Short Sleep and Catching a Cold

A deliberately unusual experiment helps separate sleep from the ordinary differences between people: investigators measure sleep, expose healthy volunteers to a cold virus, and observe who develops illness. The design gives useful evidence about susceptibility after exposure, but it does not predict who will catch the next cold at work or prove that sleep loss is the sole cause.

Why investigators deliberately expose volunteers

In an ordinary cold season, one person’s sleep is tangled with their contact patterns, work schedule, household size, stress, smoking, and health. A person who reports more colds may also have a different chance of meeting an infected person, remember symptoms differently, or have another condition. A controlled viral-challenge study narrows that problem: volunteers are screened, their sleep is recorded before exposure, and researchers administer a measured dose of a known virus under supervision.

That does not make the study equivalent to randomly assigning sleep. Sleep duration is observed, not allocated, so a short sleeper may differ from a long sleeper in other ways. But exposure is much more uniform than in the community. Investigators can also collect nasal samples to verify infection and measure objective illness signs rather than rely only on a participant’s recollection of feeling unwell.

There is a key distinction between becoming infected and developing a clinical cold. In the 2015 challenge paper, infection was detected by virus culture or a rise in virus-specific antibody; a clinical cold also required objective illness signs. This separates viral acquisition from symptomatic disease, although neither endpoint is the same as severe illness or complications.

The 2015 study: actigraphy before rhinovirus 39

Prather, Janicki-Deverts, Hall, and Cohen (Sleep, 2015) studied 164 healthy adults, 94 men and 70 women, aged 18–55 years, recruited in the Pittsburgh area. Participants wore wrist actigraphs and completed sleep diaries for seven consecutive days. Researchers then admitted them to quarantine, gave nasal drops containing rhinovirus 39, and monitored them for five days. Actigraphy estimates movement-related sleep-wake patterns; it is useful in field studies but is not a direct brain-wave measurement like polysomnography.

For participants sleeping under five hours per night, the adjusted odds of a clinical cold were 4.50 times those of the more-than-seven-hour reference group (95% CI 1.08–18.69). For those averaging five to six hours, the odds ratio was 4.24 (95% CI 1.08–16.71). The six-to-seven-hour category was not statistically distinguishable from the reference (OR 1.66; 95% CI 0.40–6.95). The confidence intervals are broad and overlap substantially; do not read these categories as precise steps on a biological dose scale.

The analysis adjusted for prechallenge antibody levels and several demographic, psychological, body-size, seasonal, and health-practice variables. Sleep fragmentation was not associated with clinical cold in this study. That last finding is specific to how fragmentation was defined and to this sample; it does not establish that awakenings never matter for infection risk.

The earlier challenge: sleep diaries and continuity

An earlier Cohen et al. study in Archives of Internal Medicine (2009) enrolled 153 healthy adults. Participants recorded sleep for 14 days before quarantine and nasal inoculation with a rhinovirus. Investigators combined viral evidence with daily symptoms to distinguish infection from a clinical cold. Participants reporting under seven hours of sleep had greater odds of illness than those at eight hours or more, and lower sleep efficiency—the fraction of time in bed spent asleep—was also associated with illness.

The 2009 and 2015 results are not contradictory simply because one implicated sleep efficiency and the other did not. The studies assessed sleep over different windows and with different methods: the first relied on diaries; the second added actigraphy and used a shorter measurement period. Definitions, sample composition, and analysis choices also differ. When results vary across studies, measurement and uncertainty deserve more attention than selecting the most dramatic estimate.

What these numbers can and cannot tell you

An odds ratio is a comparison of odds between groups in the studied setting. It is not a probability that a particular short sleeper will catch a cold, and it does not mean sleep caused every case. Because cold-challenge studies use volunteers selected for health and low existing antibody to the challenge virus, baseline immunity and eligibility rules shape who can participate. The volunteers are not a representative sample of children, older adults, people with chronic illness, or people taking immune-modifying medication.

