F1000Research F1000Research 2046-1402 F1000 Research Limited London, UK 10.12688/f1000research.167974.2 Research Article Articles The Effects of Hospital Noise and Noise Sensitivity on Patient`s Comfort, Annoyance, and Intention to Leave

[version 2; peer review: 2 approved, 2 approved with reservations]

 Abbasi Milad Conceptualization Formal Analysis Investigation Methodology Supervision Writing – Review & Editing https://orcid.org/0000-0002-0787-1533 1 Sharifpour Mahdi Data Curation Writing – Original Draft Preparation https://orcid.org/0009-0006-7781-8312 1 Mohammadi Mahtab Data Curation Investigation Writing – Original Draft Preparation 1 Manoochehri Yasin Investigation Methodology Writing – Original Draft Preparation https://orcid.org/0009-0009-6234-0143 2 Eskandari Tahereh Formal Analysis Software Supervision Writing – Review & Editing a 3 Saveh university of medical sciences, Saveh, 9898989898, Iran Islamic Azad University Sanandaj Branch, Sanandaj, Kurdistan Province, Iran Department of Occupational Health Engineering, Iran university of medical sciences, Tehran, 01010101010, Iran T.eskandari304@gmail.Com

No competing interests were disclosed.

 28 2 2026 2025 14 1250 22 2 2026 Copyright: © 2026 Abbasi M et al. 2026 This is an open access article distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Introduction

Hospitals are intended to serve as healing environments; however, they are frequently characterized by high levels of environmental noise pollution that can contradict their therapeutic purpose. This cross-sectional study aimed to investigate the complex relationships among hospital noise pollution, individual noise sensitivity, and acoustic comfort, noise annoyance, and intention to leave the hospital.

 Methods

This descriptive-analytical cross-sectional study was conducted in 2024 at a public hospital in Saveh, Iran. A stratified random sampling method with proportional allocation was used to select a sample of 226 hospitalized adult patients. Objective day-evening-night noise levels (L den) were measured over 24 hours, while subjective data on noise sensitivity, acoustic comfort, noise annoyance, and intention to leave the hospital were collected using standardized questionnaires. Bayesian Network (BN) modeling, a probabilistic graphical approach for examining complex dependencies, was applied in combination with delta-p sensitivity analysis to quantify the direct and joint effects of noise exposure and noise sensitivity on patient outcomes. Continuous variables were categorized using percentile cut-offs: low (<P25), moderate (P25–P75), and high (>P75).

 Results

The mean L den in the studied hospital was found to be 57.95 dB (±6.61). The Bayesian Network analysis revealed that under conditions of high level L den, the probability of high annoyance, low acoustic comfort and high intention to leave increased by 12.4%, 6.3% and 5%, respectively. Under conditions of high-Level Sensitivity, the probability of these variables increased by 9.1%, 6.2% and 4.7%, respectively. While these two variables are at high level, the most substantial positive variations occurred in high annoyance, low acoustic comfort and high intention to leave, with increases of 26.1%, 13.1% and 10.6%.

 Conclusion

Noise levels in the hospital exceed international standards, negatively affecting acoustic comfort, increasing annoyance, and influencing individuals’ intent to leave. Personal noise sensitivity further intensifies these effects.

 Hospital noise Noise Sensitivity Noise Annoyance Acoustic Comfort Intention to Leave The author(s) declared that no grants were involved in supporting this work. Amendments from Version 1

This article has been revised and expanded based on the comments and suggestions of the respected reviewers in the sections on the title, abstract, introduction, methodology, results, discussion, study limitations, and conclusion.

 1. Introduction

Hospitals are intended to serve as healing environments, yet they are often characterized by excessive levels of environmental noise that can contradict their therapeutic purpose. Sources of hospital noise are numerous and varied, ranging from medical equipment, ventilation systems, and staff activities, to patient movement and visitor interactions. 1, 2 In high-density hospital settings, noise levels frequently exceed internationally recommended thresholds, potentially interfering with both patient recovery and staff performance. 3 The World Health Organization suggests that noise levels in hospital wards should not exceed 35 dB during the day and 30 dB at night. However, many studies have shown that these levels are consistently surpassed, often reaching peaks above 85 dB. 4, 5

Excessive noise in healthcare facilities is more than a simple environmental nuisance—it can have complex consequences on physiological, psychological, and behavioral health. Research has demonstrated associations between prolonged exposure to noise and hearing impairments, risk of cancer, elevated heart rates, blood pressure irregularities, increased cortisol levels, and sleep disturbances in patients. 6– 8 On a psychological level, noise contributes to anxiety, reduced performance, stress, and emotional distress, complicating recovery trajectories. 9– 11 For healthcare providers, the effects are similarly detrimental, leading to higher levels of occupational burnout, fatigue, and even intention to resign. 12, 13 Such outcomes underline the imperative need to understand and mitigate the impact of hospital noise exposure.

Among patients, individual differences in noise sensitivity may moderate their responses to environmental soundscapes. Noise sensitivity is a stable personality trait that reflects how strongly a person reacts to noise, independent of the objective sound intensity. 10 Highly noise-sensitive individuals are more likely to experience discomfort and perceive sound as more disturbing than others, which may amplify their annoyance and overall dissatisfaction with the hospital experience. 14 In clinical settings, noise-sensitive patients might display lower pain thresholds, poorer sleep quality, and greater susceptibility to noise-induced stress.

One critical dimension in understanding patients’ reactions to noise is acoustic comfort—a concept referring to the subjective perception of the acoustic environment as pleasant, tolerable, and non-disruptive. While often discussed in architectural and environmental psychology literature, acoustic comfort has recently attracted attention in hospital research due to its connection to patient satisfaction and healing outcomes. 15 Factors influencing acoustic comfort include background noise levels, variability, predictability, and the perceived control over the auditory environment. 6 Poor acoustic comfort can lead to reduced trust in medical care, increased anxiety, and decreased willingness to remain in the facility.

Relatedly, noise annoyance—defined as a subjective negative reaction to noise—is one of the most reported emotional responses among hospitalized individuals. Annoyance can arise from both the intensity and the meaning attributed to noise, with many patients perceiving certain sounds (e.g., alarms, staff conversations) as intrusive or inappropriate in a care context. 1 Chronic annoyance is not merely an emotional state but has been linked to increased health risks and behavioral outcomes, such as reduced adherence to treatment, complaints, and negative evaluations of care. 9

Perhaps one of the most understudied yet consequential behavioral outcomes in hospital noise research is intention to leave, which refers to the patient’s conscious consideration or decision to discharge prematurely due to discomfort or dissatisfaction. While extensively studied among healthcare staff in relation to burnout and workplace environment, 12 little is known about how environmental noise might affect patients’ intentions to leave. This is particularly relevant given that early discharge—if not medically indicated—can negatively affect health outcomes and impose further burdens on healthcare systems. Despite growing awareness of hospital noise pollution, existing research has predominantly focused on staff-related outcomes or examined noise as an isolated environmental factor, with limited empirical attention to how individual noise sensitivity interacts with acoustic conditions to shape patient-centered outcomes such as acoustic comfort, noise annoyance, and behavioral intentions. 16, 17 This gap is particularly pronounced in low- and middle-income countries—including Iran—where healthcare facilities often rely on generalized international guidelines (e.g., WHO, EPA) in the absence of enforceable national standards tailored to local contexts, resulting in scarce region-specific data to inform acoustic design and noise reduction policies. 5

Addressing these gaps requires a multidimensional framework that integrates environmental, psychological, and behavioral domains. Conventional regression approaches are ill-equipped to capture the complex, non-linear, and interdependent relationships inherent in such a framework. To overcome this methodological limitation, the present study applies Bayesian Network (BN) modeling—a probabilistic graphical approach that explicitly models conditional dependencies, visualizes causal structures, and quantifies both direct and joint effects—enabling a more nuanced examination of how noise exposure and noise sensitivity synergistically influence patient outcomes. Accordingly, this study was designed to examine the complex relationships among hospital noise pollution, noise sensitivity, and key patient outcomes including acoustic comfort, noise annoyance, and intention to leave, thereby contributing to the evidence base for acoustically mindful hospital design and policy in settings where such data are urgently needed.

 2. Methodology 2.1 Study design and context

This descriptive-analytical cross-sectional study with a basic-applied orientation was conducted in 2024 across a public hospital in Saveh, Iran, with the aim of investigating how environmental noise pollution and individual differences in noise sensitivity impact patients’ acoustic comfort, noise-related annoyance, and intention to leave the hospital. The hospital setting, with its complex combination of medical equipment, staff movement, and patient activities, provided a real-world context in which environmental acoustics play a crucial role in shaping patient experience.

 2.2 Participants and sampling strategy

The minimum required sample size was determined using G*Power 3.1 software, assuming a medium effect size (0.3), α = 0.05, and a statistical power of 0.95. Based on these parameters, a minimum sample of 178 participants was deemed necessary. To account for potential attrition, this number was increased by 15%, yielding a final sample of 205 participants—the minimum acceptable number for sufficient statistical power and adequate generalizability. During data collection, a total of 226 eligible patients were recruited and participated voluntarily. This larger achieved sample strengthens the generalizability of the findings and is fully consistent with the sampling strategy, as no upper limit was imposed on participation.

