the forgotten branch. journaling and immune function

The arc of the expressive-writing literature most readers know is psychological. Mood improves, depressive symptoms drop, anxiety eases. The arc that runs alongside it, started in the same lab in 1988 and extended for thirty-five years across viral antibodies, vaccine response, CD4 counts and skin-tissue healing, is immunological. It is genuinely surprising, partially replicated, and almost entirely absent from the consumer journaling discourse.

the claim that started the branch

In 1988 Pennebaker, Kiecolt-Glaser and Glaser published Disclosure of traumas and immune function in the Journal of Consulting and Clinical Psychology. [5] Fifty healthy undergraduates were randomly assigned to write for twenty minutes on four consecutive days, either about the most traumatic experiences of their lives or about trivial assigned topics. Blood was drawn the day before writing, an hour after the final session, and six weeks later. Lymphocytes were stimulated with two T-cell mitogens, PHA and ConA, and proliferation measured.

The headline result was the PHA Condition × Day interaction, F(2, 80) = 3.36, p = .04. Trauma writers' lymphocytes proliferated more vigorously in response to the mitogen than controls' did, both immediately after writing and at six-week follow-up. ConA, the second mitogen, trended in the same direction but did not clear significance in the full sample. Health-center visits, tracked independently of the immune assay, showed a parallel Condition × Time interaction, F(1, 48) = 4.20, p < .05.

The results indicate that writing about traumatic experience has positive effects on the blastogenic response of T-lymphocytes to two mitogens, on autonomic levels, on health center use, and on subjective distress.

The paper is fifty undergraduates and one significant interaction on one of two mitogens. It is also, in 1988, the first time anyone had asked whether a writing assignment moved a cellular-immune marker in a randomised trial. The branch grew from there.

three studies that survived the next decade

Esterling, Antoni and colleagues ran the next move in 1994. Their paper in JCCP compared verbal disclosure, written disclosure, and a written-trivial control across three weekly twenty-minute sessions in fifty-seven EBV-seropositive undergraduates. [1] Post-intervention antibody titers against latent Epstein-Barr were ranked verbal-disclosure < written-disclosure < trivial-control, each step significantly lower than the next. Lower titers, in this assay, mean better cellular-immune control of a virus the host already carries: ninety per cent of adults are EBV-positive and rely on T-cell surveillance to keep the latent virus quiet. Disclosure restored some of that surveillance. Talking did it more than writing. The popular journaling literature usually paraphrases this study as writing lowered EBV titers and stops there. Against the trivial control it did. Against speech, it was the worse active arm.

Petrie, Booth and the Auckland group then ran the cleanest clinical-endpoint trial in the branch. In 1995 they recruited forty medical students naive to hepatitis B, randomly assigned them to four consecutive days of either trauma-disclosure writing or trivial-topic writing, and gave the first dose of a hepatitis B vaccination programme the day after the last session. [6] At four-month and six-month follow-up, anti-hepatitis-B antibody titers were significantly higher in the writing group than in controls. Vaccine seroconversion is a stricter endpoint than mitogen response: it asks whether the immune system, exposed to a real antigen, mounts a real protective response. On that measure, four days of writing produced a measurable advantage that survived for half a year.

Nine years later the same lab extended the design into a clinical population. Petrie, Fontanilla, Thomas, Booth and Pennebaker (2004) randomised thirty-seven HIV-infected patients to four 30-minute emotional or control writing sessions, with immune outcomes followed to six months. [7] Relative to drops in viral load, CD4+ lymphocyte counts in the emotional-writing condition rose more than in controls. Small N, single trial, never directly replicated. But the move from healthy undergraduates to a population whose disease is itself defined by a CD4+ count is the step the literature most needed.

the wound-healing finding

The single result a careful reader will remember from this branch is Koschwanez and Broadbent's 2013 trial in older adults. Forty-nine healthy participants aged sixty-four to ninety-seven were randomly assigned to a Pennebaker-protocol writing condition or a time-management writing control across three twenty-minute days. Two weeks after the writing protocol, every participant received a 4mm punch biopsy on the inner upper arm. Wounds were photographed daily for three weeks and rated by a clinician blinded to condition.

[2]

fully reepithelialised wounds at day 11 post-biopsy, older adults

76.2% vs 42.1%

koschwanez et al. 2013

Percentage of fully reepithelialised wounds at day 11 post-biopsy, expressive-writing arm versus time-management control

condition	% wounds fully healed at day 11
expressive writing	76.2
time-management control	42.1

The trial is the most physically tangible result in the branch. It is also a small single-site study in an unusual population, and the shape of its mediator analysis is the reason it remains interesting rather than tidy. Koschwanez and colleagues measured the candidates the psycho-immunological story would predict. Lipopolysaccharide-stimulated proinflammatory cytokine production in peripheral blood. Perceived stress. Depressive symptoms. Doctor visits. None of them differed between groups. The variable that did predict faster healing was sleep duration in the week before wounding, and it predicted faster healing in both arms equally. The expressive-writing effect on day-eleven reepithelialisation survived controlling for sleep, but the inflammatory pathway the trial was designed to find was not the pathway carrying the effect. Whatever the writing assignment did to the older skin, it did it through something the assay did not see.

the meta-analysis the wellness blogs don't cite

Mogk, Otte and colleagues ran the only meta-analysis of expressive writing on objective physical-health outcomes in 2006. [4] The pooled effect across the immune and biological-marker bucket was Hedges' g = 0.01, ninety-five percent confidence interval −0.27 to 0.29, across four trials. That number is indistinguishable from zero. Self-reported health and psychological wellbeing showed small positive pooled effects. The biological signal did not.

The 2017 follow-up to the wound-healing branch points the same way. Koschwanez and Broadbent's bariatric-surgery trial of seventy-six patients ran the same expressive-writing protocol before surgery and measured wound healing afterwards. [3] The writing arm did not heal faster. On hydroxyproline content, the biological proxy for collagen deposition at the wound, the writing arm was lower than the daily-activities control; TNF-α was higher. The first published failure-to-replicate of the lab's own most photogenic finding. The bariatric paper is rarely cited in the same sentence as Koschwanez 2013.

what the forgotten branch actually shows

Thirty-five years and roughly a dozen trials produce a reading narrower than the wellness blogs imply and broader than the sceptical pooled estimate suggests. Single trials show a real direction of effect on cellular-immune markers, vaccine antibody response, latent-viral antibody titers, CD4 counts and wound healing. The pooled effect on the same outcomes, across small samples and heterogeneous assays, is small to zero. The honest description has to hold both at once.

What the pillar of the science of journaling gains from the immune branch is a mechanistic claim rather than a clinical one. The same population variable that the rumination literature warns can deepen depression under the wrong cognitive frame can also, under a structured disclosure protocol, raise a vaccine antibody titer six months later. The branch is forgotten because the effect sizes are small and the wellness discourse needs them to be large. Read at the size they are, the studies remain one of the more peculiar survivals in the minimum-effective journaling literature.