The effect of measuring serum doxycycline concentrations on clinical outcomes during treatment of chronic Q fever

The effect of measuring serum doxycycline concentrations on clinical outcomes during treatment of... Abstract Background First choice treatment for chronic Q fever is doxycycline plus hydroxychloroquine. Serum doxycycline concentration (SDC) >5 μg/mL has been associated with a favourable serological response, but the effect on clinical outcomes is unknown. Objectives To assess the effect of measuring SDC during treatment of chronic Q fever on clinical outcomes. Methods We performed a retrospective cohort study, to assess the effect of measuring SDC on clinical outcomes in patients treated with doxycycline and hydroxychloroquine for chronic Q fever. Primary outcome was the first disease-related event (new complication or chronic Q fever-related mortality); secondary outcomes were all-cause mortality and PCR-positivity. Multivariable analysis was performed with a Cox proportional hazards model, with shared-frailty terms for different hospitals included. Results We included 201 patients (mean age 68 years, 83% male): in 167 patients (83%) SDC was measured, 34 patients (17%) were treated without SDC measurement. First SDC was >5 μg/mL in 106 patients (63%), all with 200 mg doxycycline daily. In patients with SDC measured, dosage was adjusted in 41% (n = 68), concerning an increase in 64 patients. Mean SDC was 4.1 μg/mL before dosage increase, and 5.9 μg/mL afterwards. SDC measurement was associated with a lower risk for disease-related events (HR 0.51, 95% CI 0.26–0.97, P = 0.04), but not with all-cause mortality or PCR-positivity. Conclusions SDC measurement decreases the risk for disease-related events, potentially through more optimal dosing or improved compliance. We recommend measurement of SDC and striving for SDC >5 μg/mL and <10 μg/mL during treatment of chronic Q fever. Introduction Q fever is caused by the intracellular Gram-negative bacterium Coxiella burnetii. After primary infection, approximately 60% of patients remain asymptomatic. The remaining patients develop illnesses such as flu-like symptoms or pneumonia.1 A small proportion of patients develop chronic Q fever after primary infection.1 In contrast to acute Q fever, chronic Q fever is a potential life-threatening disease. Most often, it leads to endocarditis, infection of vascular prostheses or arterial aneurysms.1 The recommended first-line treatment for chronic Q fever is doxycycline combined with hydroxychloroquine for at least 18 months, or alternatively doxycycline combined with a quinolone.2–5 Culturing of C. burnetii is not routinely performed in practice, because it is difficult and only allowed in laboratories with a Biosafety Level 3 facility.6–8 Therefore, monitoring of treatment effect mainly depends on measurement of antibody titres, monitoring by PCR for C. burnetii in blood or tissue, and follow-up of the patient’s clinical condition.1 To assess the adequacy of doxycycline dosage and compliance with therapy, measurement of serum doxycycline concentration (SDC) is a potential tool. In two clinical studies, SDC >5 μg/mL was associated with favourable serological response, and in another study, higher SDC to MIC ratios were associated with rapid decline in phase I IgG antibodies to C. burnetii.9–11 This can be explained by doxycycline being more effective at higher concentrations, or resistance leading to decreased effectiveness in patients with lower SDC.12–14 In two reports, resistance was reported in 6%–23% of isolates.9,14 The effect of measuring SDC on clinical endpoints such as complications and mortality and PCR-positivity, has not been studied before. By evaluating data from a nationwide cohort of chronic Q fever patients, we aimed to assess the effect of treatment when dosage was based on SDC, as performed in clinical practice, for these clinical endpoints in chronic Q fever patients treated with doxycycline and hydroxychloroquine. Patients and methods Data from the Dutch national chronic Q fever database were used: in this database, detailed data are recorded regarding the treatment of patients with proven or probable chronic Q fever diagnosed since the start of the Dutch Q fever outbreak (1 January 2007). Using these data, we retrospectively assessed the effect of measuring SDC in patients who were treated with doxycycline and hydroxychloroquine for at least 12 weeks. As this is a retrospective study and action taken by clinicians on SDC results is not standardized, we could not study the effect of SDC values on clinical outcomes: clinicians intervene based on SDC values and may aim for higher values in patients with more severe disease, which leads to bias. Therefore, we studied the effect of a treatment strategy in which dosage adjustment and coaching regarding compliance was based on SDC. Data collection and inclusion of patients The design of the Dutch national chronic Q fever database was approved by the Medical Ethics Committee of the University Medical Center in Utrecht. This database contains complete follow-up data of all chronic Q fever patients ≥18 years of age in the Netherlands from 1 January 2007 to 1 May 2016, from 28 hospitals in the Netherlands (see Acknowledgements). Diagnosis and classification of chronic Q fever was based on the Dutch chronic Q fever consensus group criteria.15 Clinicians identified patients based on a positive PCR for C. burnetii in serum or tissue and/or C. burnetii phase I IgG ≥1:1024. Patients with a serological profile and clinical condition matching acute Q fever were excluded. The Dutch chronic Q fever consensus group criteria discriminate between proven, probable and possible chronic Q fever. For this study, only patients with proven or probable chronic Q fever were included. Possible chronic Q fever patients were not included because the presence of clinical relevant infection in possible chronic Q fever patients is questionable and these patients do not have an indication for treatment; therefore, evaluation of treatment-related aspects and outcomes is impossible.The observation period for all patients ended on 1 May 2016. Laboratory testing Microbiological testing consisted of an indirect fluorescent-antibody assay (IFA) for phase I and II IgG against C. burnetii in plasma or serum (Focus Diagnostics, Inc., Cypress, CA, USA or Fuller Diagnostics, LLC, Anchorage, AK, USA). Titration of antibody levels was carried out at different hospital sites with dilutions on a binary scale with a cut-off of 1:32. Furthermore, PCR for C. burnetii DNA in serum or plasma and, if applicable, in tissue samples was performed (NucliSENS easyMAG; bioMérieux, Marcy l’Etoile, France). All SDC measurements were performed in the apothecary laboratory of the Jeroen Bosch Hospital in ’s-Hertogenbosch, with the exception of one measurement of a patient from the Laurentius Hospital in Roermond. SDCs were quantified by HPLC. Measurement of doxycycline concentrations was performed after extraction from serum samples. In brief, 500 μL of serum was mixed with 500 μL of internal standard (2.5 mg/L dantrolene) and 2 mL monosodium phosphate/sulphate buffer. After gently mixing for 30 s, 8 mL of dichloromethane was added for extraction. The mixture was vortexed for 1 min and centrifuged at 3200 g for 5 min. The organic phase was transferred to a clean tube and evaporated. The residue was dissolved in 200 μL of mobile phase (i.e. distilled water + 200 μL triethylamine + 1250 μL phosphoric acid 85%). The sample was passed through the chromatograph column (flow rate of 0.6 mL/min). The HPLC system was by Hitachi and equipped with a diode-array detector (DAD) set at 348 nm. The column was a LiChrospher 100-5 RP-18e, 120 × 4 mm, from Knauer. Definitions The primary outcome of this study was the first disease-related event (a new complication of chronic Q fever or chronic Q fever-related mortality) during treatment or within 1 year after the end of treatment with doxycycline plus hydroxychloroquine. The hazard for occurrence of the first event only was studied in case of multiple consecutive events, as multiple events within individuals are not independent of each other. Secondary outcomes were: (i) all-cause mortality during treatment or within 1 year after the end of treatment with doxycycline plus hydroxychloroquine; and (ii) PCR-positivity during treatment, defined as a new positive PCR, having been negative for at least 3 months or a persistent positive PCR for more than 6 months during treatment with doxycycline plus hydroxychloroquine. PCR-positivity was used as an outcome because the presence of C. burnetii is considered proven in the case of a positive PCR.1 The cut-off period of 1 year after the end of treatment was chosen because complications and disease-related mortality most often occur far inside this time frame.16 Conditions considered as complications of chronic Q fever were rupture or dissection of aneurysm; acute symptomatic aneurysm; arterial fistula; endoleak of vascular prosthesis; spondyl(odisc)itis or osteomyelitis; (cardiac) abscess; cerebrovascular accident (haemorrhagic or ischaemic)/transient ischaemic attack; cardiac arrest or tamponade during pericarditis. Cause of death was reviewed by two investigators in all cases (C. P. B.