Analysis of Three Dental Fluorosis Indexes
Used in Epidemiologic Trials

Antonio Carlos PEREIRA

Departamento de Odontologia Social, Faculdade de Odontologia de Piracicaba, Universidade de Campinas, Piracicaba, SP, Brasil

Braz Dent J (1999) 10(1): 1-60 ISSN 0103-6440

| Introduction | Material and Methods | Results | Discussion | Conclusions | Acknowledgments | References |

The objective of this paper was to compare the DEAN, T-F and TSIF dental fluorosis indexes in relation to prevalence of surfaces, teeth and locality, and to verify the statistical correlation among them. The sample consisted of 461 schoolchildren, ages 12-14 years, born and reared in 3 cities in the State of São Paulo from 2 years of age. A total of 153 were from Cesário Lange with a fluoride concentration in the water supply of 1.4 ppm F, 142 from Piracicaba (0.7 ppm F) and 166 from Iracemápolis (<0.3 ppm F). The clinical examination was conducted after tooth brushing, using a plane mirror, artificial light and air drying of the teeth for 1 min. Premolars, second molars, and occlusal surfaces were the most severely affected. The three indexes showed similar percentages of children affected in the 3 three cities: 32.7%, 16.9% and 4.2% for the DEAN index, 33.3%, 17.6% and 4.2% for the T-F index and 32.7%, 16.9% and 4.2% for the TSIF index. There were no difficulties in using the three indexes in the field trials; thus the use of any one may be recommended in regions with similar fluoride concentrations to those of this study.

Key Words:  dental fluorosis, fluoride, DEAN index, T-F index, TSIF index.


    Two distinct groups of indexes have been proposed for measuring enamel defects: a) specific fluorosis indexes and b) descriptive indexes, including all types of defects. The specific fluorosis indexes are used for checking enamel defects due to the chronic, accumulated ingestion of fluoride.
    Dean (1934) developed a fluorosis index which classified individuals into 5 categories, depending on the degree of enamel alteration, and which was based on the identification of the 2 most severely affected teeth, giving ordinal numbers as the severity of the enamel alteration increased. This index was later modified by Moller (1982). Thylstrup and Fejerskov (1978) developed an index (T-F) based on the biological aspects of dental fluorosis, classifying individuals into 10 categories characterizing the macroscopic degree of fluorosis in relation to histological aspects. In the original classification, buccal, occlusal and lingual surfaces were examined. Horowitz et al. (1984) developed a fluorosis index based on aesthetic aspects of tooth surface (TSIF) classifying individuals into 8 categories. In this index a value is given for each anterior tooth surface not restored (buccal and lingual) and three values for posterior tooth surfaces (buccal, lingual and occlusal). These dental fluorosis indexes have been compared in several scientific articles (Table 1).
    Granath et al. (1985), comparing the DEAN and T-F indexes, concluded that the latter was more detailed and sensitive because it was based on biological aspects where there is an increase in hypomineralization with a simultaneous increase in the depth of the enamel surface in direction of the amelo-dentin junction. The DEAN index (1934) emphasized the aesthetic aspects of dental fluorosis.
    Thylstrup and Fejerskov (1978), in the study which gave rise to the T-F index, considered that the DEAN index was not capable of distinguishing the different aspects of the severity of dental fluorosis in areas of high fluoride concentrations in the water supply (above 3 ppm F). This conclusion was sustained by Wenzel and Thylstrup in 1982. Furthermore, they confirmed that the DEAN classification was capable of outlining the severity of dental fluorosis in populations residing in areas having less than 3 ppm F in the water supply.
    Burger et al. (1987), comparing the T-F and DEAN indexes in primary dentition in children of 1-9 years of age in Tanzania, reported that the two classification systems reproduced similar prevalence, but the severity value varied in the scales; generally the T-F index showed a higher score than the DEAN index. The authors concluded by recommending the T-F index for future field studies, due to the facility of use and better defined criteria.
    Cleaton-Jones and Hargreaves (1990) compared the three fluorosis indexes (DEAN, T-F and TSIF) in deciduous dentition, reporting that the prevalence of fluorosis in individual teeth was more frequently diagnosed with the T-F index. They concluded that the T-F index is the most indicated for work where detailed information about the problem is required.
    Mabelya et al. (1994), comparing the DEAN and T-F indexes, verified that the KAPPA statistical values (Landis and Koch, 1977) for inter-examiner errors were better for the T-F index. The DEAN index presented difficulties in evaluating questionable and very light categories.
    Using the DEAN and T-F indexes, Gaspar et al. (1995) examined 114 children (10-14 years of age) and found similar prevalence for both indexes, in locations with optimal and low fluoride concentrations in drinking water.
    The objective of the present study was to compare the DEAN, T-F and TSIF fluorosis indexes in terms of prevalence by surface, tooth and locality and to evaluate the statistical correlation between the indexes.

