165.pdf

WELL Study
Effectiveness of promotional
techniques in environmental
health
Task No: 165
Task Management by: Dr Valerie CurtisQuality Assurance by: Dr Sandy Cairncross London School of Hygiene & Tropical Medicine, UKWEDC, Loughborough University, UK E-mail: WELL@lshtm.ac.ukInternet: http://www.lboro.ac.uk/well/ TABLE OF CONTENTS
Summary.3
Introduction: Changing behaviour; how and how much?.3
Methods.6
Results.7
Studies exhibiting desirable methodological characteristics.7 Example given of the materials or process employed.9 Adequate description of resources required to carry out programme.9 Measured outcome variable before and after the intervention.10 Period between education and outcome more than one year.10 Evidence of community participation in design of programme, goals or outcomemeasures.11 Article claimed to show positive results for intervention evaluated.11 Included discussion of possible biases or caveats.11 3.1.10 Included p-values or confidence intervals.113.2.
Discussion.13
Conclusions.15
References.16
Annex: Summary of published articles consulted.19
Box 1: Health promotion and health education……………………………………………………… 4
Box 2: Criteria for assessing the articles on health education or promotion interventions……… 7
Box 3: Five-stage process used to develop and implement education interventions…………….8
Box 4: Excerpt from Project Salsa Interventions…………………………………………………….9
Box 5: Articles found to have at least 60% of Loevinsohn's characteristics……………………. 13
Behaviour change is a key component in environmental health programmes. However behaviourchange is difficult and little evidence is available as to how much behaviour change can beexpected from health education and promotion.
• In 1987 Loevinsohn found only three methodologically sound studies with evidence of health impact. One of these showed evidence of behaviour change (increase in contraceptive use).
• We found five further studies published since then which had sufficient rigour to evaluate impact. Four demonstrated health impact. In only two was there clear direct evidence ofbehaviour change (purchase of water filters).
• Of eleven other studies with sufficient rigour to contribute lessons, only three showed useful evidence of any behaviour change. One was an intensive hand-washing and soapdistribution programme in one village, another was a village which decided to give upsmoking. The national diarrhoea control programme in Mexico increased sales of ORS 10fold in 11 years.
• In all other studies behaviour change was not reported, not found or very small.
Poor results stem either from a failure in conception, in execution, or in measurement. Only whenbetter-designed interventions and evaluations take place can we determine why so many effortsappear to fail.
• Better intervention design involves explicitly researching how to change specific behaviour from the outset and using messages about a few behaviours which are simple and cheap toput into practice. Political, social and economic barriers to behaviour change must beassessed and may need to be addressed before the intervention begins.
• Behaviour change can be slow and require intensive or prolonged intervention that may not be replicable or sustainable outside of the context of a research study. Slow, steady long-term progress may be a more appropriate outcome expectation than immediate majorimpact.
• Better evaluation design requires better process monitoring and evaluation, better measures of behaviour, and more rigour and scepticism. Cost-effectiveness evaluations of successfulprogrammes are urgently needed.
Introduction: Changing behaviour; how and how much?
The promotion of healthy behaviour is a major component of current efforts to prevent malaria,diarrhoea, STDs and AIDS, the immunisable diseases, heart and lung disease, malnutrition andother conditions such as unwanted pregnancies. However, good evidence for behaviour changein health programmes in both developed and developing countries is lacking. We do not know ifthis is because of problems with the design, implementation or evaluation of the interventions or,more seriously, if there are major misconceptions about the behaviour change process itself.
In 1990, Loevinsohn published a literature review of health education in developing countries (1).
In this review, he found only three that were rigorous enough to demonstrate their effectivenessand, at the same time, were presented with enough detail to allow the intervention to beduplicated.
Ten years later there have been a number of developments in this field: • health education has given way to health promotion (2)in theory if not in practice (Box 1); • there is growing realisation of the importance of behaviour change in environmental health programmes such as those promoting insecticide treated nets (ITN) and water-supply and sanitation; • further, potentially more rigorous studies and evaluations have been published.
Planners of environmental health programmes need to know whether the promotion of behaviourchange is effective and cost-effective and, if so, how they can maximise the impact of theirefforts. They also need to be able to estimate the amount of impact they can expect to have onkey behaviours with available resources. Whilst it is not possible to answer such questionsprecisely, it is worth combing the literature that has been published since Loevinsohn's review tosee if new light has been shone on these questions.
Box 1: Health promotion and health education
The Ottawa Charter for Health Promotion incorporates elements of traditional concern (personal
health skills) with the more recent attention given to community action and environmental and
public policy issues†. The five key action domains are:
The charter has had wide acceptance as a strategic 'checklist' for health promotion, if not actuallybeing used as a formal strategy. Health education may be seen as the 'communication sphere' ofhealth promotion‡.
