Lyme borreliosis in Belgium: a cost-of-illness analysis

Lyme borreliosis (LB) is the most common tick-borne disease in Europe. An erythema migrans (EM), a red expanding rash at the site of the tick bite, is one of the first and often the only symptom present. If left untreated the infection can disseminate causing more severe disease such as Lyme neuroborreliosis (LNB), Lyme arthritis (LA), Lyme carditis or acrodermatitis chronica atrophicans (ACA) [1, 2]. Even after appropriate antibiotic treatment, persisting non-specific symptoms such as fatigue, widespread musculoskeletal pain or cognitive difficulties have been reported by a subset of patients. When these symptoms persist for 6 months or more and are of such severity that they impact daily activities, this is often referred to as post-treatment Lyme disease syndrome (PTLDS) [3]. In Belgium, the annual incidence of LB has been estimated at 103 per 100,000 inhabitants (95% UI 87–120) for the period 2015–2017 [4]. A study on the prioritization of communicable diseases for future surveillance, performed in 2018, categorized it in its “high priority” diseases group [5]. Yet, the exact burden remains unknown. Cost-of-illness studies are an important part of the assessment of the impact of a disease on the healthcare system and society. They provide essential information for cost-effectiveness and budget-impact analyses, which in turn can be used for policy making on budget allocation and on prioritizing interventions that prevent or control disease burden. Studies assessing the economic burden of LB are limited, both in North-America and Europe [6, 7]. To the best of our knowledge, no such studies have been published yet for Belgium. The aim of this cost-of-illness study was to assess the economic burden of LB in Belgium, taking into account different clinical manifestations and PTLDS.

The total cost of LB was estimated through an incidence-based costing approach for patients with (i) an EM (including multiple EM) or (ii) disseminated/late LB, including patients who later developed PTLDS. A societal perspective, which considers costs for the national healthcare insurer, the patients and their employers, was taken. Direct medical costs (ambulatory [aka outpatient] and hospital [aka inpatient] care), direct non-medical costs (transportation) and indirect non-medical costs (productivity losses due to sick leave) were included using different data sources. Mean costs were extrapolated to the population level using LB incidence estimates [4] and the number of hospitalizations over multiple years to account for yearly fluctuations.

In order to estimate the economic burden of LB in Belgium, the current study used different data sources (a prospective cohort study and national hospital registries) which allowed the inclusion of direct medical costs, direct non-medical costs and indirect non-medical costs associated with this multisystem infectious disease. Based on an incidence approach, societal perspective and 2019 euros, the total annual cost associated with LB in Belgium was estimated at €5.59 million with a 95% UI between €3.82 and €7.98 million. Direct medical costs and productivity losses accounted for the majority of the costs, 49.8% and 46.4% of the total, respectively. Despite the much lower incidence of disseminated/late LB manifestations compared to EM, the former accounted for more than half of the total national LB costs per year (62%), because of its high mean cost of €5,148 (95% UI 3,091–7,911) per patient compared to a mean of €193 (95% UI 121–284) per EM patient. While the mean ambulatory direct medical cost was almost twice as high for patients with PTLDS after EM compared to EM patients without PTLDS, patients with PTLDS after disseminated/late LB incurred only higher productivity losses relative to patients without PTLDS after disseminated/late LB but no higher direct medical costs. The latter is unexpected, yet, these results need to be interpreted with caution as the sample sizes for the groups with PTLDS in the prospective cohort study were very small, causing uncertainty to be high. Although the majority of direct medical costs were reimbursed (75.5%), patients still pay €43.9 (EM) or €286 (Disseminated/late) of direct medical costs themselves. In addition, all direct non-medical costs are payed by the patient (i.e. travel expenses: €5 (EM) or €190 (Disseminated/late) and paid help: €50.9 (Disseminated/late)), making the cost to the patient still quite high. Furthermore, depending on the employment status and duration of sick leave there can be loss in salary.

