Professional Documents
Culture Documents
Dental Caries ARF Post Review
Dental Caries ARF Post Review
email: sithor@gmail.com
Key words
Epidemiology; rheumatic fever; dental caries; cohort study; dietary sucrose.
Word count
3,232
1
Licence for Publication
The Corresponding Author has the right to grant on behalf of all authors and does grant on
behalf of all authors, an exclusive licence (or non exclusive for government employees) on a
worldwide basis to the BMJ Publishing Group Ltd to permit this article (if accepted) to be
published in JECH and any other BMJPGL products and sublicences such use and exploit all
for-authors/licence-forms).
Competing interests
None declared.
2
Abstract
Objective
associated with the incidence of acute rheumatic fever and chronic rheumatic heart disease, in
Māori and Pacific children aged five and six years at their first dental visit.
hospital discharge and mortality records. Cox models were used to investigate the strength of
Results
A total of 20,333 children who were free of rheumatic heart disease at enrolment were available
for analysis. During a mean follow-up time of five years, 96 children developed acute rheumatic
fever or chronic rheumatic heart disease. After adjustment for potential confounders, children
with five primary teeth affected by caries were 57% (95% confidence interval: 20% to 106%)
more likely to develop disease during follow-up, compared to children whose primary teeth were
caries free. The population attributable risk for caries in this cohort was 22%.
Conclusions
Dental caries is positively associated with the incidence of acute rheumatic fever and chronic
rheumatic heart disease in Māori and Pacific children. Sugar intake, an important risk factor for
3
What is already known about this topic?
Acute rheumatic fever is strongly associated with Streptococcus pyogenes infection. A closely
Indices of a poor diet, such as low body weight and low serum albumin have been historically
Māori and Pacific children have both a high prevalence of dental caries and acute rheumatic
fever.
Dental caries in childhood is positively associated with the incidence of acute rheumatic fever,
This finding suggests that dietary sugar intake, a known risk factor for dental caries, is also likely
4
INTRODUCTION
Acute rheumatic fever (ARF) is almost absent from most developed countries, yet New Zealand
continues to carry a high burden among Pacific and Māori populations.1 The disease leads to
ethnic inequality in health status. The incidence and prevalence has declined substantially in
New Zealand, but remains similar to countries such as China, Russia, and Qatar.2
A new hypothesis is that ARF is related to sugar (sucrose) intake.3 It is plausible because
Streptococcus pyogenes, a bacterium associated with acute rheumatic fever, is able to ferment
sugar.4 This bacterium uses glucose as its primary carbon source during growth, but is also able
At the same time, a closely related bacterium, S. mutans, is associated with dental caries, so that
a relationship between ARF and dental caries would signal an ARF - sugar relationship. The
hypothesis is supported by evidence that Māori and Pacific children are most susceptible to ARF
and also have a high prevalence of dental caries, compared to other ethnic groups. 6
In a recent review of the relationship between sugar and caries, Sheiham and James7 conclude
that the main cause of caries is sugar intake. Other, less important influences that were
discussed are oral hygiene and water fluoridation. Their argument is supported by many
ecological studies which show a reduction of caries when sugar is restricted due to rationing, for
5
example.8-10 If sugar causes dental caries and caries is predominately the result of sugar intake
(figure 1; solid grey arrow), then caries is effectively a surrogate for sugar and may be used to
estimate the presence of an association between sugar intake and acute rheumatic fever (figure
1; dashed black arrow). Just as the bacteria that cause dental caries to thrive in the presence of
sugar, it is plausible that sugar will also fuel the growth of bacteria that cause streptococcal
[figure 1 here]
Auckland city has a single provider of dental care for children. All children in Auckland are
invited to participate in the service from birth to thirteen years, and are, ideally, seen annually
for review. Since these records and databases of hospital diagnoses exist for state run hospitals,
we conducted a cohort study linking these two databases to investigate the nature of the
association between dental caries and rheumatic fever incidence. Since ARF, in New Zealand, is
almost exclusively a disease of Māori and Pacific Island children (there were just four cases of
Data was extracted from the Auckland Regional Dental Service (ARDS). This service provides
preventive, restorative and surgical dental services, mostly undertaken by dental therapists.
