Youth Psychopathology in Daily Life: Systematically Reviewed Characteristics and Potentials of Ecological Momentary Assessment Applications

The initial search resulted in 1596 articles. Finally, 50 studies were included in our review (see PRISMA flowchart in Fig. 1 for an overview of the selection procedure).

In Table 1, we describe the characteristics of the patient group and, if applicable, comparison groups, and details of the EMA procedures for all included studies. Of all studies, 15 studies examined ADHD, 5 autism spectrum disorders (ASD), 12 mood disorders (4 of which also examined anxiety disorders [18‐21]), 16 anxiety disorders, 5 eating disorders, and 1 psychosis. The studies on ADHD considered 11 different samples of children and 1 sample of adolescents. The studies on ASD included 3 samples of children and adolescents with high-functioning autism spectrum disorder (HFASD). All studies on mood disorders involved children and adolescents with major depressive disorder (MDD); 1 study also included youth with bipolar disorder (BD). The studies on anxiety disorders mainly concerned children and adolescents with generalized anxiety disorder (GAD), separation anxiety, or social anxiety disorder (SAD). All 5 studies on eating disorders investigated adolescents with anorexia nervosa (AN), with (partial) overlap between study samples. The one study on psychosis considered adolescents with early-onset psychosis. Both school-age children (mostly between 7 and 12 years of age) and adolescents (mostly > 12 years) were studied in the context of the above-mentioned disorders, with the exception of eating disorders and psychosis (adolescents only). Sample sizes were variable, ranging from less than 20 (in case reports or pilot studies) to 133 youths with a mental disorder. Most studies (34) included a healthy or community-based comparison group; 3 studies also included a comparison group with other DSM diagnoses.

Youth completed EMA in 34 studies, their parents in 8 studies, and both in another 8 studies. EMA completed by both youth and their parents was used in the context of studying ADHD, and EMA completed only by parents about their child in ADHD and once in anxiety disorders. Various formats were used to obtain data: phone calls (17 studies); applications installed on a personal digital assistant (PDA) (14 studies), smartphone (9 studies), or iPod touch (3 studies); diary booklets with (3 studies) or without (3 studies) digital wristwatch or device; and online surveys on a smartphone or computer (1 study). One study also used vials to collect saliva samples for cortisol assessment. The use of phone calls and diary booklets occurred in the context of studying mood and anxiety disorders. Mobile technology applications or web surveys were also used once in this context, and were the only format used to study other disorders. Signal-contingent assessment was used most often (36 studies), followed by time-contingent assessment (14 studies) and event-contingent assessment (4 studies; with events being separations and socially distressing events in anxiety disorders, and auditory verbal hallucinations in psychosis). The 3 studies that were both time- and signal-based, and the 1 study that was both signal- and event-based were counted twice. The number of assessments ranged from 2 per day to every full hour (except sleeping hours) for periods of 2 to 28 days, resulting in 6 to 168 assessments across studies.

The diverse possibilities of using EMA are underlined by the various EMA applications that were found for studying youth psychopathology, capturing a broad age range, several types of psychopathology, and different EMA procedures with regards to format, contingency, intensity and duration. The results of our review pointed out potential gaps regarding, for example, the use of event-based EMA or parent-report EMA across the types of psychopathology. Our review also indicated gaps with regards to the psychopathology types and age groups that were studied using EMA measures. For example, even though we included the broad spectrum of ASD and eating disorders in the search terms, only studies that examined youth with HFASD and adolescents with AN were found. Also, there was but a single pilot in early-onset psychosis. In comparison, psychotic disorder has frequently been studied using EMA in adults (e.g., [68]).

Considering our findings regarding compliance, EMA methods seem generally feasible in various populations of youth with psychopathology. In addition, EMA appeared to provide measures with adequate construct validity, as indicated by findings of associations with related retrospective assessments. Besides, we noted differences in EMA data between clinical groups and comparison groups, and the ability of EMA data to predict outcome or functioning. Nonetheless, given findings in, for example, youths with HFASD [22, 23], it seems important to adapt EMA to the abilities and experiences of the specific target group in order to achieve compliance and valid assessments. Moreover, the information provided by mobile applications versus phone calls may be further evaluated, as quick probes or follow-up questions can be added in phone calls when an unclear answer is provided [28]. Despite the idea that youths may be skilled in using mobile applications, phone calls that allow for answering verbally may be suitable for youth as well, also because literacy may vary in children and adolescents. However, by using images, adapted response scales, clear instructions, and training these concerns may be overcome (for a further discussion of methodological considerations, also see Heron et al. [16]).

