- Martine Wallon, associate professor ()a,
- Christiane Liou, librarianb,
- Paul Garner, senior lecturerc,
- François Peyron, professora
- aService de Parasitologie, Hôpital de la Croix-Rousse, 69004 Lyons, France
- bService de Parasitologie, Faculté de Médecine, Université Claude Bernard, Lyons, France
- cCochrane Infectious Diseases Group, International Health Division, Liverpool School of Tropical Medicine, Liverpool
- Correspondence to: Dr Wallon
- Accepted 24 February 1999
Objective: To summarise the evidence that treating toxoplasmosis in pregnancy reduces the risk of congenital toxoplasma infection and improves infant outcomes.
Design: Systematic review of studies comparing at least two concurrent groups of pregnant women with proved or likely acute toxoplasma infection in which treatments were compared with no treatment and outcomes in the children were reported.
Subjects: Studies were identified from Medline (1966-97), Pascal (1990-7), Embase (1993-7), and Biological abstracts (1993-5) plus contact with experts in the field, including the European Research Network on Congenital Toxoplasmosis.
Main outcome measure: Proportion of infected children at 1 year born to infected pregnant women who were or were not treated.
Results: Out of 2591 papers identified, nine met the inclusion criteria. There were no randomised comparisons, and control groups were generally not directly comparable with the treatment groups. Congenital infection was common in treated groups. five studies showed that treatment was effective and four that it was not.
Conclusion: It is unclear whether antenatal treatment in women with presumed toxoplasmosis reduces congenital transmission of Toxoplasma gondii. Screening is expensive, so the effects of treatment and impact of screening programmes need to be evaluated. In countries where screening or treatment is not routine, these technologies should not be introduced outside carefully controlled trials.
Pregnant women in France and Austria are routinely screened for toxoplasmosis, and women negative for antibodies are followed up at regular intervals
The value of antenatal toxoplasmosis screening programmes depends on safe treatments that reduce the risk of congenital disease
This systematic review found no good comparative data measuring the potential harms and benefits of antiparasitic drugs used for presumed antenatal toxoplasma infection
Most control groups were not comparable, and incidence of congenital infection was high in the intervention groups
Countries considering introducing screening should do so only in the context of a controlled trial
Toxoplasmosis infection during pregnancy can cause congenital infection and manifest as mental retardation and blindness in the infant.1 Doctors prescribe spiramycin and sulphonamide for presumed infection to reduce mother to child transmission and the severity of fetal infection. These practices arose from reports of antiparasitic effects in vitro, 2 3 in AIDS patients,4 and in pregnancy. 5 6
In 1978 and 1985 Austria and France implemented nationwide programmes to detect and immediately treat all toxoplasma infections during pregnancy. Women of unknown immune status are tested during the first trimester of pregnancy. French seronegative women are advised on good hygiene to avoid infection and retested monthly to detect seroconversion. Women with evidence of acute infection are given spiramycin, amniocentesis, and ultrasound examination. If the fetus is infected the women are given sulphonamides and pyrimethamine, which is usually withheld in the absence of fetal infection because of potential teratogenicity and bone marrow toxicity for the mother and fetus.7 Parents can also opt for termination if there is evidence of fetal macroscopic lesions. 8 9 In France, an estimated 44% of pregnant women are regularly checked for seroconversion10 and between 5625 and 8850 women are treated during pregnancy every year to prevent congenital toxoplasmosis.
Other countries have decided against routine repeated screening in serologically negative women during pregnancy. In the United States, experts judged that such a programme was not warranted because of the low frequency of maternal infection and low chance of infection in the newborn.11 A UK working group of experts concluded in 1991 that “screening for acute toxoplasmosis in pregnancy should not be offered routinely.”12
Opponents to systematic screening also point out the need for improved diagnostic tests—even since the development of polymerase chain reaction tests13—and the issue of cost effectiveness. 14 15 Detection of infection will have no effect unless the treatments given as a result of the screening actually reduce congenital infection and improve infant outcomes. We conducted a systematic review of the effects on the fetus and infant of treating women who seroconvert during pregnancy.16
We included studies of pregnant women with toxoplasma infection, defined by an increase in specific IgG titres from paired sera or by a high titre of specific IgG at the first antenatal test. Studies based on specific IgM screening were excluded. Women could have been tested as part of a formal screening programme or through incidental testing carried out by general practitioners suspecting toxoplasmosis infection. All the included studies had to compare at least two groups of pregnant women, one of which received no antibiotic treatment. The drug, dose, and duration in the treatment group did not affect inclusion. Studies comparing treatments were excluded.