The study population consisted of hospitalized adult patients who had stayed for a minimum of 48 hours in one of the active hospital wards, including internal medicine, general surgery, and coronary care unit (CCU) department. Eligible participants were those who were 18 years of age or older, fully conscious, capable of verbal communication, and clinically stable during the time of data collection. Patients were excluded if they had been transferred to intensive care units, experienced acute medical deterioration, or provided incomplete questionnaire responses.

Sampling was conducted using a stratified random sampling method with proportional allocation. Initially, a comprehensive list of all hospital wards was prepared, and a proportionate number of patients was randomly selected from each ward based on its occupancy size. Within each ward, participants were selected randomly and voluntarily from different patient rooms. This approach ensured a representative distribution of the sample across the hospital’s various departments, thereby increasing the external validity of the findings.

To further enhance data reliability and contextual accuracy, data collection was carried out through interviewer-administered questionnaires. Researchers were physically present at the patients’ bedside, assisting participants in completing the questionnaires. This hands-on approach not only improved response accuracy but also enabled researchers to better understand the patients’ subjective experiences with hospital noise in real time. In accordance with the Declaration of Helsinki, ethical approval for this study was granted by the Medical Ethics Committee of Saveh University of Medical Sciences (Ethics Code: IR.SAVEHUMS.REC.1403.039). All procedures were conducted in full compliance with the approved ethical guidelines. Prior to participation, all individuals received comprehensive information regarding the study’s objectives, methodology, and potential risks. Written informed consent was obtained from all participants, affirming their voluntary participation.

 2.3 Objective measurement of day-evening-night noise levels (L <sub>den</sub>)

Environmental noise data were collected over a three-month period from 10 November 2024 to 10 February 2025, across multiple active hospital wards under the jurisdiction of Saveh University of Medical Sciences. To comprehensively capture acoustic variability throughout the day and night, equivalent continuous sound pressure levels were recorded in eight distinct time intervals: 07:00–10:00, 10:00–13:00, 13:00–16:00, 16:00–19:00, 19:00–22:00, 22:00–01:00, 01:00–04:00, and 04:00–07:00. These cycles were systematically rotated across weekdays to ensure that the acoustic conditions were representative of both weekdays (Saturday to Thursday in Iran) and weekends (Friday). These intervals were later aggregated into three standardized periods for analysis—day (07:00–19:00), evening (19:00–23:00), and night (23:00–07:00). Noise exposure levels were assessed as equivalent continuous sound pressure levels (Leq) over consecutive 15-minute intervals, with each 15-minute Leq measurement serving as a proxy for the corresponding three-hour timeframe. These 15-minute Leq values were aggregated to derive cumulative noise levels for the morning, afternoon, and nighttime periods. Over a 24-hour monitoring cycle, individual participants contributed 120 minutes of sampled data. Across the entire study, this methodology yielded a comprehensive dataset exceeding 24,600 minutes of noise measurements, collected between October 2024, and February 2025. To calculate the weighted 24-hour equivalent noise level, the Level day-evening-night (L den) values corresponding to the day, evening, and night periods were combined using the following standard formula, which accounts for increased human sensitivity during non-daytime hours 18 : L den = 10 . log 10 ( 1 24 ( 12 . 10 L day 10 + 4 . 10 L evening + 5 10 + 8 . 10 L night + 10 10 ) )

In this formula, L day, L evening, and L night represent the measured equivalent noise levels during the morning, evening, and night intervals, respectively. The weighting factors (i.e., +5 dB for evening and +10 dB for night) reflect internationally accepted corrections for human sensitivity to noise during these periods, as outlined by the World Health Organization and ISO acoustic standards.

Noise exposure was assessed using a B&K Sound Level Meter Analyzer, model 2250, which was calibrated before each measurement session using a TES 1356 calibrator at 1000 Hz and 94 dB. The sound level meter was operated in the “slow” response mode and adjusted to an appropriate range to capture real-time acoustic fluctuations in the hospital environment. While the measurement protocol generally followed principles outlined in ISO 9612:2009, adjustments were made to suit clinical conditions. Specifically, the microphone was positioned at the typical auditory height of a recumbent patient, facing dominant noise sources within each ward to ensure ecological validity.

 2.4 Questionnaires

Data collection was performed using a structured questionnaire pack including demographic data and four psychometrically validated instruments:

Background and demographic information

Basic demographic and background data were collected using a researcher-designed form. This section of the instrument gathered information on the participants’ age, gender, marital status, and the hospital ward in which they were admitted. These variables were used to provide a descriptive profile of the sample and to contextualize the interpretation of the study findings.

Weinstein Noise Sensitivity Scale (WNSS)

The WNSS is a 21-item scale developed to assess individual noise sensitivity, an internal trait influencing reactions to acoustic stimuli. Items are rated on a 6-point Likert scale (0 = “completely agree” to 5 = “completely disagree”), yielding scores between 0 and 105, with higher scores indicating greater sensitivity. 19 The Persian version, validated by Alimohammadi et al., demonstrated good internal consistency (Cronbach’s α = 0.78) and has been used in various Iranian populations. 20

Acoustic comfort scale

Perceived acoustic comfort was measured through a 7-point semantic differential scale ranging from 1 (very uncomfortable) to 7 (very comfortable). This instrument captures the subjective evaluation of soundscapes, a construct increasingly relevant in hospital design and healthcare experience research. 21

Visual Analogue Scale (VAS)

This study utilized a 100-mm Visual Analogue Scale (VAS) to measure noise annoyance and intention to leave the hospital, with endpoints anchored at 0 (“not at all”) and 100 (“extremely”). The VAS is widely recognized for its sensitivity and precision in capturing subjective experiences in healthcare settings, offering distinct advantages over categorical scales, including greater measurement granularity and reduced susceptibility to ceiling effects. 22, 23 The validity of using VAS for noise-related outcomes in hospitals has been demonstrated in prior research. Sayılan et al. (2021), for example, used the VAS to assess patient responses to varying noise levels in intensive care units, reporting statistically significant differences across acoustic conditions (p < 0.01). This finding underscores the responsiveness of the VAS to environmental factors within clinical contexts. 24 Given its methodological strengths—including ease of administration, enhanced sensitivity, and minimal cognitive burden—the VAS serves as an optimal tool for evaluating noise-related perceptions and behavioral intentions in inpatient populations.

Visual analogue scale for noise annoyance

Noise annoyance was measured using a 100-mm Visual Analogue Scale (VAS). Participants were asked the following question:

“How much do the sounds and noises in the hospital bother (annoy) you?”

They responded by marking a point on a horizontal line, where 0 indicated “not at all” and 100 indicated “extremely bothersome”. Higher scores reflected greater perceived annoyance due to hospital noise.

Visual analogue scale for intention to leave the hospital

Intention to leave the hospital was assessed using a single-item VAS. Participants were asked:

“How much does the noise in the hospital make you want to leave or change your room?”

Responses were marked on a 100-mm line anchored at 0 (“no desire to leave”) and 100 (“very strong desire to leave”). Higher scores indicated a stronger inclination to leave or avoid the hospital environment due to noise exposure.

 2.5 Statistical analysis

Subsequent to the completion of data acquisition, a preliminary phase of statistical analysis was executed using SPSS software, version 27. This encompassed the generation of descriptive statistics, including measures of central tendency (mean, median, mode) and dispersion (standard deviation, variance, range), as detailed in Reference [ 25]. These initial analyses provided a foundational understanding of the data distribution and variability, crucial for subsequent modeling.

Building upon this descriptive foundation, Bayesian Networks (BNs), a class of probabilistic graphical models pioneered by Pearl, 26 were deployed to explore the complex interrelationships within the data. BNs, by their nature, represent systems as directed acyclic graphs (DAGs), where nodes symbolize variables, and edges signify probabilistic dependencies. This framework explicitly delineates cause-effect relationships, enabling the visualization and quantification of how changes in one variable propagate through the system to influence others. Their ability to integrate heterogeneous data types, including both quantitative and qualitative variables, and to model complex interactions and outcomes, while simultaneously facilitating the exploration of trade-offs, positions BNs as particularly advantageous for modeling intricate causal systems. Moreover, BNs demonstrate resilience in handling data originating from diverse sources and are adept at managing datasets with missing values through probabilistic inference. The inherent causal graphical structure of BNs, characterized by conditional dependencies, promotes accessibility, allowing for the construction of models without requiring extensive technical modeling expertise and facilitating comprehension by non-technical stakeholders, a feature of substantial practical utility. 26, 27

The BN topology (nodes and directed links) was defined to reflect the study structure and was derived from the variables/domains captured by the questionnaires used in this study. The BN was then used for probabilistic inference (evidence propagation) to quantify interdependencies among nodes and to examine how changing the state of one or more nodes updates the probability distributions of other variables. Conditional Probability Tables (CPTs) were subsequently estimated from the observed study data using GeNIe.

The construction and analysis of the Bayesian network were carried out using GeNIe software, version 2.0. Following the development of the BN graphical structure, which involved defining nodes and edges based on domain knowledge and data relationships, Conditional Probability Tables (CPTs) were generated. 28 CPTs quantify the conditional probabilities of each node given its parent nodes, providing a complete probabilistic specification of the network. To determine the relative importance of individual variables within the network, a delta-p sensitivity analysis was conducted. 29, 30 This analysis involved systematically varying the states of input nodes and observing the resulting changes in the probabilities of target nodes. In intuitive terms, Δp quantifies how much the probability of each state of a target node (low, moderate, high) changes after setting evidence on an input node to a specific state (low, moderate, or high), while keeping the BN structure constant. For each evidence scenario, Δp for a given target state was computed as the updated probability of that state minus its baseline probability prior to setting evidence (updated − baseline). Larger |Δp| values indicate a stronger influence on that target state, and the sign indicates whether the probability of that state increases (+) or decreases (−).

