Brian O ‘Hara, M.D.
Developmental Behavioral Pediatrician in Honolulu Hawaii
Listening skills are critically important for young students since a majority of classroom time is spent “learning by listening.” Weak listening skills are not uncommon and relate to both audiological (hearing acuity, auditory processing) and non-audiological factors (attention control, cognitive –language abilities). It is thus important to have up front screening tests to sort out a student’s listening challenges differentially and allow targeted and timely clinical referrals. The Auditory Processing Domains Questionnaire (APDQ) has been designed for this purpose. It is a parent and/or teacher rating questionnaire which can identify problem listeners in the 7 thru 17 year age group and differentiate those at risk for Auditory Processing disorders (APD), Attention Deficit Hyperactivity Disorders (ADHD), and Language or Non-specific Learning Disorders (LD-NOS). It also can provide ecological validity and document behaviors associated with deficits identified by neuro-audiological tests.
Fifty two performance items were selected following a cross disciplinary literature search. Eighteen Auditory Processing Scale items reflected auditory decoding skills, 3 items prosodic skills, and 5 items auditory integration-binaural separation skills. Five sets of items rated performance in quiet vs noise. The Attention Control items included 6 of the “6 out of 9” behaviors required for the DMSV diagnosis of “Inattentive” ADHD. Language scale items included expressive and receptive communication skills as well as such verbal academic skills as reading and writing.
Three scales were constructed which conformed to the three main factors identified by factor analysis: Auditory Processing (AP) with 31 items, Attention Control (ATT) with 10 items, and Language (Lang) with 11 items. A fourth scale, Targeted Auditory Processing (TAP) was constructed with 18 items from the AP scale which are limited to auditory decoding and hearing. It has simple scoring and has correlated highly with AP scale scores (Anova p ≤ .001). It can be used as a non-differential, “quick screener” for APD. In progress is a modified TAP scale to be used down to age 4.5 years.
Parents or teachers rate their student’s “frequency of observed performance” on the 52 items which are scored on a 4 point modified Likert Scale as follows: 4 points if performed Regularly (over ¾ the time), 3 points if performed Often (over ½ the time), 1 point if performed Sometimes (under ½ the time), and 0 points if performed Infrequently (under ¼ the time).
Scale scores (SS) are reported as “percent of perfect score” which has the following calculation: (sum of item scores in scale) ÷ (4 x the number of items in scale) ÷100. For example a perfect AP scale score is 4 x 31 or 124. If an AP score of 28 occurs then that SS is 71% (28÷124 ÷100 = 71%).
Scale scores are positively distributed. Hence higher scores reflect greater competency. The mean NC group AP score was 84 ±15 %. The mean APD group AP score was 38 ± 14 %.
An Excel spread sheet program is available to do the scoring calculations and report findings including scale score rank percentiles and patterns indicative of APD, ADHD and Learning Disability-NOS risks.
The APDQ requires 10 to 15 minutes to complete and is rated at a 6th to 7th grade “fairly difficult” reading level (Flesh-Kincaid 2012). Data input for the Excel takes 6 to 7 minutes.
Normative Study Overview
Research approval was obtained from the Kaiser Hawaii Center for Health Research. After medical records were reviewed 198 Normal Control (NC) and 53 clinical subjects age 7 to 17 were enrolled. The NC subjects were without evidence of ADHD, APD, or special education services. The Clinical Group subjects included 43 students with medically diagnosed ADHD (ADHD group) and 11 students in special education without evidence of ADHD or APD (LD-NOS group). Twenty subjects with “audiologically” diagnosed APD were recruited from collaborating audiologists to make up the APD group.
The NC group had 102 girls and 98 boys with around 25 subjects for each year and 91 eleven to 18 year olds. Race and sex did not have a statistically significant effects on scale scores. Age was significant only when subjects formed 2 age groups. One hundred seven subjects were in the 7 thru 10 year group and 97 in the 11 to 17 year group. Race, sex and age distributions were similar in the clinical groups except for ADHD which had 70% male and 30 % female.
The racial distribution of the NC group was representative of the state of Hawaii but not of the entire United States (30% white, 25% Asian, 14% Hawaiian/Pacific Islanders and 32 % mixed). Sex and race did not have statistically significant effects on Scale scores.
Thirty five percent of parents had high school or some college education while 65 percent were college graduates or above. This 2:1 ratio differs from the national 1:2 (college/high school) ratio and the 1:1 ratio in the clinical groups. This educational demographic had significant correlations with AP and Lang scale scores. Hence the NC group was randomly reduced from 198 to 159 to achieve 1:1 rater education matching when comparisons were made with the clinical groups.
