What Is a Clinically Relevant Improvement in Quality of Life in Adults With ADHD?

Abstract

Objective: To estimate a minimal clinically important difference (MCID) on the adult ADHD Quality of Life (AAQoL) scale. Method: The MCID was determined from data from short-term (N = 537) and long-term (N = 440), placebo-controlled atomoxetine trials in adults with ADHD. For the anchor-based approach, change in clinician-rated Clinical Global Impressions–ADHD–Severity (CGI-ADHD-S) scores was used to derive MCID. For the distribution-based approach, baseline-to-endpoint mean (SD) changes in AAQoL scores corresponding to 0.5 SD were computed. Results: The MCID was similar (approximately 8-point difference) between the short-term and the long-term treatment groups when either the anchor-based or distribution-based approach was used. Conclusion: These results suggest that approximately 8 points in the change from baseline on the AAQoL is a MCID.

Keywords

attention deficit/hyperactivity disorder quality of life ADHD atomoxetine

Introduction

One key factor for evaluating a treatment effect in controlled clinical trials is the clinical relevance of the selected study endpoints or outcome measures, together with an understanding of what comprises a minimal clinically important difference (MCID) in these endpoints (Leidy & Wyrwich, 2005). The MCID originated from studies of patient-reported outcomes (PROs; Potter, Mathias, Raut, Kianifard, & Tavakkol, 2006; Wheaton & Pope, 2010), in which it represented the smallest improvement that was considered meaningful by the patient (Jaeschke, Singer, & Guyatt, 1989). Establishing the MCID for study endpoints allows the clinical relevance of data from published trial results to be determined. Although the success of clinical trials relies on conventional statistical applications to establish a statistical significance, there is an emerging consensus that statistical significance in the change in an outcome measure does not reflect whether that change is clinically important (Jacobson & Truax, 1991; Norman, Sloan, & Wyrwich, 2003). Jacobson and Truax (1991) proposed the use of a reliable change index (RCI) based on the mean difference between pre-treatment and post-treatment scores with a correction factor (standard error of measurement [SEM]) to account for error in pre- and post-treatment measurements. With the RCI, a clinically meaningful change is indicated when the RCI exceeds 2 standard errors (Jacobson & Truax, 1991; Lutz, Stulz, Martinovich, Leon, & Saunders, 2009; Norman et al., 2003). Indices such as RCI and effect size examine the statistical reliability or standardized degree of change, respectively, but lose the sense of actual improvement observed by the scale (Karpenko, Owens, Evangelista, & Dodds, 2009; Revicki, Hays, Cella, & Sloan, 2008; Sloan, 2005). Because the MCID is a direct reflection of how meaningful and significant a change is to the individual patient and/or to significant individuals in that patient’s life, it may be considered to be more clinically meaningful and interpretable to both clinician and patient (Karpenko et al., 2009; Revicki et al., 2008; Sloan, 2005).

The methodologies used to determine MCID are still evolving, and there are currently no definitive guidelines to provide direction (Revicki et al., 2008). Presently, the two main approaches used to determine the MCID for a specific rating scale are distribution-based and anchor-based methods (Guyatt, Osoba, Wu, Wyrwich, & Norman, 2002; Revicki et al., 2008). The distribution-based method does not directly measure a minimum change but compares the change in the study measure with a measure of variability (e.g., the SEM or standard deviation [SD]; Guyatt et al., 2002; Revicki et al., 2008). It has been argued, on the basis of empirical evidence buttressed by statistical and physiologic evidence that also substantiates improvement, that one half of the SD of the measured change represents a close approximation of the MCID (Guyatt et al., 2002; Norman et al., 2003). With the anchor-based approach, an external criterion, such as clinical laboratory or physiologic measure, which is valid and has relevance to the PRO that is being assessed, is used (Revicki et al., 2008). Based on the anchor measurements, participants can be categorized into groups representing no change, small changes, or large changes. Ideally, the MCID is best estimated when the anchor measurement can detect small but meaningful changes in health status, as it may be overestimated by large changes. It is recommended that several independent anchor measurements are used to confirm the MCIDs obtained (Revicki et al., 2008).

The MCID is of particular importance in studies that investigate psychiatric conditions, such as ADHD. ADHD is one of the major clinical and public health problems in the United States because of its associated morbidity and disability in children, adolescents, and adults (Biederman, 1998; McGough et al., 2005). The prevalence of ADHD in adults is estimated to be 4.4% in the United States, and the worldwide average prevalence is estimated to be 3.5% (de Graaf et al., 2008; Kessler et al., 2006). The characteristic symptoms of ADHD in children are inattention, impulsivity, and hyperactivity (American Psychiatric Association [APA], 2000). In adults, these symptoms manifest as psychological dysfunction and disability, which may lead to significant job impairment, drug and alcohol misuse, family conflicts, violence, traffic violations, and accidents (Adler et al., 2008; Fischer, Barkley, Smallish, & Fletcher, 2007; Klein et al., 2012). Adult patients with ADHD report lower quality of life (QoL) than healthy comparison participants, and the severity of ADHD symptoms is negatively correlated with measures of QoL (Adler, Liebowitz, et al., 2009; Mattos, Louza, Palmini, de Oliveira, & Rocha, 2013).

Regulatory agencies, such as the U.S. Food and Drug Administration and the European Union’s European Medicines Agency, have stated that patient-reported QoL data are important endpoints to measure in the assessment of pharmacological therapies (Committee for Medicinal Products for Human Use [CHMP], 2010a, 2010b; U.S. Department of Health and Human Services, 2009); thus, studies and clinical trials of ADHD medications have commonly used QoL scales as outcome measures. The QoL scales used to measure ADHD include the ADHD Impact Module Adult (AIM-A) Version, Adult ADHD QoL (AAQoL), World Health Organization QoL (WHO QOL), QoL Enjoyment and Satisfaction Questionnaire Short Form (QLESQSF), and 36-item Short Form Health Survey (SF-36; Agarwal, Goldenberg, Perry, & IsHak, 2012). The WHO QoL and SF-36 are examples of generic instruments of measure for QoL, not specifically designed for ADHD. The AIM-A is an adult ADHD-specific QoL scale; however, there are no studies comparing AIM-A with other QoL scales, and AIM-A has been assessed in only a limited number of clinical trials. The AAQoL, another adult ADHD-specific QoL scale, yields results that are consistent with other scales of QoL and is a commonly used measure of QoL changes in many clinical trials (Adler et al., 2008; Adler, Liebowitz, et al., 2009; Brod et al., 2006; Manor et al., 2012; Matza, Johnston, Faries, Malley, & Brod, 2007; Upadhyaya et al., 2013). Moreover, because the AAQoL was designed specifically for adults with ADHD, it yields larger effect sizes than more generic measures of QoL (e.g., the SF-36; Adler, Liebowitz, et al., 2009). However, investigators using the AAQoL or any QoL measure face the challenge of determining the clinical relevance of any observed differences (i.e., MCID) and communicating that relevance to clinicians and patients who will be applying the trial results.

