{"id":"a3a99655-feaa-47fe-9eb4-2c54c863a87e","arxiv_id":"2412.09973","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Naps change tinnitus in about 27 to 31 percent of people with tinnitus, worsening more often than improving, and nap-sensitive people also report more sleep, stress, and somatosensory modulation of their tinnitus.","lead":"A study of two large tinnitus databases found that roughly 27 to 31 percent of people with tinnitus say that daytime naps change their tinnitus, and worsening is reported about twice as often as improvement. The paper is worth reading because it maps which clinical traits, such as sensitivity to night sleep, stress, and jaw or neck movements, travel together with nap sensitivity, hinting at a recognizable tinnitus subtype.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The stated cross-database 'no difference' p=0.99 is incompatible with the paper's own reported counts; a recomputed chi-square is highly significant, undercutting the key robustness claim of congruent replication.","rationale":"The reader's weakest_assumption was the harmonization of differently worded nap questions across the TRI and Tinnitus Hub databases. That is a genuine threat to the construct-validity of the cross-database comparison, but the more immediately load-bearing problem is that the paper's own reported counts contradict its stated p = 0.99 for the distribution comparison. This is not a matter of interpretation or wording; it is an internal statistical inconsistency. The central claim—that a clinically identifiable subgroup reports nap-induced modulation, with worsening roughly twice as common as improvement, and with associated features such as night-sleep, stress, and somatosensory modulation—is still supported in direction by both databases and by previous smaller samples, so I would not move the verdict from CONDITIONAL to rejection. However, the claimed cross-database robustness is weaker than presented: the distributions differ significantly, and the erroneous p-value (along with the implausible H statistics in Table 1) undermines confidence in the reported statistical output. The harmonization concern and the p-value concern interact: if the two questions measure somewhat different constructs, the observed significant difference is even less surprising. The concrete test (recomputing the chi-square from the reported counts) settles whether the no-difference claim is simply a typo or a substantive misuse of statistics. If the p-value is corrected, the authors must rephrase the robustness argument; if it somehow remains 0.99, the reported percentages or group sizes must be wrong. Either way, the manuscript needs a correction, but the core descriptive finding is not overturned. I therefore keep the reader's CONDITIONAL verdict unchanged.","tokens_in":42496,"tokens_out":5003,"duration_ms":59162,"concrete_test":"Recompute the chi-square test of independence on the 3x2 table of nap-response groups from Section 3.1 and the tables: Tinnitus Hub (1404, 4204, 507) vs. TRI (642, 2651, 334). If the resulting p-value is <0.001 rather than 0.99, then the no-difference statement in Section 3.1 is false and must be corrected, along with any conclusions that rely on statistically identical distributions across databases. Running the published GitHub analysis on the raw counts (or on the two contingency tables) would also indicate whether the error is in the statistical test or in the reported table values.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.1 reports that there are 'no significant differences between the distributions of participants' modulation of tinnitus through naps in the two databases (p = 0.99).' This is internally inconsistent with the numbers given in the same section and in Tables 2 and 3. Using the reported group sizes (Tinnitus Hub: 1404 worse, 4204 no effect, 507 better; TRI: 642 worse, 2651 no effect, 334 better), a 3x2 chi-square test gives roughly chi-square(2) = 38, p < 0.001, not p = 0.99. Even collapsing to 'any influence vs. none' gives a strongly significant difference (31.1% vs. 26.9%, p < 0.001). The paper's central robustness argument is that two independently recruited samples produce congruent findings; the stated p = 0.99 is the numerical support for the claim that the nap-response distributions are statistically indistinguishable, and that support is invalid as reported. The substantive descriptive finding (worsening more common than improvement, and associations with night sleep, stress, and somatosensory modulation) may survive a corrected test, but the manuscript would need to acknowledge that the two samples differ statistically on nap modulation, and the discussion of 'no significant differences' and minimized sample bias must be revised. This concern is reinforced by Table 1, where the Kruskal-Wallis H statistics for age, loudness, and annoyance are implausibly large (e.g., H = 13465652), suggesting a reporting/calculation error that warrants an audit of the statistical output.