Some European countries, including Finland, notify microbiologically confirmed LB cases routinely but clinically diagnosed EM is greatly neglected. In our previous study, we evaluated the incidence and the geographic distribution of LB in Finland by using three national health care registers: Avohilmo, Hilmo, and NIDR (17). Here, we used a personal identification number in order to link individual LB cases in these registers, and thereby to evaluate our register data and to refine our previous LB incidence estimates. Avohilmo was found to be useful in monitoring clinically diagnosed EM cases, whereas Hilmo did not add much value next to current laboratory-based surveillance of disseminated LB. However, Hilmo revealed some shortcomings related to physician-based surveillance and improved the estimation of the total annual number of EM infections in Finland.
In our previous study, we evaluated the total annual number of LB cases in Finland (6440 cases in 2014) by summing clinically diagnosed and microbiologically confirmed LB cases together (17). This calculation included an assumption that the number of clinically diagnosed LB cases in Avohilmo does not substantially overlap with the number of microbiologically confirmed cases in NIDR. Here, the overlap between Avohilmo and NIDR registers was 6.3% in 2014 indicating that, indeed, the vast majority of the LB cases identified in Avohilmo represent clinically diagnosed EM infections. Since no laboratory testing for EM is either required or recommended, EM cases are not found in NIDR. Those 227 cases notified to both registers, NIDR and Avohilmo, presumably represent disseminated LB cases diagnosed in the primary health care and/or early seroconverted EM infections where laboratory tests were performed despite the recommendations. However, this rate of overlapping between Avohilmo and NIDR is so small that the double-counting of LB cases should not cause major bias to the estimated total number of LB cases in Finland. The number of clinically diagnosed EM cases where laboratory tests were performed and returned seronegative is unknown and would be of future research interest in order to understand the laboratory test ordering practices of physicians in the primary health care. In the recently published study from Norway, the antibody testing was performed in ~20.0% of GP diagnosed EM cases (19).
Next to laboratory-based reporting of LB, the inclusion of EM as another key indicator has been suggested to LB surveillance in Europe (20, 21). Some European countries have estimated the incidence of LB and the frequency of its different manifestations through repeated cross-sectional physician surveys or through register-based studies monitoring of notified in- and outpatient diagnoses (9, 13, 14, 16, 22). In these studies, the provided EM incidence estimates rarely represent the whole country or all the physicians involved in the diagnostics and treatment of LB within the country. In contrast, Avohilmo is countrywide and involves all GPs in the public primary health care. When a compulsory ICD-10 code is entered in the patient record after GP consultation, LB case is notified to Avohilmo automatically without the need for active reporting by the physician. Some underestimation of the number of clinically diagnosed LB cases is, however, acknowledged since Avohilmo does not yet cover occupational or private health care visits. Moreover, subjective clinical misjudgment by the physician may lead to under- or over-reporting of clinically diagnosed LB cases although typical EM is most likely well recognized in Finland. Despite these shortcomings, Avohilmo appeared to be a useful tool in monitoring EM infections in Finland.
According to the international and national guidelines, the first line treatment for EM is oral amoxicillin (3, 23). Doxycycline can be used for patients who are allergic to penicillin, but if contraindicated (e.g. in children <8 years and in pregnancy), azithromycin is the alternative. In this study, the high overall percentage of antibiotic purchases (93.3%), together with the fact that amoxicillin and doxycycline were the most often purchased antibiotics for clinically diagnosed LB, indicate a good compliance with the recommended treatment guidelines. In a recently published Norwegian study, the most often prescribed antibiotic for EM in general practice was phenoxymethylpenicillin (19). However, phenoxymethylpenicillin is not recommended for EM in Finnish guidelines. The total number of antibiotic purchases proportioned to the LB cases in Avohilmo exceeded 100% most likely because some of the patients have been prescribed other antimicrobial subgroups –perhaps due to side-effects– before finishing the first treatment. A proportion of patients may have received a second course of antibiotics after the first one although prolonged or repeated treatment is not recommended for LB(24). Here, we investigated which antibiotic regimens were chosen for the treatment of clinically diagnosed LB. In order to investigate whether the prescribed dosing and the duration of the treatment for clinically diagnosed LB also follow the recommended guidelines, more detailed data from the Finnish Social Insurance Institution is needed in the future.
