This observational study (n=45) followed people with treatment-resistant depression for up to 5 years and found that adjunctive intranasal esketamine was linked with sustained reductions in depression scores in those who stayed on treatment. No hypomania or mania was seen, and side effects were generally low.
Background
Treatment-resistant depression (TRD) remains a major clinical challenge, and long-term real-world data on intranasal esketamine are limited.
Objectives
To describe long-term dosing patterns and clinical outcomes associated with intranasal esketamine in routine clinical practice.
Design
Single-center observational cohort study with descriptive analyses.
Methods
We analyzed 45 patients with TRD treated with intranasal esketamine in a naturalistic setting, with follow-up extending up to 260 weeks. Esketamine was used adjunctively with ongoing oral antidepressants. Outcomes included the Montgomery-Åsberg Depression Rating Scale (MADRS), the Young Mania Rating Scale (YMRS), and the Glasgow Antipsychotic Side-effect Scale (GASS). Treatment patterns were summarized using person-time (person-weeks).
Results
Baseline MADRS was 40.0 (SD 4.63). Mean MADRS decreased to 22.9 (SD 7.99) at week 4 and to 9.70 (SD 5.35) at week 52, with mean scores remaining around 9-10 at later time points among patients with available follow-up. Eight patients (17.8%) discontinued treatment (lack of efficacy n = 4; intolerable adverse effects n = 3; lost to follow-up n = 1). No treatment-emergent hypomania or manic symptoms were observed; YMRS scores decreased over time. GASS scores were low overall, with median values of 0 from week 13 onwards.
Conclusion
In a real-world TRD cohort with complex comorbidities, intranasal esketamine used adjunctively was associated with sustained symptom improvement and a favorable long-term safety profile among patients who continued treatment, with dosing evolving from induction to individualized maintenance schedules. Findings are limited by the observational design, concomitant treatments, and survivor bias.
Papers cited by this study that are also in Blossom
Rhee, T. G., Shim, S. R., Forester, B. P. et al. · JAMA Psychiatry (2022)
Young, A. H., Llorca, P. M., Fagiolini, A. et al. · British Journal of Psychiatry (2024)
Reif, A., Bitter, I., Buyze, J. et al. · New England Journal of Medicine (2023)
Major depressive disorder is a common and disabling illness, and a substantial proportion of patients do not achieve remission or functional recovery with standard antidepressant approaches. The paper argues that treatment-resistant depression is associated with marked disability, relapse risk and ongoing symptom burden, while earlier trial evidence for intranasal esketamine has been mixed: some pivotal studies did not meet their main endpoints, although other studies suggested benefits for relapse prevention, maintenance treatment and safety over shorter to medium follow-up periods. The authors also note that real-world patients are often more clinically complex than those in controlled trials, with more comorbidity and concomitant medication use, so long-term routine-practice data remain limited. Against this background, Cuomo and colleagues set out to describe long-term dosing patterns, symptom trajectories and safety outcomes in a naturalistic cohort of 45 patients with treatment-resistant depression treated with adjunctive intranasal esketamine in routine clinical practice, with follow-up extending up to 5 years. The study was designed as an exploratory observational analysis to characterise how treatment evolved over time and to provide preliminary real-world information that might inform future inferential studies.
