Are Late- and Very-Late-Onset Schizophrenia Precursors of Dementia?
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsDear Authors,
Thank you for the opportunity to conduct an in-depth evaluation of your manuscript, “Are Late- and Very-Late-Onset Schizophrenia Precursors of Dementia?” This topic is of great importance to the field of geriatric psychiatry, as it carries fundamental implications for the differential diagnosis of psychoses and clinical prognosis.
Your study stands out for its unique and methodologically rigorous approach, involving a manual audit of patient records to validate the true age of schizophrenia onset (SSD). This is a major strength compared with other registry-based studies that rely solely on the date of first psychiatric contact.
The conclusion that—after adjustment for age, sex, and stroke—the risk of developing dementia in the LOS and VLOS groups is not higher than in the EOS group, is statistically robust and counterintuitive in light of large cohort studies.
To fully realize the potential of this important work and ensure maximal transparency, several detailed and critical issues need to be addressed.
Abstract
The abstract correctly reports the aim, method (record audit), sample size, and key statistical finding (no significant difference after adjustment for confounders).
However, it should more transparently describe the sample composition. It should explicitly note that the EOS group was partly composed of patients who were initially misclassified as LOS/VLOS in the registry (i.e., reclassified after record audit).
Introduction
The Introduction is substantively sound and well structured, effectively introducing the clinical debate regarding the classification of late-onset (LOS) and very-late-onset (VLOS) psychoses and their potential relationship with dementia.
However, it focuses mainly on dementia risk in LOS/VLOS compared to the general population. A stronger rationale is needed for why EOS is an appropriate reference (control) group. Several studies suggest that early-onset schizophrenia (EOS) itself may be associated with early neurodegenerative changes and cognitive impairment.
The Introduction should also prepare the reader for the fact that a model using age of onset as a continuous variable will be applied. Although less common, this is a statistically sophisticated approach. Please briefly signal this in the final paragraph of the Introduction, arguing that it avoids the arbitrariness of cut-off points and allows linear assessment of risk as a function of age.
Moreover, please emphasize more clearly that your study is likely the only one that has rigorously verified the age of SSD onset in patient records, rather than relying on unreliable registry data. This is a key element that will help you defend your findings in the Discussion.
Methodology
The main methodological concern is transparency. Your EOS group (early-onset schizophrenia, N=74) is not a population-based EOS cohort, but rather a group of patients whose originally misclassified LOS/VLOS diagnoses were corrected after record audit. Thus, this is an EOS group with delayed contact with psychiatric services in your region. Please strengthen this clarification in both the Introduction and Methods. This reference group is unique but may carry specific biases (e.g., greater stability, delayed treatment, or interrupted care outside your region) that could influence subsequent dementia risk.
Dementia diagnosis was based on DSM-IV criteria following neuropsychological and/or psychiatric evaluation (under supervision). Although this is superior to relying solely on registry codes, DSM-IV–based dementia diagnosis (used between 1997–2019) may be heterogeneous. Please explicitly address in the Discussion the potential inclusion of vascular dementia, particularly since stroke was included as a covariate. Stroke is strongly linked to vascular dementia. Briefly discuss how your patient record verification distinguished between vascular and neurodegenerative dementia. Please also comment on how documentation of stroke (the control variable) related to the diagnosis of vascular dementia in the records, as this is key for interpreting your logistic model.
In the Discussion, you acknowledge that you could not adjust for several critical variables (medication history, comorbidities, substance use, education level) because they were not systematically collected. The lack of control for antipsychotic treatment history (especially type and dose) and substance abuse constitutes a serious uncontrolled confounder. Please note this explicitly in the Methodology.
Results
Table 1 shows that raw data reveal dementia occurred ten times more frequently in the VLOS group than in LOS/EOS. Please maintain this high level of transparency in the text. In your first logistic analysis, clearly report that the crude OR for VLOS vs. EOS was 7.5 (95% CI: 1.7–32), but after adjustment, it reversed direction and became nonsignificant (OR 0.42). This dramatic shift is your study’s most important statistical finding and deserves full emphasis.
