CARDIOVASCULAR JOURNAL OF AFRICA • Vol 23, No 2, March 2012
70
AFRICA
In order to determine whether there were any gender differ-
ences in the HRV parameters, the data for each HRV parameter
for days 2, 3 and 4 were combined and analysis of variance
(ANOVA), with a Tukey
post-hoc
testing was performed to
compare males with females. The data were analysed with
STATISTICA version 8.0 (Statsoft Inc, Tulsa, Oklahoma, USA)
for any statistical significance (
p
≤
0.05).
Results
A total of 44 participants with complete data sets were analysed.
The outcomes of the reliability analysis are presented in Table 2
for day 2 versus day 3 and day 3 versus day 4.
Typical error of measurement
The absolute reliability indices for the HRV parameters are
shown in Table 2. The data presented here offer precision esti-
mates for single measurements of Suunto t6 HRV and data for
making decisions when monitoring changes or responses to
interventions in individuals.
Overall for both males and females, the results show that for
most HRV frequency-domain parameters, the second compari-
son day, day 3 versus day 4 had lower TEM values compared to
the first day, day 2 versus day 3. This slight decrease in the TEM
could be due to a familiarisation effect.
3
Females demonstrated
better absolute reliability in the HRV frequency domain as
demonstrated by lower TEMs. Specifically, the TEMs for day 3
versus day 4 were lower in females for the LF/HFnu ratio (116%
lower), 90% lower for HFnu (day 2 vs day 3) and 68% lower
for LFnu (day 2 vs day 3). The lowest typical error of measure-
ment as a percentage (TEM%) was 11.5% and was found in the
females for Hfnu (day 2 vs day 3).
Time-domain results had a low TEM% for IBIs (day 2 vs day
3: 4.8 males and females; day 3 vs day 4: 4.9 females and 4.1
males). Overall, the TEM% was relatively high in most HRV
parameters, specifically for LF/HFnu (day 2 vs day 3: 31.4%
females and 48.1% males; day 3 vs day 4: 29.7% females and
40.4% males). The TEM was similar for males and females for
the time-domain parameters, specifically, RMSSD, pNN50 and
the IBIs. However, the TEM for SDNN was 42% lower in the
males (day 3 vs day 4).
Interclass correlations
The interclass correlations (ICCs) for relationship 1 ranged
between 0.36 and 0.88 for the AR frequency domains and
0.70–0.92 for the time domains (Table 2). For day 3 versus day
4, the ICCs ranged between 0.72 and 0.86 for the AR frequency
domains and 0.72–0.93 for the time domains. The ICCs for
both day 2 versus day 3 and day 3 versus day 4 indicated good
to excellent (
>
0.8) reliability correlations for IBIs and pNN50
from the time-domain results. RMSSD reliability was good to
excellent with the exception of the males for day 3 versus day 4,
and showed substantial reliability of 0.79. SDNN variable results
depicted substantial reliability for both day 2 versus day 3 and
day 3 versus day 4.
Reliability ICCs for the frequency domain were lower
compared to the time domain. The ICCs in the frequency domain
were between 0.36 and 0.88, with the lowest value for these
correlations being the LF/HFnu ratio (ICC
=
0.36) for males.
It was found that overall the male participants had lower ICCs
when compared to females for the same HRV parameter in the
frequency-domain analysis.
The ICCs for most time-domain parameters were similar
between the males and females, with the exception of the IBIs.
Males displayed high IBI ICC values (day 2 vs day 3
=
0.90, day
3 vs day 4
=
0.93), which indicated higher relative reliability
compared to the females (day 2 vs day 3
=
0.83, day 3 vs day 4
=
0.79). Time-domain ICCs that were similar between males and
females were SDNN (day 3 vs day 4), males (0.73) and females
(0.72) and pNN50 (day 2 vs day 3), males (0.86) and females
(0.84).
