Counts describe this library, not the strength of evidence.
Comprendre ce type de preuve
Recovery is not one number. Cold-water studies separate soreness, power and strength; newer analyses make uncertainty more visible. HRV depends on breathing and measurement context, and a workload ratio is not an injury prediction.
Of 127 health care workers asked to do 5-minute daily heart rate variability (HRV) biofeedback by app, only 21 (16.5%) were at least 50% compliant, so the authors judged this design not feasible. Psychological scores improved numerically but not significantly (all P>.05), and the HRV signal came only from the compliant subgroup.
A meta-analysis of exercise trials in adults with coronary artery disease, heart failure or peripheral arterial disease found significant improvements in some heart rate variability measures, including SDNN for short-term recordings (p = 0.01), with the clearest gains in heart failure. The abstract gives only p values and does not report the number of trials or the size of the changes.
In 8 studies covering 138 949 pregnancies, greater visit-to-visit blood pressure variability was associated with more gestational hypertension (odds ratios 1.40-2.15), severe hypertension (1.40-2.20) and small-for-gestational-age infants (1.12-1.32), independent of average blood pressure. No study reported heart rate variability, so that half of the question is unanswered.
Pooling 6 studies with 242 participants, people with a history of concussion had significantly different resting MeanNN (p = 0.03) and RMSSD (p = 0.04) than those without, but not SDNN (p = 0.11). The evidence is thin: few small studies with differing populations and methods, and 4 of the 6 found no difference at all.
In a network meta-analysis of randomized trials in adults, high-intensity interval training (HIIT) ranked highest for SDNN, RMSSD and the LF/HF ratio (SUCRA 98.7%, 84.9% and 99.75%), while resistance training ranked first for HF power and combined training for LF power. Rankings show which option is most likely best, not how large the gain is, and the abstract gives no pooled effect sizes or trial counts.
In a systematic review of 12 longitudinal studies with 24,390 participants, all found a link between HRV and later cognition, with higher parasympathetic activity consistently tied to better cognition, particularly executive function. The results were summarized narratively without pooled estimates, and the abstract gives no effect sizes.
Pooling 43 case-control studies (2,359 people with depression, 3,547 healthy controls), depression was linked to lower resting HRV on most measures, for example SDNN with Hedges' g -0.87 (95% CI -1.14 to -0.60). Case-control comparisons cannot show whether depression lowers HRV or the reverse, and the abstract does not report how much the studies differed.
In three groups of 26 people (basketball players, swimmers, sedentary adults), both athlete groups finished the Trail Making Test and Letter Cancellation test faster than sedentary people, while basketball and swimming did not differ from each other in cognition or HRV. With small groups compared at one time point, it cannot show that sport improves cognition.
In a meta-analysis of 19 studies, people with metabolic syndrome had lower SDNN (-0.36; CI -0.44 to -0.28), rMSSD (-7.59), HF and LF than healthy controls. Studies were short recordings of 30 minutes or less, and sex-specific results differed, so effects in each sex remain unclear.
All 7 studies in this review found that cognitive performance declined when endurance athletes became overreached or overtrained, with slower reaction time on the Stroop colour test. Only seven studies were found among 221 screened, and the abstract gives no sample sizes or effect sizes.
In participants of the Study of Health in Pomerania, heart rate variability (HRV) from a 10 s ECG correlated with HRV from 5 min recordings (r = .52-.75), and it was the strongest predictor of the longer measure after adjusting for covariates. The comparison was against 5 min rest recordings in subsets, not against health outcomes, so it supports use as a proxy rather than proving risk prediction.
In 42 undergraduate students, those with higher resting heart rate variability (rMSSD) answered faster and more accurately on an n-back working memory test, with lower prefrontal oxy-Hb on brain imaging. The study is small, compares two groups split at the median, and cannot show that HRV causes better memory.
In 20 under-14 elite soccer players followed for a 26-week season, training monotony was higher in the second half, while chronic load, acute load and muscle soreness were higher in the first half. With only 20 players and a few scattered correlations, the study cannot show that fatigue, soreness or growth stage drive training load.
Across 26 publications covering 1537 people with depression and 1041 healthy controls, nonlinear heart rate variability measures separated the two groups with an overall effect size of 1.03 (95% CI 0.703-1.35). The authors argue these measures beat standard ones, but that comparison was made against effect sizes found in the literature, not within the same datasets.
In 47 people with long COVID and 42 healthy controls wearing a 24-hour ECG, controls had higher SDNN24, SDANNI and high-frequency HRV, while patients had higher low-frequency HRV (OR 1.002, p = 0.030). It was a single-center cross-sectional study, so it cannot show whether long COVID caused the autonomic differences.
Over 24 hours in 28 people, a Samsung Gear Sport smartwatch gave acceptable heart rate and most heart rate variability measures during sleep compared with a chest ECG monitor. While awake only average heart rate and AVNN (average beat-to-beat interval) were reasonably accurate, and the other HRV measures had high errors.
In 18 healthy, active women wearing a WHOOP strap before, during and after pregnancy, daily HRV drifted down and resting heart rate drifted up through pregnancy, then the trend reversed about 49 days before birth. With only 18 women and no comparison group, the pattern shows what happened in this small group, not what to expect individually.
In 28 healthy young adults, 10 minutes of transcutaneous auricular vagus nerve stimulation (taVNS) at the cymba conchae raised vagally mediated HRV in time and frequency measures compared with sham stimulation at the helix and with rest. The abstract reports no effect sizes or P values, and the effect was only measured during short stimulation in healthy people.