Material & Methods

Subjects: The subjects were persons of both sexes, over the age of 60 yr in a residential home for the aged in Bangalore city, south India. The total number of residents was 120; 30 of them were ill or bed-ridden. The remaining 90 persons were told about the trial. All of them expressed their willingness to participate in the trial. The protocol was approved by the ethical committee of the institution and the signed informed consent of each subject was taken.  They were screened using the electrocardiogram (all leads), fasting blood glucose, and blood pressure measurements, as well as a detailed clinical examination.
Subjects with the following health problems were excluded from the study: uncontrolled diabetes (seven participants), uncontrolled hypertension (four), neurological disorders (three), dementia (one), hearing impairment (five), and a detected case of non-infective Hansen’s disease. Sixty nine subjects were included for the study after this screening.

Randomization: The 69 subjects were stratified according to age [five-year intervals, between 60 and 65 yr (lower limit), and between 90 and 95 yr (upper limit)]. Within a particular five-year age range, subjects of each gender separately, were randomized as three groups (groups 1, 2 and 3) using a standard random number table. Allocation of a group to a particular intervention was carried out by the lottery method, as follows: The three interventions ‘Yoga’, ‘Ayurveda’ or ‘Wait-list control’ were written on three similar pieces of paper which were folded. A person who had no other part in the trial, picked up and opened the folded papers. The first intervention to be picked up was assigned to group 1, and accordingly for groups 2 and 3. Following stratified sampling and random allocation, there were 23 subjects in each group (including seven males in the Yoga group and six males each in Ayurveda  and Wait-list control groups) with average ages (± SD) of  70.1 ± 8.3, 72.1 ± 9.0, and 72.3 ± 7.4 yr,  respectively.

The Ayurveda group was selected so as to independently evaluate the effects of a system of treatment (and of lifestyle) which is complimentary to Yoga. The idea was to evolve a comprehensive programme with some aspects each from Yoga and Ayurveda, for a geriatric population. The Wait-list control group was selected to study the retest effect (as assessments were made thrice) on subjects who were equally motivated to receive Yoga or Ayurveda if the interventions were allocated to them.

Study protocol: All three groups were assessed for self rating of specific aspects of sleep at baseline, and after three and six months of the interventions (Yoga, Ayurveda, or Wait-list control) using a sleep rating questionnaire. There were seven questions, which subjects were asked to answer based on their experience during the week prior to assessment. Asking them to recall their quality and amount of sleep in the week prior to assessment was important, as recall over longer periods is especially likely to be influenced by anamnesis in older persons. The questionnaire consisted of seven questions. The questions were either dichotomous (i.e., two options: yes/no; Questions 5 and 6) or open questions (i.e., Questions 1, 2, 3, 4 and 7).

The questions were:
1. Approximately how long (in min) does it take you to fall asleep?
2. How many hours do you sleep each night?
3. How many times (if any) do you wake up during the night?
4. What are the usual reasons for wakening up, if you do so?
5. Do you feel rested in the morning (yes/no)?
6. Do you sleep in the day time (yes/no)?
7. If your answer to Question 6 was ‘yes’, for how long do your day time naps last (in min)?

The sleep rating questionnaire has been evaluated for its reliability and validity based on standard criteria. Reliability was ascertained based on (i) temporal stability, and (ii) internal consistency. To assess temporal stability the correlation coefficients were calculated using the data of the ‘no intervention’, Wait-list control group with two correlations being made, viz.: (i) baseline with three months, and (ii)baseline with six months. Of the five variables for which the correlations were made, the temporal stability was demonstrated for four (Table I). In order to evaluate internal consistency the correlation between two variables, which assessed an equivalent aspect of sleep, was calculated. The two variables were the number of hours slept each night and the feeling of being rested in the morning. The values for the three groups were as follows: Yoga (r = 0.643), Ayurveda (r = 0.578) and Wait-list control (r = 0.699). Validity was inferred based on the content and indirectly based on the test for internal consistency described above.

To eliminate the possibility of bias the examiner was kept unaware of the group to which the subject belonged.

Ayurveda - The Ayurveda group received a herbal preparation i.e., a ‘rejuvenating tonic’ ( Rasayana Kalpa in Sanskrit). The participants were given 10 g (1 tablespoon, approximately) of Rasayana Kalpa, twice a day, once in the morning (0600) and again in the evening (1800). After both doses they were asked to drink 200 ml of skimmed milk, as is prescribed in Ayurveda texts . Though the prescribed dose was 48 g per day , the present dose was suggested by Ayurveda experts who were consultants for the project. This preparation (10 g) consisted of the following herbs (the Sanskrit names are given in parenthesis):Withania somnifera (ashwagandha roots, 2 g), Emblica officinalis (amalaki, 1 g), Sida cordifolia (bala, 0.25 g), Terminalia arjuna (arjuna, 0.25 g), Piper longum (pippali, 0.5 g). The other contents were: sugar (4 g), honey (2 g), water and clarified butter (ghee) in the amount required to get the correct semi-solid consistency.

Yoga training: The Yoga session was planned to include: physical activity, relaxation, regulated breathing and philosophical aspects. This was an integrated approach of yoga, derived from principles in ancient texts which emphasize that yoga should promote health at all levels . The session was for sixty minutes daily, for six days a week. Subjects practiced breathing exercises (10 min), loosening exercises (shithilikarana vyayama, 5 min), physical postures (20 min) [i.e., tadasana (mountain posture) padahasthasana (hand- to-foot posture), ardhakatichakrasana (lateral arc posture), ardhachakrasana (half wheel posture), viparithakarani (half shoulder stand posture), matsyasana (fish posture), bhujangasana (cobra posture), shalabhasana (locust posture), makarasana (crocodile posture), vakrasana (sitting sideward twist posture), paschimothanasana (back-stretching posture), ushtrasana (camel posture), shashankasana (moon posture), vajrasana (diamond posture), (ardha)padmasana (half lotus posture), shavasana (corpse posture)], voluntarily regulated breathing (pranayama, 10 min) such as: nadishudhi (alternate nostril breathing), brahmari (bumble bee breathing), surya anuloma viloma (right nostril yoga breathing), and chandra anuloma viloma (left nostril yoga breathing) and yoga-based guided relaxation (15 min), which has been described elsewhere . There was an additional session in the evening which consisted of devotional songs (bhajans, 15 min) and lectures on theory and philosophy of yoga alternating with ‘cyclic meditation’. The last technique is derived from another ancient Indian text (the Mandukya Upanishad) and involves alternating cycles of physical postures and supine rest.

Statistical analysis: Data were analyzed using the statistical package (SPSS Version 10.0). The data at baseline, and at three and six months of all three groups were assessed with tests for normality distribution using both graphic presentations (box plot and stem and leaf plot) as well as Kolmogorov- Smirnov test.

Repeated measures Analysis of Variance (ANOVA) was used to test for (i) significant differences between the assessments (baseline, three and six months) of all three groups i.e., within-subjects factor, and (ii) differences between the groups (Ayurveda, Yoga and Wait-list control) i.e., between-subjects factor. The ‘t’ test for paired data was used to compare data at three and six months with those at baseline for each group, separately. These parametric tests were used even though the data were found to be not normally distributed as it has been shown that analyses of variance and t tests are usually robust enough to perform well even if the data deviate somewhat from the requirements of normality and homoscedasticity .
Since the ‘feeling of being rested in the morning’ was dichotomously scored, the data were binary. Hence, a nonparametric statistical test, McNemar test was used to compare data at three and six months with those at baseline for each group, separately.