Table 3

Table of studies in which HRV measures were calculated on a least one cohort with BD diagnoses. Summary of ECG recordings, HRV measures and results are provided, in addition to their interpretation and adherence to the GRAPH checklist.

Study/yearCohortsDataParametersResultsInterpretationGood Practice Checklist
Bassett et al21 201629 BD, 41 MDD, 38 HCECG during sleep, minimum 4 hoursRMSSD, pNN50, SDNN, LF, HF, LF/HF ratio, SD1, SD2Reduced HRV in BD (RMSSD) and depression (RMSSD, SDNN, SD1 and SD2)Impaired autonomic function in BD and depression during sleep through a reduction in parasympathetic activityNo ECG sampling rate. No information on R–R interval extraction.
Chang et al,22 201461 M-BD, 183 HC5 min ECG at restLog transforms of SDNN, VLF, LF, HF power and LF/HF ratioSignificantly reduced mRR, SDNN, LF, HF and LF/HF ratio in BD. Correlations between LF/HF ratio and HF and mania rating scale.ANS dysregulation is associated with mania in BD through alterations in parasympathetic activityNo information on ECG used, no information on R–R interval cleaning and artefacts.
Chang et al,20 2015116 D-BD, 421 UD, 591 HC5 min ECG at restLog transforms of SDNN, VLF, LF, HF power and LF/HF ratioSignificantly lower SDNN, LF and HF power and higher LF/HF ratio in BD compared with UD. Increased LF power and decreased HF power in BD compared with controlsSympathetic excitation and parasympathetic impairment in BD compared with controls, with HRV a possible tool to distinguish between UD and BD.No information on ECG used, no information on R–R interval cleaning and artefacts.
Cohen et al,7 200339 E-BD, 39 HCECG at rest. No length of time givenmRR, SDNN, SDANN, RMSSD, pNN50, VLF, LF, HF power and LF/HF ratioBD significantly lower RMSSD, total power, nHF and LF/HF ratioIncrease in parasympathetic activity and decrease in sympathetic activity in BD.No information to rule out psychiatric illness in controls. No information given on the length of the ECG recording. Little information on physiological meaning of HRV parameters until results.
Faurholt-Jepsen et al,24 201716 BDMobile ECG up to 11 days during different affective statesDifference between second longest and second shortest R–R interval every 30 sIncreased variability during manic state compared with euthymia and depressionAutonomic nervous system dysfunction in BDNo ECG sampling rate. No information on R-peak extraction or cleaning
Gruber et al,16 201123 E-BD, 24 HCECG recorded during stimuliHF powerIncreased HF power in BD after stimuliIncreased vagal tone in BD, which is a marker for positive emotionLittle information on recruitment and demographics. Little information on time periods HRV was calculated. Minimal information of R–R interval extraction and cleaning.
Henry et al,12 201023 M-BD, 14 SZ, 23 HC5 min ECG at restmRR, SDNN, RMSSD, pNN50, LF, HF power and LF/HF ratioReduced SDNN in BD, but not significance. Significant increase in LF/HF ratio and decrease in nHF, RMSSD and pNN50 compared with the controlsDecrease in HRV and parasympathetic activity in BDAll items on checklist met
Howells et al,29 201312 BD, 9 HCOne hour ECG at rest. Before and after cognitive behavioural therapyLog of LF and HF and peaks in the LF and HF bandsBD increased HF peaks compared with HC. HF peak reduced in BD after therapyTherapy improved emotional processing in BD as HF peak decreasedAll items on checklist met
Lanata et al,26 201510 BDContinuous ECGSample entropySample entropy is able to estimate long term changes in mental state of BD patientsSample entropy values of HRV may be able to aid clinicians with diagnosis and management of BDLittle information given on timing of ECG recordings and how R–R intervals are extracted and cleaned. No information on clinical interpretation of sample entropy measures.