Continuous variables were discretized prior to Bayesian Network modeling using sample-based percentile thresholds. Values below the 25th percentile (P25) were defined as low, values between the 25th and 75th percentiles (P25–P75) as moderate, and values above the 75th percentile (P75) as high. The same categorization rule was applied consistently across all continuous variables included in the model.

 3. Results

A total of 226 hospitalized patients participated in the study, with near-equal gender distribution (49.6% male, 50.4% female). The majority were married (73.0%), and approximately half were admitted to internal medicine wards (42.2%), followed by general surgery (37.2%) and the coronary care unit (18.6%). Detailed demographic characteristics are presented in Table 1.

Demographic characteristics of the study participants.
Variable Category Frequency Percent Cumulative
Gender Male 112 49.6 49.6
Female 114 50.4 100.0
Total 205 100.0 -
Marital Status Single 61 27.0 27.0
Married 165 73.0 100.0
Total 205 100.0 -
Hospital Ward Internal Medicine 100 42.2 42.2
General Surgery 84 37.2 81.4
Coronary Care Unit 42 18.6 100.0
Total 226 100.0 -

The mean ± SD of the dimensions of abovementioned variables is presented in Table 2.

Descriptive statistics and categorical distributions of study variables.
Variable Level Frequency Percent Total measured Mean Total measured SD
L den Low 70 31 57.95 06.61
Moderate 87 38.5
High 69 30.5
Noise Sensitivity Low 55 24.8 42.57 18.16
Moderate 113 50.9
High 54 24.3
Noise Annoyance Low 79 35 38.46 17.07
Moderate 74 32.7
High 73 32.3
Acoustic Comfort Low 94 25.2 04.73 01.45
Moderate 63 36
High 69 38.7
Intention to leave Low 58 27.5 26.06 11.78
Moderate 87 38.5
High 81 35.8

Note: Lden = day-evening-night weighted equivalent noise level. Mean and SD values represent the original continuous measurements for each variable prior to categorization. Lden, Noise Sensitivity, Noise Annoyance, Acoustic Comfort, Intention to.

The dependencies among the marginal probabilities of the studied variables based on the Bayesian network model is shown in the Figure 1.

The dependencies among the marginal probabilities of the studied variables based on the Bayesian network model.

Nodes represent variables; directed edges (arrows) indicate conditional dependencies. Lden (day-evening-night noise level) and Noise Sensitivity are exogenous predictors. Noise Annoyance and Acoustic Comfort are positioned as intermediate nodes in the probabilistic pathways from Lden and Sensitivity to Intention to Leave, consistent with a mediating structure.

Table 3 illustrates the marginal probability distribution, derived from the Conditional Probability Tables (CPT), for the examined variables. The CPTs quantify the associations among the variables.

Marginal probability distributions of study variables derived from Bayesian Network model.
Variable Level Marginal probability distribution
L den Low 0.310
Moderate 0.385
High 0.305
Sensitivity Low 0.350
Moderate 0.327
High 0.323
Annoyance Low 0.346
Moderate 0.313
High 0.341
Acoustic Comfort Low 0.350
Moderate 0.371
High 0.279
Intention to leave Low 0.326
Moderate 0.470
High 0.204

Note: Values are rounded consistently based on the original Bayesian Network outputs; therefore, each probability distribution sums to 100%.

Model updating was performed by instantiating the target node as evidence, leading to a revision of all variable probability distributions within the Bayesian Network. The updated probabilities are detailed in Table 4.

Updated probability distributions of outcome variables under five evidence scenarios.
Variable Level L den (High 100%) Noise Sensitivity (High 100%) L den (High 100%) and Noise Sensitivity (Low 100%) L den (Low 100%) and Noise Sensitivity (High 100%) L den and Noise Sensitivity (High 100%)
L den Low 0 0.310 0 1 0
Moderate 0 0.385 0 0 0
High 1 0.305 1 0 1
Noise Sensitivity Low 0.350 0 1 0 0
Moderate 0.327 0 0 0 0
High 0.323 1 0 1 1
Noise Annoyance Low 0.252 0.250 0.350 0.350 0.150
Moderate 0.284 0.320 0.300 0.350 0.250
High 0.464 0.430 0.350 0.300 0.600
Acoustic Comfort Low 0.414 0.410 0.350 0.350 0.481
Moderate 0.358 0.360 0.380 0.380 0.319
High 0.228 0.230 0.270 0.270 0.200
Intention to leave Low 0.254 0.254 0.320 0.325 0.180
Moderate 0.492 0.494 0.460 0.455 0.510
High 0.254 0.252 0.220 0.220 0.310

Note: Baseline probabilities represent the marginal distribution from the Bayesian network prior to evidence instantiation. Values are rounded consistently based on the original Bayesian Network outputs; therefore, each probability distribution sums to 100%.

Under conditions of high Level L den, the probability of high annoyance, low acoustic comfort and intention to leave variables increased by 12.4%, 6.3% and 5%, respectively. In contrast, the probabilities of low annoyance, low intention to leave and high acoustic comfort decreased by 9.4%, 7.4% and 5.1% ( Figure 2).

Sensitivity analysis on high L <sub>den.</sub>

Delta-p sensitivity analysis results under evidence scenario: Noise exposure = 100% High. Bars represent percentage point changes in the probability of each outcome variable state relative to baseline marginal probabilities. Positive values (right-directed bars) indicate increased probability; negative values (left-directed bars) indicate decreased probability.

Under conditions of high-Level Sensitivity, the probability of high annoyance, low acoustic comfort and high intention to leave variables increased by 9.1%, 6.2 and 4.7%, respectively. In contrast, the probabilities of low annoyance, low intention to leave and high acoustic comfort decreased by 9.6%, 7.2% and 5.2% ( Figure 3).

Sensitivity analysis on high sensitivity.

Delta-p sensitivity analysis results under evidence scenario: Noise Sensitivity = 100% High. Bars represent percentage point changes in the probability of each outcome variable state relative to baseline marginal probabilities. Positive values (right-directed bars) indicate increased probability; negative values (left-directed bars) indicate decreased probability.

Among the variables under conditions of 100% High Level L den and 100% Low Sensitivity, the changes were observed in: a 1.8% increase in high intention to leave, a 1.1% increase in moderate acoustic comfort and low annoyance, a 1.5% decrease in moderate intention to leave and a 1.3% decrease in low annoyance ( Figure 4). It is notable that the extent of changes in this section is negligible.

Sensitivity analysis on high Lden and low sensitivity.

Delta-p sensitivity analysis results under evidence scenario: Lden = 100% High AND Noise Sensitivity = 100% Low. Bars represent percentage point changes in outcome probabilities relative to baseline.

Concerning the variables under the stipulated conditions of 100% Low Level L den and 100% High sensitivity, the most prominent augmentations were evidenced in moderate annoyance, high intention to leave and moderate acoustic comfort, quantified at 3.7%, 1.5%, and 1.1%, respectively. Conversely, diminutions of 3.9%, 3%, and 1.5% were observed in the values of high annoyance, low acoustic comfort and moderate intention to leave ( Figure 5).

Sensitivity analysis on low L <sub>den</sub> and high sensitivity.

Delta-p sensitivity analysis results under evidence scenario: Lden = 100% Low AND Noise Sensitivity = 100% High. Bars represent percentage point changes in outcome probabilities relative to baseline.

Under conditions of 100% High Level L den and Sensitivity, analysis of change variables revealed that the most substantial positive variations occurred in high annoyance, low acoustic comfort and high intention to leave, with increases of 26.1%, 13.1% and 10.6%. Conversely, a corresponding analysis indicated decreases of 19.6%, 14.3% and 10.5% in low annoyance and intention to leave ( Figure 6).

Sensitivity analysis on high L <sub>den</sub> and sensitivity.

Delta-p sensitivity analysis results under evidence scenario: Lden = 100% High AND Noise Sensitivity = 100% High. Bars represent percentage point changes in outcome probabilities relative to baseline.

The sensitivity analysis of the studied variables is reported in Table 5, with a positive sign indicating an increase and a negative sign indicating a decrease.

Delta-p sensitivity analysis: percentage point changes in outcome probabilities under five evidence scenarios.
Variable Level L den (High 100%) Sensitivity (High 100%) L den (High 100%) and Sensitivity (Low 100%) L den (Low 100%) and Sensitivity (High 100%) L den and Sensitivity (High 100%)
Lden Low - 0 - - -
Moderate - 0 - - -
High - 0 - - -
Sensitivity Low 0 - - - -
Moderate 0 - - - -
High 0 - - - -
Annoyance Low -9.4 % -9.6 % +0.4 % +0.4 % -19.6 %
Moderate +3 % +0.6 % -1.3 % +3.7% -6.3 %
High +12.4 % + 9.1% + 1.1% - 3.9% +26.1
Acoustic Comfort Low + 6.3 % + 6.2% 0% - 3% + 13.1%
Moderate -1.3 % - 1.1% +1.1% +1.1% - 5.2 %
High -5.1% -5.2% -1.1% -0.8% - 7.8%
Intention to leave Low -7.2% -7.2% - 0.4% - 0.1 % -14.3%
Moderate +2.2% +2.4% - 1.5 % - 1.5 % +3.7 %
High + 5% +4.7% + 1.8% + 1.5% + 10.6 %

Note: Values represent the difference (in percentage points) between the posterior probability under each evidence scenario and the baseline marginal probability ( Table 4).

