Current guidelines (American Psychiatric Association [APA] 2006; National Institute for Clinical Excellence [NICE] 2004) recommend family therapy as a first line treatment for children and adolescents with anorexia nervosa (AN). Such therapy has been found to be more effective than individual psychotherapy, especially when the history of the disease is less than 3 years (Abbate Daga et al. 2011). Positive outcomes have also been found for other forms of family involvement, including short cycles of psychotherapy (Lock and Le Grange 2015), multiple family therapy (Depestele et al. 2015), and integrated approaches (Onnis et al. 2012; Zanna et al. 2017). It has been documented that parents’ involvement in their child’s treatment for an eating disorder (ED) supports the child’s emotional functioning and strengthens his or her ability to manage the ED and implement changes in the home environment during the treatment process (Pasold et al. 2010). For this reason, it has become increasingly important to deepen knowledge on the family functioning of young patients with AN and to implement new tools for evaluating their family relationships, in order to highlight the role of dysfunctional interactions in maintaining the disorder and to promote family resources (Le Grange et al. 2010).

AN usually develops during adolescence (Swanson et al. 2011), which is a developmental period in which parenting skills are already severely tested; in this period, parents must face new challenges, change parenting strategies, and renegotiate their coparenting relationship. However, although the literature on family functioning is very wide, only few studies have analyzed the family dynamics of patients with AN using observational tools, focusing specifically on the ways in which the mother–father dyad works together, coordinating and supporting parental roles. Many studies using self-report evaluations of typical characteristics in the families of AN patients have shown that such families are perceived to have lower general functioning and increased dissatisfaction compared to families in control groups (Casper and Troiani 2001; Cook-Darzens et al. 2005; McDemott et al. 2002). Such studies have not interpreted these family characteristics to pre-date the child’s AN, but they have instead considered them a consequence of the AN—a potentially fatal disease—in the family system (Eisler 2005). Identifying the ways in which families reorganize themselves around AN may help clinicians understand potential areas of intervention to restore healthy family functioning.

Studies that have focused on specific characteristics of family functioning have reported different and often conflicting results (Holtom-Viesel and Allan 2014). Some studies have shown differences in family functioning according to different subtypes of AN. Specifically, families with a child in the restrictive subtype of AN have been found to share a healthy perception of family functioning and to be satisfied with the family environment (Casper 1981; Garfinkel and Garner 1983; Kog and Vandereycken 1989; North et al. 1995). In these families, parents have been found to present their family as cohesive and with little conflict, and to enjoy satisfying relationships with their daughter (Garner et al. 1985). Patients have further perceived their family as flexible, cohesive, and communicative, particularly with their mother (Vidovic et al. 2005). In addition, it has been noted that girls with restrictive AN assess the family contribution of their mother better than the mother does, herself (Casper and Troiani 2001; Laghi et al. 2017). In contrast, patients with binge-purge AN have been found to indicate lower levels of cohesion and emotional involvement in the family and a more hostile, conflictual, and disorganized family environment (Baiocco et al. 2012; Casper and Troiani 2001; Strober and Humphrey 1987). However, some studies have found no significant differences between these AN subtypes (Dancyger et al. 2005; Vidovic et al. 2005).

