Effectiveness of psychoeducation and emotion recognition training in women with postpartum depression

The negative effects of Postpartum Depression (PD) are found in the health of the mother and the infant, as well as in the relationship between the dyad. The negative factors in the mother-infant relationship seem to be important mediators of the consequences of PD in child development. Thus, for an intervention to be considered effective, it should bring benefits to both the mother's health and to the quality of the dyad relationship. Objective: To verify the effectiveness of psychoeducation and emotion recognition training in the quality of mother-infant interaction in mothers with PD. Psychoeducation aims to address typical developmental issues, while training emphasizes the processing of maternal faces. Method: 14 mothers with PD participated in a case-control study with three comparison groups: psychoeducation, speech emotion training, and a waiting list. Pre and post-intervention evaluations were performed. Results: Positive effects of an increase in accuracy in the recognition of adult happy faces in mothers who participated in all groups were found, with a decrease in accuracy of anger and sad faces. Participation in the Psychoeducation group presented a positive effect on the maternal dimensions of proximity and relaxation evaluated in the mother-infant relationship. Conclusions: psychoeducation had an improvement in the quality of the mother-infant relationship when compared to the other groups. Further studies with larger sample sizes are needed in order to confirm the effects found.

infant interaction has been prominent among researchers on this subject in recent decades.
Studies point out that the implications of PD on mother-infant interaction occur in the first months of a child's life, considering the higher incidence of depressive conditions during this period (Campbell, Cohn, & Meyers, 1995;Kettunen, Koistinen, & Hintikka, 2014;O'hara & McCabe, 2013).
When comparing depressed to non-depressed mothers, the interaction of depressed mothers with their babies is characterized by less time spent looking, touching, and talking with their child, and the mothers have more negative expressions, less responsiveness, less spontaneity, and lower rates of activity with the infant (Cohn, Campbell, Matias, & Hopkins, 1990;Field, Healy, Goldstein, & Guthertz, 1990;Stewart & Vigod, 2019). In this way, depressed mothers often fail to respond appropriately to their infant, relieve their child's stress and are less effective at responding to their children's needs (Murray & Cooper, 1997). The quality of this mother-infant relationship is strongly related to the trajectory of child development (Milgrom & Holt, 2014), marked by deficits in emotional, social, cognitive and physical development (Arteche & Murray, 2011;Tiffany Field, 2010;Goodman, Guarino, & Prager, 2013;Lynne Murray, Halligan, Goodyer, & Herbert, 2010).
In one of the studies on PD conducted with Brazilian subjects, a survey of 165 mothers and their babies in a city in the state of Rio Grande do Sul correlated changes in risk indexes to the development of the infant and the presence of PD. The results indicate a positive correlation, mothers with higher depression scores were more at risk to negatively affect their infants' development (Carlesso, Souza, & Moraes, 2014). In another Brazilian study, over 90% of mothers reported finding it important to stimulate their infants with practices such as reading books, talking to and holding their children on their laps. However, just slightly over 60% of them actually did engage in these practices, and the depressed mothers did significantly less than the non-depressed mothers (Campos & Rodrigues, 2015).
Thus, interventions for PD should consider not only the remission of maternal symptomatology but also the effects on the mother-infant relationship (Letourneau et al., 2011). In a systematic review published in 2013, only 14.8% of treatments aimed to improve the quality of the mother-infant relationship and 18.5% were integrative, focusing on both relationship quality and the reduction of maternal symptoms (Olhaberry et al., 2013). Although there are treatments, PD is often underdiagnosed and undertreated (Stewart & Vigod, 2019).
The World Health Organization (2013) suggests that prevention, diagnosis, and treatment for PD should be carried out in an integrated manner in primary care to the family. A meta-analysis by Stephens et al. (2016) found positive results in decreasing depressive symptoms in interventions by primary care providers. However, these interventions were not intended to improve the quality of the mother-infant relationship (Stephens et al., 2016). In Brazil, there is a lack of procedures designed to achieve this.
Adding to this lack of effective interventions are the difficulties encountered by depressed mothers in the postpartum period to seek and adhere to appropriate treatment. Among these barriers we highlight displacement issues, the need for a caregiver for the child while the mother receives care, the cost of treatment, the stigma of seeking treatment for psychological and psychiatric issues, as well as the lack of information, both to recognize difficulties and to find suitable places to seek help (Ko, Farr, Dietz, & Robbins, 2012;O'Hara & McCabe, 2013).

