MRC working group report (Sept. 2002): Water fluoridation and health
Following the Systematic Review of Water Fluoridation by the NHS Centre for Reviews and Dissemination at the University of York in September 2000, the MRC was asked to:
(a) advise on the scientific evidence on the health effects of fluoridation;
(b) consider what further research might be needed;
(c) assign priorities for public health policy.
The MRC's report has good and bad points. Some of the latter are serious, although the former are far from negligible.
A. Good points
1. Confirmation that not enough is known about the health effects of water fluoridation is welcome. Even three years ago such an idea was dismissed in dental, medical and political circles.
2. Some important recommendations have been made, notably in the areas of
(i) research into:
(a) total and lifetime fluoride exposures in populations;
(b) the relative importance of water as a source of fluoride;
(c) the prevalence of dental fluorosis;
(d) cancer rates in fluoridated areas;
(e) the different effects if any of naturally and artificially fluoridated water;
(f) bone health in fluoridated areas, depending on (e).
(ii) dialogue between scientists and public over research and perception of risk.
B. Bad points
1. The MRC misrepresented
(a) York's findings over the strength of the evidence for caries reduction;
(b) the strength of the evidence for reducing dental inequalities across social groups.
2. The MRC introduced much new material, citing earlier reviews as well as primary studies. Since, unlike York's, its assessments were not demonstrably systematic nor transparent, its scope was less thorough, and any new material would by definition fall below the evidence level (already low to moderate) accepted under York's internationally recognised criteria, its judgements must suffer in comparison, except where/if York was clearly in error.
In simple terms, the MRC has a credibility problem over its assessment of the evidence.
3. The MRC, unlike York, gave little acknowledgement to the variation in quality of the studies it considered, and thus to the confidence with which conclusions could be stated. This again will have affected its judgements.
4. Some specific recommendations from York were ignored, for example the proposal to look at infant mortality, and at a concentration of 0.8 ppm. It would have been helpful to hear the MRC's reasons (which may have been good ones) for not pursuing these.
5. The MRC has not adequately absorbed York's strictures on the poor methodology of even the more recent fluoridation studies. It has therefore not emphasised sufficiently, and specifically, the need for research of the highest quality if policy is to be reliably underpinned. (Will this gap be filled by the MRC Research Board(s) referred to in 1.1?)
6. In widening the scope of enquiry to include public involvement and perceptions of risk, the MRC has omitted some key questions that need addressing.
7. Compared with York, there is a noticeable pro-fluoridation bias in the MRC's judgements, even if it is less strong than in earlier reviews.*
Two further flaws cannot be laid wholly at the working party's door.
8. Its remit encouraged a second-guessing of what York had already done to a high standard, which (despite a disclaimer) is largely what occurred.
9. Its composition was weighted towards supporters of a fluoridation policy.*
C. Questionable points
The actual recommendations for research in section 6. are sometimes unclear, and sometimes questionable in the light of both York's and the MRC's assessment of the evidence.
1. The lay summary says that the possibility of "hip fracture is the most important [effect] in public health terms". 6.1 lists dental fluorosis, bone health and cancer as the main negative outcomes of "interest and relevance". While there is a reasonable case for these, York flagged congenital defects, IQ and infant mortality, which could be considered important, and if the MRC is serious about involving the public in the research agenda these priorities may need to be further adjusted.
2. Recommendation 6. under 6.2 calls for "an estimate of the effects" of fluoridation on children, but it is not clear precisely what this means. Read literally, it suggests research to see to what extent fluoridation might reduce caries and produce harmful side-effects, which in view of the uncertainties highlighted by York will be a major (and very important) undertaking. (Will negative effects be looked for separately or together?)
The second part addresses the need to know more about effects across social classes, which the body of the report (unlike York) appeared to take as read, and to controlfor "effect modifiers". This is important.
What is surely needed in addition, in the light of the findings by York, are
(a) High-quality studies (level 'A', in York's terms) to supply the "clear confidence" (or not) in the effectiveness of fluoridation, which York found lacking, as well as in equity between social classes;
(b) Clearer guidance about how this major task should be carried out in a way which will avoid the pitfalls which have left the community with so little reliable knowledge about fluoridation after 50 years, to include reference to all potential confounding factors flagged by York (e.g. observer bias, number of erupted teeth, in addition to those mentioned in 6.).
3. Recommendation 8. speaks of learning more about fluoridation's impact on quality of life and economic indices. Presumably this refers back to 4.1 & 4.1.3, where children's experiences of toothache, abscesses and anaesthesia for extractions are raised. Two comments are relevant:
(a) while of interest to the dental profession (whose representatives nevertheless agreed the York protocol), research into more refined indices than the usual dmft/DMFT may be thought a doubtful use of resources, especially while more serious long-termharms remain uninvestigated;
(b) if research in this area is contemplated, it must clearly wait on C.2.(a) above since any positive "impact" of fluoridation must first be confirmed. And no cost-effect iveness studies will be valid until effectiveness is clearly shown.
4. Where study designs are specified (e.g. 6.2: 9, 13), they would appear inadequate to remove uncertainty and inform policy reliably. If this is so, they will be a waste of time and money. Cross-sectional studies, for example, recommended by the MRC for dental fluorosis (9), and a case-control study for hip fracture (13), are ranked by York as level 'C', i.e. "lowest quality of evidence, high risk of bias". If York's standards are not accepted in this - and any future systematic review will be likely to adopt similar criteria - there needs to be an urgent high-level meeting of the research community to agree on questions of methodology. If scientists cannot agree what is good science, the general public cannot be expected to make informed judgements on the evidence.
19th October 2002
* It is a point worth noting that the one review (York, 2000) that ensured a fair representation between proponents and opponents of fluoridation is the one review to have highlighted the weaknesses in the science.