Few studies have investigated relations in the families of AN patients using observational instruments. In one example, Minuchin et al. (1978) videotaped family tasks and identified four primary patterns of dysfunctional interaction in the families of patients with psychosomatic pathologies, including AN: (1) enmeshment, which refers to an extreme form of proximity and intensity in family transactions, whereby the boundaries between parents and children are weak and there is no separation between members; (2) rigidity, or the tendency to maintain the status quo and to not modify rules and transactional models according to family evolution; (3) over-protectiveness, characterized by a high degree of concern and mutual interest for each other’s well-being and hypersensitivity to any signs of discomfort, often accompanied by an important fear towards the outside world; and (4) avoidance of conflict, which is the tendency to leave problems unresolved because conflict is perceived as a threat to family unity. According to the authors, none of these characteristics can independently elicit or maintain psychosomatic symptoms, but all four patterns are family processes that encourage somatization. In this relational framework, patients’ symptoms get new meaning as regulators of a family system and, more specifically, the parental conflict in which the patient is involved. The authors also distinguished three types of families’ dysfunctional involvement with the patient: (1) detouring, whereby apparently united parents cover up their conflicts, assuming a protective position towards the sick child or blaming him or her; (2) triangulation, in which the patient is called by parents to side against the other; and (3) binding, in which the child is permanently united with one parent against the other. According to the authors, these models are not family classifications but methods that families (including functional families) use to deal with conflict. In the families of children with AN, however, such interactive patterns are prevalent and the children are frequently involved, taking the role of moderator in the conflict.

Similarly, Wallin and Hansson (1999) evidenced that families of AN patients show lower family competence and higher enmeshment, hierarchical organization, and rigidity with respect to control, when evaluated by independent observers during a videotaped problem solving task. In the study, families were evaluated using a set of one self-report measure and two rating scales—the Clinical Rating Scale-Turbo (CRS-Turbo; Cederblad et al. 1987) and the Beavers Family Competence and Family Style (Hansson 1989). The results highlighted that clinicians noted more dysfunctional patterns in the families than the family members did, themselves. This more positive perception of parents and daughters with regard to family interactions compared to that of an external clinical observer may derive from the families’ tendency to deny and avoid conflict, which could affect their judgment and lead them to idealize family relationships (Casper and Troiani 2001). For these reasons, Holtom-Viesel and Allan (2014) suggested that researchers studying the families of patients with AN should not only use self-report tools, but also use observational instruments, in order to collect external perceptions of family relationships and the implementation of these relationships, within the context of a structured task.

However, the coding systems used by Wallin and Hansson (1999) refer to global family functioning and do not consider subsystems or the contributions of individual family members to the ongoing interaction. The observational procedure proposed by Fivaz-Depeursinge and Corboz-Warnery (1999), named the Lausanne Trilogue Play (LTP), on the contrary, allows clinicians to assign a score to each individual, each subsystem, and the entire family. The LTP is a semi-structured task designed to assess the triadic competences of family members, in order to identify specific patterns of family interactions from the time the child is in the womb to the time the child is in adolescence. According to this framework, the family alliance is defined as the degree to which family members coordinate themselves in a shared play task (Mazzoni et al. 2018).

The LTP is divided into four parts: in the first part, one parent is asked to play with the child while the other parent is simply present. Next, the parents exchange roles. In the third part, parents play together with their child. Finally, in the last part, they engage with each other while the child plays alone. Prior to beginning the LTP, an expert clinician explains the rules to all family members. The play session is recorded and later observed by two independent clinicians who, using a coding system, categorize families into cooperative, stressed, collusive, and disordered family alliances. The LTP has been used in clinical and research contexts and has shown good psychometric characteristics (Carneiro et al. 2006; Lubrano Lavadera et al. 2007). Several studies have used the LTP to observe pre-natal and childhood triadic interactions, identifying specific ways in which parents and children coordinate actions (Lubrano Lavadera et al. 2016; Mazzoni et al. 2018; Simonelli et al. 2014) and demonstrating that these configurations remain stable over time (Favez et al. 2013). Relatively few studies have used the LTP with children in adolescence, examining the relationship between the quality of family interactive dynamics and the adolescent child’s psycho-behavioral problems (Gatta et al. 2016; Gatta et al. 2017).

