Predicting hearing loss in noise-exposed petrochemical workers: Influencing factors and a nomogram model Authors: T. Ma, J. Wu, Z. Dong, L. He, X. Jiang, X. Chen, S. Bao, J. Fan Published: 2026 ABSTRACT ObjectiveThe study aimed to develop a predictive nomogram capable of estimating individual risk within noise-exposed petrochemical workers.Study designA total of 1700 of noise-exposed petrochemical workers from a certain hospital in Lanzhou between March 1 to July 8, 2024, was surveyed for an occupational exposure and pure tone hearing tests.MethodsThe collected data were randomly divided into a training set (70%) and a testing set (30%). Logistic regression analyses were conducted on the training set to identify the influencing factors of hearing loss, and a nomogram was created to predict the risk of hearing loss.ResultsThe prevalence of hearing loss among noise-exposed petrochemical workers was 21.71%. Independent influencing factors included male sex (OR = 2.552, 95% CI: 1.444–4.510), increasing age (OR = 1.067, 95% CI: 1.040–1.095), earphone use (OR = 0.652, 95% CI: 0.457–0.932), tinnitus (OR = 2.152, 95% CI: 1.596–2.902), and a family history of hearing loss (OR = 1.420, 95% CI: 1.007–2.003). The AUC area of the nomogram in the training set was 0.743 (95% CI: 0.710–0.776).ConclusionsMale sex, older age, tinnitus, and a family history of hearing loss were associated with an increased risk of hearing loss, whereas earphone use appeared to be a protective factor in this population. The developed nomogram demonstrated good predictive accuracy and may serve as a practical tool for assessing the risk of hearing loss among noise-exposed petrochemical workers. What this study adds1. Identifies earphone use as a protective factor for hearing loss in noise-exposed petrochemical workers. 2. Develops a validated nomogram (AUC 0.743/0.783) for individual risk assessment in this group, addressing the lack of targeted prediction tools.Implications for policy and practice1. Prioritize interventions for high-risk subgroups (males, older workers, those with tinnitus/family history) via enhanced monitoring. 2. Apply the nomogram in occupational health to identify high-risk individuals for timely prevention. 3. Promote appropriate earphone use to reduce NIHL risk. What this study adds 1. Identifies earphone use as a protective factor for hearing loss in noise-exposed petrochemical workers. 2. Develops a validated nomogram (AUC 0.743/0.783) for individual risk assessment in this group, addressing the lack of targeted prediction tools. Implications for policy and practice 1. Prioritize interventions for high-risk subgroups (males, older workers, those with tinnitus/family history) via enhanced monitoring. 2. Apply the nomogram in occupational health to identify high-risk individuals for timely prevention. 3. Promote appropriate earphone use to reduce NIHL risk. Introduction Noise-induced hearing loss (NIHL) is a prevalent occupational health problem caused by long-term exposure to high-intensity workplace noise, which can cause mechanical and metabolic damage to the cochlea, leading to permanent hearing impairment [1]. The WHO notes that around 16% of hearing loss cases stem from occupational noise exposure [2]. In the petrochemical industry, intense noise arises from multiple sources like mechanical operations and chemical reactions. Studies indicate noise levels here often exceed 85 dB(A) – some even over 100 dB(A) – with workers frequently exposed for over 8 h daily [3]. Such prolonged exposure significantly raises the risk of hearing damage. Thus, studying NIHL in petrochemical workers is critical for targeted protection and prevention. Previous studies have shown that NIHL is affected by lifestyle behaviors, genetic, and environmental exposures [4]. Smoking and alcohol consumption are positively associated with NIHL [5,6]. Genetically, specific variants have been linked to increased NIHL susceptibility in the Chinese population [7]. A family history of hearing loss, as an important indirect indicator of genetic susceptibility, has been shown in studies to be closely associated with an elevated risk of NIHL [8]. Though these findings offer insights into NIHL pathogenesis, further research is needed to clarify specific contributing factors and mechanisms in high-risk groups like petrochemical workers. An increasing demand exists for a scientific, user-friendly, and practical tool to enable accurate prediction and risk