Confounder or effect modifier: what the difference means for a shift worker
A confounder is a separate factor that both shift work and poor mental health happen to share, so that removing it would weaken or erase the apparent link. An effect modifier is different: it changes the size of the link itself, so the relationship is stronger in one group than another. The distinction matters practically, because a confounder calls for statistical adjustment, while an effect modifier calls for paying attention to which workers are most vulnerable.
The young-adult cohort study is the clearest test of this question. Anxiety and depression scores did not differ between shift and non-shift workers overall, but they were markedly higher in people with a clinically significant sleep disorder: median anxiety was 7.0 versus 4.0 and median depression 9.0 versus 4.0 on the respective scales [1]. In other words, the sleep disorder tracked mental health far more closely than shift status did.
The effect concentrates in evening and night shift workers with a sleep disorder
When the analysis set aside day-shift workers and looked only at evening and night shifts, a statistical interaction appeared between shift work and sleep disorder status for anxiety, though not for depression [1]. Anxiety scores were highest in evening or night shift workers who also had a sleep disorder, with a median around 8.5 [1]. This is the signature of effect modification: the same sleep disorder carries a bigger mental-health cost when it coincides with evening or night schedules.
The scope of that finding is narrow and worth stating plainly. It comes from 22-year-olds in one birth cohort, with sleep measured by laboratory polysomnography—an overnight sleep study—and it applies to anxiety, not depression, in the interaction analysis [1]. The depression difference was driven by sleep disorder status generally rather than by an interaction with shift type.
Most of these sleep disorders are undiagnosed, which changes what to do about them
Roughly one in five young workers in the sample met criteria for a clinically significant sleep disorder, and about 80% of those cases were undiagnosed [1]. The prevalence was similar in shift and non-shift workers (18% versus 21%), which argues against the idea that shift work simply creates more sleep disorders. Instead, the disorders appear to be common across the workforce and largely invisible.
That combination—common, undiagnosed, and more strongly tied to anxiety and depression than shift work itself—is why the authors argue for prioritizing identification and treatment of sleep disorders in young workers [1]. The mental-health burden seems to sit with the sleep disorder rather than with shift work per se, and it appears greatest when evening or night shifts and a sleep disorder occur together [1].
Other shift-work studies looked elsewhere, so the modifier question rests mainly on one cohort
Several related studies examined shift work and mood but did not test sleep disorders as confounders or effect modifiers. A 17-year cohort study linked shift work with mood disorders and sleep problems and considered chronotype—a person's natural sleep-wake timing—but did not analyze clinical sleep disorders in that role [2]. A study of nurses used patterns of shift scheduling to examine anxiety-depression comorbidity and reported that confounders only partially attenuated the risk, without addressing sleep disorders as modifiers [3]. A Danish nurse cohort found increased psychiatric risk with night shift work and examined effect modification by age and sleep-related factors, but not clinical sleep disorders as the modifier of interest [4]. A prospective cohort on shift work and depressive complaints corrected for possible confounders and looked at the reverse direction, depressed mood affecting shift-work changes, without treating sleep disorders as effect modifiers [5].
Taken together, these studies are consistent with shift work mattering for mood in some groups, but they leave the specific confounder-versus-modifier question to the young-adult cohort that measured sleep disorders directly [1]. The practical reading is that a clinical sleep disorder is not just noise to be adjusted away; in evening and night shift workers it appears to amplify the anxiety associated with those schedules, while shift work alone showed no overall mental-health difference in that cohort [1].
Sources used in this answer
Shift work, clinically significant sleep disorders and mental health in a representative, cross-sectional sample of young working adults
In 660 young working adults, shift work alone was not associated with anxiety or depression, but clinically significant sleep disorders (present in about one in five, 80% undiagnosed) were linked to higher scores, and an interaction indicated the highest anxiety in evening/night shift workers with a sleep disorder.
Association of shift work with mood disorders and sleep problems according to chronotype: a 17-year cohort study
A 17-year cohort found shift work associated with mood disorders and sleep problems and examined chronotype, but did not analyze clinical sleep disorders as confounders or effect modifiers.
Latent Class Analysis of Shift Work Patterns Among Nurses and Its Impact on Anxiety–Depression Comorbidity: A Multicenter Cross‐Sectional Study
Among nurses, different shift-work patterns were associated with anxiety-depression comorbidity, with confounders only partially attenuating risk, but sleep disorders were not examined as effect modifiers.
Shift work and incidence of psychiatric disorders: The Danish Nurse Cohort study
Night shift work was associated with increased incidence of psychiatric disorders in female nurses, with effect modification examined by age and sleep-related factors rather than clinical sleep disorders.
The mutual relationship between shift work and depressive complaints—a prospective cohort study
A prospective cohort examined the mutual relationship between shift work and depressive complaints in both directions, correcting for confounders but not treating sleep disorders as effect modifiers.