There is considerable evidence that families experience negative health outcomes after a divorce (Amato, 2010, 2001; Auersperg et al., 2019). Accordingly, divorce is associated with physical health problems (Lorenz et al., 2006; Sbarra, 2015), psychological distress, and poorer mental health (Dohrenwend et al., 1978; Freeman et al., 2008; Hobson & Delunas, 2001). This includes higher levels of depression, anxiety, and somatization (Al-Krenawi & Graham, 2004; Bracke et al., 2010; Breslau et al., 2011; Hald et al., 2020b; Kendler et al., 2003; Metsä-Simola & Martikainen, 2013; Monden et al., 2015; Symoens et al., 2014; Thuen, 2000; Walid & Zaytseva, 2011). Paired with divorce rates of 35–50% in most industrialized countries (Center for Disease Control and Prevention, 2016; European Commission, 2015), divorce may be considered a public health concern (Vezzetti, 2016). Further, many societies now strive for a digital transformation, whereby digital services either complement existing face-to-face services or replace them. This also concerns services provided to couples and families where, for example, digital platforms seek to aid relationships (Øverup et al., 2022), hinder their breakdown (Doss et al., 2016), or ameliorate well-known adverse effects of divorce (Ciprić et al., 2022). Advantages of this digital transformation may include freeing up time for family counsellors or other public service providers to (1) focus on those families most in need as less challenged families may find ample help in the digital solutions offered, (2) use face-to-face time with families on actually counselling as the psycho-educational parts of the sessions may be communicated through digital learning modules, (3) reach groups of people who due to logistics, resources, or practicalities usually do not receive help, and tailor the face-to-face sessions or course of intervention more to the families. For example, in the area of divorce, one parent may need to learn more communication skills while the other may need to learn more about conflict management before meeting jointly with the counsellor. And lastly, this digital transformation may have the benefit of reaching groups of people who, due to logistics, resources, or practicalities, usually do not receive help. This may be possible via digital platforms that include diverse information/modules relevant to the family dynamics or problems in question. As such, digital platforms and tools become highly relevant for staff or systems working with families and offer new or complementary avenues of intervention (Ciprić et al., 2022).

The purpose of this study is to describe the overall design and content of the online divorce platform “Cooperation after Divorce” (CAD) and investigate its effectiveness on sick days among 778 recently divorced Danes, using a randomized control trial study (RCT). By doing so, we hope to contribute to the divorce intervention literature with a more detailed description of CAD, its theoretical basis, content, structure, and design, which is currently missing. In addition, we investigate whether CAD can effectively reduce divorcee’s number of sick days over a one-year post juridical divorce as sick days can be considered a good measure of overall health-related quality of life (Barile et al., 2013).

Intervention types

As a result of growing divorce rates and the burden it places on families and society, divorce education programs were introduced in the United States in the 1970s (Braver, Pearson, & DeLuse, 1996), and over the intervening decades, the number of divorce education programs have proliferated (Pollet & Lombreglia, 2008; Salem et al., 2013). These programs are primarily focused on improving knowledge, skills, and attitudes related to cooperative co-parenting and helping divorcees deal with the typical challenges related to divorce and supporting children during the (often difficult) transition (Geasler & Blaisure, 1999; Mulroy et al., 2013). The efficacy studies of face-to-face interventions have generally shown positive outcomes, such as increasing knowledge of the impact of divorce, improved communication skills, improved parental post-divorce adjustment, reduced inter-parental conflict, and reduced child exposure to parental conflict (Fackrell et al., 2011; Harold et al., 2016; Sigal et al., 2011). However, most of these studies face methodological shortcomings such as insufficient control groups and short follow-up periods (Fackrell et at., 2011; Greenberg et al., 2019).

Early divorce education programs were delivered face-to-face in a classroom setting, but with the spread of the internet, online divorce programs (ODPs) were introduced and continue to grow in popularity (Bowers et al., 2011, 2014; Turner et al., 2019). ODPs offer several advantages over more traditional face-to-face interventions, such as cost-effectiveness, convenience, and making receiving help less time-consuming for parents, especially for those who face logistical, geographical, or financial challenges with transportation, getting time off work, or receiving help to look after their children during teachings (Bowers et al., 2011; Turner et al., 2019). However, in contrast to face-to-face programs, few online divorce interventions have been subjected to rigorous scientific testing (Becher et al., 2018) and none to the use of RCT designs, except for the CAD intervention reviewed and used in this study.

Bowers et al. (2011) reviewed six ODPs in what seems to be the most extensive study to date, and recently a few studies have evaluated specific ODPs, such as “Parents Forever“ (Becher et al., 2018), the “Online Parenting Programs” (Bowers et al., 2014), “Utah’s Online Divorce Education Course” (Kopystynska et al., 2020; Turner et al., 2019), and “Focus on Kids” (Schramm & McCaulley, 2012). The aforementioned ODP studies have four main conclusions in common: 1) Online programs have potential because of the flexibility and scalability of the format, 2) There are strong indications of program effectiveness, but a lack of evidence of what works for whom, 3) There is a pressing need for more rigorous forms of evaluating and measuring the effectiveness of the programs, and 4) The effectiveness of the reviewed or aforementioned ODP’s interventions have never been tested using RCT setups. In regards to the latter point, Sigal et al. (2011) note: “The use of comparison groups provides a necessary tool for evaluating whether any change observed can be attributed to the program rather than other factors, such as the simple passage of time” (Sigal et al., 2011, p. 133). Further, notably lacking in the literature are detailed accounts of the development, theoretical background, functionality, implementation, and efficacy of ODP.

