Spotlight on Sleep: Sound Sleep in the Intensive Care Unit: Source–Path–Receiver Model

“A heavy clap of thunder! I awoke from the deep sleep that drugged my mind—startled, the way one is when shaken out of sleep.”

—Dante Alighieri, The Divine Comedy, Canto IV, lines 1–3

Throughout history, people have recognized the importance of sleep, and its fragility in the presence of sound.^a Being awoken abruptly can strain our physiology, reduce our health, and compromise our wellbeing. Yet our brains evolved to maintain some responsiveness to sound, without which sleep would render us vulnerable to threats in the environment. Neurologic structures—the acoustic nerve, as the sensory organ; brainstem, as the initial path; thalamus, as the gatekeeper; and auditory cortex, as the first-level cortical perceptor¹—are positioned as sentinels, with conflicting missions: to guard sleep from noise disruption while alerting the brain to threatening sounds. The vices and virtues of sound-induced sleep disruption are as old as our time on this planet.

In the modern era, we no longer find ourselves concerned merely with severe noises from lightning or other natural phenomena. We have created a myriad of synthetic devices that produce sound. To be sure, some of these sounds are desirable. For instance, an apartment building’s smoke alarm can be lifesaving by awakening sleeping tenants, and signaling them to evacuate their burning building urgently. Other noises are uniformly undesirable, like a lawnmower’s rumbling hum. Others are more ambiguous; they help some and harm others. For example, a driver on an early-morning commute might honk to avoid collision from a merging car. But the sleeping person in the house next to the road enjoyed no benefit from the noise. Competing interests abound.

In the hospital, and particularly in the intensive care unit (ICU), these competing interests are abundant. Alarms play many roles. They serve to alert providers to take action on behalf of their critically ill patients. But what about the sleeping patient whose sleep is disrupted by the alarms of the patient down the ICU ward’s hallway? What about the patients themselves? Unlike our smoke detector example, where the listener takes direct action, here there are multiple listeners: intended (providers) and unintended (patients, including the patient for whom the alarm is of critical value). Only the providers, however, take action. They are the intended listeners.

As a result, hospital alarms offer indirect benefit to patients by alerting providers to patients’ critical needs. The alarms themselves, however, can be a direct hindrance to patient health by stressing their physiology. Noise-induced sleep disruption from sounds commonly encountered in the hospital can lead to temporary elevations in heart rate.² In the ICU, in particular, frequent alarms cause anxiety, stress, and sleep disruption, straining the physiology of an already critically ill patient.

The origin of hospital alarms is undeniably well-intended: to alert providers of a problem, in order to take life-saving action. However, it is surprising that this decades-old, imperfect system continues to be a problem for patients in modern hospitals, given the established relationship of noise leading to sleep disruption, which worsens ICU outcomes.^2,3 It is even more surprising that this problem continues given technological advances that might refine the target of an audible alarm, or substitute a different technology altogether. New and old mitigation strategies are underused.

Alarms are among many sounds in the ICU, where noise levels are high. The human ear is tuned to perceive sounds between roughly 20 Hz and 20,000 Hz, with maximum sensitivity at about 3,000 Hz.⁴ In a study⁵ examining sound levels in nearly 40 ICUs across Australia and New Zealand, on average, peak sound levels were 78 dB (SD 9) and mean sound levels were 62 dB (SD 8). Other ICUs have reported peak sound levels greater than 90 dB each hour.⁶ To capture a sense of the hindrance to sleep in that environment, imagine lying in bed at home to go to sleep. Add people in your bedroom having a loud conversation, occasionally running a vacuum cleaner or two, and occasionally having a freight train pass by at 100 feet away (Figure). Such is the soundscape for a patient in the ICU.

Solutions are available. Rather than providing a one-size-fits-all recipe for success, which would not account for the needs of individual ICUs, and would be beyond the scope of this article, we share a framework of acoustic principles—the source–path–receiver (SPR) model—that can facilitate providers’ communication of specific problems, and foster appropriate solutions among hospital administrators, architectural acousticians, medical directors, nurse managers, and technology companies.

The SPR model provides a framework for sound and noise mitigation, focusing on disciplines of acoustics, neuroscience, and pharmacology, to foster solutions-based discussions for each hospital’s common and unique situations. The SPR model is used by architects, acousticians, and other engineers to map noises from the origin (source) to the final destination (receiver) and how it gets there (path).

Source–Path–Receiver Model

With rare exception, sensory perceptions of any modality can follow this simple model. We focus here on sound. In brief, the SPR model describes 3 fundamental elements.

Source

This is the origin of a sound signal: the physical object and circumstances causing vibrations in the audible spectrum. The presence of a noise source, and its characteristics, will define the type of vibration across 3 main variables: sound pressure level (sometimes just called “sound level” or “level,” and less accurately called loudness), usually described in decibels; frequency composition (often described as either broadband or narrowband, in Hertz); and time signature, usually described as either impulse (like a firearm) or continuous (like a ceiling fan).

Path

Once a sound source occurs, the direction that sound takes, and its success in continued propagation, depends on many factors, mostly friction loss and reflection.

Receiver

The receiver in our ICU scenario is the patient, the provider, or both: listeners, intended or not. The ear, acoustic nerve, brainstem, thalamus, auditory cortex, and higher order cortexes are paramount in this space, as well as the characteristics of the sensory system to perceive the sound. The state of each can influence the perceptibility of sound.

SPR Approach to Noise Mitigation

Essential to this model is an assumption that each of these 3 spaces—source, path, and receiver—offers an opportunity to intervene toward the immediate goal of noise mitigation. We illustrate the approach with some specific examples (Table).