All participants are deliberately exposed to the same virus under a research protocol. Everyday risk also depends on whether exposure occurs, how much virus is encountered, prior immunity, vaccination where relevant, ventilation, and the mix of circulating pathogens. These experiments chiefly ask: among eligible adults who receive this challenge, does sleep measured before exposure predict infection or clinical illness? They do not answer whether sleeping longer prevents all respiratory infections in ordinary life.

The observed sleep association might reflect biological effects, correlated behavior, or both. Researchers adjusted for several possible confounders, but statistical adjustment cannot remove unmeasured differences, and sleep itself was not randomized. The studies also cannot establish a threshold that applies to all people: the categories were chosen for analysis and their estimates carry uncertainty. The evidence is a meaningful clue, not a personalized forecast or clinical screening tool.

An odds ratio can look like a large fold comparison even when the absolute risk in each group is not supplied by the estimate. The study’s reference group is not a universal probability of illness, and its model conditions also matter. For that reason, converting 4.50 into “four and a half times more likely” can overstate what was measured. It is a modeled relative odds, with a wide interval, within this particular cohort.

Enrollment itself shapes the result. In the 2015 study, volunteers had to be healthy enough to pass medical screening and had to have low pre-existing neutralizing antibody to rhinovirus 39. People with chronic illness, pregnancy, regular sleep medication, or several other conditions were excluded. Participants were paid volunteers in one metropolitan region, mostly young adults. Those safeguards improve experimental control, but they reduce direct generalizability to children, older adults, people with asthma or sleep apnea, or patients using immunosuppressive therapy.

There is also a difference between infection susceptibility and onward transmission. The challenge design can identify infection in a participant and score symptoms, but it does not measure how often that person would meet a virus in daily life or whether they would transmit it to someone else. It cannot separate how sleep might relate to mucosal defenses, symptom perception, or the course of the infection. Community studies have their own difficulty: exposure dose, household contacts, pathogen strains, and reporting behavior are not standardized.

Thus, these results support asking whether sleep is one contributor to the chance of symptomatic illness after exposure. They do not establish a causal percentage reduction from adding an hour of sleep, and they do not show that a sleep intervention prevents an infection. The estimates are odds ratios from regression models, not randomized treatment effects.

The broader Sleep science overview and its introductory Sleep & immunity page give the high-level cold-challenge summary. Here the focus is the sequence of sleep measurement, standardized inoculation, virus confirmation, and illness scoring—details that explain why the result is stronger than a simple symptom survey and still limited.

Questions, answered briefly

• 🤧 Does a short night mean I will get sick? No. Group-level odds do not predict an individual outcome. Exposure, prior immunity, pathogen, and other health factors matter.
• 🦠 Did the researchers count every infection as a cold? No. They separated laboratory-confirmed infection from a clinical cold, which also required objective signs of illness.
• 📊 Why show odds ratios rather than percentages? The paper modeled odds relative to a reference group. Translating those estimates into an individual’s absolute risk would need a baseline probability that does not apply universally.
• ⌚ Is actigraphy the same as a sleep-lab test? No. It estimates sleep from movement and is practical over several nights, but quiet wakefulness can be misclassified and it does not measure sleep stages directly.
• 🧑‍⚕️ Should I seek care because I catch colds? These challenge studies are not diagnostic criteria. Frequent, severe, or unusual infections warrant a clinician’s assessment rather than a conclusion based on sleep duration alone.

How to use this evidence carefully

• 🔎 Separate infection from illness. A positive virus measure and symptomatic disease are related but distinct endpoints.
• 📝 Check the sleep tool. Diary reports and wrist movement monitors have different measurement error; neither gives a perfect view of sleep.
• 📐 Read the uncertainty. The 2015 confidence intervals are wide, which means the estimated size of the association is not precise.
• 🛌 Keep prevention broad. Sleep is one part of health; use recommended vaccines, testing, and evidence-based care rather than relying on sleep alone.

Related Topics