Table 1 (Characteristics) also shows that VLOS patients are significantly older (82.8 years) than LOS (65.8 years) or EOS (64.6 years). This is critical because age is the primary confounder. Your finding—namely, that the OR for VLOS changes from 7.5 to 0.42 after adjusting for age—provides strong statistical evidence that age, not VLOS itself, is the precursor of dementia. Please underscore this evidential strength more clearly in the Discussion.
Your reported result that stroke history significantly reduces the risk of dementia (OR 0.08; p = 0.009 in Table 2) contradicts general clinical consensus, as stroke is a well-established risk factor for dementia. Please recheck this effect and add a brief critical reflection in the Discussion. This result is likely an analytical artifact due to the small number of stroke/dementia cases (N=10 dementia cases) and should be nuanced rather than stated without qualification.
Discussion
The Discussion is well structured and compares your results reliably with key registry studies, which often show elevated risk (e.g., Ribe et al., Stafford et al.). Your unique contribution—auditing the age of onset—is rightly highlighted as a potential explanation for discrepancies with registry findings.
Please strongly argue that the complete attenuation of the LOS/VLOS–dementia association after adjusting for age is the central clinical conclusion of your study. In other words, it is not the later onset of psychosis but rather older age itself that is the precursor of dementia. Also emphasize that VLOS patients are significantly older (82.8 years) than LOS/EOS (≈65 years). This large age gap strengthens the clinical importance of your age adjustment.
The Limitations section is thorough, especially in transparently reporting the low rate of correctly registered onset age (only 28.4% for VLOS) and the absence of dementia validation (DSM-IV criteria only).
Conclusions
The conclusion is strong but should be balanced by stating that dementia occurred ten times more frequently in the VLOS group in the crude analysis (crude OR = 7.5). The conclusion should reflect that the correlation is strong but fully explained by age.
Language Evaluation
The English language is academic, accurate, and formal, with appropriate clinical and statistical terminology. The language quality poses no barrier to publication.
Summary
Your manuscript is a key contribution to the LOS/VLOS debate. The main revisions should focus on resolving the statistical anomaly regarding stroke, ensuring transparent discussion of the EOS group’s homogeneity, and clarifying the evidential strength of the age adjustment. These refinements will maximize the credibility and impact of your findings.
Author Response
We would like to thank the reviewer for the many important points that helped very much to improve our manuscript.
Comment #1. It should explicitly note that the EOS group was partly composed of patients who were initially misclassified as LOS/VLOS in the registry (i.e., reclassified after record audit).
Response #1. We fully agree and did not state this in the abstract because of the word limits. We have skipped one sentence and added in the abstract: Patients who were initially misclassified as LOS/VLOS in the registry but, after this audit, had an age of onset before age 40, were classified as EOS.
Comment #2. A stronger rationale is needed for why EOS is an appropriate reference (control) group. Several studies suggest that early-onset schizophrenia (EOS) itself may be associated with early neurodegenerative changes and cognitive impairment.
Response #2. We apologise for not stating this more explicitly and have added at the end of the introduction: We chose EOS as reference group because of the debate whether especially VLOS is not schizophrenia but a precursor of dementia.
Comment #3. The Introduction should also prepare the reader for the fact that a model using age of onset as a continuous variable will be applied. Please briefly signal this in the final paragraph of the Introduction, arguing that it avoids the arbitrariness of cut-off points and allows linear assessment of risk as a function of age.
Response #3. We fully agree and have added at the end of the Introduction: Cut-off points are arbitrary and we also analyzed our data with age of onset as a continuous variable.
Comment #4. Moreover, please emphasize more clearly that your study is likely the only one that has rigorously verified the age of SSD onset in patient records, rather than relying on unreliable registry data. This is a key element that will help you defend your findings in the Discussion.
Response #4. We thank the reviewer for this suggestion and have added in the Introduction: In contrast to lager cohort studies based on registry data where the validity of age of onset is not checked, we rigorously verified age of SSD onset in patient records.
Comment #5. Thus, this is an EOS group with delayed contact with psychiatric services in your region. Please strengthen this clarification in both the Introduction and Methods. This reference group is unique but may carry specific biases (e.g., greater stability, delayed treatment, or interrupted care outside your region) that could influence subsequent dementia risk.