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript analyzes two large existing tinnitus databases—the Tinnitus Hub web survey (n = 6,115) and the TRI tertiary-clinic database (n = 3,627)—to describe the frequency, direction, and clinical correlates of nap-induced modulation of tinnitus. Within each database, participants are classified into three groups (nap worsens tinnitus, no effect, nap improves tinnitus) via a harmonized three-category outcome, and the groups are compared on demographic, audiological, and comorbidity variables using Kruskal-Wallis and chi-square tests with Holm-corrected post-hoc tests. The two databases' results are then compared to assess replication. The main descriptive findings are that 31.1% (Tinnitus Hub) and 26.9% (TRI) report any nap influence, that worsening is about twice as common as improvement, and that nap-modulated groups more often report influences of night sleep, stress, somatosensory maneuvers, and physical activity on their tinnitus. The authors interpret the across-database consistency of these patterns as evidence of robustness and as support for nap sensitivity as a clinically distinct tinnitus subgroup, with sleep-related and serotonergic mechanisms proposed as candidate explanations.","tokens_in":42763,"tokens_out":24244,"duration_ms":231613,"significance":"The descriptive core of this paper—that roughly one-quarter to one-third of tinnitus sufferers report nap-induced modulation, with worsening predominating—is clinically useful and consistent with prior smaller studies (van den Berge et al.; Guillard et al.), so the manuscript merits consideration if its statistics hold up. Strengths include the large combined sample, separate within-database analysis with an explicit replication logic, appropriate Holm control for multiple comparisons, unusually complete group-comparison tables in the Supplementary Material, and public availability of the analysis code on GitHub. The hypotheses generated (bruxism/sleep-apnea mediation of nap-related worsening; sound-avoidance explanations for nap-related improvement; a possible serotonergic link) are falsifiable in polysomnographic follow-up studies. However, the numerical support for the central congruence claim is internally inconsistent (the stated p = 0.99 contradicts the paper's own counts), the omnibus statistics in Table 1 are implausible as printed, and the effect sizes in the group-comparison tables are computed with a nonstandard formula.","major_comments":[{"comment":"The claim in Section 3.1 that there are 'no significant differences between the distributions of participants' modulation of tinnitus through naps in the two databases (p = 0.99)' is contradicted by the counts reported in the same section and in Tables 2 and 3 (Tinnitus Hub: 1,404 worse, 4,204 no effect, 507 better; TRI: 642 worse, 2,651 no effect, 334 better). A 3×2 chi-square test on these counts gives chi-square(2) ≈ 38.3, p < 0.001; even collapsing to 'any influence vs. none' yields a significant difference (31.3% vs. 26.9%, p < 0.001). Because this p-value is the only quantitative support for the paper's repeated claim that the two independently recruited samples yield congruent nap-modulation distributions (Discussion; Conclusions: 'the risk of a sample bias is minimized'), the claim as printed is invalid. The authors should recompute the test, report the small but significant difference, and revise the congruence and sample-bias statements accordingly.","section":"§3.1"},{"comment":"The Kruskal-Wallis H statistics in Table 1 are not interpretable: H = 13,465,652 for age, H = 12,739,120 for loudness, and H = 13,115,675 for annoyance. For two independent samples of sizes 3,627 and 6,115, the Kruskal-Wallis statistic is bounded above by roughly N−1 ≈ 9,740 in the two-group case, so values in the tens of millions cannot be H statistics as described. The p-values and effect sizes in the same table may be approximately correct, but the printed statistics are wrong. Together with the inconsistency in Section 3.1, this indicates that the statistical analysis should be audited in full against the analysis code (which is publicly available on GitHub), and all omnibus statistics, p-values, and effect sizes should be recomputed and reported correctly.","section":"Table 