The intra-day dynamics of affect, self-esteem, tiredness, and suicidality in Major Depression

Highlights

• The intra-day dynamics of five symptom constructs in Major Depression (MDD) were explored.

• The MDD group had distinct dynamics to controls in affect, self-esteem, suicidality and tiredness.

• Tiredness had decreased variability and all other symptoms increased variability in MDD.

• Increased instability was found in affect, self-esteem and suicidality in MDD.

• Positive affect had a more pronounced and tiredness a less pronounced diurnal time-trend in MDD.

Abstract

Despite growing interest in the temporal dynamics of Major Depressive Disorder (MDD), we know little about the intra-day fluctuations of key symptom constructs. In a study of momentary experience, the Experience Sampling Method captured the within-day dynamics of negative affect, positive affect, self-esteem, passive suicidality, and tiredness across clinical MDD (N= 31) and healthy control groups (N= 33). Ten symptom measures were taken per day over 6 days (N= 2231 observations). Daily dynamics were modeled via intra-day time-trends, variability, and instability in symptoms. MDD participants showed significantly increased variability and instability in negative affect, positive affect, self-esteem, and suicidality. Significantly different time-trends were found in positive affect (increased diurnal variation and an inverted U-shaped pattern in MDD, compared to a positive linear trend in controls) and tiredness (decreased diurnal variation in MDD). In the MDD group only, passive suicidality displayed a negative linear trend and self-esteem displayed a quadratic inverted U trend. MDD and control participants thus showed distinct dynamic profiles in all symptoms measured. As well as the overall severity of symptoms, intra-day dynamics appear to define the experience of MDD symptoms.

Introduction

Recent research has found that dynamics in affect are an integral part of depression (Houben et al., 2015, Pe et al., 2015). This has led to calls for affective dynamics to be included in the diagnostic criteria for Major Depressive Disorder (MDD; Bowen et al., 2013). It would follow that other MDD symptoms are dynamic in nature, and that these dynamics are as important as overall symptom levels or intensity. However, very little research exists on the within-day dynamics of MDD symptoms other than affect. This study aims to obtain a descriptive picture of the daily fluctuations and rhythms of a range of symptom constructs in MDD.

By its very nature, MDD is a dynamic construct: symptoms are known to change over time, with MDD episodes fluctuating across remission, pro-dromal, and more clinically severe periods at different points over months and years (Fried et al., 2016, Iacoviello et al., 2010, Vergunst et al., 2013). Theory also suggests that depressive symptoms fluctuate at a more micro-level: for example, cognitive-behavioral as well as more recent work suggests that depressive symptoms are activated on a moment-to-moment basis over the course of daily life (Beck et al., 1979, Wichers, 2014). Alterations in the circadian rhythms of biological processes such as cortisol and melatonin secretion have also been found to be associated with depression (Peeters et al., 2003, Sundberg et al., 2016). Research and theory on the factors underlying MDD therefore outline inherently dynamic processes that occur within the context of an individual's everyday life. However, to date, research on MDD symptoms has relied mostly on measures and diagnostic conceptualizations of MDD that do not capture the daily flow of MDD experience.

We use the Experience Sampling Method (ESM; Csikszentmihalyi and Larson, 1987; Stone and Shiffman, 1994) to empirically assess the intra-day dynamics of five key MDD symptom dimensions (adapted to the momentary micro-level from macro-level DSM / ICD diagnostic criteria) in a sample of MDD and healthy control participants. At the macro-level, MDD is not a unidimensional construct; rather, it is a syndrome comprised of heterogeneous symptoms that are distinct etiologically and have differential effects on biopsychosocial functioning (Fried et al., 2016, Fried and Nesse, 2014). Indeed, an in-depth symptom-level approach to MDD research has recently been proposed (Fried and Nesse, 2015). Macro-level symptoms are expressions of micro-level, moment-to-moment symptom experience (Wichers, 2014). In the present study, we therefore study micro-level symptoms separately, rather than as part of one construct. Alongside positive affect (PA) and negative affect (NA), the dynamics of self-esteem, tiredness, and suicidality are explored. These symptoms include cognitive and affective symptoms, as well as a somatic, non-affective symptom (tiredness). These symptoms were chosen as the focus of study over others (e.g. appetite changes, psychomotor symptoms, and concentration difficulties) because they have been the focus of previous ESM research. As such, they have all been found to have within-day fluctuations and are known to be subjective experiences that can be studied effectively with ESM.