† Ottawa Charter for Health Promotion, 1986, World Health Organisation, Health and Welfare Canada, Canadian Public Health Association. Ottawa Charter for Health Promotion: Ontario, Canada.
Tannahill, A., Health education and health promotion: planning for the 1990's. Health Education Journal, 1990. 49: p. 194-198.
Behaviour change
Behaviour change and its resultant health benefits can only be expected if every one of the linksin the following chain functions effectively: • appropriate and feasible safe practices are targeted and a limited number of effective • messages are delivered through appropriate channels, and repeated often enough; • target groups take in, understand and process the messages; • target groups are motivated and able to act; Finally, any new behaviours which are adopted in the short term may, or may not be sustained inthe long term.
A full understanding of any behaviour change intervention would ideally evaluate success atevery one of these points in the process. However this is unlikely to be feasible in practice. At aminimum, measuring behaviour change outcomes is needed to determine the impact of a healthprogramme, and to find ways to modify it, if necessary (3). However, measuring behaviourchange is not always straightforward. Questionnaire surveys are often a poor way of findingreliable information on health related practices; especially those that are private or morallyloaded. Direct observation of the behaviour concerned can be expensive, intrusive and influencethe behaviour in question. It may nevertheless be the best technique available(4). Physicalevidence of behaviour change such as sales records of oral rehydration salts (ORS), condoms orvaccination records, may be a good way of assessing whether behaviour change has takenplace. However, physical signs alone can be misleading; soap may be present but not used,latrines may have been built but children still defecate on the ground, ITNs and ORS may bebought, but used incorrectly. Some direct observation may still be required to evaluate the impactof an intervention.
Health impact
Studies to evaluate health programmes have naturally tended to use health as the outcomevariable. However this can be a mistake. Health impacts are often small, meaning that largesample sizes are needed to detect them, and distinguishing the 'signal' of the programme impactfrom the background 'noise' of other events may be impossible.
Whether the outcome measured is behaviour, health, or some other indicator, the best studydesign is the randomised controlled trial (RCT). However as health promotion is often acommunity level intervention, it not suitable for individual randomisation. A design which makes a'before-and-after' comparison between one intervention and one control community has limitedvalue (5,6). Time series designs or randomisation of several communities are other alternatives.
We followed in Loevinsohn's footsteps and conducted a search of published English-languageliterature to determine: • the potential effectiveness of the various approaches to environmental health promotion; • the appropriate expectations and targets for changes in health behaviour.
We concentrated on published literature for two reasons: • we assumed published studies would be presented more rigorously; and • we wanted to review information which is widely available - practitioners must be able to access the studies if the findings are to be of any use.
Articles were included for review if they dealt predominantly with a health education or promotionintervention in a developing country and if an evaluation of the educational manoeuvre wascarried out - articles which just describe an educational intervention or discuss the theory ofhealth education are left out (1).
Loevinsohn reviewed articles up to 1987. This search was from 1987 to the present. The IRC andWELL libraries were consulted, as were the authors' personal collections. We used the followingcomputerised databases: • Medline, Popline, BIDS, and the IRC database.
and searched under the following keywords: • education, promotion, developing country, diarrhoea, HIV / AIDS, nutrition, malaria, The search results were put into an Endnote database and are available at the WELL ResourceCentre.
The computerised searches identified 242 articles of interest. Two reviewers assessed these bytitle for the characteristics described above and 31 articles were selected for review (7-38).
Of the 31 articles reviewed, 35% of the articles provided enough description to duplicate theproject.
60% of the articles included a control group of which 16 (53%) were controlled studies withsample sizes greater than 'two' clusters or 60 individuals.
Seven articles described truly randomised studies (8, 9, 20, 21, 25, 28, 34) and four othersdescribed quasi-randomised studies (7, 18, 22, 36).
Most of the articles examined either health status (37%) or health behaviour (83%) as end-points.
There was some overlap as 8 articles (26%) reviewed both.
None of the 31 articles reviewed contained all of the positive methodological attributes describedabove. The five most rigorous studies had 60% or more of the characteristics named byLoevinsohn in Box 2 (10, 12, 17, 20, 34). These are examined in detail here. A further 11 studieswere sufficiently rigorous to contribute some further evidence to this review. Brief summaries ofthe salient points of all of the papers are included in annex 1. Three review articles were alsolocated and lessons extracted (39-41).
• Adequate description of how the strategy was adapted to local conditions.
• Example given of the materials or process employed.
• Adequate description of resources required to carry out programme.
• Measured outcome before and after the intervention.
• Period between the education and outcome more than one year.