1"},{"comment":"The effect sizes labeled 'Cramér's V' in Tables 2, 3, and the Supplementary tables do not follow the standard definition V = sqrt(chi-square/(N·min(r−1,c−1))). For example, Table 3, 'Influence of a good night sleep' (3×3, chi-square = 1373.5, N = 6,115) reports V = 0.237, whereas the standard formula gives sqrt(1373.5/(6115·2)) = 0.335; the reported value corresponds to using (r−1)(c−1) = 4 in the denominator. Likewise, Table 2, 'Sleep at night and tinnitus during the day' (2×3, chi-square = 399.0, complete-case N = 2,033) reports V = 0.235, but the standard formula gives sqrt(399.0/2033) = 0.443. Because the distortion depends on the table dimensions, the reported effect sizes are not comparable across variables, which affects the Discussion's claim that sleep variables show the 'largest effect sizes'. Please recompute all effect sizes with the standard formula and the appropriate complete-case N.","section":"Tables 2, 3; Supplementary Tables 1–2"},{"comment":"The cross-database comparison underlying the congruence claim rests on the harmonization of two differently worded items: TRI asks 'Does taking a nap during the day affect your tinnitus?' with three answers, while Tinnitus Hub asks 'How do the following affect your tinnitus: waking up from napping?' with six answers, and Section 2.3 merges the 'a lot'/'a little' intensity categories. This assumes both that the intensity distinction is irrelevant and that 'taking a nap' and 'waking up from napping' elicit the same construct—an assumption that matters because the paper itself discusses awakening-linked phenomena (e.g., 'morning roar', and somatosensory modulation 'perceived upon awakening') as mechanistically relevant. The Limitations paragraph acknowledges wording differences, but the Conclusions' statements that 'very similar results were found regarding group distribution' and that 'the risk of a sample bias is minimized' go beyond what the harmonized comparison can support. I recommend reporting the unmerged Tinnitus Hub distribution, testing the sensitivity of the cross-database comparison to the merging rule, and softening the congruence claims accordingly.","section":"§2.3, §3.1, §5"}],"minor_comments":[{"comment":"The total sample size is given as 9,724 in the Abstract but as 9,742 in the body (Section 5); the sum of the two database sizes (6,115 + 3,627) is 9,742. Please make the numbers consistent throughout.","section":"Abstract; §5"},{"comment":"The reported Tinnitus Hub percentages are inconsistent with the counts in the same section: (1,404 + 507)/6,115 = 31.25% (reported as 31.1%) and 507/6,115 = 8.3% (reported as 8.1%). The 31.1% appears to be an artifact of summing the rounded components 23.0% + 8.1%. Please recompute all descriptive percentages from the raw counts.","section":"§3.1"},{"comment":"There are duplicated subsection numbers: two subsections are numbered 2.2.2 ('TRI database specific variables' and 'Tinnitus Hub survey database specific variables') and two are numbered 2.4 ('Statistical analysis' and 'Software choices'). Please renumber the sections.","section":"§2.2–§2.4"},{"comment":"The sentence 'Such differences between samples collected in a clinical environment and samples collected on an online platform on tinnitus have already been reported in the past [29]' cites reference 29, which is the Statsmodels software paper; the intended citation is presumably reference 22 (Probst et al., 2017), which compares recruitment samples from an outpatient clinic, a self-help web platform, and a mobile application.","section":"Discussion"},{"comment":"For the item 'Sleep at night and tinnitus during the day', 43–45% of each group is missing and the reported percentages are computed on complete cases only (n = 2,033 overall). Given this variable's prominence in the Discussion, please state the complete-case basis explicitly in the table caption and consider a sensitivity analysis addressing the missingness.","section":"Table 2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript fits the journal's scope as a large observational clinical study. The main reviewer concern is the integrity of the reported statistics: the paper's own counts contradict the stated p = 0.99 in Section 3.1, and Table 1 contains impossible Kruskal-Wallis H values. Because the analysis code is on GitHub, a full statistical audit is feasible and should be requested as a condition of revision. The harmonization of the two nap items is the weakest design assumption but is secondary to the reporting errors. The reference list includes several self-citations (e.g., refs 12, 18, 31), which are defensible given the niche topic; the substantive citation error is the use of ref 29 in the Discussion. Finally, the two databases differ on most measured characteristics (Table 1), so the manuscript's robustness argument should be reframed as replication of within-sample contrasts rather than statistical identity of the nap-modulation distributions."