Targeting Tularemia: Clinical, Laboratory, and Treatment Outcomes From an 11-year Retrospective Observational Cohort in Northern Sweden

Abstract Background Tularemia is an important reemerging disease with a multimodal transmission pattern. Treatment outcomes of current recommended antibiotic regimens (including ciprofloxacin and doxycycline) remain unclear. In this retrospective cohort study, we report clinical, laboratory, geographical, and treatment outcomes of laboratory-confirmed tularemia cases over an 11-year period in Northern Sweden. Methods Data from reported tularemia cases (aged >10 years at time of study) in Norrbotten county between 2011 and 2021 were collected through review of electronic medical records and participant questionnaires; 415 of 784 accepted participation (52.9%). Of these, 327 were laboratory-confirmed cases (serology and/or polymerase chain reaction). A multivariable logistic regression model was used to investigate variables associated with retreatment. Results Median age of participants was 54 years (interquartile range [IQR], 41.5–65) and 49.2% were female. Although ulceroglandular tularemia was the predominant form (n = 215, 65.7%), there were several cases of pulmonary tularemia (n = 40; 12.2%). Inflammatory markers were largely nonspecific, with monocytosis frequently observed (n = 36/75; 48%). Tularemia was often misdiagnosed on presentation (n = 158, 48.3%), with 65 (19.9%) receiving initial inappropriate antibiotics and 102 (31.2%) retreated. Persistent lymphadenopathy was infrequent (n = 22, 6.7%), with 10 undergoing surgical interventions. In multivariable analysis of variables associated with retreatment, we highlight differences in time until receiving appropriate antibiotics (8 [IQR, 3.25–20.75] vs 7 [IQR, 4–11.25] days; adjusted P = .076), and doxycycline-based treatment regimen (vs ciprofloxacin; adjusted P = .084), although this was not significant after correction for multiple comparisons. Conclusions We comprehensively summarize clinical, laboratory, and treatment outcomes of type B tularemia. Targeting tularemia requires clinical awareness, early diagnosis, and timely commencement of treatment for an appropriate duration.

Tularemia is a reemerging zoonotic disease present in most countries in the Northern Hemisphere [1,2].Outbreaks of tularemia caused by the bacterium Francisella tularensis subsp.holarctica ("type B") have occurred at irregular intervals in the northern parts of Sweden since first described in the country in 1931 [3,4].Although there is notable annual fluctuation, there has been an increasing trend in the number of cases seen across Sweden and the entire European Union over the latest 5-year period (2017-2021) [5,6].Increasing frequency of outbreaks in middle and southern parts of Sweden suggests tularemia is a reemerging infectious disease complexly linked with climate change [6][7][8][9][10][11][12][13][14].
Zoonotic, waterborne, airborne, and vector (primarily mosquito)-borne outbreaks of the disease have important health-related and socioeconomic burden on individuals and society through prolonged illness, absence from work, and healthcare costs [6].
Manifestations of the disease vary depending on inoculation dose, type of exposure (oral, ocular, pulmonary, cutaneous), and geographical location (type A in North America vs type B globally), with the latter causing debilitating symptoms, but rarely resulting in lethalities [4,15,16].
In this study, we comprehensively investigated clinical, microbiological, laboratory, and treatment outcomes in a cohort of tularemia cases in northern Sweden, which includes the second (2019) and third (2015) largest outbreaks recorded in Sweden.

Study Setting
This study was performed in Norrbotten County, the largest (98 245 km 2 ) and northernmost county in Sweden, with a sparse population of approximately 250 000 [23].

Study Design and Procedure
This is a retrospective cohort study involving electronic medical record (EMR) review of tularemia cases in Norrbotten in combination with a targeted questionnaire to further explore clinical details.Cases were identified via the National Infection Control database (SmiNet).Inclusion criteria were: (1) 4) not meeting World Health Organization presumptive or confirmed tularemia case definition [18].
Eligible participants were sent a letter containing a questionnaire, consent form, study information sheet, and web link to a secure survey platform (EvaSysV8.1,Region Norrbotten).Nonrespondents were sent notifying mobile text messages and a second paper questionnaire.