 4. Discussion

The present study elucidates the complex interplay between hospital noise pollution, noise sensitivity, and critical patient outcomes, including acoustic comfort, noise annoyance, and intention to leave. Our findings reveal several noteworthy patterns that both corroborate and extend existing literature in healthcare acoustics. The observed mean L den level of 57.95 dB (±6.61) substantially exceeds WHO recommendations (35 dB daytime/30 dB nighttime). This finding mirrors recent global studies. Nyembwe et al. (2023) documented comparable noise levels (55–72 dB) in ICUs across Congolese hospitals, and Amoatey et al. (2022) reported 24-hour averages of 63.5 dB in Omani healthcare facilities. 4, 5

Bayesian Network analysis demonstrated that high noise sensitivity amplified the negative effects of noise exposure. When both factors were present simultaneously, the probability of high noise annoyance increased by 26.1%. This finding aligns with emerging neurophysiological evidence from Zhou et al. (2020), whose fMRI studies revealed heightened amygdala activation in noise-sensitive individuals exposed to hospital sounds. 6 The observed moderated mediation pattern provides empirical support for the Stressor-Sensitivity Model proposed by Gong et al. (2022). This model posits that individual differences in sensory processing modulate environmental stress responses. 9 Noise annoyance has long been recognized as one of the primary effects of noise exposure across various environments. It is also closely correlated with individual noise sensitivity. In other words, noise annoyance is influenced by two sets of variables: noise exposure, which acts as an external factor, and noise sensitivity, which is an individual-specific characteristic. The results of the present study indicate that both of these factors contribute to increased annoyance among patients. Moreover, when both variables were set at their highest levels in the model, they demonstrated a synergistic effect, leading to a significantly greater increase in perceived annoyance. The literature has shown that noise annoyance acts as a mediator of the various health and psychological outcomes of noise exposure. 31 Numerous studies have indicated that individuals who are highly sensitive to noise experience greater annoyance when exposed to environmental noise, which in turn is associated with more adverse health outcomes. In this study, it was found that noise annoyance increased under all conditions — whether due to high noise exposure or heightened individual sensitivity to noise. This increase in annoyance was accompanied by a corresponding rise in the intention to leave the hospital. Thus, both noise exposure and noise sensitivity influence the intention to leave the hospital through pathways that may include direct effects as well as indirect effects via noise annoyance. However, formal statistical mediation testing was not conducted in this study. In all cases, increases in either noise levels or sensitivity led to higher intentions to leave hospital. It has been reported that annoying stimuli such as noise, can increase individual arousal and provide a motivating reason to leave the environment. 32 Therefore, it can be concluded that in hospital settings, noise annoyance may lead patients to decide to leave the facility prematurely, before completing their prescribed course of hospitalization.

Acoustic comfort is a subjective evaluation of the sound environment, influenced by both external acoustic stimuli and individual psychological predispositions such as noise sensitivity. Patients with heightened noise sensitivity tend to appraise hospital soundscapes more negatively, reporting lower levels of acoustic comfort even under moderate noise conditions. 6 This discomfort may stem from sensory overload, heightened vigilance, or perceived loss of control over the environment. In the present study, patients with high noise sensitivity demonstrated a pronounced decrease in acoustic comfort, particularly when ambient noise levels were also elevated. Reduced acoustic comfort, in turn, was associated with an increased intention to leave the hospital. This pattern is consistent with a mediating role for acoustic comfort, though formal mediation analysis was not performed. Studies suggest that when patients feel aurally overwhelmed, their sense of well-being declines. This decline leads to emotional fatigue and decreased willingness to remain in the hospital. 9 Thus, noise sensitivity and environmental exposure jointly reduce acoustic comfort. This reduction may, in turn, contribute to patients' decisions to leave the care environment prematurely—a pathway suggestive of mediation. Beyond its effects on comfort and annoyance, the interaction between noise exposure and individual sensitivity may directly influence patients’ behavioral responses, including the decision to leave the hospital early. In our model, high L den and elevated noise sensitivity increased the probability of moderate and high intention to leave by more than 3.7% and 10.6%, respectively. This behavior may be driven by psychological mechanisms such as avoidance coping or perceived threat, wherein patients interpret persistent noise as a signal of low-quality care or lack of control. Premature hospital discharge has been associated with adverse outcomes, including incomplete treatment, higher readmission rates, and reduced patient satisfaction. 14 While such departures are typically multifactorial, environmental discomfort—particularly in the form of uncontrolled noise—may act as a powerful yet under-recognized contributor.

According to the results presented in Table 5, noise annoyance exhibits greater variability compared to acoustic comfort, primarily due to fluctuations in both noise sensitivity and exposure levels. This finding highlights the significance of noise annoyance as a key factor within hospital environments. Numerous studies have demonstrated that noise annoyance has a substantial impact on various aspects of health, including the elevation of stress hormone levels. 9 Furthermore, it can contribute to increased anxiety and depressive symptoms in individuals. 9, 33 Considering that patients are often already in a compromised state of health, noise annoyance may exacerbate their medical conditions by heightening stress, anxiety, and depression, potentially leading them to leave the hospital prematurely—before completing the prescribed treatment process. In summary, noise annoyance, as the most prominent consequence of noise exposure and heightened noise sensitivity, can give rise to a wide range of adverse health outcomes. 33

5. Limitations

While this study provides valuable insights into the relationships between hospital noise pollution, individual noise sensitivity, and key patient outcomes, several limitations should be considered. First, the study was conducted in a single public hospital, which may limit the generalizability of the findings to other healthcare settings, particularly private institutions or those with different infrastructural and organizational characteristics. Second, the cross-sectional design restricts the ability to establish causal or temporal relationships between environmental noise exposure and patients’ psychological or behavioral responses. Third, although ward type was accounted for, other potentially influential clinical factors—such as pain levels, medication use, or individual health conditions—were not controlled. Forth, this study focused exclusively on acoustic environmental factors and did not measure other physical environmental parameters (e.g., thermal comfort, lighting, air quality, privacy) that may also influence patients' intention to leave. Moreover, the use of L den as the sole noise metric, while practical and standardized, may have failed to capture more granular temporal fluctuations in noise exposure that could influence patient responses. Lastly, the use of a single-item Visual Analogue Scale to measure intention to leave, while appropriate for this unidimensional and context-specific construct, does not permit assessment of internal consistency and may not capture the full complexity of patients' decision-making processes. More broadly, reliance on self-reported measures introduces the possibility of response bias, despite the use of validated instruments. Development and validation of multi-item instruments for patient voluntary departure intention attributable to environmental factors, as well as incorporation of objective behavioral indicators (e.g., actual against-medical-advice discharge rates, room transfer requests), are important priorities for future research.

 6. Policy and design implications

This persistent non-compliance with international standards across diverse healthcare systems suggests a systemic failure in noise control implementation that transcends geographical boundaries. Therefore, mitigating hospital noise and addressing the needs of noise-sensitive individuals is not only a matter of comfort, but also a determinant of patient retention and treatment adherence. Both noise exposure and noise sensitivity influence the intention to leave the hospital through direct and indirect pathways, with noise annoyance serving as an intermediate variable. Noise annoyance, as the most prominent consequence of noise exposure and heightened sensitivity, can give rise to a wide range of adverse health outcomes. Accordingly, addressing noise pollution—particularly through targeted interventions for noise-sensitive individuals and improved acoustic design—should be viewed as an essential dimension of quality healthcare delivery.

 7. Conclusion

Hospital soundscapes are active components of the care environment, not merely ambient features. This study demonstrates that noise exposure and individual noise sensitivity synergistically increase annoyance, reduce acoustic comfort, and elevate patients' intention to leave. The findings position acoustic comfort and noise annoyance as intermediate variables in these pathways. Addressing noise pollution—particularly through targeted interventions for noise-sensitive individuals and improved acoustic design—is therefore an essential dimension of quality healthcare delivery.

 Ethics approval and consent to participate

In accordance with the Declaration of Helsinki, ethical approval for this study was granted by the Medical Ethics Committee of Saveh University of Medical Sciences (Ethics Code: IR.SAVEHUMS.REC.1403.039). All procedures were conducted in full compliance with the approved ethical guidelines.

 Data availability statement

Figshare: The Effects of Hospital Noise Pollution and Noise Sensitivity on Patient’s Acoustic Comfort, Noise Annoyance, and Intention to Leave, https://doi.org/10.6084/m9.figshare.30373360.v1. 34

The project contains the following underlying data:

Hospital_Noise_Study_Data.sav: (A structured data file containing all anonymized patient responses and calculated scores for acoustic comfort, noise annoyance, noise sensitivity, and intention to leave.)

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).

 Acknowledgments

The authors appreciate all the patients who took part in this study.