assessment of NIHL among noise-exposed occupational populations. The nomogram, a reliable graphical tool based on clinical indicators, allows for the calculation of individual morbidity risk in a concise, visual format [9]. Such method has been widely applied to predict outcomes across various high-risk conditions [10]. Applying a nomogram to assess NIHL risk in petrochemical workers may enable the development of accurate predictive models and support personalized prevention and intervention strategies. This study aimed to identify factors associated with hearing loss in petrochemical workers exposed to occupational noise, based on the data collected in Lanzhou, Gansu Province, China. Furthermore, it sought to develop a nomogram to predict individual risk of hearing loss. The findings are expected to provide technical support for future applied research on NIHL prevention and control. Methods Research objectives This study focused on noise-exposed petrochemical workers from a certain hospital in Lanzhou, Gansu, China, who underwent medical examinations between 1 March and 8 July 2024. According to China's Occupational Health Standard “Classification of Occupational Disease Hazardous Operations in the Workplace”, noise-exposed operations are defined as those in which sound levels can cause hearing damage or other health hazards, specifically where the A-weighted equivalent sound level (LAeq) is ≥ 80 dB for 8 h per day or 40 h per week [11]. A total of 1763 individuals participated in the physical examinations and completed questionnaires. After excluding 63 individuals due to incomplete data or missing audiometric results, 1700 participants were included in the final analysis. Data collection Data were collected using a self-designed questionnaire informed by literature review and expert consultation. The questionnaire included demographic variables such as sex, age, body mass index (BMI), education level, marital status, and monthly income. Lifestyle-related factors included smoking, alcohol consumption, and physical activity. Other relevant variables comprised tinnitus, sleep duration, history of intense noise exposure, history of otitis media, exposure to ototoxic drugs, mechanical ear trauma, frequency of ear cleaning, use of earphones, and family history of hearing loss. Work-related factors included duration of noise exposure, occupation type, and use of hearing protection devices. Medical history, including hypertension and diabetes mellitus, was also recorded. Audiometric testing was performed in a soundproof room with background noise levels maintained below 25 dB(A), conducted by a certified technician. Pure-tone air conduction thresholds were measured at six frequencies: 500, 1000, 2000, 3000, 4000, and 6000 Hz using a diagnostic audiometer. Hearing loss was objectively diagnosed based on audiometric results. A binaural average threshold exceeding 25 dB(A) was considered indicative of hearing loss, while values below this threshold were classified as normal. Statistical analysis Statistical analyses were performed using SPSS version 26.0 and R version 4.4.1. Variables showing statistical significance in univariate analyses were subsequently entered into multivariate logistic regression models. A nomogram for predicting hearing loss was constructed using the rms package in R. A two-sided P value < 0.05 was considered significant. Results Comparison of basic characteristics between the training and testing sets A total of 1700 valid responses were collected in this survey and randomly divided into a training set and a testing set at a 3:1 ratio, resulting in 1275 participants in the training set and 425 in the testing set. A comparison of baseline characteristics between the two groups showed no significant differences (P > 0.05), indicating that the training and testing sets were well balanced and comparable. Detailed results are presented in Table 1.Table 1Comparison of basic features of training and testing sets (%).Table 1VariablesTraining set (n = 1275)Testing set (n = 425)PAge45(39,51)46(39,51)0.59SexFemale235 (18.43%)81 (19.06%)0.773Male1040 (81.57%)344 (80.94%)BMI24.42(22.44,26.64)24.45(24.42,26.65)0.89NationalityHan nationality1232 (96.63%)402 (94.59%)0.059Other nationalities43 (3.37%)23 (5.41%)Educational levelUndergraduate or above277 (21.73%)95 (22.35%)0.818Junior high school or below125 (9.80%)35 (8.24%)High school412 (32.31%)139 (32.71%)Vocational college461 (36.16%)156 (36.71%)Marital