Prevalence and diagnostic stability of ADHD and ODD in Turkish children: a 4-year longitudinal study

Attention-Deficit/Hyperactivity Disorder (ADHD) and Oppositional Defiant Disorder are among the most common psychiatric disorders of childhood either in community or clinical samples [1–3]. Moreover, they co-occur much more frequently than expected by chance [4]. Both disorders are related with substantial impairment and can be precursors of CD, severe delinquent behavior and substance use disorders [2, 5].

As in any medical conditions, the development of health strategies directed to early diagnosis and treatment of ADHD depends on robust epidemiological data [6]. It is important to note that a huge variability in prevalence rates are detected among studies [6]. Although previous literature clearly suggests that heterogeneity in prevalence rates of the disorder is associated with methodological differences among studies [7], there are yet concerns that ADHD prevalence might be exaggerated in some environments [8]. Faraone et al. [9] stated that, the predominance of American research into this disorder over the past 40 years has led to the impression that ADHD is largely an American disorder and is much less prevalent elsewhere. This impression was reinforced by the perception that ADHD may stem from social and cultural factors that are most common in American society. Moreover, some regions of the world were underrepresented in the most comprehensive review of the ADHD epidemiology in youths [7]. The only significant differences in prevalence rates were found between North America and both Middle East and Africa. However, as stated by the authors, these differences might be an artifact of the small number of studies in those regions (4 from Middle East and 4 from Africa) introducing instability in analyses. As stated by Polanczyk and Jensen [6], more studies conducted in areas with scarcity of data with better methodologies are needed to really expand the knowledge on the worldwide prevalence of the disorder.

Despite its clinical relevance, surprisingly very little data is available about ODD prevalence. One possible reason for this is the fact that several studies implemented the evaluation of ODD in combination with CD under a category named “conduct problems”. This situation has its roots in a usual tendency to view ODD under the umbrella of CD. With rare exceptions, the epidemiology, comorbidity pattern and course of ODD were investigated in combination with CD [10]. However recent studies showed that ODD could be separated both from CD and normal child behavior and must be studied independently to be well understood [11].

Studies on ODD prevalence produced a wide range of rates from 2 to 15% [12–14]. Nock et al. [11] and Maughan et al. [15] suggested that the important variation in ODD prevalence might be related to the use of non-representative samples and inconsistent diagnostic approaches in the studies. Very recently, the characteristics influencing ODD and CD prevalence worldwide have been investigated in a systematic review and meta-regression analysis [16]. The authors have assessed all the investigations conducted between 1987 and 2008 in their study. At the end of the assessments, 39 investigations were found to meet the inclusion criteria and among them, 25 studies with available data were included in the review. The pooled prevalence of ODD was estimated to be 3.3%. The authors reported age as the only covariate that remained significantly associated with heterogeneity of results for ODD after the successive deletion of non-significant variables in the multivariate meta-regression model. Geographical area was not found to be related with estimated ODD prevalence. However, it is important to note that only 4 of the 25 studies were from non-western countries (3 from Asia, 1 Middle East) and the remaining 21 were from Europe and United States.

One of the most important gaps in the epidemiology of ADHD and ODD is the fact that age related changes of these disorders were not thoroughly investigated in longitudinal community studies [17]. It is known that ADHD persists through adolescence and adulthood and 15-80% of children diagnosed with ADHD continue to have the disorder in adulthood, depending on the diagnostic criteria used [9, 18] However, these findings are mostly derived from follow-up studies relying on clinical samples of children with ADHD. Recently, Ramtekkar et al. [19] assessed a huge community sample of subjects from different ages documenting a decline of ADHD symptoms and consequently prevalence rates during the life cycle. However, this was a cross-sectional study. Similar to ADHD, longitudinal prevalence of ODD has been very rarely investigated in epidemiological sample. Nock et al. [11] suggested that ODD mean persistence is around 6 years. However, many adults diagnosed as ODD during childhood and adolescence continues to show ODD symptoms, although diagnostic criteria of ODD are not currently used in adulthood [11].

This study was conceptualized to provide the following specific data lacking in the literature: Firstly, to find out the prevalence rates of ADHD and ODD in a non-referred representative sample from a country in a region where no previous rates are available (Turkey). Turkey has also a specific importance with an interesting geographical position as a bridge between Europe and Asia. Secondly, to evaluate the differences in ADHD and ODD prevalence rates between four different waves with one-year gap in reassessments. So, the data obtained from the study has the potential of increasing our knowledge on the developmental epidemiology of ADHD and ODD.

