SUBJECTS, MATERIALS AND METHODS

TOP ABSTRACT SUBJECTS, MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Subject selection. We drew the subjects for this population-based cohort study from the Washington Dental Service (WDS) enrollees; WDS is a large, not-for-profit dental insurance company in the Northwest. The study was limited to the 1.5 million WDS enrollees living in Washington state, approximately 25 percent of the state’s population. We considered for inclusion all insured patients who reached the age of 15 years between Jan. 1, 1993, and Dec. 31, 1998. Our primary inclusion criterion was five years of continuous coverage by WDS subsequent to each subject’s 15th birthday. Other criteria were the availability of sex and demographic information. WDS computer programmers selected subjects and performed all data retrieval. For subjects who met the inclusion criteria, the programmers extracted the following data from WDS’ data warehouse: unique anonymous subject identification number, sex, birth date, all dental treatment during the study period (1993–2003), unique anonymous dental provider number and provider specialty. The Institutional Review Board of the Human Subjects Division at the University of Washington approved this study.

Exposure and outcome data. The primary exposure of interest was the first occurrence of third-molar extraction after the age of 15 years. Third-molar extraction was identified by tooth nos. 1, 16, 17 and 32 in conjunction with Current Dental Terminology (CDT) extraction codes 7110 through 7241. To indicate the level of difficulty, the extraction codes above were divided into five categories:

– Level 1: 7110 and 7120—routine extraction of erupted tooth;

– Level 2: 7210 and 7220—extraction of erupted or impacted tooth requiring mucoperiosteal flap;

– Level 3: 7230—extraction of impacted tooth, partial bony;

– Level 4: 7240—extraction of impacted tooth, completely bony;

– Level 5: 7241—extraction of impacted tooth, completely bony with unusual surgical complications.

The outcome of interest in this study was the first indication of TMD after age 15 years. The CDT codes used to identify TMD ranged from 7810 through 7899 and included TMD examinations, TMD splints and other TMD therapies such as occlusal adjustments, arthroscopy and other surgical procedures.

Confounding variables. We considered several variables to be potentially confounding variables; they included sex, extraction of other teeth, orthodontic treatment and dental care utilization. We considered extractions of teeth other than third molars to be possible confounders and categorized them by tooth type into one of three groups: anterior teeth, premolars, and first and second molars. We identified orthodontic treatment during the study period using a combination of CDT codes and WDS codes that had been investigated in a previous study.⁹ We used the number of prophylaxis claims from age 15 years through age 19 years as an indicator of dental care utilization (CDT codes 1110 and 1120).

Additionally, we investigated factors associated with extraction of third molars, such as training of provider (general practitioner or oral surgeon), type of anesthesia under which the procedures were performed (local anesthesia, nitrous oxide/conscious sedation, intravenous sedation or general anesthesia), arch of extraction (maxillary or mandibular), and complexity of extraction (indicated by the five categories described above), as variations in the exposure. Finally, we investigated the period of risk after third-molar extraction by calculating the relative risk (RR) of experiencing TMD in each year subsequent to exposure.

Statistical analyses. After performing descriptive statistics, we analyzed the data using a survival design with third-molar extraction as a time-varying exposure and TMD as the outcome. We placed all of the subjects in the unexposed cohort on their entry into the study. We transferred those subjects who underwent subsequent third-molar extraction to the exposed cohort at the time of the third-molar extraction. Thus, all subjects contributed person-time data to the appropriate cohort throughout the study period.

Subjects exited the study when the first of three conditions occurred: a TMD outcome was identified, their period of continuous enrollment ended (five-year minimum) or the end of the study period was reached (December 2003). We estimated the hazard ratio for exposed versus unexposed subjects, which is equivalent to the RR, by fitting a Cox proportional hazards model with age as the time scale and exposure as a time-varying covariate. We performed all of the statistical analyses using a statistical software package (SAS, Version 9.1, SAS Institute, Cary, N.C.). We also analyzed potentially confounding variables and considered them in adjusted models. We investigated parameters related to third-molar extraction (for example, number of third molars removed, type of anesthesia under which procedures are performed) further using Cox proportional hazards regression models. We established a significance level of P < .05 for all of the tests. A final area of investigation was temporal patterns of risk in the years after third-molar extraction. We added indicator variables for years since exposure to the Cox model, allowing the RR (exposed/unexposed) to vary with years since exposure.