-R. and S. E. van R.) and classification of the relationship between death and chronic Q fever was performed by reaching a consensus. Death was defined as related to chronic Q fever in case of active disease and a cause of death related to chronic Q fever. Active disease was defined as C. burnetii phase I IgG ≥1:1024 or positive PCR on serum or tissue. Cause of death related to chronic Q fever was defined as sepsis/feverish episode with no other cause; brain infarct or haemorrhage during endocarditis or due to cerebral aneurysm; arterial fistula; ruptured/dissected aneurysm; heart failure; fatal arrhythmia or cardiac arrest during endocarditis; surgical complications; side effects of antibiotic therapy; clinical deterioration during active disease with no other cause; chronic Q fever as cause of death proven by autopsy; unknown cause in the presence of chronic Q fever-related complications or unknown cause without adequate chronic Q fever treatment. Side effects and severity of side effects, reasons for discontinuing and/or switching and dosage adjustments of antibiotics were reported. Furthermore, SDC values were reported in order to determine whether clinicians were aiming for SDC values >5 μg/mL, as is currently recommended in the literature by experts.9,10 Statistical methods Data were retrieved from electronically stored patient records, or paper records if applicable, and stored anonymously in a Microsoft Access 2010 database. Continuous data were compared by independent samples t-test (Welch method) or ANOVA (if >2 categories). For ANOVA with significant results, post hoc analysis was performed by Tukey’s range test. In univariable analysis, survival curves were compared with a Log-rank or Tarone–Ware test as appropriate. Multivariable analysis was performed with a Cox proportional hazards model. Covariates were selected based on previously identified predictors for the outcomes.16 For disease-related events, age, presence of prosthetic material before diagnosis of chronic Q fever and PCR-positivity during disease were encountered as covariates in the model. Age and presence of prosthetic material before diagnosis of chronic Q fever were encountered as covariates in the model for all-cause mortality and PCR-positivity during treatment. For all-cause mortality, cause-specific HR was calculated. For first disease-related event and PCR-positivity, subdistribution HR was calculated to account for right-censoring. Left-censoring was accounted for by stratification for the presence of complications before the start of treatment. As the effect of measuring SDC on primary and secondary outcomes may vary per hospital, leading to correlation of patients from the same hospital, a random effect for hospital was included by fitting shared-frailty terms in the model (assuming a Gaussian distribution of the frailty parameter). If patients were treated in multiple hospitals consecutively, the hospital in which the patient was diagnosed was selected. An additional analysis was performed to assess whether intensity of patient care explained the results, as the effect of measuring SDC could be a proxy for intensity of patient care. The ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks was considered a proxy for intensity of patient care. We compared this ratio for patients with treatment based on SDC and patients without treatment based on SDC, and for patients with and without primary and secondary outcomes. Finally, we repeated Cox-regression analysis for primary and secondary outcomes, with adjustment for the ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks in the model. The Cox-regression models were fitted with the ‘cmprsk’ and ‘survival’ packages in RStudio, version 3.2.2.17 The proportional hazard assumption was verified with both formal tests and graphically, using Schoenfeld residuals. Level of significance was set at a P value of <0.05. Descriptive data were generated in SPSS, version 21.0. Figures were made in RStudio, version 3.2.2. Results Of 439 chronic Q fever patients in the database, we identified 201 eligible patients with proven or probable chronic Q fever treated with doxycycline and hydroxychloroquine for at least 12 weeks (Figure 1). In 167 patients (83%), SDC was measured. The remaining 34 patients (17%) were treated without measurement of SDC. Doxycycline and hydroxychloroquine was the first treatment for chronic Q fever in 188 patients (94%). Characteristics of all patients and number of primary and secondary outcomes in total and per subgroup are summarized in Table 1. Table 1. Characteristics of proven and probable chronic Q fever patients treated with doxycycline and hydroxychloroquine for at least 12 weeks Patient characteristics  All  SDCa  No SDC  Number of patients (%)  201  167 (83)  34 (17)  Male gender (%)  167 (83)  138 (83)  29 (85)  Mean age, years (SD)  68 (11)  68 (11)  71 (13)  Mean total follow-up, weeks (SD)  204 (98)  213(95)  159 (102)  Treatment   median maximum dosage of doxycycline, mg (IQR)  200 (200–300)  200 (200–300)  200 (200)b   mean total treatment duration, weeks (SD)  118 (66)  122 (65)  95 (71)   mean duration DH treatment, weeks (SD)c  77 (44)  82 (44)  54 (36)   DH as first treatment (%)d  188 (94)  157 (94)  31 (91)  Presence of complications before treatment  66 (33)  52 (31)  14 (41)  Classification   proven (%)  163 (81)  137 (82)  26 (76)   probable (%)  38 (19)  30 (18)  8 (24)  Focus of infection   vascular (%)  102 (51)  84 (50)  18 (53)   endocarditis (%)  48 (24)  40 (24)  8 (24)   combined (%)  33 (16)  28 (17)  5 (15)   other/no focus (%)  18 (9)  15 (9)  3 (9)  Clinical outcomes   all-cause mortality (%)e  28 (14)  23 (14)  5 (15)    mean time to death, weeks (SD)  90 (50)  88 (49)  101 (56)   chronic Q fever-related mortality (%)e  16 (8)  13 (8)  3 (9)    mean time to death, weeks (SD)  90 (47)  83 (44)  117 (64)   complications (%)e  41 (20)  28 (17)  13 (38)    mean time to complication in weeks (SD)  63 (51)  65 (50)  59 (56)   disease-related event (%)e,f  49 (24)  36 (22)  13 (38)    mean time to event, weeks (SD)  65 (50)  67 (48)  59 (56)   PCR-positivity (%)g  20 (10)  19 (11)  1 (3)    mean time to PCR-positivity, weeks (SD)  39 (17)  39 (18)  31h  Patient characteristics  All  SDCa  No SDC  Number of patients (%)  201  167 (83)  34 (17)  Male gender (%)  167 (83)  138 (83)  29 (85)  Mean age, years (SD)  68 (11)  68 (11)  71 (13)  Mean total follow-up, weeks (SD)  204 (98)  213(95)  159 (102)  Treatment   median maximum dosage of doxycycline, mg (IQR)  200 (200–300)  200 (200–300)  200 (200)b   mean total treatment duration, weeks (SD)  118 (66)  122 (65)  95 (71)   mean duration DH treatment, weeks (SD)c  77 (44)  82 (44)  54 (36)   DH as first treatment (%)d  188 (94)  157 (94)  31 (91)  Presence of complications before treatment  66 (33)  52 (31)  14 (41)  Classification   proven (%)  163 (81)  137 (82)  26 (76)   probable (%)  38 (19)  30 (18)  8 (24)  Focus of infection   vascular (%)  102 (51)  84 (50)  18 (53)   endocarditis (%)  48 (24)  40 (24)  8 (24)   combined (%)  33 (16)  28 (17)  5 (15)   other/no focus (%)  18 (9)  15 (9)  3 (9)  Clinical outcomes   all-cause mortality (%)e  28 (14)  23 (14)  5 (15)    mean time to death, weeks (SD)  90 (50)  88 (49)  101 (56)   chronic Q fever-related mortality (%)e  16 (8)  13 (8)  3 (9)    mean time to death, weeks (SD)  90 (47)  83 (44)  117 (64)   complications (%)e  41 (20)  28 (17)  13 (38)    mean time to complication in weeks (SD)  63 (51)  65 (50)  59 (56)   disease-related event (%)e,f  49 (24)  36 (22)  13 (38)    mean time to event, weeks (SD)  65 (50)  67 (48)  59 (56)   PCR-positivity (%)g  20 (10)  19 (11)  1 (3)    mean time to PCR-positivity, weeks (SD)  39 (17)  39 (18)  31h  a SDC, serum doxycycline concentration. b IQR of patients used 200 mg as maximum dosage; therefore, single value displayed. c DH, doxycycline and hydroxychloroquine. d Defined as first treatment started or treatment with other antibiotics prior to start of doxycycline and hydroxychloroquine less than 4 weeks. e During treatment or within 1 year after end of treatment. f Disease-related event: complication or chronic Q fever-related mortality. g During treatment. h No SD shown as only one input value. Figure 1. View largeDownload slide Flowchart of inclusion. Figure 1. View largeDownload slide Flowchart of inclusion. SDC measurement In 167 patients, 652 SDC measurements were taken (median 3 SDC per patient, IQR 2–5). The first SDC was >5 μg/mL in 106 patients (63%), all with a doxycycline dosage of 200 mg/day. In 145 patients (87%), at least one SDC of >5 μg/mL was measured. Doxycycline dose was adjusted in 68 patients (41%) in whom SDC was measured: it was increased in 55 patients (81%), decreased in 4 patients (6%) and both increased and decreased in 9 patients (13%). In patients with increased dosage and SDC measured before and afterwards (n = 59), mean SDC before the first increase was 4.1 μg/mL and mean SDC after the first increase was 5.9 μg/mL (Figure 2). Figure 2. View largeDownload slide Mean serum doxycycline concentration (SDC) before dosage adjustment and mean SDC after dosage adjustment. In case of multiple adjustments, first adjustment selected. Only cases with increase in dosage shown. Cases with no afterwards or before measurement are not shown in these graphs. Figure 2. View largeDownload slide Mean serum doxycycline concentration (SDC) before dosage adjustment and mean SDC after dosage adjustment. In case of multiple adjustments, first adjustment selected. Only cases with increase in dosage shown. Cases with no afterwards or before measurement are not shown in these graphs. Mean SDC was studied for different subgroups of patients. In both patients with and without primary or secondary outcomes, overall mean SDC was >5 μg/mL (Figure 3). Mean SDC values were higher for patients with the outcomes, compared with those without the outcomes. For disease-related events and all-cause mortality, this difference was not significant. For patients with PCR-positivity, this difference was significant (P = 0.02). Furthermore, during treatment with different doxycycline dosages, overall mean SDC was >5 μg/mL with no significant differences in SDC between dosages (P = 0.70). All measured SDC values with corresponding doxycycline dosages are shown in Figure 4, with multiple measurements within individuals included. The number of patients with and without measurement of SDC diagnosed in time is shown in Figure 5: no evident increase in the proportion of patients with SDC measurements in time was observed. In patients without SDC measurement, doxycycline dose was increased in one patient (3%); all other patients used doxycycline 200 mg/day. Figure 3. View largeDownload slide Differences in