In relation to compliance and validity, we also express some caution regarding the possibility of reactivity to the EMA protocol and EMA being perceived as burdensome (e.g., based on concerns raised by youths with psychosis [31]). Reactivity when using EMA to examine psychopathology refers to participants’ increased awareness of symptoms. This could be due to the frequent assessments, which could in turn result in a change in symptoms or behavior. Surprisingly, although some authors considered the possible influence of potentially predictable time-based EMA on behavior (e.g., [42]), we hardly found any EMA studies reporting on reactivity. The impact of participating in an EMA protocol has been investigated in adults with psychopathology, and a reduction in the severity of symptoms of posttraumatic stress over the EMA monitoring period was found, which was not found over an unmonitored control period [69]. Reactivity should be taken into account when interpreting findings, and should be reported on more often. In terms of burdensomeness, EMA can be considered more demanding of parents and their children than more conventional methods. As a result, the willingness or ability to complete assessments may decrease. When designing intensive or lengthy EMA protocols, the possible burden placed on parents and their children should be weighed against the benefit of richer data.

In line with previous research on cross-informant correspondence in clinical assessment (e.g., low-to-moderate correspondence [70]), parental EMA data of their child’s moods and behavior were found to differ from the child’s own EMA data. While parental EMA data may in some cases be more informative, both may be considered of value. The child’s age may also play a role, as adolescents generally have more self-insight than primary school-age children. For example, based on findings in children with ADHD, it could be suggested that parental momentary reports on their children with ADHD may be more reliable and valid than child momentary self-reports [26, 32]. Moreover, both in children with ADHD and in children with separation anxiety, parent reports were better predictors of the children’s emotional and behavioral difficulties and specific diagnosis than child reports [25, 29, 30]. On the other hand, among adolescent girls with ADHD, in general there was similarity between their self-reports of negative emotions and those of their mothers [27]. Ultimately, the decision to involve both parents and children in an EMA study on youth symptoms may best be based on the specific research questions of a study along with considering children’s developmental stages.

The results of this review showed several ways in which EMA studies may provide insights the phenomenology of youth psychopathology. First, we demonstrated how EMA can be used to gather data on the many daily environmental and interpersonal factors that cannot easily be imitated in a laboratory but may be important variables of interest (e.g., day-to-day social context, such as interactions with peers, and the relationship with positive affect [54], co-rumination during specific social interactions [46], and the actual occurrence and impact of fearful situations [50]). It should be noted that most studies conducted EMA outside of school hours, and relevant information could be missed. Studies should therefore consider the possibilities and added value of conducting EMA during school hours, even though doing this may be challenging or undesirable. However, the reviewed applications do confirm that by assessing youth in their natural environments, EMA may provide valuable and ecologically valid information about specific events, social and affective contexts, and the relations between them in the daily lives of youth. This information may add to the understanding of mechanisms that contribute to the development of psychopathology and may provide potential targets for prevention and treatment.

Second, results showed how EMA could minimize retrospective recall and the subsequent risk of recall biases by asking questions near real-time (e.g., in adolescents with HFASD [22]). More generally, findings suggest that EMA data may provide different and possibly less biased information than less frequent retrospective reports, and may offer researchers and clinicians an additional methodology for increasing the accuracy of the assessment and monitoring of stressors and other relevant measures in youth with developmental disorders or other mental disorders. It should be emphasized that retrospective reports are, of course, also relevant for specific research questions. Compared to ‘in the moment’ EMA reports, they may integrate multiple experiences or an evaluation of one’s overall experience.

Finally, results showed how studies used EMA to capture day-to-day or even within-day fluctuations in mood, behavior, and context, and temporal associations between these variables which may not be captured by more traditional methods (e.g., in ADHD [25, 32], and in AN [59]). Relatedly, we noted that EMA was also combined with physiological measures, such as actigraph-measured sleep to examine the bidirectional association between daily affect and sleep [19], and cortisol testing to examine cortisol diurnal activity [21] in youths with affective disorders. EMA thus appears suitable for providing information on these fluctuations and within-person variability of affect and related variables or contexts in youth with psychopathology, and allows for studying temporal associations. Thanks to the richness of within-person data, EMA could add to more conventional methods by providing insights into and predictions on individual patterns of, for example, affect (e.g., [28]).