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bottom line: this is a useful descriptive subtyping study, not a mechanistic advance. The replication across two large independent samples is genuinely valuable, but the paper's headline claim of 'no significant difference' between databases is contradicted by its own reported counts, and the statistics need a careful audit before this is publishable.\n\nWhat is actually new: earlier work already reported nap modulation prevalence; the new contribution is the systematic profile of nap-sensitive tinnitus patients across dozens of clinical variables, replicated in an online citizen-science cohort and a clinical cohort. That is a solid descriptive application, with appropriate Holm correction and effect sizes. The code is on GitHub, which is good practice.\n\nThe soft spots are real. Section 3.1 reports p = 0.99 for the comparison of nap-response distributions between databases. Recomputing a chi-square from the paper's own numbers (Tinnitus Hub: 1404 worse, 4204 no effect, 507 better; TRI: 642 worse, 2651 no effect, 334 better) gives chi-square(2) roughly 38, p < 0.001. That is not a rounding artifact. The central robustness argument—that two differently recruited samples produce congruent distributions—is numerically unsupported. Also, Table 1 lists Kruskal-Wallis H statistics like 13,465,652 for age, which are implausible for these sample sizes; something is off in the statistical output. The harmonization of differently worded questions (3-point TRI scale vs 6-point Tinnitus Hub scale) is assumed rather than tested; if the response patterns differ by question format, the cross-database congruence is partly an artifact of the merging rule. Data being available only on request limits reproducibility, though that is common for clinical databases.\n\nNone of this makes me doubt the central descriptive finding: worsening after naps is roughly twice as common as improvement, and nap-sensitive patients more often report modulation by night sleep, stress, and somatosensory maneuvers. That pattern is consistent across both samples and fits prior literature. But the manuscript must correct the statistical reporting, rerun the database comparisons, and either justify the harmonization or treat it as a sensitivity analysis.\n\nWho is this for? Tinnitus clinicians and researchers working on subtypes, and people designing sleep-intervention studies. It deserves a serious referee after revision—not a desk reject—because the core descriptive result is likely correct and the study is one of the largest on this specific phenomenon.","headline":"Useful descriptive subtyping study whose cross-database 'no difference' claim is contradicted by its own numbers; needs a statistical audit before publication.","tokens_in":43345,"tokens_out":1450,"would_cite":false,"duration_ms":19328,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Naps change tinnitus in roughly three in ten sufferers, and worsening is about twice as common as improvement.","keywords":["tinnitus","daytime napping","nap-induced modulation","sleep disorders","somatosensory modulation","stress reactivity","tinnitus subtyping","cross-sectional database analysis"],"falsifier":"A decisive test would be a prospective nap study in which people with tinnitus rate loudness immediately before and after a monitored daytime nap, with sleep stages recorded. If fewer than about one in four report any post-nap change, or if worsening does not outnumber improvement by roughly two to one, the survey-based prevalence and asymmetry would not survive direct measurement. A simpler check of the harmonization step would be to ask the same respondents both question formats and see whether the merged responses agree.","tokens_in":42297,"feed_emoji":"😴","tokens_out":10052,"duration_ms":98541,"temperature":0.7,"pith_summary":"This paper seeks to establish that daytime naps are a real, clinically usable modulator of tinnitus rather than a bedside anecdote. In two independent samples totaling 9,742 people with tinnitus, it finds that roughly 27-31% report some nap-related change, and that worsening is about twice as common as improvement (23.0% versus 8.1% in one sample; 17.7% versus 9.2% in the other). The same individuals who report nap effects also more often report that night