did blurred vision have an odds ratio of 18 in the trazodone-for-sleep-in-depression meta-analysis?
“As per side effects, blurred vision was most common with an odds ratio of 18 compared to placebo.”
a guest, Dr. Paul Sarkowski (clinician presenting the meta-analysis) said on Psychopharmacology and Psychiatry Updates, 2026-09-29, at 05:02. hear it in context → the speaker cited: a new meta-analysis.
the episode: Trazodone for Insomnia and Comorbid Depression (at 05:02). we link the show's own page; we do not re-host audio.
first, the part that holds
The number is real. A 2026 meta-analysis of trazodone for sleep problems in depression reports an odds ratio for blurred vision of 17.50, which rounds to the 18 the speaker gave.
what we found when we went and looked
The meta-analysis pooled ten randomized trials with 1,029 participants. Its abstract reports "blurred vision (OR = 17.50, 95% CI: 2.28-134.02)", alongside somnolence (OR 7.34) and sedation (OR 6.53), so "about 18" is accurate. Two parts of the framing go further than the abstract does. First, an odds ratio of 17.5 means blurred vision was the side effect most elevated relative to the comparison arm. It does not mean blurred vision was the most common side effect, and the abstract gives no absolute rates. Second, the abstract only says adverse effects were "more frequent with trazodone". It never names placebo as the comparator for this outcome. The confidence interval runs from about 2 to 134, which points to very few events behind the estimate. The FDA label's placebo-controlled trials show a much smaller gap: blurred vision in 6% vs 4% of inpatients and 15% vs 4% of outpatients.
assertion by assertion
| assertion | verdict | what the record says |
|---|---|---|
| in the meta-analysis of trazodone for sleep problems in depression, blurred vision had an odds ratio of about 18 | confirmed | The abstract reports OR = 17.50 (95% CI 2.28-134.02) for blurred vision. Ten RCTs, 1,029 participants. The interval is extremely wide, so the true elevation could be anywhere from about 2-fold to over 100-fold. The full text was paywalled, so we could not see how many trials or events fed this estimate. [1][2] |
| the comparison was against placebo | unverifiable | The abstract says only that adverse effects 'were more frequent with trazodone'. It does not say whether the comparator for blurred vision was placebo, an active drug or a mix. We could not read the full text to check. [1][2] |
| blurred vision was the most common side effect | misattributed | The 17.50 is a relative measure: the largest odds ratio among the side effects listed. It says nothing about absolute frequency, and the abstract gives no event rates. In the FDA label's placebo-controlled trials, blurred vision affected 6% (inpatients) and 15% (outpatients) of trazodone-treated patients, so it was not a dominant side effect by frequency. [1][3] |
both directions
for the speaker: The figure is accurate to the source. The abstract gives blurred vision an OR of 17.50, the largest of the side effects it lists, ahead of somnolence (7.34) and sedation (6.53). The authors' own conclusion singles out "somnolence and visual disturbances" as the main safety concerns. The FDA label also shows blurred vision more often on trazodone than placebo, especially in outpatients (15% vs 4%), so the direction is consistent across sources.
against: "Most common" mixes up relative and absolute risk: a large odds ratio can come from a rare event. The 95% interval of 2.28 to 134.02 is roughly 60-fold wide, which usually means only a handful of events in a few trials, so "18" sounds more precise than the data are. The abstract does not say the comparator was placebo. The FDA label's placebo-controlled data imply far smaller odds ratios for blurred vision: roughly 1.5 in inpatients and roughly 4 in outpatients, by our arithmetic from the label percentages. That suggests the pooled 17.5 depends heavily on the specific trials included.
what this cannot say
We read the PubMed abstract and the publisher page but not the paywalled full text. So we could not see how many trials and events contributed to the blurred-vision estimate, what the comparator arms were, how much the trials disagreed with each other, or how the authors graded certainty for adverse effects. The label figures come from older acute-treatment trials in depression generally, not specifically patients with sleep problems, so they are context rather than a direct check. The label-derived odds ratios are our own rough arithmetic from rounded percentages.
sources, each one fetched and read
- The efficacy and safety of trazodone for sleep problems in depressive patients: a GRADE-assessed systematic review and meta-analysis of clinical trials. Psychopharmacology (Berl). 2026;243(3):457-473. doi:10.1007/s00213-025-06910-y. PMID 41081841 (PubMed record via NCBI E-utilities) What we read there: "Ten RCTs (1,029 participants) were included... Adverse effects were more frequent with trazodone, including blurred vision (OR = 17.50, 95% CI: 2.28-134.02), somnolence (OR = 7.34, 95% CI: 2.91-18.50), and sedation (OR = 6.53, 95% CI: 3.59-11.87)." Fetched 2026-10-06 (HTTP 203). https://pubmed.ncbi.nlm.nih.gov/41081841/
- Same article, publisher page, Psychopharmacology (Springer), 2026, doi:10.1007/s00213-025-06910-y What we read there: Abstract only. Repeats blurred vision OR 17.50 (95% CI 2.28-134.02). The full text with per-outcome trial counts and event data is paywalled. Fetched 2026-10-06 (HTTP 200). https://link.springer.com/article/10.1007/s00213-025-06910-y?error=cookies_not_supported&code=cb8b60ba-3016-485f-8542-a3cd5b7c0bd8
- U.S. FDA, trazodone hydrochloride prescribing information, Adverse Reactions (controlled clinical studies), via openFDA drug label API What we read there: Blurred vision: 6% trazodone vs 4% placebo in inpatients (N=142 vs 95); 15% trazodone vs 4% placebo in outpatients (N=157 vs 158). Fetched 2026-10-06 (HTTP 200). https://api.fda.gov/drug/label.json?search=openfda.generic_name:%22trazodone%22&limit=1
speakers are described by role, not named. one sentence is quoted under fair use, with attribution and a link to the original at the timestamp; we do not re-host audio or video. nothing on this page is a reason to start, stop or change a medication. if you have the file behind anything marked unverifiable, the research board is where people can answer. adam