Review of Prevalence Studies of Tic Disorders: Methodological Caveats

Sample selection

Earlier estimates of the epidemiology of tic disorders were based on single or multicenter clinic samples. However, with the use of such referral samples, the detection and diagnosis of tic disorders, especially among mild cases and/or children with limited access to specialty health-care services, is not optimal, and prevalence estimates may be low.6 On the other hand, administrative databases (state-population databases or large national samples) provide wide coverage of the population, and provide an opportunity to include samples with different sociodemographic characteristics. In this regard, the use of administrative databases and/or telephone-based surveys, may provide us with estimates of the public health impact of tic disorders.6,35,36

The use of community-based samples has resulted in higher estimates of the prevalence of tic disorders. Epidemiological studies of tic disorders have been conducted in different populations, sampling different races and geographical areas, cohorts from movement disorder clinics, community-based samples, and school-based samples (including mainstream schools and special education centers [SECs]). Overall, TS, the most studied tic disorder, has been reported all over the world, and from all sociocultural, socioeconomic, religious, linguistic, and ethnic groups.37 However, while TS has been researched extensively in Western populations, the clinical characteristics and associated features are less well known in non-Western cultures.13 Lower prevalences of TS have been reported in China and sub-Saharan and other regions of Africa.7 According to Robertson,13 the following reasons may account for the apparent rarity in these regions: other medical priorities, lack of medical training, ethnic and epigenetic differences, and genetic and allelic differences in different races.

The age of the population may also influence estimates of the prevalence of tic disorders. Inclusion of persons in younger age groups in population-based studies (e.g., as young as 4 years of age) may result in lower estimates of prevalence, since the symptoms may not yet have manifested at such young ages. For example, in one study conducted in primary schools, TS was found in 0.15–1.1%.37,38 In contrast, the prevalence among older people, especially in adult psychiatric units, may also be relatively low (e.g., around 9.5%) when compared to the prevalence in younger people, as studies have shown that tics diminish with age.39

Prevalence estimates of tic disorders in populations with mental disabilities also seem to be higher, especially in children with autistic spectrum disorders, learning difficulties and behavioral problems; these estimates range from 6% to 23.4% for TS, and up to 46.3% for other tic disorders.22,31,40–44 Interestingly, in our study conducted in SECs, tic disorders were very frequent in children diagnosed with Down syndrome, and there was no gender preponderance.32

Screening and case ascertainment

There is no doubt that community-based surveys are costly and therefore difficult to perform. Yet they offer an ideal strategy for obtaining an adequate and representative sample size, especially for the study of low prevalent conditions or populations at high risk of dropout (i.e., populations of low income, of those with low education background, etc.). The logistic difficulties increase when epidemiological studies are carried out in populations diagnosed with diseases such as tic disorders that have fluctuating symptoms and signs. Validated screening tools and methodologies are needed to define each case. Ideally, screening should be a multistage procedure, which in a prior pilot study specific for that population and setting, has been validated against a gold standard (i.e., a diagnosis established by medical personnel).

Few epidemiological studies have previously reported the sensitivity of their screening procedures for tic disorders.21,22,44 Prior studies have used different methods to identify individuals with tic disorders, including direct observations at school, parent, or teacher interviews or questionnaires, student questionnaires, and clinical examination.19,37,44–47 The use of information from trained teachers is reported to have a sensitivity of up to 80% in mainstream schools.22 A study designed to validate screening procedures in a random selection of parents and teachers for 130 individuals in mainstream schools, demonstrated that this approach had a high sensitivity (92%) but low positive predictive value (18%).45 We conducted a pilot study designed to validate our screening procedures for tic disorders in different settings.48 We developed a two-stage screening tool. First, a Proxy-Report Questionnaire completed by parents and teachers, and, second, classroom observation of up to 3 hours by three trained lay observers. The training of the three observers was also validated by the combined rating of 11 individuals (six with tics and five without tics), which resulted in a high sensitivity (100%) and good specificity (60%) to identify tics when compared with the neurologist, good interobserver agreement (kappa coefficient [k] = 0.62), and good neurologist–observer agreement (k = 0.62). We found that 2 hours’ observation of the classroom offered the best strategy. However, this length of direct observation can be considered suboptimal as it seems to have lower single screening accuracy when compared with the information provided by the teachers and parents. The decision to use a screening test on a given group of patients is made on the basis of a cost–benefit analysis. Hence, if one is only able to use a single screening source, parents would be the best source in mainstream schools, given that their modest sensitivity (58%) is compensated by a high specificity (92%) and the highest likelihood ratio (7.38%) for a positive result prior to testing (Table 2). On the other hand, teachers are the best single screening source of information in SECs, with moderate sensitivity (73%) and specificity (55%), but with the highest likelihood ratio (1.63%) for a positive result prior to testing (Table 3). If multiple sources can be used, the most efficient screening source in mainstream schools is the combination of parents and a 2-hour classroom observation. By contrast, the information from teachers plus 2 hours of observation provides the best combination for screening in SECs. Overall, this screening procedure seems to be more accurate in mainstream schools than in SECs, most likely due to the presence of repetitive motor behaviors, such as stereotypes, in children with mental disability.

