Published on 18 September 2026
An azapirone added to an SSRI in vascular depression of the elderly: what does twelve weeks buy?
The essentials
A single-center trial compared, in 134 patients around 70 years old with vascular depression, escitalopram 10 mg per day combined with tandospirone citrate against escitalopram combined with placebo, over twelve weeks. On the HAMD, the difference between arms is significant only at week 2 (20.85 versus 15.94, 95% CI 3.97 to 5.85, p < 0.001). It is no longer significant at week 4 (14.28 versus 13.27, p = 0.123), nor at week 8, nor at week 12, where the two arms converge around 10. The same pattern appears for anxiety: a significant gap at week 2 (15.27 versus 12.24), then nothing afterward, and the overall group effect on the HAMA does not reach significance (p = 0.118). What persists concerns somatic complaints: the between-arm gap is significant at weeks 2, 4, 8 and 12 on both the PHQ-15 and the SSS, with a difference at week 12 of 0.95 point on the PHQ-15 (95% CI 0.54 to 1.37, p < 0.001) and 2.27 points on the SSS (1.36 to 3.18, p < 0.001). The finding therefore concerns the speed of onset and somatization, not a durable antidepressant superiority. Symmetrically, this later absence of difference does not prove the two strategies are equivalent either: the trial was not designed to establish that.
The context
Depression in older adults rarely presents the way it does in younger adults. Complaints often center on the body before they center on mood: diffuse pain, digestion, sleep, a sense of exhaustion. When a vascular factor is added, the response to serotonergic agents is thought to be slower and less complete, leaving the clinician with several weeks of uncertainty while the patient measures time differently.
Hence the recurring interest in upfront augmentation strategies. Tandospirone belongs to the azapirones, partial agonists at 5-HT1A receptors, the same family as buspirone (a pharmacological aside, outside the scope of the trial). The idea tested here is not to obtain more eventual response, but to obtain sooner what would be obtained anyway, and possibly to act on the somatic dimension.
The construct of vascular depression itself remains debated, and its operational definition shapes how the trial should be read. The authors stacked several criteria: a major depressive episode, first episode or recurrence, per DSM-5; the vascular depression criteria proposed by Alexopoulos and colleagues; a score above 17 on the 17-item HAMD; a score of 5 or more on the PHQ-15; and clear consciousness with sufficient autonomy to complete the scales unaided. Neuroimaging enters through the exclusion criteria: inability to cooperate with brain MRI excluded a patient, implying that every participant underwent imaging. Also excluded were psychiatric disorders other than depression and anxiety, epilepsy, intracranial tumors and other neurological disease, severe or unstable visceral disease, allergy to either molecule, high suicide risk, alcohol or substance dependence, poor adherence, and recent participation in another trial. The resulting population is therefore a symptomatically severe vascular depression, with a baseline HAMD around 28.8, but somatically pre-screened.
The proposed mechanism
The hypothesis is serotonergic. A partial agonist at 5-HT1A receptors, by desensitizing presynaptic autoreceptors in the raphe nucleus, is thought to lift more quickly the brake that delays the effect of reuptake inhibitors. This hypothesis is old and plausible, and the authors adopt it in their discussion, but this trial does not demonstrate it in humans.
What the study measured were platelet 5-HT1A and 5-HT1B receptors, assayed by enzyme immunoassay on fasting venous samples, before treatment and then at weeks 2, 4, 8 and 12, with instructions to avoid tryptophan- and tyrosine-rich foods in the preceding 72 hours. What it did not measure was the availability of these receptors in the central nervous system. The step from one to the other is a working hypothesis, not an established fact: the platelet is a convenient peripheral model, its correspondence with central serotonergic tone is not equivalence, and the authors themselves note that the animal literature is contradictory on how 5-HT1A receptors evolve under long-term SSRI exposure.
The authors also ran mediation analyses using the PROCESS macro, which test whether an effect observed on one dimension passes through another. The limitation lies in the measurement timing: when the presumed mediator and the presumed outcome are collected at the same time point, the causal order is assumed by the model, not established by the data. Only the exposure, arm assignment, is randomized here.