Estimates of enrolment in the service vary by age, but are between 90 to 96% of children age 3 to
4 years. Despite high enrolments, many do not attend clinics. The proportion that do not attend
The data extract included individuals at their first dental examination, between 14/5/2007 and
24/3/2015. Items available in the dental service data included date of birth, date of visit, ethnic
group (as claimed by parents: “Māori”, “Pacific” and “Other” ethnicity), gender, a unique
6
National Health Index code (NHI) and measures of dental caries. If a child’s parents claimed
both Māori and Pacific heritage, the child was classified as “Māori”. The caries measures
included decayed, missing (due to caries) and filled teeth with primary (dmft), and permanent
teeth (DMFT) coded separately. The NHI code enabled these records to be combined with
Exclusion criteria
Children with potentially life-limiting conditions were excluded by searching for any conditions
in the first five diagnosis categories in hospital records before they first appeared in the dental
database. This was carried out to reduce possible confounding by chronic disease status.
Diagnoses that were excluded were obtained from a study which specifically focused on children
with life-limiting illnesses.12 These were specifically: C00-D48 neoplasms, D60-D64 aplastic and
other anaemias, D80-89 immune disorders, E70-84, E88 metabolic syndromes, G00-G99
nervous system diseases, I42 cardiomyopathies, J40-J47 chronic lower respiratory diseases,
J80-84 interstitial respiratory diseases, J96.1 respiratory failure, N18 renal failure, Q00-Q99
congenital malformations, P27 perinatal chronic respiratory disease, and P91 cerebral
All children who otherwise met the entry criteria had records checked for codes consistent with
rheumatic fever or heart disease (any, rather than principal diagnosis) before their first visit to
the dental service. Children with such codes were excluded. Individuals without an NHI code
Inclusion criteria
The study population was restricted to Māori and Pacific children aged between five and six
years of age on the date of their first dental visit. The exclusion of children of other ethnicities
was applied as the incidence of ARF is very low in this group, with the likely effect of an unstable
statistical model, due to the presence of a strong association between ethnicity and disease
7
incidence. The age restriction was applied to limit the effects of exfoliation of carious primary
teeth.
Caries measures
The sum of decayed, missing (due to caries) and filled primary teeth (dmft), was recorded in the
dental register. Scores range from 0 to 20. If a carious lesion was found, or both a lesion and
restoration, then the tooth is labelled ‘d’, if a filling was present, the tooth is an ‘f’, and if
extracted due to caries it was ‘m’. DMFT similarly applies to permanent teeth. As a child ages,
between 5 and 12 years, the primary teeth are shed. If a child exfoliates a tooth affected by caries
(decayed), or a filled tooth, then dmft will decrease; however, an extracted tooth (‘m’) remains.
Both permanent (DMFT) and primary teeth (dmft) caries were considered as predictors of ARF.
Outcomes
The first appearance, after first dental examination, of an ICD-10 diagnosis of acute rheumatic
fever (I00-I02), or chronic rheumatic heart disease (I05-I09) in either death or hospital
discharge records (primary diagnosis only), during follow-up, was considered the outcome of
interest.13 Only primary diagnoses were used to reduce false-positive cases. The final date that
Loss to follow-up
We assumed that all children remained in New Zealand during follow-up, and none were lost, as
Ethical considerations
Ethical approval was granted by the New Zealand Health and Disability Ethics Committee
(14/STH/133).
8
Analysis
Checks were made for duplicate records and implausible data (wrongly recorded age, for
example). Differences in outcomes by caries status were assessed with Kaplan-Meier plots, by
initially splitting the distribution at the median value of dmft. Cox proportional hazard models
were used to assess the strength of association between dental caries and time until first hospital
diagnosis of ARF. If individuals had no such outcome, they were considered censored at the end
of the study period (22/04/2015), or in the event of death from another cause. Age at the time of
an ARF event, in those in whom it occurred, or at 22/04/2015 was considered the time scale for
analysis, since it is now recommended in the analysis of cohort studies in which conditions are
age-related.14
Results are presented as crude associations and adjusted for potential confounders (gender,
ethnic group and socioeconomic status) using Cox regression modelling. Restricted cubic splines
were used to check on the modelling assumption of linearity for continuous covariates.