This was a single-centre observational cohort study conducted at the University Hospital of Siena. Patients were enrolled by rolling recruitment between 18 March 2019 and 3 September 2024, with final data collection on 6 November 2024. The authors state that the sample included all eligible patients who initiated esketamine during the recruitment period; no a priori sample size calculation was performed. Recruitment was by convenience sampling from the outpatient psychiatry service, with treating psychiatrists identifying potentially eligible patients and deciding whether to offer esketamine on the basis of clinical judgement, severity, treatment resistance, functional impairment and patient preference. The reporting is said to follow STROBE. Eligible patients were aged 16-75 years and had DSM-5 major depressive disorder meeting criteria for treatment-resistant depression, defined as inadequate response to at least two antidepressant trials of adequate duration and dose. The extracted text lists medical, psychiatric and practical exclusions, including aneurysmal vascular disease, prior intracerebral haemorrhage, uncontrolled hypertension, recent myocardial infarction or unstable cardiac disease, severe hepatic impairment, active psychotic disorder other than psychotic depression, recent substance use disorder, acute suicidal ideation requiring hospitalisation, inability to attend visits, pregnancy or breastfeeding, and lack of consent capacity. Esketamine was administered intranasally as adjunctive therapy to ongoing oral antidepressants, following standard clinical guidance. Most patients started on 56 mg, with some starting on 28 mg and a smaller subset on 84 mg; dose and frequency could be adjusted according to response and tolerability. Treatment typically began with twice-weekly induction and then moved to individualised maintenance schedules, including weekly, biweekly, every 3 weeks and monthly administration. The study records 4607 person-weeks of follow-up in total, with individual follow-up ranging from 1 to 260 weeks, to describe exposure while accounting for different observation lengths. Outcomes were assessed repeatedly at baseline and at 1, 2, 3, 4, 8, 13, 26, 39, 52, 104, 156, 208 and 260 weeks. The primary outcome was change in Montgomery-Åsberg Depression Rating Scale (MADRS) scores over time. Safety and tolerability were assessed with the Glasgow Antipsychotic Side-effect Scale (GASS) and Young Mania Rating Scale (YMRS), together with clinical observation of esketamine-related adverse effects such as dissociation, blood pressure changes, dizziness, nausea and headache. The analysis was purely descriptive, using frequencies, percentages and summary statistics; no inferential hypothesis testing was performed.
The cohort had a baseline mean MADRS score of 40.0 (SD 4.63), indicating severe depression. Symptom scores fell quickly after treatment began, reaching a mean of 22.9 (SD 7.99) at week 4 and 15.6 (SD 6.37) at week 13. Further improvement was reported by week 52, when the mean MADRS was 9.70 (SD 5.35). Among the smaller groups with later follow-up available, mean scores stayed around 9-10 through years 2-5, with the week 260 mean reported as 9.33 in six patients. The authors describe substantial heterogeneity in individual response trajectories early in treatment, but most trajectories converged towards lower symptom scores by around 3 months. Treatment patterns shifted over time from induction to maintenance. The extracted text reports that by week 52, 91.9% of patients were maintained on 28 mg weekly, increasing to 100% by week 156 and remaining stable through year 5. Monthly administration was the most common overall pattern, accounting for 2693 person-weeks (58.5% of total observation time), followed by biweekly dosing at 1108 person-weeks (24.1%), every 3 weeks at 377 person-weeks (8.2%), weekly dosing at 231 person-weeks (5.0%), twice-weekly dosing at 193 person-weeks (4.2%) and every 10 days at 5 person-weeks (0.1%). Eight patients (17.8%) discontinued treatment. Three discontinued because of adverse effects, including severe headache with dissociative symptoms, a hypertensive crisis during or after administration, and intolerable dissociative symptoms. Four discontinued because of lack of efficacy and one was lost to follow-up. The paper states that no treatment-emergent hypomania or mania was observed and YMRS scores generally decreased over time. Side-effect burden measured by GASS was low at baseline and remained low overall, with median scores of 0 from week 13 onwards; mean GASS fell to 1.41 (SD 2.55) at week 13, 1.30 (SD 2.27) at week 52 and 0.333 (SD 0.816) at week 260. Among continuing patients, transient dissociation, blood pressure elevations, dizziness and nausea were commonly observed but were usually tolerated and did not lead to discontinuation.
The authors interpret these findings as supportive real-world evidence that adjunctive intranasal esketamine can be associated with sustained symptom improvement in treatment-resistant depression when patients remain on treatment, with benefit maintained for up to 5 years in this cohort. They emphasise that the pattern of use in practice largely matched regulatory positioning, since esketamine was used mainly as add-on treatment to oral antidepressants rather than as monotherapy. They also note that baseline severity in this cohort was high, with mean MADRS scores around 40, which they suggest may reflect preferential use in more severe and clinically complex patients than those enrolled in many trials. The authors position their findings as complementary to earlier controlled and real-world studies. They contrast the mixed results of the pivotal TRANSFORM trials with the more encouraging maintenance and longer-term findings from other studies, arguing that their observational data add pragmatic information about how treatment unfolds in routine care. However, they stress that the results cannot establish esketamine-specific efficacy because outcomes occurred alongside frequent concomitant antidepressant and psychotropic use, including mood stabilisers and antipsychotics, and because there was no control group. They therefore frame the study as descriptive rather than causal. Several limitations are highlighted. The authors note survivor bias, because outcomes after discontinuation were not systematically followed and the long-term data largely reflect patients who continued treatment. Rolling recruitment meant later time points were based on smaller subsets of early-enrolled patients. Safety assessment was also limited because GASS does not adequately capture esketamine-specific adverse effects such as dissociation and blood pressure changes, and some adverse effects were recorded clinically rather than with validated scales. In addition, convenience sampling and clinician-directed treatment selection may limit generalisability, and the adequacy of prior antidepressant trials was not always confirmed with therapeutic drug monitoring. In terms of implications, the authors suggest that longer, prospective studies with matched controls and more sophisticated analytical methods are needed. They state that future work should account for the ordinal nature of psychiatric rating scales and the heterogeneity of individual response trajectories, rather than relying only on mean group effects. The study is presented as preliminary evidence that may inform future research and clinical decision-making in complex real-world patients.