Serology and Microbiology
Serological analysis was performed at the regional clinical microbiology laboratory at Sunderby hospital, Luleå, Sweden, using an immunochromatographic rapid test (VIRapid, Vircell, Granada, Spain) [24]; and the national reference laboratory for tularemia at Umeå University hospital, Umeå, Sweden, using a validated in-house enzyme-linked immunosorbent assay, together with polymerase chain reaction (PCR) and cultures [25].

Statistical Analysis
Data were collected from the regional EMR (VAS 49.0, Region Norrbotten), including documentation from general practitioners (GPs), hospitals, and specialist clinics.Integrated electronic prescriptions and microbiological data were reviewed.Geographical information of reported site of infection was mapped using Epi Info 7.2.5 (Centre for Disease Control).Statistical analysis was performed in SPSS 28.0 (IBM).Statistical methods included logistic regression (categorical variables) and Mann-Whitney U test (continuous and ordinal variables) for univariate analysis; a multivariable logistic regression model using a stepwise backwards elimination protocol (Supplementary Table 1) [26].Holm-Šídák correction was applied, with a corrected P value <.05 considered significant.

Ethics
Ethics committee approval was obtained from The Swedish Ethical Review Authority (Dnr 2020-04411, 2021-02953) and Norrbotten County Research Council (Dnr 01690-2020).All study participants received written age-appropriate information about the study and provided written informed consent.Additionally, parental consent was obtained for individuals aged <15 years.

Study Participants
In total, 830 cases of suspected or confirmed tularemia were reported to the Department of Communicable Disease Control, Norrbotten County, Sweden, between 1 January 2011 and 31 December 2021 (incidence, 30.1 per 100 000 per year).Three cases of tularemia reinfection were reported, all without laboratory confirmation.Overall, there were 441 questionnaire respondents between 30 December 2020 and 16 October 2022 (56.3% response rate), with 26 declining study participation.A further 88 of 415 were excluded after EMR review (Figure 1).
Geographical distribution (Figure 2) showed predominance of tularemia cases in coastal municipalities, as well as high incidence (>80 per 100 000 inhabitants) in certain highland areas popular among tourists, with a high density of lakes and waterways.

Laboratory Manifestations
Laboratory data (Table 1) were available for most patients because of capillary point-of-care testing in all primary care facilities.Among 289 (88.4%) participants who had an initial C-reactive protein (CRP) recorded, median CRP was highest on days 7 through 9 after symptom onset (78 mg/L; IQR, 40-133), and lower among participants who presented later, suggesting a spontaneous reduction over time and/or milder illness (Figure 3B).

Multivariable Analysis of Treatment Outcomes
Participants who experienced treatment failure after receiving an appropriate antibiotic regimen were compared with those who had successful treatment (Table 2).Variables were analyzed using multivariable analysis after correction for multiple comparisons and included time until receiving appropriate antibiotics (8 [IQR,.75]vs 7 [IQR, 4-11.25]days; adjusted P = .076),doxycycline-based treatment regimen (vs ciprofloxacin; adjusted P = .084),and participant age (years; adjusted P = .270).Thus, no significant correlations were identified.