In this article, we used both a distribution- and an anchor-based methodology to determine the MCID of the AAQoL scale from three short-term (N = 537) and two long-term (N = 440), placebo-controlled trials of atomoxetine hydrochloride in adults with ADHD. Atomoxetine hydrochloride is a selective norepinephrine reuptake inhibitor approved for the treatment of ADHD in children ages 6 years and older, adolescents, and adults. Several randomized, placebo-controlled short- and long-term clinical trials demonstrated the efficacy of atomoxetine in the treatment of ADHD in children (Kelsey et al., 2004; Michelson et al., 2002; Michelson et al., 2001; Newcorn et al., 2006; Spencer et al., 2002) and adults (Adler, Spencer, et al., 2009; Buitelaar et al., 2007; Michelson et al., 2003; Spencer et al., 1998; Upadhyaya et al., 2013).

Method

Description of Samples

To provide for more accurate estimates of the MCID on the AAQoL total scores, we used two samples in the present study. Importantly, because the MCID is a measure of the patient’s perception of when a change becomes meaningful, independent of the treatment given, the atomoxetine and placebo treatment groups were pooled for the derivation of the MCID.

Sample 1 consisted of pooled data from three short-term, multi-center, randomized, double-blind, placebo-controlled trials of atomoxetine in adults with ADHD (N = 537). Study durations ranged from 10 to 16 weeks. To measure QoL, the AAQoL administration ranged from two to six times: once at baseline and one to five times depending on the study schedule. To measure ADHD disease severity, the Clinical Global Impressions–ADHD–Severity (CGI-ADHD-S) administration ranged from 7 to 10 times: twice at baseline and 5 to 8 times depending on the study schedule.

Sample 2 consisted of pooled data from two long-term, multi-center, randomized, double-blind, placebo-controlled trials of atomoxetine in adults with ADHD (N = 440). The duration for both long-term studies was approximately 6 months. The AAQoL was administered four times to measure QoL: once at baseline and three times during long-term treatment with atomoxetine. ADHD disease severity was assessed via the CGI-ADHD-S administration, which ranged from 8 to 11 times: twice at baseline and 6 to 9 times depending on the study schedule.

For both samples, studies enrolled patients with ADHD as defined by the Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV; APA, 1994; 4th ed., text rev.; DSM-IV-TR; APA, 2000), which requires adult patients to have had ADHD symptoms during childhood. Diagnoses were confirmed in all studies using structured diagnostic interviews by physicians, clinical psychologists, or other clinical researchers who were experienced in the assessment of ADHD in adults. Short-term studies in Sample 1 excluded patients with a history of bipolar, psychotic, and substance abuse disorders. Long-term studies in Sample 2 excluded patients with current major depressive disorder and current anxiety disorders, in addition to excluding patients with bipolar, psychotic, and substance abuse disorders.

Measures

The AAQoL is a patient-reported measure. It is composed of 29 items and was developed to specifically assess health-related QoL during the past 2 weeks among adults with ADHD. Each item is rated by patients on a 5-point Likert-type scale ranging from “not at all/never” (1) to “extremely/very often” (5). The AAQoL yields a total score (based on all 29 items) and four subscale scores: Life Productivity, Psychological Health, Life Outlook, and Relationships. Total and subscale scores are computed by (a) reversing scores for all items except the seven items in the Life Outlook subscale, (b) transforming all item scores to a 0- to 100-point scale (1 = 0; 2 = 25; 3 = 50; 4 = 75; 5 = 100), and (c) summing item scores and dividing by the item count to generate subscale and total scores. Higher scores on the AAQoL indicate better functioning.

The CGI-ADHD-S was administered by the clinician to assess the severity of ADHD in relation to the clinician’s total experience with patients who have ADHD. The CGI-ADHD-S is a single-item rating provided from a 7-point scale with 1 meaning normal, not at all ill and 7 meaning among the most extremely ill (Guy, 1976).

Data Analysis

All analyses were performed using data from Sample 1 (short-term treatment) and Sample 2 (long-term treatment). A single distribution-based method was used to estimate the MCID in AAQoL total scores for each sample (atomoxetine and placebo treatment groups were pooled to assess MCID). The SD of baseline-to-endpoint mean change scores in AAQoL total scores was divided by 2 to establish a 0.5-SD estimate in each sample, which may approximate a MCID for PRO measures (Norman et al., 2003).

The anchor-based approach was also used to determine the MCID. Although there is a paucity of published studies aimed at determining a MCID in patients with ADHD, guidance for the use of an appropriate anchor for the present investigation was available from studies showing weak to moderate, but significant, correlations between AAQoL scores or subscores and CGI-ADHD-S scores (Brod et al., 2015; Matza et al., 2007). For example, correlation between the AAQoL change scores and the change in the CGI-ADHD-S ranged from −.37 to −.50 (Matza et al., 2007), and thus fell above the correlation threshold of .30 to .35 (Revicki et al., 2008). Accordingly, the change in the CGI-ADHD-S was chosen as an anchor in this study. This choice of anchor is supported by a recent study in which, although determination of a MCID was not an explicit aim, the Sheehan Disability Scale was anchored to the Clinical Global Impression–Improvement (CGI-I) to detect a clinically meaningful change in adult patients with ADHD (Coles et al., 2014). For the anchor-based analyses, patients were categorized into moderate/much-improvement (−5 to −2), slight-improvement (−1), or no-improvement (0) groups based on change in clinician-rated CGI-ADHD-S scores, as described by Matza and colleagues (2007). Thus, a negative CGI-ADHD-S change score indicated improvement in disease severity. As the aim here was to determine a minimal detectable change, changes indicating moderate to much improvement were combined, whereas those indicating slight or no improvement were segregated. Baseline-to-endpoint AAQoL total score mean (SD) changes were calculated for each CGI-ADHD-S category for each sample. The MCID was calculated as the difference in AAQoL total score mean changes for CGI-ADHD-S slight- and no-improvement groups (Revicki et al., 2008).

