When a scientist uses his position of authority and trust to deceive the public, he does not become a dishonest scientist: he ceases to be a scientist.
This is a study of self-selected data with no blinding and little weight given to confounding factors. Clearly the York team would have rejected this one.
Firstly, this study breaks the fundamental rule of scientific enquiry. Both Lee and Dennison are ardent pro-fluoridationists, and selected data knowing what outcome they wanted. To avoid unconscious bias data must be selected “blind”, unlike this study. Napier and Hastings were chosen for New Zealand’s original study because of their demographic similarity, to minimise confounding factors. Wellington, by contrast, is quite unsuitable for comparison with any other community: It has the highest educational and income levels in New Zealand, both of which correlate to good dental health. It would be extremely difficult to allow for confounding factors using Wellington statistics. We only need to look at statistic from nearby Hutt Valley to see this. For Form 2 children, Hutt Valley Health records 1.10 DMFT in fluoridated areas and 1.01 DMFT in unfluoridated areas. It shows up in the Maori statistics too: 1.18 DMFT fluoridated, 0.72 DMFT unfluoridated (i.e. the unfluoridated areas have less decay than the fluoridated areas. Obviously something in that area is leading to better teeth (irrespective of fluoridation status).
In fact, Canterbury and Wellington are not a particularly good fit. For example, Wellington is much more urbanised than Canterbury, and there is a well-known correlation between urbanisation and lower rates of decay.
Amongst the serious flaws in this study are the following:
- The fluoridation history of the subjects was not known, nor whether they used fluoride toothpaste or not;
- Two of the communities discontinued fluoridation (i.e. changed their fluoridation status) during the subjects’ lifetime;
- There was no blinding;
- Data collection was not standardised;
- Examiner reliability was not ensured;
- Comparability of subjects was not assessed;
- Confounding factors were not eliminated;
- The mineral, particularly calcium and phosphorus, content of the respective water supplies was not determined
The socio-economic evaluations are completely unreliable. They use the 1996 TEFA ratings to assess the 1996 data. But the 1996 TEFAs were derived from 1991 census data. Moreover, TEFAs have been roundly criticised by schools for not accurately reflecting the socio-economic status of their students. Not only were the TEFAs based on old data, but that data was derived from just a limited sampling of students (not the entire student population).
But there is an even a more fundamental flaw. To adequately correlate socio-economical status with decay, you need to do so on a subject-by-subject basis. You need to associate a particular child’s DMFS with that child’s status, not give every child in the school the same average value (TEFA). As a specific demonstration of the unreliability of the socio-economic assessments used in the study, Te Aro School (Wellington) was designated decile 3 in 2002 (lowest group) but decile 8 in 2003 (highest group). (see table below)
The ethnic evaluations are also worthless. Despite the fact that the study was not designed to assess differences based on ethnicity, the authors make broad statements based on inadequate data: “in the non-fluoridated group, the mean dmfs score of Maori five-year-olds was double that of the “Other” group, and that for Pacific children, three times greater than the “Others”. That non-fluoridated group is Canterbury, which has a low Maori/Pacific population. Drawing conclusions from such a small sample is worthless. One must also question why South Canturbury was included in the Canturbury sample. The obvious comparison would be between Wellington and Christchurch. Including South Canterbury makes the unfluoridated sample much larger than the fluoridated sample. Why was that done? It creates a suspicion that the higher decay rates of Timaru were added to pull down the Christchurch figures.
As an indication of the readiness of the authors to mislead the public, they quote Fergusson (1986), Stockwell (1990), Treasure (1994), Slade (1996), Jones (1999) and Riley (1999) as authorities but neglect to mention that the York Review (2000) said those studies were poor quality and could not be relied on.
But ultimately their conclusions actually showed there was not much difference anyway.
“RESULTS: Caries prevalence and severity was consistently lower for children in the fluoridated area for both age groups, and within all subgroups. Five-year-olds in the fluoridated area had 2.63 dmfs (sd, 5.88), and those in the non-fluoridated area 3.80 dmfs (sd, 6.79). For 12-year-olds the respective figures were 1.39 DMFS (sd, 2.30) and 2.37 DMFS (sd, 3.46). Multivariable analysis confirmed the independent association between water fluoridation and better dental health.” See study.
This results in a saving of 1.17 decayed, missing of filled surfaces in 5 year olds and 1.09 decayed, missing or filled surfaces in 12 year olds. This is hardly any achievement.
Decile ratings of selected schools
|2002 decile||2003 decile|
|Waihao Downs School||6||10|
|Te Moana School||9||5|
|Te Aro School||3||8|
|Glen Oroua School||4||10|
|Arthurs Pass School||7||1|
|Timaru Christian School||9||4|
|Makarora Primary School||10||5|
|South Wellington Intermdiate||5||8|