Abstract
In Europe, Lyme neuroborreliosis (LNB) is the most severe manifestation of Lyme borreliosis and has recently been added to the communicable disease surveillance list for EU/EEA by the European Commission. In Northern Europe, LNB is primarily caused by the spirochete Borrelia garinii and transmitted by the tick Ixodes ricinus. This Danish observational epidemiologic case-control study includes every identified LNB patient (n = 401) on Funen, Denmark, from 1995-2014. We display spatial and temporal LNB incidence variation, seasonal distribution of cases and local spatial case clustering. Seasonal patterns show LNB symptom-onset peaking in July and a significant seasonal difference in number of cases (p < 0.01). We found no significant change in seasonality patterns over time when dividing the study period into 5-year intervals. We identified a significant local geographical hot-spot of cases with a relative risk of 2.44 (p = 0.013). Analysis revealed a significantly shorter distance to nearest forest for cases compared with controls (p < 0.001). We present a novel map of the focal geographical distribution of LNB cases in a high endemic borreliosis area. Continued studies of case clustering in the epidemiology of LNB are of key importance in guiding intervention strategies.
Similar content being viewed by others
Introduction
Lyme Borreliosis (LB) is the most common tick-borne infection in Denmark and in Europe1,2,3. It is a spirochetal infection caused by Borrelia burgdorferi sensu lato (sl), which in Denmark is transmitted by the tick Ixodes ricinus2. This tick species is found throughout the country, but it is most abundant in the eastern and central parts of Denmark1. LB can manifest with neurological symptoms, called Lyme neuroborreliosis (LNB), the most severe form of LB. A previous study has strongly suggested a correlation between variations in tick density and LNB incidence in Denmark1.
Primarily, four genospecies of B. burgdorferi sl are known to be associated with human disease. B. burgdorferi sensu stricto (ss), B. afzelii, B. garinii and B. bavariensis. In Europe, the predominant species are B. afzelii, B. garinii and B. bavariensis3,4.
Infections with different genospecies often result in different clinical manifestations. B. garinii primarily causes LNB with symptoms like lymphocytic meningitis, painful radiculitis and cranial neuropathy (particularly facial nerve palsy). B. afzelii is mostly associated with skin manifestations such as erythema migrans and acrodermatitis chronica atrophicans3,4. Different Borrelia genospecies have different preferred reservoir hosts, and thus the distribution of clincal manifestations may vary. The incidence of LB in Europe has increased over the past few years2. In a recognition of this, the European Commision has in 2018 amended LNB to the communicable disease surveillance list5, in an effort to monitor the epidemiology in order to support measures to prevent and control the disease and the following complications. In Denmark, the LNB incidence was found to be 3.2/100,000 population when the national microbiology database (MiBa) was used for surveillance6, while our research group have found a higher incidence of 4.76/100.000 in the area of Funen7.
Humans living in regions with competent hosts of I. ricinus are at higher risk of disease, as these may serve as reservoirs hosts for various pathogens that can be transmitted by tick bites to humans8. Although the distribution and abundance of ticks are highly impacted by climate and landscape9,10, abundance of host species also affect the presence and abundance of ticks10. Among other species, the European roe deer (Capreolus capreolus) is an important tick host and previous studies have found a correlation between tick abundance and roe deer abundance at a local scale. Thus, changes in the roe deer population may alter the number of ticks in the following seasons11. Local variations in reservoir host animal numbers can however also affect local difference in Borrelia genospecies domination3,12. The risk of acquiring LNB is thus a complex interplay between Borrelia reservoir host distribution and tick abundance.
The primary objectives of this observational study were to (1) describe both the spatial and temporal LNB incidence variation, and examine any change in seasonal distribution over the last 20 years, and (2) identify potential spatial patterns of LNB-cases on Funen, and quantify difference in distance to nearest forest between cases and controls based on home addresses.
Method
Study population
A former study of every available patient chart from Funen, Denmark in the period 01.01.1995 to 31.12.2014, uncovered 431 patients with a LNB diagnosis7. A diagnosis was made if the patient had clinical symptoms of LNB and a positive Borrelia intrathecal antibody index test (IgM and/or IgG) performed at the Department of Microbiology, Odense University Hospital6. Of these 431 patients, 401 were included in this study (Fig. 1).
We extracted case addresses and the date of symptom-onset from the case database. At extraction, case addresses were scrambled, by randomly changing the house number to either +1, no change, or −1, due to guidelines regarding clinical research issued by the Danish National Committee on Health Research Ethics13.