To supplement and to examine the robustness of the MCID estimates, cumulative distribution, response rate, and responsiveness were assessed. All analyses were done by treatment groups (atomoxetine and placebo).

Cumulative distribution was evaluated by assessing the percentage distribution of patients by mean change from baseline to endpoint in AAQoL total scores.

Responder analysis was conducted to show the percentage of patients who had an improvement of at least 0.5 SD on the AAQoL, comparing atomoxetine and placebo groups for each sample using the Cochran–Mantel–Haenszel test.

To further examine how improvement by 0.5 SD on the AAQoL discriminates responders from nonresponders and corresponds with the CGI-ADHD-S categories, responsiveness was examined by calculating the AAQoL mean (standard error [SE]) change for each CGI-ADHD-S category for patients showing an improvement in the AAQoL of at least 0.5 SD (responders) and for patients showing an improvement of less than 0.5 SD (nonresponders). This analysis is similar to other studies examining responsiveness of the AAQoL (Revicki et al., 2008).

Finally, the magnitude of MCID relative to the sample was examined between the distribution- and anchor-based approaches by computing the percentage of the baseline AAQoL total score represented by the MCID for each sample (atomoxetine and placebo treatment arms were pooled), that is, dividing the MCID estimate by the mean total scale score.

Results

Sample Description

A brief overview of the demographic of each sample is provided in Table 1. Patients were between approximately 18 and 65 years of age for Sample 1 and between 18 and 55 years of age for Sample 2. Most patients were White.

Table 1.

Baseline Demographic and Clinical Characteristics of Study Samples.

	Sample 1			Sample 2
Variable	Placebo N = 555	ATX N = 537	Total N = 1,092	Placebo N = 344	ATX N = 440	Total N = 784
Sex, n (%)
Female	268 (48.3)	255 (45.7)	523 (47.9)	151 (43.9)	211 (48.0)	362 (46.2)
Male	287 (51.7)	282 (52.5)	569 (52.1)	193 (56.1)	229 (52.0)	422 (53.8)
Age, years
M (SD)	31.7 (9.9)	32.2 (10.2)	32.0 (10.1)	37.6 (9.3)	37.6 (9.4)	37.6 (9.3)
Range (minimum-maximum)	17.6-64.4	18.2-64.7	17.6-64.7	18.5-54.6	18.2-54.9	18.2-54.9
Weight, kilogram
M (SD)	75.6 (20.6)	76.3 (22.5)	76.0 (21.5)	85.1 (20.3)	83.2 (19.1)	84.0 (19.6)
Ethnicity, n (%)
White	276 (49.7)	280 (52.1)	556 (50.9)	293 (85.2)	378 (85.9)	671 (85.6)
Asian	190 (34.2)	178 (33.1)	368 (33.7)	0 (0.0)	0 (0.0)	0 (0.0)
Hispanic	40 (7.2)	41 (7.6)	81 (7.4)	21 (6.1)	25 (5.7)	46 (5.9)
African/African descent	33 (6.0)	27 (5.0)	60 (5.5)	23 (6.7)	20 (4.5)	43 (5.5)
East/Southeast Asian	9 (1.6)	7 (1.3)	16 (1.5)	2 (0.6)	2 (0.5)	4 (0.5)
Western Asian	2 (0.4)	3 (0.6)	5 (0.5)	2 (0.6)	3 (0.7)	5 (0.6)
Native American	1 (0.2)	0 (0.0)	1 (0.1)	0 (0.0)	0 (0.0)	0 (0.0)
Other	4 (0.7)	1 (0.2)	5 (0.5)	3 (0.9)	12 (2.7)	15 (1.9)

Note. ATX = atomoxetine; N/n = number of participants.

Distribution-Based Evaluation

Summarizing the distribution-based results across both samples, the AAQoL mean changes for Samples 1 and 2 were 11.69 (SD = 15.79) and 11.79 (SD = 16.09), respectively. The estimates of MCIDs for the AAQoL (i.e., 0.5 SD) were a 7.89-point improvement with short-term treatment (Sample 1) and an 8.05-point improvement with long-term treatment (Sample 2).

Anchor-Based Evaluation

Mean changes in AAQoL total scores by CGI-ADHD-S criteria for disease severity for Sample 1 (short-term treatment) and Sample 2 (long-term treatment) are shown in Figure 1. Baseline-to-endpoint AAQoL mean (SD) changes in the moderate/much-, slight-, and no-improvement groups with short-term treatment were 21.31 (17.11), 12.38 (13.75), and 4.30 (12.24), respectively, and with long-term treatment were 23.84 (16.41), 11.21 (12.56), and 2.83 (11.30), respectively. Estimates of the moderate/much- and slight-improvement groups were consistently higher and in the expected direction than were those for the no-improvement group. The mean change score differences for Sample 1 showed an 8.08-point difference between the slight- and no-improvement groups. For Sample 2, an 8.37-point difference was observed between the slight- and no-improvement groups. Thus, with short- and long-term treatment, the MCID was 8.08 and 8.37 points, respectively.

Figure 1.

Mean change from baseline in adult ADHD quality of life total score by CGI-ADHD-S in Samples 1 and 2.

Robustness of Distribution and Anchor-Based Minimal Clinically Important Differences

Figures 2A and 2B show the cumulative percentage of patients in Sample 1 (A) and Sample 2 (B) by mean change from baseline to endpoint in AAQoL total score. The cumulative percentages corresponding to MCID for atomoxetine and placebo groups were 60.2% and 50.5% in Sample 1 and 58.0% and 48.0% in Sample 2, respectively. These results correspond to the responder analysis of patients with at least 0.5 SD (i.e., 7.89 points) for Samples 1 and 2. The atomoxetine groups had significantly greater response rates compared with placebo groups for both Sample 1 (9.7% difference, p < .001) and Sample 2 (10.0% difference, p = .005).

Figure 2.

Cumulative percentage of patients in Sample 1 (A) and Sample 2 (B) by mean change from baseline to endpoint in AAQoL total score.

