Submit to this Journal Review for this Journal Propose a Special Issue

Article Menu

Share Help Cite Discuss in SciProfiles

Open AccessReview

Peer-Review Record

Resting State Heart Rate Variability in Depression: An Introductory Narrative Review of Cross-Sectional and Longitudinal Evidence

J. Pers. Med. 2026, 16(2), 87; https://doi.org/10.3390/jpm16020087

by Evelien Van Assche¹

and Carmen Schiweck^2,3,*

Reviewer 1:

Marcel Alexandru Gaina

Reviewer 2: Anonymous

J. Pers. Med. 2026, 16(2), 87; https://doi.org/10.3390/jpm16020087

Submission received: 30 October 2025 / Revised: 10 January 2026 / Accepted: 14 January 2026 / Published: 3 February 2026

(This article belongs to the Special Issue Multiple Biomarkers for the Diagnosis and Precision Treatment of Depression)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear esteemed colleagues Van Assche and Schiweck,

It is with genuine appreciation and scholarly enthusiasm that I approach the review of your manuscript titled "heart rate variability as a biomarker for depression: a narrative review of cross-sectional and longitudinal evidence." your contribution arrives at a pivotal moment when precision medicine demands robust, clinically applicable biomarkers for depression, and your comprehensive synthesis addresses this need with remarkable depth and clinical relevance. The manuscript demonstrates substantial scholarly rigor in synthesizing complex physiological mechanisms, meta-analytic evidence, and methodological considerations that are essential for advancing the field. Your work has the potential to significantly influence how clinicians and researchers conceptualize and utilize HRV measurements in depression research and clinical practice.

The systematic organization of evidence from cross-sectional studies through longitudinal investigations, treatment interventions, and confounding factors provides readers with a progressive understanding that builds logically from foundational concepts to clinical applications. Your acknowledgment of both the promise and limitations of HRV as a biomarker reflects the intellectual honesty and scientific maturity that characterizes high-quality scholarship. This balanced perspective will undoubtedly guide future research directions and prevent premature clinical implementation without adequate validation.

Advisory notes for manuscript enhancement

Structural and formatting refinements

Page 1, lines 1-2: the title formatting requires standardization. Advisory: revise to ensure consistent capitalization throughout, particularly "narrative" should maintain lowercase except as first word. This aligns with mdpi journal formatting standards.

Page 1, line 5: email address contains formatting error with incorrect symbol "evelien.vanassche@)ukmuenster.de". Please correct to proper email format "evelien.vanassche@ukmuenster.de" to ensure correspondence functionality.

Page 1, line 26: keywords section exhibits inconsistent formatting with "keyword 1; heart rate variability 2; depression 3; longitudinal; 4 treatment; 5 biomarker". Advisory: standardize to either "keywords: heart rate variability; depression; longitudinal studies; treatment; biomarker" or follow mdpi-specific keyword formatting guidelines. The current numbering system appears erroneous and detracts from professional presentation.

Methodological transparency enhancements

Page 1, lines 60-64 and throughout: while the manuscript acknowledges being a narrative review highlighting "meta-analyses and key studies," the absence of explicit search strategy documentation represents a significant methodological gap for contemporary narrative reviews. Advisory: incorporate a dedicated methodology subsection (approximately 150-200 words) immediately following the introduction that addresses: (1) databases searched and date ranges, (2) search terms employed, (3) rationale for study selection, and (4) acknowledgment that this represents a narrative rather than systematic approach. This enhancement would elevate methodological rigor without compromising the narrative review format.

Page 2, line 112: the statement "we do not claim completeness of all available studies published in the literature" appropriately acknowledges the narrative approach. Advisory: expand this disclaimer to appear earlier, preferably in the introduction (around line 61-62), and articulate more explicitly the criteria guiding your selection of meta-analyses and "key studies." what constitutes "specific significance" in your evaluation framework? Providing these operational definitions strengthens transparency and allows readers to contextualize your interpretive choices.

Content depth and analytical rigor

Page 3, lines 100-114: the discussion of HRV parameters acknowledges that "hf-HRV and rmssd are often used to describe 'vagal tone'" with appropriate caveats about this interpretation being "only partially correct." advisory: strengthen this critical analysis by dedicating 2-3 additional sentences to explicating the specific assumptions that must be met (you mention "within normal breathing frequencies" but this deserves elaboration). Consider citing quigley et al. More prominently here, as this represents a crucial interpretive framework that impacts the entire manuscript's subsequent discussions.

Page 6-7, table 2: this comprehensive synthesis represents a major strength of the manuscript. Advisory: enhance the table's utility by adding a column for "heterogeneity statistics" (i²) for each meta-analysis, as you reference high heterogeneity in line 191-192 but do not systematically present these values. Additionally, consider adding brief footnotes explaining contradictory findings (e.g., why brown et al. 2018 shows only lf-HRV reduction in older adults while other populations show hf-HRV reductions). This would transform an already excellent table into an exceptional reference resource.

Page 8, lines 202-233: the longitudinal observational studies section presents crucial evidence but lacks critical synthesis. Advisory: after presenting individual study findings, dedicate a paragraph (approximately 100-150 words) to synthesizing mechanistic implications. Specifically, the finding that "baseline HRV predicted depressive symptoms but not vice versa" (lines 213-218) carries profound theoretical significance for understanding directionality of effects. Explore whether this pattern suggests HRV as a vulnerability marker versus a state marker, and discuss implications for the neurovisceral integration model mentioned earlier.

Page 8-9, lines 234-295: the treatment studies sections (pharmacological and non-pharmacological) present somewhat contradictory findings without adequate reconciliation. Advisory: create a synthesizing paragraph at line 295 that directly addresses why some studies show HRV changes with treatment while others do not. Consider whether differences in depression severity at baseline, treatment duration, or specific HRV parameters measured might explain heterogeneity. The hartmann et al. Finding that change in HRV correlated with symptom reduction (lines 246-248) versus brunoni et al. Finding no change (lines 252-255) requires deeper analytical integration rather than sequential presentation.

Page 9, lines 263-295: the biofeedback section introduces exciting intervention possibilities. Advisory: expand the discussion of mechanisms by which biofeedback might improve both HRV and depressive symptoms. Does biofeedback represent a direct intervention on autonomic dysfunction, or might improvements in HRV reflect increased self-efficacy, attention regulation, or other psychological mechanisms? Exploring these alternative explanations strengthens the critical analysis.

