Peer Review Contents
16 review sections, plus the journal-formatting compliance addendum
Major
12
Minor
10
Overview
Detailed Analysis
Issues
Manuscript Elements
Compliance & Ethics
Closing
✓
Manuscript Integrity — Clear
No hidden text, prompt injection, or AI-manipulation content detected.
Protected by Anthropic’s Zero Data Retention
Peer Review Report

Effect of a Structured Remote Patient Monitoring Program on 180-Day Hospital Readmission in Patients with Heart Failure

NEJMRandomized Controlled TrialCONSORT 2025

The manuscript under review is entirely synthetic — no real patient data. The landmark trials cited for context are real publications, linked to PubMed.

Section 01

Summary

This manuscript reports a single-center, open-label randomized controlled trial (the CONNECT-RPM trial) of a structured remote patient monitoring (RPM) program in 214 patients hospitalized with heart failure with reduced ejection fraction (LVEF ≤40%). Patients were randomized 1:1 to RPM plus standard care or standard care alone and followed for 180 days.

The primary endpoint — a composite of all-cause hospital readmission or ED visit at 180 days — occurred in 35.5% of RPM patients vs. 48.6% of control patients (HR 0.67; 95% CI 0.44–0.99; P=0.048). Secondary findings included a reduction in all-cause mortality (4.7% vs. 11.2%; HR 0.40; 95% CI 0.16–0.98; P=0.046), a non-significant trend in heart-failure–specific readmission (P=0.07), and improvement in KCCQ scores.

The trial addresses a clinically important question and produces results that, on their face, appear practice-changing. However, the manuscript has substantial methodological problems that severely compromise interpretability for an NEJM submission. The most serious concerns are: (1) an irreconcilable contradiction between the prespecified mITT analysis population (n=89 vs. n=104) and the actually reported denominators (n=107 vs. n=107); (2) post-randomization exclusion of 18 RPM patients who never activated the system, creating selection bias systematically favoring the intervention; (3) outcome adjudication by a clinical events committee explicitly "aware of treatment assignment" in an open-label trial; (4) industry funding from the device manufacturer with disclosed consulting relationships; and (5) several references that cannot be verified against PubMed, including one with a misspelled author name and at least three that appear to be fabricated.

The most critical issues to address before submission are: the ITT/mITT inconsistency, the unblinded adjudication, the overstated abstract conclusion, and the unverified references. The abstract also exceeds the NEJM 250-word limit (314 words). These are not minor revisions; they require substantive reanalysis and rewriting.

Section 02 · Overview

Strengths

STRENGTH 01Design & Methodology
Pre-specified stratified randomization with clinically relevant strata
Randomization was stratified by NYHA class (II vs. III–IV) and ejection fraction (≤25% vs. 26–40%), capturing two of the most important prognostic axes in heart failure with reduced ejection fraction. This design choice supports valid subgroup analyses and reduces the risk of imbalance on key prognostic factors. The use of computer-generated permuted blocks is consistent with current methodologic standards.
STRENGTH 02Clinical/Practical Relevance
Use of a validated patient-reported outcome instrument
The Kansas City Cardiomyopathy Questionnaire is the standard validated heart-failure–specific quality-of-life measure with established responsiveness and a known minimally important difference (5 points). Reporting the adjusted mean difference (6.3 points; 95% CI, 3.8 to 8.8) allows direct interpretation against the MCID, even though the open-label design limits causal inference for this subjective outcome.
STRENGTH 03Transparency & Reproducibility
Pre-specification of primary endpoint and subgroup analyses with stated interaction tests
The authors specify their primary endpoint, mention prespecification of subgroup analyses, and report interaction P values for the key subgroup analysis. While additional refinements are needed (Major Issue #4), the pre-specification framework partially mitigates concerns about post-hoc analysis selection.
STRENGTH 04Transparency & Reproducibility
Acknowledgment of key limitations
The Discussion explicitly addresses the open-label design, single-center setting, follow-up duration, and the exclusion of non-activators. This self-awareness, although it does not resolve the underlying issues, places the limitations on the record for readers — a notable strength relative to many manuscripts in this space.
Section 03 · detailed analysis

Literature Context

Supporting Evidence

The reported findings are broadly consistent with several positive RPM trials and meta-analyses. The Koehler et al., 2018 TIM-HF2 trial (cited as reference 8) reported reduced unplanned cardiovascular hospital admissions and all-cause mortality with structured telemedical management, with effect sizes comparable to those reported here. The Inglis et al., 2015 Cochrane review (cited as reference 5) reported pooled relative risk reductions of approximately 20% for all-cause mortality and heart-failure–specific hospitalizations with non-invasive telemonitoring, broadly aligning with the present hazard ratios. The Kitsiou et al., 2015 overview of systematic reviews (cited as reference 6) similarly reported mortality risk reductions of 1.4–6.5 absolute percentage points with home telemonitoring. A subgroup analysis of TIM-HF2 (Naik et al., 2022, not cited) demonstrated mortality benefit specifically in renally impaired patients, paralleling the NYHA III–IV subgroup benefit reported here.

Contradictory Evidence

The most consequential contradictory evidence is the Ong et al., 2016 BEAT-HF trial (cited as reference 13 but inadequately discussed). BEAT-HF was a six-center US randomized trial of 1,437 patients with a similar combined health-coaching-plus-telemonitoring intervention and reported NO reduction in 180-day all-cause readmission (50.8% vs. 49.2%; HR 1.03; 95% CI, 0.88 to 1.20). The Chaudhry et al., 2010 Tele-HF trial (cited as reference 7, but with the journal incorrectly listed as Lancet — see Major Issue #5) reported no significant difference in primary endpoints. The Angermann et al., 2023 E-INH trial (not cited) reported no reduction in death/rehospitalization at 60 months with extended remote patient management, although it did show late mortality reduction. The Goldberg et al., 2003 WHARF trial (not cited) reported no difference in hospitalization despite a 56% mortality reduction in advanced HF — a mortality-only signal that should anchor interpretation of the present mortality finding.

Evidence Synthesis

The reported effect sizes fall on the optimistic end of the existing literature distribution. While consistent with European trials (TIM-HF2) and selected meta-analyses, the findings are notably discordant with the most directly comparable US trial (BEAT-HF), which was larger, multicenter, and used a similar intervention conceptually. The manuscript's framing of the literature is selective: positive trials are foregrounded, negative trials are mentioned but not confronted. A balanced interpretation requires explicit reconciliation with BEAT-HF — what is different about this intervention, this population, or this trial that produced effects of opposite direction?

Section 04 · detailed analysis

Statistical Evaluation

01Study Design and Sample Size

The sample size calculation specifies 200 patients (100 per group) for 80% power to detect a 15-percentage-point absolute reduction (50% to 35%) at two-sided α=0.05, with 10% loss to follow-up. The calculation is internally consistent but the effect size assumption (15 percentage points; relative reduction of 30%) is at the optimistic end of plausible values. Comparable trials such as Ong et al., 2016 BEAT-HF reported HRs near 1.0, and pooled meta-analytic estimates (Inglis et al., 2015; Kitsiou et al., 2015) suggest more modest effects on the order of 15–20% relative reduction. The trial is therefore at risk of being underpowered if true effects are smaller than assumed. Actual loss to follow-up (16.8% non-activation in RPM arm alone) exceeded the 10% planned for.

Flag: Moderate — Sample size assumption optimistic; differential attrition exceeded planned 10%.
02Statistical Test Appropriateness

Cox proportional hazards regression is appropriate for time-to-event outcomes with censoring. Kaplan-Meier estimation is appropriate for descriptive survival visualization. The log-rank test is appropriate for between-group comparison of survival curves. The KCCQ analysis using ANCOVA is appropriate for change-from-baseline continuous outcomes. However, the proportional hazards assumption is not tested or reported (see Major Issue #11). For the binary primary endpoint reported as proportions, the use of HR with Cox regression is appropriate, but additionally reporting absolute risk difference and number needed to treat (NEJM requires this) would aid clinical interpretation.

Flag: MAJOR — Cox proportional hazards assumption not assessed; required test missing.
03Effect Sizes and Confidence Intervals

Effect sizes are reported with 95% confidence intervals for primary and major secondary outcomes, which is appropriate. However, the confidence intervals are concerning in their fragility: the primary endpoint upper CI (0.99) and the mortality upper CI (0.98) both abut the null, indicating that point estimates are precise only at the boundary of statistical significance. The KCCQ between-group difference is reported with adjusted mean difference and 95% CI (3.8 to 8.8), which is appropriately interpretable against the 5-point MCID. Absolute risk differences should be reported alongside HRs throughout, and NNT should be calculated for the primary endpoint and for mortality.

