A Retrospective Study Examining Predictors of Treatment Response in Over 1,300 Children and Adolescents with Attention-deficit Hyperactivity Disorder

Abstract

Objective

This study aimed to examine the predictors of treatment response in children and adolescents with attention-deficit hyperactivity disorder (ADHD).

Methods

The study included 1,314 patients with ADHD who came to the child psychiatry department over a 5-year period. The 276 patients whose treatment was changed due to treatment non-response or side effects were included twice for the first and second treatment, so a total of 1,590 patient data were used. Age, sex, initial ADHD severity, ADHD treatment, treatment response, comorbidities, additional treatments, and chronic disease status of the cases were recorded.

Results

One thousand twenty-seven (78.16%) of the patients were male and 287 (21.84%) were female. Using the combination therapy of methylphenidate (MPH) and atomoxetine (ATX) as the reference category, ATX monotherapy showed a significantly lower likelihood of responding to treatment. Moderate and marked disorder severity were associated with higher odds compared with severe disorder. Borderline intellectual functioning (BIF), mild intellectual disability, conduct disorder (CD), autism spectrum disorder (ASD), and adjustment disorder (AD) showed significantly lower odds. Older age was significantly associated with response to treatment. In the MPH monotherapy and ATX monotherapy groups, compared to moderate disorder, severe disorder was associated with significantly lower odds. BIF, CD, ASD, enuresis, communication disorders, and AD showed significantly lower odds in MPH group, while only ASD had significantly lower odds in ATX group. Higher ATX dosage was significantly associated with treatment response in this group.

Conclusion

Knowing the factors affecting treatment response can be used to predict prognosis.

INTRODUCTION

Attention-deficit hyperactivity disorder (ADHD) is a neurodevelopmental disorder characterized by symptoms of inattention and/or hyperactivity-impulsivity that impair school, social or work functioning [1]. ADHD, which is the most common neurodevelopmental disorder in the clinical practice, is not only seen in childhood, but is also highly persistent in adolescence and adulthood [2-4]. Therefore, the treatment of this disorder, which causes loss of functionality in daily life and can persist for many years gains importance.

Investigating the response to ADHD treatment is an important issue and has long been of interest to clinicians. There have been three reviews evaluating the predictors of stimulant response in ADHD at consecutive different time intervals since 1976. In the study in which Barkley examined 36 studies, the best predictors of response to stimulants in hyperactive children were measures of attention span or concentration and their correlations [5]. Gray and Kagan [6] found that lower ADHD severity, older age and anxiety comorbidity were associated with poor treatment response in their report examining 59 studies. Finally, Pagnier indicated that there is no relationship between methylphenidate (MPH) response and sex, age, socioeconomic status, ADHD subtype, MPH dosage and comorbidities after reviewing the results of 46 studies. No inferences could be made regarding ADHD severity, intelligence quotient, body weight, or presence of ADHD and depression symptoms in the parent [7].

There are limited studies on predictors of atomoxetine (ATX) response. It was reported that, variables such as age, sex, ethnicity, ADHD subtype, baseline severity, previous stimulant use, height, weight, comorbid oppositional defiant disorder, CYP2D6 genotype were not predictive for ATX treatment response [8,9]. Age older than 12 years, female sex, previous ADHD treatment other than ATX and MPH, absence of specific phobias, and absence of ADHD-hyperactive-impulsive subtype were associated with remission with ATX [10].

Knowing the predictors of treatment response in ADHD can be used to develop personalized treatment methods and predict prognosis. This retrospective study aimed to evaluate the predictors of treatment response in children and adolescents with ADHD admitted to a university hospital for 5 years.

