However, there was significant variation in terms of data frequency, accuracy and cost. There were limited integration and coordination of data flow from the identified sources with minimum to non-existing automated data entry and transmission.ConclusionThe study demonstrated how the available data sources have great potential for early warning surveillance in Tanzania. Both existing and potential data sources had complementary strengths and weaknesses; a multi-source surveillance system would be best placed to harness these different strengths.

Automated Cluster Detection of Health Care-Associated Infection Based on the Multisource Surveillance of Process Data in the Area Network: Retrospective Study of Algorithm Development and Validation.

BackgroundThe cluster detection of health care–associated infections (HAIs) is crucial for identifying HAI outbreaks in the early stages.ObjectiveWe aimed to verify whether multisource surveillance based on the process data in an area network can be effective in detecting HAI clusters.MethodsWe retrospectively analyzed the incidence of HAIs and 3 indicators of process data relative to infection, namely, antibiotic utilization rate in combination, inspection rate of bacterial specimens, and positive rate of bacterial specimens, from 4 independent high-risk units in a tertiary hospital in China. We utilized the Shewhart warning model to detect the peaks of the time-series data. Subsequently, we designed 5 surveillance strategies based on the process data for the HAI cluster detection: (1) antibiotic utilization rate in combination only, (2) inspection rate of bacterial specimens only, (3) positive rate of bacterial specimens only, (4) antibiotic utilization rate in combination + inspection rate of bacterial specimens + positive rate of bacterial specimens in parallel, and (5) antibiotic utilization rate in combination + inspection rate of bacterial specimens + positive rate of bacterial specimens in series. We used the receiver operating characteristic (ROC) curve and Youden index to evaluate the warning performance of these surveillance strategies for the detection of HAI clusters.ResultsThe ROC curves of the 5 surveillance strategies were located above the standard line, and the area under the curve of the ROC was larger in the parallel strategy than in the series strategy and the single-indicator strategies. The optimal Youden indexes were 0.48 (95% CI 0.29-0.67) at a threshold of 1.5 in the antibiotic utilization rate in combination–only strategy, 0.49 (95% CI 0.45-0.53) at a threshold of 0.5 in the inspection rate of bacterial specimens–only strategy, 0.50 (95% CI 0.28-0.71) at a threshold of 1.1 in the positive rate of bacterial specimens–only strategy, 0.63 (95% CI 0.49-0.77) at a threshold of 2.6 in the parallel strategy, and 0.32 (95% CI 0.00-0.65) at a threshold of 0.0 in the series strategy. The warning performance of the parallel strategy was greater than that of the single-indicator strategies when the threshold exceeded 1.5.ConclusionsThe multisource surveillance of process data in the area network is an effective method for the early detection of HAI clusters. The combination of multisource data and the threshold of the warning model are 2 important factors that influence the performance of the model.

Incidence of recreational sports-related sudden cardiac arrest in participants over age 12 in a general African population

BackgroundThe incidence of sports-related sudden cardiac arrest (SrSCA) in sub-Saharan Africa is unknown.ObjectiveTo determine the incidence of sudden cardiac arrest (SCA) in non-competitive athletes in an urban population of Cameroon,...

Assessing the accuracy of the death certificate injury at work box for identifying fatal occupational injuries in Michigan.

Death certificates contain a box labeled "Injury at Work" which is to be marked "Yes" for all fatal occupational injuries. The accuracy of this box in Michigan is not fully characterized. The accuracy of the Injury at Work box on the Michigan death certificate was compared to deaths identified from 2001 through 2016 by the Michigan Fatality Assessment and Control Evaluation multi-source surveillance system. The sensitivity was calculated across this time period, while specificity and positive and negative predictive values were derived for 2011-2016. Univariate and multivariate regression were used to examine differences in the sensitivity over time and across demographic variables, industry, and the type of death. We found a sensitivity for the Injury at Work box of 73.1% among 2156 deaths. The sensitivity showed a significant declining trend over the 17 years, from 79.8% to 63.1%. Sensitivity varied significantly across incident type (aircraft, animal-related, drug overdose, motor vehicle, and suicides having particularly lower sensitivities, and electrocutions, falls, and machine-related incidents having higher sensitivities), and industry sector (construction, manufacturing, public safety, transportation, and trade sectors having higher sensitivities, and agriculture and services sectors showing lower sensitivities). Across nearly all categories the sensitivity was significantly below 1. The Injury at Work box on the Michigan death certificate was often incorrectly completed and has become less accurate with time, though the degree of this inaccuracy varies by the industry of the victim and the type of incident.