 References Das A Kishore J : Noise levels and annoyance among staff in a tertiary care hospital in North India: A cross-sectional study. Arch. Environ. Occup. Health. 2023;78(9-10):471–482. 38009651 10.1080/19338244.2023.2286388 Abbasi M Tokhi MO Eyvazzadeh N : Prioritization of noise abatement methods for controlling hospital noise pollution. Journal of Low Frequency Noise, Vibration and Active Control. 2024;43(3):1126–1138. 10.1177/14613484241245002 Lima Andrade E de Cunha e Silva DC da Lima EA de : Environmental noise in hospitals: a systematic review. Environ. Sci. Pollut. Res. 2021;28:19629–19642. 33674976 10.1007/s11356-021-13211-2 Nyembwe J-PKB Ogundiran JO Gameiro da Silva M : Evaluation of noise level in Intensive Care units of hospitals and noise mitigation strategies, Case Study: Democratic Republic of Congo. Buildings. 2023;13(2):278. 10.3390/buildings13020278 Amoatey P Al-Harthy I Al-Mushaifari MA : Effect of ambient noise on indoor environments in a health care facility in Oman. Environ. Sci. Pollut. Res. 2022;17:21–29. 10.1680/jenes.20.00040 Zhou T Wu Y Meng Q : Influence of the acoustic environment in hospital wards on patient physiological and psychological indices. Front. Psychol. 2020;11:1600. 32848994 10.3389/fpsyg.2020.01600 PMC7396688 Abbasi M Derakhshan J Darabi F : The impact of noise-induced hearing loss on individual job performance: exploring the role of aggression and work-related quality of life. BMC Psychology. 2024;12(1):624. 39501410 10.1186/s40359-024-02113-w PMC11536970 Abbasi M Yazdanirad S Dehdarirad H : Noise exposure and the risk of cancer: a comprehensive systematic review. Rev. Environ. Health. 2023;38(4):713–726. 36064622 10.1515/reveh-2022-0021 Gong X Fenech B Blackmore C : Association between noise annoyance and mental health outcomes: A systematic review and meta-analysis. Int. J. Environ. Res. Public Health. 2022;19(5):2696. 35270388 10.3390/ijerph19052696 PMC8910193 Abbasi M Yazdanirad S Zokaei M : A Bayesian network model to predict the role of hospital noise, annoyance, and sensitivity in quality of patient care. BMC Nurs. 2022;21(1):243. 36050728 10.1186/s12912-022-00948-5 PMC9435418 Monazzam Esmaielpour MR Zakerian SA Abbasi M : Investigating the effect of noise exposure on mental disorders and the work ability index among industrial workers. Noise & Vibration Worldwide. 2022;53(1-2):3–11. 10.1177/09574565211052690 Tang M Liu L Cai J : Effect of Noise in the Emergency Department on Occupational Burnout and Resignation Intention of Medical Staff. Noise Health. 2024;26(121):102–106. 38904808 10.4103/nah.nah_127_23 PMC11530110 Armbruster C Walzer S Witek S : Noise exposure among staff in intensive care units and the effects of unit-based noise management: a monocentric prospective longitudinal study. BMC Nurs. 2023;22(1):460. 38057790 10.1186/s12912-023-01611-3 PMC10699060 Voitl P Sebelefsky C Mayrhofer C : Noise levels in general pediatric facilities: A health risk for the staff? Plos one. 2019;14(3):e0213722. 30865703 10.1371/journal.pone.0213722 PMC6415854 Walker L Karl CA : The hospital (not so) quiet zone: Creating an environment for patient satisfaction through noise reduction strategies. HERD: Health Environments Research & Design Journal. 2019;12(4):197–202. 30947541 10.1177/1937586719839229 Terzi B Azizoğlu F Polat Ş : The effects of noise levels on nurses in intensive care units. Nurs. Crit. Care. 2019;24(5):299–305. 10.1111/nicc.12414 Khaiwal R Singh T Tripathy JP : Assessment of noise pollution in and around a sensitive zone in North India and its non-auditory impacts. Sci. Total Environ. 2016;566-567:981–987. 10.1016/j.scitotenv.2016.05.070 Brink M Schäffer B Pieren R : Conversion between noise exposure indicators Leq24h, LDay, LEvening, LNight, Ldn and Lden: Principles and practical guidance. Int. J. Hyg. Environ. Health. 2018;221(1):54–63. 29037963 10.1016/j.ijheh.2017.10.003 Weinstein ND : Individual differences in reactions to noise: a longitudinal study in a college dormitory. J. Appl. Psychol. 1978;63(4):458–466. 701213 10.1037/0021-9010.63.4.458 Alimohammad I Azkhosh S : Reliability any Validity of the Persian Translation of the Weinstein Noise Senditivity scal The validity and validity of the Persian translation of Weinstein’s voice sensitivity test. Psychol. Res. 2006;17(9):74–87. Yang W Kang J : Acoustic comfort evaluation in urban open public spaces. Appl. Acoust. 2005;66(2):211–229. 10.1016/j.apacoust.2004.07.011 Voutilainen A Pitkäaho T Kvist T : How to ask about patient satisfaction? The visual analogue scale is less vulnerable to confounding factors and ceiling effect than a symmetric Likert scale. J. Adv. Nurs. 2016;72(4):946–957. 26689434 10.1111/jan.12875 Sung Y-T Wu J-S : The visual analogue scale for rating, ranking and paired-comparison (VAS-RRP): a new technique for psychological measurement. Behav. Res. Methods. 2018;50:1694–1715. 29667082 10.3758/s13428-018-1041-8 PMC6096654 Aydın Sayılan A Kulakaç N Sayılan S : The effects of noise levels on pain, anxiety, and sleep in patients. Nurs. Crit. Care. 2021;26(2):79–85. 32621391 10.1111/nicc.12525 Čaplová Z Švábová P : IBM SPSS statistics. Statistics and Probability in Forensic Anthropology. Elsevier;2020; pp.343–352. Sentz K Ferson S : Combination of evidence in Dempster-Shafer theory. 2002. Rafiee M Eskandari T Mahmood EA : The effect of COVID-19 anxiety on nurses’ productivity determinants through perceived workload and individual job performance: A Bayesian mediation analysis. Heliyon. 2024;10(14):e34099. 39149068 10.1016/j.heliyon.2024.e34099 PMC11324830 Liu X Huang G Huang H : Safety climate, safety behavior, and worker injuries in the Chinese manufacturing industry. Saf. Sci. 2015;78:173–178. 10.1016/j.ssci.2015.04.023 Eskandari T Mirzaei M Mohammadfam I : Fire and explosion risk analysis using bow-tie method and fuzzy-Bayesian network in process industries. 2020. Eskandari T Mohammadfam I Mirzaei Aliabadi M : Dynamic Safety Analysis CNG Stations Using Fault Tree Approach and Bayesian Network. Journal of Health and Safety at Work. 2019;9(4):250–264. Kodji MK Giorgis-Allemand L Laumon B : The Role of Noise Annoyance and Noise Sensitivity in the Effect of Aircraft Noise on Self-Reported Health: The Results of the DEBATS Longitudinal Study in France. Noise Health. 2023;25(117):92–103. 37203126 10.4103/nah.nah_1_23 PMC10301904 Andringa TC Lanser JJ : How pleasant sounds promote and annoying sounds impede health: a cognitive approach. Int. J. Environ. Res. Public Health. 2013;10(4):1439–1461. 23567255 10.3390/ijerph10041439 PMC3709327 Hahad O Kuntic M Al-Kindi S : Noise and mental health: evidence, mechanisms, and consequences. J. Expo. Sci. Environ. Epidemiol. 2025;35(1):16–23. 38279032 10.1038/s41370-024-00642-5 PMC11876073 Abbasi M : The Effects of Hospital Noise Pollution and Noise Sensitivity on Patient’s Acoustic Comfort, Noise Annoyance, and Intention to Leave.[Dataset]. figshare. 2025. 10.5256/f1000research.196838.r463664 Reviewer response for version 2 Kumar Sanjay 1 Referee https://orcid.org/0000-0001-8101-097X University of Nebraska-Lincoln, Omaha, USA

Competing interests: No competing interests were disclosed.

 27 3 2026 Copyright: © 2026 Kumar S 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve

The revised manuscript titled “The Effects of Hospital Noise and Noise Sensitivity on Patient’s Comfort, Annoyance, and Intention to Leave” demonstrates substantial improvement and now meets the standards for indexing. The authors have addressed prior reviewer comments thoroughly, particularly by clarifying the study design, strengthening the theoretical framework, and providing more robust statistical analyses. The discussion has been significantly enhanced, offering clearer interpretation of the findings and stronger connections to existing literature, which reinforces the study’s relevance to healthcare environment management and patient-centered care. Additionally, the revisions improve the manuscript’s coherence, methodological transparency, and practical implications, especially regarding noise mitigation strategies in hospital settings. Overall, the study offers meaningful and timely insights into how environmental factors influence patient experience and behavioral intentions, making it a valuable contribution to the field. I therefore recommend acceptance in its current form.

Is the work clearly and accurately presented and does it cite the current literature?

Yes

If applicable, is the statistical analysis and its interpretation appropriate?

Yes

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Yes

Are the conclusions drawn adequately supported by the results?

Yes

Are sufficient details of methods and analysis provided to allow replication by others?

Yes

Reviewer Expertise:

Hospital noise, Environmental Noise, Metamaterials, Urban Built environment

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

 10.5256/f1000research.196838.r463665 Reviewer response for version 2 De Salvio Domenico 1 Referee University of Bologna, Bologna, Italy

Competing interests: No competing interests were disclosed.