A selection bias is commonly observed in studies for which the cases are ascertained based on a positive screen.16 Using such methodology, false-negative cases may be underestimated and, likewise, the estimated prevalence of tic disorders may be low. Similarly, there is a failure to ascertain cases in some studies, or assessments that are only performed by trained technicians.44 Preferably, case ascertainment should be performed by medical personnel.21,48,49

In order to decrease the burden of conventional face-to-face interviews and to avoid dropouts, telephone interviews can be an efficient alternative. In our study, we analyzed the intra-rater (neurologist) reliability of telephone interviews compared to in-person interviews to achieve a correct diagnosis of tic disorders.48 A high k of 0.83 was found, suggesting that telephone interviews may make it easier to perform clinical interviews, and to increase participation. However, in our study, the use of telephone interviews decreased the level of diagnostic certainty by increasing the frequency of possible tics up to 19.4%.32 Unfortunately, similar reliability of data from telephone interviews has not been published in other studies, so comparisons are not possible.16,50

Definition of tic disorders

Tics can be motor, vocal, sensory or cognitive, simple or complex, isolated or multiple, and brief or more dystonic and tonic. Classically, tics are described as sudden, brief, and rapid, repetitive sounds or movements that can be controlled voluntarily during short periods of time, they may change and affect other body parts periodically, and improve or worsen from time to time. Since tics are a fluctuating condition, the neurologist may not observe the tics during the exam, but can establish the diagnosis based on patient/relative reports or videotapes.51,52 Likewise, many clinicians erroneously believe that coprolalia must be present for the diagnosis of TS to be made, and some habits, such as foot tapping or finger drumming that occur when subjects are bored or anxious, are incorrectly diagnosed as tics. Although it seems to be obvious, the definition of tic phenomenology should be clear for standardization of epidemiological research efforts. In Appendices 1 and 2, we provide an example of the screening and interview schedule for tic disorders followed in our study.48

The DSM criteria for tic disorders have also changed over time.2 Specifically, the upper age limit, which is somewhat arbitrary, has changed, and the inclusion of a tic period of at least three consecutive months for TS, CVT, or CMT has been added, although there seems to be no evidence base for this.13 By contrast, in the World Health Organization (WHO) criteria,3 there has never been an age of onset requirement, resulting in the inclusion of adult-onset tic disorders. Whereas there is a clear definition of tic disorders based on the DSM, in the International Classification of Disease and Related Health Problems 10th revision,2,3 and the Classification of Tic Disorders developed by the Tourette Syndrome Classification Study Group,4 tics represent a symptom, and their definition is not included in these diagnostic revisions. Adult-onset tics have been reported, but in contrast to child-onset tics, they are likely to be secondary to trauma, drugs, stroke, or infection.53

In addition, there also several ambiguities about the definition of tic disorders, with regards to the inclusion or not of impairment criteria, and comorbid conditions. The current version of the DSM-IV TR,2 and in the future DSM-V,54 the distress item, which had been obligatory in order to establish a tic diagnosis in previous classification schemes, has been omitted. Clearly, more restrictive criteria yield much lower prevalence estimates, and, thus, misleading results.32,44,55,56 The use of impairment criteria as a means of distinguishing cases from non-cases of tic disorders is controversial, given the fact that in most cases, the source of impairment (tics or comorbidity conditions) cannot be clearly established.57

Coverage of the eligible population

Overall, there is a lack of information about attrition (i.e., the loss of participants) in neuroepidemiological studies of tic disorders. Attrition can result in selection bias and reduce the internal or external validity of the research findings. Likewise, the response rate, defined as the number of respondents who complete a questionnaire compared to the number assigned, may also impact on the quality of the data. In our study,32 we observed different response rates for mainstream schools (70%) vs. SECs (55%). Caucasians and younger students without academic problems were more likely to participate in a survey of tic disorders. In other studies, conducted in youth cohorts, the main demographic variables associated with participation were gender (girls were more likely to participate) and a high education background of parents.57 Other variables that may contribute to dropout include the frequency of contacts, format of the survey (face-to-face interview vs. telephone interview), distances, contact reminders for initial contact failure, and gender of the parent. A detailed description and analysis of attrition variables is then of major importance for future epidemiological studies, to improve the enrollment of participants at higher risk of dropout.