2.1 Study design and setting

A retrospective design was adopted for this study. Medical records of the latest periodic medical examination of workers at two steel factories in Egypt were reviewed in November 2021. The medical examination was conducted from July to September of the same year and represents the health status of the workers in that year.

2.2 Participants

The medical records of the selected factories included data of 709 workers, all of whom were men except for four workers. The inclusion criteria set for enrollment in the current study were male workers who had a job duration equal to or more than 5 years. Women workers and those with any condition considered as confounders to NIHL were excluded. Accordingly, 103 workers were excluded because of job duration less than 5 years (n = 88), previous ear infections (n = 5), history of exposure to noise during military service (n = 3), family history of hearing impairment (n = 2), and conductive hearing loss (air–bone gap > 10 dB) (n = 1). Additionally, women workers (n = 4) were excluded.

The 606 eligible workers were divided into two groups based on noise exposure: (i) workers exposed to hazardous occupational noise, necessitating actions (action level), where the workplace area A-weighted equivalent noise level was equal to or more than 85 dB and (ii) unexposed workers employed at non-manufacturing departments at the same factories who were not exposed to hazardous occupational noise. Exposed workers were further categorized according to audiometry results into workers with NIHL or free from NIHL.

2.3 Power analysis

A power analysis was conducted (using the Open-Epi online calculator Version 3.3a, OpenEpi, Atlanta, GA, USA). It showed that enrolling 396 noise-exposed workers and 210 unexposed workers is capable of detecting the least difference in prevalence of NIHL of 17% between both groups [14], with a prevalence ratio of 1.5 at a power of 98.2% and confidence level of 0.95 (α = 0.05).

2.4 Data collection

A transfer sheet was designed to retrieve relevant data from the records. The sheet included the following:

2.4.1 Sociodemographic, medical, and occupational characteristics

Workers’ medical records were reviewed for sociodemographic data (such as age, residence, highest educational attainment, marital status, and smoking status); occupational data (including job duration, job nature, work schedule, department, and occupation); and medical condition (such as ear related medical conditions, and tinnitus).

2.4.2 Results of pure tone audiometry testing

In the periodic medical examination at the selected factories, a pure-tone audiometer was used to assess hearing acuity [15] for both ears at eight octave intervals: using ascending pure tones at frequencies of 0.5, 1, 2, 3, 4, and 6 kHz, and a range of intensity of − 10 to 120 dB. The mean threshold values at 0.5, 1, and 2 kHz were used to determine low-frequency hearing status, while the mean threshold values at 3, 4, and 6 kHz were used to determine high-frequency hearing status. On a certain test frequency, normal hearing was defined as binaural hearing level ≤ 25 dB [5].

In this study, according to the World Health Organization (WHO) noise exposure guidelines, hearing impairment was defined as a hearing threshold > 25 dB at any examined frequency (either monaural or binaural hearing impairment) [4, 5]. The audiometric ISO values (averages of values at 0.5, 1, 2, and 4 kHz) were used to categorize hearing impairment as follows: slight impairment (audiometric ISO value 26–40); moderate (ISO value 41–60 dB), severe (ISO value 61–80 dB), and profound (ISO value 81 dB or greater). NIHL was defined as a notch shown at 4 kHz (around 3 to 6 kHz) and threshold values at high-frequency worse than threshold values at low frequency [5].

2.4.3 Calculated percentage of hearing disability

The percentage of hearing disability was calculated for each worker according to the Egyptian occupational health standards (OHS) formula [16] as follows: first, the average hearing threshold level at 0.5, 1, and 2 kHz was calculated for each ear. Then, the percent impairment for each ear was calculated by multiplying the amount by which the above average hearing threshold level exceeds 25 dB by 100/65 up to a maximum of 100%, which is reached at 90 dB. Binaural assessment was calculated by multiplying the smaller percentage (better ear) by 5, adding this figure to the larger percentage (poor ear), and dividing the total by 6 [16]. For each worker, the calculated percentage of hearing disability was compared with that calculated using the American Academy of Otolaryngology and American Council of Otolaryngology (AAO-ACO) formula [17].

2.4.4 Workplace area A-weighted equivalent noise level measurement

Factory records were reviewed to obtain measurements of workplace area A-weighted equivalent noise level. At both factories, measurement was done using a sound pressure level noise meter (3 M™ Sound Detector SD-200), manufactured according to the International Standard Classifications [IEC 61,672–1 (2002), IEC 61,010–1 (2010), ANSI S1.4 1983 (R2006), ANSI S1.43 (R2007), CE]. The A-weighted network was selected, and the sound pressure level meter was calibrated before use. Multiple readings were recorded during the shift, then the average noise level was calculated (in dB) for each workplace area. At workplace areas with an A-weighted equivalent noise level ≥ 85 dB (such as the compressors room; turning workshop, welding workshop, tying machine, and mechanical maintenance workshop), workers were considered as noise-exposed workers. Whereas at workplace areas with an A-weighted equivalent noise level < 85 dB (such as the billet charging area, reheating furnace control room, repair workshop, billet storage yard, and quality control lab), workers were considered unexposed.

2.5 Statistical analysis

The SPSS v.22 (IBM Corp. Released 2011. IBM SPSS Statistics for Mac, Armonk, NY, USA) was used for data analysis. Descriptive statistics were used to present qualitative data (frequencies and percentages) and quantitative data (mean and standard deviation). Data analysis involved an initial comparison between noise-exposed workers and unexposed workers to identify the frequency of occupational NIHL. Then, among noise-exposed workers, a case–control approach analysis was carried out to determine factors associated with NIHL.

Among all workers in the study (n = 606), hearing thresholds (dB) at specified tested audiometry frequencies (Hz), and ISO values were presented using mean and standard deviation for both noise-exposed and unexposed workers, stratified by age into four groups (< 30, 30 to < 40, 40 to < 50, and ≥ 50 years). The prevalence of hearing impairment and NIHL were calculated. A case–control approach analysis using univariate logistic regression was conducted to compute odds ratio (OR) and associated 95%CI to quantify the probability of hearing impairment or NIHL (dependent variable) associated with noise exposure (independent variable).

Among noise-exposed workers (n = 396), the mean hearing thresholds was calculated among the four age groups and three job duration groups (< 10, 10 to < 20, and ≥ 20 years). Multiple linear regression analysis was used to determine predictors of hearing threshold at the tested frequencies. In addition, univariate logistic regression was conducted to compute the odds of NIHL (dependent variable) associated with each sociodemographic, occupational, and medical factor (independent variables). Subsequently, multivariate logistic regression was conducted to model NIHL as a function of the significant factors identified in the univariate analysis, namely age, job duration, and tinnitus, to study their independent effect. The adequacy of the model in data fitting was determined using Nagelkerke’s R2 and Hosmer and Lemeshow goodness-of-fit test. All statistical analyses were judged at a level of significance of 5% (α = 0.05).

As for workers with hearing disability (> 0%), a comparison was made between the mean percentage of hearing disability calculated using the Egyptian formula and AAO-ACO formula.