Lee et al,19 201233 SS-BD, 59 HC5 min ECG at restmRR, SDNN, RMSSD, pNN50, VLF, LF, HF power and log total powerSignificantly lower SDNN, pNN50, log TP and VLF in BD. Also correlations between symptom severity and HRV parameters were found.HRV is reduced in BD and may be an effective marker for the disorderNo mention of ECG sampling rate or method of R–R interval extraction or cleaning. Little physiological interpretation of HRV parameters.
Levy et al,18 201233 E-BD, 22 HC5 min recording from electrogramHeart rateSignificantly increased HR parameters in BDBD patients experience larger changes to ANS on cognitive testingNo information on extraction and cleaning of R–R intervals
Migliorini et al,13 20111 BD, 8 HCECG during four nightsmRR, SDNN, RMSSD, VLF, LF, HF, LF/HF ratio, sample entropy, Lempel-Ziv complexity, detrended fluctuation analysisDecreased mRR, RMSSD and SDNN in BD. Lempel-Ziv complexity and sample entropy correlated with level of depressionThere is dysregulation of the ANS in BD, and HRV is a promising method for measuring mood changes.No demographic information. All other items on checklist met
Moon et al,23 201341 BD, 35 SZ, 34 PTSD, 34 UD, 27 HC5 min ECG at restSDNN, RMSSD, VLF, LF, HF, TP, LF/HF ratio and approximate entropySDNN, RMSSD, TP, LF and HF all significantly reduced in BD compared with controlsNot possible to use HRV to discriminate between mental health disorders but possible to discriminate from healthy. BD showed most significant HRV changes.Little participant demographic information. No information on ECG sampling rate or methods to remove artefacts in the R–R interval series
Quintana et al,9 201533 BD, 47 SZ, 212 HC5 min of pulse oximetryHF powerSignificant reduction in HF power in BD, independent of age, BMI and medicationParasympathetic activity is altered in BD, with HRV a possible marker of cardiovascular risk in BD.Pulse oximeter not ECG, but all other items met on checklist
Tanaka et al,30 201325 BD, 22 HCNo information on ECGLF, HF and LF/HF ratioNo difference between BD and HC after stimulation to wristNo difference in HRV between groups, biological background did not influence hormonal reaction observed.No details of collection method of ECG. No information on R–R interval extraction and cleaning
Todder et al,15 200539 E-BD, 39 HCECG at rest. No length of time givenNon-linear parametersNo significant differences between any parametersThere is no disturbance in the ANSNo information to rule out psychiatric illness in controls. No information given on the length of the ECG recording. Little information on physiological meaning of HRV parameters until results.
Valenza et al,25 20158 BDECG recorded approximately 10 min during tasksPoint-process-based non-linear autoregressive integrative modelAround 90% accuracy of predicting depression or euthymia in BD.A link between ANS function and BD exists, with parameters measuring ANS able to predict mood or emotion of patientsLittle information on method of ECG recording and extraction and cleaning of R–R intervals
Voggt et al,17 201590 E-BD, 62 HC30 min ECG recordingSDNN, LF, HF and LF/HF ratioSignificantly lower SDNN, LF and HF in BDSDNN may be used to study interventions to reduce cardiovascular disease in BDMost items met, no mention of ECG sampling rate or causes of artefacts
  • ANS, autonomic nervous system; BD, bipolar disorder; BMI, body mass index; D-BD, depressed bipolar disorder; E-BD, euthymic bipolar disorder; GRAPH, Guidelines for Reporting Articles on Psychiatry and Heart rate variability; HC, healthy control; HF, high frequency power; HRV, heart rate variability; LF, low frequency power; LF/HF ratio, low frequency to high frequency ratio; M-BD, manic bipolar disorder; MDD, major depresive disorder; mRR, mean of R–R interval; pNN50, perentage of R-R intervals over 50ms; PTSD, post-traumatic stress disorder; RMSSD, root mean square of successive differences; SDANN, SD of average R–R intervals; SDNN, SD of R–R intervals; SD1, standard deviation in Poincare plot y=-x direction; SD2, standard deviation in Poincare plot y=x direction; SS-BD, subsyndromal depression bipolar disorder; SZ, schizophrenia; TP, total power; UD, unipolar depression; VLF, very low frequency.