• Evidence of community participation in design of programme, goals or outcome • Article claimed to show positive results for the intervention evaluated.
• Included discussions of possible biases or caveats.
• Included p-values or confidence intervals.
Box 2: Criteria for assessing the articles on health education or promotion interventions
Studies exhibiting desirable methodological characteristics
The desirable characteristics of a study as proposed by Loevinsohn (Box 2) are used as criteriato evaluate the five best studies.
3.1.1
Study based on explicit theory
Leovinsohn suggests that interventions are too often based only on epidemiological evidenceabout disease patterns. However, to change behaviour detailed specific research and planning isneeded. Four of the five studies reported doing this: Box 3: Five-stage process used to develop and implement education intervention
Stage 1. Formative Research.
Stage 2. Developing recommendations for behaviour change.
Stage 3. Development of educational messages.
Stage 4. Development and production of educational materials.
Stage 5. Distribution of educational materials.
• Lloyd (10) describes a five-stage process used to develop mosquito control in Mexico • Project Salsa, based on the USA/Mexican border, was organised around a seven-step sequence of planning, intervention and evaluation procedures represented by theacronym ONPRIME: organising, needs/resources assessment, priority-setting, research,monitoring and evaluation (17).
• A diarrhoea education project in Zaïre was based on keeping behaviour change messages to a minimum. This study concentrated on four behaviours (34).
• Pant (20) quotes the Alma Ata declaration suggesting that nutrition education is the most appropriate way to improve the nutritional status of at-risk populations, however nospecific behaviour change planning was reported for this study in Nepal.
• The health education intervention in Tayeh's study was carried out in close collaboration with Ghana's National Guinea Worm Eradication Programme. Health education and filterdistribution were carried out in the manner of the larger programme, while the healtheducation methods were based on assessing people's knowledge, attitudes andpractices (12).
3.1.2
Adapting the intervention
According to Loevinsohn, an effective intervention must adapt strategies to local conditions.
Three of the five studies described this process: • The first stage in the Aedes aegypti control programme was formative (10). This included qualitative and quantitative research techniques and was critical to the project. It tried toensure that all relevant issues, for both the community and the population, wereaddressed. Project staff could then prioritise behavioural factors known to affect Aedesaegypti in a context relevant to the local community.
• Project Salsa's funders allowed the community to determine which issues were perceived to be important. Project Salsa formed an advisory council of communityrepresentatives affiliated with organisations which could contribute resources andlegitimise the interventions.
• Baseline data for Project Salsa were collected from local health departments and health service agencies; other methods included a telephone survey, home interviews, interceptsurveys, key informant surveys and an observational study for point-of-purchaseassessments (17).
• Staff at the education intervention in Zaïre developed their messages using data from a comparative analysis of diarrhoeal rates established from a community survey, and fromethnographic data and observed hygiene practices. Trainers worked with communityvolunteers and planned non-formal lessons based on culture-specific experience and tobe delivered using songs, stories etc (34).
• The description of the vitamin A supplementation does not indicate that the strategy was • The report of the Ghanaian project (12) states that it worked along the lines of Ghana's National Guinea Worm Eradication Programme, but there is no description of how theeducation strategy was adapted to the local conditions.
3.1.3
Example given of the materials or process employed
• Lloyd (10) describes each of the stages in Box 3, laying particular emphasis on stage two, ie developing the recommendations for behaviour change.
• Elder provides a number of tables detailing the segments of the population at which Project Salsa was aimed and the interventions for each segment (See Box 4) (17).
Box 4: Excerpt from Project Salsa interventions
Intervention
Timetable
Incentives
Facilitators
Attendance
Monitoring
selected
participant
s
• Haggerty (34) refers to the training of the community volunteers and how they were instructed to teach by way of analogy, song and poetry.
• Pant (20) gives no examples of the education process or the materials used.
• In Ghana interviewers were trained to demonstrate how cloth filters should be placed on the household's traditional clay pots and to explain that this would filter out the sediments.
The live cyclopoids were shown moving on the filter. This was reported to be aconvincing demonstration of the need for filters (12).
3.1.4
Adequate description of resources required to carry out programme
• In the Aedes control programme, no details were given of the resources required to visit each household three times. Photo novellas were expensive but thought to be effective.
• Project Salsa was funded over five years by the Henry J. Kaiser Family Foundation and affiliated donors. Resource utilisation is not described (17).
• Haggerty (34) gives no indication of the resources required for the programme in Zaïre.
Resources needed to cover 18 separate rural sites must have been considerable.
• Costs were calculated in the Nepal programme. The education was less cost-effective than the capsule distribution programme not because it was less effective but because itwas more expensive. The cost per death avoided with capsules was $73 and witheducation $237 (20).