Confounders included gender, ethnic group (Māori, and Pacific) and socioeconomic status
of characteristics derived from the New Zealand census (car and telephone access, receipt of
benefits, unemployment, income, education, home ownership and living space). All these
variables were considered to be common causes of caries and ARF, and so were adjusted for in
the analysis.
The assumption of proportional hazards throughout the duration of follow-up was checked by
creating interaction terms with time for each covariate and calculating the Pearson product-
moment correlation between scaled Schöenfeld residuals for this term with a log transformation
of time. A significant correlation between residuals and transformed time suggests deviation
from proportionality during follow-up. Evidence for interactions was checked between ethnic
group and severity of dental caries and survival, using a likelihood ratio test. A two-sided type-1
9
error rate was set at 5%. Very few missing values were present, mainly in people without NHI,
Population attributable risk (PAR) for significant dental exposures was calculated using the
formula for multiple exposure levels15 with estimated risk ratios taken from the fitted Cox
model. PAR here is effectively the percentage reduction in ARF incidence that would result if
R software was used for all analyses (version 3.2.1).16 The rms, survival, magrittr and epicalc
RESULTS
The flow of individuals into the study is outlined in figure 2.
[Figure 2 here]
After age and ethnicity restrictions were applied, a total of 20,333 individuals were available for
analysis, totalling 103,588 person-years of follow-up. Mean follow-up time for the cohort was
five years, over which time 94 cases developed. Table 1 shows the characteristics of the study
population, divided by the level of caries in the primary teeth (dmft = 3 represents the median
value). The table shows that Pacific children had higher levels of caries than Māori, and those
subjects with lower socioeconomic status and boys had higher levels of caries. Lesions in the
primary teeth mainly consisted of decay, which are presumably treated at subsequent visits, as
did those in the permanent teeth. The proportion of children who developed ARF or chronic
rheumatic heart disease during follow-up was about doubled in the high caries group (dmft ≥
10
Table 1. Baseline characteristics, by caries status (dmft) at enrolment.
dmft
<3 ≥3 Total P-
mean (sd)* mean (sd)* mean (sd)* value†
Total 11,593 8,740 20,333
Gender (Male); n (col. %) 5,669 (48.9) 4,440 (50.8) 10,109 (49.7) 0.008
Age at first visit (years) 5.84 (0.57) 5.92 (0.57) 5.88 (0.57) <0.001
Ethnic group; n (col. %) < 0.001
Māori 5,891 (50.8) 3,408 (39.0) 9,299 (45.7)
Pacific Island 5,702 (49.2) 5,332 (61.0) 11,034 (54.3)
Deprivation quintile; n (col. %) < 0.001
1 and 2; least deprived 896 (7.7) 286 (3.3) 1,182 (5.8)
3 and 4 1,131 (9.8) 481 (5.5) 1,612 (7.9)
5 and 6 1,356 (11.7) 777 (8.9) 2,133 (10.5)
7 and 8 2,459 (21.2) 1,776 (20.3) 4,235 (20.8)
9 and 10; most deprived 5,514 (47.6) 5,189 (59.4) 10,703 (52.6)
Rural (not categorised) 237 (2.0) 231 (2.6) 468 (2.3)
Primary teeth caries
decayed 0.408 (0.709) 4.697 (3.008) 2.252 (2.947) < 0.001
missing 0.005 (0.078) 0.212 (0.893) 0.094 (0.597) <0.001
filled 0.067 (0.306) 1.010 (1.830) 0.472 (1.308) < 0.001
dmft 0.480 (0.755) 5.919 (2.660) 2.818 (3.258) < 0.001
Adult teeth caries
Decayed 0.022 (0.219) 0.100 (0.438) 0.055 (0.334) < 0.001
Missing 0.003 (0.214) 0.004 (0.151) 0.003 (0.189) < 0.001
Filled 0.002 (0.060) 0.014 (0.155) 0.007 (0.112) 0.004
DMFT 0.026 (0.313) 0.117 (0.492) 0.065 (0.403) < 0.001
Outcomes
Acute rheumatic fever; 31 (0.267) 50 (0.572) 81 (0.398) <0.001
n (col. %)
Chronic rheumatic heart 9 (0.078) 10 (0.114) 19 (0.093) 0.537
disease; n (col. %)
ARF or CRHD‡ ; n (col. %) 36 (0.311) 58 (0.664) 94 (0.462) < 0.001
Follow-up time (days) 1,865 (726) 1,852 (706) 1,861 (719) 0.224
*Mean (standard deviation) unless categorical variable, then n (column %). †Chi-square
test for categorical variables and Wilcoxon rank-sum test for continuous. ‡ARF: acute
rheumatic fever. CRHD: chronic rheumatic heart disease. dmft: sum of decayed, missing
(due to caries) and filled primary teeth. DMFT: sum of decayed, missing and filled
permanent teeth. sd: standard deviation.