Response #5. That is a very important point and we have added in the Methods: Our EOS group consisted therefore of patients with delayed contact with psychiatric services in our region. And we added in the discussion: This may carry specific biases (e.g., greater stability, delayed treatment, or interrupted care outside our region) that could influence subsequent dementia risk.
Comment #6. Please explicitly address in the Discussion the potential inclusion of vascular dementia, particularly since stroke was included as a covariate. Stroke is strongly linked to vascular dementia. Briefly discuss how your patient record verification distinguished between vascular and neurodegenerative dementia. Please also comment on how documentation of stroke (the control variable) related to the diagnosis of vascular dementia in the records, as this is key for interpreting your logistic model.
Response #6. This point indeed needed more clarification and we have changed in the Methods: The diagnoses of dementia was made by or under supervision of a psychiatrist, based on DSM-IV criteria for dementia. We also added: MRI’s and a neuropsychological investigation by a clinical psychologist where found in a limited number of patients case notes, and therefore information of the type of dementia was very often lacking. We did record a history of a stroke but this is not enough to distinguish between vascular and neurodegenerative dementia. In the discussion we added as limitation: For example, the type of dementia was lacking resulting in a heterogenous group of DSM-IV–based dementia diagnosis. And also as limitation: … and MRI’s or structured neuropsychological assessments to diagnose dementia were exceptions.
Comment #7. The lack of control for antipsychotic treatment history (especially type and dose) and substance abuse constitutes a serious uncontrolled confounder. Please note this explicitly in the Methodology.
Response #7. We added in the statistical section: We could not control for important variables as antipsychotic treatment or substance abuse, as these were not registered systematically in the patients case notes.
Comment #8. In your first logistic analysis, clearly report that the crude OR for VLOS vs. EOS was 7.5 (95% CI: 1.7–32), but after adjustment, it reversed direction and became nonsignificant (OR 0.42). This dramatic shift is your study’s most important statistical finding and deserves full emphasis.
Response #8. In the results section, we have used the word ‘reversed” instead of the more neutral ‘changed’.
Comment #9. Age, not VLOS itself, is the precursor of dementia. Please underscore this evidential strength more clearly in the Discussion.
Response #9. Thank you for this suggestion to make this more explicit. We have added in the discussion: After correction for age and other confounders, the risk reversed direction and became nonsignificant. The correlation between VLOS and dementia is strong but fully explained by age.
Comment #10. Add a brief critical reflection in the Discussion (concerning the stroke findings). This result is likely an analytical artifact due to the small number of stroke/dementia cases.
Response #10. We have missed the importance of discussing this and added: Although it was not surprising that our study confirmed age as a predictor of dementia, we were surprised by the lower risk of dementia in stroke patients. Stroke is a well-known risk factor for dementia, as we stated before.[11] We had only 10 patients suffering from dementia and 7 patients with a history of stroke and this may just be an artifact.
Comment #11. In other words, it is not the later onset of psychosis but rather older age itself that is the precursor of dementia. Also emphasize that VLOS patients are significantly older (82.8 years) than LOS/EOS (≈65 years). This large age gap strengthens the clinical importance of your age adjustment.
Response #11. Another suggestion for which we would like to thank the reviewer. We have added in the first part of the discussion: VLOS patients were significantly older (82.8 years) than LOS/EOS (≈65 years) and this large age gap strengthens the clinical importance of adjustment for age. After correction for age and other confounders, the risk reversed direction and became nonsignificant.
Comment #12. The conclusion is strong but should be balanced by stating that dementia occurred ten times more frequently in the VLOS group in the crude analysis (crude OR = 7.5). The conclusion should reflect that the correlation is strong but fully explained by age.
Response #12. See also response #9. In addition, we have added in the discussion: The complete attenuation of the LOS/VLOS–dementia association after adjusting for age is a unique result of your study.
Reviewer 2 Report
Comments and Suggestions for Authors- Dementia is mainly the leading syndrome complex for known degenerative diseases (DDs). For sch, this problem can become acute either in later stages, associated with progressive loss of memory and executive functions, or in some specific forms, when classic DDs are added. I do not specifically name specific forms. It is advisable for the authors to make corrections to the characteristics of patients in accordance with ICD-11 or ICD-10, depending on what the specialists were guided by in the retrospective study.
