Lyme disease is becoming a “hot topic” in Western countries. In Europe, an increase in Lyme borreliosis has been described in some countries, but not in others [5]. This is also the case in the countries neighboring Belgium. The Netherlands report an increase in consultations and hospital admissions for Lyme borreliosis between 1994 and 2009 [9, 10]. Eastern Germany, on the other hand, observes a decrease in Lyme borreliosis between 2009 and 2012 [13]. French reports show no increase in hospitalizations between 2004 and 2009, and also a stability in yearly Lyme borreliosis incidence rate (data from the general practitioners sentinel network) between 2009 and 2012 [14]. Switzerland describes a stable reported incidence between 2008 and 2011 [15]. However, comparisons within and between countries need to be looked at with caution, as surveillance methodology varies by country and other factors might explain some variations. For example in Belgium, a gradual increase in the number of Lyme borreliosis positive tests was observed during the beginning of the nineties, following the introduction of ELISA test reimbursement.
In Belgium, surveillance of Lyme disease occurs through a sentinel network of laboratories (SNL) reporting the number of positive laboratory tests since 1987 [21]. The reliability of the network is based on a stable participation of laboratories, covering around 50 % of all Lyme borreliosis tests carried out in Belgium (Berger N., unpublished observations), with weekly automatic extraction of data for more than 70 % of participating laboratories. The network allows to follow trends in incidence of laboratory confirmed tests, but this is only a partial picture of the incidence of the disease. In the context of increasing incidence of Lyme borreliosis in some European countries, the routine surveillance data were validated with information collected through the registration of minimum clinical (RMC) data from the hospitals.
Both data sources converge to the same result: we do not observe any increase in Lyme borreliosis in Belgium during the study period. However, early clinical manifestations of Lyme diseases, such as erythema migrans, are not covered by these surveillance systems, since laboratory testing is not recommended in an early stage and those manifestations do not require hospitalization. Yet, results of two studies carried out by the Belgian sentinel network of general practitioners did not show a significant increase of the incidence rate of tick bites and erythema migrans between 2003 and 2009 [17]. The incidence of erythema migrans per 10,000 patients in 2003–2004 and in 2008–2009 was 8.32 and 9.02 respectively (p > 0.05). Our SLN and RMC observations are congruent with those clinical results.
Despite a global stable trend during the 2003–2010/2003–2012 period, RMC and SLN data show yearly fluctuations. Climate changes, with higher temperatures during winter months and increased humidity, or the presence of snow, impact on tick abundance and can explain some of these yearly variations [22, 23]. Further studies are needed to explore the influence of climate on Lyme disease in Belgium. The observed difference in fluctuations over time between the SLN and RMC data could be due to the fact that part of the hospitalizations are due to late manifestations of Lyme borreliosis, which may occur years after the tick bite. As RMC only provides the diagnosis without any precision regarding the type of complication, we lack information to confirm or invalidate this hypothesis.
Other factors than climate can also impact on Lyme disease incidence. The number of positive laboratory tests depends on the prescription practice of physicians, which is influenced by the reimbursement of tests, the physician’s awareness of the disease, and individual insistence to be tested. The latters are both influenced by various factors, including media. However, during the study period, the positivity rate of tests carried out by the SNL remained stable (1.5 to 3.5 % of total tests) (WIV-ISP, unpublished data), and no statistical difference was observed in incidence of hospitalization and positive laboratory results between the different time periods.
The monthly distribution of RMC and SLN cases follows a seasonal pattern, with a very small number of cases during the winter, a gradual increase during the spring and summer periods, and a maximum peak during the end of the summer and the autumn. This is congruent with Ixodes ricinus phenology [23]. The number of nymphs collected in Belgium in 2009 and 2010 increased gradually during the spring, to reach a maximum in July (in 2009) and in June (in 2010) [23]. The peak of the number of hospitalizations and positive laboratory results is reached two to three months later, which corresponds to the time of onset of the first complications after a tick bite (early disseminated Lyme borreliosis).
The highest cumulative incidence of hospitalizations, and the highest incidence of inhabitants with positive laboratory tests during the study period are registered in the province of Luxemburg. This province has different geographical and climate characteristics compared to other Belgian provinces that can influence on tick density, with a higher proportion of forests, a higher altitude (around 500 meters versus sea level) and a wetter and colder climate, with snow during the winter. Other provinces that are most affected by Lyme borreliosis are Limburg, Antwerp and Flemish Brabant. These provinces comprise small field and forest areas, alternating with high densely populated urban areas, which can also impact on tick density [22, 23]. Although the geographical distribution presented here is based on the place of residence of the patient, which is not necessarily the place of infection, the distribution in our study is in line with reported results of tick collection [23]. Indeed in 2009 and 2010, high nymph densities were reported in the provinces of Flemish Brabant (49.0 to 59.4 nymphs/100 m2), Limburg (31.4 to 56.6 nymphs/100 m2) and Luxemburg (12.1 to 17.2 nymphs/100 m2).
The individual characteristics of RMC and SLN cases globally match the literature. A higher incidence in men than in women has previously been described [17], and could be due to a higher occupational risk, and also be related to certain leisure activities. Regarding the age distribution, the literature describes two groups mainly affected: children from 5 to 14 years, and adults from 50 to 64 years [5], as observed in our study.
Although the two data sources used here report comparable results, they both have strengths and limitations. The RMC database is based on compulsorily registration of all Lyme borreliosis hospitalizations from every general hospital in Belgium and is therefore expected to be exhaustive. Its main limitations are the delay needed to get the data, and a lack of precision regarding the symptoms, the ICD 9 code making no distinction between clinical manifestations. Moreover, since the data collected represent both principal and secondary diagnosis from all hospitalizations, the hospitalization incidences presented in this study do not reflect the severity of Lyme disease in Belgium but are rather used to appraise its trends.
The sentinel laboratory network is considered to be stable, and exists since more than 30 years. It nevertheless comprises several limitations. Firstly, if the network is representative at national level, its geographical repartition at provincial level is uneven. For example, the coverage of East Flanders and Flemish Brabant is high (between 80 and 90 %), whereas Namur, Liege and Limburg have a lower coverage (below 50 %). The reported incidence may thus be underestimated in the provinces with a lower coverage. Secondly, it gives a partial picture of the incidence, as laboratory tests are not recommended for patients presenting with an erythema migrans. Thirdly, a window period of undetectable antibodies should be considered if the blood analysis occurs within the first three weeks after the tick bite. Fourthly, laboratory tests only confirm the presence of anti-Borrelia antibodies, which does not necessarily means that the patient is suffering from Lyme disease: it could also be due to a previous symptomatic or asymptomatic Borrelia infection. Fifthly, the database may content some duplicates, as a patient is considered to be eligible for reinfection after a one-year period. However, as the database is repetitively used with the same parameters, this phenomenon should not have any impact on its global trend, which is of interest and regularly followed in public health.
Finally, we must note that routine SLN monitoring shows a new peak in positive tests for the year 2013, but the positivity rate remains stable. Further follow-up and surveillance remains essential.