Evidence from human longitudinal, interventional, and experimental studies There are many reasons why there are so few planned interventional human studies of diet and dental caries¾for example, the problem of persuading groups of people to maintain rigid dietary regimens for long periods of time. Although most of such studies involved providing daily sugar supplements to subjects¾a practice that would be considered unethical today¾these studies made an important contribution to dental knowledge. However, 25 to 50 years ago, at the time of these studies, the standard of oral hygiene was very poor and fluoride toothpaste was unavailable: In most industrialized countries, both caries incidence and caries prevalence were high. Vipeholm study. The Vipeholm study is indisputably unique in the annals of caries research. All previous and most subsequent human studies of the diet-caries relationship have been either epidemiologic or cross-sectional surveys, based on dietary recall. Because such studies are noninterventional, the investigator has no control over either the amount or the frequency of sugar ingestion. Hence, the
importance of the Vipeholm findings is matchless.
The study was conducted in Sweden, over a 5-year period (1946 to 1951). The aim was to clarify the relationship between sugar intake and caries incidence (Gustavsson et al, 1954). The subjects comprised 436 institutionalized, mentally handicapped, or retarded adults. At the baseline examination in 1946, their mean age was 32 years.
Because of their poor oral hygiene (fewer than 20% brushed their teeth regularly), they had abundant amounts of plaque, an important prerequisite for caries development. The subjects were therefore not representative of the general population. The stated objective was to study the relationship between sugar consumption and caries activity by varying the cariogenic substrate (present or absent), the amount of sugar (less than, equal to, or double the normal intake), the form of sugar (nonsticky or sticky), and the frequency of sugar intake (only at meals
or at and between meals).
The chronology of the study falls broadly into three categories. During the preparatory and vitamin period (1945 to 1947), all subjects received a diet relatively low in sugar (about half the normal intake) and no additional sugar at meals. The baseline caries incidence was low, about 0.34 new carious surfaces per patient per year. During the next 2 years (1947 to 1949), carbohydrate study I, most groups consumed about twice the normal amount of sugar, but only at meals. During the final 2 years (1949 to 1951), carbohydrate study II, most groups ate normal amounts of
sugar, some only at meals and others both at and between meals. Seven distinct study groups were established:
1. Control group: continued on a low-sugar diet, only at meals
2. Sucrose group: received a high-sugar diet, mostly in drinks with meals
3. Bread group: received sugar intake either half that or equal to normal, but only in
sweetened bread at meals
4. Caramel group: given 22 sticky candies, either in two portions at meals
(carbohydrate study I) or in four portions between meals (carbohydrate study II)
5. Eight-toffee group: given eight toffees in two portions at meals (carbohydrate study
I) or in four portions between meals (carbohydrate study II)
6. Twenty-four-toffee group: allowed to eat 24 toffees, at their pleasure throughout
the day, with about twice the normal total intake of sugar
7. Chocolate group: given milk chocolate in four portions between meals
(carbohydrate study II)
Sugar consumption at meals in a nonsticky form, over a wide range of total daily
intake, from 30 to 300 g (carbohydrate study I), had very little influence on the
baseline caries rate of 0.3 to 0.5 new carious surfaces per year. The addition of sugar
to the diet resulted in an increased caries incidence, but the increase varied depending
on the manner of consumption (carbohydrate study II). Sugar consumed in sweet
drinks with meals or in bread eaten at meals had little effect. The group receiving
chocolate four times daily between meals showed a moderate increase in caries.
However, a dramatic increase occurred in groups receiving 22 caramels, eight toffees,
or 24 toffees between and after meals. Thus, caries risk was greatest if the sugar was
consumed between meals, in a form that was retained in the mouth for a long time and
provided high concentrations of sugar (Fig 53). However, there were wide individual
variations. In fact, approximately 20% of the patients did not develop any caries, even
after consuming 24 toffees daily.
In the Vipeholm study the total sugar consumption by the subjects was about twice
that of the normal Swedish diet, and their plaque accumulation was far heavier than
normally is found today. Other caries-modifying variables were also different 50
years ago. The results, therefore, should not be extrapolated directly to modern
societies. The main conclusions from the Vipeholm study were:
1. Consumption of sugar, even in large quantities, is associated with only a small
increase in caries incidence, provided that ingestion is limited to mealtimes, at most
four times a day.
