Noise in the intensive care unit and its influence on sleep quality: a multicenter observational study in Dutch intensive care units

Recently, hospital noise and its potential negative influence on patient outcome has gained widespread attention among caregivers [ 1, 2]. Noise is generally expressed as sound pressure in decibels (dB), whereby often a correction is made for the frequency of the sound (called “A-weighting”) to account for the relative loudness of the sound as perceived by the human ear. Noise levels in ICUs have been found to be beyond acceptable levels with average daytime sound pressure levels of around 60 A-weighted decibels (dBA) and peak levels > 90 dBA, the equivalent of standing next to a highway [ 3, 4]. Even more relevant, nighttime sound pressure levels are only slightly lower with averages of around 50 dBA. These sound pressure levels clearly exceed those of 35 dBA recommended by the World Health Organization (WHO) for nighttime in hospitals [ 5]. In the ICU, different factors contribute to high sound pressure levels, including a large number of alarm-generating monitoring equipment, use of mechanical ventilators and around-the-clock activities by staff members [ 4, 6, 7]. Excessive noise may cause multiple auditory and non-auditory effects, of which sleep disturbances are thought to be the most deleterious [ 8]. Sleep disturbances occur frequently in the ICU, characterized by an increase in stage 2 sleep and a decrease in stage 3 and rapid eye movement (REM) sleep [ 9, 10]. Up to now, only a few studies have studied the potential relationship between excessive noise and disturbed sleep in a real ICU setting. Studies using polysomnography in ICU patients demonstrated that between 11 and 24% of arousals are caused by environmental noise [ 11, 12]. Although polysomnography is considered the gold standard for evaluating sleep, it is labor intensive and a burden for ICU patients. Moreover, it is notoriously difficult to interpret and may not adequately reflect subjective sleep [ 13, 14]. Furthermore, studies have been performed in a small number of patients and in single centers, thus limiting the generalizability of the findings. Also, the acoustic parameters that have been analyzed are conventional measures of sound level, indicative of physical changes in the sound field. To further quantify the effects of noise on humans, advanced parameters, such as loudness and restorative periods, defined as periods of relative quietness, may be more useful. We therefore set up a prospective, multicenter observational study aimed to determine the association between various acoustic parameters and subjective perceived sleep quality in ICU patients.

A total of 71 patients fulfilled the criteria between April 2013 and August 2015 and were included in this study. Data on seven patients were removed from the final analysis due to missing audio data. Baseline characteristics of the remaining 64 patients can be found in Table  1. On average, patients were 63.9 ± 11.7 years old and 48 (68%) patients were male. Most participating ICUs had single-bed rooms and the daily visiting routine occurred at similar times (see Table  2).

In this prospective multicenter, observational study, we showed that background noise was negatively associated with sleep, while gender (female) and the number of restorative periods were positively correlated with sleep quality. Patients’ perceived sleep quality was poor and did not differ between participating hospitals. Overall, noise levels were consistently above the values (< 35 dB L _Aeq day and night; < 40 dB L _AFmax night) recommended by the WHO [ 5]. This is the first study to evaluate a link between subjective sleep quality and objective noise parameters in ICUs.

Sleep disturbances are very common in ICU patients [ 13]. As a risk factor for sleep disturbances [ 13], noise in ICUs is ubiquitous and is mainly caused by staff activity, machines and alarms [ 4].

Research on the relationship between noise and sleep arousals in ICU patients, thereby using polysomnography, shows that noise peaks > 80 dBA are associated with arousal from sleep and that noise is responsible for 11 to 24% of the total number of arousals [ 11, 12, 24]. More subjectively, patients themselves consider noise in the ICU to be disturbing to sleep [ 25].