11
Table 2 highlights the differences in the distribution of socio-demographic factors and oral
health status between children that went on to become cases and those who did not. Cases were
more likely to live in deprived neighbourhoods (P = 0.002), and were more likely to have higher
levels of decay in the primary (mean dmft 50% higher in cases than non-cases) and permanent
12
The Kaplan-Meier cumulative hazard, which compared children with <3 to ≥ 3 dmft, shows a
gradual departure of the risk of ARF with age, with clear separation between the two groups
after 12 years (figure 3). By the age of 13 years, the cumulative hazard of disease in the high
decay group (10.0 per 1000 per year; 95% confidence interval (CI): 7.0 to 13.1) was just over
twice the incidence in the low decay group (5.0 per 1000 per year; 95% CI: 3.0 to 7.0).
The c-statistic for the adjusted model was 0.67 (standard error: 0.031), which indicates
Table 3. Cox proportional hazard model showing the crude and adjusted
association between dental caries and the incidence of ARF or chronic rheumatic
heart disease (n = 20,333; number of events during follow-up = 94).
Table 3 outlines the crude and adjusted hazard ratios, derived from a Cox model. The
comparisons for continuous variables are illustrated for the difference between zero and five
primary teeth affected by caries. In the model, male gender and Pacific ethnicity (compared to
Māori) were associated with increased risk of disease, but neither association was statistically
significant.
13
Of the indices of caries, decay in the primary and permanent teeth were both significantly
associated with disease incidence in both the crude and adjusted models. After adjustment for
other covariates, a five unit increment in dmft was associated with a 57% increase in risk of
disease (95% confidence interval: 20 to 106%). The presence of caries on permanent teeth was
also significantly associated with disease incidence, with a five unit increase in caries associated
with a 170% increase in risk (95% confidence interval: 32 to 453% increase). A five unit change
in dmft was chosen as this roughly approximates the change in a binary variable, with five units
being approximately the 16th and 84th centiles of the distribution of dmft values.17
Although the model did show some departure from hazard proportionality (the P-value for the
cox.zph test for the ethnicity term was 0.03), the adjusted effect estimate and confidence
interval for total caries did not change, if separate baseline hazards were estimated for different
ethnic groups. This departure from the proportional hazards assumption showed that the
with time during follow-up. The use of restricted cubic splines as a check on the modelling
A population attributable risk calculation showed that factors that lead to decay of the primary
teeth were responsible for 22% of the disease burden. This compared to 2.0% for caries of the
permanent teeth, since these teeth were relatively caries free in the study population.
DISCUSSION
This study reports a positive association between dental caries and ARF. If this association is
assumed causal, and caries were entirely eliminated, the population risk calculation suggests
that there would be a 22% drop in ARF incidence. This figure assumes the fitted Cox model,
with its exponential increase in risk with number of decayed teeth, is a reasonable
representation.
14
The study has some limitations. We relied on hospital records, rather than physician review of
charts to determine ARF status at follow-up. We believe, however, that coding of ARF is likely to
be reasonably accurate. One review of the accuracy of hospital records, compared to a physician-
curated register, to identify cases of ARF in New Zealand showed a sensitivity of 88% (481/548),
but the number of false positives was high, returning a positive-predictive value of 64%
(483/749).18 We used the principal diagnosis to identify new cases of disease during follow-up to
many cases, radiographs were not undertaken, so that under-diagnosis was likely. Another
source of mis-measurement is that of a child’s socioeconomic status, which was taken as the
Residual confounding may also explain our results, since these findings are observational. It
may be that dental caries and sugar intake are a proxy for another condition. Sugar intake and
poverty are likely to closely co-exist. It may be that sugar intake represents an effect of residual
status by using the New Zealand standard (NZdep06), and after adjusting for this, the caries–
ARF link remained. This lack of attenuation of the strength of association between caries and
rheumatic fever, after adjustment for confounders, strengthens the evidence for a causal
interpretation. In an alternative analysis of the data,19 taken over all dental check-ups of
children as they age, the results were very similar to those reported here, which are based on
15
The main strength of the study is that it is a large, population-based sample with consistently
reported exposure and outcome measures that allowed us to achieve adequate statistical power
to detect an association.