• The cost of the Guinea worm filters was about 150 cedis (then equivalent to US$0.45) in accordance with the National Eradication Programme. The authors comment that theproject was very expensive as local and qualified interviewers and health educatorsvisited each household to demonstrate the benefits of filtering drinking water; 'Suchconcentrated health education can only be afforded by a research project' (12).
3.1.5
Measured outcome variable before and after the intervention
• In the Aedes project knowledge, belief and practice (KBP) and entomological surveys were conducted before the educational intervention in both intervention and controlcommunities. Entomological surveys were conducted in both communities after theeducational intervention (10).
• Though approximately four months was spent collecting baseline information for Project Salsa, most of the evaluation entailed monitoring participation rates and other variablesappropriate for community demonstration projects. The overall impact evaluation wasconducted by an external contractor. However few projects continued to the date of theevaluation, so only a few outcome measures were available (17).
• The Zaïre project (34) conducted a baseline survey of diarrhoeal morbidity of 2,082 children aged 3-35 months. This was collected between October and December 1987 in18 geographically separate sites. 300 randomly sampled sentinel families were visitedonce for prolonged (up to 7 hours) structured observations of child feeding and hygienepractices thought to be related to diarrhoea. The results for behaviour change pre- andpost intervention were published as a book chapter (47). • The vitamin A supplementation programme used a two-stage area probability sample of more than 100 area units in seven of Nepal's seventy-five rural districts. The study lookedat the cost of the programme and its impact on health status. Annual censuses collectedinformation on community and household variables while physical and ocularexaminations were carried out on all children under 10 years old (20).
• The Guinea worm programme conducted a baseline survey in the dry season (January to June) of 1990 (12). The intervention was implemented in stages in different villages afterthe initial information was collected. The same field workers made 2 more visits to eachhousehold at two-monthly intervals. The results were combined to give a six-monthperiod prevalence. An identical survey was carried out over the same period in 1991 toassess the impact of the intervention.
3.1.6
Period between education and outcome more than one year
Only Project Salsa, the vitamin A and the Guinea worm study were evaluated more than a yearafter the intervention.
3.1.7
Evidence of community participation in design of programme, goals or
outcome measures
Community participation, though desirable, does not mean a project will be sustainable. Three ofthe five studies mention participation, though only Project Salsa made this an explicit goal.
3.1.8
Article claimed to show positive results for intervention evaluated
Each of the five most rigorous articles claims a positive result (see Box 5). Project Salsaadmitted that the results were disappointing with only one significant change in one healthindicator in a tiny proportion of the population. A more methodologically sound approach is todefine outcome measures at the start and report on this basis. Casting about for encouragingresults can throw up "significant" results which are statistical artifacts. The vitamin A project andthe diarrhoea intervention showed small but significant health impacts without demonstratingbehaviour change. The Guinea Worm programme showed large numbers of filters bought butdid not demonstrate use. The Aedes programme's results suggested some behaviour change butwere not conclusive.
3.1.9
Included discussion of possible biases or caveats
Of the five most rigorous studies only the diarrhoea and the Guinea Worm study discussed thelimitations of their results and both of these look for explanations of how the impact may havebeen underestimated. No studies considered the possibility that change was overestimated.
None evaluates the null hypothesis: that there was no effect.
3.1.10 Included p-values or confidence intervals
The p-value is the probability value that a test statistic would be as extreme as, or more extremethan, observed if the null hypothesis were true. A study result whose probability value is less than5% (p<0.05) is considered sufficiently unlikely to have occurred by chance to justify being called'statistically significant' (42). Very few of the articles reported the exact probability value butreported p as being less than 0.05 (see Box 5).
Evidence of behaviour change
Box 5 lists the outcomes of the studies. Only two, the studies on Guinea worm and diarrhoea,
show a clear and unequivocal impact on the behaviour of target populations.
• The targeted behaviour change in the Lloyd paper (10) was elimination of breeding sites for Aedes. Six months after the intervention, the number of household containers foundpositive for larvae had increased significantly in the comparison group and remained thesame in the intervention community. We do not learn if the total number of bottles andjars was affected.
• Project Salsa (17) had a wide target of behaviours to change, both in the community and in institutions. There were no measures of actual health-related behaviour saveattendance at heart disease risk-factor screenings, which was the only interventionmaintained over the long term.
• Data on the impact of the intervention in Zaïre (34) on behaviour was published elsewhere (47). Structured observation, before and three months after the start of theintervention, indicated that the intervention was responsible for about a 10% reduction in'unhygienic' behaviour, after allowing for the fact that some improvement was also foundin the behaviour of the control families.