This study conceptualizes the temporality of symptoms in three ways, as: (1) variability in symptom scores across the sampling period; (2) instability in symptom scores between moments; and (3) systematic trends in hour-by-hour symptom levels. Below, each conceptualization is defined and a brief summary is given of how such measures have been operationalized in relation to affect in MDD.

These concepts are defined comprehensively elsewhere (see Houben et al., 2015 and Jahng et al., 2008). In short, symptom variability captures the overall spread of an individual's symptom scores across a sampling period (i.e. intra-individual variance or standard deviation). Instability is a more temporally contingent measure, capturing the frequency and amplitude of symptom fluctuations between moments. It is the within-person successive difference in between-moment symptom scores.

The extent to which affect fluctuates during the day is a central part of subjective emotional experience. In the past decade, a body of research has found that greater variability and instability in NA are related to poor psychological well-being in general and MDD in particular (Houben et al., 2015, Wichers et al., 2010).

The relationship between PA instability/variability and MDD is less clear. In their meta-analysis, Houben et al. (2015) found that although PA variability/instability was positively associated with poorer psychological health, it was a less powerful predictor of this than NA variability/instability. Studies on clinical populations have typically found no significant association between PA variability/instability and MDD (Peeters et al., 2006, Thompson et al., 2012). Indeed, when the Houben et al. meta-analysis was confined to studies with clinical populations, MDD was actually associated with less variability (though not instability) in PA.

Given that depression is thought to be a dimensional construct (Prisciandaro and Roberts, 2005), it is not clear why studies have found that increased PA instability and variability are associated with depressive symptoms in the general population, but not with depression diagnosis in clinical populations (Gruber et al., 2013, Houben et al., 2015). Indeed, Houben et al. describe their finding that less variability in PA is associated with clinical depression as ‘anomalous’, while also suggesting that less PA variability may reflect the MDD experience of anhedonia and resulting lack of PA reactivity. The research in that review includes studies using different methodologies (e.g. both ESM and retrospective measures), different time-scales between reports (between-day reports and intra-day reports) and different calculations of variability/instability (e.g. multilevel vs. single level variability/instability analyses). Given these marked methodological and measurement differences, a coherent body of research on intra-day, hour-to-hour PA variability/instability in MDD has not yet formed. Indeed, the number of studies that have explored intra-day variability/instability in affect in MDD remains relatively small. As a result, the nature and relative importance of PA and NA variability and instability in MDD is unclear. Further research on clinical samples using fine-grained, multilevel ESM methods is needed to clarify these matters.

Neither variability nor instability directly account for how time itself may influence symptom levels. In contrast, the time-trend approach models symptom severity as a function of time of day. It identifies any systematic patterns of change in symptom levels across the day. Just as research on biological circadian rhythms has shown that certain circadian patterns are associated with disease outcomes (Takahashi et al., 2008), so too may diurnal rhythms in psychological experience play a key role in functioning and outcomes in depression.

Relatively little research exists on diurnal affective rhythms in MDD (indeed, diurnal time-trends were not included in the Houben et al. (2015) meta-analysis of affect dynamics). Traditional clinical descriptions of MDD incorporate a ‘morning-worse’ pattern in affect (Hall et al., 1964, Leibenluft et al., 1992). However, the few studies that have systematically investigated diurnal time-trends in affect have yielded conflicting results.

One ESM study found a ‘morning-worse’ pattern in both NA and PA (Peeters et al., 2006). MDD participants exhibited an inverted U-shaped pattern in NA with a peak in the mid-morning, while the NA of control participants did not exhibit a diurnal trend. PA displayed an inverted U-shaped pattern in both groups. MDD participants, however, showed a significantly steeper slope in PA over the day, with relatively lower morning and higher evening levels. These findings are similar to those of Daly et al. (2011), where individuals with high levels of psychological distress had a more pronounced diurnal affective rhythm than those with low distress levels, characterized by much lower morning PA and somewhat higher morning NA. In contrast, Murray (2007) found a less distinct diurnal quadratic rhythm in PA in individuals with high depression levels compared to those with low depression levels. NA was not found to have a time-trend in either group. Mata et al. (2012) found no differences between MDD and control groups in the time-trend of either NA or PA.