The interpretation of the study findings is important. One is that dental caries is a proxy
indicator of exposure to sugar and that sugar is a direct cause of ARF by allowing pathogens
associated with ARF to grow. An alternative is that untreated caries itself may predispose to
rheumatic fever incidence. We believe the former explanation to be more likely. The correlation
coefficient between sugar intake and caries is high, 0.7 in one study.8 A narrative review of a
wide range of observational and ecological evidence supports the idea that sugar intake and
caries are strongly causally linked.7 Caries is therefore a reasonable surrogate measure of sugar
intake. The latter explanation we believe to be less likely as most lesions in the community oral
service data were recorded as ‘decayed’, and were likely to be rapidly treated subsequently
according to clinical protocols. The strong association between caries status and disease
In the introduction, we speculated that group A streptococcal infection is likely to mediate the
sugar-ARF link. Since we did not collect information about these infections, it was not possible
Since this study is confined to Māori and Pacific populations, the findings do not generalise to
NZ European children. It remains to be seen whether this link between caries and rheumatic
Two other studies have reported a higher incidence of caries in cases of rheumatic fever,
showed that 132 extractions due to caries had been carried out in the 100 cases, compared to 44
in the same number of controls.20 Gingivitis, a bacterial infection of the gums which is linked
16
with dental caries, was present in 33% of cases, compared to 4% of controls. The distribution of
age and socio-economic status was similar in the two groups. A second study, carried out in
Utah (n = 264), showed similar caries levels between cases and controls, although caries levels
were generally higher in cases, compared to controls (5 to 9 year olds; mean dmft: 7.2 in cases,
compared to 6.1 in controls).21 Other studies have reported the association between ARF, ethnic
group and socioeconomic deprivation.22 Indeed, the similarity between the risk factors for caries
Our study is consistent with other epidemiological analyses that show population-based
incidence of ARF is higher among Pacific people than Māori, although age-standardised reports
showed almost twice the incidence in Pacific compared to Māori. In this study, crude incidence
was 50% higher in Pacific people, with adjusted rates about 17% lower than that.1 The annual
incidence of new ARF or chronic rheumatic heart disease in our study was 92/100,000 per year;
higher than that reported by Jaine and colleagues in 2008,1 by ethnic group for all of New
Other studies have explored dietary links to rheumatic fever, but sugar has not been implicated.
Observational studies have reported an association between low body weight, low egg intake,
low serum albumin and prevalent rheumatic heart disease in the United States23, Yugoslavia24,
Zaire22 and Bangladesh.25 26 These results suggest that nutritional status is a risk factor for
disease onset.
In New Zealand, the recent focus of preventive effort for rheumatic fever has been early use of
antibiotics in high risk populations to treat Streptococcus pyogenes throat infection. A cluster
randomised trial, however, showed no convincing difference in disease incidence between the
schools who received the early antibiotic-based intervention and those that did not.27 Our study
suggests alternative dietary interventions may be useful. While we have seen a strong
relationship between caries and rheumatic fever incidence, these findings would be
17
strengthened by an independently conducted study using different methods, for example, a
Given the importance of sugar intake in the aetiology of dental caries,7 as well as its role in
childhood obesity and other morbidities, we advocate for sugar reduction in Māori and Pacific
populations. Dietary interventions which limit sugar intake could play an important part in
Acknowledgements
We thank Dr Satha Kanagaratnam and Gene Becconsall from Auckland Regional Dental Service
for their co-operation and work in obtaining and extracting the data. The authors also thank
Dean Papaconstantinou for his work in linking the dental data with hospital records. We also
thank Dr Julia Peters from Auckland Regional Public Health Service who supported this work.
