Changes in heart rate variability during TOVA testing in patients with major depressive disorder


Correspondence: Shao-Tsu Chen, MS, MD, Department of Psychiatry, Buddhist Tzu-Chi General Hospital, 707, Sec 3, Chung-Yang Road, Hualien 970, Taiwan. Email:



The aim of this study was to identify major depressive disorder (MDD) based on heart rate variability (HRV) during tests of variables of attention (TOVA).


Forty-five MDD patients without cardiovascular disease and 45 controls matched by age and gender participated in this study.


Compared to the controls, the MDD group had lower resting HRV parameters, more omissions and variability and longer response times on TOVA, and failure of attention employment to decrease HRV.


The resting HRV parameters may provide easily measured, clinically useful ways to identify patients with MDD and to monitor their progress in treatment.

DEPRESSION IS A worldwide public health concern because of its high prevalence.[1] Major depressive disorder (MDD) is a mental illness with significant psychological, physiological, and social impact. It is well known that abnormal heart rate variability (HRV) and impaired concentration occur in patients with MDD,[2] but these parameters are not currently considered to be diagnostic markers.

Thayer et al. found that cognitive performance and prefrontal neural functions were linked to HRV.[3] It is well known that MDD significantly influences cardiac function. Researchers have suggested that negative emotion is associated with suppression of affective responses in some areas of the brain, especially the right hemisphere, which regulates parasympathetic and sympathetic activity.[4, 5] There is also evidence to show that MDD patients have higher cardiovascular mortality and morbidity rates.[6] Previous research has indicated that abnormal HRV is a risk for sudden cardiac death.[7, 8] Recent research has shown that depression and coronary heart disease (CHD) are related. Theories range from neurobiological theories of stress encompassing genetic, neuronal, hormonal, and immunologic responses to sociological theories.[9] One or more still-unrecognized underlying physiological factors may predispose a patient to both depression and cardiovascular disease. Several prospective studies have found an association between depression and the subsequent development of CHD, and depression is an independent risk factor for the progression of heart disease.[6] HRV, then, shows great potential as a biomarker to bridge depression and cardiac status.

Studies of the relationship between MDD and HRV have found a reduction in autonomic nervous system (ANS) activity in MDD patients and linked that to heart disease.[10-12] Sayar et al. studied the relationship between MDD and HRV in time domain features, and found no statistically significant difference between the study and control groups on root mean square successive difference (rMSSD), percent difference between adjacent normal R-R intervals >50 ms (pNN50), standard deviation of the R-R interval (SDNN), SDNN index, or standard deviation of the average normal R-R intervals (SDANN) on analysis of HRV.[12] No significant relationship between level of anxiety and depression and HRV measures were found and the HRV reduction was attributed to antidepressant treatment. A meta-analysis by Kemp et al., however, showed that study participants with MDD but without cardiovascular disease had lower HRV than did healthy controls, and the decrease could not be resolved by antidepressants. In addition, the severity of depression was negatively correlated with HRV.[11]

The test of variables of attention (TOVA™) is an objective neuropsychological test that measures a person's sustained and selective attention.[13, 14] Deficits in attention and executive functions in patients with MDD have been confirmed,[15-17] and were most pronounced during the first 200 ms of a task.[18] The non-verbal properties of TOVA allow cross-culture comparison, and it is a potential tool for the evaluation of attention, but HRV parameters in different stages of mental attention, especially in patients with MDD, have yet to be determined.

The aims of this study were to determine the characteristics of HRV during standard TOVA testing in patients with MDD, and to develop evaluation criteria for diagnosing MDD based on those electrocardiogram (ECG) features. We hypothesized that the attention task would trigger both sympathetic and parasympathetic changes in HRV, which would be more apparent in healthy controls than in subjects with MDD who would be unable to adapt, and that their level of cognitive function and attention deficit might even be quantified.


This investigation was carried out in accordance with the Declaration of Helsinki. The study design was approved by the Institutional Review Board of Tzu Chi Hospital. The nature of the procedure was explained to all participants, and all participants provided signed informed consent forms. We invited both inpatients and outpatients over the age of 18 who were diagnosed with MDD (ICD-9: 296.2, 296.3) in the Department of Psychiatry, Tzu Chi General Hospital, Hualien, Taiwan to participate. We collected data from 45 patients with MDD but without a history of cardiovascular disease or secondary psychiatric diagnoses, as well as from 45 controls matched by gender and age. The control group was recruited via announcements posted in the hospital, and screened for psychiatric disorders with the Mini-International Neuropsychiatric Interview. Both groups completed the Beck Depression Inventory (BDI) and the Beck Anxiety Inventory (BAI), and the MDD group also completed the Hamilton Depression Rating Scale (HAMD-17).

All experiment data were collected in the morning between 09.00 hours and 12.00 hours to eliminate possible diurnal variations. For the experimental procedure, all subjects first filled out the aforementioned scales, physically rested for 2 min, and then were tested on the TOVA program. For all subjects, ECG signals were recorded for 2 min at rest and then for 22 min during the TOVA attention test. All subjects sat in front of a computer while HRV was measured. The entire ECG recording was segmented into 2-min sessions for further short-term HRV processing. Because of the different time durations for resting and the attention test, the HRV parameters for the TOVA test were recorded as the averages of all 11 2-min sessions.