The study at a glance
| Population | Patients with vascular depression recruited at a single center, Union Hospital of Fujian Medical University, Fuzhou. 262 patients screened, 134 analyzed, 67 per arm. Mean age 69.88 ± 5.71 years in the monotherapy arm and 70.82 ± 6.84 years in the combination arm. Men: 17 (25.4%) and 23 (34.3%). Mean schooling of eight years. Hypertension in 71.6% and 82.1%, diabetes in 35.8% and 22.4%. Baseline HAMD 28.79 ± 2.38 and 28.85 ± 2.47, indicating severe depression at entry. No baseline between-arm difference (p ≥ 0.05). |
|---|---|
| Intervention | After a one-week washout, escitalopram 10 mg per day combined with tandospirone citrate, 5 mg three times daily in the first week then 10 mg three times daily from the second week onward, for twelve weeks. If an adverse effect was attributed to escitalopram, the dose was reduced to 5 mg per day for three days then returned to 10 mg. |
| Comparator | Escitalopram 10 mg per day combined with a placebo matched in size, shape and odor, at the same dosing schedule. There was neither a tandospirone-alone arm nor a pure placebo arm: the trial therefore cannot isolate what belongs to each molecule. The authors acknowledge this choice and cite an ethical rationale for the absence of a placebo-only arm. |
| Endpoints | Primary endpoints: change from baseline to week 12 in the HAMA, HAMD, PHQ-15 and SSS, the latter being the Chinese version of the Somatic Symptom Scale. Secondary endpoints: platelet 5-HT1A and 5-HT1B receptors. Response was defined as a reduction of at least 50% in total score. Four scales announced at the same rank raise a question of hierarchy and multiplicity: no single primary endpoint is designated, no multiplicity correction is described, and no sample size calculation appears in the material reviewed. |
| Design | Single-center, double-blind, randomized controlled trial against placebo, registered as ChiCTR2300075407 and approved by the institution’s ethics committee. 1:1 allocation by an independent investigator, using an algorithm generated in SPSS 25.0. Blinding was assessed with the Bang blinding index. Intention-to-treat analysis by generalized estimating equations, supplemented by Spearman correlations adjusted for age, sex and schooling, and by mediation analyses. |
Quality control
| Blinding maintenance | Bang index announced, values not published. FINDING: The use of a Bang blinding index is uncommon in this field and is a notable methodological intention. It nonetheless leads to no verifiable conclusion: the method is cited in the methods section, but no index value, by arm or by time point, appears in the results or the supplementary material. The reader knows blinding was measured, not whether it held. |
|---|---|
| Trial registration | ChiCTR registration, date not reported. FINDING: The trial carries registration number ChiCTR2300075407, and ethics approval carries reference number 2019KJCX065. The registration date does not appear in the material reviewed, leaving open the question of prospective registration and of endpoint stability between the registered protocol and the publication. |
| Size and duration | One center, 134 patients, twelve weeks. FINDING: A sample of this size is sufficient to detect a marked early gap, far less to rule out a modest late one. Twelve weeks says nothing about what happens to the difference beyond that point, and the authors themselves list both points among their own limitations. |
| Comparison arms | Neither tandospirone alone nor pure placebo. FINDING: The two-arm design answers an augmentation question, not a question of standalone efficacy. What tandospirone contributes outside a serotonergic agent remains outside the scope of this trial. |
| Endpoints | Four scales announced at the top rank. FINDING: Without an explicit hierarchy, a multiplicity correction, or a reported sample size calculation, reading one significant result among several becomes fragile, and the choice of which result to foreground becomes an editorial decision by the authors. |
