Neurocognitive and Neuropsychiatric Trajectories in a Post-COVID Cohort: A Descriptive Longitudinal Study

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Please see below 

 

The article addresses the important topic of long-term neurological manifestations following SARS-CoV-2 infection (cognitive dysfunction specifically). Given the magnitude of SARS-CoV-2 sequelae, the topic is timely and the paper is a useful addition to the existing literature. Some revisions are however necessary before the paper can be considered for publication. Some suggestions to improve the manuscript are below. I specifically address the manuscript in terms of coherence, language, flow, clarity, structure and terminology. Specialist peer review is needed for other aspects of the paper such as statistical analysis and the specific tests performed.

General comments: PACS is not the most common term used (Long Covid /LC and Post Covid condition are, for example). But it’s not wrong, so obviously it’s not erroneous to use it. Please cite Post Covid condition in the text as well, as you mention the WHO case definition which is technically for post Covid condition (although WHO also uses LC).

The paper is generally well written and clear, and grammatically correct. Clarity can be occasionally improved. The methods section can be improved. The bibliography can be improved. Please spell out acronyms and abbreviations clearly and early on in the paper, when you don’t do that. Please clarify in the methods how the assessment was conducted: please provide more details. The results and discussion can be polished for better clarity.

Line 22. Completers. Please clarify in relation to your study.

Line 24. Broadly speaking, you mean in all of your sample? Please clarify

Line 27. May: unclear. You mean it can happen in your sample? In how many patients? In general? Please clarify.

Keywords. Please spell the acronym out.

Line 39 to 40. Please add bibliography after first sentence. Please add bibliography in the following sentences about multi-system pathology.

Line 44. Conservative. Unclear. I get what you mean I presume, but please make it clear.

Line 51. It can also appear already during the acute phase in some cases?

Line 54. Following rather than recovering? Not everyone recovers.

Line 59. Please add at least a couple of papers, not only one.

Line 66. Again, more bibliography would be appreciated.

Line 88. Here you introduce another term for PACS / LC and speak about a four weeks threshold, contradicting what you said above about the three months threshold. I presume this is an early terminology and definition you used early on in the pandemic, as you imply in the following sentence. Please clarify and rephrase. When were you using those terms and thresholds? At the time of the study approval, which appears to be in 2020? Or you used initially the NICE guidelines of 2021? Did you change terminology and other relevant aspects during the study? Or for writing this article?

Line 100. Spontaneous recovery. How are you sure there is recovery and not just improvement or remission? How do know it’s spontaneous? No one was getting support or medication which could have aided such recovery process?

Fig 1. Please spell the acronym used in the figure out in the caption. Otherwise it’s unclear. I understand you explain the acronyms in the paper but readers would struggle to locate each one in the text.

Line 106 and following. Please better address this in view of what you said above about the timeline and thresholds.

Please justify why you chose the specific tests performed.

Line 148. Please add bibliography.

Line 174. I would add also the percentage of males for additional clarity.

Line around 181. No death that you know of? How did you gather the reasons for not completing the study? Follow up questions? Or are these hypotheses?

Line 193. Yes and no. There are many factors at play here e.g. those who had an overt pulmonary embolism (PE) and received treatment might have improved better than others who had micro-thrombi in the distal pulmonary vessels, were not appropriately diagnosed and treated, and remained disabled because of this, while appearing less severe on paper etc. A nuanced discussion is needed here (the PE case is just an example). Not saying you are wrong in your assessment. But it’s complicated.

Table 1. All conditions were present prior to COVID? Please clarify further.

Line 257. May like in some or most people in your study but not all?

Line 265. Please break sentence for additional clarity and better flow.

Line 267. Unclear. Please clarify.

Line 273. More severe acute illness or in general? Please see comment above.

Line around 286. These manifestations can linger in LC, not just be present in the acute phase, which could help explain the persistence of symptoms and cognitive impairment. Please rephrase.

Line 291. Did the patients use medication? Practice effects: unclear.

Line around 293. Above you gave the impression “recovery” was the norm but obviously not everyone apparently recovered or fully improved. Please double check the language around this aspect across the manuscript.

Line 294. How many patients?

Line 307. What do you mean by mental fatigue? Please clarify.

Line 314. Or are neurological and other manifestations (e.g. dysautonomia) presenting as “anxiety” and the like rather than viceversa?

Line 332. What do you mean by targeted rehabilitation? If symptoms are caused by e.g. neuroinflammation or neurovascular manifestations, would rehabilitation be enough? Do you include medication e.g. for depression and sleep disturbances in “rehabilitation”. If not, shouldn’t be these approaches be considered?