• In Nepal both interventions (food education and capsule distribution) were effective in improving measures of vitamin A status (20). 80% of the target population receivedcapsules. Knowledge about foods increased dramatically in both groups. The authorssuggest there was doubt about mother's food behaviour change because informationwas not received or understood, or was inappropriate because mothers did not have thecapacity to act on it in some cases. It is, nevertheless, hard to see how the programme Box 5 Articles found to have at least 60% of Loevinsohn's characteristics
Location
The total number of bottles and jars (p<0.01), the number of disposable bottles and jars (p<0.02) and the number of animal dishes (p<0.02) which were positive for Aedes aegypti larvae were significantly lower in intervention groups when compared withthe comparison groups.
Project Salsa never became greater than the sum of its parts as most elements were not maintained beyond the duration of external funding. The heart health and screening, counseling and referral activities achieved a reduction in cholesterol among high risk group (F{1,423} = 4.22 p<0.05).
Children in intervention communities experienced an 11% reduction in the risk of reporting diarrhoea during the peak season (p<0.025). The largest differences were seen among children aged 24-35 months. 'Unhygienic' behaviour was reduced by10%.
At 24 months after the implementation of the project the reduction of risk for xerophthalmia was greater among children whose mothers were able to identify vitamin-A rich foods (RR =0.25, 95%CI = 0.1-0.62) than among those who received mega-dose capsules (RR =0.59, 95%CI = 0.41-0.84). Therisk of mortality at 2 years was reduced for both thenutrition education (RR =0.64, 95%CI = 0.48-0.86)and capsule distribution (RR =0.57, 95%CI = 0.42-0.77) cohorts. There were doubts about theadoption of food related behaviours.
The impact of filtering was less than expected 56% of the study households bought 802 filters; but only 37% bought at least one filter for every 10 A logistic regression model with demographic and behavioural factors showed a significantly reduced risk of infection (OR=0.80) among householdmembers who bought at least one filter for10household members in comparison with thosewithout filters.
can have succeeded in improving nutritional status without having had some impact onfood behaviour.
• Tayeh's paper (12) focuses on two behaviours: the decision to buy a filter and the decision to use one. 56% of households were persuaded to buy one. Asking people ifthey used the filter was not a good measure of use, but no better measure was found.
Health improvement in the earlier intervention villages was presumably due to correctuse of the filters.
Discussion
Not one of the five studies that we identified as methodologically sound shows a clear impact onbehaviour. Do the other studies contribute anything further? Of the 11 with reasonable rigour,only two showed a clear and major impact on behaviour: • A small scale hand-washing intervention, where soap and detailed instructions on hand- washing were given at two weekly intervals for four months in one village in Indonesia,showed major and sustainable improvements in behaviour (37,38).
• Groth-Marnat et al (29) describe how a village in Fiji decided to stop smoking. As smoking levels rose due to increased marketing, the village took the decision to becomea non-smoking village. Eventually all persons in the village who smoked were able togive up smoking, with specific exceptions (elders, and a visitor) and became nationallyknown as the village that gave up smoking. Follow up evaluation at 9 and 21 monthsindicated sustained success. The authors describe how the social cohesion of the villageallowed the enforcement of a communal decision which made smoking tabu and morallyreprehensible. This was not a conventional study with measured outcomes, but there isno reason to doubt the dramatic impact of the villagers' decision.
Several other papers suggest good results: • Reported annual sales of ORS packets went from 7.6m to 80 million per year between 1984 and 1993 in Mexico, suggesting that a major behavioural change in diarrhoeamanagement had taken place (43). However, no detailed description of the way in whichthis was achieved was offered. There have undoubtedly been other comparable majorsuccesses in national health promotion programmes. However, few of these aredocumented in scientific journals.
• In Tanzania large numbers of bed nets were sold in 6 villages, but the reasons why they • Kroeger suggested that education had improved reported chloroquine utilisation despite confusing local treatment regimes in two out of three Latin American countries with amalaria education programme (14).
All of the other papers with any rigour showed little or no impact on behaviour. In many of thestudies the impression is given that authors search widely for any positive result to report. If thisis true, then at least some of the studies are likely to be reporting positive findings which are infact a result of statistical artifact. On average about 5% of reported results will be "significant" bychance alone. By looking at the results of published studies we probably also under-reportstudies with a negative result. Studies which conclude that interventions had a negligible or nileffect on behaviour are much less likely to be written up and, if written, to be published.
The review papers are hardly more encouraging: • Kloos (40) reviewed Schistosomiasis control efforts. Projects in Egypt, Ethiopia and Zimbabwe found that increasing people's knowledge about the parasite and how thedisease is transmitted did not cause people to reduce their contact with water. Intensivehealth education in St. Lucia led to a 92% reduction of water contact, but this effect wasnot sustained.