SDC for patients with and without primary and secondary outcomes. The differences in first disease-related event and all-cause mortality are not statistically significant. The difference for PCR failure is statistically significant (P = 0.02). Dosage was unknown during 11 SDC measurements in one patient. SDC = serum doxycycline concentration. Mean serum doxycycline concentration in µg/mL shown below figure. Dosage unknown during 11 SDC measurements in one patient. Figure 3. View largeDownload slide Differences in SDC for patients with and without primary and secondary outcomes. The differences in first disease-related event and all-cause mortality are not statistically significant. The difference for PCR failure is statistically significant (P = 0.02). Dosage was unknown during 11 SDC measurements in one patient. SDC = serum doxycycline concentration. Mean serum doxycycline concentration in µg/mL shown below figure. Dosage unknown during 11 SDC measurements in one patient. Figure 4. View largeDownload slide All measured serum doxycycline concentrations in µg/mL, with corresponding dosages of doxycycline in mg/day. Multiple SDC measurements within individuals all included. Dosage was unknown during 11 SDC measurements in one patient. Figure 4. View largeDownload slide All measured serum doxycycline concentrations in µg/mL, with corresponding dosages of doxycycline in mg/day. Multiple SDC measurements within individuals all included. Dosage was unknown during 11 SDC measurements in one patient. Figure 5. View largeDownload slide Percentage of patients with or without measurement of SDC. Total number of patients: 8 in 2008, 20 in 2009, 39 in 2010, 67 in 2011, 18 in 2012, 18 in 2013, 21 in 2014, 9 in 2015, 1 in 2016. Figure 5. View largeDownload slide Percentage of patients with or without measurement of SDC. Total number of patients: 8 in 2008, 20 in 2009, 39 in 2010, 67 in 2011, 18 in 2012, 18 in 2013, 21 in 2014, 9 in 2015, 1 in 2016. SDC in relation to primary and secondary outcomes The primary outcome (first disease-related event) occurred in 36 patients (22%) with SDC measurement and in 13 patients (38%) without SDC measurement. As for secondary outcomes, all-cause mortality occurred in 23 patients (14%) with SDC measurement and 5 patients (15%) without SDC measurement. PCR-positivity occurred in 19 patients (11%) with SDC measurement and 1 patient (3%) without SDC measurement. Kaplan–Meier survival curves for disease-related events and all-cause mortality are shown in Figure 6. In univariable survival analysis, there was a significant difference in disease-related events between patients with SDC measured and patients without SDC measured, with a lower risk for the outcome for those with SDC measured (P = 0.008). No significant differences in all-cause mortality were observed between patients with and without SDC measured (P = 0.64). Figure 6. View largeDownload slide Kaplan–Meier curves for all-cause mortality and disease-related events (unadjusted for covariates): comparison of patients with and without treatment based on serum doxycycline concentrations (SDC). Figure 6. View largeDownload slide Kaplan–Meier curves for all-cause mortality and disease-related events (unadjusted for covariates): comparison of patients with and without treatment based on serum doxycycline concentrations (SDC). In multivariable analysis, treatment based on SDC was associated with a significantly lower risk for disease-related events (HR 0.51, 95% CI 0.26–0.97, P = 0.04). It was not associated with a lower risk for all-cause mortality during treatment or within 1 year after the end of treatment (HR 0.93, 95% CI 0.35–2.51, P = 0.89) or PCR-positivity during treatment (HR 5.87, 95% CI 0.73–46.98, P = 0.10) (see Table 2). Table 2. HR for association between measurement of serum doxycycline concentrations and occurrence of primary and secondary outcomes Outcome  SDCa  No SDC  (S)HRb (95% CI) for SDC measurement  P value  Number of patients  167  34  –  –  Disease-related eventsc,d  36 (22)  13 (38)  0.51 (0.26–0.97)  0.04  All-cause mortalityc,e  23 (14)  5 (15)  0.93 (0.35–2.51)  0.89  PCR-positivityc,d  19 (11)  1 (3)  5.87 (0.73–46.98)  0.10  Outcome  SDCa  No SDC  (S)HRb (95% CI) for SDC measurement  P value  Number of patients  167  34  –  –  Disease-related eventsc,d  36 (22)  13 (38)  0.51 (0.26–0.97)  0.04  All-cause mortalityc,e  23 (14)  5 (15)  0.93 (0.35–2.51)  0.89  PCR-positivityc,d  19 (11)  1 (3)  5.87 (0.73–46.98)  0.10  a SDC, serum doxycycline concentration. b Subdistribution hazard ratio. c In univariable analysis non-significant differences (χ2 test, Fisher’s exact test or independent samples t-test as appropriate). For disease-related events: presence of prosthetic material prior to occurrence of complications and age were significant covariates. PCR-positivity was a non-significant covariate. For all-cause mortality and PCR failure: age and presence of prosthetic material prior to occurrence of complications were non-significant covariates. d Subdistribution HR. e Cause-specific HR. Additional adjustment for intensity of patient care The ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks was significantly higher in patients with SDC measured (0.071 versus 0.057, P = 0.02). The ratio was significantly higher in patients that died (of all causes) in comparison to patients that survived (0.093 versus 0.021, P = 0.001). No differences in this ratio were found between patients with and without disease-related events (0.076 versus 0.067, P = 0.11) or PCR-positivity during treatment (0.077 versus 0.068, P = 0.21). In Cox-regression analysis, correction for the ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks did not significantly change the HR estimates and CI for primary and secondary outcomes. Side effects Patients treated based on SDC reported side effects in 78% (n = 131), in patients without measurement of SDC, side effects occurred in 62% (n = 21); see Table 3. Mean maximum dosages were not significantly different in patients experiencing side effects and in patients stopping due to side effects in comparison to those who did not (mean 232 versus 233, P = 0.97 and mean 228 versus 235, P = 0.36). Moreover, there was no significant difference in the mean SDC value for those with side effects (mean 6.18 μg/mL) compared with those without side effects (mean 5.76 μg/mL, P = 0.37). Table 3. Reasons for stop and switch and severity of side effects Variable  All  SDCa  No SDC  Number of patients (%)  201 (100)  167 (83)  34 (17)  Reason for stop/switch   side effects (%)  80 (40)  63 (38)  17 (50)   deceased (%)  14 (7)  12 (7)  2 (6)   adequate treatment finished (%)  58 (29)  52 (31)  6 (18)   insufficient clinical response (%)  28 (14)  25 (15)  3 (9)   not stopped at end of observation period (%)  28 (14)  19 (11)  9 (26)   other (%)  30 (15)  26 (16)  4 (12)  Severity of side effects, n (%)   minor  152 (76)  131 (78)  21 (62)   hospitalization (%)  9 (4)  7 (4)  2 (6)   permanent damage (%)  14 (7)  11 (7)  3 (9)   life threatening (%)  3 (1)  3 (2)  –   potentially lethal (%)  2 (1)  2 (1)  –  Variable  All  SDCa  No SDC  Number of patients (%)  201 (100)  167 (83)  34 (17)  Reason for stop/switch   side effects (%)  80 (40)  63 (38)  17 (50)   deceased (%)  14 (7)  12 (7)  2 (6)   adequate treatment finished (%)  58 (29)  52 (31)  6 (18)   insufficient clinical response (%)  28 (14)  25 (15)  3 (9)   not stopped at end of observation period (%)  28 (14)  19 (11)  9 (26)   other (%)  30 (15)  26 (16)  4 (12)  Severity of side effects, n (%)   minor  152 (76)  131 (78)  21 (62)   hospitalization (%)  9 (4)  7 (4)  2 (6)   permanent damage (%)  14 (7)  11 (7)  3 (9)   life threatening (%)  3 (1)  3 (2)  –   potentially lethal (%)  2 (1)  2 (1)  –  a SDC, serum doxycycline concentrations. Discussion Treatment based on SDC was associated with a lower hazard for disease-related events compared with treatment with a fixed doxycycline dose. The effect may be explained through better treatment accomplished by optimized dosage of doxycycline, or by better adherence to medication in patients with measurement of SDC. We hypothesized that treatment based on SDC could be a proxy for intensity of patient care, which may explain favourable outcomes for patients in whom treatment was based on SDC. Indeed, the ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks was higher in patients treated based on SDC. Adjustment for this ratio did not lead to different conclusions. This could be due to the fact that both parameters represent exactly the same underlying variable. However, the ratio was associated with all-cause mortality while SDC measurement was not. Therefore, the distribution of this variable is different from SDC measurement and we must conclude that increased intensity of patient care cannot explain the observed effect entirely. Measurement of SDC was not associated with all-cause mortality or PCR-positivity during treatment. The lack of an association between treatment based on SDC and all-cause mortality or PCR-positivity may be due to lack of power: we only observed 28 deaths within 1 year of stopping doxycycline plus hydroxychloroquine and 20 cases with PCR-positivity during treatment. In earlier studies, SDC > 5 μg/mL proved to be beneficial in terms of serological response. Here, to our knowledge, we provide the first study of the effect of treatment based on SDC on clinical endpoints, which makes this a unique study. Due to the retrospective origin of this study, it was not possible to draw conclusions on the classical dose–response relationship. Clinicians titrate doxycycline dosage up to SDC of >5 μg/mL, in accordance with recommendations in the literature, or address incompliance in order to improve SDC.9,10 The fact that no differences in SDC were observed under different doxycycline dosages supports the hypothesis that clinicians in practice aim at values >5 μg/mL. This leads to bias when assessing the correlation between SDC values and clinical outcomes: in patients