 25 3 2026 Copyright: © 2026 De Salvio D 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

The paper investigates an important topic in healthcare environments: the relationship among noise pollution, noise sensitivity, and patient annoyance. The study is interesting, well written, and well described. The reproducibility of the method is quite ensured. While the idea of using a Bayesian Network is helpful for linking objective and subjective data, several methodological issues arise. Following my points:

- Sampling only 15 minutes to representatively describe 3 hours of noise is strongly limiting. The noise environment in hospitals is dynamic, driven by various sources with distinct spectral and temporal features that deeply affect noise annoyance assessment. Also, impulsive noises can be highly intrusive.

- I don't understand why the authors stated that the monitoring they performed doesn't allow a statistical analysis of percentile levels. 

- Noise monitoring is the only objective data underlying the analysis of the context. No information about the interval time or the features extracted is provided. 

- L_den is scarcely representative of the annoyance in such a dynamic context. All impulsive events are squeezed into a single number with no information about the nature of the disturbing noise. This metric is usually used in the context of transportation noise, where noise spectra are well-defined.

- Also, low frequencies play a key-role in annoyance. No mention has been made about it.

- In section 2.5, it is written, "This framework explicitly delineates cause-effect relationships, enabling the visualization and quantification of how changes in one variable propagate through the system to influence others". This is misleading because a cause-and-effect relationship implies causality, and a single Bayesian Network cannot assess causality within a given context. It provides conditional probabilities, not causalities.

- Continue variables like VAS score are forced to be categorized in "Low", "Moderate", and "High". This is fundamental for the use of CPT, but it strongly squeezes the data's variance. Thus, the study's sensitivity in describing patients' annoyance and intention to leave is deeply affected.

Since this is the second round of the review, my review focuses on the extension of Section 5 - Limitations. Here, all the concerns listed before should be addressed and described in the paper. While this section analyzes the limits of the work, it should also describe potential ways to deepen it. Particularly, the use of advanced methods to analyze noise data is carefully described in:

Refer to reference no. 1-4

I think that adding a broader, more conscious section on the study's limitations and highlighting its potential for future development can make the reader more aware of the work's importance. Also, providing appropriate literature that enables in-depth physiological and technical analysis for scholars is fundamental. 

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Partly

Are all the source data underlying the results available to ensure full reproducibility?

No source data required

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Yes

Reviewer Expertise:

Noise monitoring and analysis

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

References : Relating clustered noise data to hospital patient satisfaction. The Journal of the Acoustical Society of America, 154(2), 1239–1247 .2023; : Clustering analysis of noise sources in healthcare facilities. Applied Acoustics, 214, 109660 .2023; : A study on the conversion relationship of noise perceived annoyance and psychoacoustic annoyance—a case of substation noise. Journal of Low Frequency Noise, Vibration and Active Control, 41(2), 810–818 .2022; : Noise assessment within healthcare facilities: annoyance and intrusiveness of sound sources. — .2025; 10.5256/f1000research.196838.r463111 Reviewer response for version 2 Patrick Azodo Adinife 1 Referee Department of Mechanical Engineering, Federal University Wukari, Wukari, Nigeria

Competing interests: No competing interests were disclosed.

 11 3 2026 Copyright: © 2026 Patrick AA 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

The revised title appears to lose some construct precision by replacing “acoustic comfort” with “comfort” and “noise annoyance” with “annoyance.” Since the manuscript consistently examines these outcomes specifically as acoustic comfort and noise annoyance, I suggest revising the title to: “The Effects of Hospital Noise and Noise Sensitivity on Patients’ Acoustic Comfort, Noise Annoyance, and Intention to Leave.” This would preserve clarity while remaining more closely aligned with the constructs analyzed in the paper.

Table 1 still appears inconsistent with the explanation provided. In the current table, the category frequencies for Gender (112 + 114) and Marital Status (61 + 165) sum to 226, yet the corresponding Total rows are still reported as 205, while Hospital Ward is reported as 226. Since the manuscript also states that a total of 226 patients participated and that all analyses were conducted on the full sample, these Total entries appear to be residual table errors rather than missing responses. Please revise Table 1 so that all totals are reported consistently.

Mediation-oriented wording still remains in the revised manuscript in places where no formal mediation analysis was conducted. For example, in the Figure 1 caption, the manuscript states that Noise Annoyance and Acoustic Comfort are “positioned as intermediate nodes in the probabilistic pathways from Lden and Sensitivity to Intention to Leave, consistent with a mediating structure.” In the Discussion, it further states that the pattern is “consistent with a mediating role for acoustic comfort,” and later describes the relationship as “a pathway suggestive of mediation.” Because mediation was not formally tested, I recommend replacing these phrases with more neutral language referring to intermediate or associated variables within the Bayesian network, rather than mediation.

The rationale for the single-hospital setting is only partly reflected in the manuscript. The revised methods section now makes clear that the study was conducted in a public hospital in Saveh, Iran, which may be acceptable from a lenient review perspective. However, the fuller justification for why only one hospital was included is not explicitly inserted into the methods text itself.

Minor copyediting issues remain. The manuscript still contains at least one obvious typographical error, such as “Forth” instead of “Fourth” in the limitations section, indicating that copyediting was not fully completed.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Yes

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Yes

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Industrial Engineering; Ergonomics; Health, safety,  and Environmental management

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

 10.5256/f1000research.196838.r463112 Reviewer response for version 2 Secchi Simone 1 Referee https://orcid.org/0000-0002-8539-1578 Nannipieri Elisa 1 Co-referee Universita degli Studi di Firenze (Ringgold ID: 9300), Florence, Tuscany, Italy

Competing interests: No competing interests were disclosed.

 2 3 2026 Copyright: © 2026 Nannipieri E and Secchi S 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve

The paper has been significantly improved and is now suitable for publication

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

I cannot comment. A qualified statistician is required.

Are all the source data underlying the results available to ensure full reproducibility?

Partly

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Yes

Are sufficient details of methods and analysis provided to allow replication by others?

Yes

Reviewer Expertise:

Environmental and room acoustics

We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.

References : Analysis of the Acoustic Comfort in Hospital: The Case of Maternity Rooms. Buildings .2022;12(8) : 10.3390/buildings12081117 10.3390/buildings12081117 10.5256/f1000research.185125.r452548 Reviewer response for version 1 Secchi Simone 1 Referee https://orcid.org/0000-0002-8539-1578 Nannipieri Elisa 1 Co-referee Universita degli Studi di Firenze, Florence, Tuscany, Italy

Competing interests: No competing interests were disclosed.

 9 2 2026 Copyright: © 2026 Secchi S and Nannipieri E 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

The study shows that noise pollution in Iranian hospitals exceeds WHO standards. High noise levels and individual sensitivity reduce acoustic comfort, increasing discomfort and the desire to leave hospital early, thus hindering patient recovery.

The article primarily uses the Bayesian Networks (BNs) model combined with delta-p sensitivity analysis to examine the relationships between the variables studied.

to visualise and quantify how changes in one variable affect others.

The study is interesting, but I think the following items should be clarified.

In paragraph  2.3, it is unclear whether the measured level is an equivalent sound pressure level (i.e. an integration over the measurement period) or single instantaneous measurements.

In the same paragraph please specify the model of the B&K sound level meter

In my opinion, Table 2 is unclear. In the table, quantitative variables (such as Lden) are listed alongside qualitative variables (noise sensitivity, noise annoyance, etc.).

I understand that 57.95 dBA is the Lden value, but I do not understand what 42.57 represents for noise sensitivity, and the same applies to the other qualitative variables.

In Figure 1 and in the following please repalce Ldn with Lden

Please control the percentages written in tables since the sums of different percentages in many cases is not 100%.

In general, the correlation between the intention to leave and other environmental comfort variables other than noise should also be studied.

Furthermore, the equivalent level alone is not a very meaningful parameter for assessing noise disturbance, since it would also have been useful to evaluate statistical levels such as L10 and L90.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

I cannot comment. A qualified statistician is required.

Are all the source data underlying the results available to ensure full reproducibility?

Partly

Is the study design appropriate and is the work technically sound?

Partly

Are the conclusions drawn adequately supported by the results?

Yes

Are sufficient details of methods and analysis provided to allow replication by others?

Yes

Reviewer Expertise:

Environmental and room acoustics

We confirm that we have read this submission and believe that we have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however we have significant reservations, as outlined above.

Abbasi Milad saveh university of medical sciences, Iran

Competing interests: There are no competing intrestes

 18 2 2026

Response to Reviewer: Dr. Simone Secchi

Dear Dr. Secchi,

Thank you very much for your thoughtful and detailed review of our manuscript. Your comments have been invaluable in strengthening the clarity, methodological transparency, and overall quality of our work. Below, we provide a point-by-point response to each of your observations and concerns. All revisions have been made using track changes in the revised manuscript.

Comment: In paragraph 2.3, it is unclear whether the measured level is an equivalent sound pressure level (i.e. an integration over the measurement period) or single instantaneous measurements.

Response to Comment 1:

Thank you for this important comment. We confirm that noise exposure was measured as equivalent continuous sound pressure levels (Leq) over consecutive 15-minute intervals, not as instantaneous spot measurements. The text has been revised accordingly to reflect this clarification.

Comment: In the same paragraph please specify the model of the B&K sound level meter

Response to Comment 2:

Thank you for this technical comment. We have used B&K Sound Level Meter Analyzer, model 2250.