	RESULTS

TOP ABSTRACT SUBJECTS, MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
From a total of 281,074 WDS enrollees reaching the age of 15 years between 1993 and 1998, 35,479 subjects (13 percent) met the inclusion criteria of five years of continuous enrollment after turning 15 years of age. Of these subjects, 988 were dropped due to missing sex data, leaving 34,491 subjects, 49 percent of whom were female. Slightly more females than males had third molars removed during the study period, and females were more likely to develop TMD (Table 1). The exposure was common, with 50 percent of all subjects undergoing one or more third-molar extractions during the study period. Dental care utilization, as measured by prophylaxis visits, was slightly higher in subjects who had undergone third-molar extraction or who had experienced TMD. The mean age at third-molar extraction was 18.2 years, and the mean age at TMD onset was 18.7 years.

Survival analyses. The crude RR using the Cox proportional hazards model was 2.1, indicating a significantly increased risk of experiencing TMD in subjects who had their third molars extracted (Table 3). We found that sex and dental care utilization both were associated with an increased risk of experiencing TMD and, therefore, we added them to the adjusted models. When this was done, the RR decreased to 1.6, but it still was statistically significant. In secondary analyses, removing the 32 subjects with histories of TMD examinations but no further TMD treatment did not alter the risk estimate. We found that orthodontic treatment was not related to TMD (crude RR = 0.8), and, therefore, we did not include it in further analyses.

Variations in exposure. To assess parameters associated with third-molar extraction, we investigated the following factors: level of provider training, type of anesthesia under which the procedures were performed, arch of extraction and difficulty of extraction. None of these parameters exhibited significantly increased RRs for TMD (Table 3).

Temporal pattern after exposure. When we adjusted for sex and dental care utilization, we found that the risk of experiencing TMD in the first three years after third-molar extraction was significantly elevated, but there was no pattern of increasing or decreasing risk during that time. During the fourth and fifth years after exposure, the risk continued to be elevated, but not significantly (Table 3).

	DISCUSSION

TOP ABSTRACT SUBJECTS, MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
We noted a 60 percent increased risk of experiencing TMD in adolescents and young adults who had their third-molars removed, after we adjusted for sex and dental care utilization. Thus, the hypothesis of third-molar extraction as a risk factor for TMD is supported by our investigation. While the increase in RR was modest, the high frequency of third-molar extraction also must be considered in weighing the risk to the population. Approximately 50 percent of this insured population had third molars removed by the age of 20 years. This translates into a population attributable risk of 0.23, meaning that 23 percent of the total incident TMD cases in this age group might be due to third-molar extraction.

Our retrospective cohort study cannot prove that third-molar extraction is causal for TMD. However, it is useful to consider how the evidence for third-molar extraction as a risk factor for TMD fits with criteria for causality. First, biological plausibility should be considered. Various types of facial trauma have been reported as risk factors for TMD,^10–15 and oral surgery textbooks acknowledge the potential for trauma to the TMJ during extraction.^5,6 Therefore, this criterion appears to be met.

Strength of the association also should be addressed. Although our study only found a moderate increase in RR, it is based on a large, longitudinal sample with unbiased ascertainment. Molar extraction has been reported as a risk factor for TMD in other articles.^10,16,17 Threlfall and colleagues¹⁸ did not find a statistically significant association between disk displacement and third-molar extraction. However, the odds ratio was elevated (1.28), and the confidence interval (CI) was suggestive of a relationship (95 percent CI, 0.96 to 1.71). Their study assessed TMD in a much older sample (ages 18–83 years, with 72 percent of subjects older than 30 years) than our study, and their subjects may have experienced other risk factors, such as other forms of facial trauma, stress and psychosocial issues. In the relatively young population we assessed in our study, third-molar extraction may have played a greater role in the etiology of TMD, while other competing risk factors may have become more important in adulthood. We are assessing the relationship between third-molar extraction and TMD in another population, which includes subjects of all ages.

There was a 60 percent increased risk of experiencing temporomandibular disorder in adolescents and young adults who had their third molars removed.