The authors conclude that this exploratory observational study provides preliminary long-term real-world information on intranasal esketamine in a complex treatment-resistant depression cohort with substantial comorbidity and up to 5 years of follow-up. They state that the descriptive approach reveals response heterogeneity and dosing evolution over time, but that more rigorous prospective studies with control groups and suitable longitudinal methods are needed to clarify longer-term outcomes.
As an exploratory investigation, this study employs descriptive statistics to characterize the sample, treatment implementation, and outcomes without inferential hypothesis testing. The descriptive approach serves two key purposes: (1) to provide clinicians with preliminary real-world information about esketamine use in complex patients typically excluded from controlled trials, and (2) to identify patterns that can guide the design of more rigorous inferential analyses to be conducted on this expanding dataset as additional observations are collected. This naturalistic framework allows for an initial exploration of long-term treatment strategies, potential efficacy signals, and safety considerations in routine clinical practice. This preliminary observational study provides a descriptive analysis of treatment patterns and outcomes in patients diagnosed with treatmentresistant depression (TRD) undergoing esketamine therapy in naturalistic clinical settings. A total of 45 patients were included, with a followup period extending up to 5 years. No a priori sample size calculation was performed for this observational study. The sample size represents all eligible patients initiating esketamine during the recruitment period. Patients were enrolled through rolling recruitment from March 18, 2019, to September 3, 2024, with final data collection on November 6, 2024. Patients were recruited through convenience sampling from the outpatient psychiatry service. Treating psychiatrists identified potentially eligible patients with treatment-resistant depression and assessed eligibility. The decision to offer esketamine was made by the treating psychiatrist based on individual clinical judgment, considering severity, treatment resistance, functional journals.sagepub.com/home/tpp TherapeuTic advances in psychopharmacology impairment, and patient preference. No systematic protocol was used to determine which eligible patients were offered esketamine versus other options, representing potential selection bias. This convenience sampling may have resulted in the selection of patients with particular characteristics that limit generalizability. Each patient's baseline (Week 0) represents their individual treatment initiation date, not a fixed calendar date. Consequently, the maximum potential follow-up duration varied by enrollment date: patients enrolled in 2019-2020 could complete up to 5 years of observation, while those enrolled later had proportionally shorter maximum follow-up opportunities. The varying number of patients with available data at later time points reflects both actual treatment discontinuation and "administrative censoring" (insufficient elapsed time for later-enrolled patients to reach those time points before study closure). Patients were followed through regular clinical encounters for esketamine administration and routine clinical assessment. Standardized rating scales were administered during prespecified timing encounters at baseline and 11 follow-up time points (1, 2, 3, 4, 8, 13, 26, 39, 52, 104, 156, 208, and 260 weeks). Clinical assessments were conducted in a hospital setting, including both inpatients admitted to psychiatric wards and outpatients attending scheduled evaluations. The study was carried out at the University Hospital of Siena, a tertiary care academic center. The reporting of this observational study conforms to the STROBE statement.