Confounding factors: enrichment opportunities

Page 10, lines 301-324: the genetics section provides valuable context. Advisory: integrate discussion of how genetic predisposition might interact with environmental stressors (briefly mentioned in lines 319-321) more explicitly. Consider whether individuals with genetic vulnerability to low HRV might show different depression trajectories or treatment responses. This connects genetics to clinical utility rather than presenting it as purely descriptive information.

Page 10-11, lines 325-349: the sex differences discussion appropriately uses "sex" terminology for biological factors. Advisory: as you correctly note in lines 347-349, future studies should assess both sex and gender effects. Enhance this section by providing 2-3 specific examples of how gender (social/cultural factors) might influence the HRV-depression relationship independently of biological sex. For instance, gender differences in stress exposure, coping strategies, or healthcare-seeking behavior could moderate associations.

Page 11, lines 350-370: the age effects are well-documented. Advisory: integrate the non-linear relationship with age more explicitly into clinical recommendations. Specifically, the finding that associations may differ in populations over 70 years (lines 354-356) has direct implications for interpreting the an et al. Contradictory findings discussed earlier (lines 221-226). Cross-reference these sections more explicitly to demonstrate how confounding factors explain apparent contradictions in the literature.

Page 11-12, lines 371-396: lifestyle factors are comprehensively addressed. Advisory: prioritize discussion of physical activity and bmi given their modifiable nature and clinical relevance. Consider adding 2-3 sentences discussing practical implications: should depression studies routinely assess and control for physical activity? How might lifestyle interventions targeting HRV improvement compare to or complement antidepressant treatment? This moves from description to clinical translation.

Page 12, lines 397-415: antidepressant medication effects receive appropriate attention given their importance as confounders. Advisory: the discrepancy between naturalistic studies showing negative effects and meta-analyses in unmedicated patients (lines 408-411) deserves more analytical attention. Propose 2-3 testable hypotheses that might explain this discrepancy: might antidepressant effects on HRV differ in treatment responders versus non-responders? Could the timing of HRV assessment relative to treatment initiation matter? Offering these mechanistic possibilities strengthens the scholarly contribution.

Methodological considerations section

Page 12, lines 416-434: this section provides practical guidance but could be more systematic. Advisory: reorganize into a structured framework using subheadings: "participant characteristics to assess," "environmental controls," "measurement protocols," and "analytical considerations." within each subsection, distinguish between essential controls (those that must be implemented) versus recommended controls (those that enhance but are not critical). This transforms guidance from a list into a decision-making framework.

Specifically at lines 430-434: the discussion of wearable technology and ppg-derived data acknowledges higher signal-to-noise ratio. Advisory: expand with specific recommendations for minimum data quality standards (e.g., acceptable artifact percentages, minimum recording lengths for wearable data). Additionally, discuss whether wearables might be more suitable for within-subject longitudinal tracking versus between-subject comparisons, given their different measurement properties compared to ecg.

Conclusions section

Page 12-13, lines 435-451: the conclusions appropriately balance promise with caution. Advisory: strengthen the final paragraph by offering 2-3 specific research priorities that emerge from your synthesis. For example: (1) adequately powered longitudinal studies that assess HRV, depression severity, and treatment outcomes while controlling for key confounders, (2) investigation of HRV as a predictor of treatment response across different treatment modalities, (3) validation of wearable-derived HRV measurements against gold-standard ecg in clinical populations. Providing these concrete directions enhances the manuscript's impact on guiding future research.

Additionally at lines 442-444: the statement about sensitivity but weak specificity acknowledges a crucial limitation. Advisory: elaborate slightly by discussing whether HRV might be better conceptualized as a transdiagnostic marker of autonomic dysfunction or stress vulnerability rather than a depression-specific biomarker. This reframing aligns with research domain criteria (rdoc) approaches and dimensional models of psychopathology, positioning your work within contemporary psychiatric nosology discussions.

Reference quality and completeness

Throughout: the reference list demonstrates comprehensive coverage with appropriate citation of seminal works and recent meta-analyses. Advisory: verify that all references follow mdpi formatting requirements precisely, particularly for journal abbreviations. Specifically check references with unusual formatting (e.g., reference formatting at lines 463-621).

Page 14, reference 21: koenig et al. 2016 is cited but appears to be an earlier version of the baumeister-lingens et al. 2023 updated analysis. Advisory: clarify the relationship between these works in the text to avoid confusion about whether these represent independent evidence or updated analyses of overlapping data.

Tables and figures

Page 3, tables 1a and 1b (lines 115-120): these tables provide essential reference information. Advisory: consider adding a "clinical interpretation" column that briefly explains the clinical significance of elevated versus reduced values for each parameter. This enhances accessibility for clinician readers who may be less familiar with HRV technicalities.

Page 6-7, table 2 (lines 195-201): as mentioned earlier, this represents exemplary synthesis. Advisory: consider whether a supplementary table categorizing studies by key methodological features (sample size, depression diagnostic criteria, HRV recording duration, controlled confounders) might further enhance utility for readers planning future studies.

Language and clarity refinements

Page 1, line 30: the pascal quote provides engaging opening but the citation format "(pensées, 282)" may not be immediately clear to all readers. Advisory: expand to "pensées, section 282" or provide a more complete citation format.

Page 2, line 89: the phrase "considered as less adaptable" could be more precisely stated. Advisory: revise to "reflects reduced adaptability to environmental demands" for clearer mechanistic interpretation.

Page 4, line 143: "g = -0.46" - ensure consistency in reporting effect sizes throughout. Some sections report hedges' g while others report Cohen's d. Advisory: briefly note in the text when effect size metrics differ across meta-analyses and clarify that hedges' g provides bias-corrected estimates particularly valuable for small sample studies.

Page 9, line 261: "stratification by change in rsa may thus help to identify a biological subgroup" represents an intriguing finding. Advisory: expand this observation with 2-3 sentences discussing potential mechanisms and clinical implications. Might treatment-related HRV changes reflect restoration of autonomic balance specifically in certain depression subtypes? This deserves more analytical attention.

