Description of included trials
We identified 5,708 records from the literature searches and an additional ten records were identified from reference lists and contact to author. A detailed flow diagram of the study selection process is shown in Additional file 3. In total, we included 17 trials in the review. We have provided full details of the included trials in the ‘Characteristics of included trials’ table [see Additional file 5]. A list of excluded trials with brief explanation of reasons is reported in Additional file 8.
Some trials were reported in more than one report. The 17 trials were reported in 21 papers and 1 oral presentation. Only papers fulfilling the requirements for analysis are included. The trial by Maimburg et al. was reported in two papers and one oral presentation [12-14], and the trial by Werner was reported in three papers [15-17]. For the remainder of the review, only the main report for each included trial is cited.
Results from the included trials were reported between 1988 and 2014 in 20 papers and 1 oral presentation (obtained from the author). Six trials were conducted in the United States [18-23], four trials were conducted in Australia [24-27], two were conducted in Denmark [12,16], one in the United Kingdom [28], one in Canada [29], one in Finland [30], one in Mexico [31], and one multicenter trial was conducted in Spain and France [32]. In total, the trials included 6,507 randomized women and 961 men, with a range from 74 to 1,193 participants per trial.
All trials tested the effect of antenatal education in small classes; however, the content and form of the experimental condition varied between the trials. The amount of education in the experimental condition varied from a single 1-h session [24] to 24 sessions each lasting 2.5 h [22]. Some trials focused on prevention of a specific condition among participants at specific risk, e.g., women at high risk of postnatal depression [21,28,31] or women with low socio-economic status [19,32]. Other interventions were targeted at a broader population group, e.g., all primipara at a specific birth site [12]. Also, control conditions differed between trials. In most of the trials, the control group was offered standard care which varied by content and amount, e.g., individual consultations with a midwife that also the experimental condition was offered [12,31]. In four trials, the control group was offered other interventions other than antenatal classes, e.g., one-to-one contact with a medical doctor [19].
Two trials were directed towards expecting fathers [23,26], and three trials specifically addressed the couple as a unit [18,22,32]. The remainders of the trials were directed towards the pregnant women, but in some of them, the expecting fathers were welcome to join one or all sessions.
For three of the pre-specified outcomes, maternal sense of control/active decision-making during labor and birth, partner involvement at birth, and infant care abilities, no data were reported. Data on pain relief during labor, obstetric interventions, knowledge acquisition, breast feeding, social support, relationship quality and divorce/separation, and psychological and social adjustment to parenthood were reported. Within the overarching category of psychological and social adjustment to parenthood, the following outcomes have been reported: antenatal and postnatal depression, anxiety, readiness for delivery and child care, self-efficacy and locus of control, co-parenting, and parent-child interaction.
Risk of bias in included trials
We assessed the risk of bias in the 17 included trials. Full details on the risk of bias scoring can be found in the ‘risk of bias tables’ , ‘risk of bias summary’ , and ‘risk of bias graph’ [see Additional file 4]. All trials except for two [27,30] reported self-reported outcomes, and two trials additionally reported objective outcomes [12,16]. Blinding of participants was only possible in one trial [32].
All trials were scored overall ‘high risk of bias’ for the self-reported outcomes. For the objective outcomes, two trials were scored ‘overall moderate risk of bias’ [12,16]. These two trials were scored ‘overall high risk of bias’ for the self-reported outcomes since participants were not blinded. Also, the trial by Ickovics et al. was scored ‘overall high risk of bias’ for the same reason although this trial had ‘low risk of bias’ in all other domains but reported no objective outcomes [20].
Effects of interventions
Depression prevention classes versus standard care
Three trials compared a depression-preventive program in small classes with standard care [21,28,31]. Brugha et al. examined the effect of a depression prevention antenatal program for women at risk of depression and found no significant effect on depression measured with several different measurement tools, self-efficacy, or locus of control [28]. A trial conducted by Lara et al. examined effects of a psycho-educational antenatal program among women at high risk of depression and reported no effect on depressive symptoms 6 weeks postnatally [31]. Also, Le et al. reported no effect of a psycho-educational antenatal program among women at high risk of depression - neither in pregnancy nor 6 weeks postnatally [21]. All three trials were scored ‘overall high risk of bias’.
Psycho-social prevention program versus brochure on child care
One trial assessed the effect of a psycho-social prevention program for couples, compared to a brochure on child care delivered to participants in the control condition [18] on depressive symptoms, co-parenting, anxiety, and parent-child interaction for both mothers and fathers 6 months postnatally. They reported that fathers, but not mothers, in the experimental group experienced significantly higher co-parental support (MD 0.29, 0.05 to 0.53), parenting-based closeness (MD 0.35, 0.04 to 0.66), and significantly lower father-child dysfunctional interaction (MD−0.26, −0.43 to −0.09) compared to fathers in the control condition [18]. This trial was scored ‘overall high risk of bias’.
Psycho-educational classes versus letter on fear of childbirth
One trial by Rouhe et al. compared the effect a group-based psycho-educational intervention directed towards women with severe fear of childbirth to written information in the form of a letter addressing fear of childbirth delivered to the participants in the control condition [30]. They found that the intervention significantly increased the likelihood of spontaneous vaginal delivery (RR 1.33, 1.11 to 1.61). They reported no effect on the use of epidural analgesia, overall caesarean section, elective and emergency caesarean section, vacuum extraction, and induction of labor [30]. This trial was scored ‘overall high risk of bias’.