• A review of breastfeeding interventions (41) had several successes to report. Most effective was the institution of 'baby-friendly' hospitals and changing the behaviour ofinstitutional players. One hospital achieved an 80% decrease in consumption of formula.
Health education increased the duration of exclusive breastfeeding in a number of studies, especially where interventions were frequent. Social action and legislation alsohad important and measurable impacts on reported breastfeeding rates. Marketing ofbreast milk substitutes had a major impact on mothers' behaviour, unfortunatelydetrimental to the child's health.
• Ebrahaim et al (44) reviewed a series of coronary heart disease RCT interventions in the UK; these tackled multiple risk factors and included counselling, education and drugtreatments. They found poor or non-existent impact results which, they were satisfied,were not due to failures in measurement. They concluded that health protection throughnational fiscal and legislative changes should have a higher priority than health promotionapplied to general and workforce populations.
"The current concepts and practices of multiple risk factor intervention,primarily through individual risk factor counselling, must not be exported topoorer countries as the best policy option for dealing with existing andprojected burdens of cardiovascular disease. Health protection shouldbe promoted as the mainstay of preventing chronic diseases in poorercountries." (44, p1672).
If marketing has been successful in promoting smoking and formula feeding there is also somehistorical evidence that it has had a major positive impact on hygiene in the West. Soapmanufacturers have promoted and made soap easily accessible, partly with health messages,but mostly though emotional appeal (45). Social marketing taps into the techniques ofcommercial marketing to encourage behaviour change. Has it been more successful? • Social marketers claim excellent results for their efforts but only one scientific evaluation appeared in this review. The study showed no net increase in contraceptive prevalencethough the branded 'pill' did increase market share.
Marketing specialists do not, however, expect miracles, in the way some health educators seemto. Take-off curves for some consumer durables in the US shows how behaviour change can belong and slow. For example few colour TVs were sold between their introduction in 1954 and1962, when sales went over the thousand and climbed steeply from there. Answering machinestook four years to take off, as did electric shavers. Blenders took about 14 years for sales to goabove 1000 per year (46). One reason why results have been poor for most of the reportedstudies of health promotion may simply be because most interventions are too short andevaluations too early to show a clear impact.
Conclusions
Of the studies published since Loevinsohn's review in 1987, both the five best and the 11 nextbest studies of the impact of health promotion in developing countries reveal a depressingpicture of very little behaviour change for a lot of invested effort.
Whilst almost all authors save Ebrahaim, remain convinced of the value of health promotion,there do seem to be legitimate grounds for doubt. The possibility that health promotion/educationis ineffective is the null hypothesis, and should not be ruled out in future studies. However, giventhat a few interventions have demonstrated major behaviour change, there is still room foroptimism.
Changing behaviour requires a chain of events. If the chain fails at any one point then the resultswill be nil. This may be why results look so disappointing. To resolve the issue both interventionsand evaluations need to be better designed.
• Explicit design of behaviour change interventions involving formative research to develop feasible and practical replacement behaviour is needed.
• Messages should concentrate on a limited number of simple cheap and feasible • When the economic, environmental, institutional or political context is unfavourable, behaviour change may become impossible. In many cases it may be more sensible toconcentrate on removing the constraints to behaviour change before attempting topromote behaviour change directly.
• Explicit, measurable and predefined behaviour change outcomes should be set from the outset. Health outcomes can be expensive to measure, complicated to analyse and notvery useful in guiding decision making.
• Behaviour change may be slow and may require intensive interventions with many contacts. Results may take three or more years to show.
• Piloting and gradual introduction of interventions with constant monitoring helps to avoid major investments in ineffective activities.
• Marketing may have some useful techniques to teach but social marketing has yet to demonstrate its effectiveness and cost-effectiveness.
Planners and policy makers need to take into account the following points: • Cost-effectiveness data on behaviour change programmes are urgently needed, but it is only worthwhile doing such evaluations on programmes with evidence of successfuloutcomes, and there are few of these.
• If results are discouraging in small-scale intensive research-style interventions, they may be even more discouraging in the 'real world' of daily health service difficulties.
• If interventions that get good results are deemed 'expensive' in a research setting, then they may be unaffordable on a large scale in the 'real world'.
• All claims to success in effecting behaviour change should be treated with healthy References
Loevinsohn, B.P. Health education interventions in developing countries: a
methodological review of published articles. International Journal of Epidemiology. 19 (4):
788-794 (1990).
Green, L.W. & Kreuter, M.W. Health promotion planning: an educational and
environmental approach. Mountain View, CA, Mayfield, 1991.
Cairncross, S. Health impacts in developing countries: new evidence and new
prospects. Journal of the Institution of Water and Environmental Management. 4 (6): 571-
577 (1990).