sleep, stress, head/jaw/neck movements, and physical activity modulate their tinnitus. Because a clinic-based sample and an online-survey sample agree on these patterns, the paper argues that nap sensitivity identifies a distinct tinnitus subgroup that should be tracked in phenotyping studies.","feed_headline":"Naps worsen tinnitus twice as often as they improve it","feed_subtitle":"In two databases with 9,742 tinnitus patients, nap-linked changes track sleep, stress, and somatic triggers.","key_machinery":"The central tool is a three-way classification of every participant by self-reported nap response: tinnitus worsened by naps, no effect, or tinnitus improved by naps. The clinical database used an explicit three-choice question; the online survey's six-choice answers were harmonized by merging the two 'worse' categories into one and the two 'better' categories into one, with 'unsure/don't know' treated as missing. Each database was analyzed separately, comparing the three groups with non-parametric tests for rating scales and chi-squared tests for categorical variables, followed by multiple-comparison-corrected pairwise post-hoc tests. This classification does the work of turning a two-sided clinical anecdote into a stable population-level contrast that can be replicated across sampling contexts.","core_discovery":"In the paper's own terms, the central discovery is that nap-induced modulation of tinnitus is a measurable, reproducible clinical phenomenon with a consistent asymmetry: more sufferers report that a nap worsens their tinnitus than report that it improves it, and this holds in two differently recruited samples. In the online survey database, 31.1% of 6,115 respondents reported an effect of napping, with 23.0% worse and 8.1% better. In the clinical database, 26.9% of 3,627 outpatients reported an effect, with 17.7% worse and 9.2% better. The group whose tinnitus worsens after naps is more than twice as large as the group whose tinnitus improves in both samples, and both nap-sensitive groups differ from the no-effect group in their responses to night sleep, stress, head/jaw/neck movements, and physical activity. The authors conclude that nap sensitivity is a consistent clinical feature that should be considered in tinnitus characterization and subtyping.","pith_inferences":["We infer that if nap sensitivity survives direct prospective testing, the nap question could become a cheap stratification tool in clinical trials, because the worsened and improved groups may respond differently to sleep-focused or sound-based interventions.","We infer that the nap-worsened subgroup's broad reactivity to stress, sound, and movement points toward a shared central gain mechanism; measuring acoustic discomfort thresholds or startle responses in this group versus the no-effect group would be a direct way to look for it.","We infer that nap duration, timing, and sleep debt, none of which the present databases captured, may explain part of the individual variation; a diary or actigraphy follow-up could refine the binary worse/better distinction.","We infer that if the somatosensory link is confirmed by muscle or breathing recordings during naps, nap worsening would become a treatable sign: addressing bruxism or sleep apnea during daytime sleep might reduce the post-nap spike."],"forward_implications":["Nap response should be recorded as a standard item in tinnitus phenotyping, because the three groups differ consistently in sleep, stress, and somatic reactivity.","The roughly one-in-five sufferers whose tinnitus worsens after naps also more often report worsening from stress, poor sleep, sound exposure, and body movements, suggesting a broadly sensitive subgroup.","The group whose tinnitus improves after naps overlaps with hyperacusis and sound-induced discomfort, consistent with improvement from resting in a quiet environment.","The replicated link between nap modulation and night-sleep effects supports a bidirectional sleep-tinnitus interaction rather than sleep disturbance being only a consequence of tinnitus.","These associations argue for targeted sleep-recording studies of the nap-worsened group, including monitoring of teeth-grinding and snoring or apnea events during daytime sleep."],"supporting_citations":[{"why":"It supplies the earlier cluster-analysis sample whose nap-response proportions (16.4% worsened, 12.9% improved) anchor the prevalence comparison.","marker":"[12]"},{"why":"It provides the controlled polysomnographic comparison of tinnitus patients with and without nap-induced increases, the main mechanistic precedent for the subgroup claim.","marker":"[18]"},{"why":"It describes