Line 333. Which approaches?

Line 344. What do you mean by general PACS population? In your overall sample?

Line 353. Please further clarify. See comments above.

Please clarify the study is based on a 2020 sample that might not be generalised to a population first infected after being vaccinated or developing LC after infection with more recent variants.

Author Response

Response to Reviewers

 

Neurocognitive and neuropsychiatric trajectories in a Post-COVID cohort: a descriptive longitudinal study

 

We thank both reviewers for the care they have given this manuscript. The comments have led to substantive changes: the longitudinal analyses have been re-run with correction for multiple comparisons and post-hoc testing, a sensitivity analysis on the timing of the baseline assessment has been added, the Methods have been expanded so that the analysis can be reproduced, and the reference list has been extended with large international studies. All changes are marked in the tracked-changes version.

 

Two of the revisions altered what the manuscript reports, and we describe them openly at the outset rather than leaving them to be discovered in the tables.

 

First, correction for multiple comparisons reduces the number of significant longitudinal changes from six to three. Reviewer 2 was right that the Benjamini–Hochberg procedure had been applied only to the cognitive–neuropsychiatric association analyses. Applied across the twenty cognitive outcomes, Corsi span forward (FDR-adjusted p = 0.001), Corsi span backward (p = 0.013) and Digit Symbol (p = 0.013) remain significant, whereas verbal short-term learning (p = 0.128), constructional praxis (p = 0.090) and phonological fluency (p = 0.167) do not. We now report both the unadjusted and the adjusted results, treating the former as exploratory, and have revised the abstract, Results and Discussion accordingly.

 

Second, the persistence of impairment is now quantified. Where the manuscript previously stated that deficits persisted "in a proportion of individuals", it now reports that 28 of 42 patients (66.7%) still scored in the impaired range on at least one test at 12 months, and 17 (40.5%) on two or more.

 

---

 

# Reviewer 1

 

## General comments

 

Terminology (PACS, long COVID, post COVID-19 condition). Adopted. The opening sentence now introduces all three terms and attributes *post COVID-19 condition* to the WHO, whose case definition we cite. The keywords have been rewritten accordingly and the acronym is spelled out.

 

Acronyms. All acronyms are now spelled out at first occurrence, including in the keywords and in the legend of Figure 1.

 

Methods insufficiently detailed. Addressed in several places: the rationale for the test battery, the classification threshold, the criteria for data completeness and the handling of missing data, the tests for which alternate forms exist, the interval between infection and baseline assessment, and the multiple-comparison and post-hoc procedures.

 

Bibliography. Five references have been added, all to large international studies or systematic reviews, and cited at the points where the reviewer asked for support.

 

Language of "recovery". The reviewer raised this at lines 54, 100 and 293. We have gone through the manuscript and replaced *recovering from COVID-19* with *following COVID-19*, *following recovery from acute COVID-19* with *following the acute phase of COVID-19*, and *spontaneous recovery* with *natural symptom resolution*. Where improvement is described we now quantify it rather than calling it recovery.

 

## Specific comments

 

Line 22 (completers). The abstract now defines the term: patients who completed all three assessments (n = 42) compared with those who attended the first evaluation but did not complete follow-up (n = 544).

 

Line 24. Clarified: the improvements are group-level and refer to the whole sample.

 

Line 27. Quantified, as described above.

 

Lines 39–40, 59, 66, 148 (bibliography). References added, see the general comment.

 

Line 44 (conservative). Replaced with *more restrictive*, and the criterion is now stated explicitly.

 

Line 51. The reviewer is correct. The sentence now states that cognitive symptoms may already be present during the acute illness itself, as well as arising in the subacute phase or after discharge.

 

Line 54. Amended, see the general comment on the language of recovery.

 

Line 88 (four weeks versus three months). This was a genuine inconsistency in exposition. The Methods now explain that recruitment began in September 2020 under the operational definition in force at the time — symptoms persisting at least four weeks, subsequently formalised in the NICE guideline of 2021 — and that the three-month criterion later endorsed by the WHO, NASEM and CDC was published after recruitment had begun and was not applied retrospectively. Inclusion criteria were not modified during the study. Terminology in the article follows current usage while the inclusion window reflects the definition in force at enrolment.

 

Line 100 (spontaneous recovery). The word has been removed. We have also added to the limitations that concomitant treatments received during follow-up — cognitive rehabilitation, psychological support, antidepressant or hypnotic medication — were not systematically recorded, so the extent to which improvement reflects unassisted change cannot be determined.