2. In subjects with poor oral hygiene, consumption of sugar both between meals and
at meals is associated with a marked increase in caries incidence.
3. Under uniform experimental conditions, the increase in caries incidence varies
widely from person to person.
4. Caries activity subsides once sugar-rich foods are withdrawn from the diet.
5. In subjects with poor oral hygiene, carious lesions occur despite the avoidance of
sugar.
Turku sugar studies. By 1970, there was considerable evidence of variation in the rate
of acid production from different sugars by plaque microorganisms: For example, the
sweet polyalcohols (sorbitol, xylitol, and mannitol) produced virtually no acid.
Animal experiments had also demonstrated that sugars differed in their cariogenicity.
To test whether the same difference applied to humans, a clinical study was conducted
in Turku, Finland, from 1972 to 1974 (Scheinin and Makinen, 1975). The objective
was to study the effect on dental caries incidence of almost total substitution of
sucrose, with either fructose or xylitol, in a normal diet.
Because the study required the full cooperation of the subjects, including undergoing
a wide range of biochemical and microbiologic tests, the study was restricted mainly
to adults, most of whom were associated with the Turku dental or medical schools. Of
the original 125 subjects, 115 remained after 2 years. There were three groups of
subjects: sucrose (S), fructose (F), and xylitol (X). Because full cooperation was
essential, the subjects were invited to choose which group they wished to join. All
clinical caries examinations were conducted blindly by the observer throughout the
study. Two standardized bitewing radiographs were taken of each side of the mouth.
Precavitational and cavitational lesions were recorded, for both primary and
secondary caries.
The organization of the dietary regimens for the subjects in the three groups was very
complex, requiring virtually all foods that normally contain sucrose to be
manufactured with fructose or xylitol instead of sucrose. The cumulative development
of caries, diagnosed both clinically and radiographically, is presented in Fig 54 (a).
These results include both precavitational and cavitational lesions. The 24-month
caries incidences in the S, F, and X groups were 7.2, 3.8, and 0.0, respectively. These
refer to primary caries only. Inclusion of secondary caries (Fig 54 (b)) gives a
considerable increase in the magnitude of caries incidence (10.5, 6.1, and 0.9 decayed,
missing, or filled surfaces in the S, F, and X groups, respectively at 24 months),
indicating the importance of secondary caries in adults.
In 1987, to quantify changes in the size of approximal carious lesions, Rekola
reexamined the radiographs from the Turku study using a planimetric method. At
baseline, the mean size of the lesions was similar for the X and S groups, but at the
end of the 2-year study the mean size was significantly smaller in the X group (P <
0.01): Lesion size increased almost linearly by 0.12 mm2/year in the S group but
remained virtually unchanged in the X group.
Analyses of the caries data over the 2-year period showed that substitution of xylitol
for sucrose in a normal Finnish (high-sucrose) diet resulted in a markedly lower caries
incidence for both initial and manifest lesions. Although subjects in the S group
developed more initial lesions than did those in the F group, more lesions in the F
group progressed to cavitation. The X diet was clearly less cariogenic than either the
S or F diet, but it cannot be concluded that the F diet was less cariogenic than the S
diet.
Comprehensive biochemical and microbiologic tests were carried out parallel to the
caries assessments. Although a very slight fall in plaque weight was observed in the S
and F groups, a much greater decrease was recorded in the X group (P < 0.005).
Substitution of dietary sucrose with xylitol did not affect the proportion of major
bacterial groups in dental plaque but did reduce the number of most organisms,
especially the acidogenic and aciduric flora, including S mutans. Plaque from X group
subjects showed a reduced rate of sucrose hydrolysis. No adaptation by plaque
organisms to produce acid from xylitol was observed.
Experimental caries study. In experimental human studies (Von der Fehr et al, 1970),
development of buccogingival enamel caries was evaluated by the use of a dissection
microscope. Over a period of 23 days, dental students rinsing nine times daily with 10
mL of a 50% sucrose solution developed a higher Caries Index and more early lesions
than did the control group. Both groups abstained from oral hygiene. After 30 days of
oral hygiene and daily fluoride rinses, the Caries Index returned to preexperimental
levels (Fig 55).
This experiment demonstrated the rapid cariogenic effect of sucrose in combination
with dental plaque. The fact that sugar in solution proved highly cariogenic suggests
that the critical factor is the duration and frequency of sugar exposure rather than the
physical form of the sugar-containing food. However, during the 23 days without oral
hygiene, the controls also developed enamel caries.
