Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

9. Evaluating the impact of health education systems

Stanford University, Stanford, California, USA


The Mass Media and Health Practices (MMHP) project builds on a fairly extensive history of attempts to use mass media and communication campaigns to carry out health education and health behaviour change programmes. Previous efforts had used different aspects of the approach- research for planning, integrated multi-channel campaign strategy, pre-testing of messages, social marketing techniques, and the use of behavioural methodologies to analyse and generate change - but none had combined all the components and none had tackled the complex set of objectives related to introducing oral rehydration therapy. The MMHP project was thus established with the twin goals of testing and refining the methodology of this type of campaign, and of assessing the campaign's impact on health through the promotion of oral rehydration therapy in a field setting. Because success in both objectives would have implications for health behaviour efforts in other countries, a large-scale evaluation was planned. This paper describes the evaluation design, and reports early findings from data collected during the first year of the projects. Data collection continues in Honduras and the Gambia, and more extensive analyses will be conducted as the large data base becomes available.


The evaluation was for the purpose of investigating both the effects of the campaign on people's behaviours, and the effect of any behaviour change on health status. It was recognized that, in order for any changes in behaviour or health status to take place, a complex sequence of events had to occur beginning with exposure of the target audience to the campaign elements, proceeding through the learning of campaign information and adoption of new behaviours, and culminating in change in health status as a result of the behavioural changes. At each step the net impact of the project would diminish; if a failure occurred at any point along the path, no further impact would be expected.

To guide the planning of the evaluation, Stanford's Institute for Communication Research developed a detailed "Process Model" describing the sequence of events that would lead to eventual change in health status. A simplified version of that model called for the target audience to be exposed to the campaign effort, for them to learn from that exposure and remember and accept the messages, for behavioural change to occur, and for those behavioural changes to result finally in changes in health status. The model pointed clearly to the need to measure results at each level - exposure, learning, behaviour, and health status - in order to answer questions about the effects both of the campaign and of oral rehydration therapy. Hence variables were identified to track the impact at each level.

The research design developed to track the impact of the campaign involves a panel study of roughly 750 families in each country over a little more than two years. In Honduras families were recruited into the study from 20 communities under a stratified, purposive sampling plan with random selection of families within communities. In the Gambia, where communities are organized into compounds that often include multiple or polygamous families, communities were purposively selected, and compounds were randomly sampled, after which women were randomly sampled within compounds. A woman was considered eligible for the study if she was between 18 and 45 years old or was responsible for caring for small children.

Mothers are visited approximately once a month by a local field-worker who administers questionnaires and conducts observations and anthropometric measurements. A general "baseline" survey is conducted at the beginning and end of the study; in the intervening months, batteries of items relating to topics of specific interest are administered or repeated. Thus, for example, four batteries are used in Honduras, covering anthropometry, morbidity, communication, and child-care practices. Administration cycles have averaged somewhat longer than one month, so instrument batteries are repeated roughly every six months. In addition to the panel surveys, a number of much smaller ancillary investigations are conducted in order to supplement the information about exposure, learning, behavioural change, and health status change. The other studies include interviews with health workers, a community mortality study, ethnographic research in two communities, a cost-effectiveness study, and an administrative history.


Findings from the evaluation research at this point are based primarily on analysis of the survey data from the panels of families in each country through the first year of the two-year projects. As such, they are interim results and do not yet reflect the full impact of the programme. They are presented here in the order specified in the process model. First, access to the channels used by the campaign and exposure to the campaign messages are examined. Learning of campaign information is assessed, and then behavioural change is investigated. Finally, health status is examined. An attempt is made to report parallel data from both projects; however, differences in the interventions and in the data collection procedures limit the direct comparability in some cases, and in other cases, for the sake of brevity, only one site's data are presented as illustrative.