Peeters et al. (2006) is the only study we are aware of that used a clinical MDD sample in its investigation of time-trends. Mata et al. (2012) used a community sample that was diagnosed with/without MDD using structured clinical interviews, while both Murray (2007) and Daly et al. (2011) used non-clinical samples (indeed, while the PA time-trend differed across high and low depression scorers in Murray (2007), mean levels in PA were not significantly different between groups). The nuances of differences in diurnal patterns of affect may only be apparent in individuals with clinically more severe MDD. Further research on clinical samples is therefore needed to establish a consensus on the daily affective time-trend of MDD.

Taken together, research on the intra-day dynamics of affect strongly suggests that dynamic change is a core characteristic of MDD. It emphasizes the need to move beyond static conceptualizations of affect, where an individual's average levels are the only aspect considered. The dynamics revealed by these studies point to increased emotional fluctuations in MDD and difficulties in regulating daily subjective experience. It follows that other key MDD symptoms – such as self-esteem, suicidality and tiredness – may also be characterized by specific dynamic patterns. However, as discussed below, research has continued to focus on absolute levels of these symptoms in MDD and has not yet considered in detail the nature and pattern of their intra-day fluctuations.

Low self-esteem is related to both the aetiology and symptomatology of MDD. The DSM-5 specifies daily feelings of ‘worthlessness’ as part of its MDD diagnostic criteria, the ICD-10 specifically refers to ‘reduced self-esteem and self-confidence’ in MDD (World Health Organization, 2010), and most psychological models propose that MDD is maintained via pervasive negative views of the self (e.g. Beck et al., 1979). The current study uses the concept of ‘self-esteem’ rather than the specific symptom of ‘worthlessness’ to allow a comparison with previous research on self-esteem variability. Items measuring self-esteem tend to be more moderately phrased than those that measure worthlessness, and thus are more likely to detect intra-day variation (Palmier-Claus et al., 2011).

Prospective studies have consistently found that low self-esteem predicts depression in clinical and non-clinical samples (Orth et al., 2008, Rieger et al., 2016, Sowislo and Orth, 2013). A proposed mechanism underlying the relationship between low self-esteem and MDD is self-esteem variability, where relative drops in daily self-esteem trigger and maintain MDD (Kernis et al., 1998). Cross-sectional work has found higher levels of day-to-day variability in self-esteem in individuals currently experiencing depression (Franck and De Raedt, 2007, Sowislo et al., 2014), while prospective studies have found that such day-to-day self-esteem variability predicts future depression (Study 2 – Butler et al., 1994; Franck and De Raedt, 2007; Kernis et al., 1998). However, the strength of the relationship between depression and self-esteem variability is unclear, as the associations found in some of these cross-sectional and prospective studies have been weak or non-significant (Study 1 - Butler et al., 1994; Sowislo et al., 2014).

Previous work on self-esteem variability in MDD has two major limitations. Firstly, a maximum of two self-esteem measures per day have been used to derive self-esteem variability (Sowislo et al., 2014, Kernis et al., 1998). ESM studies on other populations have demonstrated that self-esteem variability can occur over a much shorter time frame, from hour-to-hour (Knowles et al., 2007, Thewissen et al., 2011). No high-frequency ESM study of self-esteem has yet been conducted on an MDD sample. This is a possible reason for the weak or non-significant associations found in some studies: as self-esteem likely fluctuates at a faster rate than day-to-day, studies with low sampling rates (i.e. one or two measurements per day) may not reliably pick up variability in this symptom.