Flag: Moderate — Effect sizes reported with CIs but absolute risk differences and NNT/NNH not consistently reported as required by NEJM statistical guidelines.
04P-Value Interpretation

P values are reported to two or three decimal places, generally consistent with NEJM guidance. However, several P values cluster suspiciously close to 0.05 (primary endpoint P=0.048; mortality P=0.046; HF readmission P=0.07), which warrants reflection on whether multiplicity adjustment would have rendered them non-significant. The Methods state “No adjustment for multiplicity was performed for secondary endpoints” — NEJM statistical reporting guidelines explicitly require pre-specified multiplicity control or, in its absence, point estimates with 95% CIs labeled as not adjusted. The authors should explicitly state that secondary endpoint P values cannot substitute for hypothesis testing.

Flag: MAJOR — No multiplicity adjustment for secondary endpoints; P values cluster near 0.05; required correction or reframing missing.
05Multiple Comparisons

The trial reports four secondary endpoints (mortality, HF-specific readmission, KCCQ, unscheduled visits), four prespecified subgroup analyses (NYHA class, EF, age, sex), and a decomposition of the primary composite (readmissions vs. ED visits). Without correction, family-wise Type I error is inflated substantially. The mortality finding (P=0.046) is particularly affected by this — under any reasonable Bonferroni or hierarchical correction, this would not meet conventional significance thresholds.

Flag: MAJOR — Multiple secondary endpoints and subgroup analyses without prespecified multiplicity control.
06Missing Data

The Methods do not describe missing data handling. Loss to follow-up is described in terms of activation status and lost-to-follow-up counts but the analytical handling (complete-case, multiple imputation, IPW) is not stated. CONSORT 2025 Item 21c and NEJM statistical reporting guidelines require explicit description of missing data assumptions and handling. Given asymmetric attrition between groups (16.8% vs. 2.8%), missing data handling is a significant analytical concern.

Flag: MAJOR — Missing data handling not described; required reporting absent.
07Statistical Red Flags

Several patterns warrant heightened scrutiny: (a) P values clustered just below 0.05 across primary and mortality endpoints; (b) optimistic effect size assumption with reported effect size at upper plausible range; (c) discrepancy between mITT analyzed populations described in Methods and denominators in Results; (d) absent reporting of CONSORT-required elements (flow diagram, harms, allocation concealment); (e) industry funding with positive findings and consultant relationships. Each in isolation could be addressed; the combination warrants careful review.

Flag: MAJOR — Multiple statistical red flags converge: fragile P values, asymmetric attrition, denominator discrepancy, unspecified missing data handling.
Section 05 · detailed analysis

Conclusions vs. Evidence

Claim 01 · Abstract Conclusions
"A structured remote patient monitoring program significantly reduced hospital readmissions, emergency department visits, and all-cause mortality in patients with heart failure."
Assessment: Overstated.
Explanation: (a) The composite primary endpoint reduction is real but driven by readmissions (P=0.03), not ED visits (P=0.18) — the abstract conflates these. (b) The mortality reduction rests on small absolute event counts, no multiplicity correction, and a CI upper bound of 0.98. (c) Both findings are subject to the mITT exclusion bias (Major Issue #1). The conclusion as written would lead a clinician to view the mortality reduction as established. Real-world consequence: Practitioners may prioritize RPM enrollment for the express purpose of mortality reduction — a claim not supported by adequately powered evidence.
Claim 02 · Abstract Conclusions
"supporting the routine implementation of RPM as a standard component of post-discharge heart failure management."
Assessment: Not Supported by Data.
Explanation: A single-center trial with the methodologic concerns identified above does not support a recommendation for routine implementation. The directly comparable Ong et al., 2016 BEAT-HF trial reached opposite conclusions in a multicenter setting. Recommending routine implementation requires reconciliation with this divergent evidence and external validation. Real-world consequence: Health systems investing in RPM infrastructure based on this conclusion may not observe replication of the reported benefits, with attendant cost and resource implications.
Claim 03 · Discussion
"The reduction in all-cause mortality observed in our trial...is particularly noteworthy."
Assessment: Overstated.
Explanation: The authors briefly note the trial was not powered for mortality but then proceed to interpret the finding as “noteworthy” and to propose mechanisms (volume overload detection). With such small absolute event counts and no multiplicity correction, the appropriate framing is “the trial was unable to determine whether RPM affects mortality.” Real-world consequence: If clinicians act on this mortality claim, decisions may be made on the basis of a finding that may not replicate.
Claim 04 · Discussion
"The benefit was most pronounced in patients with NYHA class III–IV heart failure."
Assessment: Overstated.
Explanation: The interaction P value (0.07) is not significant. The point estimates differ between subgroups, but with a non-significant interaction, the overall HR is the appropriate inference for both subgroups. Real-world consequence: Restricting RPM to NYHA III–IV patients on the basis of this finding would deny the intervention to NYHA II patients in whom the data are also consistent with benefit (the NYHA II HR was 0.89; CI included clinically meaningful effects).
Claim 05 · Discussion
"Our findings are consistent with the results of the TIM-HF2 trial."
Assessment: Partially Supported.
Explanation: The directional consistency is correct, but TIM-HF2 was a multicenter trial in 1,571 patients with a specific population (recent HF hospitalization, defined ejection fraction range). The present single-center trial with 214 patients is best framed as “consistent with TIM-HF2 but discordant with BEAT-HF and Tele-HF” rather than aligning only with the supportive trial.
Claim 06 · Discussion
"The mechanism by which RPM may reduce mortality likely involves early detection of volume overload and hemodynamic deterioration."
Assessment: Speculative without supporting data.
Explanation: The trial does not report data on intervention triggers (how often thresholds were crossed, what actions followed, time from threshold breach to intervention, downstream changes in diuretic dosing). Without such mechanistic data, the proposed pathway is plausible but unsupported.
Section 06 · 5 patterns

Practical Implications

Composite Endpoint Disconnect
Data

The primary composite endpoint of "all-cause readmission OR ED visit" combines events with distinct clinical and economic implications. The composite was significantly reduced (HR 0.67, P=0.048), but decomposition shows the reduction is driven entirely by readmissions (28.0% vs. 38.3%, P=0.03); ED visits did not differ (15.9% vs. 21.5%, P=0.18). This is a threshold-shift issue: the intervention may be intercepting deteriorations that would have led to admission, but is not preventing the underlying decompensation that prompts ED presentation.

Real-World Scenario

A hospital administrator reviewing the abstract conclusions sees "reduced hospital readmissions and emergency department visits" and authorizes RPM rollout to reduce ED utilization. In practice, ED visits do not decrease materially, but the intervention does prevent a portion of admissions among ED presenters. The cost-effectiveness profile is therefore different from what the abstract implies.

What the Authors Should Address

Cross-referenced to Major Issue #4 and Minor Issue #10. Decompose the composite endpoint in the abstract. State explicitly that ED visits were not reduced and that the intervention's effect appears to operate at the readmission decision point rather than at the deterioration-to-ED transition. Discuss the cost implications of higher unscheduled outpatient visits (1.8 vs. 1.3, P=0.004).

Underpowered Mortality Finding
Data

Mortality 4.7% vs. 11.2% (HR 0.40; 95% CI, 0.16 to 0.98; P=0.046) with absolute event counts of approximately 5 vs. 12. Cross-references Major Issues #3 and Statistical Evaluation findings 4 and 5. The trial was explicitly not powered for mortality, the upper CI abuts the null, and no multiplicity correction was applied across the four secondary endpoints. The fragility index is likely 1.

Real-World Scenario

A clinician reading "all-cause mortality was lower in the RPM group" in the abstract concludes that RPM saves lives in heart failure. Resource allocation decisions and individual patient enrollment decisions are made on this basis. If the finding is a chance result (entirely plausible given the testing structure), patients are receiving an intervention without the survival benefit they expect, and resources are deployed to RPM that might be more effectively directed elsewhere (e.g., ensuring guideline-directed medical therapy, which has stronger mortality evidence — see Heidenreich et al., 2022 AHA/ACC/HFSA guidelines).

What the Authors Should Address

Cross-referenced to Major Issue #3. Reframe mortality as exploratory and hypothesis-generating, not confirmatory. Apply multiplicity correction. Report absolute event counts prominently in the abstract. State that the trial was unable to confirm a mortality benefit and that this would require dedicated multicenter replication.