Normative Study Findings
NC scale scores were not normally distributed and thus not amenable to Standard scoring. However the robust NC subject numbers allowed accurate parametric analyses and there was solid agreement between parametric and non- parametric analyses. Non parametric analyses (e.g. Mann Whitney) only were used for the smaller Clinical groups (less than 25 subjects). Very acceptable psychometric values were noted as follows:
- Internal Reliability
Cronbach’s Alpha was .95 for AP, TAP and Lang scales and .88
- External Reliability
Test-retest Pearson R was .88 (p ≤ .01). (Twenty-six parents re- rated
their questionnaires after a 3 week interval).
- Internal Validity
Factor analysis (with oblim rotation and Kaiser normalization)
identified 3 factors with 53% of the variance. Factor items
matched AP, ATT and Lang scales 39 of 49 times
- External Validity
Very significant differences occurred between normal
controls and clinical groups for all scales. (Manova P ≤ .000).
Significant differences also occurred between ADHD and APD
groups on the AP, TAP and ATT scales (Mann Whitney p ≤.001).
Over 80% of Clinical group subjects were rated at or below the 20th percentile on all scales. Never the less, intra Clinical group scale score pattern differences were significant and suggested clear “pass-fail” cut off values when analyzed using receiver operator curve (ROC) data.
ROC scale score data readily differentiated the 3 clinical groups as indicated below. Percentiles refer to rank percentile scale scores. Sensitivity/specificity levels are noted by % / %. Subtraction is indicated by a – (minus symbol).
APD Group: ATT – AP score ≥ + 2 (90%/83%); AP scale ≤ 5th percentile (88%/80%) and Lang Scale ≥ 3rd percentile (90%/82%).
ADD Group: ATT – AP score ≤ – 10 (85%/80%); ATT scale ≤ 5th percentile (85%/85%) and Lang Scale ≥ 5th percentile (90%/82%).
LD-NS Group: Lang Scale ≤ 3rd percentile (90%/82%); ATT and AP scales both low (≤ 5th to 10th percentile).
Student’s rank percentile scores and suggested referral cut off values are recorded on an Excel spread sheet report form. Rank percentile scoring was chosen since the data had a positive kurtosis and did not follow a standard distribution curve.
Summary and Conclusions
- The APDQ is a differential screening questionnaire for APD with 52 listening skill items to be to be rated as to “frequency of occurrence” by the parents or teachers of 7 to 17 year old students. Three scales are presented which rate competent performance in hearing-auditory processing (AP), attention control (ATT) and cognitive-language skills (LD-NOS). Normal study controls numbered 198 with an additional 45 clinical subjects with ADHD, 20 with APD and 11 with non-specific learning disabilities.
- Multivariate analysis and ROC data strongly differentiated the normal controls from the clinical subjects (Manova P ≤ .000) as well as the 3 clinical groups from each other (Mann Whitney p ≤.001). Clear cut off values are available to guide referral decisions at roughly the 90% sensitivity and 80 % specificity levels.
- The psychometric values of the APDQ are very acceptable including measures of Internal and External Reliability as well as Internal and External Validity.
- Over 80% of clinical subjects were rated below the 20th rank percentile across all scales. It was as if they shared a common “listening disability” or underlying cognitive deficit. Therefore, “at risk” levels for scale scores were conservatively set at the 20th rank percentile with ROC data available to further guide clinical referrals.
- The APDQ is readily scored and scale score values reported by an Excel spreadsheet program (see Excel report form below).
- A high APD risk is suggested by an ATT minus AP score greater than +2 and an AP scale score at or below the 5th percentile, with a Lang scale score greater than 3rd percentile.
- A high ADHD risk is suggested by an ATT minus AP score less than -10, and an ATT scale score equal to or less than the 10th percentile (younger group) or 5th percentile (older group), with a Lang scale score greater than 3rd percentile.
- A strong AD-NOS or nonspecific learning disorder without APD is suggested by a Lang scale score equal to or less than the 3rd percentile and ATT and AP scale score both low (less than 10th percentile).
- The APDQ is available by emailing firstname.lastname@example.org. It will be published on line and in hard copy shortly. Copyright is 2013.
Dr. Brian O ‘Hara is a Developmental-Behavioral Pediatrician from Honolulu Hawaii. Shortly after his retirement in 2004 he followed his interest in learning issues to have a “mini fellowship” with Dr. Frank Musiek at Dartmouth University, becoming cross-trained in neuroaudiology and Auditory Processing Disorders. He started work on a multi-disciplinary screening questionnaire for APD which today is known as the Auditory Processing Domains Questionnaire or APDQ. A normative study for the questionnaire was done in 2007 at the Kaiser Center for Health Care Research in Honolulu which was affiliated with his home medical center for 30 years.
Dr. O’Hara graduated from Stanford University and the University of Washington Medical School where he also did training in general and developmental behavioral pediatrics. He is an avid grandpa, photographer and Hawaiian culture traveler and is very happy to continue to work with his colleagues in audiology.