The control addresses were obtained by extracting a list of every residential address in each of the 10 municipalities of Funen from the publicly available national address database14. Among the 238,184 extracted control addresses, we randomly chose 4001 using the RAND-function in Excel (Fig. 1).
Statistical analysis
Incidence and regional mapping
The annual LNB incidence rate (IR) of the region of Funen was calculated from the publicly available municipality population numbers15. However, as the official population numbers from 1995-96 were not available, IR could only be calculated from 1997-2014. The Edwards test was used to test for seasonality in month of symptom debut16. To test for significant differences in monthly distribution of cases in four 5-year time periods (1995–1999, 2000–2004, 2005–2009 and 2010–2014), the Kruskal-Wallis test for non-normally distributed data was used. The analyses were carried out using STATA version 15.0. A p-value <0.05 was considered statistically significant.
Local spatial clustering
We used ArcMap 10.1 ESRI. Redlands, CA, a program used to manage geographic data17, and to run an initial IDW interpolation to help us to visualize areas with potential high or low clustering of both cases and controls. The search radius was set to 5000 meters, as Funen is of limited size with many small forest areas, and we deemed this a reasonable distance that an individual would regularly travel away from their home address (going for a walk, walking the dog).
We performed a purely spatial analysis to test for and to identify local level clustering using the software SaTScan18 after transforming the address coordinates to the Universal Transverse Mercator coordinate system (UTM). The analysis included scanning for both circular and elliptic shaped clusters, containing significantly high/low rates (hot/cold spots) of cases, using the Bernoulli probability model19.
Distance to the nearest forest
We created a new 1 ×1 km raster layer of the CORINE Land Cover classification20 with only forested areas on Funen (Supplement S1). For each case and control address, we used the Spatial Analyst tool in ArcMap ESRI. Redlands, CA to calculate the Euclidian distance to the nearest forest pixel for both cases and controls. To account for spatial autocorrelation of data points, we created a 3×4 grid and overlayed it to our study area. We then extracted grid id for each of the cases and controls, and used this grid id as a random effect in a mixed model logistic regression21(see Supplement S2) to identify any increase in the probability of becoming a case rather than a control when moving one km closer to a forest area.
Ethical considerations
This study was approved by the Danish Data Protection Agency (j.nr. 2008-58-0035) and the Danish Health and Medicines Authority (j.nr. 3-3013-631/1/).
Results
Seasonal variation of symptom-onset
We found a seasonal pattern, with the number of cases starting to increase in May and peaking in July. There was a statistically significant seasonal variability of cases (p < 0.01). Divided into four seasons the majority had symptom debut in late summer in July, August and September (n = 230, 57%) compared to only 6.7% (n = 27) in the months of January, February and March (p < 0.001). When displaying seasonal variation in 5-year intervals we found the same pattern with symptom-onset peaking in July based on total numbers of cases, except in 2000–2004 where symptom-onset peaked in August, by a margin of two cases (Fig. 2). We found no statistically significant differences in monthly distribution of cases between the four 5-year intervals (all p-values > 0.05, results displayed in Supplement S3).
Incidence rate and regional mapping
We found the annual LNB incidence rates on Funen varying from 2.33 (1998) to 7.93 (2006) (Fig. 3). The incidence rate at zip-code level is displayed in Fig. 4 22. The ticks illustrate the average tick density at individual locations collected by monthly flagging from April-November 2002 in a previous study by Skarphedinsson23.
Local spatial clustering
Interpolation of the LNB-cases and controls showed areas with apparent high and low density of cases, with high incidence areas visually appearing to be correlated with the distribution of forest areas. The densely populated urban area of Odense in the middle of the main island appeared as an area with relatively few cases. West of Odense was a large elongated area with a relatively high proportion of cases (Fig. 5). The SaTScan18 analysis detected a significant local ellipsoid hot-spot with a relative risk of 2.44 (p = 0.013). This significant ellipse-shaped cluster overlapped the area west of Odense idicated by interpolation24. No other areas of apparent high or low densities of cases indicated by the IWD were detected as significant (5% significance level) clusters by the SatScan analysis.
Distance to the nearest forest
The cumulative frequency graph of the distance to the nearest forest illustrates a significantly shorter distance for cases compared to controls. A regression analysis found that the odds of being a case rather than a control increased with approximately 8% (Odds ratio 1.08) when moving one km closer to a forest (p < 0.001) (Fig. 6, Table 1).