Overall assessment considerations

Your manuscript makes a substantive contribution to understanding HRV as a potential biomarker for depression. The comprehensive coverage of cross-sectional associations, emerging longitudinal evidence, and thorough discussion of confounding factors provides readers with both current knowledge synthesis and guidance for advancing the field. The intellectual honesty in acknowledging limitations and contradictory findings enhances credibility and positions this work as a balanced, scholarly resource.

The primary enhancements needed center on methodological transparency regarding study selection, deeper analytical integration of contradictory findings, and expansion of clinical translation implications. These refinements will elevate an already strong manuscript into an exceptional reference that guides both research and clinical application.

Your attention to confounding factors represents a particular strength that distinguishes this review from simpler summaries. However, transforming descriptive presentation of confounders into analytical discussion of how they explain literature inconsistencies would substantially enhance scholarly impact. Similarly, moving from sequential presentation of treatment studies to synthetic integration that reconciles contradictory findings will strengthen the manuscript's contribution to mechanistic understanding.

The practical guidance for study design (section 6) provides valuable direction, but would benefit from more systematic organization and distinction between essential versus optional methodological controls. Researchers with varying resource levels need clarity about minimum standards versus ideal protocols.

I commend the authors for your scholarly contribution and advise you to confront any of my advisory notes if deemed outside your viewpoint.

With collegial respect,

Serving Peer Reviewer at JPM MDPI

Comments for author File: Comments.pdf

Author Response

----See also the attachment for a word format with colour.----

Reviewer 1

Advisory notes for manuscript enhancement

Structural and formatting refinements

Many thanks for your kind words. We have adjusted this, but referring to previous publications in mdpi, we have opted for keeping everything in capitals except the connecting words (see for example the publication : “Outcomes of Minimally Invasive Mitral Valve Surgery Using a Multidisciplinary Team Approach: A Single-Center Experience”

Our apologies for this formatting error. This has been corrected.

Many thanks for your comment. We have adapted the proposed changes.

Methodological transparency enhancements

We thank the reviewer for this insightful and constructive suggestion. In response, we have added a dedicated Methods subsection immediately following the Introduction to enhance transparency and methodological rigor while preserving the narrative review framework. The new section explicitly outlines the databases searched, search terms and time frame, study selection rationale, and clarifies the narrative (rather than systematic) nature of the review.

For this narrative review, the authors conducted a literature search using PubMed and Google scholar (adapted terms), with no restriction on the start date and coverage through March 30, 2025. The search strategy employed the following terms: ("heart rate variability" OR "HRV" OR “ECG”) AND ("depressi*") in the title or abstract to identify meta-analyses, and filters were set to identify meta-analysis. For longitudinal studies the terms (“follow*” OR “longitudinal”) in the title/abstract were added with an AND term in the search string and filters were set for study designs (Adaptive Clinical Trial, Clinical Study, Clinical Trial, Clinical Trial, Phase I, Clinical Trial, Phase II, Multicenter Study, Observational Study). Studies were considered eligible if they met one of the following criteria: (1) meta-analyses examining associations between heart rate variability and depression; (2) cross-sectional studies involving individuals with a diagnosis of depression or clinically relevant depressive symptoms, with a sample size greater than 50 participants and no experimental manipulation (e.g., stress tasks, cognitive challenges, pharmacological interventions or comparable); or (3) longitudinal studies including pre-post assessments. Consistent with the narrative nature of this review, smaller or methodologically diverse studies were also included when deemed informative by the authors, particularly in areas where meta-analytic evidence is lacking, specific age groups are underrepresented, or unique study designs, outcomes, or covariates provide additional conceptual insight. This approach was chosen to synthesize key findings and emerging themes rather than to provide a comprehensive or systematic appraisal of the literature. If two meta-analyses were published within a short timeframe, comprising a majority of similar studies, a consensus was found between both authors which article to include.

Many thanks, we have tried to explain this both at the end of the Introductions section and in the methods section as follows and as described above.

Introduction:

Importantly, as this is a narrative review, it is not meant to replace but rather complement existing meta-analyses. We make no claim at completeness as our aim was not to provide an extensive overview, but to provide researchers starting in the field with an introductory overview of studies that might be missed in other literature regarding the topic.

Content depth and analytical rigor

Many thanks for this important suggestion. We have done so as follows, referring to Quigley et al:

These assumptions -and limitations- have recently been summarized in a hallmark paper by Quigley et al. [16]. In brief, to be used as an indirect index of cardiac vagal modulation, recordings need to be made at rest, with minimal postural changes throughout the recording. Importantly, to be used as a proxy for cardiac vagal modulation, respiration must fall within the typical adult frequency range (approximately 7-24 breaths a minute) and respiration should be explicitly measured or controlled. Furthermore, to be reliable, only high-quality ECG recordings should be used. Provided that the phasic vagal modulation scales with the tonic parasympathetic activity, HF-HRV may serve as a reasonable proxy for individual differences or condition-related changes in cardiac vagal control, though it should not be interpreted as a direct measure of absolute vagal tone, see Quigley et al. [16]. In contrast, LF and the LF/HF ratio, which were previously used to assess the sympathetic modulation and sympatho-vagal balance, have been criticised and are not thought to reliably measure this activity, due to the LF-frequency band largely reflecting baroreflex activity and likely having mixed sympathetic-parasympathetic input. The scope of this review is therefore to focus mainly on measures reflecting cardiac vagal activity modulation.

Many thanks for your kind words. We have done so and added heterogeneity measures reported in the respective meta-analyses (I2) to the table, where available. As heterogeneity within meta-analyses is medium to high (often above 50%), interpreting differences between meta-analyses, spanning from childhood to old age, seems prone to speculation. Hence, we added this critical remark as a footnote, and offer possible interpretations for the conflicting results by Brown et al. We did however not feel comfortable with adding other explanations for critical results, as these are speculative and would require more robust methods.

The Footnotes read as follows:

¹ Overall, heterogeneity within reported meta-analyses is high I²> 50%. This complicates interpretation within meta-analyses and of findings between meta-analysis. Heterogeneity may be due to patient-related factors including age, sex, BMI, ancestry, etc. It is however notable that two meta-analyses, namely Koch et al. and Kemp et al., who both included patients with current depression and without antidepressant medication, show quite low rates of heterogeneity, particularly for HF-HRV, highlighting the important impact antidepressant medication can infer (see also section on antidepressant medication as a confounder and in particular, the NESDA study.) However Wu et al., also excluded use of antidepressants and show high levels of heterogeneity.