DISCUSSION
In this retrospective cohort study of 327 tularemia cases in northern Sweden, we demonstrate that tularemia is a multifaceted disease with a multitude of differential diagnoses and wide range of clinical manifestations, including prolonged suppurative lymphadenitis and atypical respiratory infection.We further describe successful treatment outcomes among individuals receiving early treatment with ciprofloxacin and doxycycline and highlight time until appropriate treatment as a potential cause for requiring retreatment, as previously noted [27].
Our cohort had similar demographics to previous studies performed in northern Europe [3,9,11,[28][29][30][31] and identified men and women aged 40 to 60 years as having the highest burden of tularemia, although there was high prevalence of infection throughout all age groups.Our data support ulceroglandular tularemia transmitted by mosquitos as the predominant clinical form in this geographical region [6,31], pulmonary tularemia as an important cause of pneumonia, and atypical disease often misdiagnosed as malignancy and/or tuberculosis [9,32].Immunosuppression was infrequent in our cohort (n = 8), with 4 cases admitted to the hospital, including 1 developing pulmonary tularemia and 2 cases with F. tularensis septicemia.
Our findings show secondary skin manifestations in approximately 15% of cases, consistent with previous studies [9,[33][34][35].Children aged 10 to 15 years appeared to have frequent skin involvement (54.5%), as previously described [36].Interestingly, secondary skin manifestations appeared in close relation to commencing appropriate antibiotics, and one could speculate about a correlation between the 2. We highlight that limited knowledge about these reactions can lead to potentially inappropriate allergy labeling, as evidenced by our cohort.Reported laboratory values in our cohort identified CRP (50-150 mg/L in 45.3%), procalcitonin (>0.1 ng/mL in 88.2%; <0.5 ng/mL in 94.1%) and monocytes (>0.8 × 10 6 cells/mL in 48.0%) as nonspecific markers for tularemia infection in the context of high clinical suspicion [16].Previous studies have predominantly reported normal differential counts in type B tularemia, with monocytosis >0.95 × 10 6 cells/mL present in 4% of cases compared with 30.6% in our cohort [9,37].Leukocyte differential count is not routinely performed in primary care settings in Sweden.Perhaps wider implementation could help guide diagnosis of atypical manifestations of tularemia.
Chronic suppurative lymphadenitis has previously been described in up to 30% of patients with ulceroglandular and glandular tularemia [1,9].We describe chronic suppuration in 8.8% of cases, often leading to surgical intervention and/or retreatment with multiple courses of antibiotics.Interestingly, among PCR-positive cases, culture negativity was associated with longer time since symptom onset (median, 10.5 days; IQR, 6.5-23.0vs 6 days; IQR, 4.75-8.5;P = .029).It remains unclear whether replicating bacteria are present in the ulcer after completed treatment or whether persistent lymphadenitis is primarily driven by an inflammatory response, as highlighted by few PCR-positive cases after aspiration and only a single culture positive [38].Rates of incision and drainage appear to be lower in our cohort than in studies from France (27%) and the United States (19%) [13,39], perhaps potentially related to shorter time until institution of appropriate antibiotic treatment.Even higher rates (60%) have been reported in pediatric populations [40].The clinical value of prolonged antibiotic treatment in drained suppurative lymphadenitis remains unclear and will require further studies, potentially stratified by PCR status and/or culture.
Our cohort appears to have lower retreatment rates after appropriate antibiotic regimens than previous studies (16.8% vs 38.6%) [13].This could be explained by the short duration to receiving appropriate therapy, as observed in another Swedish study with 95.3% successful treatment rate after initiating therapy early (median, 3 days from symptom onset) [19].We further show successful treatment in 3 of 4 (75%) of cases treated with moxifloxacin as a first-or second-line antitularemia agent, as supported by previous in vitro studies [17].As previously described [6,17], treatment with beta-lactams or clindamycin appears ineffective.
Comparisons between ciprofloxacin and doxycycline for treatment of tularemia have previously been limited by sample size [13,18,20,22,39].In our cohort, we show a multivariable trend toward requiring retreatment in those receiving doxycycline compared with ciprofloxacin (adjusted P = .076);however, when accounting for appropriate dosing and duration of antibiotic therapy, there was no significant difference between groups (14.2% vs 10.2%; P = .524).Although this is consistent with previous studies showing that shorter treatment periods are associated with relapse [22], our study is likely underpowered for assessment of this outcome.
Interestingly, we observed a broad range of differential diagnoses among clinicians, leading to 19.9% of participants receiving empiric antibiotics ineffective against tularemia.Higher rates were observed during nonepidemic years, suggesting that improved awareness and high clinical suspicion is required among clinicians to promote overall outcomes given that initial serological results can be negative [9].
No deaths were recorded in our cohort, and only 3 deaths were reported to the Swedish Cause of Death Register between 1997 and 2022, suggesting that, although type B tularemia can cause serious morbidity, it is rarely fatal.
Our study was limited by its retrospective design.Despite being 1 of the largest descriptive studies of clinical outcomes in tularemia, the sample size limited statistical analysis.Antibiotic treatment outcomes are difficult to interpret in those receiving retreatment because cases often received overlapping antibiotic regimens.The addition of a questionnaire to study participants improved the quality of clinical data by assessing long-term outcomes, but because of the long time between infection and time of survey in some cases, recall bias and limited memory could have influenced these results.Respondents were slightly older and more likely to be female compared with overall reported cases, introducing a selection bias that could have skewed the results.While our results should be generally applicable to type B tularemia in northern Europe, the level of relevance remains unclear in settings where alternative clinical and microbiological forms dominate.

