Community Dent Oral Epidemiol 1999; 27: 124–9 Copyright C Munksgaard 1999Printed in Denmark . All rights reserved

ISSN 0301-5661

Simon Dixon and Phil ShackleySheffield Health Economics Group, School ofEstimating the benefits ofHealth and Related Research, University ofSheffield, Sheffield, UK

community water fluoridationusing the willingness-to-paytechnique: results of a pilot studyDixon S, Shackley P: Estimating the benefits of community water fluoridationusing the willingness-to-pay technique: results of a pilot study. Community DentOral Epidemiol 1999; 27: 124–9. C Munksgaard, 1999

Abstract – Objectives: To estimate the health and non-health effects of fluoridationon well-being using the willingness-to-pay technique. Methods: A pilot study of100 people was undertaken in which the willingness-to-pay technique was usedto estimate the welfare effects of community water fluoridation in the city of Shef-field, in the United Kingdom. In view of the pilot nature of the study, it wasdecided to use quota sampling based on age and sex rather than random sampling.Reasons for respondents’ answers and their previous knowledge of fluoridationwere also noted. Results: Sixty-two percent of respondents were in favour of fluo-ridation, 31% were opposed, and 7% were unsure. Thirty-one respondents re-fused to answer the willingness-to-pay questions. The mean willingness to pay forthose in favour of fluoridation was less than that for those opposed to fluorida-

Key words: cost-benefit analysis; fluoridation;tion. The main reasons given for opposing fluoridation were its impact on freedompublic opinion; willingness to payof choice and the desire for pure water. Nineteen percent of the sample heldSimon Dixon, School of Health and Relatedincorrect information on the effects of fluoridation. Conclusions: The study il-Research, University of Sheffield, Regentlustrates how a simple referendum can miss important information regarding theCourt, 30 Regent Street, Sheffield, S1 4DA,

intensity of individuals’ preferences. An important aspect of the willingness-to- UKpay technique is its ability to identify ‘losers’ and quantify their welfare loss. Tel: π44 (0) 114 222 0724

Fax: π44 (0) 114 272 4095This issue has been ignored by all previous economic evaluations of fluoridation.E-mail: s.dixon/sheffield.ac.ukFurther investigation is required to replicate this work on a larger sample of

individuals, and to develop the willingness-to-pay technique. Accepted for publication 17 November 1998

Health care resources are scarce, and therefore notall health needs can be met. If society wishes tomake informed decisions about the allocation of re-sources between different health interventions thenthe use of economic evaluations is essential (1).Only by evaluating the costs and benefits of vari-ous interventions will society be able to assess theimpact on overall welfare (or well-being) offollowing one course of action rather than another.

It is of course essential that the information pro-vided by the relevant economic evaluations is val-id, otherwise assessment of changes in welfare willbe erroneous. Thus critical appraisal of economic

124

evaluations is as important as critical appraisal ofclinical evaluations. Despite the availability ofwell-established methods of economic evaluation(1), many published studies are of poor quality. Areview of economic evaluations of communitywater fluoridation (2) identified several problemswith the studies it identified (3–10). In this paper,we highlight the problems associated with compre-hensively assessing the effects of water fluoridationon welfare, and present the results of a contingentvaluation (CV) study to assess changes in welfareassociated with community water fluoridation.

Previous economic evaluations of fluoridation

Willingness to pay for community water fluoridation

have been of two types: cost-effectiveness analyses(CEA) and cost-benefit analyses (CBA). These twomethodologies require outcomes to be assessed indifferent ways and, more importantly, are designedto answer different ‘economic questions’.

Welfare in CEA is measured by a single naturalunit, e.g., the number of diseased, missing or filledsurfaces (DMFS). This information is then com-bined with the costs of the programme to producea cost-effectiveness ratio, i.e., cost per DMFS avoid-ed. Such ratios can then be used to identify theleast costly way to achieve an improvement inhealth outcomes, e.g., DMFS avoided (or converse-ly, the maximum number of DMFS avoided for agiven cost). Cost-effectiveness ratios do not, how-ever, allow one to assess whether the additionalhealth improvements are worth the additional ex-penditure. Only a CBA can achieve this.