Discussion
In this large retrospective study of a well-defined case population of LNB, we aimed to examine the temporal changes in LNB incidence. The average LNB incidence rate of 4.76/100 000 inhabitants found in our cohort was higher than previously described in a Danish setting6,25,26. This increase in incidence rate supports the hypothesis suggested by Dessau et al. that the incidence found in previous Danish studies has been underestimated6. Our results suggest a bias in the previous national incidence numbers due to a lack of case reporting. Since LNB can have consequences, not only for the individual patient but also socioeconomically27, correct reporting is essential to strengthen future LNB surveillance.
As for our aim to examine any seasonal changes in LNB incidence, we found a significant seasonal variation with most cases having their symptom-onset in July. This is earlier than previous findings in studies from our neighbouring countries where symptom-onset peak has been in August28,29. This finding is important for clinicians, to increase awareness of earlier seasional LNB onset.
We divided seasonal variation data into 5-year intervals to investigate if changes in the seasonal distribution could be correlated to changes in the European climate, as discussed by Lindgren et al. and Rizzoli et al.30,31,32. Seasonal distribution of casesdid not differ significantly over time during the 20-year study period. However, it is possible that 20 years of data is too short of a timeline to detect any seasonal shift in the number of monthly LNB cases due to climate change.
We aimed to identify any spatial patterns of LNB cases on Funen. We found that LNB incidence on Funen displayed a great spatial variation being higher in rural areas, particularly near forest areas. We found an elliptic shaped cluster on mid-western Funen with a significant increase in relative risk of LNB. This cluster visually appeared to be correlated to the distribution of forest areas, and may also be related to regional differences in the distribution of reservoir-hosts carrying different genospecies of Borrelia33. For the general practitioner knowledge of potiental high risk areas of LNB is of importance, and may also be utilized in prophylactic measures. These findings also underline the importance of a One Health approach to tick-borne diseases, where human disease is looked at in relation to both veterinary and the environmental factors34.
The overall LNB incidence in the main urban area was not significantly lower than the rural areas although the IDW interpolation indicated lower prevalence in the main city of Odense. Interestingly, a large number of cases were reported from the main Odense area. Although some of these cases may result from tick bites in urban areas it is likely that most tick-bites are aquired when visiting surrounding rural areas. If the risk of LNB is correlated to forest exposure-time (hours) alone, we would expect to observe a greater difference in relative risk between an urban and rural population. Given our results, one could ask if the risk of infection in urban areas is higher than previously assumed. This should call for an increased awareness of LNB symtoms among general practitioners and hospital docters not only in rural but also in urban areas.
The distance to the nearest forest was significantly shorter for cases compared to controls. The odds ratio of being a case increased with 7% when moving one km closer to a forest. This indicates, not unexpectedly, that the forest areas have an important impact on the risk of LNB infection.
The strengths of this study are the clear LNB case definition and complete data on all included patients. Other strengths are the robust surveillance data programs providing reliable geographical data from the study area. The process of adjusting the control addresses might have given a small selection bias. The authors do not know if every address included in the control group was inhabited. However, we do know that the total percentage of uninhabited addresses on Funen was low at time (6,4%)35, which limits this potential bias.
Future studies in this area should focus on identifying better local data-classification of landcover variables and variables describing temporal and spatial variation in animal abundance. This will help predicting areas with a high possibility of LNB case presence, as well as regional incidence trend variations as previously discussed by Messier et al.36. These variables can be used in models predicting areas with a higher risk of LNB infection and increase the accuracy of risk maps. A future local risk-map could be used as a tool by general practitioners in high-risk areas as well as increase awareness of LNB and LNB symptoms in people living in these areas. This could potentially shorten diagnostic and treatment delay and thereby reduce the risk of longterm sequelae7,37. Collaborative preventive actions on a European level creating a European model for predicting LNB high-risk areas would be of public health interest. Planned future changes in agriculture affecting land and forest areas could then be analysed in regards to its impact on known environmental variables e.g. roe deer density that increase/decrease the risk of tickborne infections, especially in locations close to inhabited areas.
In conclusion, we found a clear seasonal pattern of distribution of LNB cases, but no seasonal changes over the 20-year study period. we identied a significant clustering of LNB-cases in the mid-western part of the island of Funen, which represents an area with increased risk of LNB. The LNB cases resided significantly closer to forests compared with controls, indicating greater exposure to I. ricinus in forest areas.
References
Jensen, P., Hansen, H. & Frandsen, F. Spatial risk assessment for Lyme borreliosis in Denmark. Scandinavian journal of infectious diseases 32, 545–550 (2000).
