Counts describe this library, not the strength of evidence.
How to read this evidence
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.
Pooling 11 randomised trials with 549 participants, sleep deprivation was associated with lower rMSSD (p < 0.05) and higher LF and LF/HF (p < 0.05), while the changes in SDNN and HF were not statistically significant (p > 0.05). Trials differed in how HRV was measured and how long sleep was withheld, and the abstract gives no effect sizes.
In 931 adults at a US medical centre, pulse rate variability from a light-based (PPG) sensor disagreed significantly with ECG-based heart rate variability (p < 0.001 for all) and underestimated SDNN, rMSSD and pNN50 in several chronic disease groups. The sample was patients wearing one research device, so other wearables and healthy users may differ.
A review of 41 studies (from 239 identified) found consistent evidence linking reduced heart rate variability with the severity of obstructive sleep apnea (OSA) and COPD, and more pronounced changes in overlap syndrome. The authors' view that HRV is a valuable clinical tool goes beyond what a narrative literature review can prove.
In 352 women with depression tracking via the Juli app (9393 entries), mood declined gradually from 14 days before menstruation and was lowest from 3 days before to 2 days after (54.3% scored lower in that window, 95% CI 48.9% to 59.6%). Energy did not vary with cycle day, and the app-based self-ratings come from a self-selected group.
In 278 participants, those with metabolic syndrome had lower sample entropy and DFA values at rest than people without metabolic alterations, and sample entropy was already lower with one or two alterations. The abstract gives no accuracy or prediction statistics, so this does not show HRV can detect metabolic syndrome.
A scoping review found only 4 interventional studies of heart rate variability biofeedback in functional gastrointestinal disorders; 2 reported significant symptom improvement and 2 found no significant difference. The studies were mixed in patients and conditions, so the suggestion that at least 6 weeks is needed is tentative.
In 23 highly trained rowers, after periods of high versus low training load, stool frequency, short-chain fatty acids (propionic p = 0.007, butyric p = 0.003), the firmicutes to Bacteroidota ratio (p = 0.04) and diet quality (p = 0.014) all differed. The study had a small sample, and the abstract does not show that training itself caused these changes.
Journal of the International Society of Sports Nutrition
Over 24 hours of ECG recording, 52 healthy young volunteers living at about 5300 m had lower heart rate variability, higher mean heart rate (84.67 vs. 73.2 beats/min) and more sleep apnea events (Apnea-Hypopnea Index 20.14 vs. 11.36) than 56 volunteers living at about 500 m. The groups lived in different places, so altitude cannot be separated from other differences, and the Chinese cohort is small.
In a network meta-analysis of 55 randomised trials, cold water immersion for 10-15 min at 11-15°C reduced muscle soreness most (SMD -1.45, 95% CI -2.13 to -0.77), while 10-15 min at 5-10°C ranked best for jump performance and creatine kinase. The doses were compared indirectly across trials of mixed quality, so the ranking is a guide, not a proven best protocol.
In a randomized trial of 60 overweight and obese children, 3 months of omega-3 plus standard advice raised heart rate variability (RMSSD P = 0.017, SDNN P = 0.009, pNN50 P = 0.043) compared with standard advice alone. With only 30 children per group and a short follow-up, this is early evidence and says nothing about heart disease outcomes.
In a meta-analysis of 11 trials with 3177 healthy adults, cold-water immersion raised inflammation markers immediately (SMD 1.03) and lowered stress 12 hours later (SMD -1.00), but stress did not change at other time points. Only 11 mostly small trials with very different protocols (7°C to 15°C, 30 seconds to 2 hours) contributed, so the findings are tentative.
Among 355 older patients at one Chinese hospital, those with anxiety, depression and sleep disorders together had the lowest heart rate variability on most indices (P < 0.01), and each condition was independently linked to specific HRV measures. The study was cross-sectional and single-site, so it shows association, not that HRV can diagnose these conditions.
In 43 healthy adults, resting-state connectivity between the insula and the default mode network showed no correlation with heart rate variability (HRV), but left insula volume was positively associated with HRV. The abstract gives no correlation values, and a sample of 43 healthy adults at rest limits what can be concluded.
Across 50 studies with 1160 subjects, thresholds derived from heart rate variability (HRVT1 and HRVT2) showed trivial differences from reference lactate and ventilatory thresholds (SMD 0.08 and -0.06) and very strong correlations (r = 0.85 for both). The pooled agreement is for group averages from lab ramp tests, not a guarantee for an individual's own threshold.
In 34 healthy active adults randomized to cold water immersion (CWI, n = 11), percussive massage (PM, n = 11) or passive rest (n = 12), neither treatment improved performance recovery over 72 hours, and jump height was lower after CWI at one time point. The trial is very small and the abstract reports no effect sizes, so modest benefits could be missed.
This narrative review groups what lowers heart rate variability (HRV) into four sets: fixed physiological traits such as age, diseases such as heart failure, lifestyle factors such as alcohol and overweight, and external factors such as heat, noise and shift work. It is not a systematic or quantitative synthesis, so it shows which factors matter but not how much each one changes HRV.
Pooling 17 studies, women with polycystic ovary syndrome (PCOS) had lower heart rate variability (HRV) than controls, for example SDNN with an SMD of -0.763 (95% CI -1.289 to -0.237). The studies varied in quality from low to high risk of bias, and HRV was not reduced in women with obesity.
In 47 people followed for 4 weeks, several smartwatch-derived heart rate variability (HRV) measures, including RMSSD and SDNN, were significantly associated with depression scores, and SDNN, SDSD and LF with anxiety scores, with all HRV metrics correlating negatively with symptoms. Few correlations reached significance, and the sample was small and not made up of clinically diagnosed patients.