Congenital infection was defined as persisting specific IgG at age 1 year. Clinical infection was defined by the presence of: hydrocephalus, ventricle dilatation, intracranial calcifications, or chorioretinitis. Children with no clinical signs were considered disease free if seronegative at 1 year. Clinically disease free children without a negative test result were considered lost to follow up. Cases of abortion, stillbirth, or infant death with no evidence of toxoplasmosis infection were also considered as lost to follow up.
We performed an electronic search using the key words “congenital and toxoplasmosis” in four databases: Medline (1966-07/1997), Embase (1993-07/1997), Pascal (French) (1990-1997), and Biological abstracts (1993-1995). References of the papers identified were scanned for additional relevant studies. Members of the European Research Network on Congenital Toxoplasmosis and other experts were asked for relevant published or unpublished data.
The abstract and title of each of the 2591 papers thus identified were checked by two people independently and all papers dealing with animal models, biological aspects of the disease, congenitally infected children without data on pregnancy, and isolated case reports were excluded. We obtained hard copies of the studies that potentially met the inclusion criteria.
Data extraction and analysis
Papers that potentially met the inclusion criteria were scrutinised a second time by two people.Information was extracted by using a data extraction sheet, and this included the entry criteria, the source of the controls, and whether the authors stratified by stage of pregnancy when the infection occurred. We assessed the quality of studies using a scale ranging from 0 to 6, based on six equally weighted items: randomised group allocation, recruitment at the same location for both groups, recruitment during the same period for both groups, analysis based on the intention to treat, inclusion in the analysis of patients lost to follow up, and if cases lost to follow up were included, that they constituted less than 10 % of the total sample and were adequately taken into account. Each item was scored 1 for “yes” and 0 for “no” or “unknown.”
Criteria for maternal infection were divided into true seroconversion and infections that were suspected because of significant increase in IgG titres but could not be proved because the first available sample was already positive for IgG. Cases lost to follow up were included successively as non-infected and infected, and low and high transmission rates were estimated. Statistical analyses were reviewed based on the published figures (Epi-info 6.2,17).
Description of studies
We identified 305 studies to consider for inclusion, but only nine met our predetermined criteria (table). The paper by Desmonts and Couvreur in 198421 contained data previously published in 1974, so the earlier paper was excluded. Eight studies were published,18–25 and one was available only as an abstract.26 Reports were from France (four), Belgium (two), Germany (one), and Austria (one), with one French-Danish collaboration. In seven reports, women were participating in a routine screening programme. In the other two reports, subjects were recruited through screening or individual testing by their doctor. The number of included subjects ranged from 1119 to 689.26 None of the studies included randomised comparisons between women.
Five studies reported significantly lower transmission rates in children born to treated mothers (P<0.01).18–22 Maternal treatment was with spiramycin alone in four studies. Douche et al added sulphonamides to spiramycin after fetal infection was diagnosed antenatally.18 Suspected but unproved maternal infections were included in three studies.20–22 The date of maternal infection was taken into account in only one study.20 Trimester specific analysis indicates that the reduction in transmission was mainly attributable to treatment efficacy in women who seroconverted during the third trimester of pregnancy. Children were followed for at least 12 months in three studies,18–20 10 months in one21 and 6 months in one.22
Four studies reported no reduction of fetal infection in the treated groups.23–26 Sulphonamides or pyrimethamine, or both, were added to spiramycin in three studies 23 24 26 and used alone in the fourth.25 Unproved seroconversions were included in one study.25 None of the studies took into account the date of maternal infection. Only two reports mentioned the source of the controls. In one, the control group was a mixture of women diagnosed with toxoplasmosis who did not take the recommended treatment and women whose “obstetrician lacked conviction.”23 In the second, a Danish cohort of untreated women with proved seroconversion was used as controls.26 Seroconversion was not tested for until delivery, which explains the absence of antenatal treatment. Because the date of maternal infection was not known in the Danish group, there is no evidence that the controls were comparable with the intervention group. The length of postnatal follow up was at least 1 year in all four studies. In two studies the authors concluded that the treatment worked, 23 25 although our analysis of the available data did not show any difference.