Coparenting is closely linked to the triadic vision of family functioning and contributes to determining its quality. It can be defined as the coordination and mutual support between parents responsible for the care and raising of a child (McHale 2007b). McHale (2007a) identified four basic dimensions in the coparenting relationship, all of which are influenced by the quality of the marital relationship: (1) solidarity, which concerns cooperation and mutual support between the parents; (2) antagonism, which describes the presence of differences between the parents that require negotiation; and (3) the division of child care and commitment, or the level of involvement of each parent with the child. It has been demonstrated that a positive coparenting alliance influences children’s adjustment and serves as a protective factor: it fosters competence in the parental role, helps parents feel less distressed and more able to handle challenges, protects against abuse, and improves healthy development in infants and toddlers (Florsheim et al. 2003; McHale 2007b; Teubert and Pinquart 2010).

Most of the literature on coparenting has focused on the early phase of family life, involving infants and children. However, coparenting is a dynamic construct. Evidence supporting the importance of coparenting in families with younger children does not necessarily generalize to families with adolescent offspring (Feinberg et al. 2007). In fact, the characteristics of coparenting are likely to differ in families with adolescents. For example, while coparenting young children requires a high level of moment-to-moment cooperation, such teamwork is less important in coparenting pre-adolescents, who are developing independence from their parents (Margolin et al. 2001).

Coparenting disagreements disrupt parents’ ability to provide a caregiving environment—which is fundamental for supporting adolescents to develop autonomy—and undermines adolescent adjustment in crucial stage-salient tasks (Martin et al. 2017). Riina and Mchale (2014) identified two signifying coparenting dimensions in adolescence: (1) shared decision-making, relating to the adolescent child’s daily experiences, and (2) joint involvement, or time spent in shared activities with the adolescent child. They found a negative correlation between shared decision-making and youth risky behavior. In the same way, parents’ joint involvement was found to be negatively correlated with both depressive symptoms and risky behavior. These findings highlight the importance of family coordination and coparenting assessment in the diagnostic procedure of adolescent AN patients, in order to gain a better understanding of the complex nature of the patient’s family functioning and to develop an appropriate treatment plan targeting the specific aspects that the patient and his or her family needs to improve.

The present research was an exploratory study aiming at generating preliminary results about the use of the LTP for evaluating family and parenting coordination in a sample of adolescent AN outpatients. Previous research has found that, relative to the families of patients with internalizing disorders, the families of patients with AN have greater difficulty respecting roles during play, maintaining joint attention, and sharing positive affects, especially in the third phase of the LTP, in which the mother, father, and child all play together (Balottin et al. 2017). However, further research is needed to investigate the specific ways in which the parental couple cooperates and influences family coordination and the maintenance of the child’s food symptoms.

The study aimed at exploring the role of the family’s ability to repair mis-coordinations, the family’s flexibility, and the parents’ ability to cooperate and share affects. These characteristics may provide a crucial framework for planning interventions and improving outcomes for children and adolescents with AN. Additionally, the study sought to explore: (a) levels of participation, organization, focalization, affective contact, and family coordination in families with a child diagnosed with AN, in the four parts of the LTP; (b) the types of family alliances and coparenting styles in families with a child diagnosed with AN; and (c) whether family coordination in families with a child diagnosed with AN was significantly related to coparenting style and food symptoms, in terms of BMI. It was expected that these families would demonstrate low family coordination and difficulty maintaining roles and sharing affect, yet high participation and focalization in the play task. It was also expected that there would be a prevalence of dysfunctional alliances and coparenting styles in these families, with low levels of explicit competition between parents, but low warmth shown between partners and less open cooperation with the partner’s choices. Finally, a significant relationship between family coordination, coparenting style, and food symptoms was expected.

Method

Participants

The sample consisted of 24 patients with a primary diagnosis of restrictive AN and their families. All participants were observed in an outpatient setting at a children’s hospital that specialized in the treatment of eating disorders using multifocal integrated treatment (MIT; Zanna et al. 2017). The selection of patients was based on the following inclusion criteria: (a) a primary diagnosis of restrictive anorexia nervosa; (b) no mental delay or other serious organic disease; and (c) age between 12 and 18 years. Divorced parents were included, while one-parent families were excluded. The Italian version of the Schedule for Affective Disorders and Schizophrenia for School-Age Children/Present and Lifetime Version (K–SADS–PL; Kaufman et al. 1997) was used to determine Axis I psychiatric diagnoses.

