Hospitals are noisy, and noise makes it more difficult to distinguish speech; as we get older, the difficulty increases. Reading, attention, problem solving and memorization are among the cognitive effects most strongly affected by noise, even when the competing noise is not particularly loud. These effects have been studied in health care providers but not in hospitalized patients. This is concerning, as patients often receive extensive education during hospitalization.
The objectives were: 1) to examine the extent to which noise typical of nursing units reduces speech intelligibility and impairs recall in acutely ill hospitalized patients; 2) to quantify severity of reduced performance associated with age, familiarity with health care setting, hearing and health status; and 3) to measure the sensitivity of a 3-question survey in predicting need of a comprehensive audiologic evaluation.
Adult inpatients on medical/surgical nursing units at the Portland VA Medical Center were eligible to participate if their clinician gave permission. Patients who were physically or cognitively unable to participate were excluded, as were patients with documented aggressive behavior, patients undergoing detoxification, patients who were not native English speakers, and those with auditory disorders such as Meniere's disease. Individuals with normal hearing and mild to moderate hearing loss were included. Age, hearing status, number of hospital admissions, and number of medications prescribed on the day of testing were recorded for each participant. Following informed consent, subjects were asked whether they were experiencing any communication problems or difficulty hearing. This question, along with two follow-up questions, was asked to determine whether simply asking would be a useful method to learn whether a Veteran had hearing loss. Subjects then had a qualifying audiology exam and were tested in a sound proof booth.
Subjects were tested in blocks of 3 sets of 5 sentences. Participants listened to recordings of a male talker speaking at conversational level -- 60.5 decibels (dBA) -- using speech derived from the Speech-Perception-in-Noise-Revised test (SPIN-R), which incorporates high context "Stir your coffee with a SPOON" and low context "He was thinking about the RISK" sentence types. In every condition, the 15 sentences were comprised of combinations of high and low context key words.
Participants were asked to identify the last (key) word in each sentence and to remember them. After five sentences, participants were asked to recall the key words from the previous 5 sentences. The background noise normally used for the SPIN-R test was replaced first with silence, then white noise presented at low (64 dBA), medium (69 dBA), or high (74 dBA) levels, and finally with actual hospital noise recorded at the Medical Center, either with or without interfering speech presented at 3 noise levels (59, 64, and 69 dBA). All subjects performed the tasks in silence so we could determine a subject's best performance with which to compare performance in the other conditions, and in white noise so we could determine that performance was equivalent across groups. Half of the subjects were assigned to listen in hospital noise and half in hospital noise with interfering speech. Final testing in quiet allowed us to determine whether subjects were tired by the testing or had 'learned' the test.
Mild to moderate hearing impairment was a factor in study participation for the 82 Veteran inpatients who were recruited. Of those enrolled, thirteen subjects were excluded -- eight were found to be not fully eligible because of audiology history or exam results, and five could not complete testing because they didn't feel well or were called back to the hospital ward. Nineteen of the remaining 69 subjects responded 'Yes' to the question "Are you experiencing any communication problems or difficulty hearing?" Twenty-nine subjects were found to have mild to moderate hearing loss during the audiometric exam. Of those 29, only 13 had answered 'Yes'. In other words fewer than half of those subjects with measured hearing loss were actually identified with the screening question. Our findings were not consistent with previous research; this may be because our sample size was too small to be representative. Seventeen subjects who fulfilled study eligibility were excluded because they were unable to identify a minimum of 80% of the key words in the SPIN-R test in the quiet condition.
Study subjects were, on average, younger (55 versus 65 years) and had fewer prior hospitalizations than the average medical/surgical patients hospitalized during the study period. Those who had hearing loss were on average older, had more hospital admissions, and more medications prescribed on the day of testing than subjects with normal hearing, although none of those differences reached statistical significance.
There were no statistically significant differences in key word identification in the quiet condition between the subjects who listened to the hospital noise and those who listened to the hospital noise with voices. Average key word identification across groups in the quiet condition was 92.5% for the initial test and 93.7% at the final test. Recall scores were also statistically similar between initial and final tests.
Performance between groups was similar in the three white noise conditions. As anticipated, the effects of noise level on speech recognition were quite substantial (p<.001), but there were no statistically significant differences between the two groups. Average speech recognition across groups was 57% correct for the lowest noise level, 35% for the middle level, and 17% for the highest level. Recall of the words reported (regardless of whether or not they were correctly identified) was 54% in low noise, 44% in middle, and 28% at the high noise level.
When the same listeners were tested in the hospital noise, there were significant differences between the group that was tested in hospital noise alone compared to those tested in hospital noise with voices. The interaction of noise level and noise type was statistically significant on key word identification (p<.001), but not for recall (p =.625). The difference in key word identification between groups at the low level was not significant, but the differences were significant at the medium (p <.001) and high level (p<.001).
Significant impacts of noise level on recall were also obtained (p<.001), as well as significant differences between the groups (p<.01). For the low noise level, the presence of voices reduced key word identification from 85% to 79%. At the high level, those listening to noise without voices understood 36% of what was said, while those listening to noise with voices understood only 19%. Similarly large differences between the two groups were found for the recall of key words identified only minutes earlier. At the high level, those listening to noise without voices recalled 42% of the key words they had just identified, while those listening to noise with voices recalled only 30%.
Age, number of medications prescribed on testing day, and number of hospitalizations at the Medical Center were not predictors of poor performance.
This study demonstrated that medical/surgical inpatients can participate in meaningful clinical research, and that the Veterans were grateful for the opportunity. This study demonstrates that hospitalized patients may not understand and remember what they have heard in conversation taking place on hospital wards, and that unrelated speech in the background compounds their difficulty. Although word identification and recall are better at decibel levels lower than those currently measured on our hospital wards, it is obvious the acoustic environment constitutes a barrier to effective communication. Those enrolled as subjects were generally younger and healthier than the hospitalized population as a whole, so we believe that our results have, if anything, underestimated the size of the problem. The unexpected finding that almost 25% of seemingly eligible study population could not identify at least 80% of the key words in quiet bears further investigation.
Beginning in January 1, 2012 the Joint Commission on Accreditation of Healthcare Organizations will require hospitals to have documented patient-centered communication standards. Currently, the guidelines do not include requirements relevant to the acoustic environments where communication takes place. We believe that our study provides evidence that the acoustic environment where patient/provider communication takes place should be addressed.
External Links for this Project
None at this time.