Comment: In my opinion, Table 2 is unclear. In the table, quantitative variables (such as Lden) are listed alongside qualitative variables (noise sensitivity, noise annoyance, etc.).

Response to Comment 3:

Thank you for this important observation. You are correct that the original Table 2 presentation was unclear regarding the distinction between the original quantitative measurements and the categorized variables used in Bayesian Network modeling.

We wish to clarify that all mentioned variables—including Lden, noise sensitivity, noise annoyance, acoustic comfort, and intention to leave—were originally measured as quantitative variables. The means and standard deviations reported in Table 2 represent the total measured mean and total measured SD for these continuous distributions across the entire sample. However, to run the Bayesian Network model and perform delta-p sensitivity analysis, these continuous variables were converted into ordinal qualitative variables (low, moderate, high) through dichotomization. This conversion is essential for instantiating discrete evidence scenarios (e.g., setting "Lden = High 100%") within the probabilistic graphical framework.

Due to journal formatting restrictions on the number of tables, we were obligated to present both the continuous descriptive statistics and the categorical frequency distributions within a single table. To improve clarity and prevent misinterpretation, we have now revised the column headers in Table 2. The mean and standard deviation columns are now clearly labeled as:

Total Measured Mean

Total Measured SD

This revision explicitly indicates that these values correspond to the original continuous measurements for each variable across all participants, while the frequency columns present the categorized distributions used in the Bayesian Network analysis.

Comment: I understand that 57.95 dBA is the Lden value, but I do not understand what 42.57 represents for noise sensitivity, and the same applies to the other qualitative variables.

Response to Comment 4:

Thank you for this important observation. You are correct that the presentation of mean values in Table 2 requires clearer explanation. We confirm that all variables—including Lden, noise sensitivity, noise annoyance, acoustic comfort, and intention to leave—were originally measured as continuous quantitative variables. The mean and SD values reported in Table 2 represent the total measured mean and **total measured standard deviation for these continuous distributions across the entire sample, as follows:

Lden: 57.95 dB (±6.61)

Noise Sensitivity: 42.57 (±18.16)

Noise Annoyance: 38.46 (±17.07)

Acoustic Comfort: 4.73 (±1.45)

Intention to Leave: 26.06 (±11.78)

To improve clarity, we have revised Table 2 by:

we Adding a clear explanatory footnote: “Mean and SD values represent the original continuous measurements for each variable prior to categorization.

Comment: In Figure 1 and in the following please repalce Ldn with Lden

Response to Comment 5: Thank you for pointing this out. We have corrected the notation and replaced “Ldn” with the correct “Lden” throughout Figure 1 and all subsequent occurrences in the manuscript (including figure captions, tables, and the main text) to ensure consistency.

Comment: Please control the percentages written in tables since the sums of different percentages in many cases is not 100%.

Response to Comment 6: Thank you for this careful observation. The percentages reported in the tables were originally presented with three decimal places, but the values had not been consistently rounded, which caused some rows not to sum exactly to 100% due to decimal precision effects.

In the revised manuscript, the values were rounded correctly and consistently according to the original Bayesian Network outputs. As a result, the reported percentages now sum to 100% (1.000) in each probability distribution. This update only reflects proper rounding of the existing BN results, and no model outputs or analyses were changed.

Comment: In general, the correlation between the intention to leave and other environmental comfort variables other than noise should also be studied.

Response to Comment 7:

Thank you for this thoughtful comment. We acknowledge that **intention to leave** is a complex, multi-dimensional construct that may be influenced by a range of environmental, clinical, and psychological factors beyond noise—including thermal comfort, lighting, indoor air quality, spatial layout, privacy, and overall satisfaction with the physical environment. The reviewer correctly notes that these variables were not included in our Bayesian Network model, and we agree this is an important consideration.

However, we wish to clarify the scope and objective of the present study. This investigation was specifically designed to examine the isolated and interactive effects of acoustic environmental exposure (noise) and individual noise sensitivity on patient-reported outcomes. Our model was deliberately constructed to test theoretically grounded hypotheses regarding the direct and joint effects of these two primary variables, not to develop a comprehensive predictive model of all possible determinants of intention to leave.

We have now explicitly acknowledged this limitation in Section 4.5:

“Forth, this study focused exclusively on acoustic environmental factors and did not measure other physical environmental parameters (e.g., thermal comfort, lighting, air quality, privacy) that may also influence patients' intention to leave.”

Comment: Furthermore, the equivalent level alone is not a very meaningful parameter for assessing noise disturbance, since it would also have been useful to evaluate statistical levels such as L10 and L90.

Response to Comment 8:

Thank you for this important methodological observation. We agree that Lden alone does not fully capture the temporal and spectral characteristics of the hospital acoustic environment that may contribute to patient disturbance. Statistical percentiles such as L10 and L90 provide valuable complementary information about noise variability and are widely used in environmental acoustics and psychoacoustics research.

However, the primary objective of this study was to examine the relationship between overall noise exposure burden and patient-centered outcomes using a standardized, policy-relevant metric (Lden) that enables direct comparison with WHO guidelines and international studies. Lden is the most commonly reported metric in the hospital noise literature, facilitating cross-study comparability.

Furthermore, due to our measurement protocol—which assessed noise exposure over consecutive 15-minute intervals within each three-hour block rather than through continuous 24-hour monitoring—we were not able to reliably compute statistical percentiles such as L10 and L90, as these require uninterrupted high-resolution time series data.

 10.5256/f1000research.185125.r440738 Reviewer response for version 1 Patrick Azodo Adinife 1 Referee Department of Mechanical Engineering, Federal University Wukari, Wukari, Nigeria

Competing interests: No competing interests were disclosed.

 2 1 2026 Copyright: © 2026 Patrick AA 2026 This is an open access peer review report distributed under the terms of the Creative Commons Attribution Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. recommendation approve-with-reservations

Reviewer Comments to the Author

The title is clear and informative but relatively long; consider whether it can be slightly shortened without loss of meaning.

The abstract would benefit from explicitly stating the study design (cross-sectional) and the main analytical approach (Bayesian Network modeling).

In the abstract, percentage changes are reported without clarifying how variables were categorized (low/moderate/high); brief clarification would improve readability.

The introduction is comprehensive but lengthy; the final paragraphs could be streamlined to reduce repetition.

The novelty of applying Bayesian Networks could be highlighted more explicitly as part of the research gap.

The concept of “intention to leave” would benefit from clearer distinction from medical discharge or staff turnover.

There is an inconsistency in reported sample size (205 vs. 226) across sections and tables that should be corrected.

The rationale for selecting a single hospital setting should be briefly justified.

The choice of Lden as the primary noise metric should be justified relative to other commonly used indicators (e.g., LAeq).

It should be clarified whether noise measurements included weekdays, weekends, or both.

The criteria used to categorize continuous variables into low, moderate, and high levels should be explicitly stated.

The limitations of using a single-item VAS for intention to leave should be briefly acknowledged.

The Bayesian Network structure development process (expert-driven, data-driven, or hybrid) should be clarified.

A short, intuitive explanation of delta-p sensitivity analysis would improve accessibility for non-technical readers.

The Results section contains some repetition between text, tables, and figures; selective summarization would improve clarity.

Several tables are dense; clearer captions or brief interpretive notes would help guide the reader.

Figure captions should be expanded to allow standalone interpretation.

Some sentences in the Discussion are overly long and could be split for clarity.

The mediating role of noise annoyance and acoustic comfort should be framed more cautiously, as mediation was not formally tested.

Policy and design implications could be more clearly separated from theoretical discussion.

The limitations section could be strengthened by acknowledging potential clinical confounders (e.g., pain, medication).

The conclusion is appropriate but could be more concise and avoid repetition from the Discussion.

Minor grammatical, typographical, and formatting inconsistencies are present and should be corrected through copyediting.

Is the work clearly and accurately presented and does it cite the current literature?

Partly

If applicable, is the statistical analysis and its interpretation appropriate?

Yes

Are all the source data underlying the results available to ensure full reproducibility?

Yes

Is the study design appropriate and is the work technically sound?

Yes

Are the conclusions drawn adequately supported by the results?

Partly

Are sufficient details of methods and analysis provided to allow replication by others?

Partly

Reviewer Expertise:

Industrial Engineering; Ergonomics; Health, safety,  and Environmental management

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.

Abbasi Milad saveh university of medical sciences, Iran

Competing interests: There is no competing intrests

 18 2 2026

Response to Reviewer: Dr. Azodo Adinife Patrick

Dear Dr. Patrick,

Thank you very much for your thoughtful and detailed review of our manuscript. Your comments have been invaluable in strengthening the clarity, methodological transparency, and overall quality of our work. Below, we provide a point-by-point response to each of your observations and concerns. All revisions have been made using track changes in the revised manuscript.

Comment: The title is clear and informative but relatively long; consider whether it can be slightly shortened without loss of meaning.

Response to Comment 1: The Effects of Hospital Noise and Noise Sensitivity on Patient`s Comfort, Annoyance, and Intention to Leave

Comment: The abstract would benefit from explicitly stating the study design (cross-sectional) and the main analytical approach (Bayesian Network modeling).

Response to Comment 2: Thank you for this important methodological observation.

Comment: In the abstract, percentage changes are reported without clarifying how variables were categorized (low/moderate/high); brief clarification would improve readability.