| Statistical model | Week 2 absent from the primary model. FINDING: The authors state that baseline measurements and weeks 4, 8 and 12 were entered into the generalized estimating equations. Week 2, which nonetheless carries the most prominently featured result on mood, is not listed among the model’s time points, even though it appears in the results table with a time-by-group interaction. This mismatch between the model description and the data presentation is not explained. |
| Missing data | Intention-to-treat announced, attrition not reported. FINDING: The analysis is described as conducted on an intention-to-treat basis. Both arms are reported at 67 patients at every time point, without the number of dropouts or the method for handling missing data being specified in the text. The flow diagram is a figure, not reproduced in numeric form in the material reviewed. The screening count is in fact internally inconsistent: 262 patients screened, 134 reported excluded, but a breakdown of exclusions that totals 130, and 134 patients ultimately analyzed. |
| Funding and conflicts of interest | Public funding, no industry ties declared. FINDING: Funding is entirely public and regional: the Fujian joint funds for science and technology innovation, the Fujian provincial natural science foundation, accompanying funding for the national clinical specialty construction project in neurology, and the Fujian clinical research center for precision diagnosis and treatment of neurological disease. The authors state the funders had no role in design, data collection, analysis, writing, or the decision to publish, and declare no competing financial interest or personal relationship. |
The results
95% confidence interval of the between-arm difference on the HAMD at week 2, the only time point at which this difference on mood is significant. At weeks 4, 8 and 12, it no longer is.
| HAMD | Baseline 28.79 ± 2.38 versus 28.85 ± 2.47. Week 2: 20.85 ± 2.76 versus 15.94 ± 2.72, 95% CI 3.97 to 5.85, p < 0.001. Week 4: 14.28 ± 3.91 versus 13.27 ± 3.78, p = 0.123. Week 8: 10.57 ± 2.72 versus 10.22 ± 2.55, p = 0.447. Week 12: 10.24 ± 2.51 versus 10.19 ± 2.67, p = 0.920. Overall group effect p < 0.001.
READING: This is the trial’s central result, and it does not support the summary’s framing. The nearly five-point gap at week 2 is clear, it shrinks to one point at week 4 and then disappears: the two arms converge around 10. The overall group effect stays significant, but it is carried entirely by week 2, which makes it an indicator of speed rather than final level. The combination does no better than escitalopram alone on mood at twelve weeks. It does no worse either, and the trial does not establish equivalence between the two strategies any more than superiority: this symmetry cuts both ways. |
|---|---|
| HAMA | Baseline 24.13 ± 4.16 versus 24.15 ± 4.08. Week 2: 15.27 ± 4.23 versus 12.24 ± 4.13, 95% CI 1.60 to 4.46, p < 0.001. Week 4: 11.04 ± 2.92 versus 10.91 ± 2.81, p = 0.784. Week 8: 9.10 versus 8.88, p = 0.657. Week 12: 8.60 ± 2.94 versus 8.57 ± 2.99, p = 0.953. Overall group effect p = 0.118.
READING: The pattern mirrors the HAMD: an initial three-point gap that does not hold. Here, however, the overall group effect does not reach significance, which considerably weakens the claim of an anxiety improvement made in the abstract and in the authors’ conclusion. |
| PHQ-15, somatic symptoms | Baseline 12.97 ± 1.73 versus 12.99 ± 1.55. Week 2: 10.10 versus 8.06 (1.45 to 2.64, p < 0.001). Week 4: 8.16 versus 6.21 (1.34 to 2.57, p < 0.001). Week 8: 4.97 versus 4.19 (0.29 to 1.27, p = 0.002). Week 12: 4.07 ± 1.33 versus 3.12 ± 1.11 (0.54 to 1.37, p < 0.001). Overall group effect p < 0.001.
READING: Unlike mood, the gap exists at every time point and holds to the end. Its magnitude shrinks over time, however: two points at weeks 2 and 4, under one point at weeks 8 and 12, on a scale running from 0 to 30. What a gap of under one point at trial end means for an individual patient is not settled by this study, and the authors report no minimal clinically important difference for this scale. |
| SSS | Baseline 49.34 ± 4.86 versus 50.45 ± 4.60. Week 2: 44.28 versus 40.60 (2.04 to 5.34). Week 4: 37.81 versus 35.13 (1.60 to 3.74). Week 8: 30.40 versus 28.34 (0.95 to 3.17). Week 12: 27.18 ± 2.77 versus 24.91 ± 2.57 (1.36 to 3.18). All these differences are significant at p < 0.001. Overall group effect p < 0.001.