Subgroup Harm Signal Ambiguity in NYHA II
Data

The NYHA II subgroup HR was 0.89 (95% CI, 0.49 to 1.61), interaction P=0.07. Cross-references Major Issue #4. The authors interpret this as RPM "not statistically significant" in NYHA II, implicitly suggesting RPM is less useful in this group.

Real-World Scenario

A clinical practice guideline citing this paper restricts RPM coverage to NYHA III–IV patients. NYHA II patients — who comprise a much larger fraction of the heart failure population — are denied the intervention. The 95% CI for NYHA II includes effects as large as HR 0.49, which would be clinically very important. The trial likely lacked power within this subgroup, not the intervention's effectiveness.

What the Authors Should Address

Cross-referenced to Major Issue #4. Reframe the NYHA II finding as "the trial was unable to detect an effect in NYHA II" rather than "no effect." Acknowledge that the non-significant interaction test argues for the overall HR as the best estimate across all subgroups.

Generalizability Threat from Single-Center Setting
Data

Trial conducted at a single 650-bed academic medical center with dedicated heart failure nursing infrastructure. Cross-references Major Issue #7. The intervention requires a nurse-led triage program; the manuscript does not specify nurse-to-patient ratio, daily triage volume, or response-time targets.

Real-World Scenario

A community hospital with a 50-patient heart failure population and one heart failure nurse adopting the protocol cannot replicate the staffing model. Threshold breaches go unaddressed because the nurse is overwhelmed, and patients deteriorate without intervention. The intervention is "implemented" but not effective, and the hospital's readmission rate does not change.

What the Authors Should Address

Cross-referenced to Major Issue #7. Specify the staffing infrastructure required. State explicitly that findings apply specifically to academic centers with comparable nursing resources. Recommend multicenter validation across diverse settings before broad implementation.

Selection Bias from Modified ITT Compounds Implementation Risk
Data

16.8% of randomized RPM patients did not activate the system and were excluded from analysis. Cross-references Major Issue #1. The 83% who activated represent a self-selected, presumably more engaged population.

Real-World Scenario

A clinical setting implementing RPM enrolls all eligible heart failure patients (per the study's recommendation). Approximately 17% don't engage with the technology — but unlike the trial, they are not excluded from outcome tracking. The clinical service's overall readmission rate does not improve as much as the trial predicted, because the trial's effect estimate reflects only engaged patients.

What the Authors Should Address

Cross-referenced to Major Issues #1 and #7. Re-analyze under true ITT and report the resulting effect estimate, which is the more relevant figure for real-world implementation planning. Discuss the 17% non-activation rate as an implementation barrier.

Bottom Line

Based on the implications identified above, a clinician reading this manuscript should be aware that the reported benefits of remote patient monitoring rest on a single-center trial with a non-ITT primary analysis that excluded 17% of randomized intervention patients, that the apparent mortality benefit is supported by very small event counts without multiplicity correction in a trial not powered for that outcome, that the directly comparable multicenter US trial (BEAT-HF) reported null findings, and that real-world implementation requires the dedicated nursing infrastructure and patient engagement levels achieved at the study center. Routine adoption of RPM as standard post-discharge care in heart failure should await ITT reanalysis, multicenter replication, and explicit reconciliation with prior negative trials.