A second limitation is that previous research has focused on a single dimension of self-esteem dynamics: intra-individual variability (standard deviation) across day-to-day measures. To our knowledge, no research has yet directly investigated the extent of between-moment instability in self-esteem in MDD; rather, previous work has used day-to-day variability as a proxy for self-esteem instability. Furthermore, research has not yet explored diurnal time-trends in self-esteem in healthy or MDD populations. The current study thus aims to improve upon previous literature by (1) using a high-frequency ESM protocol and (2) studying multiple aspects of self-esteem dynamics to provide a more nuanced account of self-esteem fluctuations in MDD.

Understanding the dynamics of suicidal ideation is of public health importance. Lability in suicidality may indicate difficulties in regulating intrusive suicidal thoughts (indicating suicide risk), while determining suicidality's diurnal time-course would highlight micro-periods of increased risk for suicide. A small number of previous studies suggest that micro-level variation is a feature of suicidality. Day-to-day instability in suicidal ideation has been found to have a stronger association with previous suicide attempts than duration or intensity of ideation (Witte et al., 2005, Witte et al., 2006). Furthermore, suicidal behavior has been found to follow a diurnal course: research on Italian suicide statistics, where time of death is routinely reported, has found that most suicides take place in the morning-time (Preti and Miotto, 2001, Williams and Tansella, 1987).

Little evidence exists on the intra-day dynamics of subjective suicidality. Two studies have investigated the diurnal time-trend of subjective suicidality (Nock et al., 2009, Husky et al., 2014). Contrary to work on suicidal acts, neither found suicidality to have a significant time-trend. To our knowledge, no study has yet used ESM to explore intra-day variability or instability in suicidality.

Previous studies have been limited in their ability to detect within-day variation in suicidality due to low sampling frequencies (two signals emitted per day in Nock et al. (2009) and five in Husky et al. (2014)). They also investigated only one aspect of suicidality: active suicidal thoughts. These thoughts were found to be relatively rare (7.8% of reports in Husky et al. (2014) and an intra-individual average of 1.1 suicidal reports per week in Nock et al. (2009)). Other dimensions of suicidality may be more commonly experienced in daily life. Passive suicidality deals with vague thoughts of suicide, such as thoughts of death and feelings that life is not worth living. Having ‘thoughts of death’ is itself part of the DSM-5 criteria for MDD (American Psychiatric Association, 2013) and such passive suicidality is associated both with active suicidal ideation and with suicidal acts themselves (Steer et al., 1993). A high-frequency intra-day measure of passive suicidal thoughts may therefore allow for a more nuanced measurement of suicidality in MDD.

Tiredness is the most prevalent symptom of MDD, with 78% of patients reporting at least moderate daily levels (Vaccarino et al., 2008). In healthy individuals, subjective tiredness has a distinctive diurnal time-trend: a V-shape pattern, where levels decline from morning to mid-day (where they reach a trough), and then rise steadily until reaching a peak in the late evening time (Dockray et al., 2010, Stone et al., 1996, Stone et al., 2006). However, no study has yet investigated the diurnal dynamics of tiredness in MDD. Given that the diurnal pattern in tiredness appears to be an integral part of healthy experience, understanding how the pattern differs in MDD may provide a new insight into the nature of this central symptom.

Little work has investigated the intra-day dynamics of MDD symptoms other than affect. Furthermore, work on affect dynamics has largely focused on instability and variability, neglecting the importance of diurnal time-trends in experience. In this study we aimed to extend prior work by providing a fine-grained ESM account of the within-day time-trends, instability, and variability of affect, self-esteem, passive suicidality, and tiredness in MDD. We examined multiple facets of the intra-day dynamics of affective, cognitive, and somatic symptoms in tandem in order to improve our overall understanding of the daily experience of MDD by providing (1) a phenomenological insight into how symptoms are experienced across the day in MDD and (2) an understanding of the differences between psychologically ‘healthy’ and ‘unhealthy’ patterns of experience, and as such potential targets for treatment.

Given the lack of previous research, much of this study is exploratory. For example, we had no specific hypotheses in relation to self-esteem time-trends or tiredness dynamics in MDD. Nevertheless, based on the evidence that has been discussed, we hypothesized the following:

• NA, self-esteem, and passive suicidality would exhibit significantly higher levels of variability and instability in MDD.

• Affect and passive suicidality would exhibit significantly more pronounced ‘morning-worse’ diurnal time-trends in MDD.