Statistical Conceptualisation of Mood Instability: A Systematic Review

Abstract

Background/Objectives: Our understanding of mood instability as a clinically important feature of many psychiatric conditions has been increasing over the last decade, but there remains a lack of clarity around the optimal ways to calculate mood instability in real time. We conducted a systematic review in order to describe the statistical methods used in studies investigating mood instability that collected mood data using ESM (Experience Sampling Methodology). Results: From a total of 229 papers, we found 15 discrete statistical methods were used a total of 319 times. In 76 (33%) studies, more than one statistical method was used, and 39 (17%) studies employed distinct statistical methods for particular aspects of affect dynamics. Conclusions: Based on our findings, we recommend standardisation of statistical methods to strengthen future research on mood instability and ultimately support better clinical outcomes for individuals with mood difficulties.

Keywords: affective variability; experience sampling; mood instability.

Figure 1

Diagram to show weather patterns on two students’ summer holidays, based on Ebner and colleagues [12]. The top row shows student A’s weather: sunshine the first week and clouds and rain the second week. The bottom row shows Student B’s week of weather: intermittently sunny and rainy.

Figure 2

PRISMA flow diagram to show our process from record identification to study inclusion.

Figure 3

Chart to show use of statistical methods used to conceptualise mood instability across all 229 included studies.

Figure 4

Statistical methods used to describe mood instability over time. Between 1973 and 2006, there were 2 or fewer studies per year, and they predominantly used standard deviation to describe mood instability.

Figure 5

Word cloud showing the relative use of terminologies to describe mood instability. All mood constructs of interest found across the included studies are shown in various sizes depicting how frequently the constructs were used. Variability being the most commonly used, it appears the largest.

Figure 6

Bar chart to show statistical methods used to calculate specific affect dynamics (variability, instability, inertia, and extreme change). The chart shows more than one method was used to analyse the same construct of mood. For example, instability was analysed by both mean squared successive difference (MSSD) and probability of acute change (PAC) with different frequencies.