Objectives
This study aims to verify the effects of a psychoeducational intervention and expression recognition training on the quality of the mother-infant relationship. These two interventions were compared with mothers who were assigned to the waiting list group. The hypothesis of the study is that mothers who receive facial expression training and psychoeducation will improve the quality of their mother-infant relationship, and the mothers who complete the facial expression training will have an even greater improvement than mothers in the psychoeducation group. Additionally, improvements in maternal mood are expected in both the psychoeducation group and the expression recognition training group.

Methods
This study is a case-control study with three comparison groups: psychoeducation, expression recognition training, and a waiting list. Pre and post-intervention evaluations were performed.

Sample
A total of 14 dyads of women with infants between four and twelve weeks of age participated in the study and completed the post-intervention assessment. Five dyads participated in the psychoeducation group, three in the expression recognition training group and six on the waiting list. Inclusion criteria consisted of the following: a) mothers 18 years of age or older, literate and diagnosed with PD (EPDS ≥ 11 and current depressive episode in SCID), b) full-term infants with no health problems. Exclusion criteria included: a) mothers with HIV / AIDS, drug addiction, schizophrenia, postpartum psychosis, and mental retardation; b) twin babies, babies with congenital malformation, genetic syndrome or visual impairment. The use of medication during pregnancy and during the execution of the interventions was investigated, and one mother in each group was using Fluoxetine. accessed in person by research assistants during a period of 9 months in 2017. The purpose of the study was explained to the mothers and a contact telephone number were requested for those interested in participation. Using this basic information the study excluded underage mothers and infants who fit any of the exclusion criteria for the survey.
The other inclusion and exclusion criteria were verified in the pre-intervention evaluation, as well as possible health problems of the infants.
Between four and twelve weeks postpartum, telephone contact was made with all mothers, resuming the objectives and procedures of collection, in addition to performing the EPDS application. At least two contact attempts were made and a message was sent.
The mothers who did not respond to any of these attempts were excluded.
The application of EPDS was done by telephone and mothers who scored ≥ 11 were invited to participate in the study. At this time, the application of the pre-intervention evaluation was scheduled, allowing the mother to opt to do it at university or at home.
This evaluation occurred mostly at the mothers' homes (73.33%) and the others were performed at the university. Among the three groups, 60% of psychoeducation, 75% of training and 83.34% of the waiting list performed the assessment and intervention at home. In the initial evaluation, the Informed Consent Form was signed and the subjects were randomized into one of the three interventions through the block randomization software. Participants were not informed about their intervention group. Both pre and post-intervention evaluations were applied by an evaluator blinded to which group the mother had been randomized. The order of application of the instruments began with the sociodemographic questionnaire and, if the infant was awake, the mother-infant interaction was filmed first. After the post-intervention evaluation, mothers who continued to have depressive symptoms or showed interest in continuing psychological follow-up were referred for care. Seven mothers were referred for psychological care and five were referred for psychiatric care. Figure 1 shows the data collection diagram to date. Of these mothers, one did not complete the intervention due to the infant's death; one moved to another city; one preassessment video was erroneous and the mother was partially excluded because the dyad interaction could not be evaluated, and only the outcomes in the recognition of facial expressions were evaluated, and eight dropped out between the beginning of the sessions and the post-intervention evaluation. bathing; problems such as crying, acid reflux, irritability, cramps, restless sleep, excessive sleep, poor sleep; and changes in family configuration, in maternal thoughts and feelings through booklets about these themes. All sessions were attended by the mother and infant and were conducted by a psychologist and an undergraduate researcher to assist in the care of the infant, if necessary. The intervention was performed by female professionals, with previous training and who did not participate in the assessment of the mothers.

Emotion Recognition Training
The emotion recognition training targeted maternal face processing. A total of four meetings were held with this group, one meeting per week lasting 60 minutes. All sessions were attended by the mother and the infant and were conducted by a psychologist and an undergraduate researcher to assist in the care of the infant, if necessary. The training was performed by female professionals, with previous training who did not participate in the assessment of the mothers.

Waiting List
The waitlist group functioned as a control group format. The participants in the waiting list group were free to seek treatment according to their personal needs, but the mothers only sought follow-up after the post-intervention evaluation with the indication of the team. After the pre-intervention evaluation was completed, telephone contact was made with the mothers of this group, informing them that they will receive a reevaluation five weeks after the evaluation. In the case of maintenance of depressive symptoms or identification of other difficulties, a referral was made according to the participant's need.