The latter benefit was also demonstrated in studies that used EMA to study parent–child interactions by independently assessing mother and child over time (e.g., their anger levels [40]). By studying the temporal relations between the moods and behaviors of parents and their children with EMA, both favorable and unfavorable patterns in dyads or families with a child with psychopathology could be identified. This type of information may eventually assist in tailoring interventions designed to improve the child’s functioning specific to the patterns in a dyad or family.

Our review showed how EMA may allow for the close monitoring of the processes of change during treatment. Thereby, EMA could add to the understanding of the changes and factors that contribute to treatment outcome. Traditionally, treatment studies rely on pre- and post-treatment questionnaires. However, in this way, important information about the course of treatment may be missed. Several studies have used EMA to monitor and predict treatment response (e.g., [20, 64]). Eventually, individual patterns visible in the EMA data may assist in identifying which youth will likely benefit from which treatment and whether their treatment should be changed. As a final note, in addition to monitoring and predicting treatment effects, EMA may even add to treatment in itself. Ecological Momentary Interventions (EMI) [71] could provide real-time support, for example, by cuing youths to practice CBT skills in anxious situations [72] or sending out positive thought assignments when someone indicates feeling depressed. It has already been suggested that adults with depressive disorder receiving pharmacological treatment benefit from additional feedback derived from EMA, providing insight in personalized patterns and contexts of affect [73].

A strength of our review is that, to the best of our knowledge, it is the first to provide a systematic review of the use of EMA in the broad range of youth with internalizing and externalizing disorders. In addition, it has shown how EMA methods can be used to provide feasible and valid assessments of the daily presentation and course of youth psychopathology, interactions of symptoms with contextual factors, including parent–child interactions, and the course and outcome of treatment.

A limitation of this review is the use of strict study selection criteria. Some interesting EMA studies, for example, on non-suicidal self-injury, were excluded because they did not fulfill the criterion of studying a clinical sample with a DSM-IV or DSM-5 diagnosis. Also, as a consequence of excluding non-clinical samples, we could not review the potential of EMA in the context of prevention, even though frequent monitoring of mood or behavior may be very useful for detecting youth at risk for mental health problems. Another potential limitation of the current review was that it included a narrative section on potentials of applying EMA that were shown in the studies, which has possible disadvantages as it involves a less defined synthesis of findings. Yet, it did allow us to synthesize the potential of the variety of EMA applications in more detail within a structure of three topics.

Regarding future directions, we encourage other researchers to make use of the adaptability of EMA and the richness of EMA data and to further apply it in studying youth psychopathology covered, and also psychopathology not covered in this review. For example, despite being mentioned in the search terms, no studies were found on substance use disorders or posttraumatic stress disorder in youths; EMA could also be applied in studying these disorders in youth (e.g., by gathering real-life contextual data on use or symptoms), as has been done in adult populations (e.g., [74, 75]). EMA may provide more opportunities for researchers and clinicians than those already described in this review. First, some studies used EMA to examine interesting variables in the daily lives of youth but did not relate the variables to social and affective contexts. For instance, Primack et al. [45] adapted EMA to examine the media use of youths with MDD, but media use was not linked to other variables such as negative and positive affect before, during, and after use. Second, interesting applications that were not found in the included literature consider the use of event-contingent EMA to analyze externalizing problems or eating behaviors. In non-clinical samples, event-contingent designs have already been used in, for example, children with binge eating symptomatology [76]. Third, while temporal relations between maternal and child behavior and moods were examined using EMA in youth with ADHD, this was not yet found in youth with internalizing problems. Fourth, concerning data analysis, some studies computed the average of the within-person fluctuations of the assessed variables. However, EMA data allows for more elaborate analyses that consider these within-person fluctuations to reveal interesting information about youth psychopathology. We encourage all to profit from the wealth of data resulting from EMA, and to conduct more elaborate analyses of existing data when this has not yet been done.