Efficacy and Sick Days

To our knowledge, only two studies have assessed the number of sick days among divorcees. Hald et al. (2020c) report the cross-sectional associations between burnout symptoms and the number of sick days three months before legal divorce among recently divorced Danes. Here, it was found that women reported nearly double the number of sick days than men, and that for women, being a parent was associated with an increased number of sick days. Further, Bookwala et al. (2014) compared the number of sick days over a 12-month period of older adults living in Wisconsin (USA). Although the average number of sick days increased by over two days from Time 1 to Time 2, they did not find that divorced adults reported significantly more sick days than their continuously married counterparts. However, this lack of significance may be in part because of the large variance in the outcome at the latter time point.

In order to assess the efficacy of the CAD intervention, in the current study, we compare an intervention group to a control group in terms of number of sick days over the first 12 months post juridical divorce using an RCT study design. We do this as the number of sick days provides an approximation of health-related quality of life (HRQOL) and because the number of unhealthy days has been shown to be valid in evaluating adult health and HRQOL (Jia et al., 2007; Moriaty et al., 2003). Further, HRQOL is frequently employed to assess the efficacy of interventions (Guyatt et al., 1993). Thus, this measure provides an overall estimate of the global health of participants as the reason for a sick day may be related to mental or physical health.

In a Danish context, sick days differ from actual official sick leave/absence as sick leave/absence is longer than 14 days and may last up to one year with 100% financial compensation to the employee (Krane et al., 2013). Consequently, formally, sick days in a Danish context are days in which an employee, for whatever reason, is absent from work. However, when discussing sick days in a Danish cultural context or inquiring about these, the implicit understanding is ‘too sick to go to work’ or similar. In the midst of a divorce, a “sick day” may take on additional layers of complexity. While still officially a day off from work due to illness, it may also be a much-needed break to manage the stress, emotions, and logistical challenges of ending a marriage. In Denmark, employees are not only entitled to take a paid sick day for their own physical or mental well-being but also have the provision to do so if their children are sick or experience psychological distress. This additional allowance becomes especially vital and perhaps present when juggling the practicalities and emotionality of splitting a household and co-parenting. In a time of emotional distress, this policy may become a crucial support system as the sick day may transforms into more than just an absence due to illness, namely a much-needed break to refocus mentally, attend to urgent family matters, or consult with healthcare or legal professionals as required. So, in this context, the Danish definition of a sick day becomes multi-dimensional. It’s not just a day to recover from personal illness, but also an option to ensure the well-being of children or legal and logistic matters.

Description of the Cooperation After Divorce Intervention

Background

Cooperation after Divorce (CAD) was developed in Denmark by researchers and collaborators from the University of Copenhagen. The project was based on the following four key findings from an analysis of existing divorce interventions and available information materials:

  1. (1)

    There was a need for scalable and cost-effective divorce interventions, which focused on creating real-life changes. Many interventions and information materials were available, primarily in the form of traditional face-to-face therapy, mediation, courses, books, folders, and websites. However, while the information materials were often digital and scalable, they were also very general and mostly focused on increasing knowledge. On the other hand, the interventions, which were more comprehensive, individualized, and more focused on behavioral changes, were all based on face-to-face activities. This reduced outreach potentials and flexibility, and increased costs.

  2. (2)

    The existing interventions were used by a limited group of divorcees, namely, the high-resource families, who had a recognized need, could afford to pay for help, and had the time, logistic setup, and resources to engage in such activities. Families, for whom resources were low and for whom it might be less socially acceptable to seek help and counselling, were less likely to use the interventions, which posed a dilemma since they might be the families who would benefit the most. Finally, those whose divorce is characterized by a low or moderate level of conflict are often excluded from outside help and counselling, as their need for help is often not recognized to the same extent as those who experience higher levels of divorce conflict (Hald et al., 2020d).

  3. (3)

    Most of the existing digital divorce interventions in 2012-14, when the development of the CAD intervention was initiated, were conversions of face-to-face courses to online courses. Bowers et al. (2011) conclude that this was one of the biggest challenges facing existing online interventions as they were not utilizing the potential of the digital media (i.e., not utilizing the potential of interaction, rich media, and possibilities of personalization).

  4. (4)

    Finally, the analysis confirmed the need for a stronger link between theory and praxis, the lack of scientific evaluations using a control group, and evidence for the effectiveness of the intervention.

Target Group

The primary target group of the CAD research project was Danish adults, with or without children, who recently divorced. As more than half of Danish divorces have children living at home, special consideration was given to this target group during content selection and development. Access to a computer or tablet with an Internet connection and proficiency in the Danish language were prerequisites for participation during the research period. The intervention was later (in 2021) released in English, Arabic, Icelandic, and Swedish language versions, but at the time of the RCT study, only a Danish version was available.

