An emotional processing writing intervention and heart rate variability: the role of emotional approach

ABSTRACT Expressing and understanding one's own emotional responses to negative events, particularly those that challenge the attainment of important life goals, is thought to confer physiological benefit. Individual preferences and/or abilities in approaching emotions might condition the efficacy of interventions designed to encourage written emotional processing (EP). This study examines the physiological impact (as indexed by heart rate variability (HRV)) of an emotional processing writing (EPW) task as well as the moderating influence of a dispositional preference for coping through emotional approach (EP and emotional expression (EE)), in response to a laboratory stress task designed to challenge an important life goal. Participants (n = 98) were randomly assigned to either EPW or fact control writing (FCW) following the stress task. Regression analyses revealed a significant dispositional EP by condition interaction, such that high EP participants in the EPW condition demonstrated higher HRV after writing compared to low EP participants. No significant main effects of condition or EE coping were observed. These findings suggest that EPW interventions may be best suited for those with preference or ability to process emotions related to a stressor or might require adaptation for those who less often cope through emotional approach.

Emotional approach; heart rate variability; emotional processing; emotional expression; expressive writing Functionalist theories suggest that attending to emotions underlies effective self-regulation (Carver & Scheier, 1998;Frijda, 1986). Expressing and processing emotion enables "emotional disturbances [to be] absorbed and decline to the extent that other experiences and behavior can proceed without disruption" (Rachman, 1980, p. 51). Emotional approach coping involves expressing, identifying, acknowledging, and understanding emotional responses to stressful or challenging circumstances (Stanton, Kirk, Cameron, & Danoff-Burg, 2000). There is increasing interest in how emotional approach can influence physical health and behaviour (DeSteno, Gross, & Kubzansky, 2013), as empirical work supports a salutary role for processing and expressing emotions in adjustment to stressors (e.g. Pennebaker, 2012;Stanton et al., 2000). mechanism by which emotional approach influences physical and mental health.
Heart rate variability (HRV) is the degree of fluctuation in the length of intervals between heartbeats that reflects the continuous interplay between sympathetic and parasympathetic influences on heart rate that yields information about autonomic flexibility and thereby represents the capacity for regulated emotional responding (Thayer, Åhs, Fredrikson, Sollers, & Wager, 2012). HRV is a widely used measure of stress reactivity that is associated with greater ability to respond dynamically to environmental challenges and even signals better emotion regulation (Appelhans & Luecken, 2006;Shaffer, McCraty, & Zerr, 2014). HRV is thought to reflect the functional integrity of a network linking the heart to brain regions involved in coordinating contextual cues, threat and safety appraisals, and behaviours (Thayer et al., 2012).
Reduced HRV has also been associated with misguided attempts at emotional regulation, such as being linked to patterned worry underpinning anxiety disorders (e.g. Aldao, Mennin, & McLaughlin, 2013). In fact, unconstructive forms of EP have been linked to chronically reduced HRV which in turn is linked to numerous adverse physical and mental health outcomes, including anxiety disorders, chronic inflammation, and cardiovascular disease (Shaffer et al., 2014). Thus, managing emotions in a way that promotes physiological habituation and lightens allostatic load likely has powerful implications for health and well-being.
Despite the potential benefits of EP, interventions designed to promote EP have mixed results (Frattaroli, 2006). It may be that the success of EP is sensitive to the fit between the characteristics of the stressor and the individual (Austenfeld & Stanton, 2004;Lazarus, 1996), in that interventions might confer optimal benefit when matched to an individual's preferred coping style (e.g. Engebretson, Matthews, & Scheier, 1989). For instance, Stanton et al. (2000) observed that individuals who engaged in expressive writing demonstrated greater reductions in negative affect and arousal if they also reported dispositional tendencies towards emotionally expressive coping. Furthermore, a more recent randomised control trial suggested that expressive writing may actually be contraindicated for individuals who do not habitually express emotions (Niles, Haltom, Mulvenna, Lieberman, & Stanton, 2014). Yet other studies show that greater alexithymia and/or ambivalence about emotional expression (EE) predicts greater subjective and physiological benefits (e.g. Lu & Stanton, 2010). Given the potential role of EP in behavioural adaptation and physiological habituation to stressors, we sought to examine how dispositional EP affects physiological response to an EP writing exercise.
In the present study, we recruited undergraduate students for whom attending graduate, law, or medical school was an important personal goal. Participants were interviewed by a supposed "expert in graduate admissions", who provided challenging feedback regarding their potential as an applicant. Participants were then randomly assigned to complete either an EP or neutral writing task. We predicted that participants assigned to EP writing would demonstrate greater physiological benefit from the intervention (as indexed by higher HRV) if they also endorsed a greater preference for coping via emotional approach (EE and EP).