statusMarried1105 (86.67%)369 (86.82%)0.934Unmarried and others170 (13.33%)56 (13.18%)Monthly income≥6000934 (73.25%)301 (70.82%)0.330<6000341 (26.75%)124 (29.18%)SmokingNever smoked483 (37.88%)162 (38.12%)0.737Quit smoking now212 (16.63%)64 (15.06%)Occasional (≤10 cig/W)580 (45.49%)199 (46.82%)Alcohol consumptionNever535 (41.96%)174 (40.94%)0.702Quit228 (17.88%)71 (16.71%)Occasional (≤1 times/week)512 (40.16%)180 (42.35%)Physical activityAlways (≥3 times/W)318 (24.94%)115 (27.06%)0.263Occasional (1-2 times/W)864 (67.76%)288 (67.76%)Never93 (7.29%)22 (5.18%)Frequency of ear cleaningAlways (≥3 times/W)239 (18.75%)76 (17.88%)0.797Occasional (1-2 times/W)984 (77.18%)334 (78.59%)Never52 (4.08%)15 (3.53%)Earphone useNo919 (72.08%)304 (71.53%)0.827Yes356 (27.92%)121 (28.47%)Sleep duration<5h85 (6.67%)30 (7.06%)0.6495-6h324 (25.41%)113 (26.59%)6-7h521 (40.86%)179 (42.12%)7-8h282 (22.12%)89 (20.94%)>8h63 (4.94%)14 (3.29%)TinnitusNo857 (67.22%)289 (68.00%)0.765Yes418 (32.78%)136 (32.00%)History of strong noise exposureNo984 (77.18%)328 (77.18%)1.000Yes291 (22.82%)97 (22.82%)History of otitis mediaNo1137 (89.18%)381 (89.65%)0.786Yes138 (10.82%)44 (10.35%)Exposure to ototoxic drugsNo1143 (89.65%)382 (89.88%)0.890Yes132 (10.35%)43 (10.12%)Mechanical ear injuriesNo1229 (96.39%)409 (96.24%)0.881Yes46 (3.61%)16 (3.76%)Years of exposure to noise<5239 (18.75%)72 (16.94%)0.4605-10230 (18.04%)83 (19.53%)10-15223 (17.49%)64 (15.06%)>15583 (45.73%)206 (48.47%)Occupation typeOperator466 (36.55%)143 (33.65%)0.469Petrochemical622 (48.78%)212 (49.88%)Others187 (14.67%)70 (16.47%)Hearing protection equipmentNo240 (18.82)98 (23.06%)0.058Yes1035 (81.18%)327 (76.94%)Wearing hearing protection equipmentAlways (≥3 times/W)286 (22.43%)100 (23.53%)0.407Occasionally (1-2 times/W)620 (48.63%)191 (44.94%)Never369 (28.94%)134 (31.53%)Family history of hearing lossNo1026 (80.47%)330 (77.65%)0.210Yes249 (19.53%)95 (22.35%)HypertensionNo1051 (82.43%)348 (81.88%)0.797Yes224 (17.57%)77 (18.12%)Diabetes mellitusNo1225 (96.08%)404 (95.06%)0.363Yes50 (3.92%)21 (4.94%) Table 1Comparison of basic features of training and testing sets (%).Table 1VariablesTraining set (n = 1275)Testing set (n = 425)PAge45(39,51)46(39,51)0.59SexFemale235 (18.43%)81 (19.06%)0.773Male1040 (81.57%)344 (80.94%)BMI24.42(22.44,26.64)24.45(24.42,26.65)0.89NationalityHan nationality1232 (96.63%)402 (94.59%)0.059Other nationalities43 (3.37%)23 (5.41%)Educational levelUndergraduate or above277 (21.73%)95 (22.35%)0.818Junior high school or below125 (9.80%)35 (8.24%)High school412 (32.31%)139 (32.71%)Vocational college461 (36.16%)156 (36.71%)Marital statusMarried1105 (86.67%)369 (86.82%)0.934Unmarried and others170 (13.33%)56 (13.18%)Monthly income≥6000934 (73.25%)301 (70.82%)0.330<6000341 (26.75%)124 (29.18%)SmokingNever smoked483 (37.88%)162 (38.12%)0.737Quit smoking now212 (16.63%)64 (15.06%)Occasional (≤10 cig/W)580 (45.49%)199 (46.82%)Alcohol consumptionNever535 (41.96%)174 (40.94%)0.702Quit228 (17.88%)71 (16.71%)Occasional (≤1 times/week)512 (40.16%)180 (42.35%)Physical activityAlways (≥3 times/W)318 (24.94%)115 (27.06%)0.263Occasional (1-2 times/W)864 (67.76%)288 (67.76%)Never93 (7.29%)22 (5.18%)Frequency of ear cleaningAlways (≥3 times/W)239 (18.75%)76 (17.88%)0.797Occasional (1-2 times/W)984 (77.18%)334 (78.59%)Never52 (4.08%)15 (3.53%)Earphone useNo919 (72.08%)304 (71.53%)0.827Yes356 (27.92%)121 (28.47%)Sleep duration<5h85 (6.67%)30 (7.06%)0.6495-6h324 (25.41%)113 (26.59%)6-7h521 (40.86%)179 (42.12%)7-8h282 (22.12%)89 (20.94%)>8h63 (4.94%)14 (3.29%)TinnitusNo857 (67.22%)289 (68.00%)0.765Yes418 (32.78%)136 (32.00%)History of strong noise exposureNo984 (77.18%)328 (77.18%)1.000Yes291 (22.82%)97 (22.82%)History of otitis mediaNo1137 (89.18%)381 (89.65%)0.786Yes138 (10.82%)44 (10.35%)Exposure to ototoxic drugsNo1143 (89.65%)382 (89.88%)0.890Yes132 (10.35%)43 (10.12%)Mechanical ear injuriesNo1229 (96.39%)409 (96.24%)0.881Yes46 (3.61%)16 (3.76%)Years of exposure to noise<5239 (18.75%)72 (16.94%)0.4605-10230 (18.04%)83 (19.53%)10-15223 (17.49%)64 (15.06%)>15583 (45.73%)206 (48.47%)Occupation typeOperator466 (36.55%)143 (33.65%)0.469Petrochemical622 (48.78%)212 (49.88%)Others187 (14.67%)70 (16.47%)Hearing protection equipmentNo240 (18.82)98 (23.06%)0.058Yes1035 (81.18%)327 (76.94%)Wearing hearing protection equipmentAlways (≥3 times/W)286 (22.43%)100 (23.53%)0.407Occasionally (1-2 times/W)620 (48.63%)191 (44.94%)Never369 (28.94%)134 (31.53%)Family history of hearing lossNo1026 (80.47%)330 (77.65%)0.210Yes249 (19.53%)95 (22.35%)HypertensionNo1051 (82.43%)348 (81.88%)0.797Yes224 (17.57%)77 (18.12%)Diabetes mellitusNo1225 (96.08%)404 (95.06%)0.363Yes50 (3.92%)21 (4.94%) Comparison of basic features of training and testing sets (%). Comparison of basic characteristics between hearing loss and normal groups Among the 1700 petrochemical workers surveyed, 369 were diagnosed with hearing loss, yielding a prevalence rate of 21.71%. In the training set, 285 individuals had hearing loss (22.35%). A comparison between the hearing loss group and the normal hearing group revealed statistically significant differences in 13 variables (P < 0.05). In the testing set, 84 individuals had hearing loss (19.76%). A comparison between the hearing loss group and the normal hearing group revealed statistically significant differences in 12 variables (P < 0.05). Detailed results are shown in Table 2.Table 2Comparison of basic characteristics between the hearing loss group and the normal hearing group in the training and testing sets.Table 2VariablesTraining setHearing loss Normal hearing n = 285 n = 990PTesting setHearing loss Normal hearing n = 84 n = 341PAge50.0(45.0,53.0)44.0 (37.0,49.0)<0.00151.0 (47.0,54.0)45.0 (38.0,49.0)<0.001SexFemale22(7.72%)213 (21.52%)<0.0012 (2.38%)79 (23.17%)<0.001Male263(92.28%)777 (78.48%)82 (97.62%)262 (76.83%)BMI24.4(22.6,26.8)24.4 (22.4,26.6)0.56124.5 (22.9,26.5)24.4 (22.3,26.7)0.699NationalityHan nationality279 (97.89%)953 (96.26%)0.17978 (92.86%)324 (95.01%)0.424Other nationality6 (2.11%)37 (3.74%)6 (7.14%)17 (4.99%)Educational levelUndergraduate and above31 (10.88%)246 (24.85%)<0.0016 (7.14%)89 (26.10%)<0.001Junior high school and below41 (14.39%)84 (8.48%)12 (14.29%)23 (6.74%)High school127 (44.56%)285 (28.79%)38 (45.24%)101 (29.62%)Vocational college86 (30.18%)375 (37.88%)28 (33.33%)128 (37.54%)Marital statusMarried264 (92.63%)841 (84.95%)0.00178 (92.86%)291 (85.34%)0.068Unmarried and others21 (7.37%)149 (15.05%)6 (7.14%)50 (14.66%)Monthly income≤6000220 (77.19%)714 (72.12%)0.08864 (76.19%)237 (69.50%)0.227>600065 (22.81%)276(27.88%)20(23.81%)104(30.50%)SmokingNever smoked72 (25.26%)411(41.52%)<0.00119 (22.62%)143 (41.94%)0.003Quit smoking now58 (20.35%)154(15.56%)19 (22.62%)45 (13.20%)Occasionally (≤10 cig/W)155 (54.39%)425(42.93%)46 (54.76%)153 (44.87%)Alcohol consumptionNever101 (35.44%)434 (43.84%)0.01824 (28.57%)150 (43.99%)0.028Quit50 (17.54%)178 (17.98%)19(22.62%)52 (15.25%)Occasionally (≤1 times/W)134 (47.02%)378 (38.18%)41 (48.81%)139 (40.76%)Physical activityAlways (≥3 times/W)94 (32.98%)224 (22.63%)0.00232 (38.10%)83 (24.34%)0.042Occasionally (1-2 times/W)174 (61.05%)690 (69.70%)48 (57.14%)240 (70.38%)Never17 (5.96%)76 (7.68%)4 (4.76%)18 (5.28%)Frequency of ear cleaningAlways (≥3 times/W)48 (16.84%)191 (19.29%)0.39618 (21.43%)58 (17.01%)0.333Occasionally (1-2 times/W)228 (80.00%)756 (76.36%)65 (77.38%)269 (78.89%)Never9 (3.16%)43 (4.34%)1 (1.19%)14 (4.11%)Earphone useNo233 (81.75%)686 (69.29%)<0.00168 (80.95%)236 (69.21%)0.033Yes52 (18.25%)304 (30.71%)16 (19.05%)105 (30.79%)Sleep duration (h)<523 (8.07%)62 (6.26%)0.83710 (11.90%)20 (5.87%)0.1835-673 (25.61%)251 (25.35%)17 (20.24%)96 (28.15%)6-7111 (38.95%)410(41.41%)33 (39.29%)146 (42.82%)7-864 (22.46%)218 (22.02%)21 (25.00%)68 (19.94%)>814 (4.91%)49 (4.95%)3 (3.57%)11 (3.23%)TinnitusNo152 (53.33%)705 (71.21%)<0.00146 (54.76%)243 (71.26%)0.004Yes133 (46.67%)285 (28.79%)38 (45.24%)98 (28.74%)History of strong noise exposureNo208 (72.98%)776 (78.38%)0.05657 (67.86%)271 (79.47%)0.023Yes77 (27.02%)214 (21.62%)27 (32.14%)70 (20.53%)Exposure to ototoxic drugsNo253 (88.77%)890 (89.90%)0.58273 (86.90%)309 (90.62%)0.312Yes32 (11.23%)100 (10.10%)11 (13.10%)32 (9.38%)History of otitis mediaNo250 (87.72%)887(89.60%)0.36974 (88.10%)307 (90.03%)0.602Yes35 (12.28%)103(10.40%)10 (11.90%)34 (9.97%)Mechanical ear injuriesNo273 (95.79%)956(96.57%)0.53677 (91.67%)332 (97.36%)0.023Yes12 (4.21%)34 (3.43%)7 (8.33%)9 (2.64%)Years of exposure to noise<5 years31 (10.88%)208 (21.01%)<0.0019 (10.71%)63 (18.48%)0.0015-10 years40 (14.04%)190 (19.19%)7 (8.33%)76 (22.29%)10-15 years50 (17.54%)173 (17.47%)12 (14.29%)52 (15.25%)>15 years164 (57.54%)419 (42.32%)56 (66.67%)150 (43.99%)Occupation typeOperator95 (33.33%)371 (37.47%)0.16928(33.33%)115 (33.72%)0.766Petrochemical153(53.68%)469 (47.37%)40 (47.62%)172 (50.44%)Others37 (12.98%)150 (15.15%)16 (19.05%)54 (15.84%)Hearing protection equipmentNo64 (22.46%)176 (17.78%)0.07524 (28.57%)74 (21.70%)0.181Yes221 (77.54%)814(82.22%)60 (71.43%)267 (78.30%)Wearing hearing protection equipmentAlways (≥3 times/W)62 (21.75%)224 (22.63%)0.12423 (27.38%)77 (22.58%)0.099Occasionally (1-2 times/W)127 (44.56%)493 (49.80%)29 (34.52%)162 (47.51%)Never96 (33.68%)273 (27.58%)32 (38.10%)102 (29.91%)Family history of hearing lossNo212 (74.39%)814 (82.22%)0.00366 (78.57%)264 (77.42%)0.820Yes73 (25.61%)176(17.78%)18 (21.43%)77(22.58%)HypertensionNo220 (77.19%)831 (83.94%)0.00862 (73.81%)286 (83.87%)0.032Yes65 (22.81%)159 (16.06%)22 (26.19%)55 (16.13%)Diabetes mellitusNo264 (92.63%)961 (97.07%)0.00180 (95.24%)324 (95.01%)1.000Yes21 (7.37%)29(2.93%)4 (4.76%)17 (4.99%) Table 2Comparison of basic characteristics between the hearing loss group and the normal hearing group in the training and testing sets.Table 2VariablesTraining setHearing loss Normal hearing n = 285 n = 990PTesting setHearing loss Normal hearing n = 84 n = 341PAge50.0(45.0,53.0)44.0 (37.0,49.0)<0.00151.0 (47.0,54.0)45.0 (38.0,49.0)<0.001SexFemale22(7.72%)213 (21.52%)<0.0012 (2.38%)79 (23.17%)<0.001Male263(92.28%)777 (78.48%)82 (97.62%)262 (76.83%)BMI24.4(22.6,26.8)24.4 (22.4,26.6)0.56124.5 (22.9,26.5)24.4 (22.3,26.7)0.699NationalityHan nationality279 (97.89%)953 (96.26%)0.17978 (92.86%)324 (95.01%)0.424Other nationality6 (2.11%)37 (3.74%)6 (7.14%)17 (4.99%)Educational levelUndergraduate and above31 (10.88%)246 (24.85%)<0.0016 (7.14%)89 (26.10%)<0.001Junior high school and below41 (14.39%)84 (8.48%)12 (14.29%)23 (6.74%)High school127 (44.56%)285 (28.79%)38 (45.24%)101 (29.62%)Vocational college86 (30.18%)375 (37.88%)28 (33.33%)128 (37.54%)Marital statusMarried264 (92.63%)841 (84.95%)0.00178 (92.86%)291 (85.34%)0.068Unmarried and others21 (7.37%)149 (15.05%)6 (7.14%)50 (14.66%)Monthly income≤6000220 (77.19%)714 (72.12%)0.08864 (76.19%)237 (69.50%)0.227>600065 (22.81%)276(27.88%)20(23.81%)104(30.50%)SmokingNever smoked72 (25.26%)411(41.52%)<0.00119 (22.62%)143 (41.94%)0.003Quit smoking now58 (20.35%)154(15.56%)19 (22.62%)45 (13.20%)Occasionally (≤10 cig/W)155 (54.39%)425(42.93%)46 (54.76%)153 (44.87%)Alcohol consumptionNever101 (35.44%)434 (43.84%)0.01824 (28.57%)150 (43.99%)0.028Quit50 (17.54%)178 (17.98%)19(22.62%)52 (15.25%)Occasionally (≤1 times/W)134 (47.02%)378 (38.18%)41 (48.81%)139 (40.76%)Physical activityAlways (≥3 times/W)94 (32.98%)224 (22.63%)0.00232 (38.10%)83 (24.34%)0.042Occasionally (1-2 times/W)174 (61.05%)690 (69.70%)48 (57.14%)240 (70.38%)Never17 (5.96%)76 (7.68%)4 (4.76%)18 (5.28%)Frequency of ear cleaningAlways (≥3 times/W)48 (16.84%)191 (19.29%)0.39618 (21.43%)58 (17.01%)0.333Occasionally (1-2 times/W)228 (80.00%)756 (76.36%)65 (77.38%)269 (78.89%)Never9 (3.16%)43 (4.34%)1 (1.19%)14 (4.11%)Earphone useNo233 (81.75%)686 (69.29%)<0.00168 (80.95%)236 (69.21%)0.033Yes52 (18.25%)304 (30.71%)16 (19.05%)105 (30.79%)Sleep duration (h)<523 (8.07%)62 (6.26%)0.83710 (11.90%)20 (5.87%)0.1835-673 (25.61%)251 (25.35%)17 (20.24%)96 (28.15%)6-7111 (38.95%)410(41.41%)33 (39.29%)146 (42.82%)7-864 (22.46%)218 (22.02%)21 (25.00%)68 (19.94%)>814 (4.91%)49 (4.95%)3 (3.57%)11 (3.23%)TinnitusNo152 (53.33%)705 (71.21%)<0.00146 (54.76%)243 (71.26%)0.004Yes133 (46.67%)285 (28.79%)38 (45.24%)98 (28.74%)History of strong noise exposureNo208 (72.98%)776 (78.38%)0.05657 (67.86%)271 (79.47%)0.023Yes77 (27.02%)214 (21.62%)27 (32.14%)70 (20.53%)Exposure to ototoxic drugsNo253 (88.77%)890 (89.90%)0.58273 (86.90%)309 (90.62%)0.312Yes32 (11.23%)100 (10.10%)11 (13.10%)32 (9.38%)History of otitis mediaNo250 (87.72%)887(89.60%)0.36974 (88.10%)307 (90.03%)0.602Yes35 (12.28%)103(10.40%)10 (11.90%)34 (9.97%)Mechanical ear injuriesNo273 (95.79%)956(96.57%)0.53677 (91.67%)332 (97.36%)0.023Yes12 (4.21%)34 (3.43%)7 (8.33%)9 (2.64%)Years of exposure to noise<5 years31 (10.88%)208 (21.01%)<0.0019 (10.71%)63 (18.48%)0.0015-10 years40 (14.04%)190 (19.19%)7 (8.33%)76 (22.29%)10-15 years50 (17.54%)173 (17.47%)12 (14.29%)52 (15.25%)>15 years164 (57.54%)419 (42.32%)56 (66.67%)150 (43.99%)Occupation typeOperator95 (33.33%)371 (37.47%)0.16928(33.33%)115 (33.72%)0.766Petrochemical153(53.68%)469 (47.37%)40 (47.62%)172 (50.44%)Others37 (12.98%)150 (15.15%)16 (19.05%)54 (15.84%)Hearing protection equipmentNo64 (22.46%)176 (17.78%)0.07524 (28.57%)74 (21.70%)0.181Yes221 (77.54%)814(82.22%)60 (71.43%)267 (78.30%)Wearing hearing protection equipmentAlways (≥3 times/W)62 (21.75%)224 (22.63%)0.12423 (27.38%)77 (22.58%)0.099Occasionally (1-2 times/W)127 (44.56%)493 (49.80%)29 (34.52%)162 (47.51%)Never96 (33.68%)273 (27.58%)32 (38.10%)102 (29.91%)Family history of hearing lossNo212 (74.39%)814 (82.22%)0.00366 (78.57%)264 (77.42%)0.820Yes73 (25.61%)176(17.78%)18 (21.43%)77(22.58%)HypertensionNo220 (77.19%)831 (83.94%)0.00862 (73.81%)286 (83.87%)0.032Yes65 (22.81%)159 (16.06%)22 (26.19%)55 (16.13%)Diabetes mellitusNo264 (92.63%)961 (97.07%)0.00180 (95.24%)324 (95.01%)1.000Yes21 (7.37%)29(2.93%)4 (4.76%)17 (4.99%) Comparison of basic characteristics between the hearing loss group and the normal hearing group in the training and testing sets. Multivariate logistic regression analysis Thirteen variables were included in the multivariate logistic regression analysis due to their statistically significant associations with hearing loss in the training set (P < 0.05). The results indicated that male sex (OR = 2.552, 95% CI: 1.444–4.510), increasing age (OR = 1.067, 95% CI: 1.040–1.095), earphone use (OR = 0.652, 95% CI: 0.457–0.932), tinnitus (OR = 2.152, 95% CI: 1.596–2.902), and a family history of hearing loss (OR = 1.420, 95% CI: 1.007–2.003) were independent influencing factors for noise-induced hearing loss (P < 0.05). The results are shown in Table 3.Table 3Multivariate logistic regression analysis.Table 3VariablesβSEX2/ZOR (95% CI)PAge0.0650.0134.9191.067(1.04-1.095)<0.001SexFemaleReferenceMale0.9370.2913.2242.552(1.444-4.51)0.001Educational levelUndergraduate and aboveReferenceJunior high school and below0.5340.3141.71.705(0.922-3.155)0.089High school0.4480.2591.7281.565(0.942-2.602)0.084Vocational college0.310.2451.2651.364(0.843-2.206)0.206Marital statusMarriedReferenceUnmarried and others−0.4220.266−1.5880.655(0.389-1.104)0.112SmokingNever smokedReferenceQuit smoking now0.0520.2460.2111.053(0.65-1.706)0.833Occasionally0.0630.2050.3081.065(0.713-1.591)0.758AlcoholNeverReferenceconsumptionQuit−0.2560.221−1.1560.774(0.502-1.195)0.248Occasionally0.0260.1770.1461.026(0.726-1.451)0.884Physical activityAlwaysReferenceOccasionally−0.2440.166−1.4720.784(0.566-1.084)0.141Never−0.4690.32−1.4630.626(0.334-1.172)0.143Earphone useNoReferenceYes−0.4270.182−2.350.652(0.457-0.932)0.019TinnitusNoReferenceYes0.7670.1525.0282.152(1.596-2.902)<0.001Years of exposure<5 yearsReferenceto noise5-10 years0.0710.280.2521.073(0.62-1.857)0.80110-15 years0.1550.2720.5671.167(0.684-1.991)0.570>15 years0.0230.2420.0961.024(0.637-1.646)0.923Family history ofNoReferencehearing lossYes0.3510.1751.9991.42(1.007-2.003)0.046HypertensionNoReferenceYes−0.0510.185−0.2760.95(0.662-1.365)0.782Diabetes mellitusNoReferenceYes0.3040.3180.9541.355(0.726-2.527)0.340Notes: β, The regression coefficient; SE, Standard Error; OR, Odds Ratio; CI, confidence interval; P, These p-values less than 0.05 are statistically significant. Table 3Multivariate logistic regression analysis.Table 3VariablesβSEX2/ZOR (95% CI)PAge0.0650.0134.9191.067(1.04-1.095)<0.001SexFemaleReferenceMale0.9370.2913.2242.552(1.444-4.51)0.001Educational levelUndergraduate and aboveReferenceJunior high school and below0.5340.3141.71.705(0.922-3.155)0.089High school0.4480.2591.7281.565(0.942-2.602)0.084Vocational college0.310.2451.2651.364(0.843-2.206)0.206Marital statusMarriedReferenceUnmarried and others−0.4220.266−1.5880.655(0.389-1.104)0.112SmokingNever smokedReferenceQuit smoking now0.0520.2460.2111.053(0.65-1.706)0.833Occasionally0.0630.2050.3081.065(0.713-1.591)0.758AlcoholNeverReferenceconsumptionQuit−0.2560.221−1.1560.774(0.502-1.195)0.248Occasionally0.0260.1770.1461.026(0.726-1.451)0.884Physical activityAlwaysReferenceOccasionally−0.2440.166−1.4720.784(0.566-1.084)0.141Never−0.4690.32−1.4630.626(0.334-1.172)0.143Earphone useNoReferenceYes−0.4270.182−2.350.652(0.457-0.932)0.019TinnitusNoReferenceYes0.7670.1525.0282.152(1.596-2.902)<0.001Years of exposure<5 yearsReferenceto noise5-10 years0.0710.280.2521.073(0.62-1.857)0.80110-15 years0.1550.2720.5671.167(0.684-1.991)0.570>15 years0.0230.2420.0961.024(0.637-1.646)0.923Family history ofNoReferencehearing lossYes0.3510.1751.9991.42(1.007-2.003)0.046HypertensionNoReferenceYes−0.0510.185−0.2760.95(0.662-1.365)0.782Diabetes mellitusNoReferenceYes0.3040.3180.9541.355(0.726-2.527)0.340Notes: β, The regression coefficient; SE, Standard Error; OR, Odds Ratio; CI, confidence interval; P, These p-values less than 0.05 are statistically significant. Multivariate logistic regression analysis. Notes: β, The regression coefficient; SE, Standard Error; OR, Odds Ratio; CI, confidence interval; P, These p-values less than 0.05 are statistically significant. Establishment