Access and Exposure

We first set out to determine whether the three principal communication channels used by the campaign - radio, interpersonal contact through the health-care system, and print materials were feasible means of delivering messages. We found in general that the population had good access to the channels and reported high exposure to campaign components.


Ownership and functioning of radios among the sample were assessed by asking participants whether they had a radio and asking them to turn it on to demonstrate that it worked. In Honduras, 80 per cent of the families owned radios, and 71 per cent of the families had a radio that worked on the day of the interview. In the Gambia, 66 per cent of the compounds had radios, and in 60 per cent of the compounds they were demonstrated to be in working order.

In addition to the existence of radios, the amount of radio use was investigated. The results support the notion that radio has very high penetration in rural areas and is listened to frequently. in Honduras, 63 per cent of mothers reported listening to the radio on the previous day. They start listening as soon as they get up, and radio use peaks between 6 and 7 a.m., with 38 per cent of all households tuned in at that time. There is a gradual decline over the course of the day, with the exception of another peak (36 per cent) between 12 and 1 p.m., at lunch. Relatively little radio listening takes place in the evening. In the Gambia, radio listening is also high, but the pattern contrasts sharply with that of Honduras. The most common time of listening is 8 p.m., which over 70 per cent of the women say is their preferred listening time.

Women who listen to the radio also remember hearing campaign messages as measured by a 24-hour radio listening recall test. For example, in Honduras, 44 per cent of the mothers remembered hearing at least one MMHP spot advertisement on the previous day. Mothers who listened to the radio on the previous day remembered hearing an average of more than three spots. Clearly, the use of radio as an information distribution channel to rural mothers is a success.

Interpersonal Contact

Both countries train the existing infrastructure of health-care workers about diarrhoeal disease and use them to promote campaign objectives. It is therefore important to establish whether contact between rural families and representatives of the health-care system took place frequently enough to expect this component to work. The systems of primary care in rural areas in both Honduras and the Gambia rely on tiered systems of facilities, ranging from hospitals down to community volunteers. There are also well-established systems of sources of traditional care present in the communities. In Honduras, 47 per cent of mothers had had contact with some type of healthcare worker in the previous six months. Thirty-three per cent of all mothers had seen a physician or a nurse at a clinic. Twenty-one per cent of all mothers had been in contact with some level of community-based health worker, such as local volunteers, midwives, or health educators. Eleven per cent had dealt with some "traditional" source of care, such as a healer or a masseuse. Obviously, some mothers had contacted more than one source of care. In the Gambia prior to the MMHP campaign, 85 per cent of mothers said they had gone to the health centre for their child's last case of diarrhoea, while 8 per cent said they had gone to a local healer. Two out of three Gambian mothers can reach a health-care facility within 90 minutes, usually by walking. These are both very high levels of interaction with representatives of the health care system and make the planned use of interpersonal channels a very reasonable approach. It also demonstrates a conclusion from this and other data from both countries - that the system of traditional or folk medicine co-exists with the western or modern system, with care being sought from either system according to the person's self-diagnosis, and sometimes from both systems simultaneously. The strong preference in most cases is for the modern medical system.


The campaigns in both countries rely on some use of printed materials, such as posters, flyers, and pictorial instructions. Thus, we checked the literacy in our sample. Honduran mothers were able to read a complex phrase 57 per cent of the time. In households where the mother could not read, other family members were tested for literacy. The household literacy rate was an impressively high 87 per cent. In the Gambia, the situation was markedly different. Fewer than 3 per cent of mothers could pass the literacy test, and only about a third of the compounds had a reader present when we conducted our test.

The print materials used in the two projects have been designed with these limitations in mind. In Honduras, posters have been widely distributed, as have instruction sheets with pictures and writing. After a year of campaign activity, using a fairly strict test of unaided recall, nearly half (47 per cent) of Honduran mothers could describe a MMHP poster they had seen well enough for the interviewer to identify the poster. In the Gambia, a flyer that relied almost exclusively on pictures to explain the mixing of oral rehydration solution was prepared and distributed over the entire country. In our sample, 79 per cent had seen the flyer and 71 per cent could show us their copy of it. Thus, properly designed print materials can also be used as a communication channel, even in circumstances in which literacy is quite low.


