Effectiveness of a noise reduction intervention in the intensive care unit: A prospective bicenter study

Abstract

BACKGROUND

Excessive noise in healthcare environments—commonly described as "unwanted sound"—has been linked to a range of negative impacts on both patients and staff. In clinical settings, elevated noise levels have been associated with sleep disruption, heightened cardiovascular stress, and an increased risk of delirium in patients. Among healthcare workers, noise can impair focus and cognitive performance, potentially compromising care quality.

AIM

To evaluate the effectiveness of educational and behavioural interventions in reducing noise levels within intensive care units (ICUs), recognizing their potential impact on patient outcomes and healthcare effectiveness.

METHODS

A prospective interventional study in two Singaporean teaching hospitals compared peak and average sound levels between control and intervention groups. An educational and behavioural intervention comprising talks, posters, and self-audits by nurse champions was initiated in two ICUs in one hospital on November 18, 2023. Sound measurements were collected at 4 Locations within each ICU before and after intervention. Baseline measurements were taken from October 22, 2023 to October 29, 2023, and post-intervention measurements from December 21, 2023 to December 22, 2023. The hospitals served as the primary exposure variable, controlled for ICU type (medical vs surgical) and hour of the day.

RESULTS

Our analysis generated 48 pairs of peak and average sound level readings for each unit (control n = 48 readings; intervention n = 48 readings). The effect of the intervention was associated with a significant 4.8 dB decrease in average sound level (P = 0.009) and a nonsignificant 4.3 dB decrease in peak sound level (P = 0.104), adjusted for hour of day and type of ICU.

CONCLUSION

Educational and behavioural interventions successfully reduced average sound levels, emphasizing their positive impact on noise control. These findings contribute valuable insights for optimizing noise reduction efforts in critical care settings. Future studies may explore additional systemic and environmental interventions to enhance noise management strategies.

Key Words: Noise reduction; Intensive care units; Educational interventions; Behavioural interventions; Healthcare effectiveness; Patient outcomes; Prospective interventional study; Sound levels

Core Tip: Educational and behavioural interventions demonstrated success in reducing average sound levels, emphasizing their positive impact on noise control. These findings contribute valuable insights for optimizing noise reduction efforts in critical care settings.

INTRODUCTION

Noise, defined as any sound that may produce an undesired physiological or psychological effect in an individual or group[1] or simply as “unwanted sound”[2], has demonstrated detrimental effects on patient outcomes and healthcare effectiveness. Undesired consequences include increased cardiovascular risk[3], sleep disturbances[4], associations with delirium[5,6] in patients, as well as impedance in concentration and cognitive function[1,7] among healthcare workers.

Critically ill patients, particularly sensitive to noise-induced sleep disruption[1,8], often experience sound levels exceeding acceptable thresholds. Daytime sound pressures average around 60 A-weighted decibels (dBA), while nighttime levels reach 50 dBA[9,10], with peak levels above 90 dBA[9-11]. These significantly surpass the World Health Organization-recommended LAeq 30 for hospitals, leading to sleep disturbances during nighttime in hospital settings[2].

Prior studies have attempted to mitigate noise levels in the intensive care units (ICUs) through various interventions, demonstrating some success, albeit with varying effectiveness[12-15]. However, further research is necessary to expand upon these efforts due to the complexity of the ICU environment. Factors such as patient acuity, medical equipment, and staff activities contribute to noise levels, making interventions that are effective in one ICU setting potentially ineffective in another. Our study seeks to evaluate the effectiveness of educational and behavioural interventions in reducing noise levels in the ICU. We hypothesize that educational and behavioural interventions will result in a reduction in noise levels in the ICU.

MATERIALS AND METHODS

Study design

This was a prospective interventional study conducted in the ICUs of two major teaching hospitals in Singapore: National University Hospital (NUH) and Tan Tock Seng Hospital (TTSH). The requirement for ethics approval was waived by the National Healthcare Group Domain Specific Review Board (reference number 2023/00558).

Pre- and post-intervention sound measurements were taken at selected common areas (four per ICU) within the NUH Medical ICU (MICU), TTSH MICU, NUH Surgical ICU (SICU), and TTSH SICU. Baseline sound measurements were collected between October 22, 2023 and October 29, 2023 (two Sundays) and post-intervention sound measurements were taken on December 21, 2023 and December 22, 2023.