with the most severe disease, clinicians will perhaps aim for higher SDC values. Therefore, higher SDC values will be associated with worse outcomes, probably owing to reverse causality: in patients with most severe disease, SDC is titrated up to higher values. Therefore, we decided to assess the effect of a treatment strategy in which dosage and coaching of patients with regard to compliance is based on SDC, instead of evaluating the actual SDC values. Besides the problems with assessing the relationship between SDC values and clinical outcomes, there may be a risk of bias caused by confounding by indication when studying the effect of measuring any SDC as well. Clinicians may measure SDC more often in patients with more pronounced disease or more complications. However, we observed an effect in the opposite direction: in patients with no measurement of SDC, complications occurred more frequently than in patients with SDC measured. The strength of the effect and relationship between SDC measurement and the other outcomes may be underestimated by confounding by indication. Another potential issue is that there is a ‘learning effect’ among clinicians: the more patients they have treated over time, the more experience they have gained and the better they are able to manage these patients. Theoretically, this may result in an increase of the proportion of patients in whom SDC was measured over time. However, we did not observe an evident effect of time on the proportion of patients with SDC measured (although the numbers in the group of patients without measurement of SDC are small), leading to less favourable treatment outcome for these patients. Finally, for this analysis, we only included patients treated with the combination of doxycycline and hydroxychloroquine. Therefore, it is uncertain whether these results are generalizable to patients treated with doxycycline monotherapy or doxycycline combined with other antibiotics. Altogether, SDC measurement seems beneficial during treatment for chronic Q fever. Therefore, we recommend measuring of SDC during treatment of chronic Q fever with doxycycline and hydroxychloroquine. We advise aiming for SDC >5 μg/mL because clinicians strived for SDC values >5 μg/mL in this study, and this strategy was successful. No literature is available on upper target levels during treatment of chronic Q fever. From clinical experience, we advise clinicians to strive for values >5 μg/mL but <10 μg/mL. The recommendation on the upper target range is not evidence-based: there is no literature on toxic SDC values during prolonged treatment with doxycycline. In conclusion, treatment based on SDC was associated with a lower hazard for disease-related events, but not for all-cause mortality or PCR-positivity. Intensity of patient care could not entirely explain the association we found, suggesting that SDC-based dosing itself decreases the risk for disease-related events in these patients, potentially through more optimal dosing or through improved compliance. We therefore recommend measurement of SDC and to strive for SDC >5 μg/mL and <10 μg/mL during treatment of chronic Q fever. Acknowledgements A part of the content of this manuscript was presented as a poster during the ECCMID congress in Vienna (April 2017) and as a poster during the ESCCAR congress in Marseille (P2-13, June 2017). No other acknowledgements. List of participating hospitals Amphia Hospital in Breda, Atrium Medical Center in Heerlen, Bernhoven Hospital in Uden, Bravis Hospital in Roosendaal, Canisius-Wilhelmina Hospital in Nijmegen, Catharina Hospital in Eindhoven, Diakonessenhuis in Utrecht, Elkerliek Hospital in Helmond, Erasmus Medical Center in Rotterdam, Hospital Gelderse Vallei in Ede, Gelre Hospital in Apeldoorn, Groene Hart Hospital in Gouda, Jeroen Bosch Hospital in ’s–Hertogenbosch, Leids University Medical Center in Leiden, Izore Laboratory in Leeuwarden, Isala Clinic in Zwolle, Laurentius Hospital in Roermond, Maasstad Hospital in Rotterdam, Máxima Medical Center in Eindhoven, Meander Medical Center in Amersfoort, Medisch Spectrum Twente in Enschede, Onze Lieve Vrouwe Gasthuis in Amsterdam, Radboud University Medical Center in Nijmegen, Reinier de Graaf Hospital in Delft, Rijnstate Hospital in Arnhem, St. Elisabeth Hospital in Tilburg, St. Antonius Hospital in Nieuwegein and University Medical Center Utrecht in Utrecht.  The following hospitals provided cooperation but without chronic Q fever patients: Admiraal de Ruyter Hospital in Goes, Albert Schweitzer Hospital in Dordrecht, Medical Center Haaglanden Bronovo in The Hague, Diaconessenhuis in Leiden, St. Franciscus Gasthuis in Rotterdam, St. Jansdal Hospital in Harderwijk and Vlietland Hospital in Schiedam. Funding This work was supported by foundation Q-support (grant number UMCU150401-00), granted to the research team of Sonja van Roeden, Marieke de Regt, Andy Hoepelman and Jan Jelrik Oosterheert. Transparency declarations S. E. van R., A. I. M. H. and J. J. O. received funding from Q-support and Institut Mérieux for other research purposes. The organization of C. P. B.-R. received funding from Q-support for other research purposes. Besides the reported received funding, S. E. van R., M. J. A. de R., C. P. B.-R., A. I. M. H. and J. J. O. have nothing to declare. L. M. K., P. C. W. and A. V. W. have nothing to declare. Author contributions S. E. van R.: data collection, data management, data analysis and interpretation, making figures, writing of manuscript; C. P. B.-R.: data interpretation, writing manuscript; M. J. A. de R.: epidemiological advice, data analysis and interpretation, writing manuscript; A. V. W.: data interpretation, writing manuscript; L. M. K.: data collection; A. I. M. H.: data interpretation, writing manuscript; P. C. W.: data interpretation, writing manuscript; J. J. O.: supervision during data collection, data management and analysis, data interpretation, writing of manuscript. References 1 Eldin C, Melenotte C, Mediannikov O et al.   From Q fever to Coxiella burnetii: a paradigm change. Clin Microbiol Rev  2017; 30: 115– 90. Google Scholar CrossRef Search ADS PubMed  2 Maurin M, Raoult D. Q fever. Clin Microbiol Rev  1999; 12: 518– 53. Google Scholar PubMed  3 Raoult D, Houpikian P, Tissot Dupont H et al.   Treatment of Q fever endocarditis: comparison of 2 regimens containing doxycycline and ofloxacin or hydroxychloroquine. Arch Intern Med  1999; 159: 167– 73. Google Scholar CrossRef Search ADS PubMed  4 Million M, Raoult D. Recent advances in the study of Q fever epidemiology, diagnosis and management. J Infect  2015; 71: S2– 9. Google Scholar CrossRef Search ADS PubMed  5 Anderson A, Bijlmer H, Fournier PE et al.   Diagnosis and management of Q fever—United States 2013: recommendations from CDC and the Q fever working group. MMWR Recomm Rep  2013; 62: 1– 30. Google Scholar PubMed  6 Samuel JE, Hendrix LR. Laboratory maintenance of Coxiella burnetii. Curr Protoc Microbiol  2009; Chapter 6: Unit 6C.1. 7 Vincent GA, Graves SR, Robson JM et al.   Isolation of Coxiella burnetii from serum of patients with acute Q fever. J Microbiol Methods  2015; 119: 74– 8. Google Scholar CrossRef Search ADS PubMed  8 Omsland A, Hackstadt T, Heinzen RA. Bringing culture to the uncultured: Coxiella burnetii and lessons for obligate intracellular bacterial pathogens. PLoS Pathog  2013; 9: e1003540. Google Scholar CrossRef Search ADS PubMed  9 Rolain JM, Mallet MN, Raoult D. Correlation between serum doxycycline concentrations and serologic evolution in patients with Coxiella burnetii endocarditis. J Infect Dis  2003; 188: 1322– 5. Google Scholar CrossRef Search ADS PubMed  10 Lecaillet A, Mallet MN, Raoult D et al.   Therapeutic impact of the correlation of doxycycline serum concentrations and the decline of phase I antibodies in Q fever endocarditis. J Antimicrob Chemother  2009; 63: 771– 4. Google Scholar CrossRef Search ADS PubMed  11 Rolain JM, Boulos A, Mallet MN et al.   Correlation between ratio of serum doxycycline concentration to MIC and rapid decline of antibody levels during treatment of Q fever endocarditis. Antimicrob Agents Chemother  2005; 49: 2673– 6. Google Scholar CrossRef Search ADS PubMed  12 Rouli L, Rolain JM, El Filali A et al.   Genome sequence of Coxiella burnetii 109, a doxycycline-resistant clinical isolate. J Bacteriol  2012; 194: 6939. Google Scholar CrossRef Search ADS PubMed  13 Vranakis I, De Bock PJ, Papadioti A et al.   Quantitative proteome profiling of C. burnetii under tetracycline stress conditions. PLoS One  2012; 7: e33599. Google Scholar CrossRef Search ADS PubMed  14 Rolain JM, Lambert F, Raoult D. Activity of telithromycin against thirteen new isolates of C. burnetii including three resistant to doxycycline. Ann N Y Acad Sci  2005; 1063: 252– 6. Google Scholar CrossRef Search ADS PubMed  15 Wegdam-Blans MC, Kampschreur LM, Delsing CE et al.   Chronic Q fever: review of the literature and a proposal of new diagnostic criteria. J Infect  2012; 64: 247– 59. Google Scholar CrossRef Search ADS PubMed  16 van Roeden SE, Wever PC, Kampschreur LM et al.   Chronic Q fever-related complications and mortality: data from a nationwide cohort. In: ECCMID 2017 eLibrary Material of the 27th European Congress of Clinical Microbiology and Infectious Diseases, Vienna, Austria, 2017. Abstract OSO906. European Society of Clinical Microbiology and Infectious Diseases, Basel, Switzerland. https://www.escmid.org/research_projects/eccmid/past_eccmids/. 17 The Comprehensive R Archive Network. ‘cmprsk’ and ‘survival’. https://cran.r-project.org. © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please email: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Antimicrobial Chemotherapy Oxford University Press

The effect of measuring serum doxycycline concentrations on clinical outcomes during treatment of chronic Q fever

Loading next page...
 
/lp/ou_press/the-effect-of-measuring-serum-doxycycline-concentrations-on-clinical-7SVqKBv5dI
Publisher
Oxford University Press
ISSN
0305-7453
eISSN
1460-2091
D.O.I.