Curtis, V. et al. Structured observations of hygiene behaviour in Burkina Faso; validity,
variability and utility. Bulletin of WHO. 71 (1): 23-32 (1993).
Kirkwood, B. & Morrow, R. Community-based intervention trials. Journal of Biosocial
Science. 10 S: 79-86 (1989).
Blum, D. & Feachem, R.G. Measuring the impact of water supply and sanitation
investments on diarroela disease: problems of methodology. International Journal of
Epidemiology. 12: 357-65 (1983).
Brown, L.V. et al. Evaluation of the impact of weaning food messages on infant feeding
practices and child growth in rural Bangladesh. American Journal of Clinical Nutrition. 56:
994-1003 (1992).
Neyzi, O. et al. An educational intervention on promotion of breast feeding. Paediatric
and Perinatal Epidemiology. 5: 286-298 (1991).
Shoham-Yakubovich, I. et al. Infant feeding practices: an evaluation of the impact of a
health education course. American Journal of Public Health. 80 (6): 732-734 (1990).
Lloyd, L.S. et al. The design of a community-based health education intervention for the
control of Aedes aegypti. American Journal of Tropical Medicine and Hygiene. 50 (4):
401-11 (1994).
Sarti, E. et al. Development and evaluation of a health education intervention against
Taenia solium in a rural community in Mexico. American Journal of Tropical Medicine and
Hygiene. 56 (2): 127-32 (1997).
Tayeh, A. et al. The impact of health education to promote cloth filters on dracunculiasis
prevalence in the northern region, Ghana. Social Science and Medicine. 43 (8): 1205-11
(1996).
Ekeh, H.E. & Adeniyi, J.D. Health education strategies for tropical disease control in
school children. Journal of tropical medicine and hygiene. 91: 55-59 (1988).
Kroeger, A. et al. Health education for community-based malaria control: an intervention
study in Ecuador, Colombia and Nicaragua. Tropical Medicine and International Health. 1
(6): 836-46 (1996).
Makemba, A. et al. Community-Based Sale, Distribution and Insecticide Impregnation of
Mosquito Nets in Bagamoyo District, Tanzania. Health Policy And Planning. 10 (1): 50-59
(1995).
Marsh, V.M. et al. Evaluating the community education programme of an insecticide-
treated bed net trial on the Kenyan coast. Health Policy and Planning. 11 (3): 280-91
(1996).
Elder, J. et al. Project Salsa- Development and institutionalization of a nutritional health
promotion project in a Latino community. American Journal of Health Promotion. 12 (6):
391-401 (1998).
Hussain, A. & Kvale, G. Sustainability of a nutrition education programme to prevent
night-blindness in Bangladesh. Tropical Medicine and International Health. 1 (1): 43-51
(1996).
Hussain, A. et al. Impact of a health education program to promote the consumption of
vitamin A rich foods in Bangladesh. Health Promotion International. 12 (2): 103-109
(1997).
References (cont./)
20.
Pant, C.R. et al. Impact of nutrition education and mega-dose vitamin A supplementation
on the health of children in Nepal. Bulletin of the World Health Organization. 74 (5): 533-
45 (1996).
Ruel, M.T. & Habicht, J.P. Growth charts only marginally improved maternal learning
from nutrition education and growth monitoring in Lesotho. Journal of Nutrition. 122 (9):
1772-80 (1992).
Bolam, A. et al. The effects of postnatal health education for mothers on infant care and
family planning practices in Nepal: a randomised controlled trial. British Medical Journal.
316 (7134): 805-11 (1998).
Janowitz, B. et al. Impact of social marketing on contraceptive prevalence and cost in
Honduras. Stud Fam Plan. 23 (2): 110-7 (1992).
Kandeh, H.B. et al. Community motivators promote use of emergency obstetric services
in rural Sierra Leone. International Journal of Gynecology And Obstetrics. 59 (2): S209-18
(1997).
Mbizvo, M.T. et al. Effects of a randomized health education intervention on aspects of
reproductive health knowledge and reported behaviour among adolescents in Zimbabwe.
Social Science and Medicine. 44 (5): 573-7 (1997).
Olaniran, N. et al. Mobilizing the community to utilize obstetric services, Cross River
State, Nigeria. International Journal of Gynecology and Obstetrics. 59 (2): S181-9 (1997).
Opoku, S.A. et al. Community education to improve utilization of emergency obstetric
services in Ghana. International Journal of Gynecology and Obstetrics. 59 (2): S201-7
(1997).
Pauw, J. et al. A controlled HIV / AIDS-related health education programme in Managua,
Nicaragua. AIDS. 10 (5): 537-44 (1996).
