Decline in dental caries among 12-year-old children in Brazil, 1980–2005

Abstract

Objectives: This study was conducted to determine changes in values on the decayed, missing and filled teeth (DMFT) index in 12-year-old children in Brazil between 1980 and 2005, and to correlate DMFT values with human development index (HDI) values, time, population size of municipality and fluoridation of the water supply. Methods: The present study represents a retrospective ecological study using secondary data from epidemiological surveys published in indexed journals, as well as data obtained from epidemiological official surveys carried out in Brazil in 1986, 1996 and 2003, and in the State of São Paulo in 1998 and 2002. Units of study were represented by Brazilian municipalities holding average DMFT index values for 12-year-old children. Multiple regression analysis was used to examine the correlations among DMFT and HDI values, and time, population size and fluoridation of the water supply. Results: The final database included 550 records of DMFT values in 428 different towns. Regression analysis showed statistically significant correlations between DMFT index values and time (P < 0.001), fluoridation of the water supply (P < 0.001) and size of municipality (P < 0.001). Estimated mean DMFT index values were 8.36 in 1980, 6.08 in 1985, 4.45 in 1990, 3.29 in 1995, 2.46 in 2000 and 1.86 in 2005. Conclusions: Data showed a significant decrease in dental caries across the entire country, with an average reduction of 25% occurring every 5 years. General trends indicated that a reduction in DMFT index values occurred over time, that a further reduction in DMFT index values occurred when a municipality fluoridated its water supply, and mean DMFT index values were lower in larger than in smaller municipalities.

INTRODUCTION

Dental caries is not a new subject of investigation¹. The decline in the prevalence of dental caries has been well documented. Sound epidemiological data showing dental caries to be in decline led to the First International Conference on the Declining Prevalence of Dental Caries in 1982. At that time, speakers from Denmark, Ireland, the Netherlands, New Zealand, Norway, Sweden, the UK and the USA were concerned about the availability of results, but agreed on the presence of an ongoing decrease in dental decay². Subsequently, additional international conferences were held3., 4., 5., 6. and many other published papers showed this decline to be occurring in industrialised as well as developing countries7., 8., 9., 10., 11., 12.. This new pattern in the prevalence of dental caries has driven researchers to focus on the development of new epidemiological tools for detecting this disease⁷.

In 2004, Marthaler speculated on reasons for the decline in caries. The use of fluorides is by far the most important associated factor. The effectiveness of the fluoridation of public water supplies was well documented up to 1980, but in recent years the effective use of topical fluorides in toothpaste and in products that are applied topically to the teeth has been clearly demonstrated¹³. Kumar¹⁴ considered that the fluoridation of water represented the best of the available strategies for combating dental caries across a range of countries. The present challenge refers to the documenting of the current impact of the fluoridation of water supplies on rates of dental caries.

In Brazil, water fluoridation started in 1953, but was not subject to federal legislation until 1974, since when many municipalities have fluoridated public water supplies9., 10.. In 1989, dentifrices were compelled by law to be fluoridated¹⁵. These public policy measures are normally cited as the main reasons for the decline in dental caries in the country16., 17..

Many epidemiological surveys have been conducted in Brazil, in part because it is a country of continental dimensions and its population is subject to great inequalities in income. In addition, decay is not uniformly distributed, which reinforces the need for extensive studies that cover all regions and social classes. The first epidemiological surveys of dental caries in Brazil were conducted between the 1950s and 1980s. In general, population-based surveys have been limited in scope and location, and in their widest application have tended to be regional18., 19., 20..

Two types of epidemiological survey have been conducted in Brazil. The first type refers to large surveys of national or regional scope, such as those coordinated by the Brazilian Ministry of Health (which conducted epidemiological surveys in 1986²¹, 1996²² and 2003²³) or by health departments at state level (such as those conducted in the State of São Paulo in 1998²⁴ and 2002²⁵). The second type includes the many smaller surveys carried out mainly by academics.

Peres et al.¹⁷ conducted an extensive survey to investigate decayed, missing and filled teeth (DMFT) index values among 12-year-olds in all municipalities of the State of São Paulo. The authors invited the 625 municipalities of the state to participate, achieving a response rate of 20% (125 municipalities). Mean DMFT values in the municipalities ranged from 1.3 to 13.6, with an average weighted by state population equal to 4.8. Small municipalities, but in larger quantity, represented a smaller portion of the population of the state and showed poorer DMFT values.