- Taking into account the above, in the Introduction, to substantiate the research objective, a boundary should be drawn between dementia, as a syndrome of DDs, and various forms of sch. These diseases and conditions have different etiopathogenesis.
- The work shows a small percentage of dementia in sch. This fact further confirms that dementia is a distinct syndrome with distinct etiology and treatment. At the same time, sch can remain “invisible” for many years, without cognitive impairment (except for hallucinations), especially for a non-specialist.
- The Result part of the work covers only age, gender, and medical history. In the text, the authors refer only to DSM data, but this information is not analyzed in any way. It is useful to summarize the symptoms by which dementia was assessed and describe them analytically in detail. A stroke appears in the tables; there may have been information about DDs.
- It is advisable to expand the Discussion and add to the discussion modern information about the correlation of dementia with other mental illnesses.
- Of the 17 quotes from References, only 3 are from the last 5 years. They should be updated.
Author Response
Comment #1. Dementia is mainly the leading syndrome complex for known degenerative diseases (DDs). For sch, this problem can become acute either in later stages, associated with progressive loss of memory and executive functions, or in some specific forms, when classic DDs are added. I do not specifically name specific forms. It is advisable for the authors to make corrections to the characteristics of patients in accordance with ICD-11 or ICD-10, depending on what the specialists were guided by in the retrospective study.
Response #1. We agree fully with the fact that in some patients the cognitive deficits associated with schizophrenia may be severe enough to fulfil the criteria for a major neurocognitive disorder according to DSM criteria, and without the presence of a comorbid degenerative dementia as for example Alzheimer (if this is what the reviewer means). In the files of these patients, often neuropsychological tests or MRI were used to find the cause of the cognitive deficits. However, in many patients files these additional information was lacking because the diagnoses of dementia as comorbid disorder (instead of being part of schizophrenia) was based on available information (signs, symptoms, screening tests such as the MOCA). We have added this in the limitation section. The specialists in our county use the DSM, not the ICD, so therefore all diagnoses were based on DSM-criteria.
Comment #2. Taking into account the above, in the Introduction, to substantiate the research objective, a boundary should be drawn between dementia, as a syndrome of DDs, and various forms of sch. These diseases and conditions have different etiopathogenesis.
Response #2. We have added this to the introduction, as we fully agree.
Comment #3. The work shows a small percentage of dementia in sch. This fact further confirms that dementia is a distinct syndrome with distinct etiology and treatment. At the same time, sch can remain “invisible” for many years, without cognitive impairment (except for hallucinations), especially for a non-specialist.
Response #3. That is certainly true, and therefore we rigorously verified age of SSD onset in patient records. If for example family members mentioned that their relative suffered from hallucinations for 10 years and fulfilling criteria for schizophrenia without being diagnosed, we assumed that age of onset was 10 years ago. We have added the verification of age of onset in the introduction.
Comment #4. The Result part of the work covers only age, gender, and medical history. In the text, the authors refer only to DSM data, but this information is not analyzed in any way. It is useful to summarize the symptoms by which dementia was assessed and describe them analytically in detail. A stroke appears in the tables; there may have been information about DDs.
Response #4. Unfortunately, detailed information how in every patient the diagnoses of dementia was made, is not available. We have added the lack of information of, for example, MRI’s or structured neuropsychological assessments as limitation in the discussion.
.
Comment #5. It is advisable to expand the Discussion and add to the discussion modern information about the correlation of dementia with other mental illnesses.
Response #5. This is indeed found in many other studies but was not the focus of our study. We added this as a limitation of our study.
Comment #6. Of the 17 quotes from References, only 3 are from the last 5 years. They should be updated.
Response #6. This is a limitation of the 8 studies that we have found of this topic, as indeed the vast majority is older that 5 years. We repeated our literature search and could not find other studies published in the last 5 years. The most recent study, published this year, also did not include other related studies from the last 5 years. We did add one recent study based on your comment #5.
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors responded to all comments and made appropriate revisions to the text. Since the work is subject to several significant limitations, the "Strengths and Limitations" subchapter should be separated. It would be appropriate to highlight the limitations separately.
Author Response
Thank you for this suggestion, we made a separate limitation section according to your suggestion.