Groth-Marnat, G. et al. Tobacco control in a traditional Fijian village: indigenous
methods of smoking cessation and relapse prevention. Social Science and Medicine. 43
(4): 473-7 (1996).
Alam, N. et al. Mothers' personal and domestic hygiene and diarrhoea incidence in
young children in rural Bangladesh. International Journal of Epidemiology. 18 (1): 242-247
(1989).
Bhattacharya, R. et al. Impact of education in the knowledge and practices of rural
mothers and key family members on diarrhoea and its treatment at home. Journal Of
Diarrhoeal Diseases Research. 6 (1): 15-20 (1988).
Bilqis, A.H. et al. Promoting sanitation in Bangladesh. World Health Forum. 15: 358-362
(1994).
Blum, D. et al. The Imo State (Nigeria) drinking water supply and sanitation project. 2 :
Impact on dracunculiasis, diarrhoea, and nutritional status. Transactions of the Royal
Society of Tropical Medicine and Hygiene. 84: 316-321 (1990).
Haggerty, P.A. et al. Community-based hygiene education to reduce diarrhoeal disease
in rural Zaire: impact of the intervention on diarrhoeal morbidity. International Journal of
Epidemiology. 23 (5): 1050-9 (1994).
Huttly, S.R.A. et al. The Imo State (Nigeria) drinking water supply and sanitation project.
1 : Description of the project, evaluation methods, and impact on intervening variables.
Transactions of the Royal Society of Tropical Medicine and Hygiene. 84: 309-315. (1990).
Peterson, E.A. et al. The effect of soap distribution on diarrhoea: Nyamithuthu Refugee
Camp. International Journal of Epidemiology. 27: 520-524 (1998).
Wilson, J.M. et al. Hand-washing reduces diarrhoea episodes: a study in Lombok,
Indonesia. Transactions of The Royal Society of Tropical Medicine And Hygiene. 85: 819-
821 (1991).
Wilson, J.M. & Chandler, G.N. Sustained improvements in hygiene behaviour amongst
village women in Lombok, Indonesia. Transactions Of The Royal Society Of Tropical
Medicine And Hygiene. 87 (6): 615-6 (1993).
Esrey, S.A. et al. Effects of improved water supply and sanitation on ascariasis,
diarrhoea, dracunculiasis, hookworm infection, schistosomiasis, and trachoma. Bulletin of
the World Health Organization. 69 (5): 609-621 (1991).
Kloos, H. Human behavior, health education and schistosomiasis control: a review.
Social Science snd Medicine. 40 (11): 1497-511 (1995).
Wilmoth, T.A. & Elder, J.P. An assessment of research on breastfeeding promotion
strategies in developing countries. Social Science and Medicine. 41 (4): 579-94 (1995).
Last, J.M. A dictionary of epidemiology, Oxford University Press, 1988.
Gutiérrez, G. et al. Impact of oral rehydration and selected public health interventions on
reduction of mortality from childhood diarrhoeal diseases in Mexico. Bulletin of the World
Health Organization. 74 (2): 189-97 (1996).
Ebrahaim, S. & Davey-Smith, G. Systematic review of randomised controlled trials of
multiple risk factor interventions for preventing coronary heart disease. British Medical
Journal. 314: 1666-1674 (1997).
Vinikas, V. Soft soap, hard sell: American Hygiene in an age of advertisement. Ames,
Iowa, Iowa State University Press, 1992.
Golder, P. & Tellis, G. Will it ever fly? Modeling the takeoff of really new consumer
durables. Marketing Science. 16 (3): 256-270 (1997).
Haggerty PA: Community-based education to reduce diarrhoeal disease in rural Zaïre.
In: Cairncross and Kochar VJ eds. Studying Hygiene Behaviour; Methods, Issues and
Experiences. New Delhi: Sage Publications, 1994 pp 160-171.
Annex 1: Summary of published articles consulted
Interventions
Comments
Behaviour change ascribed
evidence
to project
evidence
project Hard to find good measureof 'use' of filter Annex 1: Summary of published articles consulted (cont./)
Interventions
Comments
Behaviour change ascribed
evidence
to project
evidence
contraceptives, but marketshare increased.
management major issues. Highcosts of such programmes notaddressed.
smell had more impact thanhealth messages.
disruption. 'Not cost-effective'.
Annex 1: Summary of published articles consulted (cont./)
Interventions
Comments
Behaviour change ascribed
evidence
to project
evidence
educationwith/out growthchartsReview of RCTs Annex 1: Summary of published articles consulted (cont./)
Interventions
Comments
Behaviour change ascribed
evidence
to project
evidence
but returned to pre-intervention levels after end.

Source: http://www.lut.ac.uk/well/resources/well-studies/full-reports-pdf/task0165.pdf