REF-1242 · Micheli et al., 2026 · International Journal of Sports Physiology and Performance, ahead of print, pp. 1-13 · Paper · PubMed
Can a morning check-in tell whether I am ready to train?
Not reliably yet: 14 small studies used questionnaires, HRV or resting HR, few tested them against the next session, and none beat the rest. A prompt, not a verdict.
Micheli and colleagues asked how readiness to train is defined in endurance exercise, which tools are used to assess it, and how well those tools were validated.
They searched PubMed, Scopus and Web of Science on 26 July 2025 for endurance athletes aged 18-65 assessed for readiness before or after training or competition, with readiness compared against some standard (performance, prescription accuracy, menstrual-cycle phase), and added studies by snowballing up to September 2025.
Fourteen studies with 418 participants were included, published between 2000 and 2025. No study shared a definition: readiness was used interchangeably with fatigue and recovery, and in three studies it was treated as a practical standalone measure.
Eleven studies used self-report (perceived readiness, fatigue, soreness, sleep, mood, training distress, life demands), ten used objective markers (morning or nocturnal HRV, resting heart rate, heart-rate-to-running-speed or -power indices, countermovement jump, DFA α1 in the warm-up, interval sessions), and seven used both.
The tools moved with training load, overload and recovery, but most were validated against recovery, fatigue state or training adaptation rather than against the next session's performance, and the authors conclude that no single marker has been shown to predict subsequent performance or training tolerance better than another.
They propose a four-domain framework (physiological, psychological, perceptual, contextual) and a way to validate readiness tools against the quality of the following session; both are proposals.
Our reading: a useful inventory with a sound central message, carried on small and mixed studies and on a keyword search that missed the HRV-guided trials already in this base (REF-362, REF-438); it supports reading a check-in as a description of how the athlete is carrying recent load, not as a forecast of today's session, and it adds no evidence that one instrument is better than another.
Protocol
Design: systematic review following PRISMA, registered on the Open Science Framework (10.17605/OSF.IO/BEMYA) on 18 July 2025, eight days before the search; no meta-analysis, no certainty-of-evidence grading.
Search: PubMed, Scopus and Web of Science on 26 July 2025; endurance terms (endurance sport, running, cycling, swimming, triathlon, cross-country skiing, endurance, aerobic exercise) AND readiness terms (readiness, preparedness, ready to perform, training readiness) in title, abstract or keywords; reference lists snowballed up to September 2025.
Eligibility (PICo): endurance athletes aged 18-65 of any sex or fitness level in running, cycling, swimming, cross-country skiing or triathlon; any fitness or readiness assessment compared against another standard; any training or competition preceded or followed by a readiness evaluation (morning, immediately before the session, or in the warm-up). Excluded: non-English, grey literature, conference abstracts, clinical or rehabilitation populations.
Selection: two reviewers (Micheli, D'Alleva) independently, a third (Greco) for disagreements. 4086 records (Figure 1: PubMed 516, Scopus 1069, Web of Science 2501) → 886 duplicates removed → 3200 screened → 3136 excluded on title → 64 abstracts (35 excluded) → 29 full texts → 9 included; 5 more by snowballing → 14 studies, n = 418 participants (157 women by our count from Table 3).
Data: bibliographic details, participants, design, readiness definition, readiness indicators and the reference parameter each was validated against (Tables 1 and 3).
Quality: nine items rated by two reviewers (crossover design, randomised order, carry-over, unbiased data, allocation concealment, blinding, incomplete outcome data, selective reporting, other bias; Table 2). The text calls this the revised Cochrane Risk of Bias tool; the reference given is Ding et al. 2015 on the methodological quality of crossover studies.
Included studies (Table 3): Alfonso 2025 (cycling, 28 in three arms), Bulte 2025 (swimming, 11), De Martin Topranin 2023 (five sports, 41 women), Figueiredo 2022 (running, 36), Grove 2014 (three studies: 21 well-trained male triathletes, 12 high-level triathletes, 63 age-group swimmers aged 14.4), Hill & Chtourou 2022 (cycling, 7 women), Karu 2000 and Nurmekivi 2001 (running, 15 each, the same sample description), Nuuttila 2022a (running, 30 in two arms), Nuuttila 2022b (running, 30 in two arms), Nuuttila 2024 (running, 24), Schaffarczyk 2022 (triathlon, 11), ten Haaf 2017 (cycling, 29 in two groups), Vesterinen 2014 (running, 45).Study sizes 7 to 96 (Grove's 96 across three samples; largest single sample 63).