Patients aged 16-75 years (with guardian consent for minors) with DSM-5 major depressive disorder meeting treatment-resistant depression criteria were eligible. Treatment-resistant depression was defined as an inadequate response to at least two antidepressant trials of adequate duration and dosage. "Adequate trial" meant: (1) minimum 6 weeks at therapeutic dose, (2) dose within or above therapeutic range per prescribing guidelines, (3) reasonable adherence, and (4) <50% symptom improvement. Therapeutic drug monitoring was not systematically performed. All patients had documented SSRI trials, with most also failing SNRIs and other classes. Median previous trials: 3 (range 2-8). Medical exclusions: aneurysmal vascular disease, intracerebral hemorrhage history, uncontrolled hypertension, recent MI or unstable cardiac disease, and severe hepatic impairment. Psychiatric exclusions: active psychotic disorder (except psychotic depression), active substance use disorder (within 3 months), and acute suicidal ideation requiring hospitalization. Practical exclusions: inability to attend visits, pregnancy/breastfeeding, and lack of consent capacity.
Demographic and clinical characteristics were recorded at baseline, including age, gender, body mass index (BMI), employment status, and psychiatric comorbidities. The Montgomery-Åsberg Depression Rating Scale (MADRS) was used to assess depression severity, with additional measures for suicidality and psychiatric history. Concurrent psychotropic medication use, including antidepressants, mood stabilizers, antipsychotics, benzodiazepines, and GABA analogs, was documented.
Esketamine was administered intranasally as an adjunctive treatment to ongoing antidepressant therapy, following standard clinical guidelines. The initial dose was determined based on patient characteristics, with most patients receiving 56 mg per session. Dose adjustments were made based on clinical response and tolerability, with the possibility of increasing to 84 mg or decreasing to 28 mg if necessary. Treatment frequency varied over time, with an induction phase consisting of twice-weekly administrations followed by individualized maintenance schedules.
Person-time is an epidemiological concept that measures both the number of persons and their time contribution to a study, accounting for varying follow-up durations. In this analysis, personweeks represent the cumulative time patients spent on treatment, allowing for comparison of treatment patterns while adjusting for different observation periods. The metric is calculated by summing the weeks each patient spent on treatment, providing a more accurate representation of treatment patterns than simple patient counts. Patients were followed for a total of 4607 personweeks, with individual follow-up durations ranging from 1 to 260 weeks. Esketamine dosing frequency and cumulative weekly doses were recorded throughout the follow-up period. A detailed analysis of administration frequency trends was conducted, highlighting transitions from intensive induction dosing to extendedinterval maintenance schedules.
The primary outcome measure was the change in MADRS scores over time. Descriptive statistics (mean, standard deviation, median, range, and quartiles) were calculated for MADRS scores at each assessment time point (baseline, 1, 2, 3, 4, 8, 13, 26, 39, 52, 104, 156, 208, and 260 weeks). Individual MADRS trajectories were plotted alongside mean values to illustrate response patterns and heterogeneity in treatment effects.
Safety monitoring combined systematic assessment tools with clinical observation. The Glasgow Antipsychotic Side-effect Scale (GASS; range 0-22, higher scores indicating more side effects) was administered at each follow-up time point. The GASS assesses common side effects of antipsychotic medications, including extrapyramidal symptoms, sedation, weight gain, and sexual dysfunction. Given that 60% of patients were receiving concurrent antipsychotic medications, GASS was selected to monitor potential additive or synergistic side effects in this polypharmacy context. However, GASS does not adequately capture esketamine's characteristic adverse effects, including dissociation, perceptual disturbances, and blood pressure elevations. The Young Mania Rating Scale (YMRS; range 0-60, higher scores indicating more severe manic symptoms) was used to monitor for potential mood elevation or treatment-emergent hypomanic/manic symptoms, given concerns about ketamine's effects on bipolar spectrum patients. Esketamine-specific safety parameters monitored through clinical assessment but not systematically quantified using validated scales included: (1) dissociative symptoms during and immediately following each administration, documented qualitatively in clinical notes when severe or distressing; (2) blood pressure measured before and after each administration per protocol; and (3) common adverse effects including nausea, dizziness, and headache elicited through clinical interview. All adverse events leading to treatment discontinuation or dose modifications were documented.