Material and Methods


    The sample for this study consisted of 461 schoolchildren, 276 males and 183 females (12-14 years of age), registered at public schools. A total of 153 children (73 males and 80 females) were from the city of Cesário Lange, which has a fluoride concentration of 1.4 ppm F in drinking water supplies; 142 children (81 males and 61 females) were from Piracicaba (0.7 ppm F) and 160 children (124 males and 42 females) were from Iracemápolis (<0.3 ppm F). The age range of 12-14 years was recommended by Dean and Arnold (1943), basically, because all the permanent teeth have erupted. The schoolchildren in the sample had been born and reared in the city from 2 years of age. Sample selection was done by means of a random stratified proportional from the list of schoolchildren provided by the directors of public schools, who had previously been fully informed about the study.
    The three cities chosen are all located in the State of São Paulo, Piracicaba being about 164 km from the State capital (Municipality of São Paulo), while Iracemápolis and Cerário Lange are about 20 and 95 km, respectively, from Piracicaba.
    The fluoride concentration in drinking water from these cities was determined using fluoride electrodes (model 96-09; ORION Research Inc., Boston, MA) with the reagent TISABâ (1.0 M acetate buffer, pH 5.0, containing 1.0 M NaCl; ORION). Apart from exposure to drinking water fluoride, only toothpaste was recorded.


    Fluorosis indexes used were those of Dean (1934), T-F (Thylstrup and Fejerskov, 1978) and TSIF (Horowitz et al., 1984).

Training and calibration

    A 2-month training programme was completed by the examiner. After which, the Kappa value for intra-examiner error (Landis and Koch, 1977) was >0.75. In addition, the examiner performed repeat examinations of a random 10% of the children. The diagnostic differentiation between light degrees of dental fluorosis and non-fluoridated opacities were carried out according the criteria of Russel (1961).

Clinical examination

    All the children were instructed on tooth brushing before the examination by a trained dental hygienist. The dental fluorosis examinations were carried out in a portable dental chair, with artificial light, a plane dental mirror and the teeth were dried with an air syringe for 1 min prior to the examination. The survey team was composed of an examiner, a recorder and a logistical human resource.
    The buccal, lingual and occlusal surfaces of the posterior teeth, and the buccal and lingual surfaces of the anterior teeth were examined. Erupted teeth with less than 2/3 of the crown (Opinya, 1991) or filled teeth (Horowitz et al., 1984) were excluded from the sample. The examination sequence was as follows: 1st phase: examination using the DEAN index with a one week interval; 2nd phase: examination using the T-F index with a one week interval; 3rd phase: examination using the TSIF index. The 1-week interval between examinations was made to minimize the “habits” (possible memorization) of clinical examinations.

Statistical analysis

    For calculating the prevalence of fluorosis, every individual that showed at least one surface with DEAN, T-F or TSIF index code greater than or equal to 1 (one) was defined as a “case of fluorosis”. Spearman correlation coefficient was calculated using SAS (SAS, 1987) software. The three indexes were correlated two by two, according to the locality researched.


    The percentages of affected surfaces were 14.3%, 8.1% and 1.2% for the DEAN index, while for the T-F index they were 15.1%, 8.4% and 1.4% and for the TSIF index they were 13.6%, 8.4% and 1.4% for the cities of Cesário Lange, Piracicaba and Iracemápolis, respectively (Table 2).
    The percentages of children affected in the three cities were 32.9%, 16.9% and 4.2%, for the DEAN index, 33.5%, 17.6% and 4.2% for the T-F index, and 32.8%, 16.9% and 4.2% for the TSIF index for Cesário Lange, Piracicaba e Iracemápolis, respectively (Table 3).
    The most severely affected teeth were the pre-molars and the 2nd molars in the cities of Cesário Lange and Piracicaba, while in the city of Iracemápolis the second molars were the most affected in both jaws.
    High Spearman correlation coefficient were seen in the three indexes used, mainly when the DEAN and TSIF indexes were compared (Table 4). The Kappa values for the three indexes were similarly very close (DEAN: 0.76; T-F: 0.79; TSIF: 0.78).