Stanek, G., Wormser, G., Gray, J. & Strle, F. Lyme borreliosis. Lancet (London, England) 379, 461–473, https://doi.org/10.1016/s0140-6736(11)60103-7 (2012).
Strle, F., Ruzic-Sabljic, E., Cimperman, J., Lotric-Furlan, S. & Maraspin, V. Comparison of findings for patients with Borrelia garinii and Borrelia afzelii isolated from cerebrospinal fluid. Clinical infectious diseases: an official publication of the Infectious Diseases Society of America 43, 704–710, https://doi.org/10.1086/506936 (2006).
Stanek, G. & Reiter, M. The expanding Lyme Borrelia complex–clinical significance of genomic species? Clinical microbiology and infection: the official publication of the European Society of Clinical Microbiology and Infectious Diseases 17, 487–493, https://doi.org/10.1111/j.1469-0691.2011.03492.x (2011).
European Centre for Disease Prevention and Control, https://ecdc.europa.eu/en/news-events/ecdc-comment-european-commission-updates-communicable-disease-surveillance-list-lyme.
Dessau, R., Espenhain, L., Molbak, K., Krause, T. & Voldstedlund, M. Improving national surveillance of Lyme neuroborreliosis in Denmark through electronic reporting of specific antibody index testing from 2010 to 2012. Euro surveillance: bulletin Europeen sur les maladies transmissibles = European communicable disease bulletin 20 (2015).
Knudtzen, F., Andersen, N., Jensen, T. & Skarphedinsson, S. Characteristics and Clinical Outcome of Lyme Neuroborreliosis in a High Endemic Area, 1995-2014: A Retrospective Cohort Study in Denmark. Clinical infectious diseases: an official publication of the Infectious Diseases Society of America 65, 1489–1495, https://doi.org/10.1093/cid/cix568 (2017).
Schotthoefer, A. & Frost, H. Ecology and Epidemiology of Lyme Borreliosis. Clinics in laboratory medicine 35, 723–743, https://doi.org/10.1016/j.cll.2015.08.003 (2015).
Kjaer, L. J. et al. Predicting and mapping human risk of exposure to Ixodes ricinus nymphs using climatic and environmental data, Denmark, Norway and Sweden, 2016. Euro Surveill 24, https://doi.org/10.2807/1560-7917.ES.2019.24.9.1800101 (2019).
Tack, W., Madder, M., Beaten, L. & Vanhellemont, M. Local habitat and landscape affect Ixodes ricinus abundances in forests on poor, sandy soils. Forest ecology and management 265, 30–36 (2012).
Andersen, N. S. et al. Reduction in human Lyme neuroborreliosis associated with a major epidemic among roe deer. Ticks and Tick-Borne Diseases 9, 379–381 (2018).
Vollmer, S. A. et al. Host migration impacts on the phylogeography of Lyme Borreliosis spirochaete species in Europe. Environ Microbiol 13, 184–192, https://doi.org/10.1111/j.1462-2920.2010.02319.x (2011).
The National Committee on Health Research Ethics, http://www.nvk.dk/emner/information-og-samtykke-i-forsoeg/regler-og-retningslinjer.
Suiten., A. Download af kommuners, regioners og Danmarks adgangsadresser, http://download.aws.dk/adgangsadresser (2016).
Danmarks Statisik. Befolkning og valg, Folketal, http://www.statistikbanken.dk/statbank5a/default.asp?w=1440 (2016).
Edwards, J. H. The recognition and estimation of cyclic trends. Ann Hum Genet 5, 83–87 (1961).
ArcMap. ESRI 2011. ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems Research Institute. (2011).
Kulldorff, M. A spatial scan statistics. Communications in Statistics - Theory and Methods 26, 1481–1496 (1997).
Coleman, M., Colemann, M. & Mabuza, A. Using the SaTScan method to detect local malaria clusters for guiding malaria control programmes. Malaria journal 8 (2009).
data, C. L. C. r. The European Environment Agency (EEA). (2010).
Bolker, B. glmer: Fitting Generalized Linear Mixed-Effects Models, https://www.rdocumentation.org/packages/lme4/versions/1.1-21/topics/glmer (2019).
Skarphedinsson, S. Tick-borne infections in Denmark – with special emphasis on human anaplasmosis PhD thesis, University of Southern Denmark, (2006).
Skarphedinsson, S. Tick-borne infections in Denmark, University of Southern Denmark, (2006).