Source

The fundamental question to ask is whether the sound needs to occur. If so, does it need to occur here? Can it be moved or modified at the source? In the “just-unplug-it model” (which includes moving a noise source elsewhere), we question assumptions about whether existing noise sources are necessary, and if so, whether they can be relocated. Can a noise source be turned off, moved elsewhere, or transformed?

In the age of wireless transmission, instead of an audible telemetry alarm, which seems to reach everyone and no one, perhaps a nonaudible alert can be transmitted to designated providers. It seems reasonable to provide low-frequency vibrations on smartwatches, for example, such that vibration is felt by the intended receivers and heard by no one. There are challenges in this area that need to be addressed, but technology improvements (eg, in reliability, security, interference) occur rapidly. Technology companies will be more inclinced to pursue and provide viable solutions if hosptials clearly establish and communicate the need for source accommodations.

Path

The noise path is addressed when there is an existing noise source that cannot be removed or reduced sufficiently at the source. The essential task here is to interrupt the sound path such that the level is substantially reduced before reaching the listener. In other words, something needs to be done to the sound after it occurs but before it reaches the receiver. This process, broadly referred to as sound attenuation, invokes processes such as friction and reflection. The former causes the sound to lose energy through friction loss when encountering a material; the latter causes sound to be reflected back off of a material, such as thick glass (which also offers some friction loss), rather than transmitting through it, causing the sound to be directed elsewhere. Closing gaps in doors, giving barriers appropriate mass, maintaining physical barriers, and adding sound-absorbing materials between source and receiver are helpful. A building with closed windows is a good example. As a crude rule of thumb, buildings reduce noise by about 25 dB before the sound reaches a person inside the building, if the windows are closed.⁷

Receiver

After minimizing sound exposure with source and path considerations, we can turn our attention to the receiver—in this case, the patient. Even in the absence of critical illness, the acoustic environment of an ICU is disruptive to sleep physiology. This was demonstrated elegantly in a study reporting disrupted sleep architecture, decreased melatonin levels, and increased cortisol levels in a cohort of volunteers exposed to a simulated nocturnal ICU soundscape.⁸

Drugs. Pharmacologic interventions are the most commonly used method to promote sleep in critically ill patients, despite a lack of strong evidence or recommendations from clinical practice guidelines.⁹ Part of the concept is to raise the thresholds of perception. Using medication to treat pain and anxiety is paramount to enabling natural sleep, a state that increases threshold of perception. However, cortical processing of acoustic stimuli continues during both sedation and even general anesthesia.¹⁰ Therefore, it is unlikely that pharmacologic interventions alone are sufficient to protect patients from environmental noise. Critically ill patients can be particularly susceptible to medication side effects or cross-reactivity with other drugs, or be unable to clear medications (eg, with liver or renal disease).

Earplugs. These are the most commonly studied receiver-level method of protecting patients from the ICU’s acoustic environment.¹¹ Although a simple intervention, the technical specifications of earplugs and their tolerability vary across studies.¹² Furthermore, earplugs and eye masks are often combined into a single intervention, and measured outcomes are inconsistent. For example, Demoule et al¹³ found that a combined earplug–eye mask intervention increased time spent in N3 sleep (deep sleep); Obanor et al¹⁴ reported that earplugs and eye masks improved subjective sleep quality; and Van Rompaey et al¹⁵ reported decreased incidence of delirium with earplugs alone. Despite heterogeneity in interventions and outcomes, the literature as a whole supports the use of earplugs in the ICU, either in isolation or combined with other sleep hygiene interventions.^9,11,16

Sound Masking. In addition to blocking acoustic stimuli physically, it is also possible to mask ICU sounds, rendering them less perceptible. By analogy, a lit candle in a dark room is easy to see, but once the lights are turned on, the lit candle is less noticeable. The sensory environment in which a stimulus exists profoundly influences its perception. Sound masking in particular relies on the psychoacoustic principle that perception of one sound is affected by the presence of another. Leveraging this principle, constant exposure to carefully selected nonarousing sounds can be used to increase the threshold at which sporadic ICU sounds are perceived by patients. Broadband sounds (eg, white, brown, or pink sounds), ocean sounds, and music have all been attempted to reduce arousals from sleep and increase subjective sleep quality in critically ill patients, some with excellent effect.^17-21 The challenge with masking is there is a sound level at which the mask itself is not tolerated. One cannot increase masking indefinitely without the mask itself becoming annoying.

Noise Canceling. Sound masking is often conflated with noise canceling. In contrast to sound masking—which introduces a continuous, broadband sound (eg, rain, river sounds) to make a noise less noticeable—active noise reduction (sometimes called active noise canceling) is a process in which devices monitor environmental sounds and use adaptive signal processing algorithms to produce canceling antinoise waves such that the original sound is reduced in amplitude.²² Whereas the mechanism of masking is fundamentally neurologic, the mechanism of active noise reduction lies in physics. These technologies have been deployed to the consumer market in the form of noise-reducing headphones, but have yet to be tested in the ICU, or during sleep. One experimental model reported encouraging results, showing that commercially available noise-canceling headphones could effectively reduce noise exposure in a cardiac ICU.²³

Summary

We have all experienced the annoyance of noise-induced sleep disruption, and possibly even the resulting cognitive and physical impairments. Critically ill patients have a great need for sleep, and greater vulnerability, and the noise level in the ICU environment is high. Using the SPR model, we can substantially improve the soundscape of the ICU. None of the interventions described is mutually exclusive; the best outcomes often result from combinations of them. Knowing the problems, organizing a multidisciplinary approach to them, and taking action will lead to an improved workspace for providers and improved outcomes for patients.