Assessing ADHD symptoms in clinical public practice: Is a reliable final diagnosis possible?

Context and Objectives

Attention Deficit and Hyperactivity Disorder (ADHD) presents a significant clinical diagnostic challenge. Prevalence rates of ADHD vary between 1% and 20%, depending on the diagnostic guideline used, the assessment instrument employed, psychosocial factors, and the healthcare professional responsible for assessment. This substantial variability suggests problems in standardization of diagnostic procedures. Extensive scientific debate has identified two main conceptualizations: a neurobiological model that conceives ADHD as a syndrome with neurobiological origins (hyperactivity, attention deficit, impulsivity) and a psychopathological model that understands it as a symptomatic manifestation of different types of mental functioning or temperamental factors, considering psychosocial and environmental factors as main assessment aspects.

The general objective was to describe and compare current clinical ADHD assessment processes in the public health system across two temporal cohorts and to analyze variables related to final diagnosis. Specific objectives included: (1) describing clinical ADHD assessment tendencies among different healthcare professionals (Primary Care, Mental Health, and Neuropediatrics); (2) comparing assessment tendencies between two cohorts (2004 and 2009); and (3) analyzing whether the type of test used and the professional responsible for assessment are relevant components for final ADHD diagnosis.

Method

Participants

The sample consisted of 134 Spanish clinical cases extracted from a digital database of the Health Administration of Asturias (SESPA) in Oviedo, organized into two temporal cohorts: 55 cases detected in 2004 (41%) and 79 cases detected in 2009 (59%). The total sample included 106 males (79.1%) and 28 females (20.9%), with mean detection age of 9.88 years (SD = 5.84; 95% CI = 8.88-10.88). Mean age at symptom onset was 8.69 years in the 2004 cohort (95% CI = 7.61-9.77) and 11.03 years in the 2009 cohort (95% CI = 9.48-12.57).

Inclusion criteria were: clinical registers of ADHD symptom consultation in adult or pediatric primary care (registered as P21 code in the database) during 2004 and 2009. Exclusions included occasional consultations by displaced patients not regularly residing in the sampled area.

Design

A descriptive and observational study of two temporal cohorts was conducted with retrospective-prospective follow-up of each patient's Clinical History. Cases were detected in the Primary Care database in 2004 and 2009, and subsequently their clinical pathway was traced through digital health records. The Ethics Committee of the Central University Hospital of Asturias granted authorization in May 2014. Data collection occurred during June and July 2014.

Instruments and Variables

The following variables were extracted from clinical histories:

• Sociodemographic variables: sex, age at symptom onset, date of primary care consultation
• Organizational variables: professionals responsible for assessment (Primary Care, Mental Health, Neuropediatrics or others), referrals between services
• Assessment variables: clinical professional responsible for evaluation, tests employed (cognitive performance tests, symptom checklists, neuroimaging or other unspecified tests), tests used by each professional, final diagnosis registered (for Mental Health and Neuropediatrics), and clinical professional establishing final diagnosis

Analysis

Descriptive measures of centralization and dispersion were used for quantitative variables. For qualitative variables, descriptive analyses of frequency distributions were employed. For bivariate analysis, chi-square and Fisher's exact tests were used to describe changes among variables. Rate analyses were performed using binomial tests for rate comparison. Cohen's kappa concordance was calculated for inter-professional ADHD diagnostic reliability assessment. Descriptive analyses, chi-square and Fisher's tests were performed in SPSS 15.00, while rate analyses and kappa concordances were executed in EPIDAT 3.1.

Results

Assessment and Tests Utilized

Of 134 total cases, only 43.3% (58 subjects) had assessment supported by some test. Among these 58 subjects, a variety of instruments were used: cognitive performance tests (50%, n = 58), symptom checklists (44.8%), neuroimaging (2.069%), and unspecified tests (6.70%). Different application of these tests occurred between cohorts (Figure 1) and according to health professional (Figure 2).

The use of symptom checklists (chi2, 1df, p < 0.001) and cognitive performance tests (chi2, 1df, p < 0.001) showed association with higher ADHD diagnosis confirmation rates, but only when these tests were used by Neuropediatrics professionals, not when used by Mental Health professionals.

Referrals and Services

Of 134 total cases, 29.9% were not referred to other clinical services from Primary Care, 36.6% were referred to Mental Health, 12.7% were referred to Neuropediatrics, 17.9% were referred to both professionals (Mental Health and Neuropediatrics), and 3% were referred to another clinical professional. A statistically significant increase was observed in the 2009 cohort of cases referred to Neuropediatrics services compared to 2004 (Fisher, 2-sided, p = 0.039), and a significant decrease in 2009 of cases referred to Mental Health (Fisher, 2-sided, p = 0.017) or using Mental Health assessment services (Fisher, 2-sided, p = 0.01).

ADHD Diagnosis Confirmation

Of 53 people using Neuropediatrics service, information on diagnosis confirmation/disconfirmation was obtained in 49 subjects. Of 86 people using Mental Health service, information was obtained in 80 subjects. There were 34 subjects assessed by both services, with information about ADHD confirmation/disconfirmation by both specialties in 29 cases.

In Neuropediatrics services, 57.1% of total cases were confirmed as ADHD, compared to 35% of cases in Mental Health (Table 3). Cases assessed by both services (n = 29) presented low inter-professional ADHD diagnostic confirmation concordance (kappa coefficient = 0.39). The rate of ADHD diagnostic confirmation was significantly higher (binomial, 1df, p = 0.026) when assessment was performed by public Neuropediatrics service.