Comparison of the current results with LB cost studies performed elsewhere is challenging as there is heterogeneity in the methods used, the patient populations studied and the cost data collected, as well as in the healthcare systems in place in the countries concerned. In Europe, some older cost studies have been performed for LB in Scotland (1999) [25], for LNB patients in Sweden (2000–2005) [26] and for hospitalized patients in Germany (2008–2011) [27]. Similar to our study, van den Wijngaard et al. (2017) estimated LB costs for the Netherlands, including direct medical and indirect non-medical costs [28]. For EM, a mean ambulatory cost of €136 was estimated (amount adjusted to 2019 price level [29, 30]), which is lower than our mean estimate of €193 in EM patients (including 5.9% patients with PTLDS [31]). Note that in contrast to the Dutch study, our study included OTC medication, laboratory testing and travel expenses with a cost of €40. The mean total cost for disseminated LB in the Dutch study, including ambulatory and hospital costs as well as costs for productivity losses, was estimated at €6,327 (inflated [29]), which is higher compared to the mean of €5,148 in our study (including 20.9% patients with PTLDS). The total costs for patients with persisting symptoms in the Netherlands were estimated as a separate category, which represented a substantial amount of €6,361 per patient (inflated [29]), on top of the costs for EM or disseminated LB. Out of these, €1,362 concerned ambulatory direct medical costs, which is much higher than the additional ambulatory direct medical costs found for PTLDS patients in our study (Table 2), yet sample size for the latter was very small. Also, patients with persisting symptoms were defined differently than our PTLDS patients and were included based on a GP diagnosis, so it may be that these were more seriously ill patients; the duration of persisting symptoms in that study was longer (4.6 years). Furthermore, some cost studies have also been performed in the US, but comparisons with non-US estimates are problematic due to fundamental differences in pricing and healthcare organization, as well as differences in clinical manifestations, with a higher proportional occurrence of more costly LA cases in the US [32‐35]. Further research on the costs of patients with PTLDS seems warranted.

Comparison of our study results with the cost-of-illness of infectious diseases other than Lyme borreliosis in Belgium is also not straightforward as both costs and incidence data must be taken into account. For several infectious diseases, the average cost per case is lower (e.g., influenza [36, 37], rotavirus [38]) or similar (e.g., herpes zoster [39]) than the cost of LB, but due to the higher incidence of these infections in Belgium, their total national cost is much higher than that of LB. For example, for influenza, mean ambulatory direct costs have been estimated at €51–64 per case and hospitalization cost at €2,599, which, with an estimated 400 000 cases [36, 37], leads to a national cost of more than €30 million (2011–2012, indirect costs not included). Also for human papilloma virus (HPV), where mean costs per case and incidences differ largely by outcome (e.g., cervical cancer vs. anogenital warts), overall national costs are higher than LB [40]. Some other infections, have a lower direct cost due to a low incidence (e.g. hepatitis A [41]) or due to a low cost per case (e.g. Varicella [39]).

It is important to note that, in the current study, only costs for patients with clinical LB manifestations, either EM or laboratory confirmed disseminated/late LB, were included [31]. As such, costs for consultations for a tick bite without LB development were not counted even though they might take place due to worries about LB. Based on data from the SGP network (2015─2017), the annual number of GP consultations for tick bites can be estimated at 19,422 (95% UI 16,384–22,523), which corresponds to an estimated cost of €510,806 (95% UI 430,901–592,368) (productivity losses not counted). For the same reason, only laboratory tests performed in patients with clinical LB manifestations were included, however, many more LB serological tests are performed each year in Belgium. In the period 2015─2017, in ambulatory care only, an annual mean of 220,270 IgG EIA tests and 11,405 IgG Immunoblots were performed on blood, indicating a low positivity rate (5.18%) for the EIA (first-tier test) as immunoblot is recommended to only be performed when EIA is positive or equivocal. This indicates a potential over-use of these tests without benefits for patient management. Note that these serological tests are not advised for EM diagnosis, as in this early disease stage sensitivity is too low. In addition, tests do not differentiate between active and past infection and are not useful to follow-up antibiotic treatment success [42]. A high economic burden of several million euros related to these tests can be expected; the mean annual cost for all LB serological tests (IgG, IgM, EIA, Immunoblot, on blood, on CSF) in 2015─2017 equaled €1,092,801 and in addition a forfeit by analysis by patient (not by test) of €19.84─€30.5, depending on the number of tests performed, applies. More research into these costs and the reasons why these high numbers of tests are carried out is therefore of great importance. Also, no costs for prophylaxis treatment were included in the current study.