CONCLUSION
In conclusion, we comprehensively summarize clinical, laboratory, and treatment outcomes of type B tularemia.Targeting tularemia remains challenging because of its multifaceted presentation with a wide range of differential diagnosis and unclear optimal treatment regimens.We believe the present study can provide guidance regarding targeted treatment and highlight areas for future research.Ideally, a randomized controlled trial to evaluate effectiveness of antibiotic therapies, as well as the possible need for prolonged antibiotic treatment in suppurative lymphadenitis, should be conducted to improve clinical outcomes.

Figure 1 .
Figure 1.Study method and inclusion of participants.EMR, electronical medical record. a Lack of electronical medical record documentation (n = 15); confirmed falsepositive tularemia tests (n = 2; ×1 false-positive rapid test, ×1 false-positive PCR); misdiagnosis (later confirmed as mycoplasma infection via PCR; n = 1); incidental positive serology without evident clinical symptoms of tularemia infection (n = 5); age <10 y at time of survey (n = 2).b Clinical diagnosis of tularemia without a positive tularemia serology, culture, and/or PCR and were therefore excluded from further analysis (of these, 15 had a single negative serology result and 1 had a negative PCR result).PCR, polymerase chain reaction.

Figure 2 .
Figure 2. Geographical distribution and epidemiological incidence of tularemia cases at municipality level in Norrbotten County over 11 y between 2011 and 2021 (n = 243).Incidence calculated using total number of reported cases (N = 830).Blue dots indicate self-reported location of exposure/illness by individual cases.Number in blue dots indicate cumulative number of cases in area.

Figure 3 .
Figure 3. Initial recorded (A) temperature (°C) and (B) CRP and at time from onset of tularemia symptoms (n = 289).Box diagram denotes median, interquartile range, range (extreme values not included).Line denotes mean values.CRP, C-reactive protein.

Table 2 . Comparison of Characteristics of Participants With Completed Treatment After Appropriate Antibiotic Regimens vs Those Requiring Further Treatment Courses
Logistic regression for categorical variables.Mann-Whitney U test for ordinal and continuous variables.Data reported in absolute numbers and percentages (in brackets), or median and interquartile range, as specified.Significance in multivariable analysis after Holm-Šídák correction applied for adjusted P values is displayed in brackets.Abbreviations: CRP, C-reactive protein; IQR, interquartile range.aDefined as regimen containing ciprofloxacin, aminoglycoside, or doxycycline, as per the World Health Organization.Retreatment despite a course of ciprofloxacin or doxycycline, individuals with "optimization" of antibiotics after diagnosis (n = 2) or adverse drug reactions requiring treatment with alternative regimen (n = 11) were excluded.c Value included as alternative to "initial appropriate regimen" in multivariable analysis and not part of formal multivariable model. b