This study is a 4-year longitudinal ADHD and ODD prevalence investigation. The first year assessment started with a screening phase in April 2008 when the ADHD and Disruptive Behavior Scales were sent to parents and teachers. A study sample was derived at the end of the screening phase and clinical evaluations (screening positive cases and matched controls) were performed in May 2008, configuring the first wave of the study. The sample was reassessed in May 2009, May 2010 and May 2011 in the second, third and fourth waves of the study.

Prevalence rates of ADHD and ODD were calculated according to a standard formula as suggested by Rohde et al. [26]: Pe = cd + nns (1-npv)/n, where Pe is the estimated prevalence, cd are the cases during the diagnostic phase, nns is the number of negative screenings that were not assessed at diagnostic phase, npv is the negative predictive value of screening instrument and n is the sample size. The negative predictive value is the number of times that the screening instrument said that the subject was a non-ADHD case and the diagnostic assessment confirms the subject as a non-ADHD case. The subtraction of 1-negative predictive value gives the percentage of correct diagnosis in screening negatives. Thus, multiplying per number of screening negatives that were not assessed in diagnostic phase gives the number of potential real cases that were not assessed because they were screening negatives but have ADHD. Confidence interval for the estimated prevalence of ADHD and ODD was calculated with Fleiss Quadratic approximation method [34]. For the statistical analysis, chi square test for categorical variables and paired sample t test for numeric variables were used. P values less than 0.05 were accepted to be statistically significant.

The cases were included in analyses only if both the parent and teacher scales were returned and if they were correctly filled out. Thus out of 1500 children, 1455 (between 8 to 12 years of age) were included in the screening phase and the response rate was found to be 97% (43.1% of the subjects were females and 56.9% males).

This study is a 4-year longitudinal investigation on the ADHD and ODD prevalence rates conducted in Turkey which is an interesting geographical area standing on the intersection of Europe, Asia and Middle East. It is important to note that these 3 areas have very few investigations on the prevalence of child mental disorders including ADHD and ODD. Our findings on ADHD for the 4 waves (consecutively 13.38%, 12.53%, 12.22% and 12.91%) were remarkably higher than the worldwide pooled childhood prevalence of ADHD (6.48%) [7]. On the other hand, the prevalence rates of ODD were found to be 3.77%, 0.96%, 5.41%, 5.35% respectively in the first, second, third and fourth waves and the mean ODD prevalence was found to be 3.87% which was surprisingly very close to the worldwide pooled prevalence of ODD (3.3%) [16].

The five main findings from this study were: 1) A substantially higher prevalence of ADHD in school-age children in Turkey compared to the one reported in a recent meta-analyses including studies from several different countries [7]. This difference becomes more important when we consider that diagnostic procedures followed in our study included even an independent measure of impairment; 2) The consistency of the 4-year average ODD prevalence with the literature, in spite of the fluctuations between the waves; 3) Remarkable true positivity of screening for ADHD (high positive predictive value), as well as clear stability of the diagnosis and reliable Kappa values throughout the 4 consecutive assessments with one year intervals; 4) Lower true positivity of screening for ODD and less stable ODD diagnosis with smaller Kappa values throughout the 4-years interval compared to ADHD; 5) much larger prevalence of ODD among boys than girls.

In Turkey, ADHD prevalence was not previously studied in extensive investigations, but two studies with diagnostic procedures including only parent and teacher scales reported the prevalence of ADHD in school age children between 6–12 and 6–15 years as 8.1% and 8.4 respectively [35, 36]. It was interesting to find a higher ADHD prevalence in spite of the use of stricter methodology in our study. Moreover, ODD and ADHD prevalence rates were determined with same methodology but only ADHD prevalence rates were found to be high throughout the 4 waves. The ODD prevalence was found to be much less than the previous studies conducted in Turkey that was based on scale assessments [35, 36]. These results may raise the question “whether ADHD is more prevalent in Turkish children than some other parts of the world?” Considering the migratory origin of Turkish people, which was officially registered with “sella turcica” in anatomy, and the relation between migration and DRD4 gene (one of the most important candidate genes in ADHD etiology) this hypothesis worth to be investigated [37].