Interventions

Educational sessions to address the adverse impact of noise were conducted in TTSH for MICU and SICU staff over 2 days in November 2023 These sessions included a 3-minute virtual presentation outlining the adverse effects of noise (https://www.youtube.com/watch?v=RhOvN7yoHAY, Figure 1), followed by a 5-minute question-and-answer session focusing on intervention logistics. Staff engagement was further encouraged through email access to presentation slides and posters, as well as a self-funded prize incentive of 100 USD worth of food items for ICUs showing significant noise reduction. Intervention reinforcement was achieved with prominently displayed posters in the ICUs and systematic self-audits of noise levels during each shift, overseen by a designated nurse champion.

Figure 1 Example slide of virtual presentation. ICU: Intensive care unit.

Method of sound measurements

The investigator measured noise levels at the selected ICU locations-for at least 10 seconds–at three time points: 0715 hours, 1415 hours, and 2015 hours. Doctor's ward rounds typically occurred from 0800 hours to 1100 hours, 1500 hours to 1600 hours and nursing shift changes took place from 0700 hours to 0800 hours, 1300 to 1400 hours, 2000 to 2100 hours. Four pre-selected locations were chosen to represent noise levels within the common areas of each ICU. The measurements were conducted sequentially at each location without delay. Sound levels were recorded in decibels (dB) using an A-weighted scale [dB(A)] with the Decibel X mobile device application by SkyPaw Co. The application provided a measurement frequency of 5 measurements per second, and ensured a minimum of 50 data points per 10-second measurement. The same investigator collected all noise measurements, limiting the inter-person variability in data collection.

Statistical analysis

The study's primary outcomes included changes in peak and average sound levels, before and after intervention. The main exposure variable was the comparison between the two hospitals. A difference-in-differences linear regression analysis was done using average or peak sound level as the dependent variable, phase (pre-intervention vs post-intervention) and hospital (NUH vs TTSH) as indicator variables, and (phase x hospital) as the interaction term, adjusted for hour of day and type of ICU (MICU/SICU). Two pairs of data were analyzed: Control groups (NUH MICU and SICU) where no interventions occurred, and interventional groups (TTSH MICU and SICU) where interventions were implemented. All p-values and confidence limits were calculated as two-sided. Statistical significance was taken as P < 0.05. The statistical analyses were done using Stata SE version 13.1.

RESULTS

Our analysis generated 48 pairs of peak and average sound level readings for each unit (Control n = 48 readings; Intervention n = 48 readings). The effect of intervention was associated with a significant 4.8 dB decrease in average sound level (P = 0.009) (Table 1, Figure 2), and a nonsignificant 4.3 dB decrease in peak sound level (P = 0.104) (Table 2, Figure 3), adjusted for hour of day and type of ICU.

Figure 2 Box plots for average sound level.

Figure 3 Box plots for peak sound level.

Table 1 Table for average sound level.

Factors	Coefficient (dB)	Standard error (dB)	95%CI (dB)	P value
Phase1	3.10	1.18	(0.76-5.44)	0.010
TTSH vs NUH	5.73	1.39	(2.96-8.51)	0.000
Phase#TTSH2	-4.78	1.79	(-8.35 to -1.22)	0.009
Hour of day (14th hour)	-1.79	1.08	(-3.94-0.36)	0.102
Hour of day (20th hour)	-2.38	1.08	(-4.53 to -0.23)	0.031
SICU	-0.55	1.03	(-2.59-1.49)	0.595

¹Pre-intervention vs post-intervention.

²Interaction between the phase of the study (pre-intervention vs post-intervention) and the hospital variable (Tan Tock Seng Hospital vs National University Hospital).

TTSH: Tan Tock Seng Hospital; NUH: National University Hospital; SICU: Surgical intensive care unit.

Table 2 Table for peak sound level.

Factors	Coefficient (dB)	Standard error (dB)	95%CI (dB)	P value
Phase1	-0.62	1.71	(-4.02-2.78)	0.718
TTSH vs NUH	4.61	2.02	(0.59-8.63)	0.025
Phase#TTSH2	-4.28	2.60	(-9.45-0.90)	0.104
Hour of day (14th hour)	-0.64	1.57	(-3.76-2.48)	0.684
Hour of day (20th hour)	-1.44	1.57	(-4.56-1.68)	0.363
SICU	2.09	1.49	(-0.87-5.04)	0.164

¹Pre-intervention vs post-intervention.

