IADR Abstract Archives
Retrospective Study of Deriving Periodontal Disease Diagnosis from Periodontal Findings.
Objectives: Patients’ periodontal health information stored in electronic dental records could be utilized to monitor periodontal disease (PD) changes in response to treatment. However, clinician-recorded PD diagnosis may not be available for all patients. The objective of this study was to determine the feasibility of using periodontal findings (clinical attachment loss, periodontal pocket depth, bleeding on probing) used in case-definitions for surveillance studies (American Academy of Periodontology (AAP)) to generate PD diagnosis. We determined concordance between diagnosis generated from findings, and clinician-recorded diagnosis.
Methods: We conducted a retrospective study of 28,908 patients who were 18-years and older and received comprehensive oral evaluation between January 1st, 2009-December 31st, 2014 at the Indiana University School of Dentistry Clinics. We developed a computer program to obtain PD diagnosis from periodontal findings and a natural language processing approach to extract clinician-recorded diagnosis from clinical notes. Last, we calculated percent agreement to determine the concordance between diagnosis from periodontal findings and clinician-recorded diagnosis. Disease severity was excluded in this comparison.
Results: Periodontal charting findings were available for 22,880 (80%) patients, clinician-recorded diagnosis for 13,219 (46%) patients, and both information for 10,406 (36%) patients. Among diagnoses obtained from findings, 37%* had gingivitis, 55%* of patients had moderate periodontitis, and 28%* had severe periodontitis. From the clinician-recorded diagnosis, we found 50%* had gingivitis, 18%* had mild, 14%* had moderate, and 4%* had severe periodontitis. The concordance between findings-derived diagnosis and clinician’s diagnosis for 10,406 patients was 47%.
*percentages do not add up to 100% because some patients had both gingivitis and periodontitis.
Conclusions: The results indicate underdiagnosis of gingivitis and overdiagnosis of periodontitis when using AAP classification (case-definitions) for diagnosis. We hypothesize that AAP classification alone may not be sufficient to generate PD diagnosis because clinicians may consider soft-tissue and radiographic findings also when diagnosing PD.
Methods: We conducted a retrospective study of 28,908 patients who were 18-years and older and received comprehensive oral evaluation between January 1st, 2009-December 31st, 2014 at the Indiana University School of Dentistry Clinics. We developed a computer program to obtain PD diagnosis from periodontal findings and a natural language processing approach to extract clinician-recorded diagnosis from clinical notes. Last, we calculated percent agreement to determine the concordance between diagnosis from periodontal findings and clinician-recorded diagnosis. Disease severity was excluded in this comparison.
Results: Periodontal charting findings were available for 22,880 (80%) patients, clinician-recorded diagnosis for 13,219 (46%) patients, and both information for 10,406 (36%) patients. Among diagnoses obtained from findings, 37%* had gingivitis, 55%* of patients had moderate periodontitis, and 28%* had severe periodontitis. From the clinician-recorded diagnosis, we found 50%* had gingivitis, 18%* had mild, 14%* had moderate, and 4%* had severe periodontitis. The concordance between findings-derived diagnosis and clinician’s diagnosis for 10,406 patients was 47%.
*percentages do not add up to 100% because some patients had both gingivitis and periodontitis.
Conclusions: The results indicate underdiagnosis of gingivitis and overdiagnosis of periodontitis when using AAP classification (case-definitions) for diagnosis. We hypothesize that AAP classification alone may not be sufficient to generate PD diagnosis because clinicians may consider soft-tissue and radiographic findings also when diagnosing PD.