Program using a psycho-somatic approach versus standard antenatal education program
Ortiz Collado et al. examined the effect of an antenatal psychosomatic program designed to decrease depression among women at high risk of postnatal depression compared to standard care [32]. They reported no significant effect on depression, social support, or relationship satisfaction among women. They also assessed relationship satisfaction among men and reported no significant effect [32]. This trial was scored ‘overall high risk of bias’.
Couple-focused classes versus standard care
One trial by Schulz et al. assessed the effect of a couple-focused intervention compared to standard care on marital satisfaction among both mothers and fathers 6 months and 5.5 years postnatally as well as divorce/separation 5.5 years postnatally. They reported no significant intervention effects on any of these outcomes [22]. This trial was scored ‘overall high risk of bias’.
Self-hypnosis classes versus standard care
Werner et al. compared a self-hypnosis intervention with standard care and reported no effect on the outcomes: use of epidural analgesia as pain relief during labor, spontaneous delivery, overall caesarean section, elective caesarean section, vacuum extraction, oxytocin augmentation, induction of labor, and any breast feeding 4 months postnatally [16]. However, they reported a statistically significant increased risk of emergency caesarean section (RR 1.52, 1.02 to 2.27) in the experimental group [16]. For the outcomes related to delivery, this trial was scored ‘overall moderate risk of bias’ , while the score was ‘overall high risk of bias’ for breast feeding which was self-reported.
General antenatal education classes versus standard care
One trial by Maimburg et al. assessed the effect of general group-based antenatal training among primiparous compared to standard care on a range of both pharmacological and non-pharmacological pain relief outcomes, obstetric interventions, postnatal depression, breast feeding, breast feeding knowledge, and breast feeding self-efficacy [12]. They reported a protective effect on the use of epidural analgesia (RR 0.84, 0.73 to 0.98) but no significant effect on any other kind of pain relief or obstetric interventions, e.g., caesarean section and vacuum extraction. Also, no significant effects were reported on breast feeding at 5 weeks or 6 months postnatally and breast feeding self-efficacy or postnatal depression 6 weeks after birth. They reported a higher proportion with sufficient knowledge about breast feeding 6 weeks postnatally among women attending the general antenatal training program in small classes (RR 1.08, 1.01 to 1.15) [12]. For the outcomes related to delivery, this trial was scored ‘overall moderate risk of bias’ , while the score was ‘overall high risk of bias’ for breast feeding, breast feeding self-efficacy, knowledge, and postnatal depression which were self-reported.
Group prenatal care (20 h) versus individual prenatal care (2 h)
A trial by Ickovics et al. examined the effect of a general antenatal education program in small classes compared to individual prenatal care (total amount of time: 2 h) [20]. They reported significantly higher scores on prenatal and infant care knowledge (MD 2.60, 1.68 to 3.52) and readiness for labor and delivery (MD 7.60, 3.34 to 11.86) at 35-weeks gestation among women in the experimental condition. They found no effect on readiness for infant care or prenatal distress at 35-weeks gestation [20]. This trial was scored ‘overall high risk of bias’ due to the self-report of outcomes.
Paternal education class versus standard care
Two trials examined the effect of paternal education compared to standard care [23,26]. Westney et al. conducted an intervention targeted at prospective adolescent fathers. This intervention had a significantly positive effect on paternal knowledge acquisition in relation to pregnancy, delivery, infant care, and support towards the mother (MD 9.55, 1.25 to 17.85) [23]. Maycock et al. conducted a breast feeding intervention targeted at expecting fathers. They reported a significant intervention effect on any breast feeding 6 weeks postnatally (RR 1.09, 1.00 to 1.18). There was no effect on exclusive breast feeding 6 weeks postnatally [26]. Both of these trials were scored ‘overall high risk of bias’.
Extra breast feeding sessions versus standard care
In three trials, the authors examined the effect of giving extra breast feeding sessions in small classes [24,25,29]. Duffy et al. examined the effect of an antenatal group-teaching session aimed at increasing breast feeding prevalence but also reported obstetric interventions. They reported no effect on vaginal delivery, caesarean section, vacuum extraction, or forceps. They also assessed the effect on breast feeding and reported a positive effect on exclusive breast feeding 6 weeks postnatally (RR 3.20, 1.88 to 5.46) [24]. Noel-Weiss et al. examined effects of a breast feeding education workshop and reported no significant effect on breast feeding 8 weeks postnatally. However, they found a significantly higher breast feeding self-efficacy among participants in the experimental condition 4 weeks postnatally (MD 4.60, 0.72 to 8.48) but not 8 weeks postnatally [29]. Forster et al. conducted a trial comparing two breast feeding education classes with usual care. They reported no significant effect in initiation of breast feeding or breast feeding 6 months postnatally [25]. All three trials were scored ‘overall high risk of bias’.
Breast feeding classes versus one-to-one contact on breast feeding
Kistin et al. assessed the effect of a breast feeding class with group discussion compared to 15- to 30-min one-to-one contact with a medical doctor on breast feeding topics and reported no effect on initiation of breast feeding or of any breast feeding 12 weeks postnatally [19]. This trial was scored ‘overall high risk of bias’.
Breast feeding classes versus breast feeding and childbirth pamphlets
One trial assessed the effect of a breast feeding education program compared to breast feeding and childbirth pamphlets [27]. Rossiter reported a significantly higher rate of breast feeding initiation (RR 1.86, 1.35 to 2.55) among participant in the experimental condition but found no effect on breast feeding 6 months postnatally [27]. This trial was scored ‘overall high risk of bias’.