METHODS

Participants and Procedure

This study was conducted by retrospectively analyzing the data of children and adolescents aged 6 to 18 years with ADHD who were admitted to a university hospital in Türkiye between November 2017 and November 2022. The data of 1,314 patients whose treatment responses could be evaluated were included in the study. Of these, 276 were patients who did not benefit from the first treatment or whose treatment was changed due to side effects, and their data were included twice for the first and second treatment. Thus, a total of 1,590 patient data were used. The characteristics of the data are shown in Figure 1. The patients’ age, sex, ADHD severity, ADHD treatments, treatment responses, comorbid psychiatric disorders, additional treatments, and chronic illnesses were recorded. ADHD was diagnosed by two child and adolescent psychiatrists and residents through clinical evaluation according to the Diagnostic and Statistical Manual of Mental Disorders (DSM)-5 diagnostic criteria. 28 item Conners’ teacher rating scale, 48 item Conners’ parent rating scale and Atilla Turgay DSM-IV based child and adolescent disruptive behaviour disorders rating scale were used to support clinical assessment, both to clarify the diagnosis of ADHD and to diagnose possible comorbidities. Baseline ADHD severity was determined with the clinical global impression-severity (CGI-S) scale. All patients included in the study had a CGI-S score ≥ 4. CGI-S ≥ 4 was used as an inclusion criteria in other studies involving different disorders [11,12]. Treatment response was evaluated with the clinical global impression-improvement (CGI-I) scale after 8 weeks of treatment at the adequate dosage. As in other studies, CGI-I ≤ 2 was used to define treatment responders [13,14]. Similar to previous studies, the duration was set at 8 weeks to evaluate the short effects of MPH and ATX [15-18]. At the time of inclusion in the study, only MPH and ATX were available in the country for the treatment of ADHD. Comorbid psychiatric disorders were diagnosed according to DSM-5 diagnostic criteria. Following a clinical examination conducted by child and adolescent psychiatrists and residents, which included an assessment of abilities such as abstract thinking, knowledge, judgment, and academic skills, children suspected of having borderline intellectual functioning, intellectual disability, and specific learning disorder were evaluated by three psychologists using the Wechsler Intelligence Scale for Children-Revised. Ethics approval of this study was taken from the Erciyes University ethics committee (2022/822), and the procedures conformed to Declaration of Helsinki.

Fig. 1.

Summary of data included in the study.

ADHD, attention-deficit hyperactivity disorder; ATX, atomoxetine; MPH, methylphenidate; NR, non-responder.

Measures

Conners’ teacher rating scale-28

This 28-item scale was developed to detect students’ behavioral problems at school [19]. Dereboy et al. [20] conducted the Turkish validity study.

Conners’ parent rating scale-48

This 48-item test consists of subscales questioning learning problem, conduct problem, impulsive-hyperactive, psychosomatic, and anxiety symptoms [19]. Dereboy et al. [20] conducted the Turkish validity study.

Atilla Turgay DSM-IV based child and adolescent disruptive behaviour disorders rating scale

This scale was developed to assess the presence of ADHD, oppositional defiant disorder and conduct disorder according to DSM-IV diagnostic criteria and it consists of 41 items [21]. It was translated and adapted into Turkish [22].

Clinical global impression scale

This scale was developed to evaluate the course of psychiatric disorders for use in clinical research. It consists of three dimensions assessing severity of disorder, response to treatment and severity of side effects [23].

Wechsler intelligence scale for children-revised

It was developed to measure the intelligence of children between the ages of 6 to 16 [24]. It detects verbal intelligence, performance intelligence and general intelligence. Turkish adaptation study was conducted [25].

Statistical Analysis

The IBM Statistical Program for Social Sciences was used to analyze the data. Summary statistics of the data were given as percentage (%), number, mean, standard deviation, median and 25th to 75th. In the entire sample, generalized estimating equations were used to determine the treatment response, as 276 patients were included in the data set twice. Binary logistic regression analysis was used to determine predictors of treatment response in patient groups using MPH and ATX. The statistical significance level for all analyses was p < 0.05.

RESULTS

One thousand twenty-seven (78.16%) of the patients were male and 287 (21.84%) were female. The age, drug dosage, MPH formulation, comorbidities, and treatment response rates of patients according to drug groups were presented in Table 1, while the sex, ADHD severity, additional drug use, comorbidity, and chronic disease distributions according to treatment response status were presented in Table 2. There were 276 patients whose treatment was changed due to lack of response or side effects. Ninety-one patients had not responded to MPH monotherapy in the past. There was no response to ATX monotherapy, which was started as the second treatment in 31 (34.1%) of these patients, and to combination therapy in 3 (3.3%). Similarly, of 105 patients who had not responded to ATX monotherapy in the past, 35 (33.3%) had no response to MPH monotherapy, and 3 (2.9%) had no response to combination therapy. Four of 7 patients (57.1%) who had not responded to combination therapy did not respond to MPH monotherapy.

Table 1.