Table 2 shows how improvement by ≥0.5 SD on the AAQoL discriminates responders from nonresponders by presenting mean (SE) change from baseline to endpoint in responders and nonresponders on the AAQoL by CGI-ADHD-S category for each sample. These analyses show that for Sample 1, patients with ≥0.5 SD response (≥7.89-point improvement) had a percentage change from baseline at endpoint of 63.8%, 48.9%, and 45.1% (using mean change in AAQoL total score) in the moderate/much-, slight-, and no-improvement groups, respectively. When looking at nonresponders from Sample 1 (i.e., <0.5 SD or a <7.89-point improvement), the percentage change at endpoint was 0.2%, −0.8%, and −6.1% in the moderate/much-, slight-, and no- improvement groups, respectively. For Sample 2, patients with a ≥0.5 SD response (≥8.05-point improvement) had a percentage change from baseline at endpoint of 61.0%, 44.6%, and 38.4% (using mean change in AAQoL total score) in the moderate/much-, slight-, and no-improvement groups, respectively. When looking at nonresponders from Sample 2 (i.e., <0.5 SD response or <8.05-point improvement), percentage change at endpoint was 3.2%, −0.8%, and −4.2% in the moderate/much-, slight-, and no-improvement groups, respectively. Within the group of patients classified as responders, a similar magnitude of differentiation across CGI-ADHD-S categories (moderate/much-, slight-, and no-improvement) was observed (61%-64%, 45%-49%, 38%-45%) for Samples 1 and 2. Within the group of patients classified as nonresponders, a similar magnitude of differentiation was observed as well.

Table 2.

Mean (SEM) Baseline and Change From Baseline to Endpoint in AAQoL in Responders and Nonresponders Segregated According to CGI-ADHD-S Category Following Short- and Long-Term Treatment.

		CGI-ADHD-S category
Response	Treatment	Moderate/Much improvement	Slight improvement	No improvement
Sample 1 (short-term treatment)
Responders	Pooled atomoxetine + placebo (N)	144	138	119
	Baseline	44.0 (1.2)	42.5 (1.0)	37.5 (1.2)
	Change	28.1 (1.1)	20.8 (0.8)	16.9 (0.7)
	Atomoxetine (N)	89	85	36
	Baseline	44.2 (1.5)	43.8 (1.4)	35.3 (2.3)
	Change	29.3 (1.4)	20.5 (1.0)	16.9 (1.3)
	Placebo (N)	55	53	83
	Baseline	43.6 (2.1)	40.3 (1.6)	38.5 (1.4)
	Change	26.2 (1.7)	21.3 (1.5)	16.9 (0.9)
Nonresponders	Pooled atomoxetine + placebo (N)	46	91	209
	Baseline	59.4 (1.9)	50.2 (1.4)	47.6 (1.0)
	Change	0.1 (1.2)	−0.4 (0.8)	−2.9 (0.5)
	Atomoxetine (N)	26	52	75
	Baseline	58.1 (2.9)	51.4 (1.7)	46.0 (1.7)
	Change	−0.5 (1.7)	0.3 (1.2)	−2.8 (0.8)
	Placebo (N)	20	39	134
	Baseline	61.2 (2.4)	48.6 (2.2)	48.5 (1.2)
	Change	0.9 (1.6)	−1.3 (1.0)	−2.9 (0.7)
Sample 2 (long-term treatment)
Responders	Pooled atomoxetine + placebo (N)	209	126	84
	Baseline	45.9 (0.9)	43.5 (1.2)	43.8 (1.5)
	Change	28.0 (1.0)	19.4 (0.7)	16.8 (0.8)
	Atomoxetine (N)	130	79	46
	Baseline	46.3 (1.1)	43.5 (1.4)	43.4 (1.9)
	Change	28.7 (1.2)	19.7 (0.9)	17.3 (1.2)
	Placebo (N)	79	47	38
	Baseline	45.3 (1.4)	43.4 (2.1)	44.3 (2.4)
	Change	26.8 (1.6)	18.9 (1.2)	16.2 (1.1)
Nonresponders	Pooled atomoxetine + placebo (N)	39	89	236
	Baseline	56.6 (2.1)	48.6 (1.4)	49.8 (0.8)
	Change	1.8 (1.0)	−0.4 (0.8)	−2.1 (0.5)
	Atomoxetine (N)	28	47	110
	Baseline	57.4 (2.7)	49.1 (1.9)	49.2 (1.2)
	Change	2.5 (1.2)	0.6 (0.9)	−2.1 (0.7)
	Placebo (N)	11	42	126
	Baseline	54.7 (2.5)	48.1 (2.2)	50.2 (1.0)
	Change	−0.2 (1.8)	−1.5 (1.2)	−2.2 (0.7)

Note. Response: Increase in the AAQoL total score of ≥0.5 SD. Nonresponse: Increase in the AAQoL total score of <0.5 SD or a decrease. Moderate/much improvement: Change in AAQoL total score = −5 to −2. Slight improvement: Change in AAQoL total score = −1. No improvement: Change in AAQoL total score ≥0.

Note. SEM = standard error of the mean; AAQoL = adult ADHD Quality of Life; CGI-ADHD-S = Clinical Global Impressions–ADHD–Severity; N/n = number of participants.

Another confirmation of similar magnitude of MCID between the distribution- and anchor-based approaches was examined by computing the percentage of the AAQoL total score represented by the MCID for each sample (i.e., dividing the MCID estimate by the mean total scale score; Eton et al., 2004). Mean (SD) total score at baseline was 44.44 (14.05) for Sample 1 and 47.10 (13.46) for Sample 2. For Sample 1, this score equates to 17.75% for the distribution-based MCID (7.89 points) and 18.18% for the anchor-based MCID (8.08 points). For Sample 2, this score equates to 17.09% for the distribution-based MCID (8.05 points) and 17.77% for the anchor-based MCID (8.37 points). Hence, it would appear that the MCID is similar in magnitude, between 17% and 18% of the baseline AAQoL total score.

Discussion

Interpreting changes in outcomes of clinical trials of ADHD should be viewed from a perspective broader than only the statistical significance of the findings to determine a clinically meaningful event (Lutz et al., 2009; Norman et al., 2003; Revicki et al., 2008; Sloan, 2005). The MCID in outcome measures provides a conceptual framework to assist in clinical trial interpretation and a methodology to assess the clinical relevance of study results. Although the MCID has been suggested for a wide range of health-related QoL outcomes (Norman et al., 2003), our study is the first to have applied the MCID to an ADHD outcome measure, the AAQoL.