² Brown et al. (2018) contrasts with studies reporting reduced HF-HRV. This discrepancy may reflect differences in sample characteristics and/or respiration: because HF-HRV depends on breathing frequency, values can be biased if respiration falls outside the conventional HF band. Monitoring respiratory rate is therefore especially important in older samples. Other explanations are possible.

Many thanks for the excellent suggestion, we have done so as follows and have additionally added a table with the study characteristics to facilitate an overview (Table 3).

Together, these longitudinal observational findings extend beyond cross-sectional associations by clarifying the directionality of effects between autonomic regulation and depressive symptoms. The observation that baseline HF-HRV predicts subsequent depressive symptoms, but not vice versa, suggests that it may be a vulnerability marker rather than a transient state marker or trait marker of depression. Specifically, lower baseline HF-HRV may represent ongoing limitations in autonomic flexibility and regulatory capacity that predispose individuals (particularly males) to later increases in depressive symptoms. Within the neurovisceral integration model, this pattern is theoretically meaningful: reduced HRV would reflect weakened top-down prefrontal modulation of limbic and brainstem systems involved in emotion and stress regulation. Consequently, compromised autonomic–neural coordination may be a mechanistic pathway through which the long-term risk for depressive symptomatology is increased over time.

Many thanks for your suggestion. We have added the following paragraph to synthesize the conflicting results:

In summary, evidence linking treatment to HRV changes in depression remains mixed, and patterns differ by intervention type. In pharmacological RCTs (as gold standard), samples are typically more homogeneous and follow-ups relatively short; many studies show clear symptom improvement but no or only small average changes in HRV, and associations with outcome are inconsistent, sometimes appearing only in correlational change-score analyses or responder/non-responder contrasts rather than at the whole-group level. In contrast, large naturalistic cohorts such as NESDA include a broader, more clinically representative population and longer follow-up, and they suggest an overall HRV reduction over time that is linked to antidepressant use. A straightforward way to reconcile these findings is that symptom improvement may be accompanied by increases in HRV in some individuals, while antidepressant exposure may be associated with reductions in HRV, so the average net effect is small, inconsistent, or even appears null, especially in shorter trials that mix responders and non-responders and are not powered for physiological outcomes. If this explanation fully accounted for the pattern, a clearer HRV increase would be expected in purely psychological interventions. However, the (still relatively small) CBT literature largely shows symptom improvement with little average HRV change, except in specific subgroups (e.g., more severe depression), indicating that this explanation is insufficient to explain the absence of change. In contrast, HRV biofeedback studies report increases in HRV indices alongside symptom reductions more consistently. This implies that the HRV-symptom coupling is not uniform and may depend on the intervention’s targeted mechanisms and/or on patient subgroups. More broadly, depression treatment studies often show sizeable symptom reductions even in control/placebo conditions due to non-specific influences (e.g., spontaneous remission/regression to the mean, expectancy and therapeutic contact, improved sleep/activity from study participation, and changes in concomitant treatment), further complicating detection of HRV-specific effects. Future studies should therefore carefully model medication exposure and symptom trajectories, and prospectively test stratification approaches (e.g., using low baseline HF-HRV or related profiles) to identify subgroups in whom HRV is more tightly coupled to clinical change.

We thank the reviewer for this feedback and added the following explanation on mechanisms, in line with the rest of the review, where the autonomous nervous system plays an important role:

“Through the autonomous nervous system, HRV and depression share pathways that can be modulated by adjusting breathing frequency as an example of a voluntary action. This relationship between HRV and ‘malleable’ vegetative functions, such as breathing, makes it a good candidate for biofeedback, as patients can adapt to paced breathing and other interventions. Biofeedback as an intervention enhances self-efficacy and interoception, which also affects other systems related to stress and resilience, including HRV. Studies revealed that HRV biofeedback effectuates acute improvements during biofeedback practice (Wheat & Larkin, 2010). However, long-term results and the impact on clinical outcomes, remain unclear (Wheat & Larkin, 2010). For MDD, results are mixed: though HRV changes over time are inconclusive, biofeedback with focus on HRV seems to positively affect depression symptom improvement (Wheat & Larkin, 2010). The baroreflex system, which causes large oscillations in heart rate when breathing at a particular frequency, has been suggested as a mechanism through which controlling breathing can positively influence heart rhythm and HRV (Lehrer, 2022).”

Confounding factors: enrichment opportunities

Many thanks for your valuable comment. Unfortunately, we could not find this information in the current literature as is and would like to phrase this as a recommendation for the future. We acknowledged this in the paragraph as follows:

Further investigating gene-environment interactions and the interaction of genetic predisposition for HRV in combination with particular environmental stressors, could be a valuable next step in understanding the role of HRV genetics and its potential as a biomarker for depression-subtypes. However, for now, this relationship has not been investigated to that extent, so that it can differentiate between phenotypic subtypes or depression trajectories. More research is needed for this gene-environment interplay to be extrapolated towards patient stratification.

We have added the following:

We would here like to highlight that it is crucial to assess both the effects of sex and gender in future studies, since gendered social and cultural factors likely influence HRV. Differential exposure to chronic stressors (e.g., caregiving burden, workplace harassment, minority stress) may lower resting HRV [57] and increase depressive symptoms [58], and may thus be a moderator of associations between depressive symptoms and HRV. Gender norms may also influence coping styles, including emotion suppression or rumination. Finally, gendered patterns in healthcare-seeking and diagnosis can bias the grouping of ‘depression’ toward different severity, chronicity, or comorbidity profiles that are themselves linked to HRV, thereby moderating or distorting associations independent of biology [59] Unfortunately, since most studies use the term gender incorrectly (i.e., actually referring to sex), this often leads to wrong conclusions on the effects of gender and further, dedicated research on the effects of gender is needed.