This study employed purely descriptive statistical approaches to characterize the sample and summarize treatment patterns and outcomes. No inferential statistical testing or hypothesis testing was performed, in alignment with the exploratory and descriptive aims of this preliminary investigation. Baseline demographic and clinical characteristics were presented using frequencies and percentages for categorical variables (gender, comorbidities, medication use), while continuous variables were summarized using means, standard deviations, medians, ranges, and quartiles as appropriate (age, BMI, MADRS scores). Treatment patterns were analyzed using absolute and relative frequencies, with person-time calculations (personweeks) employed to account for varying follow-up durations when describing treatment intensity and dosing strategies. Temporal changes in MADRS, YMRS, and GASS scores were visualized through line graphs displaying individual trajectories alongside mean values at each time point, with descriptive statistics (mean, SD, min, max, quartiles) calculated for each assessment interval. Side effect patterns were further characterized through frequency distributions and percentage calculations to identify predominant adverse effect categories. No substantial differences in baseline demographics or clinical characteristics were observed between patients who discontinued treatment (n = 8) and those who continued (n = 37).
Analysis of initial dosing patterns showed both adherence to and variation from recommended Long-term follow-up data showed further consolidation of this pattern. By week 52, 91.9% of patients were maintained on 28 mg weekly, and this proportion reached 100% by week 156 (3 years) and remained stable through year 5. Analysis of the administration frequency distribution (Figure) reveals that monthly dosing was predominant, accounting for 2693 person-weeks (58.5% of total observation time). Biweekly dosing represented the second most common frequency with 1108 person-weeks (24.1%), likely reflecting an intermediate maintenance phase for many patients. Administration every 3 weeks accounted for 377 person-weeks (8.2%). More frequent administration patterns were less common in terms of cumulative person-time: weekly dosing represented 231 person-weeks (5.0%), while twice-weekly administration accounted for 193 person-weeks (4.2%). The least common frequency was administration every 10 days, accounting for only 5 person-weeks (0.1%).
The temporal evolution of depressive symptoms demonstrated a marked initial improvement. By week 4, mean MADRS scores decreased substantially to 22.9 (SD = 7.99), representing a 42.8% protocols, with clinical judgment guiding individualization. The majority of patients (n = 31, 68.9%) were initiated on the standard 56 mg dose, in line with current guidelines. A subset of 9 patients (20.0%) began treatment at the lower 28 mg dose. Specific rationales for dose selection were not systematically documented. Interestingly, a small group of patients (n = 5, 11.1%) was started directly at the higher 84 mg dose. The weekly esketamine dose is calculated by multiplying the single administered dose (28, 56, or 84 mg) by the frequency of administration per week. For example, a patient receiving 56 mg twice weekly would have a total weekly dose of 112 mg, while a patient receiving 84 mg once weekly would have a total weekly dose of 84 mg. This calculation provides a standardized way to reduction from baseline. This robust early response continued through the first 3 months, with mean scores reaching 15.6 (SD = 6.37) at week 13. Further improvement was observed at 6 months (mean = 13.2, SD = 5.61), with scores stabilizing around the 1-year mark at 9.70 (SD = 5.35). Notably, these therapeutic gains were maintained through the extended follow-up period, with mean scores remaining stable between 9 and 10 points through years 2-5, though with progressively smaller sample sizes. Individual trajectories, represented by the light blue lines in the figure, reveal considerable heterogeneity in response patterns, particularly during the first month of treatment. While most patients exhibited rapid initial improvement, the slopes of decline varied substantially. Some individuals demonstrated dramatic early responses, while others showed more gradual improvement. Despite this early variability, the majority of trajectories converged toward lower severity scores by month 3, with relatively stable long-term outcomes. The visualization of individual patterns suggests that while response timing may vary, sustained improvement was achievable across diverse response patterns (Figure). There was a progressive reduction in score variability over time, with the standard deviation decreasing from 7.99 at week 4 to 3.01 at the 5-year mark (week 260, mean = 9.33, n = 6).