The ECG was sampled at 500 Hz using the Biopac MP35 system, which is an integrated solution for the measurement of physiological signals. The MP35 (Biopac Systems, Goleta, CA, USA) with the Biopac Student Lab PRO software was used for data acquisition. Beat-to-beat measurements of R-R intervals were made during the entire period. We used an IBM-compatible PC equipped with a program for signal processing and HRV analysis. We used a Fourier transformation model to estimate the power spectrum densities for the frequency domain. Time domain measurements were derived from direct measurements of R-R intervals.[19, 20]

Statistical analysis

Demographic data and characteristics including HAMD-17, BDI, BAI, and TOVA parameters are represented by group as n (%) for gender, mean ± SD for age, HAMD-17, BDI, BAI, and median (interquartile range [IQR]: Q1–Q3) for TOVA parameters, respectively. Differences between groups were compared using chi-squared test for gender; and two-sample t-test was used for age and BAI. Mann–Whitney U-test was used to compare BDI, HRV features, and TOVA parameters between groups because the data did not follow a normal distribution. HRV features including frequency and time domains are given as mean ± SD before and after TOVA testing for a given group. Because HRV features were not normally distributed, Wilcoxon signed-rank test was used to compare the difference before and after TOVA testing for each group, and Mann–Whitney U-test was used to compare the difference between groups. In addition, to better quantify the intervention effect, we calculated Cohen's d for the MDD and control groups at rest as well as during TOVA testing using effect size calculators (available at Statistical significance was set at P < 0.05. All statistical analyses were done using SPSS 15.0 (SPSS, Chicago, IL, USA).


Subject demographic characteristics are listed in Table 1. The HAMD-17 averaged 22.2 (SD = 9.9) in the MDD group. The BDI and the BAI were both higher in the MDD group. TOVA parameters indicated that the MDD group had significantly higher omission rates and longer mean response times and more variability than did the control group, but commission rates were not significantly different.

Table 1. Demographics, HAMD-17, BDI, BAI, and TOVA parameters
DemographicsMDD (n = 45)Control (n = 45)P
  1. Two-sample t-test; χ2 test; §Mann–Whitney test. BAI, Beck Anxiety Inventory; BDI, Beck Depression Inventory; HAMD-17, Hamilton depression rating scale; MDD, major depressive disorder; TOVA, tests of variables of attention.
Age (years)41.9 ± 11.742.4 ± 11.60.850
Female (%)57.857.81.000
HAMD-1722.2 ± 9.9
BDI35.9 ± 14.73.7 ± 4.8<0.001§
BAI28.6 ± 12.33.0 ± 5.4<0.001
TOVA parameters, median (IQR)
Omission rate (%)5.25 (0.62–31.79)0 (0–0.31)<0.001§
Commission rate (%)2.47 (0.62–7.12)0.93 (0.47–2.94)0.066§
Mean response time (ms)350.0 (286.0–465.5)282 (250–326)<0.001§
Mean variability (ms)107.79 (89.20–213.06)70.14 (57.69–85.56)<0.001§

Figure 1 compares HRV features including frequency domain and time domain during rest and during the TOVA test by group. At rest, the MDD group had significantly lower total power (TP), very low frequency (VLF; <0.04 Hz), low frequency (LF; 0.04–0.15 Hz), high frequency (HF) in the frequency domain and SDNN, coefficient of variation of R-R intervals (CV) in the time domain, when compared to the control group. During TOVA testing, the MDD group had significantly lower LF in the frequency domain, and SDNN, CV in the time domain, when compared to the control group. A comparison of the differences within group (rest vs TOVA testing) showed that the VLF in the frequency domain was significantly lower during TOVA testing than that at rest in MDD group. In the control group, the TP, VLF, LF in the frequency domain, and SDNN and CV in the time domain significantly decreased from rest to TOVA testing. There were no significant differences in % change of HRV from TOVA testing to rest between the MDD and control groups (data not shown).

Figure 1.

Comparison of heart rate variability (HRV) features at rest and during tests of variables of attention (TOVA) by group. Data are presented as mean ± SD (image) at rest and (image) during TOVA test in the frequency domain, including (a) total power (TP), (b) very low frequency (VLF), (c) low frequency (LF), (d) high frequency (HF), (e) LF/HF and in the time domain, including (f) mean of R-R interval, (g) standard deviation of the R-R interval (SDNN), (h) coefficient of variation of the R-R interval (CV), (i) no. consecutive R-R intervals differing by >50 ms (NN50), (j) percentage of NN50 intervals (pNN50), (k) standard deviation of standard deviation (SDSD), and (l) root mean square successive difference (rMSSD) for the major depressive disorder (MDD) and control groups. *P < 0.05, Wilcoxon signed-rank test for comparing differences within groups; †‡P < 0.05, Mann–Whitney U-test for comparing differences between groups, during rest and during TOVA testing.