10.1093/jac/dkx487
Publisher site
See Article on Publisher Site

Abstract

Abstract Background First choice treatment for chronic Q fever is doxycycline plus hydroxychloroquine. Serum doxycycline concentration (SDC) >5 μg/mL has been associated with a favourable serological response, but the effect on clinical outcomes is unknown. Objectives To assess the effect of measuring SDC during treatment of chronic Q fever on clinical outcomes. Methods We performed a retrospective cohort study, to assess the effect of measuring SDC on clinical outcomes in patients treated with doxycycline and hydroxychloroquine for chronic Q fever. Primary outcome was the first disease-related event (new complication or chronic Q fever-related mortality); secondary outcomes were all-cause mortality and PCR-positivity. Multivariable analysis was performed with a Cox proportional hazards model, with shared-frailty terms for different hospitals included. Results We included 201 patients (mean age 68 years, 83% male): in 167 patients (83%) SDC was measured, 34 patients (17%) were treated without SDC measurement. First SDC was >5 μg/mL in 106 patients (63%), all with 200 mg doxycycline daily. In patients with SDC measured, dosage was adjusted in 41% (n = 68), concerning an increase in 64 patients. Mean SDC was 4.1 μg/mL before dosage increase, and 5.9 μg/mL afterwards. SDC measurement was associated with a lower risk for disease-related events (HR 0.51, 95% CI 0.26–0.97, P = 0.04), but not with all-cause mortality or PCR-positivity. Conclusions SDC measurement decreases the risk for disease-related events, potentially through more optimal dosing or improved compliance. We recommend measurement of SDC and striving for SDC >5 μg/mL and <10 μg/mL during treatment of chronic Q fever. Introduction Q fever is caused by the intracellular Gram-negative bacterium Coxiella burnetii. After primary infection, approximately 60% of patients remain asymptomatic. The remaining patients develop illnesses such as flu-like symptoms or pneumonia.1 A small proportion of patients develop chronic Q fever after primary infection.1 In contrast to acute Q fever, chronic Q fever is a potential life-threatening disease. Most often, it leads to endocarditis, infection of vascular prostheses or arterial aneurysms.1 The recommended first-line treatment for chronic Q fever is doxycycline combined with hydroxychloroquine for at least 18 months, or alternatively doxycycline combined with a quinolone.2–5 Culturing of C. burnetii is not routinely performed in practice, because it is difficult and only allowed in laboratories with a Biosafety Level 3 facility.6–8 Therefore, monitoring of treatment effect mainly depends on measurement of antibody titres, monitoring by PCR for C. burnetii in blood or tissue, and follow-up of the patient’s clinical condition.1 To assess the adequacy of doxycycline dosage and compliance with therapy, measurement of serum doxycycline concentration (SDC) is a potential tool. In two clinical studies, SDC >5 μg/mL was associated with favourable serological response, and in another study, higher SDC to MIC ratios were associated with rapid decline in phase I IgG antibodies to C. burnetii.9–11 This can be explained by doxycycline being more effective at higher concentrations, or resistance leading to decreased effectiveness in patients with lower SDC.12–14 In two reports, resistance was reported in 6%–23% of isolates.9,14 The effect of measuring SDC on clinical endpoints such as complications and mortality and PCR-positivity, has not been studied before. By evaluating data from a nationwide cohort of chronic Q fever patients, we aimed to assess the effect of treatment when dosage was based on SDC, as performed in clinical practice, for these clinical endpoints in chronic Q fever patients treated with doxycycline and hydroxychloroquine. Patients and methods Data from the Dutch national chronic Q fever database were used: in this database, detailed data are recorded regarding the treatment of patients with proven or probable chronic Q fever diagnosed since the start of the Dutch Q fever outbreak (1 January 2007). Using these data, we retrospectively assessed the effect of measuring SDC in patients who were treated with doxycycline and hydroxychloroquine for at least 12 weeks. As this is a retrospective study and action taken by clinicians on SDC results is not standardized, we could not study the effect of SDC values on clinical outcomes: clinicians intervene based on SDC values and may aim for higher values in patients with more severe disease, which leads to bias. Therefore, we studied the effect of a treatment strategy in which dosage adjustment and coaching regarding compliance was based on SDC. Data collection and inclusion of patients The design of the Dutch national chronic Q fever database was approved by the Medical Ethics Committee of the University Medical Center in Utrecht. This database contains complete follow-up data of all chronic Q fever patients ≥18 years of age in the Netherlands from 1 January 2007 to 1 May 2016, from 28 hospitals in the Netherlands (see Acknowledgements). Diagnosis and classification of chronic Q fever was based on the Dutch chronic Q fever consensus group criteria.15 Clinicians identified patients based on a positive PCR for C. burnetii in serum or tissue and/or C. burnetii phase I IgG ≥1:1024. Patients with a serological profile and clinical condition matching acute Q fever were excluded. The Dutch chronic Q fever consensus group criteria discriminate between proven, probable and possible chronic Q fever. For this study, only patients with proven or probable chronic Q fever were included. Possible chronic Q fever patients were not included because the presence of clinical relevant infection in possible chronic Q fever patients is questionable and these patients do not have an indication for treatment; therefore, evaluation of treatment-related aspects and outcomes is impossible.The observation period for all patients ended on 1 May 2016. Laboratory testing Microbiological testing consisted of an indirect fluorescent-antibody assay (IFA) for phase I and II IgG against C. burnetii in plasma or serum (Focus Diagnostics, Inc., Cypress, CA, USA or Fuller Diagnostics, LLC, Anchorage, AK, USA). Titration of antibody levels was carried out at different hospital sites with dilutions on a binary scale with a cut-off of 1:32. Furthermore, PCR for C. burnetii DNA in serum or plasma and, if applicable, in tissue samples was performed (NucliSENS easyMAG; bioMérieux, Marcy l’Etoile, France). All SDC measurements were performed in the apothecary laboratory of the Jeroen Bosch Hospital in ’s-Hertogenbosch, with the exception of one measurement of a patient from the Laurentius Hospital in Roermond. SDCs were quantified by HPLC. Measurement of doxycycline concentrations was performed after extraction from serum samples. In brief, 500 μL of serum was mixed with 500 μL of internal standard (2.5 mg/L dantrolene) and 2 mL monosodium phosphate/sulphate buffer. After gently mixing for 30 s, 8 mL of dichloromethane was added for extraction. The mixture was vortexed for 1 min and centrifuged at 3200 g for 5 min. The organic phase was transferred to a clean tube and evaporated. The residue was dissolved in 200 μL of mobile phase (i.e. distilled water + 200 μL triethylamine + 1250 μL phosphoric acid 85%). The sample was passed through the chromatograph column (flow rate of 0.6 mL/min). The HPLC system was by Hitachi and equipped with a diode-array detector (DAD) set at 348 nm. The column was a LiChrospher 100-5 RP-18e, 120 × 4 mm, from Knauer. Definitions The primary outcome of this study was the first disease-related event (a new complication of chronic Q fever or chronic Q fever-related mortality) during treatment or within 1 year after the end of treatment with doxycycline plus hydroxychloroquine. The hazard for occurrence of the first event only was studied in case of multiple consecutive events, as multiple events within individuals are not independent of each other. Secondary outcomes were: (i) all-cause mortality during treatment or within 1 year after the end of treatment with doxycycline plus hydroxychloroquine; and (ii) PCR-positivity during treatment, defined as a new positive PCR, having been negative for at least 3 months or a persistent positive PCR for more than 6 months during treatment with doxycycline plus hydroxychloroquine. PCR-positivity was used as an outcome because the presence of C. burnetii is considered proven in the case of a positive PCR.1 The cut-off period of 1 year after the end of treatment was chosen because complications and disease-related mortality most often occur far inside this time frame.16 Conditions considered as complications of chronic Q fever were rupture or dissection of aneurysm; acute symptomatic aneurysm; arterial fistula; endoleak of vascular prosthesis; spondyl(odisc)itis or osteomyelitis; (cardiac) abscess; cerebrovascular accident (haemorrhagic or ischaemic)/transient ischaemic attack; cardiac arrest or tamponade during pericarditis. Cause of death was reviewed by two investigators in all cases (C. P. B.