Descriptive statistics for patients in the three treatment groups

Variable	Methylphenidate group (n = 1,026)	Atomoxetine group (n = 506)	Combination therapy group (n = 58)
Age (yr)	11.19 ± 3.17	11.23 ± 2.94	11.39 ± 2.38
Methyphenidate dosage (mg)	27 (18−36)	-	19 (10−36)
Methyhlphenidate ER form	922 (89.9)	-	33 (56.9)
Atomoxetine dosage (mg)	-	41.33 ± 17.29	35.37 ± 15.77
Comorbid psychiatric disorders
Mild intellectual disability	125 (12.2)	64 (12.6)	6 (10.3)
Conduct disorder	83 (8.1)	28 (5.5)	7 (12.1)
Specific learning disorder	78 (7.6)	30 (5.9)	1 (1.7)
Borderline intellectual functioning	71 (6.9)	30 (5.9)	4 (6.9)
Enuresis	31 (3.0)	49 (9.7)	9 (15.5)
Autism spectrum disorder	48 (4.7)	34 (6.7)	4 (6.9)
Anxiety disorders	38 (3.7)	28 (5.5)	1 (1.7)
Tic disorder	16 (1.6)	26 (5.1)	1 (1.7)
Oppositional defiant disorder	29 (2.8)	5 (1.0)	-
Adjustment disorder	16 (1.6)	11 (2.2)	-
Communication disorders	14 (1.4)	11 (2.2)	1 (1.7)
Moderate intellectual disability	14 (1.4)	7 (1.4)	2 (3.4)
Obssessive-compulsive disorder	6 (0.6)	7 (1.4)	1 (1.7)
Multiple	52 (5.1)	38 (7.5)	2 (3.4)
None	405 (39.5)	138 (27.3)	19 (32.8)
CGI-I ≤ 2a	619 (60.3)	237 (46.8)	38 (65.5)

Values are presented as mean ± standard deviation, number (%), or median (interquartile range).

ER, extended release; CGI-I, clinical global impression-improvement scale; mg, milligram; n, number; SD, standard deviation; -, not available.

^aTreatment responders.

Table 2.

Descriptive statistics for patients in the three treatment groups according to treatment responses

Variable	Methylphenidate group CGI-I ≤ 2/CGI- I >2	Atomoxetine group CGI-I ≤ 2/CGI- I >2	Combination therapy group CGI-I ≤ 2/CGI- I >2
Sex
Male	483/319 (60.2/39.8)	188/218 (46.3/53.7)	32/15 (68.1/31.9)
Female	136/88 (60.7/39.3)	49/51 (49.0/51.0)	6/5 (54.5/45.5)
ADHD Severity
Moderate	24/15 (61.5/38.5)	25/17 (59.5/40.5)	1/1 (50.0/50.0)
Marked	556/305 (64.6/35.4)	197/202 (49.4/50.6)	33/10 (76.7/23.3)
Severe	39/87 (31.0/69.0)	15/50 (23.1/76.9)	4/9 (30.8/69.2)
Additional treatment
AAP monotherapy	136/128 (51.5/48.5)	56/67 (45.5/54.5)	10/5 (66.7/33.3)
SSRI monotherapy	16/15 (51.6/48.4)	9/6 (60.0/40.0)	-
TAP monotherapy	3/2 (60.0/40.0)	3/4 (42.9/57.1)	-
Combination of AAP and SSRI	8/10 (44.4/55.6)	1/8 (11.1/88.9)	-
None	456/252 (64.4/35.6)	168/184 (47.7/52.3)	28/15 (65.1/34.9)
Comorbidity
Yes	331/290 (55.3/46.7)	168/200 (45.7/54.3)	24/15 (61.5/38.5)
No	288/117 (71.1/28.9)	69/69 (50.0/50.0)	14/5 (73.7/26.3)
Chronic illness
Yes	75/50 (60.0/40.0)	39/40 (49.4/50.6)	6/2 (75.0/25.0)
No	544/357 (60.4/39.6)	198/229 (46.4/53.6)	32/18 (64.0/36.0)

CGI-I ≤ 2: treatment responders, CGI-I > 2: treatment non-responders.

Values are presented as number (%).

AAP, atypical antipsychotic; ADHD, attention-deficit hyperactivity disorder; CGI-I, clinical global impression-improvement scale; SSRI, selective serotonin reuptake inhibitor; TAP, typical antipsychotic; -, not available.