    In the distribution by surface, a constant pattern of greater prevalence in occlusal surfaces in the three indexes was noted. This is justified by the fact that diagnosis of mild forms of fluorosis is characterized by a chalk white aspect on cuspid tips. Because these surfaces were examined only on posterior teeth, this does not affect the aesthetics. In this study few aesthetically compromised surfaces were verified: two buccal surfaces of anterior teeth in Cesário Lange (codes D4, T-F5 and TSIF4), while in the cities of Piracicaba and Iracemápolis there were no aesthetically affected buccal surfaces.
    With regard to the prevalence of fluorosis per tooth, it was found that the most prevalent were the pre-molars, followed by the second molars in the cities of Cesário Lange and Piracicaba, in agreement with several researchers (Wenzel and Thylstrup, 1982; Horowitz et al., 1984; Burger et al., 1987; Gaspar et al., 1995). Moller (1982) reported maxillary incisors as being the most affected. In the city of Iracemápolis, the second molars were found to be the most affected.
    With regard to the prevalence per child (Table 3), a progressive increase in children affected was noted as the fluoride concentration in the water supplies increased. The T-F index showed a slightly higher non-significant prevalence than the DEAN and TSIF indexes for the cities of Cesário Lange and Piracicaba, which may be explained by intra-examiner diagnostic differences, since other researchers (Thylstrup and Fejerskov, 1978; Wenzel and Thylstrup, 1982; Granath et al., 1985; Burger et al., 1987) also reported small or negligible differences in prevalences among indexes.
    The standardization of data from different studies in fluoridated and unfluoridated areas was researched by Clark (1994), based on studies in the 1980s, affirming that in unfluoridated areas or those with low fluoride concentrations, prevalences varied from 4.4% to 55.0%, with 5 of the 8 studies showing prevalences above 20%. In this study, in the city of Iracemápolis a prevalence was observed of 4.2% below the estimate reported by Pendrys and Stamm (1990) of 10.0% and close to the estimate reported by the American Public Health Service (USPHS, 1991) of around 6.0%; however, these estimates were considered to be very conservative by Clark (1994). With regard to fluoridated areas, prevalences varied from 11.4% to 80.9%, with 7 of the 11 studies showing percentages above 25.0% (Clark, 1994), while Pendrys and Stamm (1990) and the American Public Health Service (USPHS, 1991), calculated prevalence estimates of around 23.0% and 22.0%, respectively, above those observed in this study of 17.0% in Piracicaba, however, very close to that observed by Gaspar et al. (1995), with 20.4%.
    It was noted that prevalences found in this study are similar to those of DEAN (1934; 1943) in the decades of the 30’s and 40’s, and lower than the prevalences in the decade of the 80’s (Segreto et al., 1984). These differences may be attributed to the wide use of fluorides from an early age, by topical methods applied in communities and fluoridated toothpaste, mouth washes, varnishes, and other dental products available in the USA and other developed countries, which have raised prevalence percentages.
    In Brazil and especially in the cities in this study, only initiatives of unsupervised preventive programs were observed, mainly after the creation of preventive procedures financed by the Unified Systems of Health, including brushing and fluoridated mouth washes after the children enter school (above 6 years of age), diminishing the probability of the appearance of dental fluorosis.
    The small differences observed among the three indexes may be explained by: a) definition of “case”, b) measuring methods, and c) intra-agreement.
    The problem with the definition of “case” comes up against some methodological aspects used specifically in each study. The DEAN index used the two teeth most severely affected by fluorosis for defining “case”. However, Kingman (1994) defined DI 2, differing from the T-F and TSIF indexes which used any value different from 0 (T-F and TSIF ³1), as the standard for measuring prevalence, being defined as T-F 1 e TSIF 1. However, owing to the bilateral character of dental fluorosis, normally DI 1 is equal to or very close to DI 2, so that in this study, in order to eliminate possible differences between DI 1 and DI 2 and differentiate the criterion for definition of “case” between the different indexes, we have opted for DI 1, T-F 1 e TSIF 1.
    The epidemiological determination of clinical conditions for estimating the prevalence of dental fluorosis may be subject to variations due to the use of different methods. Originally, the DEAN index used natural light and the teeth were examined humid, while the T-F index carried out prior prophylaxy, the examination being conducted with daylight and drying of the teeth for 1 min, whereas the TSIF used artificial light to examine the teeth humid.
    In this study, in order to standardize the methods, we have opted for brushing prior to examination, because the prophylaxy recommended by Thylstrup and Fejerskov (1978) would demand a great deal of time. In addition, drying of the teeth and artificial light were used.
    With regard to intra-examiner agreement, values of the KAPPA statistic of 0.76 for the DEAN index were similar to those reported by Riordan and Banks (1991). The KAPPA values of this study can be classified as “substantial” (P<0.01) according to the Landis and Koch (1977) classification.
    The most appropriate indications for each index may be the following: 1) DEAN index - comparative studies between current prevalences and those found in the decades of the 30’s and 40’s. 2) T-F index - clinical studies or analytical epidemiological studies. 3) TSIF index - studies in which an aesthetic basis is desired for defining “case” and it may be used where risk factors are identified or when the teeth may not be cleaned and dried.
    According to Kingman (1994), many final comments may be made concerning the use of the three indexes: 1) Steps in the methodology used for conducting the clinical examination are not reported, and this omission is particularly important with respect to drying and cleaning the teeth. 2) Many studies did not cite these error estimates, which makes it difficult to trust the data presented. The confidence of the examiners (intra-examiner error) is within an acceptable range in this study. 3) The definition of a “case” of fluorosis is very important and differs among researchers. It is indispensable for this to be established in a study. 4) Finally, all studies related to fluorosis should quote their prevalences and severity, thus facilitating understanding about the subject.