Funding and interests: the PDF carries no funding, conflict-of-interest or acknowledgements statement, and Crossref lists no funder. The review cites the readiness definition and adopts the subjective/objective, on-field/off-field tool classification of a paper first-authored by its co-author Beato (its reference 7); none of the 14 included studies is by the review's authors.
Key Results
How many studies used which kind of tool — the abstract, the body text and the review's own Table 1 (our count):
| Count | Abstract | Body text | Table 1 (our count) |
|---|---|---|---|
| Studies included | 14 | 14 | 14 rows |
| Used self-report tools | 10 | 11 | 11 |
| Used objective tools | 9 | 10 | 10 |
| Combined both | 7 | 5 said to show a gain from combining | 7 |
| Objective tools on field | not given | 4 (3 running, 1 triathlon) | 5 (Nuuttila 2022a also marked on field) |
The body's counts add up (11 + 10 − 7 = 14); the abstract's do not (10 + 9 − 7 = 12) (our arithmetic).
What each study measured and found, as the review's Table 1 summarises it (no effect sizes are given; our transcription of the rotated table):
| Study | Sport, n | Readiness measure | Checked against | Result as reported |
|---|---|---|---|---|
| Alfonso 2025 | cycling, 28 | morning HRV, resting HR, well-being; band = SD of the previous 10 days around their mean | power parameters, between three guided groups | HRV + well-being + resting HR > HRV + well-being > HRV alone; all three arms HRV-guided, no unguided control |
| Bulte 2025 | swimming, 11 | morning HRV, 7-day rolling average (text: 3-day) | training phase | rMSSD/AVNN (RMSSD normalised to mean NN interval) higher in overload than in training; text: higher performers' HRV rose 5-8 days before competition, then fell |
| De Martin Topranin 2023 | five sports, 41 women | resting HR, perceived sleep, physical and mental readiness to train | menstrual-cycle phase | resting HR higher midluteal vs early follicular; sleep quality lower midluteal vs late follicular; physical readiness lower at ovulation and midluteal |
| Figueiredo 2022 | running, 36 | morning HRV (7-day rolling), DALDA questionnaire | peak velocity, 5-km time trial | questionnaire-guided > HRV-guided > traditional training (the body text reports only HRV-guided vs traditional) |
| Grove 2014 | triathlon, swimming, 96 in three studies | Training Distress Scale | time to fatigue; 20-m shuttle run; % of personal best | more distress, worse performance in all three |
| Hill & Chtourou 2022 | cycling, 7 | morning mood (POMS) | time to exhaustion | worse mood, shorter time to exhaustion |
| Karu 2000 | running, 15 | perceived readiness between repeated runs | heart-rate and lactate recovery | readiness ratings tracked heart-rate and lactate recovery |
| Nurmekivi 2001 | running, 15 | perceived readiness between interval runs | heart-rate and lactate recovery | the same as Karu 2000 |
| Nuuttila 2022a | running, 30 | nocturnal HR and HRV, readiness to train, soreness, fatigue, sleep, stress | 3-km time trial, jump, blood markers over a 2-week volume or interval block | readiness to train fell and soreness rose; HRV fell from before to week 1 in the interval group compared with the volume group; nocturnal HR fell |
| Nuuttila 2022b | running, 30 | HRV (7-day rolling), fatigue and soreness (score ≤ 5), HR–running speed index (> 2-week average − 0.50) | maximal velocity, 10-km time trial | individualized group improved more than the predefined group |
| Nuuttila 2024 | running, 24 | nocturnal and resting HR, HRV, leg soreness, fatigue, HR–running power index, jump, submaximal run, weekly 6 × 3-min intervals | 3-km time trial; overreached vs responders | nocturnal HR, subjective readiness and the HR–running power index ≥85% valid in telling overreached from responders |
| Schaffarczyk 2022 | triathlon, 11 | DFA α1 in the warm-up, jump | before vs after a light or heavy session | DFA α1 lower after the heavy run; jump up after moderate and heavy sessions (text: up after moderate, unchanged after heavy) |
| ten Haaf 2017 | cycling, 29 | POMS, a 7-item diary and resting HR (evening and morning) | peak power; functionally overreached vs acutely fatigued | 2 of 7 items (fatigue, readiness to train) after 3 days classified 78% correctly; text: resting HR alone did not separate the two |
| Vesterinen 2014 | running, 45 | weekly average of the HR–running speed index | changes in peak speed, speed at respiratory compensation, VO2max | the index rose with training and went with the gains |
Definition: no clear, shared definition in any study; readiness was used interchangeably with fatigue or recovery, often assumed (in tapering and load management) rather than defined or measured; three studies (De Martin Topranin, Nuuttila 2022a, ten Haaf) treated it as a practical standalone measure.