260 The First Australian Multi-Source Sudden Cardiac Death Registry: Methodology and Early Findings

In 2019, the first multi-source registry of sudden cardiac arrest and death launched in Victoria, Australia. Sudden cardiac arrest / death (SCA/D) affects approximately one thousand younger Victorians per year. The Unexplained Cardiac Death Project enrols all SCA/D cases aged 1-50 years, whilst also offering family screening and access to psychological support through clinical sites and creating a genetic biorepository for whole-genome sequencing. Out of hospital cardiac arrest (OHCA) patients aged 1-50 years in Victoria, Australia are identified by either the existing Victorian Ambulance Cardiac Arrest Registry or the Victorian Institute of Forensic Medicine. Case adjudication is performed, and patients and first-degree relatives with a cardiac/unascertained cause of arrest referred to a clinical site (Royal Melbourne Hospital, St Vincent’s Hospital Melbourne, the Alfred Hospital or the Royal Children’s Hospital). Family screening is conducted, and the opportunity for index cases to contribute to a genetic biorepository offered (deceased cases have samples collected at post-mortem). To date, 509 OHCA patients aged 1-50 have been included in the registry. Of the cases enrolled, 434 (85%) died as a direct result of their cardiac arrest (377 (87%) deceased at scene). Genetic samples have been collected from 17 patients to commence the genetic biorepository. The launch of Australia’s first multi-source cardiac arrest surveillance registry highlights very high mortality in young and middle-aged OHCA patients. The registry will create a genetic biorepository of sudden cardiac arrest survivors and victims that will establish an ongoing Australian resource.

Multisource surveillance video coding with synthetic reference frame

Due to the increasing growth of surveillance data, high-efficiency surveillance video coding schemes are demanded. However, the existing conventional coding framework has difficulties in handling rotation and zooming whilst the recently proposed multisource surveillance video coding method is inflexible and prone to be affected by pose errors. To this end, we propose to combine conventional surveillance video coding and multisource surveillance video coding into one unified framework. First, global knowledge in the form of 3D model and initial textures is employed to construct a knowledge based reference frame. Meanwhile, a temporal reference frame is also generated from the reconstructed frames in decoded picture buffer. Then, they are fused to obtain the synthetic reference frame to exploit global and local information. Finally, we add the synthetic reference frame into reference picture list and rearrange the list to optimize the overall efficiency. Experimental results demonstrate the effectiveness of the proposed method over state-of-the-art anchors.

Adolescent exposures to traditional and novel psychoactive drugs, reported to National Poison Data System (NPDS), 2007–2017

BackgroundSurvey data on adolescent drug use trends have limitations evaluating the impact of drug use on health and may lag current trends. The objective was to describe National Poison Data System (NPDS) trends, medical outcomes, and healthcare encounters from adolescent exposures of traditional and novel psychoactive drugs. MethodsRetrospective review of adolescent (10–18 years of age) exposure calls to all U.S. poison centers, from January 1, 2007 through December 31, 2017, using generic codes for traditional and novel psychoactive drugs. Descriptive statistics and univariate Poisson regression modeling were used for analysis. ResultsThere were 49,757 exposure calls for the included psychoactive drugs. The median age was 16 years (IQR 15,17), 64% were male, and the majority were evaluated in a healthcare facility (92%). Marijuana had the most exposure calls (36.6%), followed by synthetic cannabinoids (e.g., spice; 21.3%). There were 181 (<1%) deaths; the highest fatality rates were from fentanyl, 2C drugs (phenylethylamine derivatives), and heroin. LSD exposure calls have had the most significant increase over the past 10 years. ConclusionU.S. Poison Centers reported almost 50,000 exposure calls and 181 deaths over 10 years for adolescent exposures of both traditional and novel psychoactive drugs, demonstrating the significant health impact on this vulnerable population. Opioids and 2C drugs contributed to the highest mortality rates, moderate/major symptoms and healthcare utilization, and LSD had the most significant increase in calls. Multi-source surveillance methodology is critical in understanding the public health impact on drug abuse in the adolescent population.