Hijmans, R., Cameron, S., Parra, J., Jones, P. & Jarvis, A. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25, 1965–1978 (2005).
Hansen, K. & Lebech, A. The clinical and epidemiological profile of Lyme neuroborreliosis in Denmark 1985-1990. A prospective study of 187 patients with Borrelia burgdorferi specific intrathecal antibody production. Brain: a journal of neurology 115(Pt 2), 399–423 (1992).
Petersen, B., Møller, J. & Vilhelm, O. Season is an unreliable predictor of Lyme neuroborreliosis. Dan Med J. 62 (2015).
Mac, S., da Silva, S. & Sander, B. The economic burden of Lyme disease and the cost-effectiveness of Lyme disease interventions: A scoping review. PLoS One 14, e0210280 (2019).
Nygard, K., Brantsaeter, A. & Mehl, R. Disseminated and chronic Lyme borreliosis in Norway, 1995–2004. Euro surveillance: bulletin Europeen sur les maladies transmissibles = European communicable disease bulletin 10, 235–238 (2005).
Sodermark, L., Sigurdsson, V., Nas, W., Wall, P. & Trollfors, B. Neuroborreliosis in Swedish Children: A Population-based Study on Incidence and Clinical Characteristics. Pediatr Infec Dis J 36, 1052–1056 (2017).
Lindgren, E., Talleklint, L. & Polfeldt, T. Impact of climatic change on the northern latitude limit and population density of the disease-transmitting European tick Ixodes ricinus. Environmental health perspectives 108, 119–123 (2000).
Rizzoli, A. et al. Lyme borreliosis in Europe. Euro surveillance: bulletin Europeen sur les maladies transmissibles = European communicable disease bulletin 16 (2011).
Cappelen, J. Denmark - DMI Historical Climate Data Collection 1768–2019, https://www.dmi.dk/publikationer/ (2020).
Vollmer, S. A. et al. Host migration impacts on the phylogeography of Lyme Borreliosis spirochaete species in Europe. Environ Microbiol 13, 184–192 (2011).
Dantas-Torres, F., Chomel, B. & Otranto, D. Ticks and tick-borne diseases: a One Health perspective. Trends Parasitol. 28, 437–446 (2012).
Danmarks Statisik. BOL101: Boliger efter område, tid, anvendelse og beboertype, https://www.statistikbanken.dk/BOL101 (2016).
Messier, K. P., Jackson, L. E., White, J. L. & Hilborn, E. D. Landscape risk factors for Lyme disease in the eastern broadleaf forest province of the Hudson River valley and the effect of explanatory data classification resolution. Spatial and spatio-temporal epidemiology 12, 9–17, https://doi.org/10.1016/j.sste.2014.10.002 (2015).
Eikeland, R. A. M., Herlofson, K. & Ljostad, U. Risk factors for a non-favorable outcome after treated European neuroborreliosis. Acta Neurol Scand 3, 154–160 (2013).
Acknowledgements
This study was funded by Clinical Centre of Emerging and Vector-borne Infections (CCEVI), Odense University Hospital, Denmark and by the ScandTick Innovation project, EU Regional Development Programme InterReg OKS. The authors would like to thank Janne Fuglsang Hansen at The Department of Infectious Diseases at Odense University Hospital for advising and assisting with data analysis.
Author information
Authors and Affiliations
Contributions
S.S., F.C.K. conceived the study; F.C.K. collected primary data, A.M.A., P.B.D., were responsible for carrying out the data analysis; F.C.K. carried out the temporal analyses; L.J.K. carried out the regression analysis; A.M.A., P.B.D., L.J.K. and R.B. carried out the spatial data analysis and A.M.A., P.B.D. drafted the manuscript. All authors read and approved the final manuscript.
Corresponding authors
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Andreasen, A.M., Dehlendorff, P.B., Knudtzen, F.C. et al. Spatial and temporal patterns of Lyme Neuroborreliosis on Funen, Denmark from 1995–2014. Sci Rep 10, 7796 (2020). https://doi.org/10.1038/s41598-020-64638-5
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-020-64638-5
- Springer Nature Limited
This article is cited by
-
The relationship between the laboratory diagnosis of Lyme neuroborreliosis and climate factors in Kalmar County Sweden — an overview between 2008 and 2019
European Journal of Clinical Microbiology & Infectious Diseases (2022)
-
Evaluation of factors influencing tick bites and tick-borne infections: a longitudinal study
Parasites & Vectors (2021)