Cost-benefit analysis measures welfare changesin monetary terms. This allows several dimensionsof well-being to be assessed simultaneously, in-cluding possible non-health effects. This informa-tion can be combined with the costs of the projectto produce an estimate of the overall effect of theprogramme on welfare in monetary terms. Such acalculation allows a direct assessment of whetheror not the expenditure is worthwhile.

These characteristics of CBA are important whenevaluating community water fluoridation. Firstly,there are several dimensions of health that are notcaptured by cost-effectiveness analyses, e.g., mor-bidity associated with tooth decay and its treat-ment. Furthermore, CEA is unable to address pos-sible non-health impacts on welfare. This omissionis potentially important as fluoridation affects indi-viduals’ freedom of choice (11, 12), and mayincrease welfare by reducing inequalities or gener-ating a sense of ‘community spirit’ or altruism. Sec-ondly, the primary economic question with regardto implementing water fluoridation is whether it isworthwhile, i.e., do the effects on welfare outweighthe costs of the programme. This is quite differentfrom the ‘health maximisation question’ that CEAattempts to answer.

It should also be noted that the cost-benefit ana-lyses published to date (3–8) have failed to mea-sure all the effects of fluoridation. The reason forthis is that they have been content with estimatingthe savings associated with fluoridation in the formof a reduced number of dental interventions suchas fillings and extractions. This is not a satisfactorymeasure of welfare.

Contingent valuation is a technique that meas-

125

ures welfare effects in monetary terms. As ex-plained above, this allows a broad range of effectsto be measured including health and non-healthbenefits. It has had several successful applicationsin other areas of medicine, and a recent reviewidentified 48 contingent valuation studies that hadbeen reported (13).

Contingent valuation is based on the premisethat the maximum amount of money a particularindividual is willing to pay for a commodity is anindicator of its value to that individual. Conse-quently, CV is frequently referred to as the willing-ness-to-pay (WTP) technique. By expressing theirpreferences for fluoridation in terms of an amountthey are willing to pay, respondents are not re-stricted to a ‘narrow view’ of the effects of fluori-dation. Morbidity, aesthetics, altruism and freedomof choice are all aspects of fluoridation that can bevalued by individuals and incorporated into an in-dividual’s WTP.

It is of course possible that some individuals willnot want fluoridation, and its implementation willreduce their welfare. Contingent valuation allowstheir preferences to be estimated in a similar wayto those in favour of fluoridation. This explicit re-cognition that some individuals are opposed tofluoridation is important, since such welfare effectsare excluded in other forms of evaluation. Accord-ing to economic theory, those who will lose welfarefollowing fluoridation should have this effectmeasured by the minimum amount of compensa-tion they would be willing to accept following itsintroduction. However, this willingness-to-accept(WTA) format has produced several problems inpractice, and it has been suggested that welfareloss can sometimes be measured using individuals’WTP to prevent a programme’s introduction (14).In order to investigate this issue, the study de-scribed below uses both techniques to assess thewelfare loss associated with fluoridation. It shouldbe noted that the principal focus of this study isto assess welfare effects and thus it stops short ofincorporating the results with other cost informa-tion to produce a full cost-benefit analysis.

Methods

A questionnaire was developed for use in face-to-face interviews and designed to take less than 20minutes to administer. In view of the pilot natureof the study, it was decided to use quota samplingbased on age and sex rather than random sam-pling. One hundred interviews were carried out

Dixon & Shackley

across four areas of the city of Sheffield in theUnited Kingdom, each area having different socio-demographic characteristics. The data were collect-ed during the first 2 weeks of December 1997.