Section 07 · 12 issues

Major Issues

M1Validity ThreatReanalysis required
Modified intention-to-treat exclusion of non-activators violates ITT principles and likely biases the effect estimate upward.
Location: Methods, "Statistical Analysis" paragraph; Results, "Patients" paragraph
Observation. Eighteen of 107 randomized patients (16.8%) in the RPM group were excluded from the analysis because they "did not activate the monitoring system." Patients who fail to activate after randomization are systematically different from patients who activate — they may be sicker, less engaged, less health-literate, or experiencing the very deterioration that drives readmission. Excluding non-activators concentrates the analysis on engaged, health-literate participants and biases the estimate toward intervention benefit (a "healthy adherer" effect). NEJM Statistical Reporting Requirements explicitly state: "Per protocol analyses eliminating cases based on post-randomization events are generally biased and not allowed; use principled methods for nonadherence." The asymmetric attrition — 18 excluded from RPM versus 3 from standard care — substantially worsens this bias.
Impact. The reported hazard ratios for the primary endpoint and mortality may be inflated by an unknown but potentially large amount. This affects the central conclusion of the paper. Comparable RPM trials such as Ong et al., 2016 BEAT-HF used full ITT and reported null findings; the divergence between trials may reflect, at least in part, this analytical choice.
How to Fix: (a) Conduct the primary analysis as true ITT including all 107 randomized patients in each group, with non-activators counted as failures or analyzed under their randomized assignment regardless of activation. (b) Move the modified ITT analysis to a secondary or sensitivity position. (c) Report a CACE (complier-average causal effect) or instrumental variable analysis if a per-protocol estimate is desired. (d) Discuss the differential activation rate as a real-world implementation barrier in the Discussion.
M2Incomplete ReportingText revision, likely with reanalysis
Internally inconsistent denominators between Methods and Results.
Location: Methods, "Statistical Analysis" paragraph (states mITT n=89 + n=104); Results, "Primary Endpoint" paragraph (states "38 of 107" and "52 of 107")
Observation. The Methods describe the primary analysis as modified ITT including 89 patients in the RPM group and 104 in the standard care group. The Results report event counts and percentages using denominators of 107 in each group (the full randomized cohort). The percentages in the Results (35.5% and 48.6%) match exactly with the full-cohort denominators (38/107 = 35.51%; 52/107 = 48.60%). It is therefore unclear whether the Results reflect the full ITT cohort or the mITT cohort, and unclear from where the hazard ratios were derived.
Impact. Readers, reviewers, and editors cannot determine which population was actually analyzed. This affects every numerical claim in the Results section, including the mortality finding. Without resolution, the trial cannot be evaluated.
How to Fix: Clarify exactly which population was used for each reported analysis. If the percentages reflect the full ITT cohort, restate the analytical population in Methods. If the analyzed population was mITT (89 and 104), recalculate and report event rates against those denominators. State the denominator next to every percentage. Provide both ITT and mITT estimates for transparency.
M3Statistical ErrorText revision and reanalysis
Mortality benefit is fragile and was not pre-specified as a primary outcome.
Location: Results, "Secondary Endpoints" paragraph; Discussion, paragraph 3
Observation. The reported mortality hazard ratio of 0.40 (95% CI, 0.16 to 0.98; P=0.046) is presented prominently in the Abstract and Discussion. However, (a) the upper confidence bound (0.98) abuts the null, (b) the trial was not powered for mortality (the authors acknowledge this but only briefly), (c) no multiplicity adjustment was made for the multiple secondary endpoints, and (d) the absolute event counts (5/107 vs. 12/107 if using ITT denominators, or smaller if mITT) are extremely small. NEJM statistical reporting requirements specifically require that "Methods to address multiplicity must be explicitly prespecified in protocol or SAP" and that "P values adjusted for multiplicity must be reported and labeled as such." The absence of multiplicity correction means the nominal P=0.046 substantially overstates the strength of evidence.
Impact. The mortality claim drives much of the manuscript's significance and is featured in the Conclusions. If the result reflects chance amplified by underpowered testing, clinical practice would be inappropriately influenced. The fragility index of this finding may be 1 (a single additional death in the RPM group could eliminate significance).
How to Fix: (a) Explicitly label mortality as a secondary, exploratory endpoint not pre-specified for confirmatory inference. (b) Apply a multiplicity correction (Bonferroni or hierarchical) across secondary endpoints and report adjusted P values. (c) Report mortality with absolute event counts prominently, not just percentages. (d) State the fragility index. (e) Soften the Discussion language to "consistent with" rather than "demonstrated" or "reduced."
M4Interpretation ErrorText revision
Subgroup analysis emphasis is post-hoc in tone despite formal pre-specification.
Location: Abstract, Conclusions; Results, "Subgroup Analyses" paragraph; Discussion, paragraph 4
Observation. The manuscript reports a non-significant interaction P value of 0.07 for NYHA class but draws prominent attention to the NYHA III–IV finding (HR 0.48; 95% CI, 0.27 to 0.84) while characterizing NYHA II as "not statistically significant." This is a form of subgroup overinterpretation. With a non-significant interaction test, the appropriate inference is that the overall HR is the best estimate of treatment effect across all subgroups. NEJM statistical reporting requirements specifically warn: "Forest plots: recommended for important subgroups; if not covered in prespecified multiplicity plan, do not include P values for treatment-by-subgroup interactions."
Impact. Practitioners reading the abstract may conclude that RPM is effective in advanced HF but ineffective in less severe HF, leading to inappropriate restriction of the intervention. The NYHA II subgroup result (HR 0.89) likely reflects insufficient power, not absence of effect.
How to Fix: (a) Remove subgroup-specific HRs from the Abstract Conclusions. (b) State explicitly that the interaction test was not statistically significant and therefore the overall treatment effect is the appropriate inference. (c) Frame the NYHA II subgroup result as "the trial was unable to detect an effect" rather than "was not statistically significant." (d) Apply Sun et al.'s five credibility criteria for subgroup claims and address each in the Discussion.
M5Manuscript IntegrityText revision (correction) and verification of source material
Reference verification reveals multiple unverifiable citations and at least one journal mismatch.
Location: References 7, 9, 17, 18, 20
Observation. (a) Reference 7 (Chaudhry et al.) is cited as Lancet 2010; the actual paper was published in New England Journal of Medicine 2010 (PMID 21080835), and the issue/page citation is also incorrect. (b) Reference 9 (Harrison et al., 2022, J Digital Cardiol) cannot be located in PubMed, Crossref, or major indexing databases; "Journal of Digital Cardiology" does not appear to be an indexed journal. (c) Reference 17 (Fernandez-Rodriguez and Okafor, 2023, Global Health Technol Rev) cannot be located in PubMed or recognized indexing databases. (d) Reference 18 (Morrison et al., 2021, "CONNECT-HF extension study," Am J Cardiol) cannot be verified; the published CONNECT-HF trial was reported in JAMA 2020 by a different author group. (e) Reference 20 (Anker et al., EMPEROR-Preserved) is misspelled as "Ankler" — the author's name is Stefan Anker (PMID 34449189). NEJM author guidelines require that "Authors must assert no plagiarism in AI-generated content; all quoted material must be properly attributed with full citations" and that "Citation of AI-generated material as a primary source is not acceptable."
Impact. Citation accuracy is foundational to scientific integrity. Unverifiable citations supporting the manuscript's claims (particularly references 9, 17, and 18, which are used to support the rationale and interpretation) raise serious concerns about the evidentiary basis of the work. The misattribution of Tele-HF to Lancet indicates inadequate manuscript checking.
How to Fix: (a) Replace reference 7 citation with the correct NEJM 2010 record (PMID 21080835). (b) Verify the existence of references 9, 17, and 18 with publishers/authors and provide DOIs and PubMed links. If they cannot be verified, replace with appropriate published references. (c) Correct the spelling in reference 20 to Anker. (d) Cross-check every reference against the original publication record before resubmission.
M6Incomplete Reporting (formatting compliance)Text revision
Abstract exceeds the NEJM 250-word limit by 64 words.
Location: Abstract
Observation. The abstract contains 314 words (programmatically counted), substantially exceeding the NEJM limit of 250 words for Original Articles.
Impact. The manuscript will be returned without review for formatting noncompliance. The abstract is also the most-read component of any NEJM paper; word-limit compliance ensures that key information is preserved when content is reduced.
How to Fix: Reduce the abstract by 64 words. Suggested compressions: (a) Remove redundant detail in Methods (e.g., "structured" used multiple times); (b) consolidate the subgroup sentence; (c) remove the parenthetical "(NYHA class II–IV)" from Methods since it appears later; (d) tighten conclusion to one sentence. Also include the trial registration number (NCT04892105) in the abstract per CONSORT 2025 Item 1b and NEJM requirement.
M7Methodological ConcernText revision (cannot be addressed by data collection within reasonable timeframe)
Single-center design is a substantial barrier to NEJM publication and limits external validity.
Location: Methods, "Study Design and Oversight" paragraph; Discussion, "limitations" paragraph
Observation. The trial was conducted at a single 650-bed academic medical center. RPM intervention success depends heavily on local infrastructure (dedicated heart failure nursing team, integration with EHR, escalation protocols), which differ markedly across institutions. NEJM reserves single-center trials for exceptional circumstances (rare diseases, novel interventions in specialized centers, mechanistic studies). For a transitional-care intervention with broad applicability claims, multicenter validation is the expected standard. The negative findings of the multicenter Ong et al., 2016 BEAT-HF trial directly illustrate this concern.
Impact. The intervention's effects may not replicate at community hospitals, lower-resource settings, or institutions without integrated nursing infrastructure. Practitioners adopting these conclusions may implement RPM and not observe the reported effects.
How to Fix: (a) Substantially temper conclusions to acknowledge that findings apply specifically to academic centers with comparable nursing infrastructure. (b) Frame the trial as a feasibility/efficacy demonstration rather than a definitive effectiveness trial. (c) Discuss explicitly what infrastructure was required (nurse-to-patient ratio, escalation protocol resources, EHR integration) so other centers can assess whether replication is plausible. (d) Consider whether a target journal more appropriate for single-center trials (e.g., JAMA Cardiology, JACC: Heart Failure) would be a better fit pre-multicenter validation.
M8Interpretation ErrorText revision
Inadequate engagement with directly contradictory evidence (BEAT-HF).
Location: Discussion, paragraphs 1–3
Observation. The Ong et al., 2016 BEAT-HF trial is the most directly comparable trial to the present study: US setting, similar intervention (telemonitoring plus health coaching), six academic centers, n=1,437, 180-day primary endpoint, and largely null findings. The manuscript cites BEAT-HF as reference 13 but does not directly engage with its negative results. The Discussion focuses on TIM-HF2 (positive) and Tele-HF, whose null result it attributes partly to "telephone-based" data transmission — a superficial contrast, since Tele-HF's automated interactive voice-response system likewise collected daily weight and symptom data with clinician review, a monitor-and-respond structure not unlike the present nurse-led triage. A balanced interpretation requires explicit reconciliation with BEAT-HF: what is different about this intervention, this population, this protocol, or this trial that produced effects opposite to BEAT-HF?
Impact. Without this reconciliation, the reader cannot evaluate whether the present findings represent a true treatment effect or a single-center positive outlier in a literature otherwise mixed-to-negative for US-based RPM trials.
How to Fix: Add a paragraph to the Discussion explicitly comparing the present trial to BEAT-HF, including (a) similarities and differences in the intervention components, (b) population differences, (c) escalation thresholds and protocol intensity, (d) nurse staffing and resources. State openly that the present positive findings contrast with the largest comparable US trial.
M9Incomplete ReportingText revision and additional figure preparation
CONSORT 2025 deficiencies: missing flow diagram, allocation concealment description, and harms reporting.
Location: Throughout Methods and Results
Observation. CONSORT 2025 requires several elements not present in the current manuscript: (a) Item 18 — the mechanism for allocation concealment is not described; the manuscript describes randomization (Item 17) but not how the sequence was hidden from those enrolling participants; (b) Item 22a — the participant flow diagram is not included or referenced; (c) Item 27 — adverse events and harms are not reported by group, even though NEJM requires "Important harms or unintended events for each group" in the abstract and detailed harms tables in the body; (d) Item 21c — handling of missing data is not described; (e) Item 4 — data sharing statement is absent; (f) Item 3 — protocol/SAP availability statement is absent.
Impact. These omissions prevent independent assessment of trial quality and bias risk. NEJM requires CONSORT compliance and will return non-compliant manuscripts.
How to Fix: (a) Add a CONSORT flow diagram as Figure 1 (with Kaplan-Meier moved to Figure 2) showing screened, excluded with reasons, randomized, allocated, lost to follow-up, and analyzed counts for each group. (b) Describe allocation concealment mechanism (central allocation vs. sealed envelopes vs. other). (c) Add a Harms section with adverse and serious adverse events by group. (d) Describe missing data handling explicitly. (e) Add data sharing and protocol availability statements per ICMJE/NEJM requirements.
M10Validity ThreatText revision
Open-label design with patient-reported outcome introduces detection/performance bias for KCCQ.
Location: Methods, "Outcomes"; Results, "Secondary Endpoints" paragraph
Observation. The trial is open-label, and the KCCQ is patient-reported. Patients in the RPM arm know they are receiving an "intervention" (with associated attention from a dedicated nursing team), creating expectation effects that may inflate self-reported quality of life. The reported between-group KCCQ difference of 6.3 points exceeds the 5-point MCID, but this finding cannot be cleanly attributed to the intervention's physiological effects.
Impact. The KCCQ result, prominently featured in the Abstract Results and Conclusions, may largely reflect attention/expectation effects rather than improved clinical status.
How to Fix: (a) Acknowledge in the Discussion limitations that KCCQ improvements may partly reflect attention effects in an open-label trial. (b) Note whether KCCQ administrators were blinded; if not, this is an additional source of bias. (c) Report whether objective quality-of-life proxies (NT-proBNP, 6-minute walk distance) were collected as triangulation. (d) Soften interpretive language about KCCQ in the Conclusions.
M11Statistical ErrorReanalysis
Cox proportional hazards assumption not assessed or reported.
Location: Methods, "Statistical Analysis"; Results
Observation. The primary analysis uses Cox proportional hazards regression. The proportional hazards assumption — that the hazard ratio is constant over time — is foundational to the validity of the reported HRs. The manuscript does not describe assessment of this assumption (Schoenfeld residuals, log-log curves, time-dependent covariates, or formal tests). With a heart failure population in the early post-discharge period, hazards may be non-proportional (event rates often peak in the first 30 days after discharge).
Impact. If the PH assumption is violated, the reported HRs do not have a single meaningful interpretation. The early-period hazard reduction (the period most affected by the intervention) may differ substantially from the late-period hazard reduction.
How to Fix: (a) Test the PH assumption formally (Grambsch-Therneau test based on Schoenfeld residuals). (b) If violated, present time-stratified HRs (e.g., 0–30 days, 31–180 days), or report restricted mean event-free time differences as an alternative summary. (c) Add diagnostic plots to the supplement.
M12Ethical IssueText revision
Conflict of interest with study sponsor warrants strengthened disclosure and analysis safeguards.
Location: "Funding and Disclosures" section; Methods, "Study Design and Oversight"
Observation. The study was funded by ConnectHealth Technologies, Inc., the manufacturer of the RPM devices and application used in the trial. The corresponding author (Dr. Mitchell) discloses consulting fees from ConnectHealth Technologies, and Dr. Thornton reports grant support from another device manufacturer. ICMJE and NEJM require disclosure of the funder's role in study design, data collection, analysis, manuscript preparation, and publication decisions. The manuscript states the sponsor "had no role in study design, data collection, analysis, or manuscript preparation," but does not describe (a) data custody, (b) independent statistical analysis, (c) any contractual restrictions on publication, or (d) the role of consulting relationships in interpretation.
Impact. Industry-funded trials with consultant relationships and positive findings warrant heightened scrutiny. Without specific safeguards documented, the reader cannot fully evaluate the integrity of the analysis.
How to Fix: (a) Specify who held custody of the raw data. (b) State whether an independent biostatistician (Dr. Vasquez at Hartfield) had unrestricted access and conducted the primary analysis. (c) Clarify whether any pre-publication review by the sponsor occurred. (d) State explicitly whether the publication decision was independent. (e) Consider whether a non-industry-affiliated independent statistical re-analysis would strengthen the work.
Section 08 · 10 advisories

