No prior research has explored the extent to which geographical variation in ADHD diagnosis may be explained by geographical variation in ADHD symptoms. To address this, this study aimed to answer three research questions regarding child and adolescent ADHD:
- Does variation in the incidence rate of ADHD diagnosis between clinics exceed chance variation?
- Does variation in symptom levels of ADHD between clinics exceed chance variation?
- Does variation in the incidence rate of ADHD diagnosis between clinics, conditional on symptom levels of ADHD, exceed chance variation?
The study used municipality-level data on all new individuals registered with an ADHD diagnosis in the Norwegian Patient Registry (NPR)* between 2011 and 2016. The cumulative incidence proportion of ADHD amongst individuals aged 0–18 years was then calculated as the number of individuals newly diagnosed with ADHD divided by the number of all individuals in that population age range.
To measure ADHD symptom levels for the general population, mother-reported data from the Norwegian mother, father and child cohort study (MoBa)† for 2011–2016 were used. The proportion of the population in a clinic’s catchment areaǂ ≥95th percentile was used. As a sensitivity measure, the study used the 90th percentile cut-off to accommodate clinics that were more prone to diagnosing ADHD than others. The parent/teacher Disruptive Behaviour Disorder Rating Scale was used to measure individual symptoms, including inattention, hyperactivity and impulsivity. Data were reported when the individual was aged 8 years.
The cumulative incidence of ADHD diagnosis was 0.016 (standard deviation 0.007, minimum–maximum 0.004–0.039, interquartile range 0.01–0.02) in 2011–2016. There was large municipality-level geographical variation in the incidence rate of ADHD diagnosis. The intra-class correlation (ICC) for variance components models for the incidence of ADHD diagnosis was 50.2% [95% confidence interval 39–61]. These data suggest that half of the total variance in ADHD diagnosis can be attributed to clinic level.
The proportion of individuals scoring ≥95th percentile and ≥90th percentile on ADHD symptoms was 0.05 and 0.1, respectively. There were two clinics which reported no individuals from the MoBa study scoring ≥95th percentile, and one clinic had no cases scoring ≥90th percentile. The ICC for the ADHD symptoms was <0.01% for the proportion of individuals with symptom scores ≥95th percentile and 0.15% for proportions of individuals with symptom scores ≥90th percentile.
The authors suggested that these data point to large between-clinic variation in the incidence rate of ADHD diagnosis, and less variation in high levels of ADHD symptoms at the ≥90th percentile. There was no evidence for more than chance variation in symptoms at the ≥90th percentile.
There were several limitations to this study:
- Ecological bias may have been a concern as both symptoms and diagnosis of ADHD are individual-level data aggregated to clinic-level variables.
- Since several units of observation can be used, statistical bias could have been introduced by the modifiable areal unit problem (Buzzelli, 2020).
- The association between symptoms of ADHD and diagnosis may have been subject to confounding bias.
- Geographical variation in areas of high ADHD diagnosis may have increased the likelihood of ADHD symptom reporting due to greater awareness of the disorder, leading to a reverse causal path between rates of diagnosis and ADHD symptoms.
- MoBa may have been a source of selection bias due to its overrepresentation of individuals with high socioeconomic status (Biele et al, 2019) and underrepresentation of non-Norwegians, young females, single households, mothers with > 2 births or previous stillbirths and smokers (Nilsen et al, 2009).
- Sample size, statistical power and researcher degrees of freedom may have led to the study findings occurring by chance.
- The study focuses on individuals aged 8 years. Therefore, the study findings do not reflect ADHD symptom levels for children and adolescents aged 0–18 years.
The authors concluded that, despite free healthcare access, comprehensive welfare and comparatively low social inequality in Norway, there was considerable between-clinic variation in ADHD diagnosis when controlling for ADHD symptoms. These findings pose a challenge for clinicians since geographical residence may play a role in ADHD diagnosis, particularly in individuals with symptoms around the threshold for diagnosis. Accordingly, residing in one clinic catchment area versus another may be a determining factor in who receives ADHD treatment. The authors note that this may highlight inequality in healthcare access based on an individual’s residence. As such, future research could consider a quasi-experimental approach to explore variation in diagnosis and/or medication rates to fill knowledge gaps relating to geographic variation in ADHD.
*The NPR is a health registry containing information on all individuals who have received or have been awaiting treatment in specialised healthcare services since 2008
†MoBa is a population-based pregnancy cohort study conducted by the Norwegian Institute of Public Health; individuals were recruited from across Norway between 1999 and 2008
ǂThe clinics’ catchment areas referred to the areas where child and adolescent mental health services clinics operated, serving ≥1 municipalities (and/or city districts in Norway’s four largest cities)
Disclaimer: The views expressed here are the views of the author(s) and not those of Takeda.
Biele G, Gustavson K, Czajkowski N O, et al. Bias from self-selection and loss to follow-up in prospective cohort studies. Eur J Epidemiol 2019; 34: 927-938.
Buzzelli M. Modifiable Areal Unit Problem. International Encyclopedia of Human Geography (2nd ed) 2020; 9.
Nilsen RM, Vollset SE, Gjessing HK, et al. Self-selection and bias in a large prospective pregnancy cohort in Norway. Paediatr Perinat Epidemiol 2009; 23: 597-608.
Widding‑Havneraas T, Markussen S, Elwert F, et al. Geographical variation in ADHD: do diagnoses reflect symptom levels? Eur Child & Adolesc Psychiatry 2022; Epub ahead of print.