Since there are yearly fluctuations in the number of LB cases, this study used mean incidence estimates for a period of three years. Nevertheless, this incidence estimate was probably not representative for the number of LB cases in 2019, as 2019 was a year with an exceptionally low number of tick bites, probably due to extreme dry and warm periods during the summer [43]. It is therefore expected that the total costs for the year 2019 would be lower than the current cost estimates based on the LB incidence 2015─2017, but the latter are expected to be more representative for an average year.

As is the case for many healthcare cost data, most costs were skewed with a few patients incurring much higher costs than the majority of patients, causing estimated mean costs to be higher than median costs (Additional file 2). For example, for ambulatory productivity loss the median cost equaled €0 in EM patients and €656 for disseminated/late LB patients compared to a mean of €63.9 and €2,624, respectively. Yet, as is appropriate for cost-of-illness and cost-effectiveness analysis, we focused on mean costs, and extrapolated these to the population level to produce our overall cost burden estimates [44].

There are some limitations to the current study. First, for the ambulatory cost calculation, no children were included, hence the same costs are assumed as for adults. Second, as data are based on the incidence estimates 2015─2017 published by Geebelen et al. (2019), limitations mentioned there also apply to the current estimates [4]. As a possible underestimation of the proportion of disseminated/late LB manifestations was suspected, national costs for this group in the current study, can also be underestimated. Third, the sample size of the disseminated/late LB group and the group of patients with PTLDS in the prospective cohort study, used for ambulatory cost calculation, was small and uncertainty in the results high. Further research into their costs is needed. Fourth, hospitalizations for LB were selected based on ICD-10-coding of LB, which mainly serves for financing and was not necessarily developed for surveillance purposes. Quality of the coding could not be checked. Since there is no specific ICD-10-CM code for EM or PTLDS, all hospital costs were allocated to the disseminated/late LB group. Although no hospital costs are expected due to the mild nature of the symptoms of EM, they may be present in exceptional cases. Fifth, while a societal perspective was taken, it was not possible to include all costs that might have occurred; no ambulatory laboratory testing other than LB serology, no medical acts for emergency visits and not all non-reimbursed costs and supplements for hospitalizations could be included. On the other hand, for hospitalizations, costs for all types of medication, laboratory testing or medical acts were included, comprising also costs for comorbidities not related to LB (e.g. blood pressure medication). Finally, for EM or disseminated/late LB patients, no costs exceeding 6 months post diagnosis were included, which could result in an underestimation of the true cost of LB in Belgium. Nevertheless, after a longer period it becomes more difficult (also for the patient) to relate specific costs to the disease.

In conclusion, we provide for the first time, a comprehensive assessment of the costs related to LB in Belgium. As EM and disseminated/late LB are estimated to cause about 40% and 60% of the total national costs, respectively, both measures to prevent EM, such as prevention of tick bites and fast removal of ticks and prevention of dissemination of disease, such as informing GPs and the public on EM to improve fast treatment, are essential. Ideally, a vaccine will be developed that prevents tick bites, hence prevents LB (all manifestations), but also prevents transmission of other tick-borne pathogens. The current cost estimation can serve as an input in future cost-effectiveness analyses of such vaccines or other pharmaceutical and non-pharmaceutical interventions. While patients with PTLDS seem to have somewhat higher costs compared to patients without PTLDS this needs further research.