TABLE 2. RELIABILITY INDICES FOR FREQUENCYAND TIME DOMAIN HRV PARAMETERS
Day 2 vs day 3
Day 3 vs day 4
HRV parameters
TEM
TEM (%)
ICC
TEM
TEM (%)
ICC
LF/HFnu
Female
0.25 (0.20–0.36)
31.4 (25.1–45.2)
0.80 (0.59–0.91)
0.24 (0.19–0.35)
29.7 (23.5–43.4)
0.83 (0.63–0.92)
Male
0.54 (0.41–0.78)
48.1 (36.5–69.5)
0.36 (0.08–0.69)
0.43 (0.33–0.62)
40.4 (31.0–58.3)
0.72 (0.41–0.88)
HFnu
Females
6.63 (5.13–9.38)
11.5 (8.9–16.2)
0.88 (0.74–0.94)
7.92 (6.13–11.21)
13.4 (10.3–18.9)
0.81 (0.61–0.92)
Males
12.58 (9.62–18.17)
24.6 (18.8–35.5)
0.55 (0.17–0.79)
8.95 (6.85–12.93)
16.7 (12.8–24.2)
0.78 (0.53–0.90)
LFnu
Female 7.78 (6.26–10.38)
17.8 (14.3–23.8)
0.84 (0.70–0.92)
7.26 (5.84–9.69)
17.2 (13.9–23.0)
0.86 (0.74–0.93)
Male
13.04 (10.41–17.70) 26.6 (21.2–36.1)
0.52 (0.19–0.74)
9.40 (7.50–12.76)
20.2 (16.11–27.4)
0.75 (0.54–0.88)
IBIs (ms)
Females 44.90 (35.21–61.98)
4.8 (3.8–6.6)
0.83 (0.67–0.91) 45.89 (35.99–63.34)
4.9 (3.9–6.9)
0.79 (0.60–0.89)
Males
53.46 (40.90–77.20)
4.8 (3.6–6.9)
0.90 (0.79–0.95) 46.46 (35.54–67.09)
4.1 (3.2–5.9)
0.93 (0.85–0.96)
SDNN
Females 28.83 (22.30–40.80) 31.2 (24.1–44.1)
0.77 (0.54–0.90) 35.58 (27.52–50.36) 32.2 (28.8–52.7)
0.72 (0.45–0.87)
Males
20.70 (15.83–29.89) 19.9 (15.2–28.8)
0.70 (0.40–0.87) 20.62 (15.78–29.78) 20.2 (15.4–29.1)
0.73 (0.45–0.88)
RMSSD
Females 17.29 (13.37–24.47)
20.5 (15.8–29)
0.92 (0.82–0.97) 20.12 (15.56–28.48)
22.6 (17.5–32)
0.91 (0.80–0.96)
Males
19.12 (14.63–27.62) 18.7 (14.3–27.1)
0.83 (0.64–0.91) 20.53 (15.71–29.65) 20.3 (15.5–29.3)
0.79 (0.55–0.89)
pNN50
Females 9.56 (7.39–13.53)
19.1 (14.8–27)
0.84 (0.67–0.93)
8.80 (6.80–12.45)
18.7 (14.4–26.4)
0.87 (0.72–0.94)
Males
7.55 (5.78–10.90)
14.2 (10.9–20.5)
0.86 (0.69–0.94)
7.07 (5.41–10.21)
13.2 (10.1–19.1)
0.87 (0.72–0.95)
HRV = heart rate variability.
Frequency domain: LFnu = low-frequency normalised units, HFnu = high-frequency normalised units, LF/HFnu = low-frequency to high-frequency ratio in
normalised units.
Time domain: IBI = R–R intervals, SDNN = standard deviation of normal-to-normal intervals, RMSSD = root mean squared differences of the standard deviation,
pNN50 = percentage of beats that changed more than 50 ms from the previous beat.
Reliability: ICC = intra-class correlation and is expressed as a mean (95% CI), TEM = typical error of measurement, TEM% = typical error of measurement as a
percentage, and both are expressed as means (90% CI).