Across two data sets of patients with ADHD reporting QoL on the AAQoL, we estimated clinically relevant differences. We focused on the AAQoL total score as a targeted endpoint because it represents one of the most clinically relevant outcomes of QoL in adult patients with ADHD, as it shows good internal consistency (0.93 overall), test–retest stability (intra-class r = .86), and construct validity (Brod et al., 2015; Brod et al., 2006; Matza et al., 2007). We chose change on the CGI-ADHD-S as an anchor in this study because it is widely understood, used with face validity, and is significantly correlated with change in the AAQoL (r = −.37 to −.50; Matza et al., 2007), which exceeds the correlation threshold proposed by Revicki and colleagues of .30 to .35 between the PRO change score and the anchor (Revicki et al., 2008).

We showed that for both samples, the initial distribution-based estimates of MCID converged with the anchor-based estimates of MCID: 7.89 points versus 8.08 points for Sample 1, respectively, and 8.05 points versus 8.37 points for Sample 2, respectively. These results suggest that a MCID of approximately 8 points on the AAQoL is a clinically relevant improvement in QoL and suggest a convergence of the distribution- and anchor-based methodologies in estimating a MCID on the AAQoL. These results are consistent with MCID studies of chronic obstructive pulmonary disease (COPD), heart failure, asthma, and various cancers, which have shown convergence of distribution- and anchor-based approaches on a variety of health-related QoL outcome measures (Cella, Eton, Lai, Peterman, & Merkel, 2002; Eton et al., 2004; Sloan, 2005). In addition, we also confirmed similarities between the distribution- and anchor-based MCID estimates by directly comparing their magnitudes relative to baseline AAQoL total scores. Our comparison of relative MCID values between acute and long-term studies revealed that the MCID is likely to be between 17% and 18% of the baseline AAQoL total score. These results may have implications for generating general MCID guidelines for all ADHD QoL scales and aggregate scores (Norman et al., 2003). It should be emphasized that a change of 17% to 18% from baseline represents the minimum change that patients report as being meaningful to them and is not meant to indicate a marked improvement in the disorder. Clearly, there is room for further improvement in the subjective and objective measures of ADHD beyond the MCID.

Improvement by ≥0.5 SD on the AAQoL was used to discriminate responders from nonresponders. In one analysis, we demonstrated that change in AAQoL total scores, based on ≥0.5 SD response criteria, discriminated among groups of patients with different levels of change, as indicated by clinician ratings on CGI-ADHD-S (i.e., moderate/much-, slight-, and no-improvement groups). In addition, consistency was demonstrated in the percentage change at endpoint for the responder groups between Sample 1 and Sample 2. When examining robustness, the percentage of patients showing improvement of at least 0.5 SD on the AAQoL was compared between atomoxetine and placebo for each sample. The percentage of patients showing improvement of at least 0.5 SD on the AAQoL was significantly higher in atomoxetine- versus placebo-treated patients for Sample 1 (60.2% vs. 50.5%; p < .001) and for Sample 2 (58.0% vs. 48.0%; p = .005). The results of the latter analysis are consistent with both short- and long-term, placebo-controlled studies of atomoxetine, which showed improved functioning as measured by the AAQoL following atomoxetine treatment (Adler et al., 2008; Adler, Liebowitz, et al., 2009; Upadhyaya et al., 2013). The placebo response rates reported in the present investigation are consistent, and within the range, of placebo responses reported in the ADHD literature, as placebo response rates tend to trend higher in ADHD studies (Waxmonsky, Waschbusch, Glatt, & Faraone, 2011). For example, in one recent 10-week study of adults with ADHD, the placebo response rates were 54% and 45% for patients evaluated with Conners’ Adult ADHD Rating Scale–Investigator-rated: Screening Version (CAARS-Inv:SV) and Conners’ Adult ADHD Rating Scale–Self-report: Screening Version (CAARS-S:SV), respectively (Lee et al., 2014). The placebo response rate varies considerably among different psychiatric disorders, and it is believed that a high placebo response rate is associated with the level of subjective distress felt by the patient (Khan et al., 2005; Waxmonsky et al., 2011). It should be noted that the present investigation utilized a subjective assessment and is designed to determine a minimal response felt by the patient, including placebo response.

The variability in placebo responses may also explain, in part, why a fairly large number of individuals receiving placebo in the short-term sample were considered responders on the AAQoL, but showed no improvement on the CGI-ADHD-S. The AAQoL is a PRO that addresses how the patient feels, and consists of 29 items, and as such is more sensitive to change compared with the CGI-ADHD-S, which consists of a single item scored by a clinician, based on observations of the patient.

Collectively, our analyses suggest that the threshold for discrimination for changes in health-related QoL, as determined by the AAQoL for ADHD in adults in the anchor-based analyses, appears to be approximately 0.5 SD. Our findings are consistent with those of Norman et al. who examined health-related QoL literature in patients with chronic disease to determine whether there is consistency in the magnitude of minimally clinically important differences (Norman et al., 2003). Across 38 studies, Norman et al. showed that, in most cases, the threshold of discrimination for changes in health-related QoL appears to be approximately 0.5 SD. They cited George Miller’s 1956 report to explain this consistency across the literature, which showed that the limit of people’s ability to discriminate over a wide range of tasks is approximately one part in seven, which is very close to 0.5 SD (Miller, 1956; Norman et al., 2003). Sloan has also noted that many of the competing methods for MCID seem to converge into the same general area in terms of proportion of SD of the PRO instrument under study (Sloan, 2005).

Some limitations of the study should be noted. First, only one distribution-based method and one anchor-based method were chosen in this study to determine the MCID. While the methods chosen to determine the MCID for this study are similar to others (Norman et al., 2003; Revicki et al., 2008), some authors have recommended that to estimate the MCID, several anchor-based methods, with relevant clinical- or patient-based indications, and several distribution-based methods (i.e., effect size, standardized response mean, SEM) should be used to determine convergence on a single value or small range of values for the MCID (Leidy & Wyrwich, 2005; Matza et al., 2011). Because MCID estimates based on a measure of dispersion (such as 0.5 SD) are subject to some bias due to sample heterogeneity, it has also been suggested that distribution-based criteria that are less dependent on sample characteristics must be introduced, such as the standard error of the mean (Hays, Farivar, & Liu, 2005). Nevertheless, the debate over the relative merits of distribution- and anchor-based approaches continues to be discussed in the PRO literature. Although some researchers who have investigated the MCID have expressed concerns over the validity of distribution-based estimates (Hays et al., 2005), others defend the use of distribution-based methods as an important validating element of the derivation of the MCID (Leidy & Wyrwich, 2005; Revicki et al., 2008). Despite those limitations, our results confirmed the MCIDs by several different types of analyses.