In 2004, Abreu et al.²⁶ performed a systematic review of publications on dental caries in rural populations in Brazil, in which the authors retrieved articles from the MEDLINE, LILACS [Literatura Latino-Americana e do Caribe em Ciências da Saúde (Literature in the Health Sciences in Latin America and the Caribbean)] and BBO [Portal de Revistas de Odontologia (Brazilian Bibliography of Dentistry)] databases. Initially, 32 studies were identified; after critical analysis, only nine papers were selected²⁶. The authors realised that the adoption of any criterion with a possible bias was sufficient to have excluded all but two of these papers. This study shows the non-standardisation and lack of specific methodological criteria in most of the surveys. Thus, the authors recommended that findings of a review of this nature should be interpreted with caution²⁶.

In 2004, using data from 131 cities and towns obtained in a survey performed in 1998 in the State of São Paulo, Antunes et al.²⁷ analysed the inequalities in the distribution of dental caries in the study population. The results showed a strong correlation between the prevalence of dental caries and socioeconomic status, and confirmed a trend towards polarisation in the distribution of DMFT in 12-year-old children²⁷.

The present research was motivated by the diversity of data on DMFT values in children aged 12 years and their evolution over time. Thus, this study aimed to assess the development of DMFT values in this age group based on retrospective data obtained in epidemiological surveys conducted using the DMFT index as a measure of dental caries in Brazil, from 1980 to 2005, and to relate DMFT data to human development index (municipal) (HDI-M) values, time, population size of the municipality, and fluoridation of the water supply.

MATERIALS AND METHODS

This is a retrospective ecological study based on secondary data obtained in epidemiological surveys published in indexed journals and official data obtained in epidemiological surveys conducted in Brazil in 1986, 1996 and 2003 (Ministry of Health), and the State of São Paulo in 1998 and 2002 (State of São Paulo Department of Health). Units of study were Brazilian municipalities with mean DMFT data for 12-year-old children.

A municipality in Brazil is an autonomous population entity with a government elected by its inhabitants, located in an area incorporated into a state. Brazil includes a total of 5,565 municipalities in 26 states and one federal district.

The search for papers reporting epidemiological surveys of dental caries using the DMFT index was performed using the MEDLINE, LILACS and BBO databases. To ensure that the information fulfilled certain quality-related criteria, three types of paper were not accessed: articles published in non-indexed journals; academic works (dissertations, theses, etc.), and epidemiological surveys conducted by municipal and state public administrators without disclosure to the scientific community. The terms ‘DMFT’ and ‘DMF’ (decayed, missing, filled) were used as keywords, limited by the word ‘child’. These terms were used to retrieve items on dental caries (DMFT) in children aged 12 years. The inclusion criteria required papers to describe a survey in Brazil that obtained mean DMFT values in children aged 12 years at the municipality level. Exclusion criteria ruled out papers that included data obtained prior to 1980, data not specific to 12-year-old children, data not reported at municipality level, the use of any dental caries index other than the DMFT index based on Klein and Palmer (1937)²⁸, and data described in previous papers already identified in the present research.

Official data on mean DMFT index values, by municipality, were obtained from the following surveys:

• •Oral Health National Survey 1986: the report for this survey did not disclose data on mean DMFT values by municipality because its purpose was to describe the data by region (north, northeast, southeast, south and centre–west) and for all of Brazil²¹; data by municipality were calculated from the database, which included primary data by individual, directly by the survey coordinator, Professor Vitor Gomes Pinto
• •Oral Health National Survey 1996: data by municipality were calculated from the database (www.datasus.gov.br), which included primary data by individual²²
• •Oral Health National Survey 2003 (SB Brazil 2003): the report for this survey gave results by region and size of municipality; data by county were calculated from the database (www.saude.gov.br), which included primary data by individual²³
• •Oral Health São Paulo State Survey 1998: the results of this oral health survey were released by the Regional Health Division of the State of São Paulo²⁴; data for each city were calculated from the database, which included primary data by individual, by the State Health Department of São Paulo (SES-SP)
• •Oral Health São Paulo State Survey 2002: data by city were taken directly from the report ‘Oral Health Conditions in the State of São Paulo in 2002’²⁵.

A municipality was deemed to have a fluoridated water supply if fluoridation had been in place for ≥5 years prior to the start of data collection.

The population of the municipality was identified as the number of inhabitants in the year of the survey. Population information was obtained from the website www.datasus.gov.br.

