Objectively assessed outcomes

Expressive writing results in significant improvements in longer-term physical health outcomes such as illness-related visits to the doctor (Reference Pennebaker and BeallPennebaker & Beall, 1986; Reference Pennebaker, Kiecolt-Glaser and GlaserPennebaker et al, 1988; Reference Pennebaker and FrancisPennebaker & Francis, 1996; Reference King and MinerKing & Miner, 2000), blood pressure (Reference Davidson, Schwartz, Sheffield, Lepore and SmythDavidson et al, 2002, citing Crow et al), lung function (Reference Smyth, Stone and HurewitzSmyth et al, 1999), liver function (Reference Francis and PennebakerFrancis & Pennebaker, 1992) and number of days in hospital (Reference Norman, Lumley and DooleyNorman et al, 2004). Expressive writing has also produced significant benefits in a number of measures of immune system functioning (Reference Pennebaker, Kiecolt-Glaser and GlaserPennebaker et al, 1988; Reference Esterling, Antoni and FletcherEsterling et al, 1994; Reference Booth, Petrie and PennebakerBooth et al, 1997; Reference Petrie, Booth and PennebakerPetrie et al, 1995, Reference Petrie, Fontanilla and Thomas2004).

Significant benefits have also been found for such objective outcomes as students’ grade point average (Reference Pennebaker and FrancisPennebaker & Francis, 1996; Reference Cameron and NichollsCameron & Nicholls, 1998), absenteeism from work (Reference Francis and PennebakerFrancis & Pennebaker, 1992), re-employment after job loss (Reference Spera, Buhrfeind and PennebakerSpera et al, 1994), working memory (Reference Klein and BoalsKlein & Boals, 2001) and sporting performance (Reference Scott, Robare and RainesScott et al, 2003). In addition, writing about emotional topics changed the way that participants interacted with others, suggesting that writing may also have an impact on objectively assessed social and linguistic behaviour (Reference Pennebaker and GraybealPennebaker & Graybeal, 2001).

Self-reported physical health outcomes

Expressive writing also produces longer-term benefits in self-reported health outcomes such as visits to the doctor (Reference Cameron and NichollsCameron & Nicholls, 1998), physical symptoms (Reference Park and BlumbergPark & Blumberg, 2002) and number of days out of role because of illness (Reference Pennebaker and BeallPennebaker & Beall, 1986; Reference Smyth, True and SoutoSmyth et al, 2001).

In general, expressive writing does not affect health-related behaviours such as exercise, diet or drug/alcohol use (Reference Pennebaker, Kiecolt-Glaser and GlaserPennebaker et al, 1988).

Self-reported emotional health outcomes

Some studies have also found longer-term benefits of expressive writing for emotional health outcomes, including mood/affect (Reference Pennebaker, Kiecolt-Glaser and GlaserPennebaker et al, 1988; Reference Páez, Velasco and GonzalezPáez et al, 1999), psychological well-being (Reference Park and BlumbergPark & Blumberg, 2002), depressive symptoms before examinations (Reference LeporeLepore, 1997) and post-traumatic intrusion and avoidance symptoms (Reference Klein and BoalsKlein & Boals, 2001). However, the findings for emotional health are not as robust or as consistent as those for physical health.

Meta-analyses

A meta-analysis of 13 studies using expressive writing with healthy participants (Reference SmythSmyth, 1998) found a significant overall benefit (d = 0.47, P<0.0001) and specific benefits in objective or self-reported physical health, psychological well-being, physiological functioning and general functioning outcomes. Smyth's review suggests that, for physically and psychologically healthy individuals, the effects produced by expressive writing are substantial and similar in magnitude to the effects of other psychological interventions, many of which are more involved, time-consuming and expensive.

In clinical populations, a meta-analysis (Reference Frisina, Borod and LeporeFrisina et al, 2004) of nine expressive writing studies also found a significant benefit for health (d = 0.19, P<0.05), although when analysed separately the effects for physical health outcomes in medically ill populations were significant (d = 0.21, P = 0.01) but those for psychological health outcomes in psychiatric populations were not (d = 0.07, P = 0.17). Although the benefits are more modest than in studies with healthy participants, this meta-analysis suggests that expressive writing nevertheless has positive effects in clinical populations.