Because Hilmo contains nationwide linkable data on hospital discharges, we expected that LB cases in Hilmo represent mainly disseminated LB infections. Such cases should be microbiologically confirmed and hence, almost every LB case in Hilmo should have an entry found in NIDR. However, surprisingly, over two thirds of the LB cases identified in Hilmo (999 out of 1420 cases) were not microbiologically confirmed (Figure 2). When a wider time period 2004–2015 was used in NIDR, 685 cases remained without microbiological confirmation. Most of the microbiologically unconfirmed hospital cases were notified to NIDR years earlier which means that all the Hilmo entries in 2014 were not new infections during the same year. The same was noted after evaluating the patient records of the subgroup of microbiologically unconfirmed hospital cases (n=396) in 2014. Around 10% of those Hilmo cases were old notifications in the register, where microbiological testing for LB was performed years earlier, but the diagnoses were repeated in subsequent hospital visits without a suspicion of a new infection. These cases were never notified to NIDR by the laboratories because the microbiological result at the time of investigation had been either negative or borderline reactive. In addition to old notifications, some over-reporting of the total annual number of LB cases in Hilmo is caused by the so-called working diagnoses (suspected LB) notified before the microbiological confirmation was obtained. However, <10% of analysed microbiologically unconfirmed hospital cases were eventually reported by other ICD-10 code than “A69.2”.
In 2014, laboratories missed to notify 2.8% of positive LB findings in the subgroup of microbiologically unconfirmed hospital cases. More under-reporting of LB cases in inpatient hospital setting occurs due to insufficient reporting practices by physicians which is reflected by the fact that 44.3% of 1696 microbiologically confirmed LB cases in 2014 were reported in Hilmo with other ICD10-codes than “A69.2” (Figure 3). These cases probably represent true LB cases where the physicians missed entering ICD-10 code for LB to the patient records. Similarly, a recently published Swedish study found that less than half of all microbiologically confirmed LNB patients received the recommended combination of ICD-10 codes for LNB and only two thirds of the patients received the code for LB at all (13).
All in all, Hilmo seems to be a heterogenic and imperfect register and it does not add much value next to NIDR to estimate the number of disseminated LB infections in Finland. Both over- and under-reporting of LB cases occur due to above-mentioned reasons related to reporting practices. On the other hand, it gives valuable information considering the total number of EM infections in our country. Of the analysed 396 microbiologically unconfirmed hospital cases, 40.4% were clinically diagnosed EM infections (Figure 2). This finding indicates that the actual annual number of clinically diagnosed EM is probably a few hundred higher than what Avohilmo illustrates: some EM cases are missed because they are only reported to Hilmo and some are missed because they are diagnosed and treated in the occupational and private health care. As a limitation of this study, we do not have actual LB case numbers from the occupational or private health care.
Despite the serological result, around one third of the analysed 396 microbiologically unconfirmed hospital cases were clinically diagnosed LB (other than EM) (Figure 2). According to the patient records, neurological symptoms and joint complaints were the most often mentioned manifestations leading to LB diagnosis and treatment, but in some cases, the clinical reasoning did not become fully clear to us. However, we cannot assess the validity of these diagnoses retrospectively, but it is noteworthy that a relatively big proportion of LB cases (other than EM) seem to be diagnosed clinically due to suggestive symptoms and, at the most, some reactivity in the serology. The heterogeneity of Hilmo stresses the need for clear, uniform LB case definitions, improved LB diagnostics, and increased awareness among physicians to improve the LB surveillance in Finland.
The lack of data from the occupational and private health care cause some underestimation of LB cases. In 2014, 13.3% of microbiologically confirmed LB cases could not be linked to any records in Avohilmo or Hilmo (Figure 3). The most logical explanation would be that these cases derive either from occupational or private health care where Avohilmo and Hilmo are not used. A small proportion of these 226 cases could be such cases where laboratory diagnostics for LB is requested in the primary health care but despite the positive finding, other ICD-10 code than “A69.2” is used in Avohilmo. However, in the case of microbiologically confirmed LB cases, ~13% could roughly represent the utilization of occupational and private health care. This proportion likely varies between regions depending on the availability of the private health care.
This register-linkage allowed the assessment of case numbers and frequency of clinical diagnoses of LB in the study year 2014, and thereby also illustrated the reporting practices considering LB in Finland. Since this study, Avohilmo has been introduced in the routine surveillance of LB next to laboratory-based surveillance of microbiologically confirmed LB (NIDR) including LNB (available in Finnish in: Borrelioosin seuranta Suomessa. NIHW(25)). Avohilmo is updated on a weekly basis reflecting the epidemiological situation of LB infections and the seasonal variation of tick exposure timely. As a register for microbiologically confirmed disseminated LB infections, NIDR is suitable for the long-range temporal and geographical distribution analyses concerning LB incidence in Finland. Overall, the LB surveillance based on both registers jointly gives a comprehensive picture of LB incidence in our country considering both the early stage EM infections as well as the disseminated LB infections.