Within the sample, 21% of the parental couples were divorced and 79% were still united. Both parents had, on average, a medium-high level of education. Parents’ median age was 48 years for mothers (SD = 4.07, range 40–54) and 51 years for fathers (SD = 4.76, range 44–60). Patients were almost entirely female (22 female vs. 2 males), with a median age of 15 years (SD = 1.16, range 13–17) and median BMI of 15.9 (SD = 1.67, range 12–18).

Procedure

Families were recruited during the first day of MIT. All families received information on the aim of the research from the psychiatrist responsible for the project and agreed to participate in the research project and the videorecorded play session. No family refused to participate. The entire family was observed throughout the LTP procedure: every session was videotaped in a separate room and later coded by two independent judges, who had completed specific training. The judges obtained a two-way mixed ICC value of uniformity of 0.87 (excellent) for the LTP encodings and 0.98 (excellent) for the coparenting evaluation.

Measures

Lausanne Trilogue Play

The Lausanne Trilogue Play (LTP; Fivaz-Depeursinge and Corboz-Warnery 1999; Malagoli Togliatti and Mazzoni 2006) is a semi-structured procedure used to observe family interactions. As widely explained in a recent work (Mazzoni et al. 2018), during LTP, the family is asked to sit around a table and play together with Duplo®—or, when the child is in adolescence, to write a story about a weekend without the parents. In the present study, the adolescent participants were asked to write a story and the parents were instructed to support and guide their child during the task. In the first two parts, parents took turns interacting with the adolescent while the other one observed; in the third part, both parents were active with their child and helped him or her write the story. Finally, in the fourth part, the adolescent wrote the story alone, while the parents had their own discussion. The parents’ task was to guide their adolescent child in the writing task, encouraging him or her in moments of impasse and promoting his or her autonomy. The adolescent completed the task by following his or her parents’ encouragement but also making choices independently.

Two independent and reliable judges, who were blind to the goals of the research, observed the videotaped task and assigned scores ranging from 0 (not appropriate) to 2 (appropriate) to each member of the family in each part of the task. The entire task sequence lasted 15–20 min, but parents could decide how much time they spent together in each part. Time spent in each part and for the whole task was also recorded. The LTP is evaluated for four functional levels: participation, organization, focalization, and affective contact. The functional level of participation assesses the extent to which each member is included and self-included within the play. This function is constant across all four parts of the task, and assessed accordingly. Organization assesses whether each member maintains his or her assigned role in the four parts of the task. In each part, it is expected that the parents will be supportive and guide the adolescent, maintaining a clear but flexible boundary (Minuchin 1974; Fivaz-Depeursinge and Philipp 2014). The functional level of focalization assesses the degree to which everyone pays attention to the task, regardless of their role; it is consequently evaluated across all parts. Affective contact assesses whether each family member appears relaxed and playful throughout the procedure, as well as whether they encourage and appreciate the other members and contribute to the overall experience. The evaluation of this function is also present across all parts. Summing up these scores, an overall score of global family functioning, ranging from 0 to 40, can be calculated.

