There is a kind of composure that comes from understanding a feeling, and another that comes from making sure nobody can see it. From the outside, the two can look identical. Inside the person, they are different psychological events.
A widely cited 12-year study suggests that the second pattern may have consequences beyond a difficult conversation or a lonely evening. Researchers affiliated with the University of Rochester, Harvard and Columbia followed a nationally representative group of American adults and found that greater self-reported emotional suppression was associated with a higher risk of death from any cause.
The finding deserves attention, but not the frightening shorthand it often receives online. It was an observational study published in 2013, not an experiment showing that unexpressed feelings directly cause death. The famous 35 per cent figure is real. Its exact meaning is narrower, and more interesting, than “bottling things up will kill you”.
The 35 per cent figure came from 729 adults
Benjamin Chapman and colleagues used the 2008 General Social Survey-National Death Index cohort. In 1996, 729 participants completed a six-item emotional-suppression scale. Researchers then linked those records to deaths registered through 2008. By the end of follow-up, 111 people had died, including 37 from cardiovascular causes and 34 from cancer.
The sample’s average age at the beginning was 44. Fifty-four per cent were women, and 81 per cent were white. The researchers estimated mortality risk after accounting for age, sex, years of education, minority racial or ethnic status and self-rated health.
In the published study, the key comparison was a 75th-percentile suppression score versus a 25th-percentile score. The higher point was associated with a hazard ratio of 1.35 for all-cause mortality. Put plainly, the model estimated a 35 per cent higher rate of death during follow-up for that difference in suppression.
This was not a comparison between the single most emotionally closed person and the single most expressive person. Nor does a 35 per cent relative increase mean that 35 per cent of suppressors died. Fifteen per cent of the whole sample died during follow-up. A hazard ratio compares the rate at which events occurred over time after the model’s adjustments.
The uncertainty was substantial. The 95 per cent confidence interval for all-cause mortality ran from 1.00 to 1.82, with a p value of .049. Cancer mortality produced a hazard ratio of 1.70, also with a wide interval. The estimated cardiovascular increase was 1.47, but it was not statistically significant. These are suggestive estimates close to the conventional significance boundary, not settled biological constants.
The questionnaire measured a pattern, not one private moment
The six questions did not ask whether someone had ever hidden a feeling. They asked about a broader tendency to keep emotions private, conceal anger, avoid telling friends something upsetting, remain pleasant to prevent distress, and hide anxiety so that other people would not worry.
Those behaviours overlap, but they are not identical. Keeping a confidence is different from being unable to tell a partner that something hurts. Waiting until a safer time to discuss anger is different from never acknowledging it. The scale compressed those distinctions into one continuous score, which is useful for statistical analysis but cannot reconstruct each participant’s emotional life.
Psychologists also distinguish suppression from reappraisal. Reappraisal means changing how a situation is understood so that its emotional force changes. Expressive suppression happens after the emotion has formed and inhibits its outward signs. Across five studies, James Gross and Oliver John found that habitual reappraisal and suppression had different associations with emotion, relationships and wellbeing. Looking calm therefore tells us very little about the process that produced the calm.
A hidden feeling can still leave the body working
One possible pathway is physiological. Preventing an expression does not necessarily switch off the arousal that preceded it. The mortality study’s authors discussed stress responses as one plausible mechanism, alongside behavioural and social routes. They did not measure a chain from suppression to chronic stress to death, however.
A smaller laboratory experiment helps explain why the hypothesis exists. Emily Butler and colleagues paired women who did not know one another and asked them to discuss an upsetting topic. One member of each pair was randomly assigned to suppress emotional behaviour, respond naturally or reappraise the situation.
Suppression disrupted communication and reduced rapport. In one experiment it magnified blood-pressure responses in the suppressors’ partners; in another it increased blood pressure in both people. That is useful causal evidence about a short conversation. It cannot demonstrate that repeated episodes produced the mortality pattern observed over 12 years.
This distinction protects the evidence from a tempting story. A momentary rise in blood pressure is not the same thing as cardiovascular disease, and the original study’s cardiovascular mortality estimate was not statistically significant. The laboratory result offers a possible piece of a mechanism, not proof of the entire pathway.
The social cost may accumulate quietly
Suppression also changes the information available to other people. Friends cannot respond to distress they never see. A partner who receives “I’m fine” every time may eventually stop asking, not necessarily because they are uncaring, but because the relationship has trained them to accept the answer.
Prospective research by Sanjay Srivastava and colleagues followed students through the transition to university. Stable differences in suppression, and increases specific to the new college setting, predicted lower social support, less closeness and lower social satisfaction. That study did not measure mortality, but it shows how an internal strategy can alter the social environment that surrounds a person.
I have written about a neighbouring pattern in Silicon Canals’ look at adults who build an identity around carrying everything alone. Self-sufficiency can prevent disappointment, but it can also remove the signal that care is needed. The overlap is not complete. Someone may avoid asking for practical help while speaking openly about emotion, or hide feelings while readily accepting assistance.
That is also why childhood history should not be assigned automatically. Our recent examination of childhood adversity and adult social connection described one route by which trust can become harder. Emotional suppression can develop for many reasons, including family norms, culture, work demands, privacy, temperament and immediate safety. Behaviour alone cannot reveal its origin.
The study found an association, not a sentence
The design had real strengths. Suppression was recorded before the deaths, the sample was designed to represent US adults, and mortality came from a national index rather than participants’ memories. Removing deaths in the first year did not erase the association, which reduced one concern that people already seriously ill might simply report more suppression.
Its limits are just as important. Suppression was self-reported once at baseline. Only 111 deaths were available for the all-cause model, with far fewer in each disease category. The researchers adjusted for several obvious differences, but an observational analysis cannot eliminate every unmeasured influence. Health behaviour, access to care, chronic stress, depression, social support and underlying illness could be causes, consequences or fellow travellers of emotional suppression.
The authors deliberately did not control for every possible behavioural or social mediator because they wanted the total association. That is defensible, but it means the model cannot tell us what carried the risk. A later Japanese cohort study also found that links between coping styles and mortality varied by sex and by the coping strategy measured. Evidence across settings is not uniform enough to turn one hazard ratio into a universal rule.
Most importantly, the result cannot calculate one person’s future. A questionnaire score is not a diagnosis, and a population-level hazard ratio is not a countdown. Someone reading this should not infer that years of reserve have caused hidden fatal damage.
Healthy expression is not constant disclosure
The alternative to suppression is not saying everything immediately. Discretion can protect privacy, timing and other people. Staying composed during an emergency may be exactly what the moment requires. Different cultures also attach different meanings to restraint, harmony and direct expression.
The more useful distinction is flexibility. Can a person postpone a feeling and return to it later, or does it disappear from every conversation? Can they choose whom to trust, or does trust never feel available? Can they recognise anger without either exploding or pretending it is absent?
Habitual suppression becomes concerning when it is the only permitted response. At that point, the person is not simply choosing privacy. They may be losing access to feedback, comfort, repair and help. If that pattern causes persistent distress or makes support feel impossible, a GP or licensed therapist can help explore it without demanding indiscriminate disclosure.
The 12-year study does not prove that expression lengthens life. It does make permanent emotional invisibility harder to dismiss as a harmless personality quirk. The practical question is not whether someone occasionally keeps a feeling to themselves. It is whether they have any safe, workable way to acknowledge what they feel and allow another person to know when it matters.
