**2. Methodology**

#### **2.1 Literature search and search strategy**

PubMed and Google Scholar were used to review the existing literature and identify a gap in the literature on the prevalence of all mental disorders among children and adolescents in Europe. A search on Prospero [9] confirmed that there were no reviews registered in this area, so a protocol for this study was then registered there (Registration number: CRD42020210451). A search strategy was developed using the SPIDER model and conducted on MEDLINE, Embase, and PsychInfo on the 30th April 2020. The search was limited to studies with a title and abstract in English.

#### **2.2 Eligibility criteria**

Studies were considered eligible if they were original epidemiological studies that determined the prevalence of a mental disorder as defined by ICD-10, DSM-IV, or DSM-V criteria, among 5–18-year-olds in European countries. Studies were excluded if they did not include the general population, for instance, by focusing on minority groups, or if they were published before 2015.

#### **2.3 Study identification and selection procedure**

Studies found by the search were screened independently by title and abstract by RS. The studies that met inclusion criteria were screened independently by full text by RS. DNB screened 20% of all the studies at title/abstract and at full text review stage. Reference lists and gray literature were searched manually by RS.

#### **2.4 Quality analysis**

The reliability, validity, and bias of each eligible study were assessed using the Appraisal Tool for Cross-Sectional Studies (AXIS) [10] and the Risk of Bias in Prevalence Studies Tool (RBPS) [11].

#### **2.5 Data analysis**

Only the eligible studies that estimated the prevalence of autism were included in the analysis for this chapter. Median and average estimates and ranges of autism prevalence rates for young people in Europe were determined. Comprehensive meta-analysis software [12] was used to analyze prevalence data from the eligible studies. A random effects model was used to determine the random effects pooled prevalence rate of autism in Europe. Prevalence rates obtained from population and register-based studies were analyzed separately since the two study types have non-homogeneous populations, and there may be significant discrepancy of factors at many levels of the variable of interest. Data from the two study designs were therefore analyzed separately to avoid Simpson's paradox [13]. The standardized residual values (SRV) were evaluated from the forest plots and a cutoff of +/−3 at 95% confidence interval was used to identify outliers [14].

Cross-national prevalence comparisons were made across countries, gender, and level of education. The latter was done to compare prevalence rates between young children who attend primary school, to older children who attend secondary school. Prevalence rates were not compared according to specific age groups since the eligible studies presented results for a mixed range of age groups, which were incomparable. The contribution of specific cofactors to heterogeneity could not be evaluated through a meta-regression analysis because results would be insignificant due to the low number of eligible studies [15].

### **3. Results**

Nine eligible studies were identified that provided prevalence estimates for 11 European countries as illustrated in **Table 1**. The AXIS and RBPS tools indicated lowlevel bias among all the eligible studies.

#### **3.1 The Prevalence of Autism among 5–18-year-olds in Europe**

Based on the eligible studies, the prevalence ranged from 0.3% in the West Pomeranian and Pomeranian regions of Poland to 14.3% in Romania (**Figure 1**). The median prevalence was 1%, and the average prevalence was 1.97%.


**Table 1.**

*Eligible studies and their characteristics.*

**6**

*The Prevalence of Autism Spectrum Disorder in Europe DOI: http://dx.doi.org/10.5772/intechopen.108123*

**Figure 1.** *The prevalence of autism among 5–18-year-old young people in Europe.*