Theoretical Cornerstones

Children are the ones most affected by the negative consequences of divorce and parental conflicts (Amato, 2000; McIntosh & Tan, 2017; Ottesen et al., 2017). At the same time, issues around the children, such as custody, alimony, and upbringing, are often the causes of parental conflicts (Johnston, 1994; Ottesen et al., 2017). It was therefore a fundamental principle of the CAD intervention to put the needs of the children at the core of the intervention, and throughout the CAD intervention, this fundamental principle guided content inclusion. Specifically, materials were included to provide the parents with practical advice and tools on how to understand, help, and communicate with their children, how to establish good cooperation between the parents, and how to help themselves. The inclusion of this material aimed to minimize the negative impacts of divorce on children by maintaining parents’ focus on the one thing with which most divorcees agree: That their children should not pay the price for their parents’ divorce and disagreements.

Consequently, CAD is based on three theoretical cornerstones, focused around the needs of the divorcees and their children; see Fig. 1.

Fig. 1
figure 1

CAD theoretical cornerstones

Full size image
The Divorce-Stress-Adjustment Perspective (DSA)

The first cornerstone is based on the Divorce-Stress-Adjustment (DSA) perspective introduced by Paul R. Amato (Amato, 2000). This perspective generally does not view marital dissolution or breakups as a discrete event, but rather a year(s)-long process that is preceded and followed by several stressful events within the relationship and post-relationship dissolution. These stressful events increase the risk of negative emotional, behavioral, and health outcomes following divorce among both children and adults. According to the DSA, two main factors impact post-divorce adjustment: Mediators (Stressors) and Moderators (Protective Factors). Mediators represent factors that may lead to greater post-divorce distress. As an example, a divorcee who loses desired contact with his or her children, experiences a high level of conflict with the former spouse, economic decline, or loss of social status and emotional support, will be at increased risk of experiencing negative outcomes. Conversely, moderators, which are factors that may mitigate or exacerbate the severity of the (negative consequences) of a divorce, may include the divorcees’ personal resources, demographic characteristics, and perception of the divorce. That is, according to the DSA perspective, divorcees who have more personal resources, such as good coping and social skills, a high income, and strong support from friends and family, may experience less post-divorce distress, as these characteristics will serve as protective factors, which may lessen the negative impact of the divorce. This means that the negative outcomes and successful adjustment vary from person to person, depending on the differing mediating and moderating factors (Amato, 2000, 2010).

Accordingly, CAD was designed to address stressors in the divorce process, to promote protective factors, and provide knowledge and tools, which increase the probability of successful adjustment in the affected families. For example, CAD provides knowledge about the grief process and children’s reactions to their parents’ divorce and practical tools and exercises for conflict management and coping with the new identity and roles post-divorce.

Divorce as a Crisis

The second cornerstone is to view divorce as a crisis since a divorce process often includes several potentially stressful events and losses. These may include loss of custody and time with children, conflict with the former spouse, a decline in disposable income, loss of emotional support, loss of social status and identity, moving to a new home, being a sole parent, and functioning in new roles (Amato, 2000, 2010; Fabricius & Luecken, 2007; Kołodziej-Zaleska & Przybyła-Basista, 2016). Accordingly, divorce and marital dissolution have consistently been identified as one of the most stressful life events across cultures, gender, and time (Hobson et al., 1998; Holmes & Rahe, 1967; Rahe & Arthur, 1978). According to Holmes & Rahe’s Social Readjustment Rating Scale (1967), divorce is even more stressful than being imprisoned, experiencing the death of a close family member, or personal injury or illness, and is rated the second-most stressful life event overall in adulthood (Holmes & Rahe, 1967; Miller & Rahe, 1997). Exposure to stress, particularly over prolonged periods, has consistently been linked with adverse physical and mental health outcomes (Budtz-Lilly et al., 2015; Cohen et al., 1997; Danese & McEwen, 2012; Fink & Rosendal, 2015; Kessing et al., 2003; Nielsen et al., 2014; Sutin et al., 2010). CAD was therefore specifically designed to support divorcees in the process of alleviating stress and crisis to avoid long-term exposure to stress and by covering material related to coping with various emotions, such as grief and anger, and to break the tendency for rumination.

Conflict as a Primary Source of Maladjustment

The third cornerstone is a focus on conflict as a factor that greatly influences the well-being of divorcees and their children. 20–80% of divorcees report some degree of parental conflict, while 5–25% report high levels of conflict (Bonach, 2005; Hald et al., 2020d; Ottosen et al., 2017; Symoens et al., 2014). Higher levels of conflict have been associated with reduced well-being and poorer mental and physical health among both adults and children (Amato, 2000, 2010; Davidson et al., 2014; Escapa, 2017; Lamela et al., 2016). Some scholars emphasize the importance of high levels of parental conflict as the main source of maladjustment for parents and children as opposed to the divorce itself (Amato, 1993; Gähler & Palmtag, 2015; Jekielek, 1998; Nunes-Costa et al., 2009; Roberson et al., 2011; Troxel & Matthews, 2004).