The interview began with some general intro-ductory information. The concept of contingentvaluation was explained to respondents and it wasemphasised that the valuation exercise was hypo-thetical and that there was no question of respon-dents actually being asked to pay. Respondentswere then given a description of fluoridation andits effects (see Appendix) and asked if they wouldbe in favour of fluoride being added to drinkingwater in Sheffield. Those respondents who were infavour were then asked if they would be willing tocontribute anything in extra taxation for fluorida-tion to go ahead. If respondents were not willingto pay, they were asked to state why. Those whoindicated a WTP were then asked to state the maxi-mum amount they would be willing to contributeeach year for fluoridation to go ahead. Respon-dents were asked to indicate their maximum WTPvalue on a payment card (15), and state why theywere willing to contribute.

If respondents indicated that they were not in fa-vour of fluoridation, they were asked questions toestablish the value of any loss in welfare to themfrom fluoridation going ahead. As indicated al-ready, the value of this welfare loss can be meas-ured in two ways, and in order to investigate pos-sible differences in the values given, half the ques-tionnaires contained a section in whichrespondents were asked ‘WTA compensation’questions while the other half contained ‘WTP toprevent’ questions. In the WTA version, respon-dents were asked if they would be willing to acceptan annual tax rebate as compensation for fluorida-tion going ahead. In the WTP version, respondentswere told to imagine that it would be possible fora device to be fitted to their water supply whichwould remove the fluoride from their drinkingwater and then asked if they would be willing tocontribute anything in extra annual taxation tohave the device fitted and maintained. These sec-tions were structured in the same way as the sec-tion answered by those who were in favour of flu-oridation, and the same payment card was used.

At the end of the interview, all respondents wereasked several socio-demographic questions to-gether with some simple questions about their useof dental services. Finally, the respondents wereasked whether they had been aware of fluoridationprior to the interview and whether they knew of

126

any effects of fluoridation not mentioned in the de-scription used.

Results

The response rate was 66%. Of the 100 respon-dents, 62 were in favour of water fluoridation, 31were opposed, two were indifferent and five wereunsure. The socio-economic breakdown of the sam-ple together with their utilisation of dental servicesare shown in Table 1.

Of the 93 respondents who had a clear preference

Table 1. Socio-economic characteristics of the sample

%*

GenderMale 53Female 47

Age18–29 2630–49 2550–69 3470 and over 14Unknown 1

Education‘A’ level and higher 27Below ‘A’ level 63

IncomeLess than GBP 6999 34GBP 7000–GBP 10 999 24GBP 11 000–GBP 16 999 13GBP 17 000–GBP 26 000 7More than GBP 26 000 11Unknown 11

Employment statusPaid employment 35Housework 11Unemployed 16Student 6Retired/pensioner 32

Number of dependent children0 781 62 133 2Unknown 1

Currently registered with a dentistYes 79No 20Don’t know 1

Visited a dentist within the last yearYes 62No 38

* nΩ100 for all questions.

Willingness to pay for community water fluoridation

Fig. 1. Summary of responses to fluoridation questions.

for or against fluoridation, 31 were classified as‘protests’. These individuals did not participate inthe WTP exercise, citing reasons such as ‘payingenough taxes/water rates already’ or ‘the watercompanies should pay’. In addition to the protests,a further nine respondents were excluded fromfurther analysis. Of these, eight did not know ifthey would be willing to pay or accept compensa-tion and one indicated that no amount of moneywould be sufficient to compensate for fluoride be-ing added to drinking water. This left 53 responsesto be analysed, of which 40 were in favour of fluo-ridation and 13 were opposed.

Of the 40 cases in favour of fluoridation, 10 gavezero values. Of these, seven indicated they couldnot afford to pay and three indicated they did notconsider the benefits to be sufficiently large to war-rant payment. All 30 who were willing to pay indi-cated they were willing to do so because of the ex-pected health benefits of fluoridation. Some re-spondents made particular reference to benefits tochildren or the notion of community-wide benefits.Of the 13 respondents who were opposed to fluo-ridation, two gave a zero WTP stating they couldnot afford to pay. The main reasons that peoplewere willing to pay (or accept compensation) wereviolation of freedom of choice and the desire tohave pure water. The responses of the sample aresummarised in Fig. 1.