of a predictive model for NIHL among petrochemical workers The nomogram was developed based on the multivariate logistic regression results from the training set (Table 3). The regression coefficients (β) of each independent predictor were used to assign points to each variable level. The scoring system was automatically generated using the rms package in R, which linearly converts the regression coefficients into a 0–100-point scale, with the variable showing the highest β value (male sex) receiving the maximum weight. The constructed nomogram visually incorporates gender, age, earphone use, tinnitus, and family history of hearing loss as risk factors (Fig. 1). To apply the nomogram, the points corresponding to an individual's risk factor levels are summed to derive a total score, which is then projected onto the bottom axis to obtain the predicted probability of hearing loss. For example, a male worker (89 points), aged approximately 48 years (69 points), who does not use earphones (60 points), experiences tinnitus (82 points), and has a family history of hearing loss (71 points), would have a total risk score of 371. According to the nomogram, this corresponds to an estimated 47% probability of developing hearing loss.Fig. 1Noise exposure petrochemical workers hearing loss prediction model nomogram.Note. Points: scoring axis; Family_history: blue box ‘No’ for no family history of hearing loss, ‘Yes’ for family history of hearing loss; Tinnitus: blue box ‘No’ means no tinnitus, ‘Yes’ means tinnitus; Earphone_use: ‘Yes’ means earphone use, blue box ‘No’ means no earphone use; Age: blue waveform shows different age distribution; Gender: blue box ‘Female’ means female, ‘Male’ for male; Total points: blue waveform shows the distribution of total score; Pr(outcome): predicted probability; ∗ denotes P< 0.05; ∗∗∗ denotes P < 0.001. Noise exposure petrochemical workers hearing loss prediction model nomogram. Note. Points: scoring axis; Family_history: blue box ‘No’ for no family history of hearing loss, ‘Yes’ for family history of hearing loss; Tinnitus: blue box ‘No’ means no tinnitus, ‘Yes’ means tinnitus; Earphone_use: ‘Yes’ means earphone use, blue box ‘No’ means no earphone use; Age: blue waveform shows different age distribution; Gender: blue box ‘Female’ means female, ‘Male’ for male; Total points: blue waveform shows the distribution of total score; Pr(outcome): predicted probability; ∗ denotes P< 0.05; ∗∗∗ denotes P < 0.001. Assessment of the nomogram model Based on the nomogram analysis, a total risk score was calculated for each study participant. Fig. 2A (training set) and Fig. 2B (testing set) displayed the receiver operating characteristic (ROC) curve analysis for the model. In the training set, the area under the curve (AUC) was 0.743 (95% CI: 0.710–0.776), while in the testing set, it was 0.783 (95% CI: 0.734–0.833). Both AUC values exceed 0.7, indicating good predictive performance of the model.Fig. 2ROC plots and Calibration curves of the nomogram.Note. Diagonal: stochastic prediction reference line, Blue area: area enclosed by ROC curve and reference line, Fig. 2: ROC plots (A-B) and calibration curves (C-D) for training/testing sets. ROC plots and Calibration curves of the nomogram. Note. Diagonal: stochastic prediction reference line, Blue area: area enclosed by ROC curve and reference line, Fig. 2: ROC plots (A-B) and calibration curves (C-D) for training/testing sets. Fig. 2C and D displayed the calibration curves for the model in the training set and testing set, evaluating the agreement between predicted probabilities and observed outcomes. In both cohorts, the calibration curves closely follow the diagonal line, demonstrating that the model is well-calibrated, with predicted risks aligning closely with actual incidence. Fig. 3 presents the decision curve analysis (DCA) for both datasets. DCA assesses the clinical utility of the nomogram by quantifying the net benefit of using the model across a range of threshold probabilities. The DCA curves lie above the “treat all” and “treat none” lines over most thresholds, indicating that the model provides a meaningful net benefit and shows promise for clinical application.Fig. 3DCA curves in the training (A.) and testing (B.) sets. DCA curves in the training (A.) and testing (B.) sets. Discussion The study aimed to develop a predictive nomogram capable of estimating individual risk within this occupational population, by investigating