Multisource surveillance for non-fatal work-related agricultural injuries

ABSTRACTObjectives: Development of a state-wide comprehensive surveillance system for non-fatal work-related farm injuries, since non-fatal injuries that occur to the self-employed (i.e., many farm owners/operators), family workers, federal government workers and small farms with fewer than 11 employees are not included in the Bureau of Labor Statistics employer-based survey used to produce the U.S. National statistics of work-related injuries.Methods: In 2015 and 2016, inpatient discharge summaries, emergency department, and hospital-based outpatient clinic records from all 134 of Michigan’s hospitals with ICD-9 codes 989.0-.1, E827.0-.9, E849.1, E906.8, E919.0 or ICD-10 codesT65.0-.1, V80, Y92.7, W55.1-.4, W30 were reviewed to identify non-fatal work-related farm injuries.Results: We identified 1,559 non-fatal work-related farm injury incidents that occurred in 1,525 individuals, with 74% being among men. The most common parts of the body injured were an upper limb (38.2%) and a lower limb (23.7%). The most common types of injury were contusions (26.4%) and fractures (19.9%). Owners/operators accounted for 44.1% and hired hands for 42.9% of individuals injured. Injuries caused by cows were the predominant cause: 472 (31.5%) of all the injuries. Dairy farms accounted for 39.6% of all cases for which the farm type was recorded.Conclusion: A comprehensive system to identify non-fatal work-related farm injuries among all individuals who work on a farm, including owner/operators, family members and migrant and seasonal farm laborers, was implemented using hospital, emergency department and hospital-based outpatient clinic medical records. Such a system is important to be able to identify hazards and target prevention.

Trends in documented co-occurring conditions in children with autism spectrum disorder, 2002–2010

BackgroundAutism spectrumdisorder (ASD) commonly presents with co-occurring medical conditions (CoCs). Little is known about patterns in CoCs in a time of rising ASD prevalence. AimsTo describe trends in number and type of documented CoCs in 8-year-old children with ASD. MethodsWe used Autism and Developmental Disabilities Monitoring Network (ADDM) data, a multi-source active surveillance system monitoring ASD prevalence among 8-year-old children across the US. Data from surveillance years 2002, 2006, 2008, and 2010 were used to describe trends in count, categories, and individual CoCs. ResultsMean number of CoCs increased from 0.94 CoCs in 2002 to 1.06 CoCs in 2010 (p < 0.001). The percentage of children with ASD with any CoC increased from 44.5% to 56.4% (p < 0.001). CoCs with the greatest increases were in general developmental disability (10.4% to 14.5%), language disorder (18.9% to 23.6%), and motor developmental disability (10.5% to 15.6%). Sex modified the relationship between developmental (P = 0.02) and psychiatric (P < 0.001) CoCs and surveillance year. Race/ethnicity modified the relationship between neurological conditions (P = 0.04) and surveillance year. ConclusionsThe increase in the percentage of children with ASD and CoCs may suggest the ASD phenotype has changed over time or clinicians are more likely to diagnose CoCs.

Historical perspective on folic acid and challenges in estimating global prevalence of neural tube defects.

Neural tube defects (NTD) are major congenital malformations affecting births worldwide. NTD are associated with life-long disability, significant medical care costs, and child mortality. Their prevalence varies worldwide. We conducted a review of published literature and surveillance systems to examine challenges in estimating an overall global prevalence estimate for NTD. Our review showed that most low- and middle-income countries do not track NTD and indicate a high prevalence of these malformations where data are available. Challenges in global NTD prevalence estimation include (1) quality of surveillance methods, (2) existing risk factors (including geographic or socioeconomic factors, availability and use of folic acid, and racial-ethnic and genetic factors), and (3) limitations in education and access to care. We recommend population-based surveillance systems tracking all pregnancy outcomes and major risk factors. Countries should invest in sustainable, multisource surveillance systems, in parallel to folic acid interventions, for gaining a more accurate knowledge of global prevalence of NTD than we currently have. Such efforts will assist in both global prevention of NTD and periodic evaluation of folic acid interventions for NTD reduction. Global NTD prevalence data can drive political will and accelerate the implementation and evaluation of NTD prevention programs.