We found that a higher L ₉₀ led to a moderate decrease in sleep quality. L ₉₀ is considered to be indicative of the background noise level, generated by, for example, air conditioning or computer ventilators [ 26]. The average nighttime value of L ₉₀ was 38.1 dBA, indicating that levels of background noise even exceed the threshold for average sound pressure levels stated by the WHO. Because the participating hospitals were different in design and layout, considerable differences were found in L ₉₀ between hospitals. Given that within-hospital L ₉₀ was not different between day and night, the observed differences cannot be attributed to differences in procedures and staff movement but must be due to the building characteristics. This finding underlines the importance of taking building properties into account when designing a new ICU. Average day and nighttime noise levels were comparable with others studies, whereby differences between day and night were only marginal [ 3, 4].

This is the first study showing a positive association between restorative periods and better sleep. Restorative periods occurred on average only 2.4 times per hour during the night, and the average duration of a restorative period was 14.1 min, which is indicative of the high number of peak noises. Since restorative periods are most frequently ended by high-level noisy events due to staff activity or speech [ 4], interventions aimed at reducing staff-generated noise appears to be a reasonable and achievable goal to improve sleep quality in critically ill patients. Additionally, since noise coming from alarms and monitors were found to be the most disturbing noise sources in our qualitative analysis, nighttime alarm modification may also be of additional value in improving sleep quality in critically ill patients.

Female patients expressed that they had better sleep compared to the male patients. Gender differences in subjective sleep quality in ICU patients have not been reported previously in the literature, however, studies in the general population generally indicate worse subjective sleep quality in women, compared to men [ 27]. Interestingly, in a large study on the effects of traffic noise on objective sleep and subjective sleep quality in healthy people, there were larger effects of noise on objective sleep parameters in men compared to women, whereas no clear differences were found in subjective sleep quality [ 28]. Future research on gender differences in sleep quality should further elucidate this finding.

In addition to pure physical properties of sound, we also analyzed the impact of noise peaks on human perception (also called loudness) by using a validated model [ 17]. Loudness peaks with a magnitude of at least 10 sone occurred significantly less during the night than during the day. It is noteworthy that the nighttime hourly rate of these peaks differed substantially between the participating hospitals, varying from 2.5 to 8.8. Without a source-specific analysis, it is difficult to pinpoint the source of these loud peaks and what causes this difference between participating hospitals.

Some limitations need to be addressed. First, we used the RCSQ as a measure of sleep quality instead of polysomnography. Although RCSQ is validated for the use with ICU patients, subjective evaluation of sleep quality is subject to different forms of bias, including recollection and response bias, which may specifically be true for ICU patients who are recovering from critical illness. However, the results of the sleep evaluations are in line with previous studies and the repetitive design of up to five measurements helps in reducing the impact of outliers. Moreover, the RCSQ is easily applicable and interpretable and is therefore easier to use in daily practice in contrast to polysomnography. Second, we included a relatively small group of ICU patients who were awake and able to communicate, thus selecting only a subgroup of less severely ill patients, which makes it difficult to generalize the findings of this study to the whole ICU population. Although sleep evaluation in sedated and/or delirious patients remains difficult, interventions aimed at improving sleep quality by addressing noise may also have beneficial effects in this other patient group. Third, we did not take the long-term outcomes of the patients into account. Therefore, negative effects on sleep may not necessarily lead to worse outcomes. However, other studies have clearly demonstrated an association between sleep deprivation and the development of delirium in ICU patients, which has a multitude of negative long-term consequences [ 29]. Moreover, even brief periods of sleep deprivation in the general population can have long-term negative consequences in immune and cognitive function and in hypertension and obesity [ 30].

Associations between various noise parameters and subjective sleep quality were found in this multicenter study, confirming the negative consequences of noise on the sleep quality in ICU patients and thereby strengthening the usefulness of noise-reducing strategies. Sleep quality in general was poor and did not differ between participating ICUs. Noise levels were high and periods of relative quietness occurred only rarely. Increasing the number of nighttime restorative periods appears to be a reasonable goal for improving patients’ sleep.