Evaluation Instruments
These instruments were used in both pre and post-intervention evaluation, except for the demographics questionnaire.
Sociodemographic Questionnaire: questionnaire developed specifically for this research. Used only during the pre-intervention evaluation.
Edinburgh Postpartum Depression Scale (EPDS) (Cox, Holden, & Sagovsky, 1987;Santos, Martins, & Pasquali, 1999): consists of 10 items on a 3-point Likert scale for depressive symptoms frequently observed in the postpartum period. The scale has good internal consistency indices (Cronbach's alpha α = .87). The EPDS was originally adapted for Brazil by Santos et al. (1999) (Santos et al., 1999) who from the evaluation of 69 women proposed a cutoff of ≥ 11, with 84% sensitivity and 82% specificity. More recently, using a sample from the city of Pelotas-RS, 378 women were evaluated in the third postpartum month, and the results indicated ≥ 11 as the best cutoff point for PD screening, with 82.6% sensitivity and 65,4% specificity (Santos, 2007). This research used as a cutoff point a score equal to or greater than 11 for screening for PD. Accuracy and intensity scores follow the pattern of infant images, the first being averaged from correct assignments (one point) and the second by scores assigned on a five-point Likert scale. In this study, the emotions of happiness, fear, anger and sadness were presented by the relevance described in the literature for PD.
Global Rating Scale for Mother-infant Interaction -GRS (Murray, Fiori-Cowley, Hooper, & Cooper, 1996): Mother-infant interaction was assessed from direct observation of play between mother and infant through the GRS instrument designed to assess the quality of mother-infant interaction. The instrument has been used to assess the quality of the mother-infant relationship in mothers with postpartum depression and has been shown to have good validity in discriminating interaction (Murray et al., 1996). In addition, the GRS has cross-cultural validation, with studies in Europe, Africa and South America. Mothers and babies were filmed interacting for five minutes at their residence.
The infant was positioned on a child basket, crib, mother's lap or sofa. The interaction was filmed freely, the mother positioned her child as she preferred, as long as it was possible to make eye contact with him. The mother was invited to interact with her child and could play and talk to the infant, as she pleased, with or without the aid of objects or toys, for 5 minutes. The camcorder was positioned to capture the entire face of the infant and its mother. Interactions were assessed every minute on a 5-point Likert scale in the To assess the effect of training on the quality of interaction, the average score given by evaluator one and two in each of the dimensions within the five minutes evaluated was recorded. A repeated measures analysis was performed, considering the dimensions of the mother-infant interaction as a dependent variable and the mother group as an independent variable and covariate for the improvement in the intensity of depressive symptoms by EPDS. For all analyses, the significance value of p ≤ 0.05 was used.

Maternal Symptomatology
There was a significant difference in maternal symptoms of depression assessed  shows the assigned intensities.  Regarding the dyad dimensions, the pairs of the three groups presented averages close to or below three, indicating relatively low levels of interactivity. There were no significant time or group * time effects in this dimension.