 

Figure 1. All acronyms are now defined in the caption.

Rationale for the tests. Added to the Methods: the battery was assembled to cover the five domains most consistently reported as affected after SARS-CoV-2 infection, using tests with Italian normative data allowing conversion into equivalent scores and, where possible, tests available in alternate forms.

 

Line 174 (percentage of males). Added: 15 participants (35.7%) female and 27 (64.3%) male. We also corrected an error the reviewer did not see: the comparison with non-completers stated that completers were more frequently female, whereas the figure quoted (64.3%) refers to males. This has been corrected.

 

Line 181 (reasons for non-completion). The reviewer is right to ask. These were established retrospectively from the clinical records of the service and from the impressions of the clinicians managing follow-up, not through structured contact with patients who discontinued, and the text now says so and describes them as indicative rather than quantified. The same applies to the absence of recorded deaths.

 

Line 193 (severity). A nuanced passage has been added. Documented acute severity may not reflect true severity: patients with overt complications such as pulmonary embolism were identified and treated, whereas undiagnosed distal microvascular involvement would leave a patient classified as less severe while potentially contributing to persistent impairment. Recorded severity should be read as a proxy for the intensity of acute care received rather than as a direct measure of biological insult.

 

Table 1 (comorbidities). The note now states that the listed comorbidities were documented prior to SARS-CoV-2 infection and were derived from the hospital admission records, and that pulmonary embolism and oxygen therapy refer to the acute episode.

 

Lines 257, 265, 267, 273. The modal verbs have been replaced with quantified statements where the data allow; the long sentence at line 265 has been split; the clause at line 267 has been rewritten as a separate sentence; and the reference to severity now points to the nuanced discussion added at line 193.

 

Line 286. Rephrased: these manifestations are not confined to the acute phase, and neuroinflammatory, endothelial and microvascular alterations may persist into the post-acute period.

 

Line 291 (medication, practice effects). Medication is addressed in the limitations, as above. The claim about practice effects has been softened: alternate forms were available only for the Rey Auditory Verbal Learning Test, the copy of the Rey-Osterrieth Complex Figure and Trail Making Test A and B, and procedural familiarity with the testing situation may still contribute to improved performance.

 

Line 293. Addressed in the general comment on the language of recovery.

 

Line 294. The claim that some individuals worsened could not be quantified from the available data and has been replaced with a statement that is: in one domain — verbal short-term span — the proportion of impaired patients increased across assessments, from 7.1% at t0 to 11.9% at t1 and 16.7% at t2.

 

Line 307 (mental fatigue). Defined in the text as the subjective difficulty of sustaining cognitive effort over time, with an explicit statement that it was drawn from the literature and not measured with a dedicated instrument in this study.

 

Line 314 (anxiety or dysautonomia). An excellent point, now addressed. Several BAI items capture autonomic symptoms — palpitations, breathlessness, dizziness, feelings of instability — that are also cardinal features of the dysautonomia described in PACS, so anxiety scores in this population may partly index autonomic dysfunction rather than anxiety as such.

 

Lines 332–333 (targeted rehabilitation, approaches). Both specified. Cognitive rehabilitation targeting the domains involved may be considered, although its efficacy in PACS is not established; rehabilitation alone is unlikely to suffice where symptoms are sustained by neuroinflammatory or neurovascular mechanisms, and pharmacological treatment of comorbid depression, anxiety and sleep disturbance, as well as management of autonomic symptoms, should be part of an integrated approach.

 

Line 344 (general PACS population). Replaced with a reference to the broader population attending the Post-COVID service, of which our completers are a small and selected subgroup.

 

Line 353 and generalisability. Added to the limitations: the cohort was infected in 2020, before vaccination was available and before the emergence of the Omicron and subsequent variants, and the findings should not be generalised to patients infected under those conditions.

 

We are grateful for the time both reviewers have given it.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents the results of a longitudinal observational study on the dynamics of neurocognitive and neuropsychiatric symptoms in 42 patients with post-acute COVID-19 syndrome (PACS). The topic of the work is highly relevant, and the use of in-person standardized neuropsychological testing at 3 time points enhances the scientific value of the study. The manuscript is written in academic language and is generally logically structured. The cited literature is mostly relevant.