Many thanks we have done so as follows:

Taken together, these findings indicate that age–HRV associations should not be assumed to be linear across older adulthood, and that the age-range of ~70+ may represent an important attenuation of previously observed age-related declines. Clinically, this supports age-stratified interpretation and/or explicit non-linear modelling of age (e.g., splines or age-band interactions), particularly in cohorts enriched for adults over 70 years. This is directly relevant to interpreting the seemingly contradictory associations discussed earlier (lines 221–226): differences in cohort age and the inclusion of very old participants can plausibly shift the observed direction and magnitude of associations, providing a confounding-based explanation for findings such as those reported by An et al. [30]. Accordingly, comparisons across studies should explicitly test and report age distributions and consider separate estimates for older age groups where feasible.

Many thanks for this valuable suggestion. We have emphasized the modifiable nature of physical activity, BMI and smoking by adding the following phrase:

Furthermore, we discussed future recommendations in this regard as follows:

These lifestyle factors have been linked to depression and depression severity, independent from HRV. BMI is a covariate often taken into account for depression studies, including physical activity as a confounding variable, but as part of an intervention is also becoming increasingly popular for depression research [add reference]. In particular future research on HRV and depression, should consider thoroughly assessing physical exercise, as well as smoking and drinking habits to include in the analyses and increase understanding of HRV in the context of depression, and the observed heterogeneity between studies. Furthermore, lifestyle interventions (e.g., regular aerobic activity, sleep regularization, paced-breathing/HRV biofeedback) targeting BMI normalization and/or smoking cessation/healthier habits may complement antidepressants by targeting physiological regulation more directly, potentially offsetting medication-associated HRV reductions while supporting symptom improvement. Clinically, they may prove useful as add-ons for partial responders and for longer-term recovery/relapse prevention, with additional cardiometabolic benefits.

Many thanks we have done so as follows:

There is an obvious discrepancy between the longitudinal, naturalistic studies in larger samples finding negative effects (i.e., antidepressant medication explained most of the effects on HRV) and the meta-analyses in unmedicated patients [18,19], who do find significant reductions of HRV in depression, even in the absence of medication. This discrepancy could reflect differences in depression subtype or comorbidities and other factors differing between unmedicated meta-analytic samples and naturalistic cohorts. Additional explanations include time-varying medication effects (i.e., when was treatment initiated/titration of the medication), or could reflect response-dependent effects, where medication-related HRV reductions may mix with recovery-related increases in HRV for responders. It is also possible that naturalistic samples capture more complex and cumulative antidepressant exposure (e.g., prior antidepressant failures, switching/augmentation, recent discontinuation) and polypharmacy, which may add to, or prolong HRV reductions. Therefore, future studies should take care to include detailed lifetime/recent medication histories (dose, duration, switching) and explicit polypharmacy indicators. Ideally, HRV changes would be measured in the context of randomized controlled clinical trials to examine HRV trajectories around medication changes in a controlled environment. Finally, an argument often made is that patients receiving antidepressant medication may suffer from more severe depression than those without antidepressant medication; particularly in light of some studies reporting a correlation of depressive symptoms and HRV, this factor should be considered.

Methodological considerations section

Thank you very much for your useful suggestion. We aligned the paragraph along your suggestions and find it much improved. The paragraph now reads as follows:

From the overview discussed in the manuscript, we list the most important aspects to acknowledge when designing your own study in patients with depression and attempt to make a list of essential versus recommended conditions. Importantly, these strongly depend on your research question and should not be taken over as is but rather be used to consider important covariates. More general guidelines on factors to considered for HRV can be found elsewhere ([15,16]).

7.1. Participant characteristics to assess

Essential

For studies in participants with depression it is mandatory to carefully collect participant characteristics known to influence HRV, and which may moderate or mediate the association with depression. These include age, sex, BMI, antidepressant medication intake (at least current medication and dosage, and whether polypharmacy is present), as well as any medication influencing the cardiovascular system (e.g., particularly beta-blockers, and other antihypertensives, stimulants, benzodiazepines, antipsychotics, and thyroid medication,) and, depending on your study design, exclude or document cardiac disease.

Recommended

In addition, it is advisable to include menstrual phase (and contraception intake) either as a controlled variable or document it; gender, and (psychiatric) comorbidities, but also sample characteristics generally associated with a lower HRV in psychiatric populations, including chronic and current stress, depression severity and anxiety as well as smoking status (or, if smoking is restricted, to document this carefully). Finally, it is useful to include a measure of physical fitness.

7.2. Environmental controls

Essential

In a laboratory environment, it is necessary to keep room temperature stable and noise to a minimum in order to reduce bias; furthermore, exercise should not be allowed on the day of testing (also consider participants cycling to the appointment) and ideally, no strenuous exercise should be permitted on the day before the appointment. Also consider that if research facilities are to be reached by stairs, an elevator should be used if possible and an acclimatization phase should be considered to control for breathing and other physiological parameters linked to physical activation, allowing these to normalize before starting the investigation. Furthermore, caffeine consumption and smoking should be restricted before testing (at least in the 1-2 hours prior to recording).

Recommended controls (enhances but not critical)

Where feasible, it can further reduce heterogeneity to standardize and report how acclimation is implemented (e.g., seated posture and minimizing talking or other distractions), especially when comparing across sessions or cohorts; alcohol consumption on the day prior to the testing should be limited. Finally, food intake should be kept stable (i.e., not test sober people together with those who just had a large lunch).

7.3. Measurement protocols

Essential controls

Measurement protocols can induce heterogeneity, which can be minimized and controlled through good preparation and strict application of the protocol. Given the important diurnal variations in HRV (e.g., increase of RMSSD in the night until morning, then decrease), if HRV measures are performed in the morning it is essential to take into account sleeping patterns and time since getting up, as well as the time of administration. This becomes particularly important for repeat-measures, where it is essential to keep the time of the second assessment identical.

Recommended controls (enhances but not critical)

Because HRV measurements largely rely on a normal breathing rhythm, breathing patterns should be taken into account or measured, if possible, particularly if there is interest in HF-HRV.