One of the most important aspects of ADHD epidemiology is the course of the disorder across the life span. The ADHD prevalence tends to decrease with age and 60% of the cases diagnosed as ADHD during childhood continue to be diagnosed with ADHD in adulthood [38]. But most of the data supporting this idea is derived from follow-up studies of clinical samples. There are scarce data from longitudinal non-referred samples [17]. In the study of Cohen et al. [39] which was an 8 year follow-up study, ADHD prevalence was found to be 12.8%, 9% and 6% in the same cohort in the following age ranges: 10–13, 14–16 and 17–20 year. In another study conducted in Spain among children at 8, 11 and 15 years old, ADHD prevalence was found to be 14.4%, 5.3% and 3%, respectively [40], and in another study 6.8% in the same country [41]. A study from Canada, reported ADHD prevalence as 5.5%, 4% and 2.5% among 6–8, 9–11 and 12–14 years-old subjects, respectively [42]. When the results of these studies are evaluated, a 0.5 to 1% decrease in ADHD prevalence rates in average is found for each 1-year increase in age. In our study, we detected a decrease of 0.47% in the prevalence rates of ADHD between the first and fourth waves which is completely consistent with previous studies. The future follow up of the subjects of this study will provide more information about the lifelong prevalence of ADHD.

One interesting finding was the fluctuation of ODD prevalence rates among the three waves. It has been known that variations like presence or absence of diagnosis are frequent in longitudinal studies [43]. Lavigne et al. [44] evaluated 510 children aged 2–5 years old with ODD for a period of 48–72 months at 5 separate waves and they stated that part of the fluctuations among the waves involves a move from exceeding a diagnostic threshold to a subthreshold status and vice versa. In our study these fluctuations may be explained similarly.

While the mean ODD prevalence found in this study is very consistent with the literature; distribution of the cases according to sex is quite different than previous literature findings. Female/male ratio of ODD was around 1/10 in all three waves of our study. This is a quite large difference, although statistical significance was not obtained. ODD prevalence was reported to be higher in boys than in girls in most of the previous studies, especially in preadolescent period. Among the previous literature, there is only one study with similar findings to our results in terms of male/female ODD ratio; but nevermore only in a subgroup of the study (Female/Male = 1/4) [45]. In that study, disruptive behavior disorders prevalence rates were compared among the Puerto Ricans living in Puerto Rico and New York and male–female difference was found to be quite large only among the cases between ages 5–9 living in New York; while in other residency places and age groups, large differences were not found in spite of the dominancy of males [45]. Although cultural variations may be suggested as an explanation for this result from our study, the findings from another study conducted previously in Turkey reporting a very close distribution of ODD cases among boys and girls (55.5% in boys vs. 44.5% in girls) does not support this view [36]. It is important to note that there is an active debate on the applicability of DSM-IV ODD criteria to girls [30, 46]. Some authors have suggested that aggression might be expressed among girls in a different way than boys; covert aggression may be a more frequently used style among girls instead of overt aggression [47]. In terms of ADHD diagnosis, the dominance of boys compared to girls was an expected finding. The male/female ADHD ratio was 3.2 times higher in male than females in the first wave and 3.4 times higher in the second wave and 3.5 times higher in the third waves.

Parent education and neighborhood were the only investigated socioeconomic variables in the study and no statistically significant difference was found between controls and both ADHD and ODD groups. In an epidemiological survey from Germany, ADHD was reported to be more frequent among subjects with low socioeconomic status whereas there were no statistically significant differences with regard to geographical characteristics (e.g. urban vs. rural) [48]. It has been reported that ODD and CD occurs most frequent in lower socioeconomic groups [13, 46]. But it has also been stated that prevalence rates of disruptive behavior disorder in the disadvantaged neighborhoods compared with advantaged inner-city neighborhoods have not been sufficiently documented, and the current evidence on possible differences in the prevalence of ODD and CD in rural and urban environments is mixed [2, 13].

Our findings describe the ADHD prevalence rates and stability for the first time in a region where no previous study was conducted implementing a careful and extensive diagnostic procedure. Surprisingly, even after using strict criteria with impairment assessments, we found a high ADHD prevalence rate for children at 8 to 10 years of age. In addition, ADHD diagnosis was quite stable in reassessments one, two and three years after. Our results argues against the hypothesis that ADHD is a cultural construct that is uniquely associated with the United States or any particular culture. More studies from regions with no previous epidemiological information on ADHD will improve our knowledge on the real impact of cultural diversity in the disorder.

In addition, to the best of our knowledge, this study was the first longitudinal prevalence study which evaluated ODD in particular by assessing it separately from CD. This study was designed with the aim to investigate ODD prevalence in a non-referred school sample, with a well defined and sufficient sample size, by three stage evaluation and by considering impairment criteria and employing best estimate procedure for diagnosis. An ODD prevalence which was very consistent with the worldwide pooled prevalence was found in the study but diagnostic stability was found to be much lower compared to ADHD diagnosis. The results obtained from this longitudinal study should be confirmed with future studies using state of the art methodologies in different regions of Turkey.