In CAD, this perspective is integrated in the intervention design in three important ways: (1) by viewing some level of conflict as a basic (and normal) condition and challenge for most divorcees; (2) by including knowledge and tools, which directly support conflict prevention and management, such as the digital modules “Avoiding typical pitfalls” and “Dealing with conflicts”; and (3) by informing about the benefits of co-parenting cooperation and teaching proven strategies for establishing or re-establishing post-divorce cooperation.

Content and Duration

The CAD digital intervention consists of 17 digital modules and supporting functionality and contents, which are accessed online from a computer or tablet. The digital modules take approximately 15 to 50 min to complete (see also Table 2), and participants can freely choose which and how many modules they want to use. The learning modules address challenges relevant for divorcees and are arranged into three main themes presented in the flowchart below (see Fig. 2A–C).

Fig. 2
figure 2

CAD modules and content topics

Full size image

Geasler and Blaisure (1999) and Braver et al. (1996) identified several content areas that would be important to include in a divorce intervention: (1) Adult-focused, (2) Child-focused, (3) Court-focused, and (4) Coparenting under special circumstances. The three main themes and the sub-domains of CAD are generally in line with these content categories, though there were no main themes related to court-focused content, and the co-parenting theme covered little related to “special circumstances” (please see the discussion). The curriculum of each of the elements is described in the following Table 1. The content and exercises were not created from a single psychological school or theoretical framework, but rather integrate elements of Cognitive Behavioral Therapy (CBT), Narrative Therapy techniques, and Acceptance and Commitment Therapy (ACT).

Table 1 Summary of CAD module content
Full size table

Multimedia

Cooperation after Divorce uses a variety of different audio-visual media strategies to ensure effective and engaging communication and to increase the likelihood that messages and concepts are remembered and applied in practice; please see Table 2 for an overview.

Table 2 Content type, scope, and duration
Full size table

The average CAD module consists of 43 screens. Each screen is individually designed with a combination of illustrations, pictures, and short texts. Two of the core design principles were to use a minimum of text and support all written or visual content with a voice-over. The purpose of this multimedia strategy is both to make the content more accessible and attractive to everyone, regardless of reading ability, and to support memory encoding and recall (Goldstein, 2015). Videos were employed when telling personal stories or stories with emotional content. This includes video-based documentary case stories, role-play videos in which actors exemplify typical situations and dilemmas, and expert videos with divorce professionals. Animations and motion graphics (animated graphical illustrations) were used to explain and visualize data, facts, theoretical models, and more complex relationships between causes and effects. CAD includes an average of 5 min of video and animation per module.

Engagement and Design Strategies

In their review of six online divorce programs, Bowers et al. (2011) identify several limitations to extant programs and conclude that the biggest challenge seems to be the adaption to the online format: “For the most part, it appeared that face-to-face programs had been put online with little consideration of online instructional strategies that facilitate learning.” (p. 786). Instead of being a simple adaption of a face-to-face program, CAD was “born digital” and designed to take full advantage of the digital technologies and techniques from the beginning. On average, every three minutes, the user is asked to make a choice, answer a question, or complete an interactive exercise. This is part of CAD’s active teaching and engagement strategy. Choices and questions help users reflect and personalize content and messages to their own, relevant context. The interactive exercises help users relate their own situation to the material, consider alternative options of thinking and actions and thereby increase the likelihood of translating knowledge into behavior. Parallel with this, “Specific, Measureable, Achievable, Realistic, and Timely (SMART)” objectives (Doran, 1981) for each of the 17 modules were defined. These SMART objectives describe what the end-user is expected to know or be able to do as a result of engaging in a learning and intervention activity. By being aware of the objective, the probability of reaching it increases. At the same time, the objectives guide the team planning, development, and evaluation of the content. This approach was inspired by Bloom’s Taxonomy Model (1956) and Dreyfus and Dreyfus Model for Skills Acquisition (1980). Other principles, regarding the tone of voice and the severity of the content, were inspired by the early user-tests of the prototype and the intention of CAD to communicate in a respectful manner.

Method

Development Process

The prototype version of the intervention was tested from July 7th, 2015 to November 15th, 2015. During this time, more than 50 technical and content issues were detected and rectified. An updated Beta-version with no known errors was released on November 16th, 2015, and beta-testing was conducted with the modifications from the prototype testing integrated. During beta-testing, less than 10 technical or usability issues were detected and rectified. The final version was released on January 1st, 2016. Data from participants from prototype and Beta-version testing were excluded from the final study.

Participants

A total of 1882 recently divorced individuals accepted the invitation and 98.6% (N = 1856 individuals; 66.8% women, and 33.2% men) completed the baseline survey. A full description of this complete sample can be found in Cipric et al. (2020); Hald et al. (2020a); Øverup et al. (2020); Sander et al. (2020). The sick days variable of interest in the current analyses was added to the study after the official start of data collection; thus, only responses from participants enrolling in the RCT after July 12th, 2016 were recorded (see Fig. 3). Thus, this reduced the sample to 778 respondents on the sick days variable (see description of the calculation of the sick day variable), of which 67% identified as women. The average age of the 778 participants in the current sample was 45.97 years (SD = 8.23) and the average length of marriage before the divorce was 13.59 years (SD = 7.95). 93.1% of participants had children and 89% were first-time divorcees. Participants responded to the baseline survey on average about 5 days after they obtained their juridical divorce. For further sample details, please see Table 3. Moreover, a thorough breakdown of the response rates for the sick day variables can be found in the supplemental materials (Table 1s).