²Interaction between the phase of the study (pre-intervention vs post-intervention) and the hospital variable (Tan Tock Seng Hospital vs National University Hospital).

TTSH: Tan Tock Seng Hospital; NUH: National University Hospital; SICU: Surgical intensive care unit.

DISCUSSION

Given the well-documented impact of noise on patient outcomes and healthcare worker well-being[4-9,12,16,17], there is a pressing need to optimize noise management strategies in ICU settings. This interventional study assessed baseline noise levels and implemented educational and behavioural interventions to mitigate noise within the ICU.

The substantial reduction of 4.8 dB in average sound levels post-intervention, surpassing the threshold of 3 dBA as outlined by Darbyshire et al[17] for clinical relevance, indicates the effectiveness of our approach. Referring to previous studies on education and training in general ICUs, three studies demonstrated a significant noise decrease of 0.9 dBA[13] to 9 dBA[12,15]. However, these studies were conducted in a general ICU without controlling for ICU type. Another study in a general ICU, focusing on training and staff awareness with noise warning devices, found no noise reduction[14]. Discrepancies between studies may arise from differences in ICU layout, patient case mix, and the specific interventions implemented. Future research should consider real-time monitoring of patient acuity and workload metrics to better account for these external influences.

The non-significant change in peak noise levels (P = 0.104) aligns with prior ICU studies, which found that while structural or behavioral modifications reduce background noise, transient peaks from alarms, staff communication, or patient distress remain difficult to control[11,18]. These events are often sudden, brief, and unpredictable, contributing to greater variability and requiring larger sample sizes for statistical significance. Future studies should explore targeted interventions, such as alarm management protocols or sound-absorbing materials, and utilize larger datasets or alternative statistical methods to improve sensitivity in detecting changes.

Education and noise warning devices have shown effectiveness as single-intervention strategies for short-term noise reduction[16,19-24]. However, since most disruptive noises in the ICU stem from staff and visitor behaviour, structural changes in behaviour are needed. Complex interventions with multiple interacting components have been proposed to gradually lower sound levels. Evidence from implementation science suggests that multicomponent strategies are more effective in changing professional behaviour in the long term[17,25]. Therefore, initiatives such as appointing a nurse champion, as demonstrated in our study, could offer sustained benefits by maintaining regular noise level surveillance and reminders to manage noise for future staff.

Our study's strengths include its prospective interventional design, rigorous sound measurement, and robust difference-in-differences analysis. However, limitations exist, and future research should address confounders, enhance patient stratification, and improve generalizability with larger, more diverse samples.

Environmental factors such as patient acuity, equipment use, and staff workload likely contributed to noise variability. Higher acuity necessitates more bedside interventions and alarms[26], while staff workload fluctuations impact noise through increased communication and movement[10]. Despite standardized measurement times and locations, residual confounding remains. Future studies should incorporate real-time workload metrics and patient severity indices to better isolate intervention effects.

Expanding data collection—such as 24-hour surveillance, intra-room sound levels, patient-provider interactions, and patient comfort—can provide deeper insights into ICU noise perception. Controlling for confounders like census, comorbidities, and staff variability is essential for reliability.

Conducted in two major Singaporean teaching hospitals, our findings may not be fully generalizable. ICU noise is shaped by institutional policies, staff practices, patient demographics, and ICU layouts, which vary across settings. While our structured approach provides a framework for future studies, further research in diverse ICUs is needed to confirm its broader applicability.

Behavioural and educational interventions, while effective in instigating change, are susceptible to the fadeout phenomenon[27], and represent only a subset of available methods[28]. Recognizing this, we recommend future studies explore diverse intervention methods and their cost, effectiveness, and duration. To assess noise level reduction efficacy comprehensively, future studies should enable a direct correlation with patient outcomes such as mood, length of stay and sleep quality. Integrating simultaneous data collection on parameters influenced by noise would offer more applicable insights into healthcare performance. Future interventions should be effective, practical, and cost-effective.

CONCLUSION

In summary, our educational and behavioral interventions significantly reduced average sound levels in the ICU. Moving forward, tailored strategies in diverse ICU environments and assessment of long-term effects on patients will be required to pave the way for more targeted and effective noise reduction measures in critical care settings.