Mood disorders in everyday life: A systematic review of experience sampling and ecological momentary assessment studies

Abstract

In the past two decades, the study of mood disorder patients using experience sampling methods (ESM) and ecological momentary assessment (EMA) has yielded important findings. In patients with major depressive disorder (MDD), the dynamics of their everyday mood have been associated with various aspects of their lives. To some degree similar studies have been conducted in patients with bipolar disorder (BD). In this paper we present the results of a systematic review of all ESM/EMA studies in MDD and BD to date. We focus not only on the correlates of patients' everyday mood but also on the impact on treatment, residual symptoms in remitted patients, on findings in pediatric populations, on MDD/BD specificity, and on links with neuroscience. After reviewing these six topics, we highlight the benefits of ESM/EMA for researchers, clinicians, and patients, and offer suggestions for future studies.

Introduction

In clinical practice symptoms are usually assessed retrospectively. It has been argued, however, that mood disorder patients' symptom recall is biased by their dysfunctional attitudes about themselves and their surroundings. This may lead them to selectively attend to negative aspects of their everyday lives (Beck, 1963). A study by Mokros (1993) highlights the effect this recall bias might have in clinical practice. Seven adolescents were diagnosed with a major depressive episode by an experienced clinician. A week later all patients were still considered depressed, based on the clinician's ratings of their recalled symptoms. Crucially, however, this finding was not in agreement with patients' real-time reporting of these symptoms. During the week patients had repeatedly completed short, standardized forms in response to frequent, irregularly occurring pager signals. There were clear discrepancies between patients' reporting of symptoms during the week and their recall of these symptoms at the end of the week.

In psychology, the method used by Mokros (1993) to repeatedly assess people in real-time is best known as experience sampling (Larson & Csikszentmihalyi, 1983). Experience sampling aims to systematically obtain self-report data on participants' everyday lives at many points in time. To this end participants are generally required to carry pagers signaling at unpredictable intervals, usually in the range of 1–2 h, and to complete forms as soon as possible after each pager signal. This is known as signal-contingent data recording. With up to 10 signals per day over multiple days, and usually only a few missed signals, the number of repeated measurements per participant is much higher than the number obtained with more traditional self-report measures. Experience sampling has been extensively validated (Csikszentmihalyi & Larson, 1987).

In medicine, experience sampling is probably better known as ecological momentary assessment (Stone & Shiffman, 1994). While the term was coined less than two decades ago, ecological momentary assessment has its roots in the development of clinical research diaries in the 1940s, ecological studies of behavior in the 1960s and 1970s, and ambulatory devices for continuous monitoring of cardiovascular activity in the 1980s (for a review see Shiffman, Stone, & Hufford, 2008). Traditionally there were some differences between experience sampling and ecological momentary assessment. Experience sampling was designed to measure people's internal affective states and associated activities at random time points during the day and so data recording has generally been signal-contingent. Ecological momentary assessment has focused more on actual behaviors and has been more likely to include the concurrent measurement of physiological variables such as blood pressure (Kamarck et al., 2002). Additionally, data recording has been more likely to be event-contingent, i.e. participants provide data right after the occurrence of certain events of interest, which may vary depending on the research question. Notably, events of interest may be more likely to occur in some people than in others, and their frequency may also change from day to day within individual persons. For example, when studying affect in relation to cigarette smoking, participants would be asked to record data every time they smoke (Shiffman, 2009).¹ In spite of these historical differences between ‘experience sampling methods’ (ESM) and ‘ecological momentary assessment’ (EMA), the two terms are increasingly used together (Stone and Shiffman, 2002, Trull and Ebner-Priemer, 2009, Wenze and Miller, 2010). We will do the same.