A validation study showed a cut-off score of 23 to differentiate between functional and dysfunctional alliances (Mazzoni and Lubrano Lavadera 2013). Specifically, the LTP classifies families into four types of alliances. First, in cooperative (A) alliances, the family plays together as a team in a cooperative and happy atmosphere. The parts follow one another naturally and fluidly and, in the case of interactive errors, the parents are able to repair and re-establish a positive affective climate. Second, in stressed (B) alliances, father, mother, and child play together but meet obstacles in their interaction that create “stresses” in the otherwise enjoyable affective climate. Despite the ups and downs, the parents are able to re-establish a collaborative environment. Third, in collusive (C) alliances, families fail to achieve the goal of shared play and fun. Parents cannot provide help and support to the child because their interactions are characterized by hidden or overt competition. The families may fail to complete all parts in the task and transitions might be abrupt or confusing. The affective climate is constantly in tension, with no repair. And finally, in disordered (D) alliances, the four parts of the task are confused and overlapping, creating constant tension and ambiguity. Sometimes family members are outright excluded from the triad; roles are not well defined and are subject to continuous interference. Play progresses in a chaotic fashion or, on the contrary, there is a rigid adherence to the four parts. So, the affective climate is clearly negative, even though there may be an apparent “pseudo-positivity” (Fivaz-Depeursinge and Corboz-Warnery 1999).

Coparenting and Family Rating System

Coparenting was evaluated using the Coparenting and Family Rating System—Toddler age and above (CFRS; McHale et al. 2001; McHale et al. 2004) during the LTP procedure. The CFRS is a coding system that assesses coparenting style through an analysis of parents’ interactive patterns, focusing on the degree of cooperation and competition in the parental couple, investment and warmth towards the child and between parents, child centerness (i.e., the centrality of parental guidance), and the presence of verbal sparrings between parents. Scores range from 1 (absence) to 5 (elevated presence) for all scales, aside from Investment and Warmth Towards the Child, which range from 1 to 7. The coding is based on video observations of the family interaction during the third part of the LTP by two independent and reliable judges, who are blind to the goals of the research. The judges assign scores to each scale and then determine the overall coparenting style. The CFRS was originally developed for the parents of preschool-age children, but it has also been used with families of school-age children (McConnell and Kerig 2002). It has never been used with families with adolescents.

The CFRS identifies four coparenting styles: cohesive, excluding, competitive, and child-at-center. The cohesive style is characterized by balanced parenting involvement, with high levels of warmth and cooperation and low levels of competition. These families work better, also in relation to the quality of the marital relationship and each parent–child relationship. The child-at-the-center style is characterized by excessive involvement in the child by both parents and a strong disconnection towards the partner. In the competitive coparenting style, parents are equally involved with their child but in competition with each other and lacking in cohesion. Finally, in the excluded style, parents are disconnected from each other and there is a large discrepancy in each partner’s level of involvement with the child. The authors also indicated a shifting coparenting style, in which a shift from a functional to a dysfunctional style is demonstrated in the same part.

Data Analysis

Due to the clinical nature of the sample, a continuous normal distribution could not be assumed. Thus, non-parametric statistical analyses were used. The distributions of categorical measures were analyzed using chi-square tests. Freidman’s chi-square and Wilcoxon signed-rank tests were used to identify any differences in function levels for the four phases of the LTP. Finally, correlations were investigated using Spearman’s rho.

Results

Functional Levels, Family Coordination, and Types of Alliances

Means and standard deviations for the functional levels are reported in Table 1. Significant differences were found in the use of functions (Friedman’s chi-square = 52.188; p < 0.001). Follow-up pairwise comparisons using a Wilcoxon signed-rank test showed that, for different functions, participation was the highest category (Wilcoxon signed-rank test with organization = −4.219, p < 0.001; focalization = −4.224, p < 0.001; affective contact = −4.249, p < 0.001). There were significant differences between organization and focalization (Wilcoxon signed-rank test = 3.346, p < 0.001) but not between organization and affective contact (Wilcoxon signed-rank test = 0.351, p = 0.726). Focalization also differed significantly with affective contact (Wilcoxon signed-rank test = -2.423, p < 0.05).

Table 1 Mean scores of group for LTP functions and phases
Full size table

The coding system enabled an additional categorical measure to be evaluated, differentiating between high (24–40) and low (0–23) global family coordination. Families were classified as demonstrating low family coordination in 16 cases (66.7%) and high family coordination in 8 cases (33.3%). In particular, there was a significant presence of collusive alliances (C; scores 16–23; n = 13; 54.2%), followed by stressed alliances (B; scores 24–31; n = 8; 33.3%), and disordered alliances (D; scores 0–15; n = 3; 12.5%). There were no functional alliances (scores 32–40).

