We screened all patients (approximately 1,200) who underwent audiometry during 16 consecutive days (7/18/07–8/10/07) in the audiology department of a large academic medical center. Patients qualified to be a part of the study if they had a disabling binaural hearing loss that was very likely to exist in the 2 years before audiometry. We excluded patients with mild or moderate conductive losses, including losses related to cerumen impaction. The only patients with conductive losses included in the study were those whose loss was explicitly noted in the audiology report to be of long duration. Any patients who reported a rapidly progressive sensorineural loss during the previous 2 years also were excluded.
We defined disabling hearing loss as binaural impairment sufficient to prevent complete recognition of spoken words in a medical office visit. We used a criterion based on the Speech Intelligibility Index (SII), which relates audiometric findings to the percentage of spoken words that individuals can recognize correctly at any given loudness.4
Importantly, the SII is a prediction of the best possible performance. Most people with hearing loss actually perform more poorly than the calculated value.5
We included patients for whom the SII predicted that their bilateral hearing thresholds would allow less than 90% word recognition at 50 dBHL (moderate speech).
Of all the charts for patients who received audiometry testing from July 18, 2007 to August 10, 2007, 680 were adults who had searchable EMRs, and of these, 254 (37%) met our hearing criterion. Our sample consisted of the first 100 patients (of the 254) who had recent, sufficient, and comprehensive EMR notes to review for documentation of hearing loss. We defined a recent note as one created no earlier than July 2005, within 2 years before the patients’ audiometry. Pilot surveys looking at shorter time frames found relatively few patients with comprehensive visits documented in the EMR, prompting us to expand the time window to 2 years. We defined a sufficient and comprehensive EMR entry as the single most recent outpatient visit that had a physician note containing both the headings “History” (or “Review of Systems”) and “Physical Exam.” Notes did not always specify whether the physician was the patient’s primary care physician, but we excluded notes from visits for specialty care focused on one issue, emergency department visits, and notes from inpatient stays. We also did not review history and physical examination reports recorded by physician assistants, nurses, or other nonphysicians. None of the 100 patients was deaf from birth, and all used oral language plus their residual hearing to communicate. Therefore, none of the survey patients would have benefited from the use of a sign language interpreter.
One author (CH) reviewed all 100 audiologic evaluations and the index EMR note, using an abstraction form to record age and sex, severity of the patients’ loss, whether the patient had a hearing aid or cochlear implant, and the physician’s documentation of hearing status. Any mention of hearing loss, hearing device, any in-office test, referral, or other notation that would imply hearing loss was counted as a documentation of loss. This was true regardless of which section of the report the notation was found. The reviewer assumed that the acronym “HEENT” (head, eyes, ears, neck, and throat) referred only to the appearance of the pinnas, auditory meati, and tympanic membranes. Thus, we interpreted the phrase “HEENT normal” as meaning “hearing loss not mentioned,” rather than “hearing normal.” The common statement “CNI-XII normal” was interpreted as indicating that the sensory function of CNVIII was normal. This abbreviation produced many of the notations of normal hearing that we report. The chart review also separately recorded whether patients had hearing aids at their audiology visit to allow us to perform an analysis of the group with hearing aids versus those without. We did not know whether patients wore their hearing aids during their physician visit unless that was documented in the EMR.
A subsample (n
25) of both the audiograms and EMR notes was reviewed by an independent observer who was blinded to the original chart review results. This blinded reviewer judged whether hearing loss was mentioned, not mentioned, or was noted to be normal by the same criteria described earlier. She also was asked to identify any case that may not have had substantial hearing loss at the time of the physician’s note. Agreement between the independent observer and the original reviewer was 100% for the 25-patient subsample. The independent review of the audiograms also revealed that all 25 cases likely had hearing loss at the time of the physician note.
The study was approved by the Massachusetts Eye and Ear Infirmary Institutional Review Board [# 07-07-050X].