7.4. Considerations for studies using wearables

As wearables take over and contribute significantly to research on physiological measures such as HRV in an every-day context, it is necessary to be aware of the advantages and limitations of these technologies. To our knowledge, no consensus exists on the preprocessing requirements, but an overview of commonly used methods can be found here https://pmc.ncbi.nlm.nih.gov/articles/PMC12737534/). In our experience, when using wearable-derived data (PPG), phases of physical activity should be clearly separated from sedentary phases, because HRV can be biased by motion-related artifacts and changes in ambient light (https://pmc.ncbi.nlm.nih.gov/articles/PMC12737534/), leading to reduced signal quality. For HF-HRV, ECG-based standards generally assume clearly identifiable R-peaks and recommend stationary 5-min segments with rigorous beat editing/artefact handling, often complemented with manual control of said artefacts; however, these procedures are not always directly transferable to PPG because the pulse waveform is more motion-sensitive and wearables are typically recorded over extended periods (e.g., 24 h or longer), making manual artefact correction extremely time-consuming. If wearables are used for short-term recordings (i.e., 5-min segments), caution is warranted to ensure participants are correctly instructed regarding the recording situation (similar to laboratory conditions as described above), and recordings with a high number of invalid IBIs (e.g., >5% of total beats) should be treated with caution. Finally, due to the large inter-individual variation, the main strength of wearable HRV measurement lies in within-subject settings (i.e., change from baseline; longitudinal/interventional designs), permitting control for important covariates, rather than between-subject comparisons without stringent standardization and may provide a valuable and well-acceptable method to obtain data from patients with depression.

Thank you for your valuable input. This is a rapidly evolving field, and no clear recommendations can be made, but we have attempted to give recommendations as follows:

7.4. Considerations for studies using wearables

As wearables take over and contribute significantly to research on physiological measures such as HRV in an every-day context, it is necessary to be aware of the advantages and limitations of these technologies. To our knowledge, no consensus exists on the preprocessing requirements, but an overview of commonly used methods can be found here https://pmc.ncbi.nlm.nih.gov/articles/PMC12736534/). In our experience, when using wearable-derived data (PPG), phases of physical activity should be clearly separated from sedentary phases, because HRV can be biased by motion-related artifacts and changes in ambient light (https://pmc.ncbi.nlm.nih.gov/articles/PMC12736534/), leading to reduced signal quality. For HF-HRV, ECG-based standards generally assume clearly identifiable R-peaks and recommend stationary 5-min segments with rigorous beat editing/artefact handling, often complemented with manual control of said artefacts; however, these procedures are not always directly transferable to PPG because the pulse waveform is more motion-sensitive and wearables are typically recorded over extended periods (e.g., 24 h or longer), making manual artefact correction extremely time-consuming. If wearables are used for short-term recordings (i.e., 5-min segments), caution is warranted to ensure participants are correctly instructed regarding the recording situation (similar to laboratory conditions as described above), and recordings with a high number of invalid IBIs (e.g., >5% of total beats) should be treated with caution. Finally, due to the large inter-individual variation, the main strength of wearable HRV measurement lies in within-subject settings (i.e., change from baseline; longitudinal/interventional designs), permitting control for important covariates, rather than between-subject comparisons without stringent standardization and may provide a valuable and well-acceptable method to obtain data from patients with depression.

Recommended for Wearable measurements (PPG)

Although wrist-based monitors are commonly used, finger-based methods (i.e., rings) may provide a better PPG signal (doi: 10.3389/fphys.2019.00198). We have had good experiences focusing on sedentary periods such as sleep or resting episodes, which can be detected by most modern sensors (e.g., via accelerometry), and using automated filtering in KUBIOS software (medium correction setting), which produced results that compared well with manual spot-checks (unpublished data). Importantly, when removing motion artefacts, it is advisable to record several days of data if circadian rhythm is relevant, to avoid large phases of missing data.

Conclusions section

Many thanks we have done so:

To move beyond correlational findings, several research priorities follow from our synthesis: (1) adequately powered, preregistered longitudinal studies that repeatedly assess HRV and depression severity while rigorously controlling for key confounders (e.g., antidepressant use/duration, physical activity/fitness, age, sex, BMI and smoking behaviour, comorbid disorders, and cardiovascular/metabolic factors), and that explicitly model within-person change; (2) studies testing whether HRV predicts treatment response and relapse risk across modalities (pharmacotherapy, psychotherapy, exercise), ideally incorporating HRV as a prospective stratification variable rather than only an outcome; and (3) validation of wearable-derived HRV against the gold-standard ECG in clinical samples with depression. Progress on these priorities would clarify temporal ordering, strengthen causal inference, and determine whether HRV can be translated into a robust, clinically meaningful tool rather than a correlate of depression

Many thanks for the nice suggestions. We have done so as follows:

This challenge is not unique to HRV; it reflects a broader issue within the field of biological psychiatry, where many proposed biomarkers for depression and stress-related disorders show strong sensitivity but weak specificity (e.g., HRV has also been considered a “biomarker” for anxiety disorders and post-traumatic stress disorder potentially reflecting autonomic dysregulation or stress vulnerability across disorders rather than a depression-specific biomarker).

Reference quality and completeness

Many thanks we have checked the references and formatting again. We applied the mdpi zotero template for the references.

Many thanks, we have added this in the study section of the table.

Tables and figures

We appreciate the recommendation. We chose not to include a “clinical interpretation” column because there is no single clinically universal meaning of “high” or “low” for many HRV metrics, and interpretations vary by indication, patient characteristics, and measurement protocol. Including brief interpretations in a tabular format could unintentionally be read as normative clinical guidance. To avoid speculative statements, we would therefore like to keep Tables 1a–1b as reference tables.

Many thanks for your comment. As table 2 reflects the meta-analytic findings it is not possible to retrieve this information for all articles that were included in each of the meta-analyses. We have however added a major new table, (table 3) detailing the required information for the observational/longitudinal studies. We hope we have understood this point correctly.

Language and clarity refinements

We thank the reviewer for this suggestion and adjusted the citation accordingly.

Many thanks we have adapted it as suggested.

We thank the reviewer for this suggestion and consequently used ‘g’ throughout the manuscript for hedges’ g as the abbreviation is explained at the end of the manuscript. Furthermore, we added a corresponding statement at the end of the paragraph “4. HRV Measurement and Parameters”:

“We provide an overview of studies and the HRV-parameters investigated and report estimated effect sizes as they were reported in the meta-analyses (standardized mean difference (SMD), Cohen’s d or bias-corrected Hedges’ g, particularly suitable for smaller studies).”