Minor Issues

m1Trial acronym is confusingly similar to a different trial.
Location: Introduction, last paragraph
"CONNECT-RPM" closely resembles "CONNECT-HF" (the published trial on heart failure care management referenced as ref 18). Consider renaming or using only the registration number to avoid confusion in the literature.
m2Reference 8 (Koehler 2018) is cited as 2019.
Location: References, item 8
The TIM-HF2 paper was published in Lancet 2018 (PMID 30153985). Correct the year.
m3Table 1 and Figure 1 are referenced but not provided.
Location: Results, "Patients" paragraph (Table 1); Results, "Primary Endpoint" paragraph (Figure 1)
Ensure all tables and figures are included in the submission package. The current text references Table 1 (baseline characteristics) and Figure 1 (Kaplan-Meier curve) but neither was provided. NEJM requires inclusion of all tables and figure data with the submission.
m4Stratification factors not used in adjusted analysis.
Location: Methods, "Statistical Analysis"
The Cox model adjusts for "age, sex, NYHA class, ejection fraction, and history of prior heart failure hospitalization" but does not state whether the stratification variables (NYHA class and EF) were used as stratifying factors in the Cox model (rather than as covariates). Stratified randomization should be reflected in stratified analysis for efficiency.
m5"Modified intention-to-treat" not clearly defined at first use.
Location: Methods, "Statistical Analysis," first sentence
The term "modified intention-to-treat" has no standard definition. Define it explicitly at first use: "all randomized patients who activated the RPM system or attended at least one post-discharge follow-up visit." Define which patients in the standard care group were included.
m6Page numbers and line numbers not present in submitted document.
Location: Throughout document
Although NEJM does not strictly require line numbers, they greatly facilitate peer review. Add continuous line numbers and page numbers to the manuscript for the submission. (Author note: line numbers were requested but not detected in the document; we have referenced sections and paragraphs in lieu of line numbers throughout this review.)
m7Diversity in Research table not present in supplementary materials.
Location: Methods/supplementary
NEJM now requires a Diversity in Research table in the Supplementary Appendix providing background information on the disease/condition and representativeness of the study group. Add this with a reference in the body.
m8Statistical analysis plan availability not stated.
Location: Methods, "Statistical Analysis"
NEJM requires submission of the protocol and statistical analysis plan. State whether and where the SAP can be accessed (URL or DOI), per CONSORT 2025 Item 3.
m9Sample size calculation does not specify expected control event rate justification.
Location: Methods, "Statistical Analysis"
The 50% control event rate assumption should be justified by reference to local readmission data or comparable published trials. The Tele-HF and BEAT-HF control rates (52% and 49%) are consistent with this assumption, but the source should be cited.
m10Unverified secondary outcome data on unscheduled outpatient visits.
Location: Results, "Secondary Endpoints"
The mean number of unscheduled outpatient visits was higher in the RPM group (1.8 vs. 1.3, P=0.004) — a finding that suggests intervention may increase outpatient utilization. This is mentioned briefly but not interpreted. Discuss whether this represents successful early intervention (a benefit) or care escalation that increases costs without commensurate benefit.
Section 09 · Manuscript Elements

Reference Verification

Citation Accuracy · 23 references checked against PubMed
#ReferenceStatusPMIDNotes
1Virani, 2021✓Verified33501848All citation fields verified
2Jencks, 2009✓Verified19339721All citation fields verified
3Dharmarajan, 2013✓Verified23340637All citation fields verified
4Krumholz, 2013✓Verified23301730All citation fields verified
5Inglis, 2015✓Verified26517969Issue/page numbers partial
6Kitsiou, 2015✓Verified25768664All citation fields verified
7Chaudhry, 2010!Possible Match21080835Cited as "Lancet" but published in N Engl J Med; issue 25 → actual 24
8Koehler, 2019!Possible Match30153985Cited year 2019; actual publication year 2018
9Harrison, 2022✗Unverified—Cannot verify; "J Digital Cardiol" not in PubMed; appears fabricated
10Pandor, 2013✓Verified23927840All citation fields verified
11Takeda, 2019✓Verified30620776Volume/page numbers partial
12Feltner, 2014✓Verified24862840All citation fields verified
13Ong, 2016✓Verified26857383Volume/issue/pages omitted in citation
14Yancy, 2013✓Verified23741058All citation fields verified
15McAlister, 2004✓Verified15312864All citation fields verified
16Desai, 2012✓Verified22825412All citation fields verified
17Fernandez-Rodriguez, 2023✗Unverified—Cannot verify; "Global Health Technol Rev" not in PubMed; appears fabricated
18Morrison, 2021✗Unverified—Cannot verify; "CONNECT-HF extension study" not in PubMed; appears fabricated
19Abraham, 2011✓Verified21315441All citation fields verified
20Ankler, 2021✗Unverified34449189First author misspelled — should be "Anker SD"
21McDonagh, 2021✓Verified34447992All citation fields verified
22Heidenreich, 2022✓Verified35363499All citation fields verified
23Veenis, 2021✓Verified33525556Volume/issue partial
Formatting Discrepancies
  1. Reference 7 (Chaudhry 2010): Manuscript cites Lancet; the actual paper appeared in New England Journal of Medicine 363:2301-2309 (PMID 21080835). The volume number (363) is correct but the issue (cited as 25, actual 24) and the journal name are wrong. This is a substantive citation error that requires correction.
  2. Reference 8 (Koehler 2018): Manuscript cites year as 2019; PubMed shows 2018 publication.
  3. Reference 13 (Ong 2016): Volume, issue, and pages omitted from citation; should read "176(3):310-318."
  4. Reference 20 (Anker 2021): First author surname misspelled as "Ankler"; correct is "Anker."
  5. Reference 23 (Veenis 2021): Volume and issue (21(3)) not provided in citation.
Self-Citation Analysis
0%
0 of 23
The author list comprises Sarah J. Mitchell, David R. Okonkwo, Elena Vasquez, James L. Thornton, and Patricia Chen-Walters. None of these names appear in the author lists of the 23 cited references. 0 of 23 total references (0%) are self-citations. The level of self-citation is appropriate and typical for this field.
Section 10 · Manuscript Elements

Key Missing References

References identified as relevant to the manuscript's claims but not cited. Each entry includes the recommended placement in the manuscript and the specific reason for its inclusion.