Our study accounted for temporal sequence by identifying the dates of extraction and TMD outcomes. Thus, we considered only outcomes that occurred after third-molar extraction to be associated with the exposure. We found the risk of experiencing TMD to be significantly elevated in the first three years after third-molar extraction, after adjusting for sex and dental care utilization, which lends support for causality. Some may propose that third molars are removed as a treatment for TMD, thus confounding the relationship we investigated. However, it has been reported that TMD as an indication for third-molar extraction is rare, occurring in only 0.1 percent of approximately 10,000 subjects seeking third-molar extraction.¹⁹

We investigated dose-response—the final criterion for consideration—by analyzing parameters associated with extractions. We found no significant associations due to provider training, type of anesthesia under which the procedures were performed, arch of extraction or severity of impaction. This, however, may have been the result of a relatively uniform exposure, as 83 percent of all extractions were performed by oral surgeons, and of these, 92 percent were performed under general anesthesia, 85 percent had all four third molars removed at one time, and 91 percent were partial or full-bony impactions. Thus, the power to discriminate between these extraction parameters was limited.

The annual incidence of TMD in this population was 0.2 percent, which is low compared with the range that has been reported in the literature (0.7–18 percent).²⁰ There are several factors, however, which could explain this discrepancy. Many definitions of TMD have been used in the literature, and some are based on signs rather than symptoms. Also, the outcome in our study required that a subject sought TMD care in the WDS system. The estimates reported in other studies for subjects with TMD symptoms who sought care have varied widely, from as low as 3 to 7 percent^21,22 to as high as 25 to 35 percent.^20,23 Only about 50 percent of WDS enrollees have TMD coverage, and some of these enrollees may have sought care under their medical benefits. Finally, some surgeons may have provided minor palliative TMD care without filing an insurance claim that indicated a TMD outcome.

If we assume an intermediate value for the percentage of patients seeking care (16 percent), along with a 50 percent capture of outcomes owing to coverage, we estimate the true incidence of TMD in our study population to be about 12.5 times higher, or 2.5 percent. Using this value for the overall incidence, the attributable risk of experiencing TMD after third-molar extraction is 1.2 percent, which is similar to the postoperative risk of developing infection, trismus and paresthesia.^7,8,19

There were some limitations to our study. There was the possibility of misclassification of the exposure, as some subjects may have had third-molar extraction before the age of 15 years. When planning our study, we found that less than 3 percent of all third-molar extraction occurred before the age of 15 years. We also knew that the large majority of third-molar extractions were performed between the ages of 16 and 18. Thus, we felt that enrolling subjects at age 15 years and requiring continuous enrollment through the age of 19 years would be the optimal method of capturing exposures and outcomes. Using an earlier enrollment age would have reduced the number of eligible subjects due to the requirement of continuous coverage. Similarly, the selection of our study period took into account a significant drop-off in coverage after the age of 20 years. Misclassification of exposure would have resulted in an underestimation of the true risk of TMD due to third-molar extraction.

Identification of the outcome was another potential concern. Diagnostic codes from the International Classification of Diseases, Ninth Edition, would have provided more accurate TMD information, but these were not available in the WDS dataset. However, the CDT treatment codes in the WDS database have been shown to be a reliable mechanism for identifying dental treatment, with almost 97 percent agreement between the database and chart notes.²⁴ We used only codes specific to TMD to indicate outcomes, and we performed secondary analyses, excluding subjects who had only TMD examinations. We were aware that less than 3 percent of TMD claims occurred in subjects younger than 15 years, but for the reasons stated above, we decided to accept this level of misclassification. It is difficult to predict the impact of this small amount of outcome misclassification.

While some advocate routine removal of third molars,²⁵ others believe that prophylactic removal of third molars is of limited value.²⁶ Our study did not attempt to address this controversy. It did, however, address a risk related to third-molar extraction that dentists and oral surgeons might be overlooking. Thus, our study identified the need for dental professionals to be aware of the association between third-molar extraction and TMD, to take precautions to minimize trauma to the joint during extraction, to assess joint health in the subsequent follow-up period and to conduct research to further clarify this relationship.

One strength of our study is that it is based on a large, insured population with unbiased ascertainment of exposure and outcome information. The use of a survival design maximized the use of the data, and we adjusted the analyses for both sex and dental care utilization. When we subjected the results to subanalyses that took into account a stricter definition of TMD, we found that they were stable. Finally, our study quantified the risk of experiencing TMD in a population of people during the period when third-molar extraction occurs most commonly. These results should be generalizable to other insured populations in the U.S.

	CONCLUSIONS

TOP ABSTRACT SUBJECTS, MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
In this insured population of adolescents and young adults, third-molar extraction appears to be associated with a 60 percent increased risk of experiencing subsequent TMD after adjusting for sex and dental care utilization. Owing to the high rate of third-molar extraction in the U.S., dentists performing these procedures should be aware of this risk and minimize trauma to the joint during extraction.