The results of the generalized estimating equations analysis, which included 1,590 data to determine predictors of treatment response, were presented in Table 3 and Table 4. ADHD treatment type, disorder severity, comorbidities, and age were found to be significant predictors in this analysis (Table 3). Using combination therapy of MPH and ATX as the reference category, ATX monotherapy showed a significantly lower odds of treatment response (odds ratio [OR] = 0.438, 95% confidence interval [CI]: 0.248−0.771, p = 0.004). Moderate and marked disorder severity were associated with higher odds compared to severe disorder (OR = 4.634, 95% CI: 2.551−8.415, p < 0.001; OR = 3.568, 95% CI: 2.477−5.140, p < 0.001, respectively). When the group without comorbidities taken as the reference category, borderline intellectual functioning, mild intellectual disability, conduct disorder, autism spectrum disorder (ASD), and adjustment disorder showed significantly lower odds (OR = 0.478, 95% CI: 0.313−0.729, p = 0.001; OR = 0.687, 95% CI: 0.486−0.970, p = 0.033; OR = 0.471, 95% CI: 0.296−0.752, p = 0.002; OR = 0.378, 95% CI: 0.245−0.584, p < 0.001; OR = 0.221, 95% CI: 0.095−0.516, p < 0.001, respectively). Age was significantly associated with treatment response (OR = 1.053, 95% CI: 1.017−1.091, p = 0.004) (Table 4).

Table 3.

Tests of model effects from generalized estimating equations

Variable	Wald χ2	df	p
Intercept	4.628	1	0.031
Sex	0.474	1	0.491
ADHD treatment	23.244	2	< 0.001
Disorder severity	48.559	2	< 0.001
Chronic diseases	0.331	1	0.565
Additional treatment	1.818	4	0.769
Comorbidity	48.694	14	< 0.001
Age	8.316	1	0.004

ADHD, attention-deficit hyperactivity disorder.

Table 4.

Generalized estimating equations^a analysis of factors associated with treatment response

Variable	B	SE	95% Wald confidence interval			Hypothesis test			Exp(B)	95% Wald confidence interval for Exp(B)
			Lower	Upper		Wald χ2	df	p		Lower	Upper
Intercept	−0.692	0.4446	−1.563	0.179		2.422	1	0.120	0.501	0.209	1.197
Male sex	0.092	0.1332	−0.169	0.353		0.474	1	0.491	1.096	0.844	1.423
Use of MPH	−0.281	0.2843	−0.838	0.276		0.978	1	0.323	0.755	0.432	1.318
Use of ATX	−0.826	0.2893	−1.393	−0.259		8.161	1	0.004	0.438	0.248	0.771
Moderate disorder	1.533	0.3044	0.937	2.130		25.368	1	< 0.001	4.634	2.551	8.415
Marked disorder	1.272	0.1862	0.907	1.637		46.657	1	< 0.001	3.568	2.477	5.140
Absence of chronic diseases	−0.086	0.1499	−0.380	0.208		0.331	1	0.565	0.917	0.684	1.231
Use of AAP	−0.013	0.1395	−0.286	0.261		0.008	1	0.927	0.987	0.751	1.298
Use of SSRI	0.011	0.3458	−0.666	0.689		0.001	1	0.974	1.012	0.514	1.992
Use of TAP	−0.119	0.4906	−1.080	0.843		0.058	1	0.809	0.888	0.340	2.323
Use of AAP and SSRI	−0.606	0.4599	−1.507	0.295		1.736	1	0.188	0.546	0.221	1.344
Obsessive-compulsive disorder	0.119	0.7150	−1.282	1.520		0.028	1	0.868	1.126	0.277	4.574
Borderline intellectual functioning	−0.738	0.2156	−1.161	−0.316		11.727	1	0.001	0.478	0.313	0.729
Mild intellectual disability	−0.376	0.1762	−0.721	−0.030		4.546	1	0.033	0.687	0.486	0.970
Specific learning disorder	0.036	0.2190	−0.393	0.466		0.028	1	0.868	1.037	0.675	1.593
Conduct disorder	−0.752	0.2380	−1.218	−0.285		9.978	1	0.002	0.471	0.296	0.752
Autism spectrum disorder	−0.972	0.2215	−1.407	−0.538		19.278	1	< 0.001	0.378	0.245	0.584
Enuresis	−0.381	0.2475	−0.866	0.104		2.368	1	0.124	0.683	0.421	1.110
Communication disorders	−0.592	0.4479	−1.470	0.286		1.746	1	0.186	0.553	0.230	1.331
Adjustment disorder	−1.508	0.4322	−2.355	−0.661		12.177	1	< 0.001	0.221	0.095	0.516
Anxiety disorder	−0.509	0.2889	−1.076	0.057		3.109	1	0.078	0.601	0.341	1.058
Moderate intellectual disability	0.901	0.5697	−0.216	2.017		2.499	1	0.114	2.461	0.806	7.516
Oppositional defiant disorder	−0.217	0.3636	−0.930	0.496		0.356	1	0.551	0.805	0.395	1.641
Tic disorder	−0.398	0.3427	−1.069	0.274		1.346	1	0.246	0.672	0.343	1.315
Multiple	−0.449	0.2325	−0.904	0.007		3.725	1	0.054	0.638	0.405	1.007
Age	0.052	0.0179	0.017	0.087		8.316	1	0.004	1.053	1.017	1.091