Many analytical studies have sought to relate DMFT values with socioeconomic indicators, including the HDI-M, an index that combines life expectancy, educational attainment and income at municipality level. The HDI-M index was created by the United Nations Development Programme (UNDP). Values on this index range between 0 and 1; a score of 0 indicates no human development and a score of 1 indicates the presence of total human development. Data for the 2000 HDI-M index were accessed at the website http://www.pnud.org.br. This page is maintained by the UNDP²⁹.

Multiple regression analysis was used to link HDI-M value, time, population size of municipality and fluoridation of the water supply to the mean DMFT value for each municipality, with the DMFT value as the dependent variable. The regression analysis used a backward stepwise model in which all variables are included in the initial model and the most non-significant variable is removed in each step. A significance level of 5% (P < 0.05) was adopted.

An attempt was made to fit the multiple regression model to a linear model. The coefficient of determination (R²) increased substantially when an exponential effect was adopted for the variable of time. For the variables of fluoridation, population size and HDI-M value, a linear effect was adopted.

Because sample sizes differed among the various epidemiological surveys, and sample size is related to the accuracy of the DMFT value calculated, the sample size of each survey was used as a weighting variable. Sample size was taken into account because the larger the sample size, the more accurate the mean DMFT value. The sample size was used in the loss function of the regression analysis; thus the largest sample size has the biggest influence on the loss function, making it a weighted loss function.

All calculations were performed using statistica Version 7.0 (StatSoft, Inc., Tulsa, OK, USA).

RESULTS

The search for articles reporting DMFT data in 12-year-olds by municipality generated an initial list of 670 articles, of which 358 were retrieved from LILACS, 222 from BBO and 90 from MEDLINE. When these articles had been filtered according to the study inclusion and exclusion criteria, only 67 remained. The large reduction in the initial number of papers occurred for several reasons: papers did not include data specific to the age group of interest; papers were based on primary data from official surveys; papers used data sourced from other articles, and papers used indices other than the DMFT index.

These 67 articles contributed 111 records of DMFT index values according to municipality. The national surveys of 1986, 1996 and 2003 accounted for 292 records (16 from 1986, 27 from 1996, 249 from 2003). The official survey of the State of São Paulo contributed 147 records, of which 131 referred to the 1998 survey and 16 to the 2002 survey.

The final database comprised 550 records of DMFT values from 428 different municipalities (Table 1). Although the data had been collected by municipality, and any single epidemiological survey might cover several municipalities, to facilitate the use of terminology, municipal surveys were referred to as ‘epidemiological surveys’.

Table 1.

Survey distributions by Brazilian state and geographic region in relation to population per region

State	Surveys per state		Region	Surveys per region		Population per region, %
	n	%		n	%
Distrito Federal	5	0.9	Central–west	78	14.2	6.9
Goiás	27	4.9
Mato Grosso do Sul	12	2.2
Mato Grosso	21	3.8
Tocantins	13	2.4
Acre	5	0.9	North	47	8.6	7.6
Amazonas	12	2.2
Amapá	3	0.6
Pará	20	3.6
Rondônia	5	0.9
Roraima	2	0.4
Alagoas	5	0.9	Northeast	74	13.5	28.1
Bahia	17	3.1
Ceará	7	1.3
Maranhão	7	1.3
Paraiba	9	1.6
Pernambuco	15	2.7
Piauí	8	1.5
Rio Grande do Norte	3	0.6
Sergipe	3	0.6
Paraná	19	3.5	South	65	11.8	14.8
Rio Grande do Sul	26	4.7
Santa Catarina	20	3.6
Espírito Santo	3	0.6	Southeast	286	52.0	42.6
Minas Gerais	35	6.4
Rio de Janeiro	7	1.3
São Paulo	241	43.8
Total	550	100	Total	550	100	100

Table 1 shows the distribution of the 550 records according to state, geographic region, and the proportions of surveys and Brazilian population in each region.

Mean DMFT values by year of epidemiological survey as obtained from the 550 records are shown in Figure 1.

Figure 1.

Distribution of mean decayed, missing and filled teeth (DMFT) index values from 550 epidemiological surveys, by year of data collection.

Overall, 53.3% of records referred to cities in which the water supply was fluoridated and 46.7% referred to cities without a fluoridated water supply.

Of the total 550 records, 24 did not give information on the number of children in the sample. The 526 records that did include this information referred to a total of 72,637 children.