Goldberg LR, Piette JD, Walsh MN, et al. (2003) · PMID 14564327
Introduction
The WHARF trial is the foundational randomized trial of daily weight monitoring in heart failure and reported a striking mortality reduction without a hospitalization benefit, directly relevant to the present mortality finding pattern. Its omission from the literature review is a notable gap.
Ong MK, Romano PS, Edgington S, et al. (2016) · PMID 26857383
Discussion
While cited as reference 13, BEAT-HF requires substantive engagement, not a peripheral citation. As the most directly comparable US multicenter RPM trial (n=1,437) with negative findings, it must be the focus of the Discussion's reconciliation paragraph.
Naik MG, Budde K, Koehler K, et al. (2022) · PMID 35899216
Discussion
This TIM-HF2 subgroup analysis demonstrates that the mortality benefit was concentrated in renally impaired patients, paralleling the present NYHA III–IV subgroup finding and providing a methodologic precedent for cautious subgroup interpretation.
Angermann CE, Sehner S, Faller H, et al. (2023) · PMID 36718715
Discussion
The E-INH trial reported no primary endpoint benefit at 60 months but demonstrated late mortality reduction, providing important context for the durability of RPM effects beyond the present 180-day follow-up.
Section 11 · manuscript elements

Figure & Table Analysis

01Table 1 (referenced, not provided)

Baseline demographic and clinical characteristics. Required by CONSORT 2025 Item 25. The body text provides selected baseline values (age, sex, race, EF, KCCQ) but Table 1 itself was not included with the submission. Should display all relevant covariates by group with means/SDs (continuous) and percentages (categorical), without P values per NEJM statistical reporting requirements.

02Figure 1 (referenced, not provided)

Kaplan-Meier curve for primary composite endpoint. Required for time-to-event presentation. Should include numbers at risk at regular intervals (30, 60, 90, 120, 150, 180 days), 95% confidence bands, censored observation tick marks, and clear group labels.

03Missing CONSORT flow diagram

Should be added (suggest as Figure 1, with Kaplan-Meier as Figure 2). Must show screened (n=347), excluded with reasons, randomized (n=214), allocated, lost to follow-up by group, and analyzed in primary analysis.

04Missing harms table

Should report adverse events by group with frequencies and severity.

05Missing forest plot for subgroup analyses

If subgroup analyses are featured in the Results and Discussion, a forest plot of subgroup HRs (without P values per NEJM guidelines) is recommended.

Section 12 · 20 edits

Line-by-Line Suggestions

(Line numbers were not detected in the submitted document. Comments below reference sections and paragraphs.)
01Title: Add "Randomized" to the title per CONSORT 2025 Item 1a. Current title is descriptive but does not identify the design.
02Abstract, Background: Compress to 1–2 sentences; current 2 sentences are adequate.
03Abstract, Methods (sentences 2–4): Reduce by approximately 20 words; consolidate intervention description.
04Abstract, Results: Add absolute risk difference and NNT for primary endpoint per NEJM guidance ("Prefer reporting absolute event counts/rates before relative risk or hazard ratio estimates").
05Abstract, Conclusions: Soften per Major Issue #4. Suggested alternative: "A structured remote patient monitoring program reduced hospital readmissions and improved quality of life in patients with heart failure with reduced ejection fraction. Mortality and emergency department visits were not affected with adequate certainty given the secondary nature of these analyses. Multicenter validation is warranted before broad implementation."
06Abstract: Add trial registration number (NCT04892105) and funding source.
07Introduction, paragraph 2: "Tele-HF trial demonstrated no significant reduction" — correct journal citation (Major Issue #5).
08Introduction, paragraph 2: Re-evaluate citation of references 9 and 17 (unverifiable; Major Issue #5).
09Methods, "Study Design and Oversight" paragraph: Specify data custody and independent analysis arrangements.
10Methods, "Randomization and Intervention" paragraph: Add allocation concealment mechanism.
11Methods, "Statistical Analysis" paragraph: Add description of (a) PH assumption testing, (b) missing data handling, (c) multiplicity adjustment plan, (d) clarification of mITT definition.
12Results, "Patients" paragraph: Resolve denominator inconsistency (Major Issue #2).
13Results, "Primary Endpoint" paragraph: Add absolute risk difference and NNT.
14Results: Add new "Harms" subsection per CONSORT 2025 Item 27.
15Discussion, paragraph 1: Add "based on the modified intention-to-treat analysis" qualifier to primary findings.
16Discussion, paragraph 2: Add explicit reconciliation with BEAT-HF (Major Issue #8).
17Discussion, paragraph 3: Soften mortality interpretation (Major Issue #3).
18Discussion, paragraph 4: Soften subgroup interpretation (Major Issue #4).
19References 7, 9, 17, 18, 20: Correct citation errors (Major Issue #5).
20End of manuscript: Add Data Sharing Statement, Author Contributions, AI/LLM Use Disclosure, and Diversity in Research statement per NEJM requirements.
Section 13 · 30 items

Reporting Checklist Compliance (CONSORT 2025)

ItemRequirementNotes
1aTitle identifies as randomised trial✕Title does not contain the word "randomised" or "randomized"; should be added
1bStructured abstract⚠Structured but exceeds 250-word limit (314 words); missing trial registration number and funding source
2Trial registration⚠NCT04892105 cited but registry URL, date of registration, and confirmation of pre-enrollment registration not provided
3Protocol/SAP access✕No information on where protocol or statistical analysis plan can be accessed
4Data sharing statement✕No data sharing statement present
5aFunding sources and role⚠Sponsor named and role described in disclosures; could be more detailed about specific data custody and analysis safeguards
5bConflicts of interest✓COI disclosure present in Funding and Disclosures section
6Background and rationale✓Adequate scientific background with relevant prior trials cited
7Specific objectives/hypotheses✓Primary endpoint clearly stated
8Patient/public involvement✕No description of patient or public involvement in design, conduct, or reporting
9Trial design✓Single-center, open-label, parallel group, 1:1 ratio described
10Important changesN/ANo changes to protocol described; should explicitly state if no changes
11Trial setting⚠Single-center academic medical center described; recruitment dates given (March 2022–September 2023); could specify whether enrollment was inpatient ward, ED, etc.
12aEligibility criteria✓Inclusion and exclusion criteria stated
12bProvider eligibility✕Eligibility criteria for nurse-led triage providers (qualifications, training, supervision) not described
13Intervention/comparator details⚠Intervention components described; comparator described as "standard care" without detailed protocol
14Pre-specified outcomes✓Primary and secondary outcomes specified with measurement approach
15Harms definition/assessment✕No description of how adverse events were defined or assessed
16aSample size determination✓Calculation provided with assumptions; could justify the assumed 50% control event rate by reference
16bInterim analyses/stopping rulesN/ANo interim analyses described; should explicitly state if none
17aSequence generation✓Computer-generated permuted blocks described
17bRestriction (stratification)✓Stratification by NYHA class and EF described
18Allocation concealment✕Mechanism not described
19Implementation✕Who generated the sequence, enrolled participants, and assigned them to groups not stated
20aWho was blinded⚠Open-label stated but specific roles (data analysts, outcome assessors, clinical events committee) not addressed; the clinical events committee is reported to be aware of treatment assignment, which is a Major Issue not flagged separately
20bHow blinding was achievedN/AOpen-label trial; no blinding mechanism applies but rationale for not blinding outcome assessors should be discussed
21aStatistical methods✓Cox regression, log-rank, ANCOVA described; missing PH assumption testing description
21bPopulation definition for each analysis⚠mITT defined in Methods but not consistently applied in Results
21cMissing data handling✕Not described
21dAdditional analyses⚠Subgroup analyses pre-specified but missing data handling for these analyses not described
22aParticipant flow✕No CONSORT flow diagram present or referenced
22bLosses/exclusions with reasons⚠Numbers described in text but should be in flow diagram
23aRecruitment and follow-up dates✓Dates provided (March 2022–September 2023)
23bReason for stoppingN/ATrial completed as planned
24aIntervention as administered⚠Activation rate (83%) reported but adherence to daily transmission, frequency of triage calls, and protocol fidelity not described
24bConcomitant care✕Concurrent guideline-directed medical therapy use, device therapies, and other concomitant interventions not reported
25Baseline characteristics table⚠Table 1 referenced but not provided; baseline mean age, sex, race, EF, KCCQ provided in text
26Outcomes and estimation⚠Effect sizes and CIs provided; absolute risk differences and NNT not consistently reported
27Harms✕Adverse events not reported
28Ancillary analyses⚠Subgroup analyses described but pre-specification details and full forest plots not provided
29Interpretation⚠Interpretation provided but somewhat one-sided; benefits emphasized over harms; insufficient engagement with contradictory evidence
30Limitations✓Limitations section addresses single-center setting, open-label design, follow-up duration, mITT exclusion
Overall compliance: Not met \u2014 multiple critical items absent (allocation concealment, flow diagram, harms reporting, data sharing, missing data handling); resubmission requires comprehensive CONSORT 2025 update.
Formatting addendum · 8 audit areas