AAP, atypical antipsychotic; ATX, atomoxetine; MPH, methylphenidate; SSRI, selective serotonin reuptake inhibitor; TAP, typical antipsychotic; SE, standard error.

^aGeneralized estimating equations = Probability distribution: binomial, link function: logit, working correlation matrix structure: exchangeable.

Following this analysis, it was planned to investigate predictors for each drug group. No analysis was performed for the combination therapy group due to insufficient number of events per variable. To enable analysis within the MPH group, the typical antipsychotic group, which had a low number of patients, was combined with the atypical antipsychotic group to form the antipsychotic group. For similar reasons, the obsessive-compulsive disorder (OCD) group had a low number of patients and was therefore excluded. The results of this analysis were presented in Table 5. Although p < 0.05 for the MPH dosage, it was not considered significant because the 95% CI included 1. Compared with moderate disorder, severe disorder was associated with significantly lower odds of treatment response (OR = 0.237, 95% CI: 0.102−0.548, p = 0.001). When the group without comorbidities taken as the reference category, borderline intellectual functioning, conduct disorder, ASD, enuresis, communication disorders, and adjustment disorder showed significantly lower odds (OR = 0.404, 95% CI: 0.237−0.690, p = 0.001; OR = 0.535, 95% CI: 0.294−0.977, p = 0.042; OR = 0.405, 95% CI: 0.211−0.779, p = 0.007; OR = 0.239, 95% CI: 0.107−0.535, p < 0.001; OR = 0.221, 95% CI: 0.071−0.682, p = 0.009; OR = 0.242, 95% CI: 0.082−0.719, p = 0.011, respectively) (Table 5).

Table 5.

Results of multivariable binary logistic regression analysis^a for treatment response in MPH group

Variable	B	SE	Wald	df	p	Exp(B)	95% Wald confidence interval for Exp(B)
							Lower	Upper
Age	0.028	0.027	1.106	1	0.293	1.028	0.976	1.083
Male sex	0.038	0.168	0.051	1	0.822	1.038	0.748	1.442
Use of ER MPH	0.323	0.245	1.741	1	0.187	1.381	0.855	2.231
MPH dosage	0.012	0.006	4.120	1	0.042	1.012	1.000	1.024
Use of AP	−0.221	0.180	1.517	1	0.218	0.801	0.563	1.140
Use of SSRI	−0.316	0.450	0.493	1	0.483	0.729	0.301	1.762
Use of AAP and SSRI	−0.598	0.550	1.182	1	0.277	0.550	0.187	1.616
Marked disorder	−0.170	0.367	0.215	1	0.643	0.844	0.411	1.732
Severe disorder	−1.441	0.428	11.323	1	0.001	0.237	0.102	0.548
Borderline intellectual functioning	−0.906	0.273	11.036	1	0.001	0.404	0.237	0.690
Mild intellectual disability	−0.395	0.232	2.904	1	0.088	0.674	0.428	1.061
Specific learning disorder	−0.067	0.273	0.060	1	0.806	0.935	0.547	1.598
Conduct disorder	−0.625	0.307	4.145	1	0.042	0.535	0.294	0.977
Autism spectrum disorder	−0.903	0.333	7.353	1	0.007	0.405	0.211	0.779
Enuresis	−1.431	0.410	12.146	1	< 0.001	0.239	0.107	0.535
Communication disorders	−1.511	0.576	6.893	1	0.009	0.221	0.071	0.682
Adjustment disorder	−1.418	0.555	6.522	1	0.011	0.242	0.082	0.719
Anxiety disorder	−0.529	0.392	1.820	1	0.177	0.589	0.273	1.270
Moderate mental retardation	0.059	0.647	0.008	1	0.927	1.061	0.298	3.774
Oppositional defiant disorder	0.083	0.444	0.035	1	0.851	1.087	0.456	2.593
Tic disorder	−0.685	0.557	1.517	1	0.218	0.504	0.169	1.500
Multiple	−0.308	0.336	0.839	1	0.360	0.735	0.381	1.420
Presence of chronic diseases	0.187	0.222	0.710	1	0.400	1.206	0.780	1.864
Constant	0.174	0.489	0.127	1	0.721	1.190	-	-