Figure 2 presents the mean WTP and WTA re-sponses for the sample. Intensity of preference isgreater for those opposed to fluoridation, as shownby the higher mean responses. Mean WTP for thosein favour of fluoridation (GBP 12.63) was less thanthat for those opposed to fluoridation (GBP 29.38),although this difference was not statistically signi-ficant using the Mann-Whitney test (PΩ0.102).There was a statistically significant difference (PΩ

127

Fig. 2. Mean values to the willingness-to-pay/accept ques-tions.

0.001) between the mean WTP of those in favourof fluoridation and the mean WTA of those op-posed (GBP 76.00).

In terms of their knowledge of fluoridation, 86%of the sample had heard of fluoridation prior tothe interview. Of these, 22% had heard of effects offluoridation not mentioned in the description giv-en. All of the effects mentioned were erroneous,with several people simply claiming that it waspoisonous.

Discussion

This study has demonstrated the feasibility of usingcontingent valuation to investigate the benefits ofcommunity water fluoridation. The issue of the va-lidity of this technique and its results is more diffi-cult to assess, but it is encouraging that the respon-dents did seem to be considering relevant aspects offluoridation, as highlighted by their answers whenasked why they would be willing to pay.

There are some potential problems with the tech-nique. A positive association between WTP and in-come is typically observed in contingent valuationstudies, and this can bias the aggregated results infavour of the preferences of richer respondents. Forexample, if richer people are more likely to opposefluoridation, then the aggregate WTP would behigher even if the numbers for and against werethe same, and their intensity of preferences werethe same. This possibility was investigated bystandardising the income structure of those in fa-vour, to the same structure as those against. Thisprocedure reduced the aggregate WTP for those infavour of fluoridation.

The issue of how to define and deal with protestsis also important. In this study, any protests wereexcluded from the analysis on the basis that such

Dixon & Shackley

individuals did not provide an estimate of theirvaluation of fluoridation. Consequently, 31% of thesample was excluded. Some of the problems withprotests are generated by people in publicly fund-ed health systems being unfamiliar with beingasked to pay for health interventions. Even whenthe hypothetical nature of the payments is stressed,some people refuse to participate ‘on principle’. Inthis study, those opposed to fluoridation weremore likely to protest than those in favour (48% vs26%). This may have the effect of producing a bi-ased sample for the valuation exercise in terms ofseveral characteristics. Even at the most simplisticlevel, the exclusions changed the ratio of those infavour from around 2:1 in the referendum questionto 3:1 in the valuation exercise.

An important aspect of the WTP technique is itsability to identify ‘losers’ and quantify their wel-fare loss. This study shows that although a minori-ty of respondents were opposed to fluoridation,their intensity of preference (as measured by meanwillingness to pay or accept) was greater thanthose in favour of fluoridation. This important is-sue has been ignored by all previous economicevaluations of fluoridation.

It should also be noted that a full cost-benefitanalysis of fluoridation would need to take into ac-count the costs of fluoridation, together with anysavings it produced. Summing these amounts withthe estimate of welfare change from a WTP exercisewould give a societal estimate of the net benefits(or costs) of fluoridation. The study presented herefocused purely on benefits and does not in itselfconstitute a CBA.

Finally, although the vast majority of the samplewere previously aware of fluoridation, a numberof respondents held incorrect information about it.These beliefs will undoubtedly have reduced theirWTP for fluoridation. However, these individualswere not excluded from the analysis because theywould still suffer real welfare losses in the event offluoridation going ahead, no matter how misplacedtheir beliefs. It would seem, however, that publiceducation needs to continue in order to allay mis-placed fears and so that better informed judge-ments can be made.