Response to Comment 3: Thank you for this helpful comment. We agree that reporting percentage changes in the abstract without briefly indicating how “low/moderate/high” categories were defined may reduce readability. In the revised abstract, we have added a short clarification stating that continuous variables were categorized using percentile-based thresholds: low (P75). This concise note improves interpretability while keeping the abstract within the word limit.

Comment: The introduction is comprehensive but lengthy; the final paragraphs could be streamlined to reduce repetition.

Response to Comment 4: The introduction was revise according to this comment

Comment: The novelty of applying Bayesian Networks could be highlighted more explicitly as part of the research gap.

Response to Comment 5: The introduction was revise according to this comment

Comment: There is an inconsistency in reported sample size (205 vs. 226) across sections and tables that should be corrected.

Response to Comment 6: Thank you for this comment. I clarify the sample size reporting as follows:

“The minimum required sample size was determined using G*Power 3.1 software, assuming a medium effect size (0.3), α = 0.05, and a statistical power of 0.95. Based on these parameters, a minimum sample of 178 participants was deemed necessary. To account for potential attrition, this number was increased by 15%, yielding a final sample of 205 participants —the minimum acceptable number for sufficient statistical power and adequate generalizability. During data collection, a total of 226 eligible patients were recruited and participated voluntarily. This larger achieved sample strengthens the generalizability of the findings and is fully consistent with the sampling strategy, as no upper limit was imposed on participation.”

Importantly, the Methods section (Section 2.2) clearly states that 205 was the target sample size and that this number was increased from the minimum required 178 to account for attrition. The text does not state that 205 was a fixed maximum, nor does it preclude the inclusion of additional participants. The achieved sample of 226 is therefore fully consistent with the stated sampling strategy. All analyses were conducted on the full sample of 226 participants. Tables and frequency distributions have been verified and corrected to consistently reflect N = 226 throughout the manuscript.

Comment: The rationale for selecting a single hospital setting should be briefly justified.

Response to Comment 7: Thank you for this comment This study was conducted at a public hospital in Saveh, Iran. Although Saveh has three hospitals, two are affiliated with the private sector and, due to their internal policies, were unable to participate in this research. Accordingly, data collection was necessarily limited to the single hospital that granted approval. For ethical reasons, the names of the institutions that declined participation are not disclosed.

Comment: The choice of Lden as the primary noise metric should be justified relative to other commonly used indicators (e.g., LAeq).

Response to Comment 8: Thank you for your comment. We selected Lden (day-evening-night weighted equivalent level) as our primary noise metric because it applies empirically derived +5 dB (evening) and +10 dB (night) penalties that reflect increased human sensitivity during rest periods—a critical consideration in hospital settings where sleep disruption directly impacts patient recovery. Unlike unweighted LAeq, Lden is the standard indicator recommended by the World Health Organization and the European Environmental Noise Directive for health-based noise impact assessment, ensuring policy relevance and comparability across studies. Furthermore, Lden is explicitly calibrated to exposure-response relationships for annoyance and sleep disturbance, our primary outcomes of interest. Thus, Lden was not a default choice but a deliberate selection aligned with our research questions, regulatory context, and the unique temporal sensitivities of the hospital acoustic environment.

Comment: It should be clarified whether noise measurements included weekdays, weekends, or both.

Response to comment 9: Thank you for this important methodological clarification. We confirm that the noise measurements included weekdays and weekends.

Measurements were conducted over a 12-week period from November 2024 to February 2025, with a full 24-hour measurement cycle per week in each participating ward. These cycles were systematically rotated across weekdays to ensure that the acoustic conditions were representative of both weekdays (Saturday to Thursday in Iran) and weekends (Friday). This approach was necessary because hospital activity patterns—including staffing levels, visitor attendance, non-emergency procedures, and patient movements—differ significantly between weekdays and weekends, and measurement exclusively on weekdays would overestimate the true level of noise exposure.

Comment: The criteria used to categorize continuous variables into low, moderate, and high levels should be explicitly stated.

Response to comment 10: Thank you for this important comment. We agree that the criteria used for categorizing continuous variables should be clearly reported. In the revised manuscript, we now explicitly state that continuous variables were discretized using sample-based percentile thresholds. Specifically, values below the 25th percentile (P25) were classified as low, values between the 25th and 75th percentiles (P25–P75) were classified as moderate, and values above the 75th percentile (P75) were classified as high. These thresholds were calculated from the distribution of the study data and were applied consistently across all variables used in the Bayesian Network analysis.

Comment: The limitations of using a single-item VAS for intention to leave should be briefly acknowledged.

Response to comment 11: While thanking you for this valuable comment, this was added to the limitations of the study.

Comment: The Bayesian Network structure development process (expert-driven, data-driven, or hybrid) should be clarified.

Response to comment 12: Thank you for this comment. We agree that the Bayesian Network (BN) structure development process should be clarified. In the revised manuscript, we explain that the BN topology (nodes and directed links) was constructed based on the study structure and the variables/domains captured by the questionnaires used in this study. The purpose of using the BN was not to “predict” the structure, but to quantify the interdependencies among nodes and to evaluate how changes in the state of one or more nodes (evidence instantiation) update the probability distributions of other variables (probabilistic inference / evidence propagation). Accordingly, the conditional probability tables (CPTs) were estimated from the observed study data using GeNIe.

Comment: A short, intuitive explanation of delta-p sensitivity analysis would improve accessibility for non-technical readers.

Response to comment 13: Thank you for this helpful suggestion. We agree that a short, intuitive explanation of Δp sensitivity analysis would improve accessibility for non-technical readers. In the revised manuscript, we added a brief clarification explaining that Δp represents the change in probability of each state of a target node (low/moderate/high) after setting evidence on an input node to a specific state (low/moderate/high). We also clarified that Δp is computed as updated probability minus baseline probability for each target state, and that larger |Δp| indicates a stronger influence, while the sign indicates an increase or decrease in that target-state probability.

Comment: The Results section contains some repetition between text, tables, and figures; selective summarization would improve clarity.

Response to comment 14: Thank you for your valuable comment. We have deliberately summarized the descriptive findings to maintain conciseness. Regarding the Bayesian analysis, given that not all readers may be fully familiar with this methodological approach and its interpretation, we aimed to present only the essential interpretations—including one detailed example—while avoiding excessive elaboration. A comprehensive interpretation of all results would have substantially increased the length of the manuscript without proportional benefit to clarity.

Comment: Several tables are dense; clearer captions or brief interpretive notes would help guide the reader.

Response to Comment 15: Thank you for this excellent practical suggestion. The brief interpretive notes were added.

Comment: Figure captions should be expanded to allow standalone interpretation.

Response to Comment 16: Thank you for this excellent practical suggestion. The brief interpretive captions were added.

Comment: Some sentences in the Discussion are overly long and could be split for clarity.

Response to Comment 17: Thank you for this important observation regarding readability. We have conducted a careful, line-by-line review of the entire Discussion section and have revised according to this comment.

Comment: The mediating role of noise annoyance and acoustic comfort should be framed more cautiously, as mediation was not formally tested.

Response to Comment 18:

Thank you for this methodologically rigorous and important observation. You are absolutely correct. While our Bayesian Network model  implies mediation by positioning noise annoyance and acoustic comfort as intermediate nodes between Lden/sensitivity and intention to leave, we did not conduct formal mediation analysis (e.g., bootstrap tests of indirect effects, Bayesian mediation with credible intervals, or counterfactual-based approaches). Our use of terms such as "mediator," "mediating role," and "mediating factor" throughout the Discussion and Conclusion overstates the evidentiary support for mediation given our analytical approach.

We have revised the manuscript to replace all claims of mediation with more cautious language that accurately reflects what our Bayesian Network demonstrates: probabilistic dependence structures consistent with mediation, not confirmed causal pathways.

Comment: Policy and design implications could be more clearly separated from theoretical discussion.

Response to Comment 19:

Thank you for your valuable comment. In accordance with your request, all sentences and phrases related to policy and design implications that were scattered throughout the Discussion section have been extracted and consolidated into one dedicated and focused subsection titled "6. Policy and Design Implications ".

Comment: The limitations section could be strengthened by acknowledging potential clinical confounders (e.g., pain, medication).

Response to Comment 20:

Thank you for this important methodological observation. These confounders were mentioned as follows:

“Third, although ward type was accounted for, other potentially influential clinical factors—such as pain levels, medication use, or individual health conditions—were not controlled. These unmeasured confounders may have influenced patients' noise perceptions and tolerance.”

Comment: The conclusion is appropriate but could be more concise and avoid repetition from the Discussion.

Response to Comment 21:

Thank you for this important observation. The conclusion was revised as follows:

Hospital soundscapes are active components of the care environment, not merely ambient features. This study demonstrates that noise exposure and individual noise sensitivity synergistically increase annoyance, reduce acoustic comfort, and elevate patients' intention to leave. The findings position acoustic comfort and noise annoyance as intermediate variables in these pathways. Addressing noise pollution—particularly through targeted interventions for noise-sensitive individuals and improved acoustic design—is therefore an essential dimension of quality healthcare delivery.

Comment: Minor grammatical, typographical, and formatting inconsistencies are present and should be corrected through copyediting.

Response to Comment 21:

Thank you for this important observation. We have conducted a thorough, line-by-line copyediting review of the entire manuscript. All identified errors have been corrected.