Sport-related sudden cardiac arrest in black African non-competitive athletes: Incidence in a general population of Cameroon

Incidence estimates of sudden cardiac arrest (SCA) in athletes living in sub-Saharan Africa (SSA) are unknown. Over 12 months, the household administrative office and health community committee within neighbourhoods in two health areas of Douala (Cameroon) registered all deaths among 86,188 inhabitants aged under 18 years. As part of an extended multi-source surveillance system, all sport fields, the Emergency Medical Service (EMS), local medical examiners and district hospital mortuaries were also surveyed. Whereas two physicians investigated every natural death, two cardiologists reviewed all unexpected natural deaths. There were 288 all-cause deaths and 27 (9.4%) were SCD. The crude incidence rate was 31.3 [95% confidence interval (CI): 20.3–40.6]/100,000 person–years. The age-standardized rate by the African standard population was 33.6 (95% CI: 22.4–44.9)/100,000 person–years. Death occurred at night in 37% of cases, including 11% of patients who died while asleep. Out-of-hospital SCA occurred in 63% of cases, 55.5% of which occurred at home. Of the 88.9% cases of witnessed CA, 63% occurred in the presence of a family member and cardiopulmonary resuscitation was attempted only in 3.7%. While running, death occurred in 1 (3.7%) non-competitive female athlete aged 35 years. Sport-related SCA in this African population claimed an annual incidence of 1.15 cases per 100,000 person–years, which is in accordance with reported rates in the literature. Long-term cohort studies in many African countries are warranted to better understand incidence estimates of SCA in black athletes.

Multi-source surveillance for work-related crushing injuries.

Work-related crushing injuries are serious but preventable. For 2013 through 2015, the U.S. Bureau of Labor Statistics' (BLS) Survey of Occupational Injuries and Illnesses (SOII) reported 1260 crushing injuries in Michigan. In 2013, Michigan initiated multi-data source surveillance of work-related crushing injuries. Records from all 134 of Michigan's hospitals/emergency departments (EDs), the Workers Compensation Agency (WCA) and Michigan's Fatality Assessment Control and Evaluation (MIFACE) program were used to identify work-related crushing injuries. Companies, where individuals were hospitalized or had an ED visit for a crushing injury, potentially had an OSHA enforcement inspection conducted. From 2013 through 2015, there were 3137 work-related crushing injury incidents, including two fatalities. The Michigan OSHA program completed inspections at 77 worksites identified by the surveillance system. The Michigan multisource surveillance system identified two and a half times more crushing injuries than BLS and was useful for initiating case-based enforcement inspections.

Epidemiology of sudden cardiac death in Cameroon: the first population-based cohort survey in sub-Saharan Africa.

BackgroundIncidence estimates of sudden cardiac death (SCD) in sub-Saharan Africa (SSA) are unknown.MethodOver 12 months, the household administrative office and health community committee within neighbourhoods in two health areas of Douala, Cameroon, registered all deaths among 86 188 inhabitants aged >18 years. As part of an extended multi-source surveillance system, the Emergency Medical Service (EMS), local medical examiners and district hospital mortuaries were also surveyed. Whereas two physicians investigated every natural death, two cardiologists reviewed all unexpected natural deaths.ResultsThere were 288 all-cause deaths and 27 (9.4%) were SCD. The crude incidence rate was 31.3 [95% confidence interval (CI): 20.3–40.6]/100 000 person-years. The age-standardized rate by the African standard population was 33.6 (95% CI: 22.4–44.9)/100 000 person-years. Death occurred at night in 37% of cases, including 11% of patients who died while asleep. Out-of-hospital sudden cardiac arrest occurred in 63% of cases, 55.5% of which occurred at home. Of the 88.9% cases of witnessed cardiac arrest, 63% occurred in the presence of a family member and cardiopulmonary resuscitation was attempted only in 3.7%.ConclusionThe burden of SCD in this African population is heavy with distinct characteristics, whereas awareness of SCD and prompt resuscitation efforts appear suboptimal. Larger epidemiological studies are required in SSA in order to implement preventive measures, especially in women and young people.