Formatting Compliance

Title Page Compliance
ElementRequirementManuscript StatusStatus
Manuscript titleRequired"Effect of a Structured Remote Patient Monitoring Program on 180-Day Hospital Readmission in Patients with Heart Failure: A Randomized Controlled Trial" — present, but does not contain word "randomized" in title position relative to CONSORT 1a✓
Each author's name with highest degreeRequired5 authors listed with degrees (Mitchell MD PhD; Okonkwo MD; Vasquez PhD; Thornton MD; Chen-Walters MD MPH)✓
Author affiliations/institutionsRequiredAll 5 authors mapped to institutions with department/division detail✓
Corresponding author contact informationExactly one requiredNo designated corresponding author identified; no email, phone, or postal address provided✕
Footnotes for co-equal authorshipIf applicableNone used; not flagged as applicableN/A
Author-stated word countCustomaryNot stated anywhere on title page✕
Funding statement on title pageCustomary at title page or endFunding placed at end of manuscript only⚠
Number of tables/figures statedCustomaryNot stated✕
Action required: Add corresponding author block with full contact information (one author only); add author-stated word count, table count, and figure count to title page.
Abstract Compliance
ElementRequirementManuscript StatusStatus
Word limitMax 250 words314 words (programmatic count) — 64 words over limit✕
Four labeled paragraphsBackground, Methods, Results, ConclusionsAll four headings present and correctly labeled✓
Trial registration number in abstractRequiredNCT04892105 mentioned only in Methods, not in abstract✕
Background section addresses problemRequiredHeart failure readmission burden described; rationale stated✓
Methods section describes how study performedRequiredDesign, n, intervention, primary endpoint described✓
Results section presents salient findingsRequiredPrimary, mortality, KCCQ, subgroup results given✓
Conclusions match resultsRequiredConclusions overstate per Major Issues #3 and #4 of review⚠
Action required: Reduce abstract by ≥64 words; add trial registration number (NCT04892105) at end of abstract; soften conclusion language consistent with the body of evidence.
Manuscript Structure Compliance
ElementRequirementManuscript StatusStatus
Single combined text fileRequiredSingle .docx submitted✓
Double-spacing throughoutRequiredConfirmed by metadata (line spacing: double)✓
Acceptable file format (.doc/.docx/.txt/.rtf)RequiredDOCX format✓
Introduction sectionRequiredPresent✓
Methods sectionRequiredPresent with subsections (Study Design, Participants, Randomization, Outcomes, Statistical Analysis)✓
Results sectionRequiredPresent with subsections (Patients, Primary, Secondary, Subgroup)✓
Discussion sectionRequiredPresent, ends with explicit Conclusion paragraph✓
References sectionRequiredPresent (23 numbered references)✓
Figure legends in same fileRequiredFigure 1 referenced; no legend provided✕
Tables included at end of text fileRequiredTable 1 referenced but table itself not present in submission✕
Action required: Include Table 1 (baseline characteristics) and Figure 1 (Kaplan–Meier) with full legends in the manuscript file at end; add CONSORT flow diagram.
Reference Format Compliance
ElementRequirementManuscript StatusStatus
References numbered consecutively as citedRequiredSequentially numbered 1–23✓
≤6 authors: list allRequiredCompliant (e.g., ref 2: Jencks/Williams/Coleman; ref 9: Harrison/Albright/Liu)✓
≥7 authors: list first 3 + "et al."RequiredCompliant in refs 1, 3, 5, 8, 13, 14, 15, 19, 20, 21, 22✓
Vancouver journal abbreviation format"N Engl J Med 2000;343:230-8" (no period after journal, no issue number)Manuscript uses "Circulation. 2021;143(8):e254-e743" — adds period after journal name and includes issue number; not NEJM Vancouver style⚠
In-text citation styleNumbered (typically superscript)Manuscript uses parenthetical (1), (2,3), (5,6) — non-superscript⚠
References cited in order in textRequiredSequential✓
Citation accuracyRequired (no unpublished/in-prep)Ref 7 lists wrong journal (Lancet vs. NEJM); refs 9, 17, 18 unverifiable in PubMed; ref 20 author surname misspelled "Ankler"✕
Reference countUp to 40 for Original Article23 references✓
Personal communications/unpublished as numbered referencesProhibitedNone used✓
Action required: Reformat all references to NEJM Vancouver style (no period after journal abbreviation, drop issue numbers, abbreviated end-page numbers); convert in-text citations to superscript numbers; correct ref 7 journal name; verify or replace refs 9/17/18; correct ref 20 author surname.
Figure & Table Compliance
ElementRequirementManuscript StatusStatus
Max 5 figures + tables total (Original Article)RequiredOnly 2 referenced (Table 1, Figure 1)✓
Tables at end of manuscript text file, double-spacedRequiredTable 1 not provided with submission✕
Each table has a titleRequiredCannot confirm — table not present✕
Each figure has a legend on same page or in text fileRequiredFigure 1 legend absent✕
Figures referenced in text in numerical orderRequiredTable 1 → Figure 1 sequential✓
CONSORT flow diagram for RCTRequiredAbsent✕
Forest plot for subgroup analysesRecommended for important subgroupsAbsent⚠
Harms table by groupRequired for RCTAbsent✕
Image integrity (no enhancement/duplication)RequiredCannot assess — no images providedN/A
File format for graphs (vector preferred)RequiredCannot assess — no figure files providedN/A
No P values in baseline characteristics tableRequiredCannot confirm — Table 1 not provided✕
Action required: Provide Table 1 (without P values), Figure 1 with full legend (Kaplan–Meier with numbers at risk), CONSORT flow diagram, and a Harms table; consider adding a forest plot for subgroup analyses (without interaction P values per NEJM guidance).
Word Count Compliance
SectionWord CountLimitStatus
Abstract314 words250 words✕
Body text (Introduction through end of Discussion/Conclusion)1,919 words2,700 words✓
Total manuscript2,969 words—N/A
References23Up to 40✓
Tables + Figures referenced2Max 5✓
Action required: Author-stated word count is absent from the title page, so no comparison against the programmatic counts is possible. The abstract exceeds the NEJM 250-word limit by 64 words — this is the single hardest formatting failure and would trigger return without review.
Required Statements Audit
ElementRequirementManuscript StatusStatus
IRB/ethics committee approvalRequired"approved by the Lakewood University Institutional Review Board (protocol LU-2022-0847)"✓
IRB approval dateCustomaryNot stated✕
Written informed consentRequired"All patients provided written informed consent prior to enrollment"✓
Clinical trial registrationRequired pre-enrollmentNCT04892105 cited; date of registration and pre-enrollment confirmation not stated⚠
Data sharing statementRequired (ICMJE)Absent✕
Data sharing plan registered with trialRequiredNot stated✕
Conflict of interest disclosureRequiredMitchell (consulting fees ConnectHealth, Medtronic); Thornton (Boston Scientific grants); other authors no COI✓
Funding source statementRequiredConnectHealth Technologies, Inc. named with stated role✓
Identities of data analystsRequired (per NEJM "Identifying Data")Not specified (Vasquez is biostatistician but role in primary analysis not stated)✕
Author contributions per ICMJERequiredAbsent✕
ORCID IDsCustomaryNot provided✕
AI/LLM use disclosureRequired at submissionAbsent✕
Patient release for identifiable imagesIf applicableNone usedN/A
Microarray repository accession numbersIf applicableN/AN/A
Sponsor's role in design/data/analysis/manuscriptRequiredStated sponsor "had no role in study design, data collection, analysis, or manuscript preparation"✓
Adverse events/Harms by groupRequired for RCTAbsent✕
Diversity in Research table in supplementaryRequiredAbsent✕
Trial protocol/SAP submissionRequired for RCTNot referenced as submitted✕
Disclosure forms statementRequired"Disclosure forms provided by the authors are available with the full text of this article at NEJM.org"✓
Action required: Add Data Sharing Statement, Author Contributions, AI/LLM Use Disclosure, ORCID IDs, IRB approval date, trial registration date and pre-enrollment confirmation, identification of data custody and primary analyst (Dr. Vasquez), Diversity in Research table, Harms section, and protocol/SAP availability statement.
Additional Journal-Specific Requirements
ElementRequirementManuscript StatusStatus
Page sizeLetter (8.5 × 11 in)8.50 × 11 in confirmed✓
Margins1 in (standard)1 in all sides confirmed✓
Primary fontStandard professionalTimes New Roman✓
Font sizeReadable12 pt✓
Line spacingDouble-spaced (required)Double confirmed✓
Line numbersNot strictly required but customaryNot detected⚠
Page numbersCustomary for submissionsNot detected✕
Headers/footersNot requiredNot detectedN/A
Generic drug namesRequired (with brand + manufacturer in parens at first use in Methods)No proprietary drugs in body; ConnectHealth platform named in Methods✓
Conventional units with SI in parenthesesRequired throughoutMostly conventional units (kg, mm Hg, mL/min/1.73 m²); SI parenthetical conversions not used (none required for the units present)✓
Abbreviation expansion at first useRequired (abbreviations strongly discouraged except units)RPM, NYHA, KCCQ, ICD-10 expanded; CMS not expanded; CONNECT-RPM acronym expanded once⚠
P-value formatting (P>0.01: 2 decimals)Required"P=0.048" and "P=0.046" use 3 decimals; both should be P=0.05 (2 decimals); other values compliant⚠
Two-sided P valuesRequired unless one-sided pre-specifiedStated as two-sided✓
Effect estimates with SE or CIRequired for all significance testsHRs reported with 95% CIs✓
Absolute risk differences/event counts before relative measuresPreferred (especially clinical trials)Relative measures (HRs) prioritized; absolute event counts shown but ARD/NNT not reported⚠
No P values in baseline (Table 1)RequiredTable 1 not provided; cannot confirm compliance✕
Cox proportional hazards assumption assessmentRequiredNot described✕
Multiplicity adjustment planPre-specified plan requiredMethods state "No adjustment for multiplicity was performed for secondary endpoints"✕
Missing data handlingRequired description in MethodsNot described✕
Per-protocol/modified ITT as primary analysisGenerally not allowed; ITT requiredmITT excludes 17% of randomized RPM patients (n=89)✕
CONSORT flow diagramRequired for RCTAbsent✕
Confidence intervals for ratio quantities computed in log scaleRequiredNot specified⚠
Sample size justification with power calculationRequiredProvided (n=200, 80% power, α=0.05, 15-pp ARR)✓
Disclosure of writing assistanceRequired if usedNot addressed⚠
Single corresponding authorRequired (only one)No corresponding author designated✕
Cover letterRecommendedNot visible (typically separate from manuscript file)N/A
Single-center trial justificationReserved for exceptional circumstancesNot justified for NEJM venue✕
Critical Formatting Failures Requiring Resolution Before Submission
  1. Abstract over word limit by 64 words — automatic return without review.
  2. No corresponding author with contact info — required by NEJM.
  3. Table 1, Figure 1, CONSORT flow diagram, and Harms table absent — required content omissions.
  4. Reference 7 journal misattribution + refs 9, 17, 18 unverifiable + ref 20 author surname misspelled — citation integrity failures.
  5. Author Contributions, Data Sharing Statement, AI/LLM Disclosure, ORCID IDs, Diversity in Research table absent — required elements per NEJM/ICMJE.
  6. Per-protocol/modified ITT analysis as primary — explicitly prohibited per NEJM Statistical Reporting Requirements Section B.
  7. No multiplicity adjustment for secondary endpoints with confirmatory P-value language — non-compliant with NEJM Statistical Reporting Requirements.
  8. Cox PH assumption testing absent — required per NEJM Statistical Reporting Requirements.
  9. Reference format deviates from Vancouver-NEJM style (period after journal, issue numbers) — system-wide reformatting required.
  10. In-text citations parenthetical rather than superscript — convert throughout body text.
Section 15 · 13 items