READING: Agreement with the PHQ-15 reinforces the somatic reading of the benefit, across two distinct instruments and at every time point. Both arms move from moderate-to-severe somatization to mild somatization by the thresholds the authors use for the Chinese version of the scale, with the between-arm gap remaining around two points on scores ranging from 25 to 50. |
| Platelet 5-HT1A receptors | Increase in both arms. Gap favoring the combination arm at week 8 (9.97 ± 2.01 versus 14.28 ± 1.85 ng/mL) and week 12 (10.98 ± 1.78 versus 14.89 ± 2.40 ng/mL), p < 0.001 at both time points. No difference at baseline, week 2 or week 4.
READING: Consistent with the pharmacological hypothesis, on an exploratory peripheral marker. It provides neither a monitoring criterion nor a surrogate for the clinical outcome, especially since the biological gap appears at week 8, when the clinical gap on mood has already disappeared. One further caution is needed: the publication’s abstract reports for this marker at week 12 a confidence interval of 0.08 to 0.29, incompatible with the 3.91 ng/mL difference and the interval of minus 4.63 to minus 3.18 carried by the results table. This inconsistency belongs to the source. |
| Platelet 5-HT1B receptors | Decrease in both arms, from approximately 9.55 ng/mL at baseline to 3.70 ± 0.80 and 3.64 ± 0.80 ng/mL at week 12. No between-group difference at any time point, overall group effect p = 0.718.
READING: The decline affects both arms and is of large magnitude. It therefore does not isolate an effect specific to tandospirone and could reflect the course of serotonergic treatment in general, or the natural course of the episode. |
| Mediation analysis | The treatment arm acts on the PHQ-15 with the HAMD as mediator: total effect minus 1.14 (minus 1.71 to minus 0.58, p = 0.0001), direct effect minus 0.60 (minus 0.87 to minus 0.33), indirect effect minus 0.54 (minus 1.03 to minus 0.04), proportion mediated 47.41%. The same model toward 5-HT1A receptors: indirect effect 0.17 (0.02 to 0.33), proportion mediated 10.34%.
READING: The model’s direction is explicit: arm assignment to the combination is the exposure, the HAMD the mediator, somatization the outcome. The authors conclude that tandospirone improves somatization by relieving depression. A mediation figure is not, however, a causal demonstration: the indirect effect’s interval brushes zero at its upper bound, and the model imposes a temporal order between mediator and outcome that concurrent measurements do not provide. The subsequent moderated-mediation model, with the 5-HT1B receptor as moderator, belongs to the exploratory register. |
| Tolerability | 10 adverse events in the monotherapy arm (14.9%, 10 of 67) and 12 in the combination arm (17.9%, 12 of 67), no significant difference (p = 0.641). No notable abnormality in hematology, urinalysis, liver or renal function, electrocardiogram, or electroencephalogram.