the clinical database that contributes one of the two analyzed samples.","marker":"[19]"},{"why":"It describes the online survey database that contributes the other analyzed sample.","marker":"[20]"},{"why":"It defines the original clinical nap-influence question used for the clinic-based sample's outcome.","marker":"[10]"},{"why":"It establishes nap as a standard modulatory factor in the tinnitus screening questionnaire tradition the study extends.","marker":"[11]"},{"why":"It compares recruitment channels and supports the design rationale that combining clinical and online samples reduces selection bias.","marker":"[22]"},{"why":"It reports a smaller-sample replication of the nap-response distribution (23.5% worsened, 11.8% improved) that reinforces the prevalence estimate.","marker":"[31]"}],"fun_headline_variants":["Naps twice as likely to worsen tinnitus than improve it","Nap-time tinnitus worsening is more common than improvement","Tinnitus sufferers: napping often makes the ringing worse","Study: naps aggravate tinnitus more often than they help","Napping worsens tinnitus for more patients than it improves"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the differently worded nap questions in the two databases—three answer choices in one, six in the other—measure the same thing once the 'little' and 'a lot' responses are merged; if the question formats trigger systematically different response patterns, the cross-database agreement could be an artifact of the merging rule.","fun_headline_variants_meta":{"raw":{"variants":["Naps twice as likely to worsen tinnitus than improve it","Nap-time tinnitus worsening is more common than improvement","Tinnitus sufferers: napping often makes the ringing worse","Study: naps aggravate tinnitus more often than they help","Napping worsens tinnitus for more patients than it improves"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000399,"raw_usage":{"total_tokens":2107,"prompt_tokens":986,"completion_tokens":1121,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":602,"completion_tokens_details":{"reasoning_tokens":1042}},"tokens_in":602,"tokens_out":1121,"duration_ms":13124,"temperature":1.0,"reasoning_tokens":1042,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T16:29:20.949673+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive test would be a prospective nap study in which people with tinnitus rate loudness immediately before and after a monitored daytime nap, with sleep stages recorded. If fewer than about one in four report any post-nap change, or if worsening does not outnumber improvement by roughly two to one, the survey-based prevalence and asymmetry would not survive direct measurement. A simpler check of the harmonization step would be to ask the same respondents both question formats and see whether the merged responses agree.","supporting_citations":[{"cited_title":"Cluster Analysis to Identify Possible Subgroups in Tinnitus Patients","cited_arxiv_id":null,"evidence_quote":"It supplies the earlier cluster-analysis sample whose nap-response proportions (16.4% worsened, 12.9% improved) anchor the prevalence comparison."},{"cited_title":"The Tinnitus Research Initiative (TRI) Database: A New Approach for Delineation of Tinnitus Subtypes and Generation of Predictors for Treatment Outcome","cited_arxiv_id":null,"evidence_quote":"It describes the clinical database that contributes one of the two analyzed samples."},{"cited_title":"The Presence of Physical Symptoms in Patients With Tinnitus: 2023, 12, x FOR PEER REVIEW 14 of 37 International Web-Based Survey","cited_arxiv_id":null,"evidence_quote":"It describes the online survey database that contributes the other analyzed sample."},{"cited_title":"Standardised Profiling for Tinnitus Research: The European School for Interdisciplinary Tinnitus Research Screening Questionnaire (ESIT-SQ)","cited_arxiv_id":null,"evidence_quote":"It establishes nap as a standard modulatory factor in the tinnitus screening questionnaire tradition the study extends."},{"cited_title":"Outpatient Tinnitus Clinic, Self-Help Web Platform, or Mobile Application to Recruit Tinnitus Study Samples? Front","cited_arxiv_id":null,"evidence_quote":"It compares recruitment channels and supports the design rationale that combining clinical and online samples reduces selection bias."},{"cited_title":"Validated French Translation of the ESIT-SQ Standardized Tinnitus Screening Questionnaire","cited_arxiv_id":null,"evidence_quote":"It reports a smaller-sample replication of the nap-response distribution (23.5% worsened, 11.8% improved) that reinforces the prevalence estimate."}],"review_version":1}