Linking Death Certificates, Postmortem Toxicology, and Prescription History Data for Better Identification of Populations at Increased Risk for Drug Intoxication Deaths

Our objective was to assess and improve the completeness of drug-related information, to evaluate the risk for drug intoxication deaths associated with specific drugs among different demographic groups, and to show the benefit of multi-source surveillance to better target interventions and to inform policy evaluation. This study linked death certificate, prescription drug monitoring program, and postmortem toxicology data for Kentucky residents aged ≥18 years who died from drug intoxication during 2013–2014. The most frequently listed contributing drugs were pharmaceutical opioids (60%), alprazolam (35%), and heroin (29%) (not mutually exclusive). The highest death rates from fentanyl (4 per 100,000 Kentucky population aged ≥18 years) and heroin (13 per 100,000) intoxication were among those aged 25–34 years, whereas the highest death rates associated with prescription opioids (excluding morphine and fentanyl) were among those aged 35–54 years (15 per 100,000). Appalachian county residents had a significantly higher intoxication fatality rate (34 per 100,000) than residents from non-Appalachian counties (27 per 100,000). One-quarter of death certificates did not list any specific contributing drug, but postmortem toxicology reports were available for 89% of these cases and provided information on drugs used before death. Analysis of decedents’ controlled substance prescription records suggested drug diversion (e.g., only 36% of the decedents who tested positive for oxycodone had recently filled prescriptions). The abuse and diversion of drugs significantly impact the health and safety of Kentucky citizens. This study demonstrated the benefit of integrated multi-source drug intoxication fatality surveillance to identify drug misuse and diversion and population groups at high risk for drug poisoning mortality and to inform targeted prevention programs.

Knowledge-Based Coding of Objects for Multisource Surveillance Video Data

Global object redundancy (GOR), as opposed to local spatial/temporal redundancies in a single video clip, is a new form of redundancy common in multisource surveillance video data (MSVD). GOR is induced by the repetition of foreground objects across multiple cameras, and becomes influential as the number of objects increases. Eliminating GOR considerably improves MSVD coding efficiency. In an effort to accomplish this, this study first proposes a knowledge-based representation of objects based on careful analysis of GOR composition. The representation contains a constant part and a variational part: the former is used to represent the common knowledge shared by an object across multiple cameras, while the latter is used to represent local variations on the object's surfaces. Based on the proposed representation, a knowledge-based coding (KBC) method is then proposed in which each foreground object is encoded with a hybrid prediction scheme, where the constant part of the object is generated via global prediction from a model library and the variational part is predicted via local reference frames with pose-based, short-term prediction. Experimental results showed that the KBC method saves more than 39% bits on average for encoding foreground objects in high-resolution video clips (compared to 16% for the entire videos). Applying the proposed coding method to surveillance videos in large spatial and temporal scale allows storage savings at the PB level.

Epidemiological situation of Lyme borreliosis in Belgium, 2003 to 2012

BackgroundSome studies show that the incidence of Lyme borreliosis is increasing in different European countries. In order to evaluate if this is also the case in Belgium, different data sources were consulted to describe the epidemiology of Lyme borreliosis in the country during the last decade.MethodsData from two databases were analyzed for the time period 2003–2010 and 2003–2012 for respectively: the registration of minimal clinical data from Belgian hospitals (principal and secondary diagnosis), and a sentinel laboratory network reporting positive laboratory results.ResultsThe number of hospitalized cases per year remained stable between 2003 and 2010, ranging from 970 (in 2008) to 1453 (in 2006), with a median of 1132.5 cases per year. Between 2003 and 2012, yearly fluctuations in the number of positive tests were reported by the sentinel laboratory network (with a minimum of 996 positive tests in 2007 and a maximum of 1651 positive tests in 2005), but there is no increasing trend over the study period (median = 1200.5 positive tests per year). The highest incidence rates of hospitalization and the highest reported incidence of positive laboratory results are registered in the provinces of Luxemburg, Limburg, Flemish Brabant and Antwerp, with a typical seasonal pattern (peak in September). The age groups affected most are those from 5 to 14 years and 45 to 69.ConclusionBased on hospital records and laboratory results, no increasing trend in Lyme disease was observed over the 2003–2012 period in Belgium. These results are in line with the stable incidence of erythema migrans reported by a sentinel network of general practitionners between 2003 and 2009. Multi-source surveillance of vector-borne diseases should be further implemented.