Validation: readiness was checked against indirect indicators of recovery or adaptation more often than against later exercise performance; the authors conclude that the evidence does not support the superiority of any single readiness marker for predicting later endurance performance or training tolerance, and that only a limited number of studies examined prediction at all.
Self-report: perceptual measures moved with load (readiness down and soreness up during a 2-week intensified block), with overload and with menstrual-cycle phase; training distress went with lower time-to-fatigue and time-trial results. The authors judge them sensitive to fatigue and recovery but of unclear specificity as readiness indicators, with contrasting results on telling training groups or phases apart.
Objective markers: HRV, measured at rest in the morning or over about 4 h of night recording and usually averaged over 3- to 7-day rolling windows, served mainly as a proxy of recovery and autonomic balance; daily resting HR alone did not separate functionally overreached from acutely fatigued cyclists; the on-field indices (HR–running speed, HR–running power, interval sessions, jump) were used as markers of training adaptation and their reading as readiness is, in the authors' words, "largely indirect".
Quality (Table 2): blinding high-risk in all 14; incomplete outcome data high-risk in one (Alfonso), low in 13; unbiased data, selective reporting and other bias low in all; randomised order rated in 5 studies (4 low, 1 unclear), carry-over in 4 (all unclear), allocation concealment in 3 (all unclear), all otherwise not applicable; the crossover item not applicable to any.
Framework (authors' proposal): readiness as a transient psychophysiological state, the momentary training capacity emerging from physiological, psychological, perceptual and contextual factors; presented as a conceptual synthesis, not an empirically validated model.
Proposed validation (authors' proposal): first check that morning readiness (subjective and objective) agrees with on-field measures, then test whether readiness before a session predicts its quality — holding target power or speed in intervals, lower heart-rate or RPE drift and decoupling in long sessions, smaller velocity or power loss, time in target zones.
Where the abstract and the body disagree: self-report studies 10 vs 11 and objective studies 9 vs 10 (Table 1 sides with the body); the abstract's "seven studies combined subjective and objective measures" matches Table 1, while the body says five studies showed that combining "enhanced training prescription ... compared with objective data alone" — by Table 1 only Alfonso compared a combination against HRV alone, Figueiredo had no combined arm (its questionnaire-only group ranked first), Nuuttila 2022a and 2024 compared no guidance methods, and Nuuttila 2022b compared a combined rule with a predefined plan, not with objective data alone.
Where the body disagrees with itself or its tables: the Discussion says HRV-guided interventions "consistently yield superior adaptations across endurance disciplines", while the Results say markers showed positive and negative associations and a few sentences later the same Discussion paragraph says the evidence does not support the superiority of any single marker; the guided trials behind it are three, in cycling and running, one without an unguided control and one where questionnaire guidance ranked above HRV.The text says HRV-guided cyclists improved "compared with control groups"; Table 1 and Table 3 show three HRV-guided arms (and Table 1 misprints the top arm as "GHR+WB+RHR" for GHRV+WB+RHR). Bulte's HRV is a 3-day rolling average in the text and a 7-day one in Table 1. Schaffarczyk's jump rose after moderate and heavy sessions in Table 1, after moderate only in the text.The text counts four on-field studies; Table 1 marks five with objective on-field tools, labels Schaffarczyk's laboratory warm-up "On field (laboratory)", and the jump test is filed as on-field in the abstract and Results but as off-field in the Discussion. The text describes Nuuttila 2022a as "several weeks" of training and also as a 2-week block.
Numbers that do not add up or break the rules: the selection text has 64 abstracts screened, 35 excluded and 29 full texts read, of which 9 were included; Figure 1 labels all 64 as full articles assessed and 55 as full articles excluded, which is the text's 35 abstract exclusions plus 20 full-text exclusions (our arithmetic) — the numbers agree, the labels do not: by the text only 29 full texts were read.Both reach 9 + 5 = 14; the sentence on snowballing ("5 more studies ... together with one of the 5 retrieved from citation tracing") does not parse. Grove's third study (63 swimmers, mean age 14.4) is outside the 18-65 eligibility; Hill & Chtourou's seven kinesiology students are Tier 1, not endurance athletes.Karu 2000 and Nurmekivi 2001 share authors and an identical sample (15 college middle-distance runners, 19.6 ± 2.6 y), so the 418 may count one group twice (inference). The 418 itself adds up from Table 3 (our arithmetic).