Quality of Family Interactions in the Four LTP Phases

Freidman’s chi-square test was run on the various phases of the LTP (mother/child + father; father/child + mother; three together; mother + father + child). Seven families did not participate in the third part of the game, and one family did not participate in the fourth part; for this reason, all eight families were excluded from the following analysis. Means and standard deviations for the LTP phases are reported in Table 1. Significant differences were found for the different phases (Friedman’s chi-square = 10.277; p < 0.05). Follow-up pairwise comparisons using a Wilcoxon signed-rank test showed that the first two parts of the dyadic play showed no significant differences (Wilcoxon signed-rank test = −0.833, p > 0.05); the third part significantly differed from the first (Wilcoxon signed-rank test = 2.222, p < 0.05); and the fourth part scored significantly lower than the first (Wilcoxon signed-rank test = 2.457, p < 0.05) and the second part of the game (Wilcoxon signed-rank test = 1.989, p < 0.05). Families’ ability to play the third part of the LTP was found to be positively associated with BMI (rho = 0.558, p < 0.01).

Coparenting Relationship and Global Family Coordination

Using the CFRS, scores were assigned to families on the basis of observations of the third part of the LTP, in which both parents were asked to interact together with the adolescent. In this phase, coparental interactions were most evident because parents had to coordinate their interventions. Consequently, the seven families that did not complete this phase were excluded from the analysis. The coparenting coding showed the following coparenting styles: child-at-center (n = 7; 41.2%) and excluding (n = 6; 35.3%) were most prevalent, followed by competitive (n = 3; 17.6%) and shifting (n = 1; 5.9%). Additionally, there was a marginally significant correlation between coparenting and global family coordination, as evaluated by the LTP (rho = -0.364, p = 0.080).

Discussion

The literature underlines that family interactions play an important role in the maintenance of AN symptoms in children and adolescents and the successful treatment of AN in these persons (Le Grange et al. 2010). Nevertheless, few studies have analyzed the family dynamics of patients with AN using observational tools and focusing on the specific ways in which the mother–father dyad works together, coordinating and supporting their parental roles. The present research was an exploratory study on co-parenting and family coordination in the families of adolescents with AN, using a semi-structured observational task to evaluate family interactions.

The first objective was to analyze functioning in the families of adolescents with AN, in terms of family coordination and functioning. The results of the present study showed high participation and focalization and low organization and affective contact in these families. In other words, the families tended to demonstrate strong physical and psychological participation but poor definition of roles and low emotional sharing. This finding is aligned with the results of previous research with families of patients with AN that has found that these families demonstrate great difficulty in respecting roles during play, sharing positive affects, and maintaining joint attention, especially in the third phase of the LTP (Balottin et al. 2017).

The families in the present sample also showed low coordination, particularly in the third and fourth phases of the LTP. In these phases, parents are meant to work together to achieve the goals of shared play and then start a dialog without the mediation of the child. Difficulty was particularly evident in the seven families that did not participate in the third phase of the LTP and the single family that did not participate in the fourth phase.

Several studies on clinical populations have found similar results, and it is likely that these findings are more common in families of clinical populations, as well as families suffering from divorce or high conflict (Carneiro et al. 2006; Favez et al. 2012; Fivaz-Depeursinge and Philipp 2014; McHale 2007a). In the LTP, families test themselves on their ability to manage boundaries; it is plausible that, in situations of conflict or child pathology, family members are more stressed and therefore less inclined to cooperate and respect roles and rules (Le Grange et al. 2010). This is a two-way process: from the parents’ side, they may not avert attention from their child; from the child’s perspective, he or she may intervene in a protective manner in parental interactions perceived as at risk (of conflict) (Lubrano Lavadera et al. 2011).