Despite its merits, the work has several significant limitations. First, the experimental design is characterized by a small sample size, the absence of a control, and considerable heterogeneity in the timing of baseline assessments. Second, the methodological section is insufficiently detailed, which limits the reproducibility of the results. Finally, some conclusions regarding persistent impairments are not supported by quantitative estimates. The authors correctly acknowledge most of these limitations in the Discussion; however, this does not obviate the need to account for them when interpreting the findings.

Conclusion. The manuscript contains valuable longitudinal data on a pressing issue, but it requires substantial revision to enhance its scientific rigor and practical significance. Responses to the questions and comments presented below should be addressed in the relevant sections of the manuscript.

Critical Comments and Questions:

1. Selection bias. Since the patients who completed the study ("completers") were initially more severely ill than those who dropped out, could the observed improvement reflect statistical regression to the mean rather than genuine recovery? This factor should be taken into account.
2. Heterogeneity of baseline timing. The initial assessment was conducted within an interval ranging from 1 to 7 months after acute infection. Was any analysis performed with time from infection resolution to t0 included as a covariate? If not, how can the effects of PACS be distinguished from natural post-hospital recovery?
3. Multiple comparisons. Was correction for multiple comparisons (e.g., FDR) applied to all primary cognitive outcomes in the analysis of changes over time, or only to the association tests with psychiatric scales? Post-hoc analysis is needed to identify the specific time points at which significant differences occurred.
4. Lack of improvement in certain domains. How can the absence of significant changes in categorical fluency and MFTC accuracy be explained - by low test sensitivity or by persistent deficits in these areas? This should be discussed.
5. To ensure reproducibility, please clarify in the Methods section: the threshold values used for classifying results; the criteria for data completeness and the method for handling missing data; and whether alternative test forms were used to minimize practice effects.
6. It is recommended to expand the Reference list to include independent large-scale international studies in order to strengthen the objectivity of the bibliographic review.
7. Epidemiological heterogeneity. The patient enrollment period spanned more than two years, during which various dominant SARS-CoV-2 strains successively replaced one another (from the original variant through the Delta and Omicron lineages), differing substantially in virulence, neurotropism, and the spectrum of clinical manifestations. This may have influenced both the severity of the acute phase and the structure and severity of post-COVID neurocognitive symptoms. It is recommended to conduct an additional stratified analysis and also to indicate the proportional distribution of patients across periods corresponding to the circulation of different strains, in order to assess the possible impact of this factor on the results.
8. Technical note. In the author list, the line "...Antinori Andrea and MD" should be corrected by replacing "and" with a comma.

Author Response

Response to Reviewers

 

Neurocognitive and neuropsychiatric trajectories in a Post-COVID cohort: a descriptive longitudinal study

 

We thank both reviewers for the care they have given this manuscript. The comments have led to substantive changes: the longitudinal analyses have been re-run with correction for multiple comparisons and post-hoc testing, a sensitivity analysis on the timing of the baseline assessment has been added, the Methods have been expanded so that the analysis can be reproduced, and the reference list has been extended with large international studies. All changes are marked in the tracked-changes version.

 

Two of the revisions altered what the manuscript reports, and we describe them openly at the outset rather than leaving them to be discovered in the tables.

 

First, correction for multiple comparisons reduces the number of significant longitudinal changes from six to three. Reviewer 2 was right that the Benjamini–Hochberg procedure had been applied only to the cognitive–neuropsychiatric association analyses. Applied across the twenty cognitive outcomes, Corsi span forward (FDR-adjusted p = 0.001), Corsi span backward (p = 0.013) and Digit Symbol (p = 0.013) remain significant, whereas verbal short-term learning (p = 0.128), constructional praxis (p = 0.090) and phonological fluency (p = 0.167) do not. We now report both the unadjusted and the adjusted results, treating the former as exploratory, and have revised the abstract, Results and Discussion accordingly.

 

Second, the persistence of impairment is now quantified. Where the manuscript previously stated that deficits persisted "in a proportion of individuals", it now reports that 28 of 42 patients (66.7%) still scored in the impaired range on at least one test at 12 months, and 17 (40.5%) on two or more.

# Reviewer 2

 

1. Selection bias and regression to the mean. Accepted. Because completers were more severely affected during the acute phase, their baseline scores were on average more extreme, and regression to the mean is a plausible partial contributor to the improvement observed. Our design cannot separate it from genuine recovery, and the limitations now say so.