-R. and S. E. van R.) and classification of the relationship between death and chronic Q fever was performed by reaching a consensus. Death was defined as related to chronic Q fever in case of active disease and a cause of death related to chronic Q fever. Active disease was defined as C. burnetii phase I IgG ≥1:1024 or positive PCR on serum or tissue. Cause of death related to chronic Q fever was defined as sepsis/feverish episode with no other cause; brain infarct or haemorrhage during endocarditis or due to cerebral aneurysm; arterial fistula; ruptured/dissected aneurysm; heart failure; fatal arrhythmia or cardiac arrest during endocarditis; surgical complications; side effects of antibiotic therapy; clinical deterioration during active disease with no other cause; chronic Q fever as cause of death proven by autopsy; unknown cause in the presence of chronic Q fever-related complications or unknown cause without adequate chronic Q fever treatment. Side effects and severity of side effects, reasons for discontinuing and/or switching and dosage adjustments of antibiotics were reported. Furthermore, SDC values were reported in order to determine whether clinicians were aiming for SDC values >5 μg/mL, as is currently recommended in the literature by experts.9,10 Statistical methods Data were retrieved from electronically stored patient records, or paper records if applicable, and stored anonymously in a Microsoft Access 2010 database. Continuous data were compared by independent samples t-test (Welch method) or ANOVA (if >2 categories). For ANOVA with significant results, post hoc analysis was performed by Tukey’s range test. In univariable analysis, survival curves were compared with a Log-rank or Tarone–Ware test as appropriate. Multivariable analysis was performed with a Cox proportional hazards model. Covariates were selected based on previously identified predictors for the outcomes.16 For disease-related events, age, presence of prosthetic material before diagnosis of chronic Q fever and PCR-positivity during disease were encountered as covariates in the model. Age and presence of prosthetic material before diagnosis of chronic Q fever were encountered as covariates in the model for all-cause mortality and PCR-positivity during treatment. For all-cause mortality, cause-specific HR was calculated. For first disease-related event and PCR-positivity, subdistribution HR was calculated to account for right-censoring. Left-censoring was accounted for by stratification for the presence of complications before the start of treatment. As the effect of measuring SDC on primary and secondary outcomes may vary per hospital, leading to correlation of patients from the same hospital, a random effect for hospital was included by fitting shared-frailty terms in the model (assuming a Gaussian distribution of the frailty parameter). If patients were treated in multiple hospitals consecutively, the hospital in which the patient was diagnosed was selected. An additional analysis was performed to assess whether intensity of patient care explained the results, as the effect of measuring SDC could be a proxy for intensity of patient care. The ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks was considered a proxy for intensity of patient care. We compared this ratio for patients with treatment based on SDC and patients without treatment based on SDC, and for patients with and without primary and secondary outcomes. Finally, we repeated Cox-regression analysis for primary and secondary outcomes, with adjustment for the ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks in the model. The Cox-regression models were fitted with the ‘cmprsk’ and ‘survival’ packages in RStudio, version 3.2.2.17 The proportional hazard assumption was verified with both formal tests and graphically, using Schoenfeld residuals. Level of significance was set at a P value of <0.05. Descriptive data were generated in SPSS, version 21.0. Figures were made in RStudio, version 3.2.2. Results Of 439 chronic Q fever patients in the database, we identified 201 eligible patients with proven or probable chronic Q fever treated with doxycycline and hydroxychloroquine for at least 12 weeks (Figure 1). In 167 patients (83%), SDC was measured. The remaining 34 patients (17%) were treated without measurement of SDC. Doxycycline and hydroxychloroquine was the first treatment for chronic Q fever in 188 patients (94%). Characteristics of all patients and number of primary and secondary outcomes in total and per subgroup are summarized in Table 1. Table 1. Characteristics of proven and probable chronic Q fever patients treated with doxycycline and hydroxychloroquine for at least 12 weeks Patient characteristics  All  SDCa  No SDC  Number of patients (%)  201  167 (83)  34 (17)  Male gender (%)  167 (83)  138 (83)  29 (85)  Mean age, years (SD)  68 (11)  68 (11)  71 (13)  Mean total follow-up, weeks (SD)  204 (98)  213(95)  159 (102)  Treatment   median maximum dosage of doxycycline, mg (IQR)  200 (200–300)  200 (200–300)  200 (200)b   mean total treatment duration, weeks (SD)  118 (66)  122 (65)  95 (71)   mean duration DH treatment, weeks (SD)c  77 (44)  82 (44)  54 (36)   DH as first treatment (%)d  188 (94)  157 (94)  31 (91)  Presence of complications before treatment  66 (33)  52 (31)  14 (41)  Classification   proven (%)  163 (81)  137 (82)  26 (76)   probable (%)  38 (19)  30 (18)  8 (24)  Focus of infection   vascular (%)  102 (51)  84 (50)  18 (53)   endocarditis (%)  48 (24)  40 (24)  8 (24)   combined (%)  33 (16)  28 (17)  5 (15)   other/no focus (%)  18 (9)  15 (9)  3 (9)  Clinical outcomes   all-cause mortality (%)e  28 (14)  23 (14)  5 (15)    mean time to death, weeks (SD)  90 (50)  88 (49)  101 (56)   chronic Q fever-related mortality (%)e  16 (8)  13 (8)  3 (9)    mean time to death, weeks (SD)  90 (47)  83 (44)  117 (64)   complications (%)e  41 (20)  28 (17)  13 (38)    mean time to complication in weeks (SD)  63 (51)  65 (50)  59 (56)   disease-related event (%)e,f  49 (24)  36 (22)  13 (38)    mean time to event, weeks (SD)  65 (50)  67 (48)  59 (56)   PCR-positivity (%)g  20 (10)  19 (11)  1 (3)    mean time to PCR-positivity, weeks (SD)  39 (17)  39 (18)  31h  Patient characteristics  All  SDCa  No SDC  Number of patients (%)  201  167 (83)  34 (17)  Male gender (%)  167 (83)  138 (83)  29 (85)  Mean age, years (SD)  68 (11)  68 (11)  71 (13)  Mean total follow-up, weeks (SD)  204 (98)  213(95)  159 (102)  Treatment   median maximum dosage of doxycycline, mg (IQR)  200 (200–300)  200 (200–300)  200 (200)b   mean total treatment duration, weeks (SD)  118 (66)  122 (65)  95 (71)   mean duration DH treatment, weeks (SD)c  77 (44)  82 (44)  54 (36)   DH as first treatment (%)d  188 (94)  157 (94)  31 (91)  Presence of complications before treatment  66 (33)  52 (31)  14 (41)  Classification   proven (%)  163 (81)  137 (82)  26 (76)   probable (%)  38 (19)  30 (18)  8 (24)  Focus of infection   vascular (%)  102 (51)  84 (50)  18 (53)   endocarditis (%)  48 (24)  40 (24)  8 (24)   combined (%)  33 (16)  28 (17)  5 (15)   other/no focus (%)  18 (9)  15 (9)  3 (9)  Clinical outcomes   all-cause mortality (%)e  28 (14)  23 (14)  5 (15)    mean time to death, weeks (SD)  90 (50)  88 (49)  101 (56)   chronic Q fever-related mortality (%)e  16 (8)  13 (8)  3 (9)    mean time to death, weeks (SD)  90 (47)  83 (44)  117 (64)   complications (%)e  41 (20)  28 (17)  13 (38)    mean time to complication in weeks (SD)  63 (51)  65 (50)  59 (56)   disease-related event (%)e,f  49 (24)  36 (22)  13 (38)    mean time to event, weeks (SD)  65 (50)  67 (48)  59 (56)   PCR-positivity (%)g  20 (10)  19 (11)  1 (3)    mean time to PCR-positivity, weeks (SD)  39 (17)  39 (18)  31h  a SDC, serum doxycycline concentration. b IQR of patients used 200 mg as maximum dosage; therefore, single value displayed. c DH, doxycycline and hydroxychloroquine. d Defined as first treatment started or treatment with other antibiotics prior to start of doxycycline and hydroxychloroquine less than 4 weeks. e During treatment or within 1 year after end of treatment. f Disease-related event: complication or chronic Q fever-related mortality. g During treatment. h No SD shown as only one input value. Figure 1. View largeDownload slide Flowchart of inclusion. Figure 1. View largeDownload slide Flowchart of inclusion. SDC measurement In 167 patients, 652 SDC measurements were taken (median 3 SDC per patient, IQR 2–5). The first SDC was >5 μg/mL in 106 patients (63%), all with a doxycycline dosage of 200 mg/day. In 145 patients (87%), at least one SDC of >5 μg/mL was measured. Doxycycline dose was adjusted in 68 patients (41%) in whom SDC was measured: it was increased in 55 patients (81%), decreased in 4 patients (6%) and both increased and decreased in 9 patients (13%). In patients with increased dosage and SDC measured before and afterwards (n = 59), mean SDC before the first increase was 4.1 μg/mL and mean SDC after the first increase was 5.9 μg/mL (Figure 2). Figure 2. View largeDownload slide Mean serum doxycycline concentration (SDC) before dosage adjustment and mean SDC after dosage adjustment. In case of multiple adjustments, first adjustment selected. Only cases with increase in dosage shown. Cases with no afterwards or before measurement are not shown in these graphs. Figure 2. View largeDownload slide Mean serum doxycycline concentration (SDC) before dosage adjustment and mean SDC after dosage adjustment. In case of multiple adjustments, first adjustment selected. Only cases with increase in dosage shown. Cases with no afterwards or before measurement are not shown in these graphs. Mean SDC was studied for different subgroups of patients. In both patients with and without primary or secondary outcomes, overall mean SDC was >5 μg/mL (Figure 3). Mean SDC values were higher for patients with the outcomes, compared with those without the outcomes. For disease-related events and all-cause mortality, this difference was not significant. For patients with PCR-positivity, this difference was significant (P = 0.02). Furthermore, during treatment with different doxycycline dosages, overall mean SDC was >5 μg/mL with no significant differences in SDC between dosages (P = 0.70). All measured SDC values with corresponding doxycycline dosages are shown in Figure 4, with multiple measurements within individuals included. The number of patients with and without measurement of SDC diagnosed in time is shown in Figure 5: no evident increase in the proportion of patients with SDC measurements in time was observed. In patients without SDC measurement, doxycycline dose was increased in one patient (3%); all other patients used doxycycline 200 mg/day. Figure 3. View largeDownload slide Differences in SDC for patients with and without primary and secondary outcomes. The differences in first disease-related event and all-cause mortality