READING: The detail is consistent with the expected profile of a serotonergic agent: six digestive disorders, two headaches, one episode of dizziness and one of drowsiness under monotherapy; seven digestive disorders, four episodes of dizziness and one paresthesia under the combination. Events cluster in the first two weeks and are described as mild and self-resolving. Two caveats temper this reassuring reading: with 67 patients per arm, a trial of this size can only detect a frequent signal, and data are missing on the points of greatest concern at this age, sodium level, falls and QT interval, with biological and electrocardiographic monitoring reported only in aggregate, without figures. |
Critical appraisal
| Randomization | Sequence generated, allocation concealment not described. FINDING: Allocation is described: 1:1 ratio, algorithm generated in SPSS 25.0, executed by an investigator independent of the care team. What is missing is downstream: concealment of allocation until enrollment, whose procedure is not specified. |
|---|---|
| Protocol deviations | Matched placebo, blinded raters. FINDING: The placebo is matched in size, shape and odor, at the same dosing schedule, and raters were unaware of arm and treatment. The intention to measure blinding with a Bang index points the same way, but it remains without a published result, which prevents concluding that blinding actually held for twelve weeks. |
| Missing data | Attrition undocumented. FINDING: In patients of this age over twelve weeks, attrition is never negligible, and its extent bears on how the curves should be read. Yet the reported numbers stay at 67 per arm at every time point, with no dropout and no imputation method reported. |
| Outcome measurement | Two trained neurologists, blind to arm. FINDING: The HAMD and HAMA require a trained rater. Here, assessment was entrusted to a senior attending neurologist and a deputy chief neurologist, trained for scoring agreement and kept unaware of arm and treatment. The inter-rater agreement coefficient is not, however, reported. |
| Selective outcome reporting | Four endpoints, hierarchy unspecified. FINDING: The registration number is provided, but the registered protocol is not reproduced in the material reviewed. It is therefore not possible to know whether foregrounding somatization was planned or decided after the fact, nor whether all four scales had been declared at that rank. |
| Causal inference of mediation | Concurrent measurements, causality not established. FINDING: This is the weakest point. A mediated proportion calculated from measurements collected at the same time point describes a correlational structure, it does not demonstrate a sequence. Yet the authors write that the mediation analysis confirms that tandospirone attenuates somatization by relieving depression: the verb is too strong for the design. |
| External validity | One center, older Chinese population. FINDING: Transposing this to a French patient population is a matter for discussion rather than assertion: comorbidity profiles, prescribing habits, cultural expression of somatic complaints and drug availability all differ. Adding to this, a mean schooling of eight years and a broad exclusion of unstable somatic comorbidities are two features that move the sample away from the ordinary polymorbid older patient. |
| Match between claim and evidence | Abstract more affirmative than the results. FINDING: The conclusion announces a significant improvement in anxiety, depression and somatization. On anxiety, the overall group effect is not significant (p = 0.118), and the gap disappears from week 4 onward. On depression, the gap also disappears from week 4 onward. Only somatization sustains the claim at twelve weeks. This gap between claim and data is exactly why the abstract should not be the last word. |
Level of evidence
A single randomized trial of modest size, whose results have not been replicated by an independent team. Confidence is relatively high on two points, the existence of an early gap favoring the combination followed by its extinction on mood and anxiety, and the persistence of a gap on both somatization scales, because the design is randomized, controlled against a matched placebo, and the raters were trained and blinded. It is weak on the rest: the clinical significance of a sub-one-point gap on the PHQ-15 at trial end, the proposed mechanism, the reach of the platelet marker, and transposition to another population. On tolerability, the data exist and point in a reassuring direction over the short term, but a trial of 134 patients over twelve weeks cannot settle either rare events or the risks specific to older age, namely hyponatremia, falls and QT prolongation, none of which is given a figure.
One further caution of principle should be added. A twelve-week trial in 134 patients lacks the power to rule out a late difference of small magnitude. Saying the combination is not superior at twelve weeks amounts to saying this trial did not detect superiority, which is not the same as saying none exists.
The colleague test
What an experienced colleague would say if shown this study in two minutes, between two consultations.
“An older patient, vascular depression, somatic complaints in the foreground, fine, I recognize the picture and I get the logic of adding a 5-HT1A partial agonist to the SSRI. But I don’t read this as antidepressant superiority: by twelve weeks the two arms meet, at 10 on the HAMD. What I take away is speed of onset and somatization. Tolerability looks comparable, around fifteen percent of adverse events on each side, mostly digestive, but with 67 patients per arm that tells me nothing about the problems I actually worry about at this age. And since the molecule tested isn’t one I could prescribe here, I’d want to read the full paper before drawing anything from it.”