    1. The percentages of affected surfaces by dental fluorosis determined by each index were similar for each of the cities of Cesário Lange, Piracicaba and Iracemápolis.
    2. The most severely affected teeth were the pre-molars and the 2nd molars in the cities of Cesário Lange and Piracicaba, while in the city of Iracemápolis the second molars were the most affected in both jaws.
    3. There was a progressive increase in children affected as the fluoride concentration in the water supply of the three cities increased.
    4. The three indexes used showed a high Spearman correlation coefficient, mainly when the DEAN and TSIF indexes were compared.
    5. The three fluorosis indexes had similar prevalences using the same measuring methods for clinical examination.


    We gratefully acknowledge the financial support of CAPES-CNPq.


Burger P, Cleaton-Jones P, du Plessis J, de Vries J: Comparison of two fluorosis indices in the primary dentition of Tswana children. Community Dent Oral Epidem 15: 95-97, 1987

Clark DC: Trends in prevalence of dental fluorosis in North America. Community Dent Oral Epidem 22: 148-152, 1994

Cleaton-Jones P, Hargreaves JA: Comparison of three fluorosis indexes in a Namibian community with twice optimum fluoride in the drinking water. J Dent Assoc South Afr 45: 173-175, 1990

Dean HT: Classification of mottled enamel diagnosis. J Am Dent Assoc 21: 1421-1426, 1934

Dean HT, Arnold FH: Endemic dental fluorosis or mottled teeth. J Am Dent Assoc 30: 1278-1284, 1943

Gaspar MR, Pereira AC, Moreira BHW: Opacidades de esmalte de origem não fluorótica e fluorose dentária em áreas com baixa (0,2 ppm F) e ótima (0,7 ppm F) concentrações de flúor nas águas de abastecimento público. Revta Bras Odont 52: 13-18, 1995

Granath L, Widenhein J, Birkhed D: Diagnosis of mild enamel fluorosis in permanent maxillary incisors using two scoring systems. Community Dent Oral Epidem 13: 273-276, 1985

Horowitz HS, Driscoll WS, Meyers RJ, Heifetz SB, Kingman A: A new method for assessing the prevalence of dental fluorosis - the Tooth Surface Index of Fluorosis. J Am Dent Assoc 109: 37-41, 1984.

Kingman A: Current techniques for measuring dental fluorosis: issues in data analyses. Adv Dental Res 8: 56-65, 1994

Landis JR, Koch GG: The measurement of observer agreement for categorical data. Biometrics 33: 159-174, 1977

Mabelya L, König KG, van Palenstein Helderman WH: Comparison of two indices of dental fluorosis in low, moderate and high fluorosis Tanzanian populations. Community Dent Oral Epidem 22: 415-420, 1994

Moller IJ: Fluorides and dental fluorosis. Int Dent J 32: 135-147, 1982

Opinya R: Fluorosis of deciduous teeth and first permanent molars in a rural Kenyan community. Acta Odont Scand 49: 197-202, 1991

Pendrys DG, Stamm JW : Relationship of total fluoride intake to beneficial effects and enamel fluorosis. J Dent Res 69 (special issue): 529-538, 1990

Riordan PJ, Banks JA: Dental fluorosis and fluoride exposure in Western Australia. J Dent Res 70: 1022-1028, 1991

Russel AL: The differential diagnosis of fluoride and non-fluoride enamel opacities. J Pub Hlth Dent 21: 143-146, 1961

Segreto VA, Collins EM, Camann D, Smith, CT: A current study of mottled enamel in Texas. J Am Dent Assoc 108: 56-59, 1984

SAS Institute: SAS/STAT Guide for Personal Computers. SAS Institute, Cary 1987

Thylstrup A, Fejerskov O: Clinical appearance of dental fluorosis in permanent teeth in relation to histologic changes.
Community Dent Oral Epidem 6: 315-328, 1978

USPHS (United States Public Health Service): Report on fluoride benefits and risks: Review of fluoride benefits and risks. Report of the Ad Hoc Subcommittee on Fluoride. 1991

Wenzel A, Thylstrup A: Dental fluorosis and localised enamel opacities in fluoride and non-fluoride Danish communities. Caries Research 4: 340-348, 1982

Correspondence: Dr. Antonio Carlos Pereira, Departamento de Odontologia Social, Faculdade de Odontologia de Piracicaba, UNICAMP, Av. Limeira, 901, 13414-018 Piracicaba, SP, Brasil. Email:

Accepted October 17, 1998
Electronic publication: September, 1999