A second potential limitation that should be considered is that Sample 1 had a much higher proportion of Asians than did Sample 2. It is generally accepted that there are considerable cultural differences that can affect the diagnosis and therapy of mental disorders, including ADHD (Ang et al., 2009; Hodgkins et al., 2012; Leu, Wang, & Koo, 2011; Moon, 2011; Timimi & Taylor, 2004). Interestingly, individuals in Western cultures tend to associate positive feelings with success and self-esteem, and there is a strong negative correlation between positive and negative feelings, whereas in Asian cultures, there is a greater tendency to report simultaneous good and bad feelings (Leu et al., 2011), which may have implications in determining a MCID. However, in the present investigation, the outcomes were similar for both samples, suggesting that the different ethnic makeup of the cohorts most likely did not affect the results of the investigation.

A third limitation one might consider is that the study used for Sample 2 excluded individuals with major depressive disorder and anxiety disorders. As atomoxetine has shown efficacy against anxiety disorders present alone or comorbid with ADHD (Geller et al., 2007; Ravindran, Kim, Letamendi, & Stein, 2009), it is possible that individuals in Sample 1 who had anxiety might perceive improvement before those with ADHD alone. However, again, the outcomes between the two groups were similar, suggesting that the possibility of patients with anxiety in one study did not alter the determination of MCID. There is little evidence that atomoxetine is an antidepressant; hence, it is unlikely that improvement in depressive symptoms may have affected outcome.

Clearly, clinicians need a systematic way to assess the perceived benefit of ADHD treatment based on individual patient improvement relative to both cost and risk of complications. A MCID would ideally provide a specific threshold to serve as a treatment goal, and MCID is already used in that regard in other chronic diseases. The success of treatment would then be measured by the proportion of patients who reach MCID, as opposed to the average change of a group of patients. Beyond the values presented here, additional studies are needed to define MCID thresholds for ADHD QoL.

Footnotes

Acknowledgements

We would like to thank Maria Rovere, MTSC, Angela Lorio, and Michael Ossipov of inVentiv Health Clinical, LLC, for their help with editing and formatting. Eli Lilly and Company contracted inVentiv Health Clinical, LLC, for writing and editorial services.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Yoko Tanaka and Himanshu Upadhyaya are employees of Eli Lilly and Company. Meryl Brod and Jeannine Lane have been paid consultants to Eli Lilly.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The trials were funded and sponsored by Eli Lilly and Company, Indianapolis, IN, USA, and/or any of its subsidiaries.

Author Biographies

Yoko Tanaka is a research advisor in global statistics sciences with Eli Lilly & Company. She completed her PhD in Statistics from the University of Florida. She joined Eli Lilly & Company in 1998 and has led many neuroscience projects. Her research of interest has been in neuroscience therapeutic area and statistical considerations in multi-regional clinical trials.

Meryl Brod is the President of the The Brod Group, a Health Outcomes and Phase IV strategic planning consultancy with an emphasis on patient centered research. She is an expert in bridging clinical and marketing needs both in the United States and Europe, has designed and successfully implemented numerous clinical trial piggyback and stand-alone outcomes research studies. Dr. Brod is an internationally recognized expert in the development of patient-reported measures (PROs) to evaluate the impact of disease and treatment as well as the strategy and design of Late Phase studies including Registries, Comparator Trials, and Effectiveness Studies.

Jeannine R. Lane is a consultant principal medical writer with Ardea Biosciences. She completed her PhD in Behavioral Neuroscience from the University of Washington. She began consulting for Eli Lilly & Company in 2008 and has written many regulatory submissions and publications. Her expertise is in neuroscience, diabetes, and gout multi-center Phase 2-4 clinical trials.

Himanshu Upadhyaya is a medical fellow, Biomedicines core team with Eli Lilly & Co, and an adjunct clinical associate professor of Psychiatry at the University of Texas Health Science Center at Houston. He completed his medical training at the University of Bombay, India, and a Masters in Neuropharmacology at the University of Texas at Austin. He completed his general psychiatry residency at the University of Cincinnati, Ohio, and fellowships (Child and Adolescent Psychiatry, and Addiction Psychiatry) at the Medical University of South Carolina (MUSC). He was on the faculty at the MUSC for 10 years teaching, conducting research, clinical practice, and directing the addiction psychiatry fellowship program. He left MUSC as a tenured associate professor to join Eli Lilly & Co medical team in August of 2008 and has been the global lead physician for Strattera since January 2011. He has published many papers and book chapters.

References

Adler

L. A.

Liebowitz

Kronenberger

Qiao

Rubin

Hollandbeck

. . . Durell

(2009). Atomoxetine treatment in adults with attention-deficit/hyperactivity disorder and comorbid social anxiety disorder. Depression and Anxiety, 26, 212-221.

Adler

L. A.

Spencer

Brown

T. E.

Holdnack

Saylor

Schuh

. . . Kelsey

(2009). Once-daily atomoxetine for adult attention-deficit/hyperactivity disorder: a 6-month, double-blind trial. Journal of Clinical Psychopharmacology, 29, 44-50.

Adler

L. A.

Spencer

T. J.

Levine

L. R.

Ramsey

J. L.

Tamura

Kelsey

. . . Biederman

(2008). Functional outcomes in the treatment of adults with ADHD. Journal of Attention Disorders, 11, 720-727.

Agarwal

Goldenberg

Perry

IsHak

W. W.

(2012). The quality of life of adults with attention deficit hyperactivity disorder: A systematic review. Innovations in Clinical Neuroscience, 9, 10-21.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.). Washington, DC: Author.

Ang

Q. Q.

Wing

Y. K.

Sulaiman

A. H.

Chiu

N. Y.

Shen

Y. C.

. . . Dossenbach

(2009). Association between painful physical symptoms and clinical outcomes in East Asian patients with major depressive disorder: A 3-month prospective observational study. International Journal of Clinical Practice, 63, 1041-1049.

Biederman

(1998). Attention-deficit/hyperactivity disorder: A life-span perspective. Journal of Clinical Psychiatry, 59(Suppl 7), 4-16.