The quality score was 2 in two studies, 20 24 3 in two studies, 21 22 and 4 in five, 18 19 23 25 26 ;no studies scored 6 (table). Three of the five studies suggesting a beneficial effect of treatment scored less than 4,20–22 whereas only one of the four studies that found no effect scored less than four.24
Ideally, policies for managing women who develop toxoplasmosis in pregnancy should be based on good research. Studies should randomly allocate women who seroconvert in pregnancy to an intervention or control group, should be large enough to give sufficient power, and should follow up children to a point when congenital toxoplasmosis can be excluded. None of the 2591 papers published over the past 30 years fulfilled all these criteria.
Despite our broad inclusion criteria, only a few studies included controls. The controls that were used were often not directly comparable with the treatment group: some controls were women who failed to comply with the antibiotic instructions; women who seroconverted in late pregnancy; or women from populations unrelated to those in the intervention group. In other studies, it was not clear where the controls came from. Date of seroconversion is likely to influence the effects of the infection and of treatments, but this was taken into account in only one paper.20 Only five studies were based on pregnant women with proved seroconversion. None of these studies gave details of the delay between infection and onset of treatment, and details of the treatment were not always available. The outcome assessment was satisfactory in seven studies. Even in the intervention groups, treatment failures are relatively common, with congenital infection occurring in a third to a fifth of cases. Although the rate of infection was higher in the control groups, our inability to assess or otherwise quantify the lack of comparability of the control groups makes interpretation difficult.
Implications for policy
In our opinion, current evidence is insufficient to confirm that treating mothers who seroconvert during pregnancy prevents fetal infection and improves infant outcomes. We are not saying that there is no such benefit but that current research is inadequate to assess whether the putative benefits outweigh the potential harm of the drugs to the fetus.
We did not examine potential indirect benefits of screening. Serological testing early in pregnancy allows health professionals to advise seronegative women on avoiding infection. Immune women can be reassured and excluded from further testing. Identifying acute infection through repeated antenatal tests facilitates fetal diagnosis through polymerase chain reaction amplification of DNA in amniotic fluid and ultrasonography to monitor fetal development. If infection is confirmed, the parents have the options of termination of pregnancy in the case of morphological lesions or experimental, in utero treatment with sulphonamides and pyrimethamine. Treatment and surveillance can be continued immediately after birth.
Implications for research
Further studies are needed to evaluate the benefits of maternal treatment schedules. They should use standardised interventions and outcome measures and be based on proper randomisation.
Countries which do not currently perform systematic screening should conduct appropriate research before introducing screening. A large study could randomise healthcare clinics to no screening (existing practice) or screening with follow up of seronegative patients and treatment if they convert. Only then will it be possible to know whether the package of care is effective in preventing congenital toxoplasmosis.
In countries where screening is routine, assessment will be more difficult because of the conviction of health professionals and the public's belief in screening. Congenital infection remains a problem despite intervention, and it would be possible to conduct randomised trials of different treatment regimens. France and Austria have been expected by many other countries to produce evidence of the benefits of national screening programmes. Ironically, they will depend on other countries to demonstrate the cost effectiveness of their screening programmes.
This review has been submitted to Pregnancy and Childbirth collaborative review group and a longer version of it will be maintained in the Cochrane Library.
Contributors: FP and PG were responsible for the conception of the study and the original protocol draft. MW, CL, and FP were responsible for identifying the papers, extracting and synthesising the data, and writing the first draft. All authors took part in its revision and approved the final version. FP is the study guarantor.
Funding: This study was supported by a grant from the European Union Directorate General XII, and the Cochrane Infectious Diseases Group is supported by the Department for International Development (UK). The funding bodies take no responsibility for the data presented and the views expressed.
Conflict of interest: None declared.