Three patients (6.7%) discontinued treatment due to adverse effects. Specific adverse events were as follows: (1) severe headache and dissociative symptoms at 2 weeks (n = 1, 2.2%), () hypertensive crisis during/after administration at 2 weeks (n = 1, 2.2%), and (3) intolerable dissociative symptoms at 1 month (n = 1, 2.2%). All adverse events had a clear temporal relationship to esketamine administration, were consistent with the known adverse effect profile, and resolved after treatment discontinuation without sequelae. No serious adverse events requiring hospitalization or resulting in persistent disability occurred. Among patients who continued treatment, transient dissociative symptoms, blood pressure elevations (managed through monitoring), dizziness, and nausea were commonly observed during acute administration periods but were generally tolerated and did not lead to discontinuation. These common but tolerated adverse effects were not systematically quantified using validated measures. showing some variability in the early weeks, demonstrated remarkable consistency in their downward trend, with no cases of significant manic elevation observed throughout the follow-up period. Descriptively, we observed variance compression in YMRS scores over time, with standard deviations decreasing from 2.33 at baseline (n = 45) to 0.516 at week 260 (n = 6; Figure). This pattern likely reflects the small sample size at later time points, floor effects of the scale, and survivor bias, rather than representing a statistically comparable change. These descriptive observations are presented to characterize the available data and inform future inferential analyses. This pattern of decreasing variance and non-normal distribution at later time points would complicate traditional parametric longitudinal analyses, suggesting that alternative approaches would be required for formal inferential testing in future analyses. Formal side effect assessment utilized the Glasgow Antipsychotic Side-effect Scale (GASS; range 0-22) and Young Mania Rating Scale (YMRS; range 0-60). GASS was selected given high rates of concurrent antipsychotic use (60%), but does not capture esketamine-specific adverse effects (dissociation, blood pressure changes), limiting the comprehensiveness of our tolerability characterization. YMRS monitored for potential mood elevation or treatment-emergent manic symptoms. Our analysis of side effects using the GASS revealed a favorable long-term tolerability profile (Figure). At baseline, patients demonstrated a relatively low side effect burden (mean GASS = 2.60, SD = 2.86, median = 2.0), which remained stable during the initial treatment phase (weeks 1-4, median consistently at 2.0 despite slight fluctuations in means). The distribution of GASS scores was notably right-skewed throughout the study period, with a substantial proportion of patients reporting minimal side effects, while a smaller subset experienced more YMRS, Young Mania Rating Scale. 12 journals.sagepub.com/home/tpp TherapeuTic advances in psychopharmacology pronounced adverse reactions. This skewness makes the median a more appropriate measure of central tendency than the mean, which is disproportionately influenced by the few higher values. A notable pattern emerged between week 4 and week 13, where individual trajectories revealed considerable crossing of paths-some patients showing improvement, while others experienced transient worsening of side effects-before converging toward lower scores. This heterogeneity in individual responses during the transition from acute to maintenance treatment would have been obscured in analyses focused solely on aggregate measures. Despite this initial variability, a significant improvement was observed by week 13, with median GASS scores decreasing to 0 and mean scores to 1.41 (SD = 2.55). The long-term follow-up demonstrated further improvements in tolerability. By week 52, the mean GASS score had decreased to 1.30 (SD = 2.27), with continued reduction through extended follow-up (mean = 0.333, SD = 0.816 at week 260). The distribution of scores showed that while many patients experienced no side effects (median = 0 from week 13 onward), there was persistent individual variability in susceptibility to adverse effects, as evidenced by maximum scores of 6-8 points through the first 2 years. This pattern of predominantly zero values with occasional higher scores creates zero-inflated distributions at later time points, further complicating traditional statistical approaches that assume normality.
This real-world study provides valuable evidence on the use of intranasal esketamine in patients with treatment-resistant depression (TRD), including those with complex psychiatric and medical comorbidities, complementing the existing empirical literature.Our treatment approach, with esketamine predominantly used as adjunctive therapy to ongoing antidepressants (82% of patients), aligns with regulatory approval conditions in Europe and North America, where esketamine is indicated as adjunctive treatment rather than monotherapy. This mirrors the design of pivotal trials (TRANSFORM, SUSTAIN) that evaluated esketamine plus antidepressant versus placebo plus antidepressant, and reflects typical realworld implementation, in which esketamine is added to inadequate-response regimens rather than replacing existing treatments. Despite these difficulties, esketamine was associated with GASS, Glasgow Antipsychotic Side-effect Scale. sustained clinical improvement in patients who continued treatment as shown by a decrease from a 40.0 baseline MADRS score to around 9-10 after 1 year, with the therapeutic effect maintained for a maximum of 5 years. Moreover, it is noteworthy that the mean MADRS score at baseline was 40, whereas in most clinical trials it is approximately 35. This likely reflects the fact that, in our real-world setting, esketamine was preferentially selected for patients presenting with particularly severe depressive symptoms. A critical consideration for interpreting our findings is that the observed clinical improvements occurred in the context of polypharmacy, with 82.2% of patients receiving concurrent antidepressants and 60%-64% receiving other psychotropic medications (mood stabilizers, antipsychotics). This treatment context has several important implications. First, observed improvements reflect the combined effect of the entire treatment regimen rather than esketamine in isolation, and without a control group receiving identical concurrent treatments, we cannot determine the specific contribution of esketamine to outcomes. Second, our treatment approach aligns with both regulatory approval (esketamine as adjunctive therapy to antidepressants) and real-world clinical practice, making our findings generalizable to typical implementation contexts. Third, this confounding underscores why our purely descriptive approachrather than causal or efficacy claims-is methodologically appropriate. We characterize real-world outcomes when esketamine is added to existing treatment regimens, providing clinically useful prognostic information, but we cannot establish esketamine-specific efficacy. The high rate of concurrent psychotropic medications also reflects the clinical complexity and treatment resistance of our cohort. Many patients required multiple agents to manage severe depression, comorbid anxiety or personality disorders, previous partial responses, or medication-related side effects. This polypharmacy, while complicating attribution of treatment effects, accurately represents the patient population most likely to receive esketamine in routine clinical practice: those with complex, treatment-resistant presentations requiring multifaceted pharmacological approaches.