Many thanks for your suggestions we have implemented it as follows:

Stratification by change in RSA may thus help to identify a biological subgroup [35]: it may be possible that RSA change reflects a biological “responsiveness” phenotype (e.g., preserved capacity for adaptive autonomic regulation) rather than a depression subtype per se. If validated, RSA trajectories might contribute to risk stratification or early treatment monitoring, complementing symptom-based assessments. These hypotheses remain exploratory in the context of the small sample and warrant confirmation in adequately powered studies.

Overall assessment considerations

I commend the authors for your scholarly contribution and advise you to confront any of my advisory notes if deemed outside your viewpoint.

With collegial respect,

Serving Peer Reviewer at JPM MDPI

Dear Reviewer,

We kindly thank you for your excellent suggestions and hope to have addressed your concerns adequately. It was a pleasure to receive your feedback to enhance our manuscript.

With best regards

Drs Van Assche & Schiweck

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for the opportunity to review this paper. It addresses the interesting topic of HRV and its relationship with depression.

The paper is well written in terms of sentence structure, wording etc. There are several concerns about what this paper really brings to the literature.

This is not a standard type of literature review. The purpose of the paper is vaguely stated, and the authors are straight forward about not including all papers that exist related to HRV and depression. This paper would have much more impact if the authors engaged in some of the basic strategies typically used in a review paper. The authors could easily explicitly state the questions they sought to answer. Transparency about the search terms/databases used and methods for determining which papers to include would also increase the manuscript’s contribution to the discussion of depression and HRV. Additionally, there is no reason the authors not cannot consider bias in the studies they selected for this review. There are instruments and guidelines to provide information about bias in the studies which would give the reader a better insight into the weight that shoudlb e placed on the conclusions from each study. This is particularly important given that some clinicians feel strongly that HRV is particularly valuable for a range of reasons while others give it much less credence.
In section 4.1.1 the authors discuss a paper in which SDNN and RSA were not significant and with smaller effect size in patients taking antidepressants. This calls into question the value of these markers for HRV and at least suggests that antidepressants affect HRV measures and should be at least reported in other HRV studies – in fact there may be enough evidence that presence of antidepressants should be considered in analyses of HRV. The presence of antidepressants is reported sporadically throughout the paper. This could be easily added to the paper – if not in the narrative at least in the tables reporting the papers.
The authors offer little evaluation of study quality in any of the sections. This is important because some studies (see section 4.4) had small sample sizes of 30-50 patients. There is also no discussion of observational studies vs more rigorous designs.
Unfortunately, because the purpose of this work and the methods employed lack the rigor of typical reviews, it is difficult to identify the value of the conclusions. This does not mean this paper couldn’t have value, rather there are important components missing that are needed for it to make a valuable contribution to the literature.

Author Response

----See also the attachment for a word format with colour.----

Reviewer 2

Thank you for the opportunity to review this paper. It addresses the interesting topic of HRV and its relationship with depression.

The paper is well written in terms of sentence structure, wording etc. There are several concerns about what this paper really brings to the literature.

This is not a standard type of literature review. The purpose of the paper is vaguely stated, and the authors are straight forward about not including all papers that exist related to HRV and depression. This paper would have much more impact if the authors engaged in some of the basic strategies typically used in a review paper. The authors could easily explicitly state the questions they sought to answer. Transparency about the search terms/databases used and methods for determining which papers to include would also increase the manuscript’s contribution to the discussion of depression and HRV. Additionally, there is no reason the authors not cannot consider bias in the studies they selected for this review. There are instruments and guidelines to provide information about bias in the studies which would give the reader a better insight into the weight that shoudlb e placed on the conclusions from each study. This is particularly important given that some clinicians feel strongly that HRV is particularly valuable for a range of reasons while others give it much less credence.

We appreciate the reviewer’s careful reading and agree that the scope and contribution of the manuscript should be stated more explicitly. This article was intentionally designed as a narrative synthesis, not a systematic review, with the goal of integrating evidence from meta-analyses, longitudinal studies, and key methodological considerations (including confounding factors) to provide an accessible overview and clinically oriented understanding of HRV in depression. We also aimed for the review to serve as a starting point for researchers new to the field, given the volume and heterogeneity of the literature A specific contribution of the manuscript is that it maps findings from existing meta-analyses and longitudinal studies, enabling readers to compare the scope, samples, and conclusions across these higher-level and temporally informative sources. We agree that greater transparency regarding study identification and selection is needed and have therefore added a Methods section describing the databases consulted, guiding questions, and the logic used for study selection and inclusion. To also make the scope of the review clear from the start we have adapted the title to the following and hope that this will draw the right readership.:

“Resting state Heart Rate Variability in Depression: An Introductory Narrative Review of Cross-Sectional and Longitudinal Evidence”

Regarding risk of bias, we recognize its central importance in systematic reviews and meta-analyses, where homogeneous inclusion criteria allow direct comparisons and pooled inference. In contrast, our narrative review intentionally includes heterogeneous study designs (e.g., RCTs, observational and population-based studies, case–control studies, longitudinal studies, and meta-analyses). A single risk-of-bias tool applied across such disparate designs would be misleading, while multiple design-specific tools would produce outputs that are not readily comparable within a narrative synthesis. We therefore did not conduct a cross-design risk-of-bias assessment, but acknowledge this as a limitation

In summary, we have clarified the guiding questions and rationale for study selection, added a concise methodological description, acknowledged limitations of the narrative approach, and a specific limitation on the missing risk of bias assessment. The changes are as indicated below:

Introduction:

Furthermore, existing meta-analyses fail to provide an integrative overview limited by their own strengths: e.g. strict inclusion and exclusion criteria. The focus of such work often considers cross-sectional or longitudinal evidence separately from each other, thereby not providing an integrative synthesis of both. In this narrative review, we therefore aim to provide an accessible starting point for researchers who may feel overwhelmed by the breadth of work on heart rate variability (HRV) and depression, by summarizing evidence for and against cross-sectional alterations of HRV in depression, on its predictive value for future depression occurrence and on depression-related interventions (pharmacological and non-pharmacological) and their impact on HRV. We first summarize selected studies and meta-analytic evidence examining HRV both cross-sectionally and longitudinally. We then outline key methodological considerations, including potential confounding factors and practical guidance for designing new studies. Importantly, as this is a narrative review, it is not meant to replace but rather complement existing meta-analyses. We make no claim at completeness as our aim was not to provide an extensive overview, but to provide researchers starting in the field with an introductory overview of studies that might be missed in other literature regarding the topic.