Brod

Adler

L. A.

Lipsius

Tanaka

Heinloth

A. N.

Upadhyaya

(2015). Validation of the adult attention-deficit/hyperactivity disorder quality-of-life scale in European patients: Comparison with patients from the USA. Attention Deficit and Hyperactivity Disorders, 7, 141-150.

10.

Brod

Johnston

Able

Swindle

(2006). Validation of the adult attention-deficit/hyperactivity disorder quality-of-life scale (AAQoL): A disease-specific quality-of-life measure. Quality of Life Research, 15, 117-129.

11.

Buitelaar

J. K.

Michelson

Danckaerts

Gillberg

Spencer

T. J.

Zuddas

. . . Biederman

(2007). A randomized, double-blind study of continuation treatment for attention-deficit/hyperactivity disorder after 1 year. Biological Psychiatry, 61, 694-699.

12.

Cella

Eton

D. T.

Lai

J. S.

Peterman

A. H.

Merkel

D. E.

(2002). Combining anchor and distribution-based methods to derive minimal clinically important differences on the Functional Assessment of Cancer Therapy (FACT) anemia and fatigue scales. Journal of Pain and Symptom Management, 24, 547-561.

13.

Coles

Coon

DeMuro

McLeod

Gnanasakthy

(2014). Psychometric evaluation of the Sheehan Disability Scale in adult patients with attention-deficit/hyperactivity disorder. Neuropsychiatric Disease and Treatment, 10, 887-895.

14.

Committee for Medicinal Products for Human Use. (2010a). Guideline on the clinical investigation of medicinal products for the treatment of attention deficit hyperactivity disorder (ADHD). London: European Medicines Agency. Retrieved from www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2010/08/WC500095686.pdf

15.

Committee for Medicinal Products for Human Use. (2010b). Reflection paper on the regulatory guidance for the use of health-related quality of life (HRQL) measures in the evaluation of medicinal products. London: European Medicines Agency. Retrieved from www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC500003637.pdf

16.

de Graaf

Kessler

R. C.

Fayyad

ten Have

Alonso

Angermeyer

. . . Posada-Villa

(2008). The prevalence and effects of adult attention-deficit/hyperactivity disorder (ADHD) on the performance of workers: Results from the WHO World Mental Health Survey Initiative. Occupational & Environmental Medicine, 65, 835-842.

17.

Eton

D. T.

Cella

Yost

K. J.

Yount

S. E.

Peterman

A. H.

Neuberg

D. S.

. . . Wood

W. C.

(2004). A combination of distribution- and anchor-based approaches determined minimally important differences (MIDs) for four endpoints in a breast cancer scale. Journal of Clinical Epidemiology, 57, 898-910.

18.

Fischer

Barkley

R. A.

Smallish

Fletcher

(2007). Hyperactive children as young adults: Driving abilities, safe driving behavior, and adverse driving outcomes. Accident Analysis and Prevention, 39, 94-105.

19.

Geller

Donnelly

Lopez

Rubin

Newcorn

Sutton

. . . Sumner

(2007). Atomoxetine treatment for pediatric patients with attention-deficit/hyperactivity disorder with comorbid anxiety disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 46, 1119-1127.

20.

Guy

(1976). ECDEU assessment manual for psychopharmacology (Rev. ed.). Bethesda, MD: U.S. Department of Health, Education, and Welfare.

21.

Guyatt

G. H.

Osoba

A. W.

Wyrwich

K. W.

Norman

G. R.

(2002). Methods to explain the clinical significance of health status measures. Mayo Clinic Proceedings, 77, 371-383.

22.

Hays

R. D.

Farivar

S. S.

Liu

(2005). Approaches and recommendations for estimating minimally important differences for health-related quality of life measures. COPD, 2, 63-67.

23.

Hodgkins

Arnold

L. E.

Shaw

Caci

Kahle

Woods

A. G.

. (2012). A systematic review of long-term outcomes in ADHD: Global publication trends. Frontiers in Psychiatry, 2, Article 84.

24.

Jacobson

N. S.

Truax

(1991). Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. Journal of Consulting and Clinical Psychology, 59, 12-19.

25.

Jaeschke

Singer

Guyatt

G. H.

(1989). Measurement of health status: Ascertaining the minimal clinically important difference. Controlled Clinical Trials, 10, 407-415.

26.

Karpenko

Owens

J. S.

Evangelista

N. M.

Dodds

(2009). Clinically significant symptom change in children with attention-deficit/hyperactivity disorder: Does it correspond with reliable improvement in functioning? Journal of Clinical Psychology, 65, 76-93.

27.

Kelsey

D. K.

Sumner

C. R.

Casat

C. D.

Coury

D. L.

Quintana

Saylor

K. E.

. . . Allen

A. J.

(2004). Once-daily atomoxetine treatment for children with attention-deficit/hyperactivity disorder, including an assessment of evening and morning behavior: A double-blind, placebo-controlled trial. Pediatrics, 114, e1-e8.

28.

Kessler

R. C.

Adler

Barkley

Biederman

Conners

C. K.

Demler

. . . Zaslavsky

A .M.

(2006). The prevalence and correlates of adult ADHD in the United States: Results from the National Comorbidity Survey Replication. American Journal of Psychiatry, 163, 716-723.

29.

Khan

Kolts

R. L.

Rapaport

M. H.

Krishnan

K. R.

Brodhead

A. E.

Browns

W. A.

(2005). Magnitude of placebo response and drug-placebo differences across psychiatric disorders. Psychological Medicine, 35, 743-749.

30.

Klein

R. G.

Mannuzza

Olazagasti

M. A.

Roizen

Hutchison

J. A.

Lashua

E. C.

. . . Castellanos

F. X.

(2012). Clinical and functional outcome of childhood attention-deficit/hyperactivity disorder 33 years later. Archives of General Psychiatry, 69, 1295-1303.

31.

Lee

S. I.

Song

D. H.

Shin

D. W.

Kim

J. H.

Lee

Y. S.

Hwang

J. W.

. . . Treuer

(2014). Efficacy and safety of atomoxetine hydrochloride in Korean adults with attention-deficit hyperactivity disorder. Asia-Pacific Psychiatry, 6, 386-396.

32.

Leidy

N. K.

Wyrwich

K. W.

(2005). Bridging the gap: Using triangulation methodology to estimate minimal clinically important differences (MCIDs). COPD, 2, 157-165.