Medical conditions

In comparisons with controls, expressive writing produced significant benefits for individuals with a variety of medical problems (Box 3). Study participants with asthma or rheumatoid arthritis showed improvements in lung function and physician-rated disease severity respectively, following a laboratory-based writing progamme (Reference Smyth, Stone and HurewitzSmyth et al, 1999), although people with rheumatoid arthritis using a home-based videotaped programme showed no benefit (Reference Broderick, Stone and SmythBroderick et al, 2004). Some studies found that patients with cancer reported benefits such as better physical health, reduced pain and reduced need to use healthcare services (Reference Rosenberg, Rosenberg and ErnstoffRosenberg et al, 2002; Reference Stanton, Danoff-Burg, Lepore and SmythStanton & Danoff-Burg, 2002), although others failed to find any benefits (Reference Walker, Nail and CroyleWalker et al, 1999; Reference de Moor, Sterner and Hallde Moor et al, 2002). Patients with HIV infection showed improved immune response similar to that seen in mono-therapy with anti-HIV drugs (Reference Petrie, Fontanilla and ThomasPetrie et al, 2004) and individuals with cystic fibrosis showed a significant reduction in hospital-days over a 3-month period (Reference Taylor, Wallander and AndersonTaylor et al, 2003). Women with chronic pelvic pain reported reductions in pain intensity ratings (Reference Norman, Lumley and DooleyNorman et al, 2004) and poor sleepers reported shorter sleep-onset latency (Reference Harvey and FarrellHarvey & Farrell, 2003). Benefits have also been found for post-operative course after papilloma resection (Reference Solano, Donati and PecciSolano et al, 2003) and for primary care patients (Reference Klapow, Schmidt and TaylorKlapow et al, 2001; Reference Gidron, Duncan and LazarGidron et al, 2002).

Psychological conditions

Other studies have investigated expressive writing in preselected groups of trauma survivors and individuals with specific psychological difficulties, with mixed results. Students with a trauma history have shown improvements in physical health (Reference Greenberg, Wortman and StoneGreenberg et al, 1996; Reference Sloan and MarxSloan & Marx, 2004a ), post-traumatic stress disorder (PTSD) symptomatology and other aspects of psychological health (Reference Schoutrop, Lange, Brosschot, Vingerhoets, van Bussel and BoelhowerSchoutrop et al, 1997, Reference Schoutrop, Lange and Hanewald2002; Reference Sloan and MarxSloan & Marx, 2004a ), although not all studies find benefits (Reference Deters and RangeDeters & Range, 2003).

Limited benefits were obtained for male psychiatric prison inmates (Reference Richards, Beal and SeagalRichards et al, 2000), victims of natural disaster (Reference Smyth, Hockemeyer and AndersonSmyth et al, 2002) and individuals who had experienced a recent relationship breakup (Reference Lepore and GreenbergLepore & Greenberg, 2002).

Expressive writing was beneficial, but not significantly more so than control writing, for females writing about body image (Reference Earnhardt, Martz and BallardEarnhardt et al, 2002), children of alcoholics (Reference Gallant and LafreniereGallant & Lafreniere, 2003), caregivers of children with chronic illness (Reference Schwartz and DrotarSchwartz & Drotar, 2004), students screened for suicidality (Reference Kovac and RangeKovac & Range, 2002) and individuals who had experienced a bereavement (Reference Range, Kovac and MarionRange et al, 2000; Reference O'Connor, Nikoletti and KristjansonO'Connor et al, 2003).

Compared with controls, expressive writing was detrimental for adult survivors of childhood abuse (Reference Batten, Follette and Rasmussen HallBatten et al, 2002) and for a small sample of eight Vietnam veterans with PTSD (Reference Gidron, Peri and ConnollyGidron et al, 1996).

Our review of the literature shows that psychological health benefits tend to be more often found when participants’ traumas and/or symptoms are clinically more severe, although results are inconsistent. One explanation for this inconsistency may be that many of the studies with null findings instructed participants to write about the specific traumatic event they were selected for, rather than using the standard instructions (Box 1), which allow them to write about events of their choosing. In studies where expressive writing was beneficial, many participants wrote about topics other than their particular physical illness or psychological problem, but still showed improvements in that area (Reference Smyth, Pennebaker and SnyderSmyth & Pennebaker, 1999).

Individual differences

In addition to studying specific health populations, researchers have explored various individual difference indices to identify those subgroups for whom expressive writing is most beneficial. Results have been inconsistent. Variables generally found to be unrelated to outcome include age, trauma severity, baseline physical and psychological health levels, negative affectivity and measures of inhibition and prior disclosure.

Reference SmythSmyth's (1998) meta-analysis found that the effects were greater for males than for females. Expressive writing is more beneficial for those high in alexithymia (Reference Páez, Velasco and GonzalezPáez et al, 1999; Reference BaikieBaikie, 2003; Reference Solano, Donati and PecciSolano et al, 2003) and high in splitting (Reference BaikieBaikie, 2003), characteristics often seen in patients with psychosomatic disorders and borderline personality disorder respectively, suggesting potential for the use of expressive writing in these populations.

Conclusion

Overall, studies examining expressive writing demonstrate some beneficial effects in physical and/or psychological health. Although the empirical findings are at times equivocal and further research is required to clarify populations for whom writing is clearly effective, there is sufficient evidence for clinicians to begin applying expressive writing in therapeutic settings with caution. Indeed, Reference SpiegelSpiegel (1999) noted that a drug intervention reporting medium effect sizes similar to those found for expressive writing (Reference SmythSmyth, 1998) would be regarded as a major medical advance.

Emotional catharsis

There is little support for the initial hypothesis that expressive writing operates through a process of emotional catharsis or venting of negative feelings. Writing only about the emotions associated with a trauma is not as beneficial as writing about both the event and the emotions (Reference Pennebaker and BeallPennebaker & Beall, 1986). Furthermore, expressive writing results in immediate increase in negative affect rather than immediate relief of emotional tension, and the obtained health benefits are unrelated to the amount of negative emotion or distress either expressed or reported just after writing (Reference SmythSmyth, 1998).

