Every resident in Denmark is provided with a unique civil registration number that enables individual-level linkage between national registries. Data from health records at obstetric and infectious disease departments were thereby linked to civil registration numbers from women (18–45 years of age) who had positive or equivocal tests at the Statens Serum Institut for antibodies to C. burnetii during 2007–2011. Using these data, we identified pregnant women who could be included in the study on the basis of positive serologic test results for C. burnetii and availability of titers from throughout pregnancy to enable evaluation of infection in paired samples.

We identified 12 women with equivocal and positive antibody titers for C. burnetii infection who underwent 19 pregnancies during the 5-year study period. All women were farmers or veterinarians and resided in rural areas of Denmark. Obstetric complications were recorded in 9 (47%) of the 19 pregnancies (Table 1). None of the women were found to be IgM positive for pathogens regarded as classic causes of infection of the developing fetus during pregnancy; these pathogens included Toxoplasma gondii, parvovirus B19, rubella virus, cytomegalovirus, and herpes simplex virus. However, not all women were exhaustively tested during pregnancy. For all women tested during pregnancy, C. burnetii serologic test results were available from no later than pregnancy week 13 (Table 2).

Two patients (1 and 10) reported dry cough and short episodes of fever; both had antibody titers during the first trimester consistent with acute infection. These patients were treated with cotrimoxazole beginning in gestational week 15 or 10, respectively; patient 1 was treated throughout pregnancy and patient 10 until gestational week 39. Patent 1 had a PCR-positive urine sample in gestational week 10, but results of PCR on a bone marrow biopsy from gestational week 15 and PCR on amniotic fluid and placenta were negative. For patient 10, no C. burnetii DNA was detected by PCR from placental tissue or breast milk; for her second pregnancy, serologic test results for C. burnetii were negative, PCR on placenta was not performed, and the pregnancy had a healthy outcome.

Patient 2 reported weeks of dry cough without fever during the weeks just before the first pregnancy that ended with miscarriage. She had 3 miscarriages and an extrauterine pregnancy within 2 years, and the titers that were found indicate that she was acutely infected weeks before the first miscarriage. Her antibody titers reached a maximum level after the second miscarriage. No embryo material was tested from any of her miscarriages.

Patients 3, 5, and 12 had serologic profiles with IgG phase I titer of 1,024 at the end of pregnancy; 2 of the patients were treated during pregnancy, but none had symptoms, received a diagnoses of endocarditis, or received long-term postpartum treatment. Patient 12 had a serologic profile indicating reactivation of C. burnetii infection, but her antibody titers had been negative in a previous pregnancy 2 years earlier.

Patient 4 seroconverted before her second pregnancy, during which she experienced a single fetal death around gestational week 8 and a surviving co-twin. She had a decrease in IgG phase I during pregnancy; the surviving twin was delivered healthy and at term.

Patient 5 was treated for acute Q fever before her second pregnancy. Her antibody titers were stable during the second pregnancy, and her baby was full-term but slightly small for gestational age. After a short interpregnancy interval, she had a significant increase in IgG phase II titers during her third pregnancy; because of fetal growth restriction and oligohydramnion noted during gestational week 28, she had a cesarean section during gestational week 38. The placentas from these 2 pregnancies were not tested.

Patient 8 had rising antibody titers during her first pregnancy, and because of bleeding and contractions, she had an acute cesarean section in week 27 and gave birth to a severely growth retarded and malformed infant who lived only a few hours. The fetus and placenta were not tested for C. burnetii, but results of testing for toxoplasmosis, cytomegalovirus, and parvovirus B19 were negative, as were results of genetic testing for neuromuscular diseases. Her titers decreased slightly postpartum, but during her second pregnancy, titers increased significantly, and treatment with cotrimoxazole was initiated around gestational week 22. In gestational week 30, she spontaneously went into labor and gave birth to a healthy baby. Treatment was terminated immediately postpartum, and her antibody levels decreased, indicating that she was not chronically infected. Thus, patients 5 and 8 had serologic indication of a reactivation of infection with C. burnetii: a postpartum decline in antibody titers followed by a >4-fold increase in titers during the next pregnancy.

Three of the remaining pregnancies (in patients 6, 9, and 11) had an uncomplicated course with a healthy pregnancy outcome. Patient 7 had an acute cesarean because of rupture of the uterus; she had had a cesarean in her first pregnancy.

In summary, 3 patients (1, 2, and 10) reported symptoms of acute Q fever. At least 1 (patient 8) appeared to have seroconverted without symptoms close to the beginning of her first pregnancy.

For 7 of the 19 pregnancies, treatment with cotrimoxazole was initiated in the patient; 6 of these pregnancies resulted in healthy, full-term babies, and no mention of severe side effects was found in the mothers’ medical records. One of the women who had obstetric complications received treatment with cotrimoxazole (patient 8, in her second pregnancy). By comparison, among the 12 pregnancies in which no treatment was given, 8 resulted in obstetric complications. The effect of treatment with cotrimoxazole on complications was tested, but the difference was not significant (p = 0.057 by Fisher exact test; data not shown).

PCR was performed on placentas from 10 pregnancies and in 4 of these, breast milk was also tested; no results were positive. For 7 of the 10 pregnancies in which placentas were tested, the woman had received treatment with cotrimoxazole during pregnancy.