The rate of ADHD confirmation in Neuropediatrics services increased significantly between 2004 (33.3% of the sample assessed by NP) and 2009 (62.9% of the sample assessed by NP) (binomial, 1df, p = 0.049). The rate of ADHD confirmation by Mental Health showed no significant change in 2009 (37.2%) compared to 2004 (27.9%).

Final Diagnoses

Of 53 people using Neuropediatrics service, information on final diagnosis was obtained in 45 subjects. Of 86 people using Mental Health service, information was obtained in 74 subjects. The diagnosis linked to ADHD symptoms differed significantly according to the assessing professional (Figure 3).

In Neuropediatrics, diagnoses were distributed as: no diagnosis (8.16%), organic disorder (10.20%), developmental disorder (22.45%), ADHD (51.02%), behavioral disorder (8.16%), Z-codes (0%), personality disorders (0%), anxiety disorders (0%), and other disorders (0%).

In Mental Health, diagnoses were: no diagnosis (8.75%), organic disorder (2.50%), developmental disorder (5%), ADHD (33.75%), behavioral disorder (16.25%), Z-codes (20%), personality disorders (5%), anxiety disorders (5%), and other disorders (3.75%).

Discussion and Conclusions

Main Findings

Results demonstrate that the majority of analyzed cases (56.7%) lacked assessment supported by tests. Although tests are not necessary for ADHD diagnosis, some such as cognitive performance tests or executive function tests constitute a good complement for more precise differential diagnosis. The lack of supporting tests is associated with higher ADHD incidence rates, according to previous literature. Professionals who depend heavily on symptom checklists are primarily Primary Care and Neuropediatrics professionals. Symptom checklists have been associated with higher ADHD rates as they are based on parent and teacher criteria.

However, in the present study, ADHD rates were not related to the type of test employed in assessment but rather to the professional using them. This finding suggests that final diagnosis does not depend as much on the test employed but on the professional implementing them.

Temporal Changes and Professional Trends

Changes were observed from 2004 to 2009: an upward trend of Neuropediatrics and Primary Care professionals assessing ADHD, and a downward trend of Mental Health professionals doing so. The increase of Neuropediatrics professionals responsible for ADHD assessments can be related to the prevailing neurobiological model for ADHD. This predominant model also affects the ADHD conception sustained by parents and educational systems, their needs, expectations, and demands in the health system.

Previous literature indicates that ADHD incidence rate is higher when assessment is performed by Primary Care. However, results of the present study demonstrate a significant increase in ADHD rate when Neuropediatrics service is responsible for assessment. Neuropediatrics tends to diagnose more ADHD and developmental disorders for the same symptom profile, while Mental Health professionals distribute their main diagnoses among ADHD, behavioral disorder, or Z-codes (presence of factors influencing health status).

Alternative Explanations

The data could indicate that different types of patients are referred to different services for diagnosis (suspected organic disorders tend to be referred to Neuropediatrics, other disorders to Mental Health). Another hypothesis is that different professionals sustain different ADHD conceptions when assessing identical symptoms. In relation to previous literature, a neurobiological conception tends to be sustained by Neuropediatrics professionals (considering mainly symptom presence), while a psychopathological conception could be sustained by Mental Health professionals (considering psychosocial and environmental factors as main aspects in assessment and treatment process). Different conceptual models underlying different diagnoses can be related to the low ADHD diagnostic kappa concordance results found in the present study.

Clinical Implications

These findings have important implications for Primary Care professionals. Final diagnosis and treatment, when referring a child with ADHD symptoms for assessment, will vary significantly depending not only on instruments and diagnostic criteria considered in assessment, but also on the healthcare professional to whom the child is assigned. It would be interesting to consider as a first step in ADHD assessment to begin with a social assessment, usually conducted by mental health professionals. This way, social and environmental factors would be considered from the beginning of the process, avoiding in some cases an ADHD diagnosis and pharmacological treatment in childhood.

Significance and Contribution

This study provides critical empirical evidence regarding inter-professional variability in ADHD diagnosis in a real clinical practice context of the public health system. Low inter-professional diagnostic concordance (kappa = 0.39) between Neuropediatrics and Mental Health, combined with the increasing trend of ADHD diagnoses by Neuropediatrics between 2004 and 2009, suggests that ADHD diagnostic criteria appear unclear in clinical practice. Results indicate that ADHD diagnosis must be used with caution to ensure good quality clinical standards when assessing and treating ADHD symptoms. Assessments supported by symptom checklists and performed by Neuropediatrics or Primary Care could be contributing factors to an ADHD over-diagnosis tendency.

The study warns that ADHD diagnosis and prescription of psychotropic drugs do not appear to be rooted in clear criteria. Gold clinical standards must be warranted in all cases, independently of the professional who diagnoses and treats the case. Further investigation needs to be developed for establishment of clear and functional criteria to be used when diagnosing and treating ADHD.

Limitations

The study was based on a natural clinical context sample without unified ADHD detection and assessment procedure, presenting good external validity but weak internal validity. Additionally, although symptom data were collected, impairment/functional disability data were not sufficiently available in clinical histories (most of the sample lacked registers about disability caused by symptoms). Also notable was the appearance of statistically atypical cases of ADHD onset in adulthood in 2009, which could indicate an upward trend of ADHD diagnosis in adults requiring consideration.

---