are not statistically significant. The difference for PCR failure is statistically significant (P = 0.02). Dosage was unknown during 11 SDC measurements in one patient. SDC = serum doxycycline concentration. Mean serum doxycycline concentration in µg/mL shown below figure. Dosage unknown during 11 SDC measurements in one patient. Figure 3. View largeDownload slide Differences in SDC for patients with and without primary and secondary outcomes. The differences in first disease-related event and all-cause mortality are not statistically significant. The difference for PCR failure is statistically significant (P = 0.02). Dosage was unknown during 11 SDC measurements in one patient. SDC = serum doxycycline concentration. Mean serum doxycycline concentration in µg/mL shown below figure. Dosage unknown during 11 SDC measurements in one patient. Figure 4. View largeDownload slide All measured serum doxycycline concentrations in µg/mL, with corresponding dosages of doxycycline in mg/day. Multiple SDC measurements within individuals all included. Dosage was unknown during 11 SDC measurements in one patient. Figure 4. View largeDownload slide All measured serum doxycycline concentrations in µg/mL, with corresponding dosages of doxycycline in mg/day. Multiple SDC measurements within individuals all included. Dosage was unknown during 11 SDC measurements in one patient. Figure 5. View largeDownload slide Percentage of patients with or without measurement of SDC. Total number of patients: 8 in 2008, 20 in 2009, 39 in 2010, 67 in 2011, 18 in 2012, 18 in 2013, 21 in 2014, 9 in 2015, 1 in 2016. Figure 5. View largeDownload slide Percentage of patients with or without measurement of SDC. Total number of patients: 8 in 2008, 20 in 2009, 39 in 2010, 67 in 2011, 18 in 2012, 18 in 2013, 21 in 2014, 9 in 2015, 1 in 2016. SDC in relation to primary and secondary outcomes The primary outcome (first disease-related event) occurred in 36 patients (22%) with SDC measurement and in 13 patients (38%) without SDC measurement. As for secondary outcomes, all-cause mortality occurred in 23 patients (14%) with SDC measurement and 5 patients (15%) without SDC measurement. PCR-positivity occurred in 19 patients (11%) with SDC measurement and 1 patient (3%) without SDC measurement. Kaplan–Meier survival curves for disease-related events and all-cause mortality are shown in Figure 6. In univariable survival analysis, there was a significant difference in disease-related events between patients with SDC measured and patients without SDC measured, with a lower risk for the outcome for those with SDC measured (P = 0.008). No significant differences in all-cause mortality were observed between patients with and without SDC measured (P = 0.64). Figure 6. View largeDownload slide Kaplan–Meier curves for all-cause mortality and disease-related events (unadjusted for covariates): comparison of patients with and without treatment based on serum doxycycline concentrations (SDC). Figure 6. View largeDownload slide Kaplan–Meier curves for all-cause mortality and disease-related events (unadjusted for covariates): comparison of patients with and without treatment based on serum doxycycline concentrations (SDC). In multivariable analysis, treatment based on SDC was associated with a significantly lower risk for disease-related events (HR 0.51, 95% CI 0.26–0.97, P = 0.04). It was not associated with a lower risk for all-cause mortality during treatment or within 1 year after the end of treatment (HR 0.93, 95% CI 0.35–2.51, P = 0.89) or PCR-positivity during treatment (HR 5.87, 95% CI 0.73–46.98, P = 0.10) (see Table 2). Table 2. HR for association between measurement of serum doxycycline concentrations and occurrence of primary and secondary outcomes Outcome  SDCa  No SDC  (S)HRb (95% CI) for SDC measurement  P value  Number of patients  167  34  –  –  Disease-related eventsc,d  36 (22)  13 (38)  0.51 (0.26–0.97)  0.04  All-cause mortalityc,e  23 (14)  5 (15)  0.93 (0.35–2.51)  0.89  PCR-positivityc,d  19 (11)  1 (3)  5.87 (0.73–46.98)  0.10  Outcome  SDCa  No SDC  (S)HRb (95% CI) for SDC measurement  P value  Number of patients  167  34  –  –  Disease-related eventsc,d  36 (22)  13 (38)  0.51 (0.26–0.97)  0.04  All-cause mortalityc,e  23 (14)  5 (15)  0.93 (0.35–2.51)  0.89  PCR-positivityc,d  19 (11)  1 (3)  5.87 (0.73–46.98)  0.10  a SDC, serum doxycycline concentration. b Subdistribution hazard ratio. c In univariable analysis non-significant differences (χ2 test, Fisher’s exact test or independent samples t-test as appropriate). For disease-related events: presence of prosthetic material prior to occurrence of complications and age were significant covariates. PCR-positivity was a non-significant covariate. For all-cause mortality and PCR failure: age and presence of prosthetic material prior to occurrence of complications were non-significant covariates. d Subdistribution HR. e Cause-specific HR. Additional adjustment for intensity of patient care The ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks was significantly higher in patients with SDC measured (0.071 versus 0.057, P = 0.02). The ratio was significantly higher in patients that died (of all causes) in comparison to patients that survived (0.093 versus 0.021, P = 0.001). No differences in this ratio were found between patients with and without disease-related events (0.076 versus 0.067, P = 0.11) or PCR-positivity during treatment (0.077 versus 0.068, P = 0.21). In Cox-regression analysis, correction for the ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks did not significantly change the HR estimates and CI for primary and secondary outcomes. Side effects Patients treated based on SDC reported side effects in 78% (n = 131), in patients without measurement of SDC, side effects occurred in 62% (n = 21); see Table 3. Mean maximum dosages were not significantly different in patients experiencing side effects and in patients stopping due to side effects in comparison to those who did not (mean 232 versus 233, P = 0.97 and mean 228 versus 235, P = 0.36). Moreover, there was no significant difference in the mean SDC value for those with side effects (mean 6.18 μg/mL) compared with those without side effects (mean 5.76 μg/mL, P = 0.37). Table 3. Reasons for stop and switch and severity of side effects Variable  All  SDCa  No SDC  Number of patients (%)  201 (100)  167 (83)  34 (17)  Reason for stop/switch   side effects (%)  80 (40)  63 (38)  17 (50)   deceased (%)  14 (7)  12 (7)  2 (6)   adequate treatment finished (%)  58 (29)  52 (31)  6 (18)   insufficient clinical response (%)  28 (14)  25 (15)  3 (9)   not stopped at end of observation period (%)  28 (14)  19 (11)  9 (26)   other (%)  30 (15)  26 (16)  4 (12)  Severity of side effects, n (%)   minor  152 (76)  131 (78)  21 (62)   hospitalization (%)  9 (4)  7 (4)  2 (6)   permanent damage (%)  14 (7)  11 (7)  3 (9)   life threatening (%)  3 (1)  3 (2)  –   potentially lethal (%)  2 (1)  2 (1)  –  Variable  All  SDCa  No SDC  Number of patients (%)  201 (100)  167 (83)  34 (17)  Reason for stop/switch   side effects (%)  80 (40)  63 (38)  17 (50)   deceased (%)  14 (7)  12 (7)  2 (6)   adequate treatment finished (%)  58 (29)  52 (31)  6 (18)   insufficient clinical response (%)  28 (14)  25 (15)  3 (9)   not stopped at end of observation period (%)  28 (14)  19 (11)  9 (26)   other (%)  30 (15)  26 (16)  4 (12)  Severity of side effects, n (%)   minor  152 (76)  131 (78)  21 (62)   hospitalization (%)  9 (4)  7 (4)  2 (6)   permanent damage (%)  14 (7)  11 (7)  3 (9)   life threatening (%)  3 (1)  3 (2)  –   potentially lethal (%)  2 (1)  2 (1)  –  a SDC, serum doxycycline concentrations. Discussion Treatment based on SDC was associated with a lower hazard for disease-related events compared with treatment with a fixed doxycycline dose. The effect may be explained through better treatment accomplished by optimized dosage of doxycycline, or by better adherence to medication in patients with measurement of SDC. We hypothesized that treatment based on SDC could be a proxy for intensity of patient care, which may explain favourable outcomes for patients in whom treatment was based on SDC. Indeed, the ratio of number of phase I IgG antibody titre measurements and follow-up duration in weeks was higher in patients treated based on SDC. Adjustment for this ratio did not lead to different conclusions. This could be due to the fact that both parameters represent exactly the same underlying variable. However, the ratio was associated with all-cause mortality while SDC measurement was not. Therefore, the distribution of this variable is different from SDC measurement and we must conclude that increased intensity of patient care cannot explain the observed effect entirely. Measurement of SDC was not associated with all-cause mortality or PCR-positivity during treatment. The lack of an association between treatment based on SDC and all-cause mortality or PCR-positivity may be due to lack of power: we only observed 28 deaths within 1 year of stopping doxycycline plus hydroxychloroquine and 20 cases with PCR-positivity during treatment. In earlier studies, SDC > 5 μg/mL proved to be beneficial in terms of serological response. Here, to our knowledge, we provide the first study of the effect of treatment based on SDC on clinical endpoints, which makes this a unique study. Due to the retrospective origin of this study, it was not possible to draw conclusions on the classical dose–response relationship. Clinicians titrate doxycycline dosage up to SDC of >5 μg/mL, in accordance with recommendations in the literature, or address incompliance in order to improve SDC.9,10 The fact that no differences in SDC were observed under different doxycycline dosages supports the hypothesis that clinicians in practice aim at values >5 μg/mL. This leads to bias when assessing the correlation between SDC values and clinical outcomes: in patients with the most severe disease, clinicians will perhaps aim for higher SDC values. Therefore, higher