Conclusion

This paper has highlighted how previous economicevaluations of water fluoridation have not takeninto account all relevant impacts of fluoridation onwelfare. The pilot study presented has demon-

128

strated the feasibility of using contingent valuationto investigate all the effects of fluoridation on wel-fare. Although some issues with regard to the tech-nique need to be resolved, the study has shown thatthere are important non-health impacts on individ-ual welfare. The study also illustrates how a simplereferendum can miss important information re-garding the intensity of individuals’ preferences.Further investigation is therefore required to repli-cate this work on a larger sample of individuals, andto develop the technique in order to resolve some ofits problems, such as the high number of protests.

Appendix

Description of fluoridation and its effectsFluoride is a naturally occurring substance whichis already present in very small quantities in watersupplies. Fluoride has been shown to be beneficialin reducing tooth decay and is included in manytoothpastes. Another way in which fluoride can re-duce tooth decay is to add it to drinking water.Where this happens, fluoride is added to the waterbefore it reaches residential areas. This means allhouseholds would receive water containing addi-tional fluoride. Adding fluoride to water does notaffect its taste. Fluoride also has no effect on house-hold equipment such as kettles and washing ma-chines.

In Sheffield, children typically have 2 or 3 teethwhich are decayed, missing or filled. For adults,the number of decayed, missing or filled teeth isaround 17.

Adding fluoride to water in Sheffield wouldhalve the number of decayed, missing or filledteeth for children and adults in the future. Theseimprovements will be greater in the poorer areasof Sheffield.

There is a very small chance that, for a few peo-ple, adding fluoride to water could cause smallwhite patches to appear on some teeth. Anyonetaking fluoride supplements, such as tablets ordrops, should stop taking them. If they do not, theywill be at greater risk of developing discolouredteeth. It may be advisable for young children tostop using toothpaste containing fluoride.

References

1. Drummond MF, O’Brien B, Stoddart GL, Torrance GW.Methods for the economic evaluation of health care pro-grammes. Oxford: Oxford Univ Pr; 1997.

2. White BA, Antczak-Bouckoms AA, Weinstein MC. Issues

Willingness to pay for community water fluoridation

in the economic evaluation of community water fluori-dation. J Dent Educ 1989;53:646–57.

3. Davies GN. Fluoride in the prevention of dental caries:a tentative cost-benefit analysis. Br Dent J 1973;135:131–4.

4. Dowell TB. The economics of fluoridation. Br Dent J1976;140:103–6.

5. Nelson W, Swint JM. Cost-benefit analysis of fluoridationin Houston, Texas. J Public Health Dent 1976;36:88–95.

6. Niessen LC, Douglass CW. Theoretical considerations inapplying benefit-cost and cost-effectiveness analyses topreventive dental programs. J Pubic Health Dent1984;44:156–68.

7. Carr SM, Dooland MB, Roder DM. Fluoridation II: aninterim economic analysis. Aust Dent J 1980;25:343–8.

8. Doessell DP. Cost-benefit analysis of water fluoridationin Townville, Australia. Community Dent OralEpidemiol 1985;13:19–22.

9. Horowitz HS, Heifetz SB. Methods for assessing the cost-

129

effectiveness of caries preventive agents and procedures.Int Dent J 1979;29:106–17.

10. Manau C, Cuenca E, Martinez-Careeterro J, Salleras L.Economic evaluation of community programs for theprevention of dental caries in Catalonia, Spain. Com-munity Dent Oral Epidemiol 1987;15:297–300.

11. Boughton BJ. Compulsory health and safety in a free so-ciety. J Med Ethics 1984;10:186–90.

12. Isman R. Public views on fluoridation and other preven-tive dental practices. Community Dent Oral Epidemiol1983;11:217–23.

13. Diener A, O’Brien B, Gafni A. Health care contingent val-uation studies: a review and classification of the litera-ture. Health Econ 1998;7:313–26.

14. Mitchell R, Carson R. Using surveys to value publicgoods. Washington (DC): Resources for the Future; 1989.

15. Donaldson C, Thomas R, Torgerson DJ. Validity of open-ended and payment scale approaches to eliciting willing-ness to pay values. Appl Econ 1997;29:79–84.