Omnibus tests of model coefficients: chi-square = 113.097, df = 23, p < 0.001, Nagelkerke R square = 0.143, Hosmer and Lemeshow test: p = 0.773.

AAP, atypical antipsychotic; AP, antipsychotic; ER, extended release; MPH, methylphenidate; SSRI, selective serotonin reuptake inhibitor; SE, standard error; -, not available.

^aBinary logistic regression analysis = Enter method.

In the analysis of the group using ATX, a single antipsychotic group was similarly formed. Patients using selective serotonin reuptake inhibitors were excluded from the analysis due to their limited number. Due to the low number of patients in the moderate intellectual disability group for comorbidities, the groups were combined to form a single intellectual disability group. Patients with OCD and oppositional defiant disorder were excluded due to the small number of observations which prevented reliable regression modeling. ATX dosage was significantly associated with treatment response (OR = 1.016, 95% CI: 1.002−1.030, p = 0.023). Compared with moderate disorder, severe disorder was associated with significantly lower odds of treatment response (OR = 0.195, 95% CI: 0.074−0.511, p = 0.001). When the group without comorbidities taken as the reference category, ASD showed significantly lower odds (OR = 0.291, 95% CI: 0.118−0.717, p = 0.007) (Table 6).

Table 6.

Results of multivariable binary logistic regression analysis^a for treatment response in ATX group

Variable	B	SE	Wald	df	p	Exp(B)	95% confidence interval for Exp(B)
							Lower	Upper
Age	−0.063	0.041	2.332	1	0.127	0.939	0.866	1.018
Male sex	−0.052	0.246	0.045	1	0.832	0.949	0.587	1.536
ATX dosage	0.016	0.007	5.163	1	0.023	1.016	1.002	1.030
Use of AP	0.460	0.262	3.074	1	0.080	1.584	0.947	2.648
Use of SSRI	0.651	0.662	0.967	1	0.325	1.917	0.524	7.019
Marked disorder	−0.483	0.364	1.760	1	0.185	0.617	0.302	1.259
Severe disorder	−1.635	0.491	11.074	1	0.001	0.195	0.074	0.511
Borderline intellectual functioning	0.274	0.426	0.413	1	0.520	1.315	0.570	3.032
Intellectual disability	0.034	0.330	0.011	1	0.917	1.035	0.542	1.976
Specific learning disorder	0.301	0.420	0.515	1	0.473	1.351	0.594	3.076
Conduct disorder	−0.954	0.578	2.726	1	0.099	0.385	0.124	1.195
Autism spectrum disorder	−1.236	0.461	7.196	1	0.007	0.291	0.118	0.717
Enuresis	0.468	0.344	1.851	1	0.174	1.597	0.814	3.133
Communication disorders	0.434	0.658	0.435	1	0.510	1.544	0.425	5.609
Adjustment disorder	−1.672	0.853	3.841	1	0.050	0.188	0.035	1.000
Anxiety disorder	−0.397	0.448	0.787	1	0.375	0.672	0.279	1.617
Tic disorder	0.083	0.449	0.034	1	0.853	1.087	0.451	2.619
Multiple	−0.531	0.410	1.683	1	0.195	0.588	0.263	1.312
Presence of chronic diseases	0.246	0.280	0.775	1	0.379	1.279	0.739	2.214
Constant	0.510	0.574	0.789	1	0.374	1.666	-	-

Omnibus tests of model coefficients: chi-square = 46.502, df = 19, p < 0.001, Nagelkerke R square = 0.122, Hosmer and Lemeshow test: p = 0.741.