Fig. 3
figure 3

CONSORT diagram

Full size image

We obtained socio-demographic data from Statistics Denmark on all people who were legally divorced during the study period in Denmark. The study participants were representative in terms of marriage duration (p > .05) but differed from the national divorced population in terms of age (participants were, on average, less than one year older: t(777) = 2.74, p = .006, Cohen’s d = 0.10), gender (more women participated: χ2 (1, n = 778) = 92.32, p < .001), educational level (study participants were more highly educated: χ2 (2, n = 778) = 578.63, p < .001), income (study participants earned slightly more : t(777) = 2.08, p = .04, Cohen’s d = 0.08, and number of previous divorces (respondents had on average fewer previous divorces: t(777) = − 7.28, p < .001, Cohen’s d = 0.26).

Table 3 Participant sociodemographicinformation (N = 778)
Full size table

Procedure

During the study period, which ran from January 2016 to January 2018, divorce procedures in Denmark were initiated by one spouse submitting an application to the Danish State Administration (DSA). Divorce was granted immediately without a mandated separation period in divorce proceedings in which both spouses agreed to the divorce and its terms. In cases in which one or both spouses disagreed with the divorce or its terms, divorce was granted after a 6-month separation period, which according to the DSA occurred in approximately 30% of divorces.

The CAD RCT was a 12-month online parallel-group intervention study that assessed the CAD intervention’s efficacy in reducing self-perceived stress, anxiety, depression, hostility, somatization, overall mental and physical health, the number of sick days, and parental reports of children’s health-related quality of life. Here, we report on the number of sick days. The effectiveness of the CAD solution on the other outcomes can be found in Cipric et al. (2020); Hald et al. (2020a); Øverup et al. (2020); Sander et al. (2020).

Recruitment was done in cooperation with DSA, which included an invitation in a message sent to the e-Boks (online digital mailbox used in Denmark to communicate with the Public Sector and Governmental Agencies), along with the official divorce decree. All Danish adults are required to have an e-Boks and less than 3% of the entire Danish population is exempt from receiving digital mail from government agencies. The DSA was not able to provide the exact number of invitations sent out in the study period and the response rate therefore cannot be calculated.

Study participation was initiated by interested participants by creating an account on the CAD webpage utilizing the enclosed link sent in the invitation letter by the DSA. The link directed people to a simple homepage with information about the project, contact and support information, a 2-minute information video, and a log-in section for registered users (www.ses.dk). Once on the CAD platform, participants provided informed consent by clicking a box, completed the initial questionnaire, and were subsequently randomized into the intervention or treatment-as-usual control group. A randomization schedule was set up so that, during the study inclusion period, participants were assigned sequentially over a 2-week period to either the intervention or control group, resulting in a total of 27 recruitment rounds for the intervention group and 27 recruitment rounds for the control group (i.e., 108 weeks). The assignment schedule was blinded to the researchers during the inclusion period. Intervention group participants received access to the digital intervention upon completion of the baseline survey and had access for the duration of the study. Participants were able to use as many or as few of the modules as they wished; that is, they were able to select those modules that they felt were of relevance to their situation.

The 3-, 6-, and 12-month follow-up questionnaires were sent via email. Reminder e-mails containing summaries of popular content, such as “children’s typical reactions to divorce” or “how to deal with crisis or grief” were sent to all registered users, who had been inactive (had not visited the website) for 30 days. This was done manually on a monthly basis. The reminder e-mails included a link to the intervention website and encouragement to revisit the intervention for more in-depth knowledge and tools. Survey responses for the baseline, and 3-, 6-, and 12-month follow-up questionnaires were stored anonymously on a secure server. The study was approved by the Danish Data Protections Agency. The study was exempt from further ethical evaluations following the rules and regulations as set forth by the Scientific Ethical Committees of Denmark.

We would like to acknowledge that the development of the digital intervention was financially supported by the Egmont Foundation; moreover, this project was supported by the Carlsberg Foundation’s ‘Distinguished Associate Professor Fellowship’ (given to the last author) under Grant no. F16-0094. For due diligence, we would like to declare that two of the co-authors (First Author and Last Author) own the digital intervention through the Company ‘Cooperation after Divorce’ (Samarbejde Efter Skilsmisse ApS); these authors did not have access to the data nor did they decide on or complete the data analyses. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Measures