 

2. Heterogeneity of baseline timing. The interval between resolution of the acute infection and the baseline assessment is now reported: median 2.9 months (IQR 2.8–5.6, range 1.3–7.6). Because the Friedman test does not accommodate covariates, we examined this factor in a sensitivity analysis correlating the interval with the magnitude of change between baseline and 12 months in the three outcomes that survived correction. No association was found (Corsi span forward rho = −0.28, p = 0.124; Corsi span backward rho = −0.21, p = 0.259; Digit Symbol rho = −0.32, p = 0.074). Heterogeneity in the timing of the first evaluation therefore does not account for the observed change, which is now stated in the Results.

 

3. Multiple comparisons and post-hoc analysis. Addressed as described in the opening section. Post-hoc pairwise comparisons were performed with McNemar's test, Bonferroni-corrected for the three contrasts, and locate the change entirely in the first interval: all three robust outcomes improved between baseline and 12 months (p = 0.001, 0.022 and 0.010), Corsi span forward and Digit Symbol improved between baseline and 6 months (p = 0.002 and 0.038), and none changed between 6 and 12 months (all p = 1.000). We have also noted in the Methods that, because the longitudinal variables are dichotomous, the Friedman test is algebraically equivalent to Cochran's Q.

 

4. Lack of improvement in categorical fluency and MFTC accuracy. Now discussed. Categorical fluency was not only stable but showed a higher proportion of impaired patients at 12 months than at 6 months, which argues against low test sensitivity and points to a genuinely persistent deficit in lexical-semantic retrieval. MFTC accuracy showed a ceiling pattern, with most patients within normal limits at every assessment, leaving little room to detect improvement.

 

5. Reproducibility. All three points are now in the Methods: performance was classified as impaired when the equivalent score equalled 0, that is below the outer tolerance limit of the Italian normative distribution; only patients with complete data at all three assessments were analysed, with no imputation and no partial contributions, so every analysis rests on the same 42 individuals; and alternate forms were used for the four tests for which they exist, the remainder being readministered in the same version.

 

6. Reference list. Five references have been added, all large international studies or systematic reviews, and cited where the persistence and scale of post-COVID cognitive impairment are discussed.

 

7. Epidemiological heterogeneity and viral variants. We have examined this and can now answer it precisely. Although enrolment extended over more than two years, every participant for whom the date of acute infection is documented was infected between March 2020 and January 2021, that is before the emergence of the Delta and Omicron lineages and before vaccination was available. No patient was infected during the Delta or Omicron periods. Stratification by dominant variant is therefore neither possible nor informative: the extended enrolment window reflects the timing of referral to the Post-COVID service rather than heterogeneity of the infecting variant. This is now stated in the Results, and the restriction of the cohort to pre-vaccination, ancestral-strain infection is acknowledged as a limitation on generalisability.

 

8. Author list. Checked. The line reads "Mazzotta Valentina, MD, Antinori Andrea, MD." with a comma in the submitted file; we will verify that the version rendered by the editorial system matches.

 

---

 

We are grateful for the time both reviewers have given it.

 

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

See below 

I thank the authors for addressing my comments and reviewing extensively the manuscript, which greatly improved the article.

A few more comments for improvement, based on language, structure, coherence, etc. Statistical analysis and specific analyses would need specialised peer review.

Please double check the numbering of the citations. e.g. line 83

Please double check font and use of bold. I see in bold some parts of the paper.

Line 75. Repetition of may (although not grammatically wrong). Long sentence. Consider rephrasing for better flow.

Line 129. Current usage. Please clarify given you use PACS while the organisations you mentioned use LC or PCC. You mean as regards duration of symptoms for diagnosis?

Fig 1 flow chart. I suggest using lighter colours because it’s difficult to read the text on my device with dark purple etc.  Also would it be possible to have the image or at least the bubbles a bit bigger?

Line 163. Five according to which studies? Bibliography

Line 288. Add bibliography on microvascular damage and the like.

Line 363. You can just use the acronym here. Double check acronym use across the paper.

Line 514. Please double check grammar and use of the

Comments for author File: Comments.pdf

Author Response

Response to Reviewer 1 — second round

Neurocognitive and neuropsychiatric trajectories in a Post-COVID cohort: a descriptive longitudinal study

We thank the reviewer for the generous assessment and for a further set of comments that were, again, precise and easy to act on. All points have been addressed; the changes are marked in the tracked-changes version.

One of the comments led us to find an error of our own, which we describe under point 1.

1. Citation numbering (e.g. line 83).

The reviewer was right, and the problem was more extensive than the single instance noted. The five references added at the previous revision had been appended at the end of the list rather than inserted at their position of first citation, so the numbering no longer followed order of appearance. The whole reference list has now been renumbered accordingly, and the in-text citations updated to match; the list runs from 1 to 51 with no gaps.