Once it was established that the channels used by the campaigns were available to the mothers and could, in fact, be used to deliver campaign messages, the evaluation effort focused on measuring learning through those channels. Learning has been measured using recall, recognition, open-ended and true-false items about the specific content of the instruction given in the radio spots and programmes and by health workers. Items relating to specific topics have been combined into indices to estimate learning gains. The result is an impressive level of learning, both in terms of absolute levels of knowledge and in terms of improvement. This paper will present a few examples rather than try to summarize all the learning data.

In Honduras, the oral rehydration solution promoted is called Litrosol, and it did not exist before the programme began. By the end of one year of campaign activity, 92.5 per cent of mothers could name Litrosol as the medicine being promoted by the campaign. One-third of the mothers could define dehydration, a concept that pretesting had shown to be absent before the campaign. When prompted with the first lines, more than 70 per cent of mothers could complete the campaign jingle having to do with giving liquids during diarrhoea. An average of 62 per cent of jingle-completion items were correctly answered by mothers. Sixty-five per cent of mothers know of Dr. Salustiano, the main character in one group of radio spots. Ninety-five per cent of mothers named breast-milk as the best milk for babies (up from 62 per cent after six months of campaign). Five indices were constructed to measure learning from Dr. Salustiano spots in general, Dr. Salustiano's specific spots about dehydration, Dona Chela spots (another character used in the radio campaign), spots about dehydration, and spots about breast-feeding. Learning was assessed after six months of broadcasting and again after twelve months. The performances at the six-month period averaged 11 per cent of the index value, and after a year almost 18 per cent, an average gain of over 60 per cent. The range of gains on the five indices was 35 per cent to 180 per cent.

The campaign in the Gambia relies heavily on longer radio programmes and other promotional activities, such as a lottery, to instruct and motivate mothers. Hence assessment of learning was targeted on more general tests of knowledge than message-specific tests as in Honduras. One of the most important content areas has to do with teaching the method of mixing water-sugarsalt solution for home administration of oral rehydration. At the very start of the campaign, only one mother out of about 800 could correctly give the formula for mixing water-sugar-salt solution (WSS). By December of 1982, nine months into the campaign, 64 per cent of the mothers could given the entire formula. This represents an astonishingly powerful learning activity, since the information itself is fairly complex and the means used to teach it involve voluntary exposure to intermittent inputs. At that point in time, only 20 per cent of the mothers had no heard of the water-sugar-salt medicine that you can make at home.

One of those inputs was the "Happy Baby Lottery," which was promoted for a few months by radio and actually involved only 72 of the Gambia's hundreds of villages. None the less, half the mothers in the sample had heard of the lottery and 71 per cent actually had the instructional flyer that was used as an entry requirement. An analysis of other items on a wide variety of topics covered in the campaign shows an average gain of 35 per cent between the sixth and ninth months of the campaign.

There is no doubt that the specific content being taught by both campaigns is being learned and retained by the mothers. The general pattern seems to be one of fast initial gains followed by slower gains as time goes by.


The analysis to date of behaviour changes associated with the project has concentrated on the adoption of Litrosol and water-sugar-salt therapies for dehydration. ln Honduras, the use of Litrosol packets in the home is a new behaviour. The proportion of cases of diarrhoea treated with Litrosol rises from zero prior to the campaign, to nine per cent after three months of broadcasting (at which point supplies of packets were just beginning to be available in the community), to 26 per cent after a year of the effort. At the end of the first year, half the mothers reported that they had used Litrosol at some point. About two-thirds of the women who have used it report that they obtained it in their own community; the rest tend to get it at clinics or at the hospital. Knowledge on how to mix the solution is high - 95 per cent of the women know they should use a litre of water and put in the whole packet. Most women (58 per cent) reported that they learned how to mix it from the packet (which comes with an instruction sheet). Fewer women (43 per cent) learned from interpersonal sources, and far fewer from the radio (14 per cent). There is a positive relationship between the severity of the case and likelihood that it will be treated with Litrosol.