AP, antipsychotic; ATX, atomoxetine; SSRI, selective serotonin reuptake inhibitor; SE, standard error; -, not available.

^aBinary logistic regression analysis = Enter method.

DISCUSSION

In this study, predictive factors associated with response to ADHD treatment in children and adolescents were examined. A wide range of variables have been investigated in many studies, but no definitive results have been found. In our study, the first analysis was made with all patient data, and then separate analyses were made in MPH and ATX groups. Thus, response to treatment was evaluated in three different groups. Despite the retrospective design of the study, it differs from other studies due to its large sample size, the separate analysis of treatment response predictors for different drug groups, assessment of comorbidities, as well as additional psychotropic use and chronic diseases, and the inclusion of two different drug responses in 276 patients in the analyses. The presence of both similar and different results across the three separate analyses conducted to identify predictors suggests that different factors may influence treatment response across treatment modalities.

There are different results in the literature regarding age and treatment response [5,6,9]. The increase in treatment response with age in our first analysis may be related to the maturation of dopamine and norepinephrine pathways in the brain, a decrease in drug metabolism rate compared to early childhood, and improved medication compliance. In addition, the fact that more detailed information about ADHD symptoms can be obtained from the child themselves as they get older, while teacher and parent assessments are more prominent in younger children, may also have contributed to this result.

Consistent with the literature, sex was not associated with treatment response in any group. Barkley, Gray and Kagan as well as Pagnier found no predictive effect of sex on stimulant response [5-7]. While studies have reported no sex differences in treatment response with ATX [8,9], one study found that female sex was associated with better treatment response [10].

The relationship between treatment response and initial ADHD severity was common in all three analyses. There are different results in the literature on this subject. Gray and Kagan [6] indicated poor MPH response in low ADHD severity. In the study involving 207 children, patients who did not respond to treatment had more severe hyperactivity-impulsivity symptoms [26]. Higher baseline ADHD severity was associated with lower treatment response in the MTA study [27]. MPH treatment response was stronger in low-severity ADHD [28]. No relationship was found between ATX response and ADHD severity [9]. The fact that dopamine-norepinephrine pathways in the etiology of ADHD are more affected in severely ill patients and the mechanism of action of drugs is through these neurotransmitters may be a factor that reduces the response to treatment. Related to this, participants with lower norepinephrine transporter availability had higher ADHD symptom severity [29]. It has been found that low dopamine synthesis in subcortical regions correlates with the attention deficit symptom severity [30].

In our study, the response rate to MPH was 60.3% and to ATX was 46.8%. Similarly, the MPH response rate was found to be 56% and the ATX response rate was 45% in a randomized-controlled study [31]. Treatment response was found to be lower in patients using ATX monotherapy than in patients using combination therapy. It is known that the effect size of ATX is lower than stimulants [32,33]. The fact that combination therapy was associated with a better response than ATX therapy may be due to the added MPH.

Three analyses related to comorbidities yielded different results. This may be due to both the number of patients in the groups and the selection of medications in some patients due to comorbidities (e.g., the preference for ATX in cases of tic disorder, anxiety disorder, and enuresis). The presence of anxiety, elimination disorders, oppositional defiant symptoms, and depressive symptoms accompanying ADHD was found to be associated with lower MPH response [28,34-37]. The analysis including all patients and the MPH group showed that borderline intellectual functioning and mild intellectual disability are negative predictors of treatment response, indicating that low IQ negatively affects response. Studies indicated that high IQ is associated with better MPH response [28,34]. In addition, due to low mental capacity, it may be difficult to evaluate treatment response in these patients, particularly in terms of attention. Again, in these two groups, ongoing stress due to adjustment disorder may worsen attention processes and impair academic functioning, thereby reducing treatment response. The developing brain may be more susceptible to the neurobiological and cognitive impacts of daily stress [38]. Similarly, conduct disorder can also reduce medication compliance. Enuresis, and communication disorders were identified as negative predictors in the MPH group and as positive predictors in the ATX group, although not statistically significant. This result can be interpreted as ATX being preferred in cases of enuresis comorbidity and MPH potentially increasing anxiety and negatively affecting functioning in cases of communication disorders. The fact that ASD is a negative predictor in all three groups indicates that this disorder plays an important role in treatment response. Children with ASD were found to have a lower response to MPH and more side effects [39]. In this disorder, ongoing neurodevelopmental processes and impaired medication compliance due to drug side effects may reduce the response to treatment.

