26 healthy subjects (13 boys and 13 girls) aged 16.1 ± 0.2 (mean ± SD) years were recruited to the shavasan group and 17 age and gender matched healthy subjects (9 boys and 8 girls) aged 15.8 ± 0.6 years to the control group (P=0.07). The BMI of the shavasan and the control groups were 20.8 ± 2.8 kg/m2 (mean ± SD) and 19.3 ± 3.5 respectively (P=0.16). ECG (a bipolar chest lead) was continuously acquired at a rate of 1000 samples per second for five minutes using the BIOPAC® MP 100 hardware (BIOPAC Systems Inc., USA) and the Acknowledge® 3.7.1 software (BIOPAC Systems Inc., USA) and a Microsoft Windows-based PC. Blood pressure was measured by an automated non-invasive blood pressure monitor (Colins Press-Mate BP 8800, Colin® Corporation, Japan). Recordings were obtained in the Polygraph laboratory, between 10.00 am and 12.00 pm, 3 h after a light breakfast. The environment was quiet, the laboratory temperature 25°C, and the lighting subdued. After familiarizing the subject with the procedure and at least 10 minutes of rest in the supine position, ECG was recorded with subjects in the supine position. They were instructed to breathe quietly at about 12 breaths per minute during the recording. None of the subjects were taking any medication influencing autonomic function. Under similar conditions, we determined the resting BP, heart rate (HR) and HRV, before and within 5 days after 6 weeks of shavasan training. Shavasan group was taught shavasan by a trained yoga teacher and subjects practiced the same for 15 minutes a day, four days a week for a total duration of six weeks. The control group did not receive any shavasan training. The technique of shavasan is given elsewhere (8). The local ethics committee approved the study protocol. Written informed consent was taken from the parents of all subjects. 

HRV analysis was done conforming to established standards (9) using the Acknowledge 3.7.1 software. Briefly, ectopics and artifacts in the ECG were edited and a 256-second long RR interval tachogram obtained by using a rate detection algorithm. The RR interval tachogram was re-sampled at 4 Hz, its mean and trend removed, a Hanning window applied and transformed by fast Fourier algorithm to obtain a power spectrum of RR intervals. Low frequency power (LF power) and high frequency power (HF power) were obtained by integrating the spectrum from 0.04-0.15 Hz and 0.15-0.40 Hz respectively (9). Total power was calculated as the sum of LF and HF powers (5). 

The shavasan and control groups were compared by unpaired ‘t’ test. After the training period, the spectral powers of the two groups did not follow a normal distribution and these were compared using a non-parametric test (Mann-Whitney test). Changes within the shavasan and control groups after the training period were analyzed by paired ‘t’ test A P value less than 0.05 was taken as indicating a statistically significant difference between the compared means.