Given the focus on people's internal affective states, activities, and behaviors, it is not surprising that ESM/EMA is increasingly used in mood disorders research. Ebner-Priemer and Trull (2009) reviewed relevant studies in adults with major depressive disorder (MDD), bipolar disorder (BD), or borderline personality disorder. They focused their discussion on data supporting the advantages of ESM/EMA: mood is assessed in real-time, repeatedly, and in everyday life situations. This way recall bias can be avoided, the dynamics of mood can be revealed, and the data can be easily generalized. Mood can be assessed in specific contexts and over the course of treatment so the factors that alter mood can be elucidated and treatment progress can be studied closely. Also, mood can be assessed together with actual behavior as well as with physiological variables. Ebner-Priemer and Trull (2009) did not include studies in pediatric populations in their review. This is unfortunate because studying the everyday lives of youth with mood disorders is valuable given age-related differences in the clinical presentation of these disorders (e.g., Kovacs, 1996, Leibenluft et al., 2003). It is conceivable that mood may be influenced by, and may influence, other aspects of everyday life in children and adolescents than in adults. One of the benefits of ESM/EMA is that the variables under study may be changed to suit the population of interest.

Another review on the use of ESM/EMA in mood disorders research did include studies in pediatric populations (Wenze & Miller, 2010). Like Ebner-Priemer and Trull (2009), the authors reviewed the types of questions relevant to mood disorders that ESM/EMA is especially suitable to answer. Moreover, they discussed the feasibility of using ESM/EMA in patients with MDD or BD and compared the different kinds of data-collection formats used to date (i.e. paper forms versus hand-held computers versus investigator-administered phone interviews). Together, these reviews provide valuable insight into the opportunities ESM/EMA offers to mood disorder researchers and form important additions to reviews focusing on the pros and cons of ESM/EMA in general (e.g., Moskowitz et al., 2009, Shiffman et al., 2008). However, one limitation of these reviews is that they mostly focus on the methods of the selected studies and are therefore more relevant to researchers than to clinicians. A systematic review of the specific results and practical applicability of relevant ESM/EMA studies will be useful for both researchers and clinicians interested in the everyday dynamics of mood disorder symptoms.

Providing this review now seems especially timely for multiple reasons. Firstly, it has been 20 years since the publication of deVries' (1992) book on the use of ESM/EMA in clinical populations. deVries has since been involved in multiple relevant studies in MDD and BD. Over the years ESM/EMA has become much more sophisticated, especially in its data analytic approaches, with promising potential for both research and clinical practice. Secondly, and in line with these rapid developments, there has been a notable boost in the number of publications on ESM/EMA in mood disorder patients. Many of the currently available data have been published in the past few years and have not previously been reviewed. Thus, we feel it is time for a systematic review of the insights gained from ESM/EMA studies in mood disorder patients.

Specifically, and in line with our interest in the clinical insights that are emerging from this growing body of literature, we structure this review using six main topics. Firstly, we focus on the dynamics of mood in symptomatic MDD patients, in other words what contextual factors are associated with patients' affective states? This may provide insight in the potential causes of MDD symptoms and perhaps contribute to the future development of new forms of (more personalized) treatment. Secondly, we focus on the impact of existing forms of treatment on MDD patients' internal affective states, activities, and behaviors. This may provide insight into how exactly depressed patients undergoing treatment improve clinically over time. Thirdly, we focus on the characterization of residual symptoms that may continue to exist in MDD patients even when clinical improvement has been obtained. Mood disorders are considered chronic illnesses and characterizing patients' residual symptoms may provide insight into why many remitted patients relapse; this might in turn contribute to the development of new relapse prevention strategies. Fourthly, we focus on pediatric populations of MDD patients. There are age-related differences in the clinical presentation of MDD (Kovacs, 1996) and ESM/EMA may provide insight in how the everyday lives of children and adolescents with MDD might influence and be influenced by their mood in ways that are unique to this age group. Fifthly, we focus on the differences from MDD and BD that have been revealed by ESM/EMA. This seems particularly relevant in the context of the publication of the fifth version of the Diagnostic and Statistical Manual of Mental Disorders (DSM) in 2013, for which specific categories for “Depressive Disorders” and “Bipolar and Related Disorders” have been proposed. Sixthly and lastly, we focus on the relatively recent inclusion of neurobiological variables (e.g. cortisol) in ESM/EMA studies of mood disorder patients. These types of studies may provide valuable insight into the neuroscience of mood disorder patients' everyday lives.

Before describing the available data for each of these six topics, we clarify the methods used to select the relevant studies and summarize the methodological characteristics of these studies. Afterwards, we briefly mention the advantages and disadvantages of using ESM/EMA for the study of mood disorder patients and then move to discussing the implications for clinical practice and providing multiple suggestions for future research.