The drug dosage was found to be a positive predictor in patients using ATX. Since ATX was titrated to 1.2 mg/kg/day in patients, this may suggest children with higher body weight—who are likely to be older—exhibit a better treatment response. As previously explained, this situation can be interpreted similarly to the age-related increase in treatment response.

Our study found that the additional treatments used by patients were not a significant predictor of treatment response. This can be explained by the fact that additional psychotropic medications are more effective on comorbidities and not on core ADHD symptoms.

The presence of chronic diseases was not a predictor of treatment response. Patients with psychiatric disorders and chronic diseases are likely to have other comorbid conditions, drug-drug interactions and impaired psychosocial functioning [40-42]. According to our results, it can be concluded that these or other factors do not affect the treatment response in ADHD patients with chronic disease.

It has been reported that no response or remission occurs in 20−40% of individual with ADHD [43]. In our study, 34.1% of patients who had not responded to MPH also did not respond to ATX treatment. Among the patients whose first treatment was ATX and who had not responded to treatment, 33.3% were non-responders to the MPH treatment. There were also studies that found higher rates than our data. In one study, 43% of patients who did not respond to MPH later responded to ATX. 42% of patients who did not respond to ATX had responded to an OROS treatment in the past [31]. In the study including 75 patients, 68% of patients who did not respond to ATX did not respond to MPH also, and 49% of patients who did not respond to MPH did not respond to ATX [44].

Limitations

The study has some limitations. First, its design was retrospective, and data were obtained from patient files. No semi-structured interviews were used to diagnose both ADHD and comorbidities for all patients. Due to missing data, subtypes and IQ could not be included in the analysis. Since the analysis was limited to data obtained from a single institution, it is difficult to generalize the findings to other cultural contexts or healthcare systems. Additionally, patient treatment compliance and long-term follow-up data for patients were not included. Finally, one of the most significant limitations of this study was the use of only the CGI-I scale to assess treatment response. Although the CGI-I is widely used in clinical research due to its simplicity and clinical relevance, it remains a clinician-based assessment and may therefore be influenced by subjective judgments. In a large naturalistic data set with psychiatric evaluations performed by multiple clinicians, differences in clinical assessment practices can sometimes lead to interpersonal variability, which may result in misclassification of outcomes. This type of variability could attenuate genuine differences in comparisons across medication groups, comorbidity profiles, and severity levels. Therefore, since relying on a single assessment tool and an outcome measure dependent solely on the clinician may limit the robustness of the findings, the results should be interpreted carefully. Future studies incorporating additional standardized symptom scales or outcomes reported by patients, teachers, or parents may help provide a more comprehensive assessment of treatment response.

Conclusions

The primary treatment method for ADHD, whose symptoms begin in childhood and continue into adulthood, is still medication. Although there are very effective drugs used in treatment, the number of patients who do not respond to treatment is considerable. Knowing which factors are associated with response and resistance can be used to predict the patient’s prognosis. Since ADHD is a heterogeneous condition in terms of clinical features, severity and possible biology involved in etiology, individualized treatments are very important [45]. Individualization of ADHD treatment may be one of the important steps to improve treatment response. Since each individual may respond differently to treatment, assessing and knowing the factors affecting treatment response may help to understand which treatment modalities are more effective individually and help patients by identifying and modifying changeable factors that negatively affect treatment response. Although the study found that treatment response was associated with more unmodifiable factors such as age, severity of disorder, and comorbidities, the results may provide clinically meaningful information for treatment planning. Patients who initially present with severe symptoms may require earlier assessment of treatment effectiveness, closer clinical monitoring, and faster treatment adjustments. Younger patients may require closer monitoring and possibly additional psychosocial or behavioral support to optimize treatment outcomes. Furthermore, interventions to improve medication adherence especially in specific comorbidities, and effective treatment of comorbidities may increase treatment response. In patients with comorbidities where the response to MPH may be less favorable, ATX may be considered a reasonable alternative. At the same time, the lower treatment response to ATX treatment compared to combination therapy may raise the question of whether MPH should be preferred, at least in the absence of comorbidities, which are the most common reason for initiating ATX. Although not addressed in the study, family functioning and the improvement of psychosocial factors may also have an impact on prognosis. However, it is clear that studies evaluating treatment response in combination with clinical, biological and environmental factors are needed.