AUDIOVISUAL EDUCATION AND BREASTFEEDING PRACTICES : A PRELIMINARY REPORT

A randomized control trial was conducted at the Coronation Hospital, to evaluate the tffect of audiovisual breastfeeding education. Within 72 hours trfter delivery, 340 women who affved to participate were allocated randomly to view one of two video programmes, one of which dealt with breastfeeding. To determine the effect of the programme on infant feeding a structured questionnaire was administered to 108 women who attended the six week postnatal check-up. Alternative methods, such as telephonic interviews (24) and home visits (30) were used to obtain information from subjects who did not attend the postnatal clinic. Comparisons of mother-infant relationships and postpartum depression showed no significant differences. Similar proportions of each group reported that their baby was easy to manage, and that they felt close to and could communicate well with it. While the overall number of mothers who breast-fed was not significantly different between the two pmips, there was a trend towards fewer mothers in the study group supplementing with bottle feeding. It was concluded that the effectiveness of aidiovisual education alone is limited, and attention should be directed towards personal follow-up and support for breastfeeding mothers.


INTRODUCTION
To determine the effect of the study program m e on infant feed in g , a structured q u estion n aire was administered at the 6 week postnatal visit.All interviews were done by two registered midwives who administered a structured questionnaire at the postnatal clinic (108), telephonically (24) or at home visits (30).The midwives were unaware of the group allocation, of the mothers.The questionnares were either conducted in Afrikaans or English as all the mothers could understand one of these two languages.Despite a concerted attempt made to follow up those who failed to attend the post natal clinic, we were able to follow up only 162 (47.6%) of subjects within 12 weeks after delivery.
Although the follow-up rate was only 47.6%, re-analyses of the baseline variables showed no differences between the two groups which were followed up and we have no theoretical basis on which to expect that there should have been any bias introduced by the low follow-up rate; nevertheless, this remains a matter of concern.
All com pleted questionnaires were coded and statistical calculations were performed using the CMS statistical package through the Institute for Biostatistics, South African Medical Research Council, as well as Epi Info (World Health Organisation).Statistical comparisons of continuous variables were analyzed using the Student t-test, Mann Whitney U test and proportions were com pared by m eans of 95% confidence interval (Cl) of the odds ratio and the chi-squared test.To determine w hether the d iffe r en ce betw een outcomes might be the result of chance, statistical tests of chance were used where a p value is often given.All this indicates is the likelihood that the difference between outcomes occurred by chance.Thus a p value less than 0.05 indicates a less than 5% likelihood that the difference is due to chance.It is important not to confuse the degree of certainty indicated by the p value with the clinical importance of the difference.It is possible for a study using large numbers to have a high significant value when the actual size of the difference is too small to be clinically meaningful although more often the opposite happens.Because of small numbers studied, a difference between groups which might be very important cannot be proved to be statistically significant.

For this reason many researchers prefer to use confidence intervals to indicate the range within which the true difference between alternative forms of care is likely to lie. Confidence intervals provide more information than p values because a confidence interval is defmed by two numbers rather than the single one which denotes a p value. If the values do not
include 1, the 95% confidence interval of the estimate of the difference between two observations confirms that the observed difference is unlikely to be due to chance; in addition, it provides potentiaUy important information about which of two methods is liklier to lead to a reduction or increase in the outcome.The odds ratio thus reflects the chances of reduction or increase of the effect under study.Instead of giving a p-value, we say that we are 95% confident that the true difference lies within this range (Chalmers, 1989).

Socio-demographic characteristics.
There were no significant differences betw een the so cio -d em o g ra p h ic characteristics of the two groups confirming that the randomization was effective and that similarity of the groups was not compromised by the loss to follow-up (Table 1).you think your baby is special, in-between or ordinary.P itt's D ep ressio n Questionnaire (Pitt,1968) was used to measure the depression scores and is a scale designed specifically to assess postpartum depression.This instrument was selected because it is reliable and valid.It includes all the major clinical features associated with postpartum depression and does not specify low moderate or high depression ratings but  M oth er-in fa n t r e la tio n sh ip and depression scores.2).
While the overall number of mothers who breastfed was not significimtly different between the two groups (Table 3), there was a tendency for more mothers in the control group to supplement with bottle feeding.When asked about the reasons for introducing the bottle, similar p rop ortion s rep orted that their breastmilk was inadequate.A more flexible approach to feeding was adopted by the study group.The duration of 'breastfed only' was similar in both groups.
G agliano (1988) has reviewed 25 m ethodologically sound studies on in-patient video education and one of her con clu sion s was that audiovisual stim u lation in creases short-term knowledge.As a once-off intervention we chose the postnatal period as this would be the time when preoccupations with labour and birth would be over and the mothers would be involved maximally with their babies and be most receptive to information on breast-feeding.They also would be able to put information into p ractice im m ediately w hereas information provided in the antenatal

Baby problems and advice or treatment sou^t
M others in the con trol group e x p erien ced sligh tly m ore baby problems, such as vomiting, colds and diarrhoea (Table 4).Significantly more mothers in the control group sought medical advice or treatment for their babies (Cl .21-.87).The main cause of high infant mortality rates in communities without access to safe alternative feeding methods is gastroenteritis resulting directly from bottle-feeding (Irwig & Ingle 1984).More mothers in the control group mentioned that their babies had had diarrhoea already and that they had sought medical advice before 6 weeks (Table 4).It is important to consider using audiovisual education if it reduces the use of bottle-feeding even by a small margin.

CONCLUSIONS
Because of the great importance of breast-feeding, the marginal benefits suggested by the trends in this study may be sufficient to justify the routine use of audiovisual education, particularly as it is not labour intensive and can be used on a large scale.However we would strongly recommend that such a programme be backed up by individual help, and support with breast-feeding education to fam ily m em bers and ongoing breast-feeding support services.Future research should be directed towards evaluation of the effectiveness of such an integrated programme.

RECOMMENDATIONS FOR FUTURE RESEARCH
This study piu-posefully investigated the isolated effect of audiovisual education from many parts of the world has shown low and declining rates of breastfeeding success.(Martin & White 1988).At Coronation Hospital only 40% of first time mothers are exclusively breastfeeding their babies at 6 weeks, another 33% are supplementing their babies' needs with bottle feeds and 27% are giving only bottle feeds to their b ab ies.T hirty-six p ercen t Questions that were used to measure m other-infant relationship did not constitute a single measuring instrument but were essentially rating scales eg.Do rather facilitates the location of scores on a continuum of postpartum depression.The groups did not differ significantly with resp ect to m easures o f the mother-infant relationship, or to the depression scores as measured with the Pitt's depression inventory (Table objective of this study was to determine whether meaningful improvements in breastfeeding success could be achieved by means of single exposure to a video programme.The programme aimed to create a positive attitude to breastfeeding and to give in form ation about b