In the Gambia, various techniques of mixing and using water-sugar-salt solutions had been promoted by the Ministry of Health in the past. However, none of these used the same formula as the one in the current campaign. The use of water-sugar-salt solutions rose from 4 per cent of all cases of diarrhoea at the beginning of the campaign to 48 per cent of all cases after nine months. If the analysis is restricted to those mothers who treat their child's diarrhoea themselves, the rise in adoption of solution is even more marked. The proportion of mothers treating cases themselves rose from 18 per cent at the start to 54 per cent after nine months. The percentage of those mothers treating with solution rose from 21 per cent to 89 per cent over the same period. Use of other, often inappropriate, treatments falls correspondingly.

The third step in the sequence of events that the evaluation's process model postulates, the adoption of the promoted behaviours, is also attained. Many mothers have acquired the skills and are introducing the behaviour into their patterns of response to diarrhoeal disease.


The health status variables being monitored in the evaluation include anthropometric measurements, morbidity, and mortality.

Morbidity has been measured in various ways. Prevalence of diarrhoea among children under five on the day of the interview in Honduras has shown a high of 14.6 per cent (in June) and a low of 8.7 per cent (in March) in the two assessments analysed thus far. Incidence in the previous two weeks was 36 per cent in June and 21.5 per cent in March. These differences are probably seasonal variations, as treatment with Litrosol is unlikely to affect the incidence of the disease. There appears to be a rapid drop-off in memory of cases; reported prevalence on the fourteenth day prior to the interview is roughly half the day-of-interview rate. None the less, by combining data from various sources it is possible to estimate that a minimum of half the children have had an episode of diarrhoea in the six months prior to the interview.

Incidence in the Gambia is roughly the same and, because more closely spaced assessments are available, can be shown to exhibit a clear seasonal variation. Two week period prevalence ranged from a low of 10.3 per cent of children during the dry season to a high of 33.5 per cent during the rainy season.

The baseline anthropometric measurements of nutritional and growth status in Honduras have been analysed. Approximately 30 per cent of children under five are "stunted" (i.e. less than 90 per cent of height for age), while only slightly more than 1 per cent are "wasted" (i.e. less than 80 per cent weight for height). Thus, while a cumulative growth deficit is occurring, the children did not appear to be acutely malnourished in June of 1982. Subsequent analyses will examine the change in health status and its relation to illness and oral rehydration therapy.

Mortality data are being collected on the entire population of the communities from which the study sample is drawn, thus giving a substantially larger population base. Data from Honduras show a significant drop in the proportion of deaths that involve diarrhoea among children up to 24 months old. For a comparable set of seasons, diarrhoea was mentioned as a cause of death in 45.3 per cent of the deaths for one and a half years before the start of the intervention and in 25.0 per cent of the deaths in the year and a half following the availability of oral rehydration solution. This decline is significant at the p< .005 level. Mortality data are also being collected in the Gambia but are not yet complete.


This paper has presented the evaluation plan and some of the early findings of the evaluation of the Mass Media and Health Practices projects in Honduras and the Gambia. The findings indicate that: the target populations have good access to the communication channels used by the campaigns; the target populations are exposed to and aware of the campaign activity; the audiences are learning a great deal of the campaign information; they are adopting the treatment behaviours of oral rehydration therapy on a large scale; and the proportion of mortality among children of less than two that involves diarrhoea has fallen significantly in one country.

Subsequently analyses will follow these same issues through the data from the second years of each campaign and relate the behaviour to changes in health status of the treated children.

Contents - Previous - Next