Translation for practice: this trial does not provide an argument for adding an azapirone to an antidepressant upfront in an older depressed patient. It provides a hypothesis worth watching, a gain in time-to-effect and an action on the somatic dimension, in a specific subgroup, with a molecule that is not available in France.
What you can do with this
- Measure the somatic dimension separately from mood in the older depressed patient. This is the trial’s most solid message: the two dimensions do not move at the same pace, and an improving HAMD can mask unchanged somatic complaints, or the reverse.
- When a patient asks why it is taking so long, be honest about the limits of the evidence: the question of accelerating antidepressant onset remains open, and no strategy is established to the point of changing first-line practice in older adults.
- Do not transpose directly. The proposed transposition toward buspirone rests on shared membership in the azapirone class, but it changes molecules, and this substitution was not tested by this trial. It is a hypothesis, not a result, and a molecule’s actual market availability is a separate question from its authorization status, which is itself separate from reimbursement. As a concrete illustration: as of 1 September 2026, tandospirone holds no authorization in France or the European Union, its authorization being Japanese only, while buspirone is authorized and actually marketed in France, under two generics listed as marketed in the public medicines database, schedule I, reimbursed at 65 percent. Its authorized indication, however, is the treatment of anxiety, so using it for augmentation in depression of the elderly would be an off-label use, which itself presupposes the absence of an appropriate authorized alternative. In patients over 65, the escitalopram summary of product characteristics also caps the maximum dose at 10 mg per day, a limit regulators reiterate because of dose-dependent QT prolongation. The example here is French, but the structure of the problem is not: authorization, marketing status and reimbursement are three distinct questions that vary from one country to another and change over time, so readers practicing outside France should verify all three for buspirone, or for any azapirone considered as a substitute, against their own jurisdiction’s current regulatory and reimbursement sources.
- If a serotonergic augmentation strategy is considered in an older patient outside the setting of this trial, keep in mind the points of vigilance specific to this age group: antidepressant dosing adapted to age, sodium monitoring, fall risk, drug interactions, and electrocardiographic monitoring depending on context and concomitant treatment. The trial did carry out biological and electrocardiographic monitoring, but publishes no numeric values for it.
- Value the method as much as the result. A trial that plans to measure its own blinding is worth more than one that merely asserts it, provided the result is actually published, which is not the case here. This is a reusable reading criterion for any placebo-controlled trial.
- Management is detailed in the NICE decision tree for depression in adults.
- The course of action is set out in the NICE decision tree for depression in adults.
Does this trial justify adding an azapirone upfront to an antidepressant in an older depressed patient?
No. The benefit on mood and anxiety does not last beyond the first few weeks, the trial is single-center and of modest size, and the molecule tested is not available in France. What is demonstrated here is limited to an early gap and a sustained, though shrinking, gap on two somatic-symptom questionnaires.
Can tandospirone be replaced with buspirone?
This is a hypothesis based on shared membership in the same pharmacological class, not a conclusion of the study. Two molecules within the same class do not necessarily share the same pharmacokinetic profile or the same clinical efficacy, and no data in this trial concern buspirone. In France, as of 1 September 2026, tandospirone is authorized only in Japan, while buspirone is authorized and marketed, but for anxiety, so using it here for this purpose would be off-label. Authorization, marketing status and reimbursement are three separate questions, and all three vary by country and change over time: readers outside France should verify each one against their own jurisdiction’s current sources before considering this substitution.
What does a benefit on somatization actually mean?
It refers to change on scales of somatic complaints, the PHQ-15 and the Chinese version of the Somatic Symptom Scale. These instruments measure reported symptom burden, not functional status or an objectively documented somatic course. The gap on the PHQ-15 at week 12 is 0.95 point, with a 95% confidence interval of 0.54 to 1.37, on a scale of 0 to 30: it is statistically significant, but what it represents for an individual patient is not established by this study.
Was tolerability measured?