the factors influencing hearing loss among noise-exposed petrochemical workers. Our data suggest that gender, age, earphone use, tinnitus, and a family history of hearing loss are independent factors affecting NIHL. The nomogram, constructed based on these variables, demonstrated strong predictive capability for hearing loss in this occupational population. Our findings further revealed that 21.71% (277/1700) of noise-exposed petrochemical workers experienced hearing loss—an unacceptably high rate that raises major concerns about the adequacy of hearing protection in the working environment. Thus, proactive measures should be urgently implemented to safeguard the health of these workers. The literature reports a wide range of NIHL prevalence, from 5.39% [12] to 76.97% [13], with a mid-range estimate of 21.3% [14] in similar populations. These variations may be attributed to differences in diagnostic thresholds [e.g., our use of >25 dB(A) compared to >40 dB(A)], as well as differences in occupational environments (significant differences in noise intensity, protective conditions in different occupational environments), warrant further investigation. We found that the rate of NIHL was significantly higher in men than in women, female hormones may offer some protection against noise-induced damage to cochlear function [15]. One possible underlying mechanism is that oestrogen enhances the expression of anti-oxidant enzymes and suppresses the production of reactive oxygen species (ROS), thereby reducing damage to cochlear hair cells [16,17]. It also modulates nitric oxide synthase activity and improves cochlear microcirculation, offering further protection to auditory structures [18]. However, this proposed explanation requires further experimental validation, and its potential industrial implications should also be explored. Moreover, we also found that age was a significant risk factor for NIHL, likely related to the duration of exposure in the working environment (i.e., the number of years spent in such conditions relative to the workers' age). Furthermore, this approach could provide more objective insights into the underlying mechanisms. Our findings are consistent with those of Kobrina and Rahimi et al. [19,20], Kobrina demonstrated through animal experiments that aging aggravates noise-induced hearing impairment, and Rahimi confirmed through population studies that the incidence of hearing loss shows a significant upward trend with age. All these further indicate that age is also a key influencing factor in the occurrence and development of NIHL. It should also take the age factor into consideration, as organ degeneration when people get old, when NIHL is determined, because as individuals age, the cochlea's antioxidant defence capacity declines, reducing its ability to counteract oxidative stress [21]. Additionally, ageing is associated with slower neural conduction and reduced synaptic plasticity in the auditory pathway, further impairing auditory function [22]. When combined with prolonged noise exposure, this leads to cumulative damage to cochlear hair cells and an increased risk of hearing loss [23]. This study found that prolonged earphone use was associated with reduced hearing loss. A possible explanation is that earphone users may be more likely to adopt self-protective behaviours, such as controlling volume levels and limiting usage time, thereby indirectly reducing the risk of noise-induced damage [24]. Furthermore, tinnitus has been identified as an independent risk factor for NIHL, consistent with previous findings [25]. Tinnitus is believed to contribute to oxidative stress, which may damage hair cells, vascular structures, and neural components within the auditory system [26]. Family history of hearing loss was identified as another independent risk factor for NIHL in the current study. Genetic influences may account for up to 50% of the variability in susceptibility to noise-induced hearing damage [27,28]. Inherited variations may increase cochlear vulnerability or impair cellular repair mechanisms [29], suggesting that individuals with a family history of hearing loss may be genetically predisposed to NIHL. Low compliance with hearing protection devices was noted in this study. A key reason was that workers rely on real-time verbal communication during operations, and hearing protectors were perceived to reduce speech clarity and hinder effective communication, thus limiting consistent use. