Validity of the Autism Dysmorphology Measure (ADM) for Assessment of Generalized Dysmorphology

Background:  

ADM is a statistically valid measure originally designed to allow non-geneticist physicians to assign dysmorphology status of autism patients presenting for diagnosis or participation in research protocols. The ADM assesses 12 physical structures (height, hair growth pattern, ear size, structure and placement, nose size, face size and structure, philtrum, mouth and lips, teeth, hand size, fingers and thumbs structure, nails and feet size and structure); each structure is labeled dysmorphic or non-dysmorphic based on comparisons with non-affected parents. Using the 12 results the ADM algorithm leads the physician to a decision of dysmorphic or non-dysmorphic. This report provides information on the success of various learning groups using the ADM.

Objectives:  

Determine whether non-geneticist physicians can accurately assign dysmorphology status of autism patients and what training and testing modalities are most effective.

Methods:  

More than 100 individuals received training on the use of the ADM, including receiving the Training Manual, ADM Training PowerPoint, in person instruction, practice scoring sessions and 10 test cases.  Learners seeking validation of their training were asked to submit results of one or more case studies. Results from 16 individuals (11 physicians, 5 non-physicians (RNs, psychologists) who submitted a total of 34 ADM cases are presented here.  Each ADM case was graded on 51 responses; ability to convert growth measures to centiles using standard growth charts, ability to judge the dysmorphology status of 12 physical structures and conclude whether each structure fell within or outside the range of normal, based on comparison with parents, and use of the ADM scoring algorithm to formulate the final dysmorphology decision.

Results:  

Seventy-five percent of test takers obtained a passing score of 85%.  The primary predictor of success was physician status. Physicians passed 81% of cases, non-physicians only 25%. The average score for physicians was 88% and 74% for non-physicians. Analysis by question type revealed learners plotted growth curves with 95% accuracy (physicians 96%; non-physicians 90%).  Physicians were 84% accurate assigning dysmorphology status to the 12 structures; non-physicians 76%. Assignment of final dysmorphology status was 88% correct for physicians and 61% for non-physicians. Each of the 12 structures assessed by the ADM measure was analyzed individually to identify areas of difficulty. Though nose size, philtrum, and face size/structure were correctly classified less often, overall the 12 structures presented similar challenges, based on low SDs (84% ± 7% for MDs; 79% ± 17% for non-MDs). Physician geneticists scored only minimally higher than geneticists and both groups assigned dysmorphology status with 88%.

Conclusions:  

With minimal training on the ADM physicians can correctly assign dysmorphology status 88% of the time. Extending training time to 2+ hours and providing 3 dimensional photographs raised accuracy to 95%. Non-pediatricians who don’t routinely plot growth curves require extra training in this area.  Comparisons of results indicate training targeted specifically for MDs versus non-MDs could improve results especially for the non-physicians. Though 3 dimensional face photographs were available for most cases, the inherent difficulty of assessing dysmorphology via photographs is considered the main limitation.