Sociodemographic Variables

The following variables were used to evaluate the sociodemographic characteristics of the participants: (1) gender was evaluated with the question “Are you a man or a woman?”. The response options were: 0 = man and 1 = woman, (2) age at divorce (in years), (3) education level (the highest level of completed formal education). Responses were categorized so that 0 = low level of education (e.g., primary school, high school, business high school, vocational education), 1 = medium level of education (e.g., medium-cycle tertiary education, bachelor’s degree), and 2 = high level of education (e.g., master’s degree or higher). (4) Monthly income was given employing a 9-point scale of 10,000 DKK intervals (approximately 1,500 USD). 0 = below 10,000 DKK (less than 1,500 USD) to 8 = more than 80,000 DKK (more than 12,000 USD) and was used for the main analyses. Monthly income was also divided into three categories, with 1 = “Below national average” (1–3; less than 30,000 DKK), 2 = “At the national average” (4–5; 30,001–50,000 DKK), and 3 = “Above the national average” (6–9; above 50,001 DKK) (1 DKK = 0.15 USD). These were categorized so that 0 = below the Danish national average monthly salary, 1 = at the Danish national average monthly salary, and 2 = above the Danish national average monthly salary.

Marriage and Divorce-Related Variables

Several questions were used to obtain information about participants’ divorce and former marriage: (1) the number of previous divorces (number), (2) marriage duration (in years from marriage date to legal divorce date), (3) legal divorce duration (in days from legal divorce date to study initiation date), and (4) the number of children (number).

Sick Days

The number of sick days was measured with the following question: “In the last three months, how many days have you been sick?” Responses were given in whole days using drop-down response options ranging from 0 to 30 + days. Responses of 30 + days were conservatively recoded to ‘30’ days. To calculate the total number of sick days per participant, we summed the reported number of days at 3, 6, and 12 months (we did not use baseline scores as they include sick days 3 months prior to being able to use the intervention). Missing data points for individual cases were treated by applying the following formula: 3 m + 6 m + 12 m = X; X/3 = Y; Y * 4 = the total numberFootnote 1.

Data Analysis

We report the means and frequencies of the socio-demographic and divorce-related variables in Table 3. We used two-sample t-tests and chi-squared analyses to assess group differences in this table.

The overall sample attrition was roughly 70% from T1 to T2 (see Cipric et al., 2020; Hald et al., 2020a; Øverup et al., 2020; and Sander et al., 2020 for a full description). For the sick day specific analyses reported here, the sample attrition was 65% from T1 (n = 697) to T2 at 3 months (n = 243) in the intervention group and 65% in the control (T1 = 583, T2n = 206), but stabilized in subsequent waves at 6 and 12 months (intervention group: T3 = 249 and T4 = 225; control group: T3 = 206 and T4 = 187). This attrition rate is consistent with the dropout rates expected in online health evaluations (Donkin et al., 2011; Eysenbach, 2005; Geraghty et al., 2013). In previous research, the attrition rate ranged from 0 to 83% (i.e., Donkin et al., 2011), though one study found that 61% of participants in internet smoking cessation interventions were lost to the first follow-up and 78.4% of participants were eventually lost to follow-up completely (Geraghty et al., 2013), suggesting that typical attrition rates are likely at the higher end of the attrition range specified above (Eysenbach, 2005). In order to account for the drop-out, the number of sick days was imputed for each of the 778 participant by calculating an average number of sick days for individuals who participated in at least one study wave and multiplying that number with 4 (the number of study waves; see above in measure description). Given the large attrition rate, we examined whether those that reported on at least 1 sick day variable at the 3, 6, or 12-month time point differed from those that only completed the sick day variable at baseline in terms of sick days at baseline. We found that those that reported on sick days during the follow-up surveys had slightly fewer sick days at baseline (M = 3.22, SD = 6.54) than those that only reported on sick days at baseline (M = 4.04, SD = 7.54), t(1264.8) = 2.07, p = .039, Cohen’s d = 0.12.

With respect to the main data analyses, we then used two-sample t-tests to compare the total number of sick days reported over the one-year follow-up by those in the intervention group with those in the control group. The assumptions for the two sample were met as (a) the variable is of continuous measurement, namely the number of days being sick (dropdown menu with a number, ranging from 0 to 30+), and (b) observations from the two group were independent from each other (i.e., persons in the intervention group were unrelated to persons in the control group, to the best of our knowledge) and identically distributed. However, (c) the sample variances were not homogenous, and therefore, we used the Welch’s approximate t-test estimates. These analyses were performed on an intention-to-treat principle. A series of negative binomial regressions were conducted to examine whether covariates (i.e., gender, parental status, education level, and number of modules used) predicted the number of sick days. Negative binomial regression is used when the outcome is a count variable, as is the case with sick days.

Results

Two-sample t-tests assessed the number of sick days over the first 12 months of divorce. Participants in the intervention group reported significantly fewer sick days (M = 9.57, SD = 17.74) than those in the control group (M = 15.39, SD = 29.96), t(766) = -3.60, p < .001, Cohen’s (d) = 0.26. Additional analyses, in which we limit the analytic sample to examine the robustness of the findings, can be found in the main results section and Table 3s in the supplemental materials.

