Real-World Study Finds Melatonin Starters Had Higher, Not Lower, Dementia Rates

A global records analysis built to compare new melatonin users against new benzodiazepine and zolpidem users for dementia and Parkinson's disease outcomes found the reverse of the hoped-for result: melatonin starters had higher rates of diagnosis, not lower.

Adults aged 50 and older who were newly prescribed melatonin for a sleep disorder went on to be diagnosed with all-cause dementia, vascular dementia, Alzheimer's disease, and Parkinson's disease at higher rates than matched adults newly prescribed benzodiazepines or zolpidem, according to a retrospective cohort study published in Scientific Reports on June 24, 2026. Researchers led by Hui-Chin Chang used a method called target trial emulation, built to mimic a randomized trial using routine health records, drawing on the TriNetX Global Collaborative Network between 2015 and 2024. For all-cause dementia, the hazard ratio for melatonin starters versus benzodiazepine starters was 2.09 (95 percent confidence interval, 2.01 to 2.16); versus zolpidem starters it was 1.74 (95 percent CI, 1.67 to 1.82). Similar patterns appeared for vascular dementia, Alzheimer's disease, and Parkinson's disease, and held up in sensitivity analyses that pushed the start of follow-up out as far as four years after the first prescription.

Melatonin is sold over the counter in many countries and has been studied for possible effects on circadian rhythm and, in laboratory and animal work, on markers of oxidative stress and inflammation linked to neurodegeneration. Those laboratory findings have fed a hope, discussed in prior research, that melatonin might be neuroprotective compared with sedative drugs. This study, comparing real-world prescribing patterns rather than testing melatonin against placebo in a trial, found the opposite direction of association, a reminder that an observational comparison between two active drugs answers a narrower question than a trial designed to test whether a treatment prevents disease.

The study's own authors caution against reading the result as evidence that melatonin causes dementia or Parkinson's disease. People who go on to develop these conditions frequently have disrupted sleep for years before a diagnosis is made, a phenomenon researchers call the prodromal period; if melatonin is more often reached for at that early stage than a benzodiazepine or a sleeping pill prescribed by a physician, the medication itself may simply be a marker of that early disease process rather than a cause of it. The authors describe residual confounding, confounding by indication, and prodromal neurodegenerative disease as more likely explanations than a pharmacologic effect of melatonin.

The researchers call for prospective studies that can better separate early disease-driven sleep changes from any effect of the medications themselves. Until then, the finding stands as a caution against treating any one sleep aid as a clear safer choice on neurological grounds.