Mechanisms
Measured (in the included studies, as the review reports them): subjective ratings fell and soreness rose with intensified training; HRV fell after heavy sessions and within an interval block and rose with adaptation in responders; nocturnal heart rate fell with adaptation; heart rate at a given running speed or power fell as fitness rose; resting heart rate rose in the midluteal phase.None of these was tested as a forecast of the next session in more than a few studies.
Authors' interpretation: readiness is a separate construct from fatigue and recovery — the momentary capacity to tolerate and perform a coming task — so tools that track the response to past load (wellness, HRV, RPE) do not by themselves capture it; integrating subjective, physiological and neuromuscular measures with context (stress, travel, the menstrual cycle) is the most promising route.For prolonged aerobic work they suggest HRV and heart-rate indices with sleep quality, perceived fatigue and stress; for intermittent work, HRV and jump tests with soreness and life demands.
Inference: the tools are validated on responsiveness (they move after load), which is the property a monitoring signal needs; a forecast needs validation against the next session, which almost none had. A check-in can therefore say "you are carrying more than usual" with some support and "you will have a bad session" with almost none.
Inference: the search required readiness-type words in the title, abstract or keywords, so the evidence base is defined by vocabulary: HRV-guided trials in this base (REF-362, REF-438), the meta-analysis of HRV-guided training (REF-994) and the submaximal-test timing study (REF-753) are not among the 14, although they are the kind of evidence the review already holds three of (training guided by a marker, judged by adaptation over weeks, which it calls indirect), not tests of whether a morning reading predicts the session that follows; and a meta-analysis could not have been one of the 14 primary studies in any case.
Practical Application
Can a morning check-in tell whether I am ready to train? Not reliably yet: 14 small studies used questionnaires, HRV or resting HR, few tested them against the next session, and none beat the rest. A prompt, not a verdict.
Read the check-in as how the athlete is carrying recent load, not as a forecast of today's session. The tools in the review track fatigue, recovery and adaptation; whether a low morning predicts a worse session was rarely tested (B+, systematic review of small studies). The base's own practice on what to do with a low morning is SYNTH-30.
The athlete's own report stays first. Self-report was the most used tool (11 of 14 studies) and moved with load; in one study fatigue and readiness-to-train ratings after 3 days of hard cycling correctly classified 78% of riders as functionally overreached or acutely fatigued while resting heart rate alone did not separate them, and in one trial in 36 recreational runners the review's table ranks questionnaire-guided training above HRV-guided and fixed training (the review's summaries; no effect sizes given).This agrees with REF-368 (A); the review itself calls the specificity of self-report for readiness unclear.
Read devices as a trend, not a morning. The HRV studies averaged over rolling windows of 3 to 7 days (one read a band of the previous 10 days' mean ± SD), which the authors say "appears crucial" for separating a trend from day-to-day noise; this matches REF-361 (B+).Resting heart rate on its own did not separate functional overreaching from acute fatigue in the one study that tested it (ten Haaf), and the authors judge isolated resting heart rate short on prescription accuracy.
Combining measures is the authors' recommendation, tested head to head in one included study only (Alfonso, all arms HRV-guided: HRV + well-being + resting HR ranked above HRV + well-being, above HRV alone, in 28 cyclists); the meta-analysis of HRV-guided training found no significant performance gain (REF-994, A-).
What would show a readiness rule works (the authors' proposal; the base's own practice can use it): compare the morning reading with the quality of the session that follows — target power or speed held in intervals, heart-rate or RPE drift, time in the target zone — rather than with how the athlete felt or how much load came before.
Context counts but is untested: psychological stress, travel and the menstrual cycle are named as factors that change how the markers read; in the one cycle study resting heart rate was higher in the midluteal phase. How much each should move a decision is not known.Inference: a rise in resting heart rate in the luteal phase may be the cycle rather than fatigue — but in the same study perceived physical readiness to train also fell at ovulation and in the midluteal phase, so the athlete's own report moves with the cycle too; read both against the athlete's own earlier cycles.
Limitations
Small, mixed studies, no pooling. 14 studies of 7 to 96 people, only three of which guided training by readiness markers and compared groups; the rest are observational, questionnaire validations, single sessions or within-session ratings, some from 2000-2001. No effect sizes are extracted, so the review can describe, not weigh.
A vocabulary-defined search. Only papers using readiness-type words were found, so the HRV-guided trials and the submaximal-test timing study already in this base (REF-362, REF-438, REF-753) are absent; the meta-analysis REF-994 is not cited either (as a review it could not have been one of the 14). The conclusions are about studies that called their measure readiness, not about all daily monitoring.