Method
In addition to a full description of the utilised study methods, below we report how we determined our sample size, all data exclusions (if any), all manipulations, and all measures in the study.

Participants
Participants (n = 98) were university undergraduates recruited from a student subject pool who were fluent in English and reported a moderate-to-strong intent to apply to medical, law, or other doctorallevel studies. This was determined by a single-item screening question: "How important to you is applying to medical school, law school, or a Ph.D. program upon completion of your undergraduate studies?" rated on a 5-point scale ranging from 0 (Not at all important) to 4 (Extremely important). Only those answering 3 or 4 were invited to participate.
Participants were mostly female (n = 64), with a mean age of 20.02 years (SD = 2.58). Approximately 46% were upper-class students and the majority (75%) identified as non-White ethnic minorities (Table 1). Sample size estimates were determined by G*Power (Faul, Erdfelder, Lang, & Buchner, 2007), in which a sample size of 89 participants was estimated to achieve .80 power.

EE and EP
EE and EP were measured by the Emotional Approach Coping scales (Stanton et al., 2000). Higher scores on each 4-item subscale indicate greater dispositional preference for that form of coping in relation to stressful occurrences. Items were rated in response to how they usually respond to stressors on a 4-point response scale (1 = I don't do this at all; 4 = I do this a lot). Cronbach's alpha was .77 for EP and .90 for EE. EE and EP were positively correlated (r = .61, p < .001).

Heart rate variability
HRV was calculated as the standard deviation of normal to normal (SDNN) interbeat intervals. SDNN has been recommended as a good measure of overall variability in heart rate (e.g. Allen, Chambers, & Towers, 2007). Decreased HRV can reflect either parasympathetic withdrawal or "a saturatingly high level of sympathetic input" (Task of the European Society of Cardiology, 1996) and SDNN summarises total variability as a function of both parasympathetic and sympathetic influences. Whether HRV analyses should account for concurrent respiration has been a subject of debate (Allen et al., 2007). However, during resting state recordings when respiration is not expected to vary significantly, controlling for respiration rate is likely unnecessary. SDNN is also less subject to respiratory influence than other measures of HRV (Allen et al., 2007).
The electrocardiogram (EKG) was recorded using two Ag/AgCl electrodes (EL503; Biopac Systems, Inc., Goleta, California, USA), connected to an ECG100C amplifier for the MP36R System (Biopac Systems, Inc.). Electrodes were placed in a modified lead-II configuration. The EKG was acquired continuously in Biopac Student Lab (Biopac Systems, Inc.) and digitised at a sampling rate of 1 kHz, with high-and low-pass filters at 0.5 and 35 Hz.
EKG data during the baseline and post-writing period were extracted from the continuous recording and exported to QRSTool and CMetX software programs (Allen et al., 2007; http://psychofizz.org). QRSTool interpolates the series of interbeat intervals (IBIs) in a graphical user interface for semi-automatic R-peak detection and cleaning. CMetX calculates timedomain metrics of HRV from the cleaned IBI series. Both QRSTool and CMetX have shown good field validity (Hibbert, Weinberg, & Klonsky, 2012). In order to minimise the likelihood of artefact contamination recordings with excessive motion or poor electrode contact were excluded from analysis. Data were manually inspected and removed if HRV values exceeded three standard deviations from the mean and CMetX identified artefacts in the cleaned IBI series (when the difference between successive IBIs falls under a threshold of 300 ms). Six samples were excluded due to artefacts. One individual's data were removed due to the presence of ectopic beats, and the remainder excluded due to artefacts from movement or poor electrode contact (these tend to co-occur).

Sociodemographic and biobehavioural factors
Participants self-reported social and demographic variables (e.g. age, class year, and ethnicity) and relevant biobehavioural factors (e.g. medication use, recent caffeine intake, and cigarette use).