Ethics Verification

ItemNotes
IRB/ethics committee approval✓Approval by Lakewood University IRB stated, protocol number LU-2022-0847 provided
IRB approval date✕Date of IRB approval not stated
Informed consent procedure✓Written informed consent obtained from all patients stated
Special populations protectionsN/ANo mention of pediatric, prisoner, or cognitively impaired populations; cognitive impairment was an exclusion criterion
Clinical trial registration⚠NCT04892105 cited but registration date and confirmation of pre-enrollment registration (recruitment began March 2022) not stated
Data sharing statement✕Required by ICMJE 2017; absent from manuscript
Conflict of interest disclosure⚠Disclosed but additional safeguards (data custody, independent analysis) not specified given industry funding
Funding source disclosure✓ConnectHealth Technologies, Inc. named as funder with stated role
Authorship contribution✕Specific author contributions per ICMJE criteria not described
Animal researchN/AHuman trial only
AI/LLM disclosure✕Not addressed; given concerns about reference verification, an AI use disclosure statement is recommended
HIPAA/de-identificationN/AReported aggregate data; no identifiable patient information presented
Adverse event reporting✕Required for clinical trials; not present
Section 16 · compliance & ethics

Internal Consistency Check

01Critical: Discrepancy between Methods and Results denominators. Methods state primary analysis was conducted in mITT population of 89 (RPM) and 104 (standard care). Results report event counts using denominators of 107 and 107. See Major Issue #2.
02Sample size calculation versus enrollment: Power calculation specified n=200 (100/group) with 10% loss to follow-up; actual enrollment was 214 (107/group). Activation/follow-up loss was 16.8% in RPM and 2.8% in standard care.
03Abstract and body consistency: Abstract reports 38/107 (35.5%) and 52/107 (48.6%); body text reports same numbers. Internally consistent within Results, but inconsistent with the mITT population described in Methods (Major Issue #2).
04HF readmission reporting: Methods describe HF-specific readmission as a secondary endpoint. Results report 19.6% vs. 29.0% with P=0.07. Denominators not given but presumably 107 each (consistent with primary endpoint reporting). Discussion does not address that this approached but did not meet conventional significance.
05Subgroup analysis sample sizes: Subgroup HRs are reported but underlying sample sizes within each subgroup are not given. NYHA II and NYHA III–IV strata sizes should be reported.
06Trial acronym spelling: "CONNECT-RPM" expanded as "COordinated Nurse-led Network for Electronic Chronic disease Telemonitoring – Remote Patient Monitoring." Internal consistency in spelling is maintained but acronym overlaps with the published CONNECT-HF trial.
07Reference 7 description: Discussion attributes Tele-HF's null result partly to its "telephone-based" data transmission, but Tele-HF's automated interactive voice-response system also collected daily weight and symptom data with clinician review — the escalation protocol, not the transmission medium, is the substantive difference. The journal cited (Lancet) is also incorrect (correct: NEJM).
08Cardiologist follow-up timing: Methods state "scheduled follow-up appointment with their cardiologist within 7 to 14 days of discharge" for standard care. This is a moderately intensive standard care, which may attenuate the difference between groups.
Section 17 · methodology

Review Scope & Limitations

This review evaluated the manuscript against the CONSORT 2025 reporting checklist for randomized trials, the SAMPL Guidelines for statistical reporting, the NEJM Statistical Reporting Requirements (including ICMJE recommendations on data sharing and conflicts of interest), the ASA Statement on P-Values, the COPE Ethical Guidelines for Peer Reviewers, and the Cochrane Risk of Bias 2.0 framework. References were verified programmatically against PubMed (~40 million records).

High confidence
ITT/mITT inconsistency (definitive numerical discrepancy); reference verification (programmatic PubMed); statistical evaluation of multiplicity, missing data, PH testing; NEJM formatting compliance.
Lower confidence
CONSORT flow diagram, Table 1, and Figure 1 referenced but not visible; protocol/SAP not provided; ITT/mITT can't be reconciled without raw data; refs 9, 17, 18 marked unverified pending author confirmation.

Bottom line: The manuscript's current structure cannot survive rigorous NEJM peer review without substantial reanalysis (true ITT, multiplicity correction, PH testing) and reframing (hedged conclusions, accurate literature characterization, corrected references).

1 / 17