Adverse pregnancy outcome was observed in 9 of 19 pregnancies among 4 of the 12 pregnant women in which equivocal or positive tests for C. burnetii antibodies were found. One woman had 3 miscarriages and an extrauterine pregnancy, 1 experienced preterm delivery, 1 had a single fetal death with a surviving co-twin, and 1 delivered a small-for-gestational-age baby. Oligohydramnion and fetal growth restriction were found in 2 pregnancies; 1 had a healthy outcome, but in the other, the baby died a few hours postpartum.

The observed complication rate of 47% may seem high, but the causal relationship of this finding may not be clear. For example, none of the tested placentas were examined by histopathology, and all of the 10 placentas tested by PCR were negative for C. burnetii. Furthermore, not all of the women were thoroughly tested for other infections. One possible explanation for the lack of findings related to the placenta in 7 of the cases could be the patient’s treatment with cotrimoxazole or focal placental infection. Nonetheless, we did observe adverse pregnancy outcome in 8 (67%) of 12 pregnancies in which the women were not treated with cotrimoxazole, which supports the beneficial effects of treatment.

Among the cases we reviewed, none of the 4 breast milk samples tested by PCR were positive for C. burnetii. This bacterium has been found in human milk (22,23), but the implications for the breastfed child are unclear. Because of the lack of evidence, breastfeeding has been deemed safe according to the obstetric guidelines in Denmark for the treatment of C. burnetii–seropositive pregnant women and their newborns.

Two studies, in the Netherlands (24) and Canada (25), have suggested a low rate of placental C. burnetii infection in asymptomatic women and that obstetric complications in symptomatic cases may be explained by massive placental necrosis following a higher bacterial load in the placenta, systemic infection, or both. A study in France (26) found that untreated Q fever in 1 pregnancy may be reactivated in a subsequent pregnancy, a result found in 3 patients in our study (patients 5, 8, and 12).

The evidence for an adverse pregnancy outcome in humans in relation to Q fever mainly originates from case studies from France of referred, infected, pregnant patients, as well as pregnancies in which a diagnosis of Q fever was reached retrospectively, after an adverse outcome (9–11). Carcopino et al. reported clinical symptoms in 32 (60.4%) of 53 cases and a chronic serologic profile in more than half of patients and concluded that Q fever in pregnancy may cause severe complications (10). The Netherlands has recently experienced an unprecedented Q fever outbreak that has prompted 2 large studies in pregnant women. One study, a population-based study of 1,174 serum samples collected at the twelfth week of pregnancy, found no association between antibodies to C. burnetii and adverse pregnancy outcome among women living in the area with the highest Q fever incidence. The other study, a randomized, controlled trial, tested 1,229 pregnant women living in high-risk areas during the outbreak; 15% of the women were seropositive in both the intervention group and the control group, and no difference was found in obstetric complications (13). Hence, the findings from France were not reproduced in the Netherlands. Likewise, a recent study in Denmark assessed the association between presence of antibodies to C. burnetii, seroconversion, and pregnancy outcome and found that seropositivity was not associated with miscarriage, preterm birth, or low birthweight (15).

Differences in findings among these various studies may be explained in part by differences in study design. The indication for serologic testing is a crucial point; in our study, the indication for testing was exposure to livestock for most of the women. In contrast, the women in the case series in France were primarily tested because of pathologic conditions during pregnancy or clinical symptoms (e.g., fever, hepatitis) or retrospectively because of an adverse pregnancy outcome. Angelakis et al. (11) found that 17 of 30 C. burnetii seropositive pregnant women were asymptomatic; only 2 of these had an uncomplicated pregnancy, but no placentitis or isolation of C. burnetii was found in 14 available biopsy specimens. These authors suggested that the different rates of obstetric complications found in various geographic areas could be related to strain specificity, potentially because of differences in plasmid types.

All the women in this case series had contact with livestock in Denmark, and it is reasonable to assume that these women were occupationally exposed to endemically occurring C. burnetii infections among cattle. However, the observations by Angelakis et al. suggest that strains of C. burnetti in Denmark, and possibly cattle in general, might be less virulent than that seen when the infection is acquired from other animal reservoirs (e.g., goats). In France, goats and sheep have been the main source of C. burnetti infection. The recent outbreak in the Netherlands was detected shortly after a large number of dairy milk farms had changed from cattle to goats as production animals. In Denmark, goat and sheep farms are rare, and despite high clinical awareness during the past 7 years, no reports have described a microbiologically verified outbreak of Q fever in humans or a case of chronic Q fever that was definitely acquired in Denmark. However, a large percentage of dairy cattle in Denmark shed C. burnetii, and a high prevalence of antibodies has been found among pregnant women who had exposure to cattle (16,17). These observations could, in part, be an explanation for the discrepancies in rates of serious adverse pregnancy outcomes among studies from different countries.

In conclusion, we evaluated risks and implications in 19 pregnancies with positive or rising titers against C. burnetii from Denmark, a country that has high seroprevalence of C. burnetii but low prevalence of clinical Q fever and for which cattle are the primary bacterial reservoir. In this case series, almost half of the women had obstetric complications, which is comparable to previous case series. We found complications in 8 out of 12 untreated pregnancies; 7 pregnant women received long-term treatment with cotrimoxazole. In this study, serologic signs of Q fever were associated with adverse pregnancy outcome. However, in none of the cases could we identify a definite causal relationship between C. burnetii seropositivity and adverse pregnancy outcome. Because only 9 cases of adverse pregnancy outcome were found over 5 years, despite increased awareness among the relevant risk groups, and because community studies in Denmark and the Netherlands have failed to confirm this association, the overall risk for a Q fever–associated adverse pregnancy outcome in Denmark is likely to be low.