To better quantify the intervention effect, we compared the magnitude of the outcome differences between groups by effect size. The overall effect sizes in the frequency domain between the MDD and control groups at rest and during TOVA testing were (−0.56 to −0.24) and (−0.41 to −0.08), respectively. The overall effect sizes in the time domain between the MDD and control groups at rest and during TOVA testing were (−0.44 to −0.26) and (−0.39 to −0.18), respectively (Table 2).

Table 2. Effect size for multiple HRV features
HRV featuresEffect size (Cohen's d)
MDD vs control (at rest)MDD vs control (at TOVA)
  1. CV, coefficient of variation of R-R interval; HF, high frequency (0.15–0.5 Hz); LF, low frequency (0.04–0.15 Hz); NN50, no. consecutive R-R intervals differing by >50 ms; pNN50, percentage of NN50 intervals; rMSSD, root mean square of standard deviation; SDSD, standard deviation of standard deviations; SDNN, standard deviation of the R-R interval; TP, total power; VLF, very low frequency (<0.04 Hz).
Frequency domain
TP (ms2)−0.4195−0.3138
VLF (ms2)−0.2880−0.2155
LF (ms2)−0.555−0.4064
HF (ms2)−0.3247−0.2509
Time domain
Mean of R-R interval (ms)−0.2595−0.1768
SDNN (ms)−0.4427−0.3608
CV (ms)−0.3855−0.3674
NN50 (times)−0.3704−0.3938
pNN50 (%)−0.3654−0.3723
SDSD (ms)−0.3391−0.3096
rMSSD (ms)−0.3644−0.3082


In the present study, HRV was found to decrease from the resting state to TOVA testing. The MDD group had significantly higher omission rates and longer mean response times and more variability than did the control group during TOVA testing. In the MDD group, only VLF significantly decreased from resting to TOVA, while in the control group, VLF, LF, SDNN, and CV significantly decreased from resting to TOVA. This indicated that MDD patients could not adjust HRV as well as the controls could. These data support the hypothesis that the attention deficit of MDD patients could be measured with a standardized test.

In the time domain, in comparison with the age-matched control group, MDD patients had significantly lower SDNN and CV both at rest and during TOVA testing. In the frequency domain, TP, VLF, LF and HF at rest and LF during the attention test were lower. LF for controls was significantly higher on both mental tasks and clearly differentiated the controls from MDD patients, but HRV variance at rest was better for demonstrating that differentiation.

Within-group comparisons showed that, for control subjects, the HRV features including RR mean, SDNN, CV, TP, VLF, LF, HF, and LF/HF were distinguishable in different mental states. The phenomenon was not repeated in the age- and gender-matched MDD group; therefore, the ANS of control subjects adapted during the attention test. MDD subjects did not. LF, which corresponds primarily to the sympathetic system, was the clearest distinguishing factor between the two groups. In general, MDD subjects had lower regulation of sympathetic or parasympathetic activities than did controls. The controls not only had larger HRV over time but also had better control during transitions between the two mental states. The study confirmed that, even at rest, MDD had a great impact on the ANS.

These results are consistent with those of a recent study by Udupa et al., who used different measurements of ANS function.[21] Tonhajzerova et al. also reached the same conclusions, with the additional observation that the position in which the ECG was recorded was also a variable.[22] Kikuchi et al. found that their MDD group also had a lower response to regular deep breathing.[23] Kemp et al. noted that treatment with antidepressant medication improved the symptoms of depression but did not resolve the HRV changes.[11] The present results support this finding.

The present subjects also had significant differences on the TOVA itself, with more omissions and variability and longer mean response times. Although significant in its own right, the attention deficit did not add power to the HRV differences between groups. HRV changes have also been noted in schizophrenia[24, 25] and bipolar mania.[25] HRV changes are characteristic of MDD and provide an easily measured, clinically useful way to identify these patients, but more study is needed to determine the range of variables involved, for example, age,[26] gender,[27] genetic background,[28] and various medical conditions[29] and the physiological patterns that may distinguish major mental illnesses from each other.

There are limitations to this study. The study sample size was small and the SD for the HRV indices was relatively high. Whether there is a unique diagnostic sensitivity in this group requires further investigation. All MDD patients were on medication but these were not differentiated because Kemp et al. have previously indicated that medication did not affect HRV.[11] HRV and impaired concentration are symptoms of other psychiatric diseases such as schizophrenia and bipolar disorder. Further study is needed to clarify the differences, if any.

This study is unique because of its introduction of TOVA within the context of HRV. By incorporating multiple parameters of HRV including time and frequency domains, we hoped to provide a reference for other researchers in this field. This study has shown that simple, non-invasive measurements of attention deficit and HRV could reliably identify patients with MDD and might therefore be used as routine clinical assessment tools as well as to monitor patient progress over time.


This study was supported by grants from the Buddhist Tzu-Chi General Hospital, Hualien, Taiwan (TCRD98-42) and Tzu Chi University (CMRC-P-97006). There is no conflict of interest.