Discussion
This study aimed to analyze the effectiveness of two interventions for women with PD in the quality of the mother-infant relationship. The main hypothesis of the study was that expression recognition training would increase maternal accuracy in facial expression recognition and, consequently, there would be an increase in the quality of the mother-infant relationship. Mothers in the psychoeducation group would not have improved facial expression recognition, but to a lesser extent have improved the quality of their mother-infant relationship. Both intervention groups would have benefits compared to the waiting list. Although this study provides preliminary and limited data, the results suggest the effectiveness of interventions developed in different components.
There was a high percentage of mothers not adhering to participate in the study after applying the EPDS by telephone (50%). This is an important fact that corroborates the difficulties found in the literature regarding the search for care and adherence to it by mothers with PD (Bauer, Ofner, Pottenger, Carroll, & Downs, 2017;Grote et al., 2014), given that it is a project that proposed to follow the mothers through home visits.
Interventions that can be carried out by the primary health care team, as recommended by the WHO (2013) All three groups improved in the maternal symptoms of PD by EPDS, indicating a tendency to decrease in symptom intensity over time, ie a spontaneous remission. This result may have been influenced to the extent that the three groups all received some form of follow-up, including the mothers in the waiting list, with pre and post evaluation. In these meetings, the mothers were welcomed, had space to talk about their difficulties and were diagnosed with PD, normalizing the difficulties encountered by them. In addition, health services provided through home visits, as were performed in this study, are related to an increase in the perception of social support (Milani et al., 2017). Lack of social support is considered an important risk factor for PD and the decrease in this factor may be associated with decreased symptoms of depression. Regarding the effects of interventions on the assessed outcomes, preliminary data indicated that the maternal accuracy for babies' faces was quite high and this effect may have occurred because the babies' faces used in this study were of high emotional intensity, facilitating their identification. Another hypothesis, corroborated by the study by Gil et al. (2011) (Gil, Teissèdre, Chambres, & Droit-Volet, 2011), is that the recognition of babies' facial expressions is mainly influenced by anxiety symptoms.
Mothers with higher postpartum anxiety had a more negative perception of child emotional expressions, but this was not maintained with symptoms of depression.
Therefore, in future studies, it is suggested to include lower-intensity stimuli and also the evaluation of maternal anxiety symptom effects.
Regarding the faces of adults, the emotions that presented the highest accuracy were anger and sadness, in contrast to happiness with the lowest accuracy. Although the purpose of this study was not to investigate the effects of PD on the ability to recognize facial expressions, previous studies indicate that women with PD have lower accuracy rates on adult faces of happiness (Flanagan, White, & Carter, 2011) and this result was maintained in the present study. Difficulty in recognizing happy faces may indicate patterns linked to depressive symptoms, suggesting that this bias occurs through a selective process that maximizes negative stimuli and filters or minimizes positive ones (Gotlib, Krasnoperova, Neubauer Yue, & Joormann, 2004;Joormann & Gotlib, 2007;Stein et al., 2010). Even with this pattern, all groups had improved recognition of happy faces after the intervention, even the waiting list group.
In addition, other studies conducted with adults diagnosed with depression indicate that they have a higher reactivity to negative emotions and lower reactivity to positive facial emotions (Gollan, Hoxha, Getch, Sankin, & Michon, 2013), requiring higher emotional intensity stimuli on happy face recognition training to improve accuracy. In our study, the accuracy of anger recognition decreased at 1000ms in all groups, suggesting that there is a modification at a more conscious level. In the same direction, all groups decreased the accuracy of sadness at 200ms. This may be associated with improved depression symptoms and, therefore, a lower reactivity to negative emotions.
Regarding the increase in the quality of the mother-infant relationship, in mothers who participated in psychoeducation, there was an increase in the maternal dimensions of approximation and relaxed. Possibly participation in a more directive intervention, providing information about the normal development of the infant, may have facilitated maternal behaviors directed towards the infant. Other maternal dimensions that did not obtain positive results may have a relevant cultural learning factor, requiring interventions focused on these behaviors to make a significant difference. Role-play training where a professional can serve as a role model and giving feedback on maternal behavior can assist in this process.
Testing the main hypothesis of this study, although it found a positive effect in improving the accuracy of recognition of facial expressions of happiness in mothers who participated in all groups, no improvement in the quality of their mother-infant relationship was identified in the training group as expected. The main hypothesis for this result is that the emotions of sadness, considered in the literature as one of the most impacted by the bias of PD, did not have positive results by training. In addition, as the training group didn't receive a directive intervention, this may help to maintain less satisfactory maternal relationship patterns.

Conclusions
PD has a high prevalence in Brazil and there a significant amount of evidence of its negative consequences on mother and infant health, as well as impaired interaction (Moehler, Brunner, Wiebel, Reck, & Resch, 2006). In addition, there is a gap in the preparation of Brazilian professionals for both proper diagnosis and interventions in this population, making this a public health issue. Among the community sample of 15 mothers participating in this study, none of them were identified with a diagnosis of PD by the healthcare system and only one was seeking psychological treatment. However, they were all clinically depressed and all had difficulties in their moods and their relationships with their infants.
The main limitation of this study is the small sample size, making it impossible to perform more sophisticated analyses and for the results to have a larger effect size. This limitation made it impossible to perform correlational studies between the recognition of facial expressions and the quality of the mother-infant relationship. To confirm the patterns of the results found, the collection continues in progress, seeking an increase in the sample size. Correct identification of the effects of interventions on the accuracy of facial expression recognition will allow adaptation to the procedures used. In addition, this study performed only a single post-intervention assessment not long after the intervention or after four weeks on the waiting list, and further follow-up was required to assess the maintenance of the changes found. To the extent that one of the negative effects of PD occurs on the child's development, follow-up is needed to assess the actual impact on the trajectory of these children.
There is a need to create interventions that invest in the quality of the motherinfant relationship, adapted to the Brazilian reality and that can be applied through home visits, increasing their adherence. Future studies are needed to determine how much these interventions can prevent negative effects on mother, infant and dyad health, and followup in the long term after the intervention.
The authors report that there is no conflict of interest.