Almost a third (32.4%) of the records came from municipalities with fewer than 10,000 inhabitants. More than half (55.5%) of the epidemiological surveys had been conducted in municipalities in which HDI-M values ranged from 0.700 to 0.799, which is considered to indicate a range of medium human development, reflecting the general profile of Brazil.

Municipalities were classified in size according to the criteria used by the SB Brazil 2003 epidemiological survey as having: (i) ≤ 5,000 inhabitants; (ii) 5,001–10,000 inhabitants; (iii) 10,001–50,000 inhabitants; (iv) 50,001–100,000 inhabitants, and (v) > 100,000 inhabitants.

The results of the regression analysis of the four independent variables showed statistically significant relationships with DMFT index values. There was also a statistically significant correlation between HDI-M values and the size of the city (r = 0.49, P < 0.001). When HDI-M was removed from the model, the overall coefficient of determination (R²) did not change and therefore HDI-M was not retained in the final model (Table 2).

Table 2.

Multiple regression analysis using the mean decayed, missing and filled teeth (DMFT) index value as the dependent variable by municipality

Source	B-value		P-value	R-value	R2-value	P-value
	Estimated value	Standard error
Constant	2.708	0.049	<0.000*	0.73	0.54	< 0.001*
Time (year)	7.974	0.039	<0.001*
	−0.067†	0.001
Water fluoridation, 0 = no; 1 = yes	−0.940	0.009	<0.001*
Municipality size‡	−0.347	0.004	<0.001*

^*Statistically significant (P < 0.05).

^†Power constant.

^‡Population sizes: ≤ 5,000, 5,001–10,000, 10,001–50,000, 50,001–100,000, > 100,000 inhabitants.

The final regression function was:

$\begin{array}{cc}\mathrm{DMFT}\hfill & \mathrm{=}2\mathrm{,}708\mathrm{+}7\mathrm{,}974\mathrm{\times }{\mathrm{e}}^{\mathrm{-}0\mathrm{.}067\mathrm{\times }\mathrm{time}}\mathrm{-}0\mathrm{.}94\hfill \\ \hfill & \mathrm{\times }\text{fluoride}\mathrm{-}0\mathrm{.}347\mathrm{\times }\text{municipality size}\mathrm{.}\hfill \end{array}$ The residual analysis of estimates showed that the adopted model met the basic assumptions of the regression analysis; thus the adopted model was considered valid for data analysis.

Using the function obtained by regression, the evolution of DMFT index values in Brazil was estimated (Figure 2). In addition to estimates, Figure 2 shows the values obtained in the general surveys of Brazil in 1986, 1996 and 2003.

Figure 2.

Curves showing mean estimates for decayed, missing and filled teeth (DMFT) index values in Brazilian municipalities by year, population dimension and presence of public water fluoridation.

Using the function, it was possible to quantify the reductions in DMFT index values over the 25 years to 2005. For this estimate a standard medium-sized municipality (50,001–100,000 inhabitants) with a fluoridated water supply was used. The results are shown in Table 3.

Table 3.

Estimated reductions in mean decayed, missing and filled teeth (DMFT) index values between 1980 and 2005

Period	Estimated DMFT value
	Start of period	End of period	Reduction	Reduction, %
1980–1985	8.36	6.08	2.28	27.3
1985–1990	6.08	4.45	1.63	26.8
1990–1995	4.45	3.29	1.16	26.1
1995–2000	3.29	2.46	0.83	25.2
2000–2005	2.46	1.86	0.60	24.4

DISCUSSION

This is an ecological study and is thus weak for inferring a cause-and-effect relationship. It should also be noted that ecological studies are valid for the population as a whole, but any particularisation for individuals may be false²⁷.

The only quality assessment criterion applied to articles to be included in this review related to whether the research had been published in an indexed journal. Therefore, this paper cannot be considered a systematic review. If we had adopted typical criteria for selecting articles for a systematic review, almost all of the papers referred to here would have been excluded from the present research.

All estimates described here should be viewed with caution because several factors not assessed in this study may influence the incidence of dental caries and, consequently, DMFT index values; however, from the present findings, it is possible to obtain an overview of the evolution of dental caries in the 25 years to 2005 in Brazil.

Of the 550 surveys identified in this study, 63.8% were carried out in the southern and southeastern regions of the country; 57.4% of the Brazilian population lives in these regions. These regions are the most developed in the country. Furthermore, 22.0% of the surveys referred to the northern and northeastern regions, which are less developed and which are home to 35.7% of the population. As the prevalence of dental caries is highly variable from municipality to municipality, the concentration of information in a single region indicates that more epidemiological surveys, mainly in economically less privileged areas of Brazil, are required. This can potentially cause a bias in the estimate and thus a socioeconomic index (HDI-M) was used to minimise any possible bias.