Subsequently, the sucrose rinsing experiment was repeated for 3 weeks. This time, the
subjects employed chemical plaque control by rinsing twice a day with 0.2%
chlorhexidine solution but used no fluoride; no caries developed (Loe et al, 1972).
These two short-term human experimental studies showed that:
1. Sugar is not an etiologic factor for caries development, but it is a modifying risk
factor.
2. Dental plaque is an etiologic factor for caries development.
3. Despite frequent sugar intake, clean teeth do not develop caries, even in the
absence of fluoride.
These early longitudinal, interventional, and experimental studies in Scandinavian
adults clearly showed sugar to be an external modifying risk factor for caries
development. However, 25 to 50 years ago, both caries incidence and caries
prevalence in Scandinavia were very high. Moreover, in the designs of both the
Vipeholm study and the experimental caries study, the frequency of sugar intake
(eight to 24 times per day) was extreme, and the major modifying preventive factors
(plaque control and fluoride administration) were absent. In other words, the etiologic
factor (thick, undisturbed dental plaque) was continuously in situ on most tooth
surfaces and there was no intermittent supply of fluoride to modify the fall in plaque
pH.
For ethical reasons, under the Helsinki Declaration, such interventional human studies
would no longer be permitted. Therefore, the relative role of sugar as an external
modifying risk factor for caries development under present conditions in Scandinavia
is unknown. Because of the excellent standard of plaque control and associated use of
topical fluorides, particularly in toothpaste, both caries incidence and caries
prevalence in children are very low, despite an increase in the consumption of sticky,
sugar-containing products over the past 30 years.
Observational studies. Although interventional human longitudinal studies with
frequent sugar administration are no longer permitted, longitudinal observational
human studies are still allowed, and a few have been conducted during the past two
decades. The studies by Wendt et al (1992), Wendt (1995), Wendt and Birkhed
(1996), and Wendt et al (1996), described earlier, documented the relationship
between oral hygiene and dietary habits and caries development from the ages of 1 to
3 years.
In a region of Egypt where the water fluoride concentration was higher than 1 mg/L,
Axelsson and El Tabakk (2000b) followed caries incidence in relation to dietary
habits in a 2-year study of 685 12 year olds with very poor oral hygiene habits (fewer
than 10% brushed their teeth daily). The diet was evaluated according to a
cariogenicity point scale. The results showed that a diet rich in sugar was a risk factor
for caries development, albeit a weak one. The caries incidence per individual over 2
years, related to cariogenicity scores 1 to 8, 9 to 13, and 14 to 17, was 0.8, 1.0, and 1.9
new carious surfaces, respectively.
Two other important large-scale observational longitudinal studies were conducted in
schoolchildren in Northumberland, England, by Rugg-Gunn et al (1984) and in
Michigan by Burt et al (1988). Some data from the two studies are compared in Table
5. To avoid confounding effects, both investigations were conducted in communities
with low concentrations of fluoride in water. In the English study, from 1979 to 1981,
the subjects were initially aged 11.5 years; in the American study, from 1982 to 1985,
the subjects were initially aged 11 to 15 years. Dietary analyses differed: Rugg-Gunn
et al used 3-day diet diaries on five separate occasions, each followed by an interview,
using models to assess portion size. Burt et al used 24-hour recall interviews,
conducted with the aid of food models, on three or four occasions. To assess the
importance of frequency of eating, the timing and grouping of food intakes were
noted in both studies. Caries was scored by clinical and partial radiographic
examination (Rugg-Gunn et al) or clinically only (Burt et al). In both studies, pit and
fissure caries and approximal caries were scored separately.
In the Rugg-Gunn et al study, caries incidence was related to a wide range of dietary
variables and by examining groups of children with extremes (high versus low) of
sugar intake and caries incidence (0 versus more than 7 new decayed, missing, or
filled surfaces). For total daily intake of sugars and total caries incidence, the
coefficient of correlation was low, but increased when the incidence of fissure caries
was considered alone. The overall incidence for smooth-surface caries may have been
too low to give statistically significant results.
In a later (1987) analysis of the data, Rugg-Gunn et al examined the possible
interaction between starch and sugar in the development of caries. The subjects were
divided into a high-sucrose/low-starch group, and a low-sucrose/high-starch group.
