Study population
This descriptive and cross-sectional study was carried out on 209 nurses in 2021 in Imam Reza hospital, Tehran, Iran. Required data was gathered in a four months period from April 10, 2021, to August 10, 2021, in Imam Reza hospital, Tehran, Iran. After taking preliminary measures such as receiving the ethics code from the research ethics committee of Aja University of Medical Sciences (IR.AJAUMS.REC1400.060) and taking permission from the hospital management, study executive steps were started. To achieve data related to study goals, questionnaires including demographic information, noise sensitivity, noise annoyance scale, and quality of patient care were used. Concerning that this study was conducted during the COVID-19 pandemic, to comply with preventive measures, the questionnaires were presented to the nurses based on their request as an online link or printed questionnaires. The questionnaires link was prepared and delivered to the hospital management. They had access to the nurses' contact information and they were requested to send the link to occupied nurse. In the online questionnaires, the goals and way of participating in the study was expressed and nurses were invited to complete questionnaires voluntarily. Finally, nurses with less than one year of work experience, incompletely completed questionnaires, nurses working in other hospitals at the same time, and individuals with restrictive diseases were excluded from the study.The online questionnaire was provided and sent to all nurses occupied in the studied hospital. To attract public participation, also printed questionnaires were provided and delivered in person to those who did not complete the online questionnaire.
Questionnaires
The demographic questionnaire was developed to collect information such as age, experience, gender, marital status, education, and working department. Following the properties of mentioned questionnaires are expressed.
Weinstein noise sensitivity scale
To determine noise sensitivity, Weinstein noise sensitivity scale was used. This scale comprising items related to attitude and affective reactions to noise in the various situations. It is consisted of 21 items with 6-points Likert scale ranging from “agree strongly” to “disagree strongly”. The total score ranging from 0 to 105 can be categorized into three groups of lowly sensitive (scores less than 25th percentile), moderately sensitive (scores ranging from 25th percentile to 75th percentile), and highly sensitive (scores more than 75th percentile). Alimohammadi et al investigated and confirmed the reliability and validity of the Persian translation of this scale [15].
Noise annoyance scale
Moreover, A 100-point graphical rating scale with two verbally labelled poles ranging from “not at all annoying” and “extremely annoying” was used to assess nurses noise annoyance [16]. They were asked to mark their annoyance perceived due to exposure to hospital noise on a 100-point graphical rating scale. The total score of noise annoyance was stratified as lowly annoyed (scores less than 25th percentile), moderately annoyed (scores ranging from 25th percentile to 75th percentile), and highly annoyed (scores more than 75th percentile).
Quality patient care scale
Quality patient care scale was used to assess quality of patient care. This tool consists of 65 questions in three dimensions including psychosocial (28 items), communicational (13 items) and physical (24 items). The items were answered based on a 4-point Likert scale ranging from 1 to 4 as rarely (1), sometimes (2), often (Score 3), and Always (4). The score specified to the items that were not applicable was not applicable (0). The total score of quality of patient care ranged from 65 to 260 and the higher the score the higher the quality of patient care. Scores less than130 were considered as undesirable, 130- 195 as partly undesirable, and 195-260 as desirable. Regarding the psychosocial aspect, scores less than 56 were considered as undesirable, 56-84 as partly undesirable, and 84-112 as desirable. Regarding the communicational aspect, scores less than 26 were considered as undesirable, 26-39 as partly undesirable, and 39-52 as desirable. Moreover, physical aspect score was categorized as undesirable (scores less than 48), partly undesirable (scores from 48 to 72), and desirable (scores from 72 to 96). The reliability of this questionnaire was checked and Cronbach’s alpha coefficient of 0.94 was determined.
Noise measurement procedure
To plot the noise map, measurement points (1510 points five meters apart) were determined on the hospital building map and the noise was measured at all points in different departments. Measurement was performed in accordance with ISO 9612 using KIMO DB 300/2 sound level meter and analyzer. Measurement was started at 08:00 and 1-minute equivalent continuous sound pressure level (Leq,1 min) was measured at each point. To cover all noise level changes during the whole day, measurement was repeated at each point at two-hour intervals except for the nighttime shift from 00:00 to 08:00 which two measurements were performed as a representative of the noise level in this shift. Therefore, to determine noise level at each point, measurement repeated ten times for each point and in total 250 hours noise measurement was performed. The logarithmic mean for each point was calculated and noise map was plotted.
To determine nurses’ noise exposure, nurses who were participated in the study were asked to specify their workstations and the duration of their work at that point. Using these data and referring to the ISO 9612, the equivalent continuous sound pressure level (LAeq) was calculated by the LAeq equation. Finally, the daily personal noise exposure level (LEP,d) was calculated for each nurse. Nurses' noise exposures with values less than 50 dBA, 50 to 60 dBA, and higher than 60 dBA were considered as low exposure, moderate exposure, and high exposure, respectively.
Statistical analysis
The statistical tests were carried out using SPSS software version 24. Descriptive statistics were calculated. Then the expectation-maximization method was applied to calculate and replace the missing values.
Bayesian Networks (BNs), as probabilistic and graphical models, were introduced by Pearl [17]. In this study, GeNIe academic software version 2.3 was used to analyze Bayesian network.After drawing the BN graphical structure, a Conditional Probability Table (CPT) was obtained by the model with the Expectation-Maximization algorithm [18]. Then, delta p sensitivity analysis was applied to rank the parameters [19]. Finally,10-fold cross-validation analysis was exploited to examine the model validity. Dataset was randomly divided into ten equal folds, nine folds (9 subsamples) were applied to train the Bayesian network model, and the remaining fold (1 subsample) was used to validate the model. A sensitivity analysis also was conducted to examine the effects of the variables [20].