When stratified by gender, women in the intervention group had significantly fewer sick days (M = 12.06, SD = 20.80) than women in the control group (M = 17.05, SD = 29.32), t(521) = − 2.25, p < .001, Cohen’s (d) = 0.20. Similar results were found among men; men in the intervention group had significantly fewer sick days (M = 5.21, SD = 8.98 days) in the control group (M = 11.22, SD = 19.35 days), t(253) = − 3.34, p = .001, Cohen’s (d) = 0.43 (see Fig. 4). An independent samples t-test found that women reported more sick days (M = 14.50, SD = 25.43) than did men (M = 7.61, SD = 14.35), t(760) = − 4.82, p < .001, Cohen’s (d) = 0.31. However, a negative binomial regression did not yield an interaction effect between gender and group assignment (p = .127), supporting the pattern of findings from the stratified t-tests that the effect of group assignment was similar for men and women.

Fig. 4
figure 4

The average number of sick days over the first 12 months post-divorce in the intervention and control groups by gender

Full size image

An additional negative binomial regression also examined parental status, education level, and number of modules used as covariates; group assignment and gender were also included in the analyses. The results revealed that only gender remained a significant predictor of sick days (p < .001); parental status (p = .263), educational level (p = .663), and number of modulesFootnote 2 started and completed (p = .116) were unrelated to the number of sick days. A complete results table can be found in Table 4s in the supplemental materials.

Discussion

The current manuscript details the development of a digital intervention aimed at helping divorcees cooperate post-divorce, particularly with regards to parenting. Parents and children alike may experience negative outcomes post-divorce, in part because of parental conflict regarding parenting practices (Amato, 2014; Ottosen et al., 2017). The use of a digital intervention, in which divorcees receive psychoeducation and complete therapeutic activities separate from their former partner, may be beneficial because of the inherent distance from the object (i.e., the former partner; Hald et al., 2023). By working through issues separately, hostility may be reduced (Øverup et al., 2020), allowing for improved collaboration between former partners. Indeed, many of the interactive activities in the intervention focused on highlighting how maladaptive interactional patterns with the former partner may negatively affect both the children and the adults; the content also sought to aid the user to develop new interactional schemas, including thinking about when and how to communicate with the former partner in a way that does not escalate conflict. There is considerable evidence documenting that this digital post-divorce intervention also improves individual mental and physical health (Sander et al., 2020). Thus, the positive effects of a digital intervention may be both intrapersonal and interpersonal.

The CAD Intervention Effects on Sick Days

At the time of the study (2016–2018), there were approximately 8.5 workdays lost due to sickness per person in Denmark (Stewart, 2022). In the current study, the intervention group participants reported an average of 5.82 fewer sick days during the first 12-months post-divorce than the control group did (intervention group = 9.57 sick days per year; the control group = 15.39 sick days per year) with this difference being statistically significant. More healthy days have important implications for the individual’s health and health-related quality of life, as even one more healthy day is a meaningful change for an individual (Moriarty et al., 2003).

One might speculate that the CAD intervention, due to its efficacy in addressing symptoms of stress, depression, anxiety, and somatization, contributes significantly to the overall health improvement of the participants. This would explain the reduction in sick days observed in the intervention group compared to the control group. However, another dimension to consider is the potential psychosocial impact of the CAD intervention. By enhancing parents’ focus on, comprehension of, and coping strategies for their children’s emotions and behavior post-divorce, the CAD intervention might indirectly be reducing the emotional and psychological burdens on parents arising from parenting. This, in turn, might make parents less likely to feel the need to take “mental health days” off. Finally, with its focus on conflict management and promotion of good post-divorce co-parenting cooperation, the difference in sick days between the intervention and the control group may also be a reflection of less need to take time off to meet with layers or the juridical system to resolve disagreements with a former partner. We hope that future studies and designs will allow for more in-depth understanding of the mechanisms, whereby digital interventions, like CAD, may affect users to further the understanding of the advantages and limitations of digital interventions in the family or related areas of study.

We also speculate that a decrease in the number of sick days may be beneficial to the parents themselves, allowing for time spent on activities that lead to increases in quality of life such as physical activities, being with friends, or other hobbies and personal interests. Further, more healthy days for parents may have indirect effects on children’s wellbeing especially if these sick days are related to parental mental health concerns, like depression and anxiety, as decreased parental mental health is related to poorer well-being among children, including physical health, parent-child relations, behavioral problems, educational attainment, anxiety, and depression (Bennett et al., 2012; Panter-Brick et al., 2014; Pearson et al., 2016; Turney, 2011). Thus, in the intervention group, there may have been downstream positive effects of the intervention for the children, as families in this group may have had fewer mental health-related sick days. On a societal level, a decrease in sick days may lessen the healthcare burden associated with divorce in terms of less use of the health care system and less loss of productivity for the workforce, and savings for the workplaces. Future research should conduct health economics analyses of the intervention effects to qualify these economic impacts.

We found that women reported more sick days than their male counterparts did, and a study on general workforce sick day work absences in Denmark has also found gender differences, although they were smaller (Thorsen et al., 2018). This is most likely related to parenthood status, as Hald et al. (2020) found that parenthood status was significantly associated with more reported sick days in the three months prior to divorce. Even in a relatively egalitarian society such as Denmark, women are often the primary person responsible for childcare, and this may explain the gender differences found in this study.

Intervention Content and Design

Bowers and colleagues (2011) proposed a checklist for evaluating the content of divorce interventions. Employing their checklist to assess CAD, we found that CAD included all of the key content areas regarding Child-Focused, Adult-Focused, and Parent-Focused Contents. However, CAD was not exhaustive with respect to Court-Focused Content. Material regarding legal procedures and the law was not included in the program as this was not considered relevant in the Danish cultural context. However, CAD did contain modules and exercises in terms of co-parenting, such as content about best interests of the child and parenting plans from Bowers et al. (2011) checklist. Finally, CAD did not include content related to Co-parenting Under Special Circumstances, for example, in situations of intimate partner violence or co-parenting children with special needs.

Schramm and colleagues (2018) developed a conceptual framework for the content of divorce interventions, based on a literature review. According to this framework, there are three content tiers: (1) Core content (child focused), (2) Strategic content (adult focused), and (3) Unique circumstances. In reference to tier 1 (child focused content), research into divorce education program content concluded that there are four key focus areas that, if included in divorce intervention programs, are likely to improve healthy post-divorce adjustment for children: impact of divorce on children, reducing inter-parental conflict, positive co-parenting skills, and parenting strategies in divorce (Shramm et al., 2018). CAD includes content and exercises focusing on these four key areas, as well as 18 of the 19 core-content sub-areas described in this article (Schramm et al., 2018). The only area not covered was dating and introducing new partners.

Tier 2 (Adult focused content) is divided into three areas: adult self-care issues, managing divorce issues, and moving forward in life (Schramm et al., 2020). CAD contains content and exercises focusing on these three areas, based on psychoeducation, mindfulness exercises, techniques from Cognitive Behavioral Therapy (CBT), Narrative Therapy, and Acceptance and Commitment Therapy (ACT). Regarding Tier 3 (Unique circumstances), CAD did not have specific content regarding co-parenting children with special needs, domestic violence, military/long-distance parenting, and being left vs. leaving the marriage. However, the module “Ask the experts” allowed for those in unique circumstances to ask questions regarding these or similar issues.

Limitations and Future Directions

While this intervention program and the sick days study has many strengths, including the large sample size, RCT study design, its participant recruitment strategy through governmental collaboration with the DSA, and being “born digital” (vs. being an adaptation of a face-to-face program), there are limitations to take into account. First, the measure of sick days does not necessarily inform about work absenteeism, which limits the comparability with other studies that assess work absences, nor does it specify being “sick in bed” as one of the few other studies on sick days and divorce published (Bookwala et al., 2014). These differences in measurement and psychometric considerations limit the comparability of the outcome measure and the contextualization of these results. Nonetheless, self-reports of the number of sick days have been found to be valid measures (Stapelfeldt et al., 2012) and are a straightforward and easily comprehendible measure of overall health and health-related quality of life, and in a Danish context, as noted in the introduction, ‘sick days’ are often equated with being away from work, that is, work absenteeism. Further, supplementary analyses revealed great stability across sick days results (see also supplementary materials) when comparing sick days of participants with different response patterns, for example, when comparing participants who have responded to two or more measures of sick days. This indicates that the sick days results were not prone to influence of the frequency of participants responses over the duration of the trial.

Secondly, the intervention content did not meet international researchers’ recommendations of including Court-Focused Content (Bowers et al., 2011). However, these recommendations were developed in the American legal and cultural context, which differ from the Danish or most other cultural contexts. Concretely, this area of court-focused/legal counsel would not be relevant in Denmark, particularly during the time of the study, as the divorce procedures for mutual agreement divorces were easy to navigate (Fallesen, 2021). While this content area may be important in the United States and other countries with more opaque or convoluted legal structures, we find that evaluation criteria for divorce educational programs may need to be more universal and/or more culturally sensitive. Further, such additional culturally relevant modules can be developed and added to CAD or other interventions on a culturally-specific “need to” basis which is an advantage of CAD or similar solutions.

Another area that is not explicitly covered in CAD is dating and introducing new partners. While there is content on “bonus families” (the literal translation of the Danish term for “stepfamilies”), perhaps the CAD program would have benefited from targeted material on dating and introducing new partners. Finally, as the CAD program was designed to be scalable and nationally implementable, unique circumstances, such as parenting special needs children and co-parenting among divorced military families, were not included. While customizability was possible through the “ask the experts” feature, future developments and additions to CAD could and should focus on tailored content that is available for divorcees in unique circumstances or uniquely required in certain cultural settings and as CAD is thematic module-based, this would be fully possible. The last limitation is that in this manuscript, we compare the content areas to international empirical research standards for divorce education, but we do not have third-party experts assessing the quality of the content in each of the areas, however, external experts were used in the development and approved all of the content.

Conclusions

The CAD digital divorce platform is a nationally implementable divorce platform that was developed digitally for a digital age. The material included in CAD meets international recommendations for the contents of divorce education programs. In this study, CAD was demonstrated to be effective in reducing the number of sick days in the first year post juridical divorce by nearly 6 days on average. A reduction of this size has positive human and economic implication at both an individual and societal level.