Yes. Adverse events were recorded continuously on a form given to each patient, using the TESS severity scale, and supplemented with laboratory tests, an electrocardiogram and an electroencephalogram. Rates were comparable, 14.9% versus 17.9% (p = 0.641), and events were mostly digestive and clustered in the first two weeks. This comparability holds for frequent events: the sample size cannot support conclusions about rare events, and no figures are published for sodium level, falls, or QT interval.
Is measuring platelet receptors useful in practice?
Not today. It is an exploratory peripheral marker whose link to central neurotransmission remains a hypothesis. It has not been validated as a decision aid or as a monitoring criterion. Notably, the between-arm gap on this marker appears only at week 8, by which point the clinical gap on mood has already disappeared.
Does the absence of a difference at twelve weeks mean the combination adds nothing for mood?
No, and this distinction matters in both directions. This trial did not detect a difference on mood at twelve weeks, which is not the same as demonstrating that none exists: with 134 patients, a difference of small magnitude would go unnoticed. Symmetrically, this result does not demonstrate that the two strategies are equivalent either, since the trial was not designed to establish equivalence.
Annotated bibliography
Source study. Zeng G, Wu S, Zhu C, Xie S, Huang Y, Wei X, Zhang J, Xiao Y, Liu N, Chen H, Chen R. Efficacy of escitalopram plus tandospirone in treating vascular depressive mood disorder: A 12-week single-center randomized controlled trial. Journal of Affective Disorders, 2026, volume 399, article 121133. DOI 10.1016/j.jad.2025.121133. PMID 41500315. Single-center, double-blind, randomized controlled trial against placebo, conducted at Union Hospital of Fujian Medical University, 262 patients screened and 134 analyzed, 67 per arm, twelve weeks, registration ChiCTR2300075407, ethics approval 2019KJCX065, intention-to-treat analysis by generalized estimating equations. Public regional funding: Fujian joint funds for science and technology innovation (2025Y9348), Fujian provincial natural science foundation (2023J01648 and 2025J01117), accompanying funding for the national clinical specialty construction project in neurology (0802714), Fujian clinical research center for precision diagnosis and treatment of neurological disease (2022Y2005); the funders declare no role at any stage. Conflicts of interest: the authors declare no financial interest or personal relationship that could have influenced this work. The first three authors are declared equal contributors and share first authorship.
The team’s prior work. This trial belongs to a series conducted by the same authors on tandospirone in vascular depression, which they cite themselves: Chen H et al., Journal of Psychiatric Research, 2019, volume 114, pages 133 to 140, a single-center randomized pilot trial combining tandospirone citrate and escitalopram; Chen R et al., Journal of Psychiatric Research, 2023, volume 159, pages 274 to 282, tandospirone augmentation in patients with mild cognitive impairment; Chen H et al., CNS Neuroscience & Therapeutics, 2024, volume 30, issue 3, an open-label combination of venlafaxine and tandospirone; Chen H et al., CNS Drugs, 2025, volume 39, issue 7, pages 669 to 683, escitalopram and tandospirone in vascular depression with chronic insomnia. This continuity is both an asset of consistency and a limitation of independence: the week-2 result is presented as confirming a prior observation from the same team, and no replication by an outside group is cited.
Construct framework. The vascular depression definition used draws on Alexopoulos GS et al., Archives of General Psychiatry, 1997, volume 54, issue 10, pages 915 to 922, complemented by the consensus update from Aizenstein HJ et al., BMC Medicine, 2016, volume 14, issue 1. The blinding measurement method draws on Qin Z et al., General Psychiatry, 2024, volume 37, issue 6, article e101578. These references are cited by the authors and verified against the publication’s bibliography; they were not read in full for this analysis.
Editorial collections
Tags
Verified on September 1, 2026 against the full text of the publication and its supplementary material where available. This analysis underwent an independent double reading. The English version was checked for conformity on September 18, 2026, against the figures of the French version and against the source. How we verify what we publish
This analysis is intended for healthcare professionals. It does not constitute a prescribing recommendation and does not replace individual clinical judgment.
Analysis from Psychiatry Evidence Base, evidence-based psychiatry, explained with rigor.