While checking this we also found that one of the references we had added in the previous round duplicated an entry already present in the list (Hampshire et al., N. Engl. J. Med. 2024, formerly numbered both 46 and 52). The duplicate has been removed and the citation now points to the single retained entry.

Because renumbering affects nearly every citation in the manuscript, it has been carried out as a clean edit rather than as tracked changes, which would have rendered the document unreadable. All other changes remain tracked.

2. Font and use of bold.

Corrected. Eighty-two passages of body text in the Methods, Results and Discussion had been set in bold, apparently through an inherited paragraph style; these have been returned to regular weight, with bold retained only for section headings and structural labels. The change is tracked as a formatting change and will appear as such in Word.

3. Line 75 — repetition of "may", long sentence.

Rephrased. The sentence has been split in two and the second modal removed:

Cognitive symptoms may already be present during the acute illness itself. They can also emerge in the subacute phase or shortly after hospital discharge, and persist for several months following the acute phase of COVID-19.

4. Line 129 — "current usage".

The reviewer's reading is correct: we meant the duration criterion, not the choice of term, and the sentence did not say so. It now reads:

Throughout this article we use PACS for consistency with the wider neuropsychological literature, while recognising that the WHO, NASEM and CDC refer to the same clinical picture as post COVID-19 condition or long COVID; the divergence from those definitions concerns the symptom duration required for inclusion, not the construct itself, and the inclusion window used here reflects the case definition in force at the time of enrolment.

5. Figure 1 — colours and size.

The flow chart has been redrawn. The saturated purple and burgundy fills have been replaced with a pastel palette (soft rose, lavender, amber, teal) carrying dark grey text, which should read clearly on any display. The nodes and type are substantially larger, the dotted connectors have been replaced with directional arrows so that the flow is unambiguous, and the image has been re-exported at 2832 × 1851 pixels, against 1025 × 753 previously, so that it remains sharp when enlarged. All numbers are unchanged. The figure is also supplied as a separate file.

6. Line 163 — "five according to which studies?"

References added. The statement that the battery covers the five domains most consistently reported as affected after SARS-CoV-2 infection is now supported by the cohort studies and meta-analyses cited elsewhere in the manuscript for that claim.

7. Line 288 — bibliography on microvascular damage.

References added to the passage discussing undiagnosed distal microvascular involvement, pointing to the mechanistic literature already cited in the Introduction.

8. Line 363 — acronym already defined.

Corrected: PACS is no longer spelled out again in the Discussion, having been defined at first occurrence.

9. Acronym use across the paper.

A systematic check was carried out and produced seven corrections beyond the one noted:

• The delayed-recall RAVLT index appeared as RAVLT-RD in Table 2 but as RAVLT-DR in the abbreviation list and in the text; RAVLT-DR is now used throughout.
• MFTC-ACC, MFTC-ERR and ROCF-DR were used in Table 2 but were missing from the abbreviation list, and have been added.
• LC and LMWH were used without being listed, and have been added.
• PACC appeared in the abbreviation list but nowhere in the text; it has been replaced with PCC (post COVID-19 condition), which the manuscript does use, and the term is now defined at first occurrence in the Introduction.
• In the Results, three tests were named in full where the manuscript's own convention is domain name followed by the acronym in parentheses. These have been brought into line, so that CSF, CSB, DS and DSF are used consistently after first definition.

 

10. Line 514 — grammar.

Corrected: "The author gratefully acknowledges" is now "The authors gratefully acknowledge".

We are grateful to the reviewer for a second reading as careful as the first, and in particular for the comment on citation numbering, which led us to a duplicated reference we would otherwise have carried into print.

 

Reviewer 2 Report

Comments and Suggestions for Authors

All comments (1-8) have been thoroughly addressed, and the corresponding revisions have been incorporated into the manuscript. It is particularly noteworthy that the authors not only responded formally to the comments but also conducted additional statistical analyses, the results of which altered the interpretation of the data, and they have openly acknowledged this in the revised text. The manuscript has been substantially improved in terms of methodological rigor, reproducibility, and transparency of interpretation.

Recommendation: The manuscript is acceptable for publication after final copyediting. No further substantive revisions are required.

Author Response

We thank the reviewer for this assessment, and for two rounds of comments that materially improved the manuscript. The point on multiple comparisons in particular changed what we are able to claim from these data, and the paper is more honest for it. We will address any remaining copyediting requests at proof stage.