Regarding the second objective, the study found a significant presence of collusive alliances in the sample. Malagoli Togliatti and Mazzoni (2006) suggested that, in collusive alliances, parents divert their unexpressed conflict to their child, competing to capture his or her attention or to maintain a leadership role, through covert or overt action. Consequently, competitiveness between parents may create a constant and unrepaired tension, often manifested through apparent serenity. In line with these data, the study also found a prevalence of a child-at-center coparenting style, followed by an excluding coparenting style. Both of these patterns suggest a tendency of the parents to show low levels of not only overt competition and verbal sparring, but also cooperation and warmth. In the child-at-center coparenting style, parents appear as largely passive with regard to offering guidance and support to the child, establishing a role reversal dynamic (Kerig 2005). In contrast, the excluding coparenting style indicates a pattern in which one parent shows a greater involvement in the child than the other. These patterns support clinical observations that have shown that families of patients with AN are characterized by the child’s involvement in conflictual parental dynamics, parental leadership struggles, and parental over-control (Minuchin et al. 1978; Selvini Palazzoli 1981).

The present study was the first to capture competition using the LTP coding system, rather than the CFRS. A hypothesis was that families of patients with AN would tend to show formal agreement that would be contradicted by verbal and non-verbal communication, indicating the absence of real cooperation. The CFRS calculates overt competition events (i.e., the number of times that parents take parallel and different initiatives, with risks of overlapping). However, the LTP rating scale also includes covert competitive behavior, such as depreciation, sarcasm, and attempts to hinder or interrupt interactions between the child and the partner, including through body language. For this reason, the LTP is likely to be a more accurate observational tool for this kind of family dynamic.

Finally, the third goal of the study was to identify a relationship between global family functioning, coparenting styles, and food symptoms, in terms of BMI. A marginally significant correlation was found between coparenting and global family functioning, in terms of family coordination. Several studies have found that coparenting influences—albeit indirectly—family functioning, with marital satisfaction as a mediator. For example, a negative correlation has been found between marital conflict and cooperative coparenting behaviors, while marital conflict has been found to be positively related to triangulation and conflict coparenting (Margolin et al. 2001). Furthermore, Katz and Gottman (1996) found several significant associations between dyadic marital and triadic family levels of interaction. The present study found that, in triadic interactions, marital hostility was linked to higher levels of intrusiveness and lower levels of positive involvement for fathers, and higher levels of derisive or sarcastic humor for mothers. Further studies should investigate this spillover phenomenon in families with adolescents with AN, exploring the role of marital satisfaction in cogenitorial functioning and the entire family system.

A positive association was found between family coordination and BMI, in line with recent work using the LTP in the psychopathological assessment of children and adolescents (Gatta et al. 2016; Gatta et al. 2017). In particular, the severity of the clinical presentation of the adolescent AN seems to have been connected to the adolescent’s specific difficulties in triadic, rather than dyadic, interactions (Balottin et al. 2018). Moreover, adolescents’ better functioning in the triadic phase, at first assessment, has been found to be associated with a better outcome, as well as the parents’ ability to positively interact within the triadic relationship (Balottin et al. 2018). These findings underline the importance of evaluating the role of the coparenting relationship in maintaining eating pathology, separate to the parent–child relationship and global family functioning. Moreover, adolescents may play an active role in family and couple dynamics, and administration of the LTP upon admittance would be helpful to evaluate the adolescent’s individual functioning within the different family relational configurations.

The limited sample of the present study and the absence of a control group represent important limitations of this preliminary research. A larger sample would have enabled the use of more sophisticated statistical analyses, including investigations of the distribution of interactive family patterns within different eating disorder typologies in childhood and adolescence. It also would have enabled the analyses to differentiate the sample according to age and severity of eating symptoms and to identify specific patterns of family functioning.