Procedures
Following informed consent, participants completed self-report measures and a 10-minute resting physiological baseline. They were introduced to an "authority on graduate admissions" who conducted a uniform, 10-minute interview regarding participants' plans and preparedness for graduate study, while providing standardised feedback intentionally highlighting their weaknesses and a general high bar for admission. A guided imagery exercise followed, in which participants vividly imagined receiving an ultimate admissions rejection.
Participants were then randomised into one of two writing conditions using writing instructions adapted from studies of expressive writing (Pennebaker & Beall, 1986): (1) emotional processing writing (EPW) condition directed participants to identify and attempt to understand any emotions elicited during the tasks, and to consider the meaning of their current affective state; and (2) fact control writing (FCW) participants were instructed to describe a factual account of activities performed in the prior 24-hour period with little to no emotional content. Instructions in both conditions were to write continuously for 10 minutes without concern for grammar, spelling, or punctuation. After writing, participants underwent a second 10-minute physiological recording (the "post-writing" period). During physiological recording (baseline and post-writing), participants were seated upright, asked to make as few movements as possible, and were left alone in the testing room. There was an approximate one-minute lead in time at each recording. Participants were fully debriefed at the close of the study. Participants were compensated with course credit. All study procedures were approved by the university Institutional Review Board.

Sample characteristics at baseline
Participants in the two conditions did not differ significantly on age, race/ethnicity, or year in school (Table 1)

Manipulation check
At the conclusion of the experimental session, participants were asked to rate the extent to which they experienced "real" emotions about their graduate school goals during the study on a scale from 1 (no real emotions) to 7 (very real emotions). Generally, participants reported experiencing moderately real emotions, M = 4.17 and SD = 1.43. Ratings did not differ significantly between the FCW (M = 4.11, SD = 1.62) and EPW (M = 4.22, SD = 1.25) conditions [t(96) = −.34, p = .72]. All writing samples were independently coded by two raters unaware of study assignment to identify condition, and accuracy was 100%.
To consider the possibility that EP inadvertently occurred in the FCW group, all essays were coded by a team of four independent coders with adequate interrater reliability using a coding rubric for evidence of both constructive and unconstructive forms of EP (Hoyt, Austenfeld, & Stanton, in press). All writing in the control condition had negligible evidence for EP.

Heart rate variability
Neither baseline nor post-writing HRV was significantly associated with age, cigarette use, caffeine intake, gender, or medication status. Baseline HRV was significantly correlated with EE (r = .27, p < .01), but not EP (r = .09, ns). However, post-writing HRV was significantly correlated with both EE (r = .21, p < .05) and EP (r = .20, p < .05). We regressed post-writing HRV on condition, EP, and their interaction by multiplying the dummycoded condition variable (FCW = 0, EPW = 1) with mean-centred EP scores using methods outlined by Aiken and West (1991) for categorical by continuous variable interactions. The model [F(3,97) = 3.42, p Figure 1. Association between HRV and emotional processing in emotional processing and FCW conditions. Notes: FCW, fact-control writing condition; EPW, emotional processing writing condition; HRV, heart rate variability (in standard deviation of interbeat intervals). "Low" and "High" labels on the graph depict directionality of the continuous scale.
= .02, R 2 = .098] was on the border of significance for an effect of condition (B = .62, p = .06) and no significant main effect of EP was observed (B = −.19, p = .27). However, the EP × condition interaction was significant (B = .90, p = .02). For individuals in the EPW condition, EP was positively related to HRV (B = .21, p = .04); however, the simple effect was not significant for FCW (B = .17, ns) (Figure 1).
In post hoc analyses, models were re-examined when controlling for baseline HRV. However, largely due to strong correlations between baseline and post-writing HRV, no significant effects were detected.