Methods:

Limitations:

First, while we attempted to give a good overview of the field, our approach was not systematic, and it is possible that important studies were overlooked. Because this narrative review integrates heterogeneous study designs, we did not conduct a formal risk-of-bias or quality assessment across all included studies, and the strength of evidence should not be interpreted as equivalent across designs.

In section 4.1.1 the authors discuss a paper in which SDNN and RSA were not significant and with smaller effect size in patients taking antidepressants. This calls into question the value of these markers for HRV and at least suggests that antidepressants affect HRV measures and should be at least reported in other HRV studies – in fact there may be enough evidence that presence of antidepressants should be considered in analyses of HRV. The presence of antidepressants is reported sporadically throughout the paper. This could be easily added to the paper – if not in the narrative at least in the tables reporting the papers.

We appreciate the reviewer’s emphasis on this important point. We agree that antidepressant medication is a key confounding factor in HRV research and that its effects deserve more consistent and explicit treatment. In addition to the discussion in the text and the section on antidepressant medication, we have added the use of antidepressant medication to the relevant table (table 2(specific column) and table 3 (study design)) and emphasize the discussion section regarding antidepressant use and HRV indices. As a recommendation, we also added a statement that medication status should be systematically reported and, where possible, accounted for in HRV analyses.:

As briefly mentioned above, it is well known that antidepressant medication can decrease HRV. In the previously mentioned NESDA study [17], it was shown that antidepressant medication (tricyclic antidepressant medication, but also SSRIs/SNRIs) reduced the association between depression and RMSSD significantly. The longitudinal analysis with a focus on antidepressant medication showed that the negative association between RSA and antidepressant medication was sustained over a 2-year follow-up period. These findings are also consistent with cross-sectional associations from the Irish Longitudinal Study on Ageing (TILDA), which found that in 317 elderly participants with depression, who were not taking antidepressants, HRV did not differ from controls on any measures of HRV, but all antidepressants were associated with lower measures of HRV [60]. Unfortunately, depression severity was not controlled for in this analysis [61,62]. There is an obvious discrepancy between the longitudinal, naturalistic studies in larger samples finding negative effects (i.e., antidepressant medication explained most of the effects on HRV) and the meta-analyses in unmedicated patients [18,19], who do find significant reductions of HRV in depression, even in the absence of medication. This discrepancy could reflect differences in depression subtype or comorbidities and other factors differing between unmedicated meta-analytic samples and naturalistic cohorts. Additional explanations include time-varying medication effects (i.e., when was treatment initiated/titration of the medication), or could reflect response-dependent effects, where medication-related HRV reductions may mix with recovery-related increases in HRV for responders. It is also possible that naturalistic samples capture more complex and cumulative antidepressant exposure (e.g., prior antidepressant failures, switching/augmentation, recent discontinuation) and polypharmacy, which may add to, or prolong HRV reductions. Therefore, future studies should take care to include detailed lifetime/recent medication histories (dose, duration, switching) and explicit polypharmacy indicators. Ideally, HRV changes would be measured in the context of randomized controlled clinical trials to examine HRV trajectories around medication changes in a controlled environment.

The authors offer little evaluation of study quality in any of the sections. This is important because some studies (see section 4.4) had small sample sizes of 30-50 patients. There is also no discussion of observational studies vs more rigorous designs.

We thank the reviewer for this important comment. We agree that study quality varies substantially across the included literature and is very important for conclusions drawn; however, because our review deliberately synthesizes fundamentally different study designs (meta-analyses, observational cross-sectional/longitudinal studies, and interventional/clinical trials), a single “study quality” appraisal or ranking would be difficult to apply consistently and risks giving readers a misleading comparison. In particular, the small-sample studies in Section 4.4 (now 5.4) were included because they represent unique interventional/clinical trial designs that address treatment effects/mechanisms in ways observational studies cannot, and they are therefore not directly comparable on the same quality scale. However, we agree that the information on study design/sample size etc. is important for the reader to assess especially the longitudinal studies and needs some different presentation to allow readers to draw a judgement on this. Therefore, we have summarized the studies in the format of a table (see table 3), with a specific column also mentioning limitations of each study design. We further acknowledge the lack of a systematic study quality as a limitation and have added the following:

Limitations:

While we attempted to give a good overview of the field, our approach was not systematic, and it is possible that important studies were overlooked. Because this narrative review integrates heterogeneous study designs, we did not conduct a formal risk-of-bias or quality assessment across all included studies, and the strength of evidence should not be interpreted as equivalent across designs. The readers are advised to carefully consider study limitations when comparing the different study designs.

Unfortunately, because the purpose of this work and the methods employed lack the rigor of typical reviews, it is difficult to identify the value of the conclusions. This does not mean this paper couldn’t have value, rather there are important components missing that are needed for it to make a valuable contribution to the literature.

We appreciate the reviewer’s candid assessment and constructive framing of this concern. We agree that greater clarity regarding the purpose, scope, and methodological choices will strengthen the manuscript’s contribution. We now clearly state that our aim was not to provide a systematic review of the literature but rather to complement them, providing an overview for starting researchers. In our experience with young researchers, we often get asked the questions where to start, but given the narrow focus of most systematic reviews, we feel that there is currently no review that assesses observational/cross-sectional, longitudinal and/ or intervention studies while also permitting a focus on different ages, and groups with depressive symptoms as well as a diagnosis of major depressive disorder. To make this clearer we have added the respective explanations in the introduction, methods and discussion sections and have changed the title as explained above

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript is much improved. Inclusion of search criteria strengthens the manuscript so the reader knows how manuscripts were selected.

An extremely interesting topic.

Article Menu

Resting State Heart Rate Variability in Depression: An Introductory Narrative Review of Cross-Sectional and Longitudinal Evidence

Further Information

Guidelines

MDPI Initiatives

Follow MDPI