33.

Leu

Wang

Koo

(2011). Are positive emotions just as “positive” across cultures? Emotion, 11, 994-999.

34.

Lutz

Stulz

Martinovich

Leon

Saunders

S. M.

(2009). Methodological background of decision rules and feedback tools for outcomes management in psychotherapy. Psychotherapy Research, 19, 502-510.

35.

Manor

Ben-Hayun

Aharon-Peretz

Salomy

Weizman

Daniely

. . . Adler

L. A.

(2012). A randomized, double-blind, placebo-controlled, multicenter study evaluating the efficacy, safety, and tolerability of extended-release metadoxine in adults with attention-deficit/hyperactivity disorder. Journal of Clinical Psychiatry, 73, 1517-1523.

36.

Mattos

Louza

M. R.

Palmini

A. L.

de Oliveira

I. R.

Rocha

F. L.

(2013). A multicenter, open-label trial to evaluate the quality of life in adults with ADHD treated with long-acting methylphenidate (OROS MPH): Concerta Quality of Life (CONQoL) study. Journal of Attention Disorders, 17, 444-448.

37.

Matza

L. S.

Johnston

J. A.

Faries

D. E.

Malley

K. G.

Brod

(2007). Responsiveness of the adult attention-deficit/hyperactivity disorder quality of life scale (AAQoL). Quality of Life Research, 16, 1511-1520.

38.

Matza

L. S.

Van Brunt

D. L.

Cates

Murray

L. T.

(2011). Test-retest reliability of two patient-report measures for use in adults with ADHD. Journal of Attention Disorders, 15, 557-563.

39.

McGough

J. J.

Smalley

S. L.

McCracken

J. T.

Yang

Del’Homme

Lynn

D. E.

. . . Loo

(2005). Psychiatric comorbidity in adult attention deficit hyperactivity disorder: Findings from multiplex families. American Journal of Psychiatry, 162, 1621-1627.

40.

Michelson

Adler

Spencer

Reimherr

F. W.

West

S. A.

Allen

A. J.

. . . Milton

(2003). Atomoxetine in adults with ADHD: Two randomized, placebo-controlled studies. Biological Psychiatry, 53, 112-120.

41.

Michelson

Allen

A. J.

Busner

Casat

Dunn

Kratochvil

Harder

(2002). Once-daily atomoxetine treatment for children and adolescents with attention deficit hyperactivity disorder: A randomized, placebo-controlled study. American Journal of Psychiatry, 159, 1896-1901.

42.

Michelson

Faries

Wernicke

Kelsey

Kendrick

Sallee

F. R.

Spencer

(2001). Atomoxetine in the treatment of children and adolescents with attention-deficit/hyperactivity disorder: A randomized, placebo-controlled, dose-response study. Pediatrics, 108, Article E83.

43.

Miller

G. A.

(1956). The magical number seven plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63, 81-97.

44.

Moon

S. Y.

(2011). Cultural perspectives on attention deficit hyperactivity disorder: A comparison between Korea and the U.S. Journal of International Business and Cultural Studies, 6, 1-11.

45.

Newcorn

J. H.

Michelson

Kratochvil

C. J.

Allen

A. J.

Ruff

D. D.

Moore

R. J.

(2006). Low-dose atomoxetine for maintenance treatment of attention-deficit/hyperactivity disorder. Pediatrics, 118, e1701-e1706.

46.

Norman

G. R.

Sloan

J. A.

Wyrwich

K. W.

(2003). Interpretation of changes in health-related quality of life: The remarkable universality of half a standard deviation. Medical Care, 41, 582-592.

47.

Potter

L. P.

Mathias

S. D.

Raut

Kianifard

Tavakkol

(2006). The OnyCOE-t questionnaire: Responsiveness and clinical meaningfulness of a patient-reported outcomes questionnaire for toenail onychomycosis. Health and Quality of Life Outcomes, 4, 50.

48.

Ravindran

L. N.

Kim

D. S.

Letamendi

A. M.

Stein

M. B.

(2009). A randomized controlled trial of atomoxetine in generalized social anxiety disorder. Journal of Clinical Psychopharmacology, 29, 561-564.

49.

Revicki

Hays

R. D.

Cella

Sloan

(2008). Recommended methods for determining responsiveness and minimally important differences for patient-reported outcomes. Journal of Clinical Epidemiology, 61, 102-109.

50.

Sloan

J. A.

(2005). Assessing the minimally clinically significant difference: Scientific considerations, challenges and solutions. COPD, 2, 57-62.

51.

Spencer

Biederman

Wilens

Prince

Hatch

Jones

. . . Seidman

(1998). Effectiveness and tolerability of tomoxetine in adults with attention deficit hyperactivity disorder. American Journal of Psychiatry, 155, 693-695.

52.

Spencer

Heiligenstein

J. H.

Biederman

Faries

D. E.

Kratochvil

C. J.

Conners

C. K.

. . . Potter

W. Z.

(2002). Results from 2 proof-of-concept, placebo-controlled studies of atomoxetine in children with attention-deficit/hyperactivity disorder. Journal of Clinical Psychiatry, 63, 1140-1147.

53.

Timimi

Taylor

(2004). ADHD is best understood as a cultural construct. The British Journal of Psychiatry, 184, 8-9.

54.

United States Department of Health and Human Services, Food and Drug Administration. (2009). Guidance for industry: Patient-reported outcome measures: use in medical product development to support labeling claims. Retreived from www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM193282.pdf

55.

Upadhyaya

Ramos-Quiroga

J. A.

Adler

L. A.

Williams

Tanaka

Lane

J. R.

(2013). Maintenance of response after open-label treatment with atomoxetine hydrochloride in international European and non-European adult outpatients with attention-deficit/hyperactivity disorder: A placebo controlled, randomised withdrawal study. European Journal of Psychiatry, 27, 187-205.

56.

Waxmonsky

J. G.

Waschbusch

D. A.

Glatt

S. J.

Faraone

S. V.

(2011). Prediction of placebo response in 2 clinical trials of lisdexamfetamine dimesylate for the treatment of ADHD. Journal of Clinical Psychiatry, 72, 1366-1375.

57.

Wheaton

Pope

(2010). The minimally important difference for patient-reported outcomes in spondyloarthropathies including pain, fatigue, sleep, and Health Assessment Questionnaire. Journal of Rheumatology, 37, 816-822.