Our purely descriptive analytical approach, while limiting inferential claims, was chosen deliberately given the methodological challenges inherent in this preliminary investigation. Standard inferential tests (e.g., paired t-tests, Wilcoxon tests) would be inappropriate given: (1) repeated measures violating independence assumptions,(2) differential attrition introducing survivor bias, and (3) the ordinal nature of rating scales.The observed heteroskedasticity, floor effects, and non-normal distributions at later time points further violate assumptions of parametric tests, potentially leading to biased estimates and invalid inferences.More sophisticated approaches-such as mixedeffects models or Bayesian hierarchical models incorporating item response theory (IRT) to model scale-level measurement characteristics-would be needed for confirmatory analyses. We are developing Bayesian IRT models to properly handle ordinal scale properties, longitudinal dependencies, and missing data; these will be presented in future work. For this preliminary report, transparent descriptive presentation allows examination of individual trajectory heterogeneity that would be obscured by population-averaged effects.Our visualization revealed substantial individual variation in response patterns-with some patients showing rapid improvement while others demonstrated gradual changes-underscoring the importance of personalized treatment approaches.These individual differences, while complicating grouplevel statistical inference, provide clinically meaningful information for treatment individualization that aggregate analyses might mask. Our findings must be interpreted within the context of mixed and debated evidence from randomized controlled trials. The failure of two of three TRANSFORM trials to achieve statistical significance,combined with questions about the clinical meaningfulness of statistically significant findings in TRANSFORM-2,indicates that esketamine's efficacy profile remains uncertain. Our observational data cannot resolve this uncertainty, as we lack a control group and cannot isolate esketamine effects from concurrent treatments (82% receiving antidepressants). However, our findings provide complementary real-world evidence that, in routine clinical practice where esketamine is added to existing regimens in complex treatment-resistant patients, 82% of patients continue treatment with observed journals.sagepub.com/home/tpp TherapeuTic advances in psychopharmacology clinical improvements over extended follow-up. This pragmatic effectiveness information-while not establishing esketamine-specific efficacymay inform shared decision-making about treatment trials in similar patient populations.
Strengths of this study include the naturalistic clinical cohort characterized by a high burden of psychiatric and medical comorbidities, capturing patients who are frequently excluded from randomized controlled trials. This enhances the ecological validity of the findings and provides clinically relevant insights into the real-world implementation of intranasal esketamine in treatment-resistant depression. A major strength is the extended follow-up duration of up to 5 years, which allows characterization of long-term treatment patterns, dosing evolution, and clinical outcomes that remain largely unexplored in the existing literature. Unlike controlled trials with fixed protocols and limited observation windows, this study reflects routine clinical decision-making, including individualized maintenance strategies over time. The use of a person-time (person-weeks) analytical framework represents an additional methodological strength, enabling an accurate description of treatment exposure while accounting for heterogeneous follow-up durations resulting from rolling recruitment. Furthermore, the visualization of individual symptom trajectories alongside group-level descriptive statistics allows detailed characterization of response heterogeneity. This approach highlights clinically meaningful differences in response timing and durability that would be obscured by analyses focused solely on mean effects. Together, these strengths provide complementary real-world evidence on the long-term use of intranasal esketamine in complex patients with treatment-resistant depression and offer a valuable foundation for future hypothesis-driven and inferential investigations.