In terms of the distribution of surveys over time, the years showing the highest occurrence of surveys are coincident with the years of public surveys. The largest survey to be conducted in the country was performed in 2003. The second highest peak in the administration of surveys occurred in 1998, when a large survey was conducted in the State of São Paulo²⁴. As these surveys were planned on a regional basis, our samples represent all regions for the cited years. Because government-coordinated national surveys employed the same methodology and calibrated examiners and the selection of samples, we believe that estimates of DMFT values are more accurate for the years to which these surveys refer. The increased number of surveys in specific years (reflecting national surveys) had an unavoidable effect on the results of the regression analysis, leading to a better estimate of DMFT index values.

In a study of the national health survey of 1996, Moysés³⁰ found a correlation between HDI-M and DMFT index values; however, municipalities of greater size and with fluoridated water supplies usually generate higher HDI-M values and, consequently, the influence of HDI-M in this study may have been minimised by the higher correlation between these factors.

The outcome of the coefficient of determination shown in Table 2 (R² = 0.54) means that the model proposed here explains 54% of the variation in DMFT index values. In other words, 54% of the variation in DMFT index values among municipalities can be explained by the year of assessment, water fluoridation status, and population size of the municipality.

The regression analysis indicates a statistical relationship between fluoridation of the water supply and DMFT index value, with a mean difference of 0.94 in DMFT values in favour of municipalities with public water fluoridation compared with those without. A reduction also occurred over time in cities with non-fluoridated supplies; this may reflect better living conditions, participation in education and prevention programmes, and the use of topical fluorides among these populations. Several studies reported in the literature have shown the influence of fluoride in reducing DMFT index values. Bastos et al.³¹ concluded that the fluoridation of water supplies in Brazil was especially crucial in the 1980s and 1990s in reducing prevalences of dental caries. In an evaluation of caries in more recent periods, the influence of fluoridated water was considered to have possibly declined as the use of other fluoridated products such as dentifrices rose³².

Population size also impacted mean DMFT index values: the larger the size of the municipality, the lower the mean DMFT value. Time was also negatively related to the DMFT index, indicating a trend towards a reduction in DMFT index values over the years. Clearly, neither time nor population size have a direct cause-and-effect relationship with DMFT index values. Other factors not assessed in these surveys, including oral health education, access to dental treatment and the provision of information on hygiene in general, among others, have improved in recent years, and the fact that the inhabitants of larger cities have greater access to these factors also underlies the relationships between time and size of municipality with DMFT values.

The mean DMFT index value obtained in the 1986 survey lies in the centre of the curves (Figure 2); this survey was conducted in 16 municipal capitals, most of which did not have fluoridated water supplies. The mean DMFT index value obtained in the 1996 survey is close to the curve for municipalities of > 100,000 inhabitants with fluoridated water supplies. The survey was conducted in 27 cities, about half of which had water fluoridation. However, the mean DMFT index value obtained in the 2003 survey is located near the centre of the curves, and is a little closer to the curves for medium and large cities, which are where most of the Brazilian population live. The values obtained in the national surveys were compared with the estimated curves to determine the feasibility of the curves. It should be emphasised that the regression analyses used data from the national surveys and thus much of the information used in the construction of these curves was contributed by these surveys (Figure 2).

The average annual reduction in the absolute DMFT value has fallen in recent years. This is expected because a decline in the absolute DMFT value also decreases the possibility of reduction. However, it is interesting to note the consistency in the percentage reduction of DMFT values, which show an average reduction of about 25% every 5 years from 1980 (Table 3).

Notwithstanding the drawbacks of the present study’s design, we conclude that historical data for DMFT index values in children aged 12 years show a reduction in the prevalence of dental caries across the entire country. Statistically significant relationships between DMFT index values and time, fluoridation of water supply and municipality size were apparent.

This analysis of the last national surveys to be conducted in Brazil is also intended to contribute to discussions about the performance of new national epidemiological surveys.

Acknowledgements

The authors wish to thank Professor Vitor Gomes Pinto, Coordinator of the Health Knowledge Unit of SESI (Industry Social Services), Brasília, Federal District, Brazil, for the provision of primary data from the Brazilian Dental Caries National Survey 1986, which hugely aided this research.

Conflicts of interest

None declared.