The quality appraisal does not fit the designs. The tool is cited to a crossover-trial checklist but called the revised Cochrane tool; its crossover item applied to none of the 14 and most randomisation items were not applicable; blinding was high-risk in all 14. The text's "mainly low or unclear" for randomised order, carry-over and allocation concealment describes the few studies rated, not most.
Internal inconsistencies. The abstract's counts of studies by tool (10, 9) differ from the body and Table 1 (11, 10) and do not add up to 14; the body's claim that five studies showed combining measures "enhanced training prescription ... compared with objective data alone" is not supported by its own Table 1 for four of the five (only Alfonso had an HRV-alone arm); Figure 1 labels 64 records as full texts assessed where the text says 29 were read; Bulte's averaging window and Schaffarczyk's jump result differ between text and table; the Discussion's claim that HRV-guided interventions "consistently yield superior adaptations across endurance disciplines" rests on three guided trials in two sports (one with no unguided control, one where questionnaire guidance ranked above HRV), sits uneasily with its own finding that no marker is superior, and conflicts with REF-994.
Eligibility not kept. Grove's 63 swimmers had a mean age of 14.4 against an 18-65 criterion; seven kinesiology students (Tier 1) are counted as endurance athletes; Karu 2000 and Nurmekivi 2001 may be one sample counted twice (inference).
Who. Mostly recreational runners and cyclists (Tier 2); 157 of 418 participants are women, 41 of them in one study; no masters-specific or elite-only analysis.
The framework and the validation model are proposals by the authors, not tested; the practical recommendations (which tools for which kind of exercise) are opinion built on the review.
Funding and interests: none stated in the PDF; the review builds on a definition and a tool classification by its co-author Beato; no included study is by the review's authors.
How this entry was checked and changed
Why this level
read from the full text (publisher PDF). The rubric's A- line is an SR/MA "with serious heterogeneity or mostly small trials"; graded one step lower because only three of the 14 studies guided training by readiness markers and compared groups (one of them with no unguided control, per the review's Table 1), the rest being a randomised comparison of two training blocks not guided by readiness (Nuuttila 2022a) and observational, questionnaire-validation, single-session or within-session studies of 7 to 96 people; there is no pooled effect, confidence interval or certainty rating; the quality tool, called the revised Cochrane Risk of Bias tool, is cited to a crossover-trial checklist (Ding 2015) whose crossover item was not applicable to any of the 14, and blinding was high-risk in all 14; the abstract's counts of studies by tool do not match the body or Table 1, and one of an included study's three samples (Grove 2014: 63 swimmers, mean age 14.4) falls outside the stated 18-65 eligibility. Its main conclusions (no shared definition, no validated tool) are descriptive and hold on its own tables; its framework is explicitly the authors' conceptual synthesis
Full text (checked 2026-10-27)
A total of 4086 articles were retrieved; after removing 886 duplicates, 3200 records remained; nine articles met the inclusion criteria and 5 more came from snowballing, for 14 studies and 418 participants (full text)
Eligible: endurance athletes aged 18–65 years; protocol registered on the Open Science Framework on July 18, 2025; search on July 26, 2025 (full text)
Subjective tools were used in 11 studies and 10 studies examined objective tools; the abstract says ten and nine (full text)
Four studies examined on-field tools, 3 in running and 1 in triathlon (full text)
The body credits 5 studies with showing that adding subjective well-being tools to objective ones improved training prescription and explained performance adaptations better than objective data alone (full text)
HRV came from resting morning measurements or nocturnal recordings of around 4 hours, usually averaged over 3- to 7-day rolling windows (full text)
Readiness was checked against indirect indicators of recovery or adaptation more often than against subsequent performance, and no single readiness marker was shown superior for predicting later performance or training tolerance (full text)
ten Haaf, in Table 1: 2/7 items (fatigue, readiness to train) after 3 days of cycling classified 78% correctly as functionally overreached or acutely fatigued (the text says "the majority"); in the text, daily resting HR alone could not differentiate the two (full text)
Nuuttila 2024, in Table 1: nocturnal HR, subjective readiness and the HR–running power index showed ≥85% validity in discriminating overreached from responder states (full text)
In swimmers the text gives a 3-day rolling average of RMSSD and higher performers' HRV rising 5 to 8 days before competition; Table 1 gives a 7-d rolling average for the same study (full text)
Grove study (c), in Table 3: 63 high-level age-group swimmers, 38 of them female, mean age 14.4 (full text)
Table 2: every included study was rated high risk of bias for blinding (full text)
The PDF carries no funding, conflict-of-interest or acknowledgements statement (full text)