Several important limitations warrant acknowledgment. First, while the rolling recruitment design reflects real-world clinical practice, treatment discontinuation introduces survivor bias. Although the discontinuation rate was relatively modest (17.8%), patients who discontinued treatment due to lack of efficacy or intolerable adverse effects likely experienced less favorable outcomes than those who continued treatment. As outcomes after treatment discontinuation were not systematically assessed, the present findings primarily characterize outcomes among treatment continuers (82.2% of initiators) rather than providing population-level intent-to-treat effectiveness estimates. This may result in overestimation of clinical benefit and underestimation of adverse effects. Second, heterogeneity in follow-up duration resulting from rolling recruitment limits the interpretability and generalizability of long-term outcomes. Data from later time points (years 3-5) reflect a progressively smaller subset of earlyenrolled patients, such that long-term estimates primarily represent treatment continuers rather than the full cohort. Third, the single-center observational design and absence of a control group preclude causal inference regarding esketamine-specific efficacy. Esketamine was administered adjunctively with ongoing antidepressants and, in many cases, additional psychotropic medications, including mood stabilizers and antipsychotics. Consequently, observed clinical improvements reflect the combined effect of the entire treatment regimen rather than esketamine in isolation. It is not possible to disentangle the effects of esketamine from synergistic pharmacological interactions, delayed treatment effects, spontaneous symptom fluctuation, or increased clinical attention. This confounding is inherent to naturalistic observational designs and underscores the rationale for adopting a purely descriptive analytical approach in this study, which aims to characterize real-world outcomes rather than establish efficacy. Fourth, safety assessment has important methodological limitations. The GASS, selected due to the high prevalence of concurrent antipsychotic use, does not adequately capture esketamine-specific adverse effects, such as dissociative symptoms, perceptual disturbances, and blood pressure changes. Validated dissociation measures, including the Clinician-Administered Dissociative States Scale, were not employed, and many adverse effects were monitored through clinical observation and spontaneous reporting rather than systematic elicitation using validated instruments. This incomplete safety characterization limits direct comparison with controlled trial safety profiles and underscores the need for esketamine-specific tolerability measures in future observational research. Fifth, convenience sampling and cliniciandirected treatment selection introduce potential selection bias. The atypical male predominance observed in the sample may reflect referral patterns, unmeasured clinical factors, or chance, and limits generalizability to broader treatment-resistant depression populations. Finally, documentation of prior treatment adequacy has limitations. Therapeutic drug monitoring was not systematically performed, and information regarding prior treatment duration and adherence partially relied on patient self-report. As a result, some cases classified as treatment-resistant depression may have reflected suboptimal exposure to previous antidepressant trials.
This observational study offers several notable strengths, including its focus on a naturalistic cohort with high psychiatric and medical comorbidities and an extended follow-up period of up to 5 years. The exploratory, descriptive nature of our analysis serves a dual purpose: primarily to guide subsequent inferential approaches on this dataset, and to provide valuable preliminary clinical insights in an area where long-term naturalistic data are scarce. While not designed for hypothesis testing, our findings illuminate treatment patterns and outcomes in real-world settings that complement controlled trial data, capturing the complexity of clinical decisionmaking in routine practice, which may inform clinical considerations. Our descriptive analytical approach with visualization of individual trajectories reveals response heterogeneity and temporal patterns that would be obscured by aggregate analyses alone, highlighting methodological limitations in the existing literature where meanbased analyses predominate. Our exploratory findings suggest that more sophisticated methodological approaches may be necessary to fully capture and model the complex patterns observed in long-term psychiatric treatment data. Future research would benefit from larger prospective studies with matched control groups, pre-specified analytical plans that account for the distributional properties of psychiatric rating scales, and statistical approaches that can accommodate both group-level trends and individual response trajectories.
The study was approved by the CET Comitato Etico Regione Toscana -Area Vasta Sud Est (protocol no. 23809). All participants gave written informed consent after a detailed explanation of the research objectives, procedures, and possible side effects. This study adhered to the tenets of the Declaration of Helsinki and the guidelines of Good Clinical Practice.
Create a free account to open full-text PDFs.