SDC values will be associated with worse outcomes, probably owing to reverse causality: in patients with most severe disease, SDC is titrated up to higher values. Therefore, we decided to assess the effect of a treatment strategy in which dosage and coaching of patients with regard to compliance is based on SDC, instead of evaluating the actual SDC values. Besides the problems with assessing the relationship between SDC values and clinical outcomes, there may be a risk of bias caused by confounding by indication when studying the effect of measuring any SDC as well. Clinicians may measure SDC more often in patients with more pronounced disease or more complications. However, we observed an effect in the opposite direction: in patients with no measurement of SDC, complications occurred more frequently than in patients with SDC measured. The strength of the effect and relationship between SDC measurement and the other outcomes may be underestimated by confounding by indication. Another potential issue is that there is a ‘learning effect’ among clinicians: the more patients they have treated over time, the more experience they have gained and the better they are able to manage these patients. Theoretically, this may result in an increase of the proportion of patients in whom SDC was measured over time. However, we did not observe an evident effect of time on the proportion of patients with SDC measured (although the numbers in the group of patients without measurement of SDC are small), leading to less favourable treatment outcome for these patients. Finally, for this analysis, we only included patients treated with the combination of doxycycline and hydroxychloroquine. Therefore, it is uncertain whether these results are generalizable to patients treated with doxycycline monotherapy or doxycycline combined with other antibiotics. Altogether, SDC measurement seems beneficial during treatment for chronic Q fever. Therefore, we recommend measuring of SDC during treatment of chronic Q fever with doxycycline and hydroxychloroquine. We advise aiming for SDC >5 μg/mL because clinicians strived for SDC values >5 μg/mL in this study, and this strategy was successful. No literature is available on upper target levels during treatment of chronic Q fever. From clinical experience, we advise clinicians to strive for values >5 μg/mL but <10 μg/mL. The recommendation on the upper target range is not evidence-based: there is no literature on toxic SDC values during prolonged treatment with doxycycline. In conclusion, treatment based on SDC was associated with a lower hazard for disease-related events, but not for all-cause mortality or PCR-positivity. Intensity of patient care could not entirely explain the association we found, suggesting that SDC-based dosing itself decreases the risk for disease-related events in these patients, potentially through more optimal dosing or through improved compliance. We therefore recommend measurement of SDC and to strive for SDC >5 μg/mL and <10 μg/mL during treatment of chronic Q fever. Acknowledgements A part of the content of this manuscript was presented as a poster during the ECCMID congress in Vienna (April 2017) and as a poster during the ESCCAR congress in Marseille (P2-13, June 2017). No other acknowledgements. List of participating hospitals Amphia Hospital in Breda, Atrium Medical Center in Heerlen, Bernhoven Hospital in Uden, Bravis Hospital in Roosendaal, Canisius-Wilhelmina Hospital in Nijmegen, Catharina Hospital in Eindhoven, Diakonessenhuis in Utrecht, Elkerliek Hospital in Helmond, Erasmus Medical Center in Rotterdam, Hospital Gelderse Vallei in Ede, Gelre Hospital in Apeldoorn, Groene Hart Hospital in Gouda, Jeroen Bosch Hospital in ’s–Hertogenbosch, Leids University Medical Center in Leiden, Izore Laboratory in Leeuwarden, Isala Clinic in Zwolle, Laurentius Hospital in Roermond, Maasstad Hospital in Rotterdam, Máxima Medical Center in Eindhoven, Meander Medical Center in Amersfoort, Medisch Spectrum Twente in Enschede, Onze Lieve Vrouwe Gasthuis in Amsterdam, Radboud University Medical Center in Nijmegen, Reinier de Graaf Hospital in Delft, Rijnstate Hospital in Arnhem, St. Elisabeth Hospital in Tilburg, St. Antonius Hospital in Nieuwegein and University Medical Center Utrecht in Utrecht.  The following hospitals provided cooperation but without chronic Q fever patients: Admiraal de Ruyter Hospital in Goes, Albert Schweitzer Hospital in Dordrecht, Medical Center Haaglanden Bronovo in The Hague, Diaconessenhuis in Leiden, St. Franciscus Gasthuis in Rotterdam, St. Jansdal Hospital in Harderwijk and Vlietland Hospital in Schiedam. Funding This work was supported by foundation Q-support (grant number UMCU150401-00), granted to the research team of Sonja van Roeden, Marieke de Regt, Andy Hoepelman and Jan Jelrik Oosterheert. Transparency declarations S. E. van R., A. I. M. H. and J. J. O. received funding from Q-support and Institut Mérieux for other research purposes. The organization of C. P. B.-R. received funding from Q-support for other research purposes. Besides the reported received funding, S. E. van R., M. J. A. de R., C. P. B.-R., A. I. M. H. and J. J. O. have nothing to declare. L. M. K., P. C. W. and A. V. W. have nothing to declare. Author contributions S. E. van R.: data collection, data management, data analysis and interpretation, making figures, writing of manuscript; C. P. B.-R.: data interpretation, writing manuscript; M. J. A. de R.: epidemiological advice, data analysis and interpretation, writing manuscript; A. V. W.: data interpretation, writing manuscript; L. M. K.: data collection; A. I. M. H.: data interpretation, writing manuscript; P. C. W.: data interpretation, writing manuscript; J. J. O.: supervision during data collection, data management and analysis, data interpretation, writing of manuscript. References 1 Eldin C, Melenotte C, Mediannikov O et al.   From Q fever to Coxiella burnetii: a paradigm change. Clin Microbiol Rev  2017; 30: 115– 90. Google Scholar CrossRef Search ADS PubMed  2 Maurin M, Raoult D. Q fever. Clin Microbiol Rev  1999; 12: 518– 53. Google Scholar PubMed  3 Raoult D, Houpikian P, Tissot Dupont H et al.   Treatment of Q fever endocarditis: comparison of 2 regimens containing doxycycline and ofloxacin or hydroxychloroquine. Arch Intern Med  1999; 159: 167– 73. Google Scholar CrossRef Search ADS PubMed  4 Million M, Raoult D. Recent advances in the study of Q fever epidemiology, diagnosis and management. J Infect  2015; 71: S2– 9. Google Scholar CrossRef Search ADS PubMed  5 Anderson A, Bijlmer H, Fournier PE et al.   Diagnosis and management of Q fever—United States 2013: recommendations from CDC and the Q fever working group. MMWR Recomm Rep  2013; 62: 1– 30. Google Scholar PubMed  6 Samuel JE, Hendrix LR. Laboratory maintenance of Coxiella burnetii. Curr Protoc Microbiol  2009; Chapter 6: Unit 6C.1. 7 Vincent GA, Graves SR, Robson JM et al.   Isolation of Coxiella burnetii from serum of patients with acute Q fever. J Microbiol Methods  2015; 119: 74– 8. Google Scholar CrossRef Search ADS PubMed  8 Omsland A, Hackstadt T, Heinzen RA. Bringing culture to the uncultured: Coxiella burnetii and lessons for obligate intracellular bacterial pathogens. PLoS Pathog  2013; 9: e1003540. Google Scholar CrossRef Search ADS PubMed  9 Rolain JM, Mallet MN, Raoult D. Correlation between serum doxycycline concentrations and serologic evolution in patients with Coxiella burnetii endocarditis. J Infect Dis  2003; 188: 1322– 5. Google Scholar CrossRef Search ADS PubMed  10 Lecaillet A, Mallet MN, Raoult D et al.   Therapeutic impact of the correlation of doxycycline serum concentrations and the decline of phase I antibodies in Q fever endocarditis. J Antimicrob Chemother  2009; 63: 771– 4. Google Scholar CrossRef Search ADS PubMed  11 Rolain JM, Boulos A, Mallet MN et al.   Correlation between ratio of serum doxycycline concentration to MIC and rapid decline of antibody levels during treatment of Q fever endocarditis. Antimicrob Agents Chemother  2005; 49: 2673– 6. Google Scholar CrossRef Search ADS PubMed  12 Rouli L, Rolain JM, El Filali A et al.   Genome sequence of Coxiella burnetii 109, a doxycycline-resistant clinical isolate. J Bacteriol  2012; 194: 6939. Google Scholar CrossRef Search ADS PubMed  13 Vranakis I, De Bock PJ, Papadioti A et al.   Quantitative proteome profiling of C. burnetii under tetracycline stress conditions. PLoS One  2012; 7: e33599. Google Scholar CrossRef Search ADS PubMed  14 Rolain JM, Lambert F, Raoult D. Activity of telithromycin against thirteen new isolates of C. burnetii including three resistant to doxycycline. Ann N Y Acad Sci  2005; 1063: 252– 6. Google Scholar CrossRef Search ADS PubMed  15 Wegdam-Blans MC, Kampschreur LM, Delsing CE et al.   Chronic Q fever: review of the literature and a proposal of new diagnostic criteria. J Infect  2012; 64: 247– 59. Google Scholar CrossRef Search ADS PubMed  16 van Roeden SE, Wever PC, Kampschreur LM et al.   Chronic Q fever-related complications and mortality: data from a nationwide cohort. In: ECCMID 2017 eLibrary Material of the 27th European Congress of Clinical Microbiology and Infectious Diseases, Vienna, Austria, 2017. Abstract OSO906. European Society of Clinical Microbiology and Infectious Diseases, Basel, Switzerland. https://www.escmid.org/research_projects/eccmid/past_eccmids/. 17 The Comprehensive R Archive Network. ‘cmprsk’ and ‘survival’. https://cran.r-project.org. © The Author(s) 2018. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please email: journals.permissions@oup.com.

Journal

Journal of Antimicrobial ChemotherapyOxford University Press

Published: Apr 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve Freelancer

DeepDyve Pro

Price
FREE
$49/month

$360/year
Save searches from
Google Scholar,
PubMed
Create lists to
organize your research
Export lists, citations
Read DeepDyve articles
Abstract access only
Unlimited access to over
18 million full-text articles
Print
20 pages/month
PDF Discount
20% off