Discussion
Identifying and understanding one's emotional responses to negative events, particularly those that challenge attainment of important life goals, is thought to confer psychological and physiological benefit for some people. While examining the physiological impact of an EP writing task in response to a laboratory stressor designed to challenge an important life goal, the present study identifies dispositional preference for coping through EP as an important individual difference that influences the intervention impact. Importantly, condition alone did not predict post-writing HRV, but participants with higher dispositional EP exhibited higher HRV following EP writing.
EP interventions, particularly those that rely on written expression and exploration of emotion, may be most beneficial when matched with individual differences in coping and emotional style and abilities (Engebretson et al., 1989;Stanton et al., 2000). Identification of "for whom" such interventions are beneficial helps to clarify inconsistent results across studies. However, the question of whether differences in preference for EP versus ability of EP accounts for results remains. Notably, we did not find an effect for coping by EE. Also, the EP by condition interaction explained only about 4% of the variance in HRV; larger effects might be observed on self-reported outcomes (e.g. Tamagawa, Moss-Morris, Martin, Robinson, & Booth, 2013).
Our findings add to the growing literature that suggests dispositional factors are important moderators to effects of written emotion-focused interventions. However, no previous study has specifically focused on the processing of emotion. The finding that EP, but not EE, interacted with condition highlights the importance of individual and intervention match. Previous findings that have found a moderating role for EE have utilised a writing intervention designed to enhance expression (and not processing) of emotion (Austenfeld, Paolo, & Stanton, 2006) and participants demonstrated both constructive and unconstructive EP when instructed to merely express (Hoyt, Austenfeld, & Stanton, in press). Unconstructive processing attempts might thwart resolution of stress and maintain or amplify negative responses, including goal failure (Brosschot et al., 2006;Jones, Papadakis, Orr, & Strauman, 2013). Future work should distinguish EP forms and other potential individual difference factors (e.g. depressive history, neuroticism, and avoidance) should be explored.
The bivariate correlations suggested a positive, though modest, relationship between EE and HRV at baseline. The law of initial values (LIV) (Wilder, 1967) suggests that participants who exhibit the lowest (or highest) baseline values would exhibit the most (or least) amount of change to a stimulus. Thus, it may be that those higher in EE exhibit a more consistent pattern of higher HRV and, therefore, have less opportunity for physiologic gains from expressive writing. Jin (1992) offered an extension of the LIV and suggested that reversed responses might occur when initial values reach an upper extremity. However, no known "upper extremity" for SDNN has been determined. Past reviews of normative HRV values suggest a mean SDNN of 50 (Nunan, Sandercock, & Brodie, 2010); which is significantly lower than observed baseline values in our study. As Jin notes, systematic group differences in initial values are generally not expected with random assignment. Thus, controlling for baseline values in a randomised experiment alters the research focus to potential differences if subjects were at the same basal level. Nevertheless, to the extent that EE involves recognising and labelling emotions, this pattern of association is consistent with prior work demonstrating a relationship between emotion recognition and HRV in both psychiatric and non-psychiatric populations (Bal et al., 2010;Quintana, Guastella, Outhred, Hickie, & Kemp, 2012). However, the psychological benefit from expressive interventions may be more pronounced and immediate with less necessary physiological concurrent change for those higher on EE.
As reported, no effects were detected when baseline HRV was controlled. Additional research is needed to better characterise the potential of expressive writing to alter HRV patterns. Although no baseline differences in HRV or other variables of interest were observed between groups, it remains possible that observed results are accounted for by an unknown third variable. Furthermore, in the absence of longitudinal data, the long-term impact of EP writing cannot be determined. Distinguishing immediate and distal consequences on health and physiology is warranted. In the present study, we used a control condition designed to distract (or turn attention away) from emotions, assuming that distraction would obstruct EP. It should be considered that distraction can be effective at reducing negative emotions in the short term (Uusberg, Thiruchselvam, & Gross, 2014). This study also utilised a novel stress task in a controlled setting. These relationships should be tested in clinical populations and samples more diverse in age, education, and stress-related characteristics.
We observed that participants who endorsed greater tendency to engage in EP demonstrated greater physiological habituation following an EP intervention for a personally relevant stressor, whereas those with lower dispositional EP had lower HRV overall but showed little difference by condition. It is important to interpret these findings in the context of the short duration of elapsed time. It is also possible that the EPW was more stressful for people lower in EP but could yield long-term benefits. Regardless, these findings underscore the influence of important individual differences, which can moderate the benefits of emotion-focused interventions. Future EP writing interventions might attempt to match individual preferences or skills to maximise utility and efficacy or might require adaptation for those who less often cope through emotional approach.