Sleep and suicide share a title but not a table
- In a cross-sectional study at the Republican Clinical Psychiatric Hospital in Ufa, 390 inpatients with paranoid schizophrenia were assessed once, between the 14th and the 21st day of admission. On the Pittsburgh Sleep Quality Index 230 of them (59.0%) scored above 5, 124 (31.8%) scored 6 or more on the Calgary Depression Scale, 71 (18.2%) scored 18 or more on the Hamilton Anxiety Rating Scale, and 74 (19%) had made at least one suicide attempt.
- The sleep score correlated with three positive items on PANSS – delusions (r = 0.126, p = 0.013), hallucinatory behaviour (r = 0.179, p < 0.001) and suspiciousness (r = 0.132, p = 0.01) – and with the positive subscale total (r = 0.159, p = 0.002). No negative item reached significance, and the negative subscale total gave r = -0.044, p = 0.388.
- Intensity of suicidal ideation, scored on a subscale of the Columbia Suicide Severity Rating Scale, correlated with hallucinatory behaviour (r = 0.138, p = 0.006), suspiciousness (r = 0.12, p = 0.017), the positive total (r = 0.132, p = 0.009), emotional withdrawal (r = 0.117, p = 0.021) and passive-apathetic social withdrawal (r = 0.147, p = 0.004). The abstract leaves out emotional withdrawal; these figures follow Table 3.
- Sleep and suicidality were each analysed only against PANSS items. No analysis in the paper relates the sleep score to suicidal ideation or to past attempts, so the data neither show a link between the two nor rule one out.
The 390 patients were on the wards of the Republican Clinical Psychiatric Hospital in Ufa, each between the 14th and the 21st day of an admission for paranoid schizophrenia. They were 18 to 60 years old, 40.8 on average, 215 men and 175 women. The exclusion criteria ruled out legal incapacity, substance dependence other than nicotine, obstacles to verbal contact, a second psychiatric diagnosis and severe somatic illness in exacerbation. By the time of assessment all of them were on antipsychotic treatment, and the authors list this among their limitations because it could have shifted the symptoms and the sleep complaints alike. Each patient was assessed once and nobody was treated as part of the study. The authors call the design cross-sectional and write that it allows no conclusion about cause.
The PANSS rows
Five tables carry the results. In every one the rows are the fourteen PANSS items and the two subscale totals, and only the second variable changes: sleep in Table 1, depression in Table 2, suicidal ideation in Table 3, a past attempt in Table 4, anxiety in Table 5.
Sleep went with the positive side. The largest coefficient in Table 1 is 0.179, for hallucinatory behaviour, then the positive total at 0.159, suspiciousness at 0.132 and delusions at 0.126. Every negative item stayed below significance. The abstract names delusions, hallucinations and the positive total; the table and the results text add suspiciousness. Suicidal ideation reached into both subscales, the highest coefficient being 0.147 for passive-apathetic withdrawal. The 74 patients with a past attempt averaged 17.84 on the positive subscale against 15.66 in the 316 without one (p = 0.001), and the two groups sat level on the negative subscale.
PSQI and the Columbia scale
The sleep questionnaire and the suicide rating came from the same 390 people, and no table sets one against the other. The PSQI appears only in Table 1, the Columbia subscale only in Table 3, the attempt history only in Table 4. Both variables correlate with hallucinatory behaviour and suspiciousness, but two correlations with a third variable do not give the correlation between the first two. Three studies in the reference list carry titles that pair sleep problems or insomnia with suicidal ideation in psychosis or in people at clinical high risk for it. The introduction cites them for the link between sleep and PANSS scores.
The sleep measure is one questionnaire filled in once. The authors call a PSQI total above 5 clinically significant insomnic disturbance, with no criterion for frequency or duration and no diary, actigraphy or polysomnography. The mean total was 6.8.
For a clinician on an admission ward the Ufa figures hold two separate facts. Poor self-rated sleep went with delusions, hallucinations and suspiciousness, and one patient in five had attempted suicide at least once. They give no ground for reading a sleep complaint as a suicide signal, and no ground for ruling it out as one.
Sleep enters Table 1, suicidal ideation Table 3 and past attempts Table 4, and no table in the paper holds two of them.
This is a cross-sectional study at a single hospital, and every figure is a correlation or a group difference measured at one point, so it reports co-occurrence and not prediction or cause. Patients were assessed on the 14th to 21st day of admission, already on antipsychotic treatment whose drugs and doses are not reported; the authors note that treatment may have altered symptoms and sleep complaints. Sleep was measured by a single self-report questionnaire, the Pittsburgh Sleep Quality Index, completed once, and the label "insomnic disturbance" means a total above 5 rather than a diagnosis with criteria for frequency and duration. No sleep diary, actigraphy or polysomnography was used. The recruitment period is not stated; the ethics approval is dated February 2019 and the manuscript was received in December 2025. The analyses are bivariate, with no adjustment for age, sex, medication or somatic illness; the authors list the unexamined external factors and somatic comorbidity among their limitations. The paper contains no analysis of sleep against suicidal ideation or attempts. The absence of that analysis is not evidence that no association exists; it means these data cannot answer the question. Patients with substance dependence, a second psychiatric diagnosis or acute somatic illness were excluded, so the sample is narrower than a general admission ward.