The former developed more new carious lesions than did the latter, but only the
difference for fissure caries approached significance. No significant correlations were
found between starch intake and any measure of caries incidence. The highsugar/
low-starch group ate more frequently than did the low-sucrose/high-starch
group (7.8 versus 5.7 times a day).
Burt et al (1988) did not report correlation coefficients between caries incidence and
dietary variables but divided the subjects into groups corresponding to contrasting
caries incidences or dietary practices. Table 6 shows the selected dietary variables for
their 0-increment group compared to those from children with 2 or more new carious
approximal surfaces during the 3-year study. Only the comparisons of energy derived
from carbohydrate (or sugar) in snacks and the percentage of total energy from sugars
in snacks reached conventional levels of significance. In contrast to the conclusions of
the English study, no difference was observed in the energy intakes from total sugars
or from meals and snacks with one or more high-sugar foods. These differences were
unaffected when baseline age was taken into account. Social factors had a highly
significant relationship with caries incidence but did not confound any of the
relationships with dietary factors.
When the subjects were grouped on the basis of high and low sugar intake, those
consuming high levels of total sugars developed more new carious lesions, but the
difference was not pronounced and approached significance only for approximal
surfaces. There was no change in significance when the subjects were grouped
according to frequency of eating or high intake of sugary snacks. However,
significantly more new approximal carious lesions developed in those eating large
amounts of sugars between meals, and here the differences in total caries incidence
approached significance. In contrast to the findings of Rugg-Gunn et al (1984), none
of the analyses disclosed any differences with respect to fissure caries.
Differences in methods of collecting and analyzing the data preclude any direct
comparison of these two well-conducted longitudinal trials. Nevertheless, in general,
both studies confirmed that, while there is a relationship between caries development
and aspects of dietary sugar consumption, it does not explain intersubject variance.
Both studies also cast considerable doubt on the importance of the frequency of eating
as a caries predictor.
Differences between the findings have already been indicated. The English study
found that the fissures were more sensitive than the approximal surfaces to intake of
sugars and other dietary influences, although the number of fissures attacked was
greater and would be expected to show significance more readily. In the American
study, although more than 80% of the new carious lesions developed in pits and
fissures, caries at these sites was apparently unaffected by sugar intake or eating
frequency.
Furthermore, the American study found a relationship between caries and the amount
of sugar in snacks but not between caries and total sugar intake or intakes with one or
more high-sugar items. The English data revealed high degrees of correlation between
caries and total sugar intake and a significant relationship between caries and highsugar
items. In both reports, the findings were interpreted in the light of the relatively
low caries incidence observed (> 1 new carious surface per individual per year).
Rugg-Gunn et al (1987) attributed the unimpressive correlation coefficients between
caries and many of the sugar-related variables to problems of data collection
(especially dietary data), intrasubject variation, and the low caries incidence and
suggested, in hindsight, that a longer study and intergroup comparisons would have
been preferable. These recommendations were adopted by the American workers,
without an apparent increase in the sensitivity of the trial. In this context, it is of
interest that comprehensive epidemiologic data collected annually from the total
population in the county of Varmland, Sweden, disclosed that in the corresponding
age groups, caries incidence decreased from 1.2 new DS in 1979 to 0.1 new DS in
1997 (Axelsson, 1998), even though there was no concurrent reduction in
consumption of sugar-containing products.
Both animal experiments and human intra- oral biochemical tests (plaque pH, which
will be discussed further) strongly suggest that the most critical predictor of dietary
cariogenicity is frequency of intake and that foods containing sugars generally have
the greatest potential to raise the acidogenicity of plaque and thus the cariogenic
challenge. The two aforementioned clinical studies, in which the methods were
optimized, indicate that these two factors (especially frequency) are not of overriding
importance, and this brings into question the likely efficacy of current dietary advice
to patients. For example, in the Rugg-Gunn study (1987), the low-sugar/high-starch
group (the subjects presumably complying with recommendations to reduce sucrose
and increase dietary starch intake) achieved only a 31.7% reduction in caries¾a
modest result compared to the groups whose dietary practices would certainly not be
approved by dental health educators.
The most significant predictors of caries risk identified in the study by Burt et al
(1988) were social factors: parental education and income. The role of educational
level as an external modifying factor will be discussed later in this chapter.