Emotional inhibition and confrontation

Reference PennebakerPennebaker's (1985) theory proposed that actively inhibiting thoughts and feelings about traumatic events requires effort, serves as a cumulative stressor on the body and is associated with increased physiological activity, obsessive thinking or ruminating about the event, and longer-term disease. Confronting a trauma through talking or writing about it and acknowledging the associated emotions is thought to reduce the physiological work of inhibition, gradually lowering the overall stress on the body. Such confrontation involves translating the event into words, enabling cognitive integration and understanding of it, which further contribute to the reduction in physiological activity associated with inhibition and ruminations (Reference PennebakerPennebaker, 1985).

This theory has intuitive appeal but mixed empirical support. Studies have shown that expressive writing results in significant improvements in various biochemical markers of physical and immune functioning (Reference Pennebaker, Kiecolt-Glaser and GlaserPennebaker et al, 1988; Reference Esterling, Antoni and FletcherEsterling et al, 1994; Reference Petrie, Booth and PennebakerPetrie et al, 1995; Reference Booth, Petrie and PennebakerBooth et al, 1997). This suggests that written disclosure may reduce the physiological stress on the body caused by inhibition, although it does not necessarily mean that disinhibition is the causal mechanism underlying these biological effects. On the other hand, participants writing about previously undisclosed traumas showed no differences in health outcomes from those writing about previously disclosed traumas (Reference Greenberg and StoneGreenberg & Stone, 1992) and participants writing about imaginary traumas that they had not actually experienced, and therefore could not have inhibited, also demonstrated significant improvements in physical health (Reference Greenberg, Wortman and StoneGreenberg et al, 1996). Therefore, although inhibition may play a part, the observed benefits of writing are not entirely due to reductions in inhibition.

Development of a coherent narrative

A computerised text analysis system, Linguistic Inquiry and Word Count (LIWC; Reference Pennebaker, Francis and BoothPennebaker et al, 2001), was specifically designed to determine whether certain linguistic markers might be associated with improvements in health. The LIWC program analyses the writing tasks by calculating the percentage of words in the text matching each of 82 predefined language categories. The most consistent finding has been that, over the course of writing, participants whose health improved used more positive-emotion words, a moderate number of negative-emotion words and an increased number of ‘cognitive mechanism’ words (the latter include insight words such as understand, realise and causal words such as because, reason) (Reference PennebakerPennebaker, 1997b ).

The subsequent suggestion that the beneficial effect of expressive writing is the development of a coherent narrative over time, reflecting increasing cognitive processing of the experience, is consistent with the literature on traumatic memory and trauma treatment (e.g. Reference Van der Kolk, McFarlane and Weisaethvan der Kolk et al, 1996). In addition, recent linguistic studies have shown that session-to-session variations in pronoun use are related to health improvements, which may reflect a transformation in the way people think about themselves in relation to others and the world (Reference PennebakerPennebaker, 2002).

Cognitive processing

More direct investigations of cognitive processing as a potential mechanism suggest that writing may help the writer to organise and structure the traumatic memory, resulting in more adaptive, integrated schemas about self, others and the world (Reference Harber, Pennebaker and ChristiansonHarber & Pennebaker, 1992).

Although the cognitive processing hypothesis has been difficult to evaluate empirically owing to the difficulty of measuring cognitive changes, there is evidence that narrative formation and coherence are necessary for expressive writing to be beneficial (Reference Smyth, True and SoutoSmyth et al, 2001) and that expressive writing increases working memory capacity, which may reflect improved cognitive processing (Reference Klein and BoalsKlein & Boals, 2001).

Exposure

The effectiveness of prolonged exposure as a treatment for post-traumatic stress (Reference Foa and RothbaumFoa & Rothbaum, 1998) led to the suggestion that the writing paradigm may produce extinction of negative emotional responses through repeated writing about traumatic memories (Reference Lepore, Greenberg, Bruno, Lepore and SmythLepore et al, 2002). This hypothesis has met with mixed support. There is some evidence that exposure may underlie the emotional health benefits of expressive writing (Reference Sloan and MarxSloan & Marx, 2004a ; Reference Sloan, Marx and EpsteinSloan et al, 2005). However, many participants benefit from writing regardless of whether they write about the same traumatic experience or different experiences at each writing session. Furthermore, writing sessions are usually considerably shorter than the 45–90 minutes deemed necessary to facilitate emotional habituation. Some of the benefits of expressive writing may be a result of repeated exposure to negative emotional experiences.

Conclusion

As can be seen, each of the proposed theories has supporting and contradictory evidence (Reference Sloan and MarxSloan & Marx, 2004b ). The mechanism of action appears to be complex, with the demonstrated benefits potentially resulting from some combination of immediate cognitive and/or emotional changes, longer-term cognitive and/or emotional changes, social processes and biological effects, rather than being accounted for by any single factor (Reference PennebakerPennebaker, 2004).