Why can both short and long sleep look risky?
The clearest answer comes from studies that modeled sleep duration as a curve rather than a straight line. In a large U.S. sample of 17,635 adults, researchers found a U-shaped association between self-reported sleep duration and CRP, with 6–7 hours appearing most optimal and both under 6 and over 7 hours linked to higher CRP [3]. A study of Korean women reported the same U-shape for high-sensitivity CRP, with both short (≤5 hours) and long sleep associated with higher levels [1].
A U-shape means the middle of the sleep range sits lowest and the two ends rise. If that is the true pattern, then a study that only compares "short" against "normal" will find short sleep looks bad, while a study that lumps long sleep in with normal sleep will miss the long-sleep end entirely. The disagreement can therefore be a matter of which part of the curve a study was positioned to see.
Who is being studied changes the answer
Population matters. The Korean study was restricted to women [1], and a U.S. study of extreme sleep durations found that the link to elevated CRP differed across sex and ethnoracial groups [4]. That means a result from one group does not automatically transfer to another, and pooling very different groups can blur a pattern that is clear within each one.
Age may matter too. In adolescents, one study reported a consistent inverse relationship, with shorter sleep tied to higher CRP, rather than a U-shape [5]. That is a different age group from the adult samples showing both ends elevated [1][3], so the two findings are not necessarily in conflict; they may describe different life stages.
How "short" is defined decides what gets found
The cutoff for short sleep is not standardized. The Taiwanese study defined short sleep as ≤5.5 hours per day and reported a clear link to elevated high-sensitivity CRP [2]. The Korean study used ≤5 hours [1], and the U.S. study treated under 6 hours as the short end of its curve [3]. A study using a stricter cutoff captures only the most extreme short sleepers, while a looser cutoff pulls in people whose sleep is only mildly reduced.
This is a measurement issue, not just a statistical one. Self-reported sleep is also an estimate, and the U.S. findings rest on self-reported duration [3]. Two studies can both be internally sound and still disagree simply because their definitions of "short" and "long" carve the same population at different points.
Does the pattern tell us sleep causes inflammation?
These studies are observational, so they show association, not cause. The U.S. analysis reported that people sleeping under 6 or over 7 hours had higher cardiovascular death risk attributable to inflammation after accounting for confounders, and its authors describe the findings as suggesting an interplay between sleep, inflammation, and cardiovascular risk [3]. That is a suggestion of a pathway, not proof that changing sleep would change CRP.
The practical takeaway is that the studies are less contradictory than they first appear. A U-shaped result [1][3], a short-sleep-only result [2], a group-specific result [4], and an inverse result in adolescents [5] can all be true at once if the shape of the relationship, the population, and the sleep cutoffs differ. The safest reading is that very short sleep is the most consistently flagged end, while the long-sleep end shows up mainly in studies that model the full curve and in adults [1][3].
Sources used in this answer
U-shaped association between sleep duration, C-reactive protein, and uric acid in Korean women
In Korean women, both short (≤5 hours) and long sleep durations were associated with higher high-sensitivity C-reactive protein, a U-shaped pattern.
Short duration of sleep is associated with elevated high-sensitivity C-reactive protein level in Taiwanese adults: a cross-sectional study
In Taiwanese adults, short sleep duration of ≤5.5 hours per day was associated with elevated high-sensitivity C-reactive protein.
Sleep duration, baseline cardiovascular risk, inflammation and incident cardiovascular mortality in ambulatory U.S. Adults: National health and nutrition examination survey
In 17,635 U.S. adults, sleep duration showed a U-shaped association with C-reactive protein, with 6–7 hours appearing most optimal.
Extreme sleep durations and increased C-reactive protein: effects of sex and ethnoracial group
Extreme sleep durations were associated with elevated C-reactive protein, with patterns that differed by sex and ethnoracial group.
Sleep duration and emerging cardiometabolic risk markers in adolescents. The AFINOS study
In adolescents, short sleep duration was inversely associated with C-reactive protein, without a reported long-sleep end.