Funding
REFERENCES
1. Jaine R, Baker M, Venugopal K. Epidemiology of acute rheumatic fever in New Zealand 1996-
2. Seckeler MD, Hoke TR. The worldwide epidemiology of acute rheumatic fever and rheumatic
3. Thornley S, Sundborn G, Schmidt-Uili SM. Rheumatic fever in New Zealand: what are the teeth
18
4. Chassy BM, Beall JR, Bielawski RM, et al. Occurrence and distribution of sucrose-metabolizing
5. Shelburne SA, Keith D, Horstmann N, et al. A direct link between carbohydrate utilization and
virulence in the major human pathogen group A Streptococcus. Proceedings of the National
6. Ministry of Health. Our Oral Health: Key findings of the 2009 New Zealand Oral Health Survey.
7. Sheiham A, James WP. A new understanding of the relationship between sugars, dental caries and
fluoride use: implications for limits on sugars consumption. Public Health Nutr
8. Okuya Y. The epidemiological study of the relation between caries incidence and sugar
9. Jamel H, Plasschaert A, Sheiham A. Dental caries experience and availability of sugars in Iraqi
children before and after the United Nations sanctions. Int Dent J 2004;54(1):21-25. doi:
10.1111/j.1875-595X.2004.tb00248.x
10. Takeuchi M. Epidemiological study on dental caries in Japanese children before, during and
11. Bolton P. Oral Health in ADHB and WDHB: Waitemata and Auckland District Health Boards,
2015.
12. Chang E, MacLeod R, Drake R. Characteristics influencing location of death for children with life-
13. Craig E, Adams J, Oben G, et al. The health status of children and young people in the northern
District Health Boards. Dunedin: NZ Child and Youth Epidemiology Service, 2011.
14. Thiébaut ACM, Bénichou J. Choice of time-scale in Cox's model analysis of epidemiologic cohort
19
15. Vander Hoorn S, Ezzati M, Rodgers A, et al. Estimating attributable burden of disease from
exposure and hazard data. Comparative quantification of health risks: global and regional
burden of disease attributable to selected major risk factors Geneva: World Health
Organization 2004:2129-40.
16. R Development Core Team. R: A language and environment for statistical computing. Vienna: R
2012;3:e111.
18. Moxon T, Lennon D, Reed P, et al. Is a rheumatic fever register the best way to evaluate the
19. Win Le SSE. Is dental caries a risk factor for acute rheumatic fever in New Zealand? Master of
20. Entine M. A survey of dental diseases as a diagnostic aid in rheumatic fever. J Am Dent Assoc
1949;38(3):303-8.
21. Wilcox EB, Greenwood DA, Galloway LS. Dental caries experience of a group of school children
at Ogden, Utah, with and without rheumatic fever. J Dent Res 1952;31(6):849-53.
22. Longo-Mbenza B, Bayekula M, Ngiyulu R, et al. Survey of rheumatic heart disease in school
23. Coburn AF. The concept of egg yolk as a dietary inhibitor to rheumatic susceptibility. Lancet
1960;1(7129):867-70.
24. Vlajinac H, Adanja B, Marinkovic J, et al. Influence of socio-economic and other factors on
20
25. Zaman MM, Yoshiike N, Rouf MA, et al. Association of rheumatic fever with serum albumin
concentration and body iron stores in Bangladeshi children: case-control study. BMJ
1998;317(7168):1287-8.
26. Zaman MM, Yoshiike N, Chowdhury AH, et al. Nutritional factors associated with rheumatic
27. Lennon D, Stewart J, Farrell E, et al. School-based prevention of acute rheumatic fever: a group
10.1097/INF.0b013e3181a282be
21
Figure legends
Figure 1. A plausible causal diagram which explains the nature of the association
between dental caries and acute rheumatic fever. Solid arrows indicate the
proposed direction of causation (but unobserved in this study; grey colour), while
the dashed arrow shows the observed (black) association, expected from the
unobserved causal relationships.
22
Figure 2. Study population flow.
23
Figure 3. Cumulative hazard of developing acute rheumatic fever or chronic
rheumatic heart disease during follow-up, comparing those with ≥ 3 primary teeth
affected by caries compared with those who had < 3, by age (years). 95%
confidence intervals surround the point estimate.
24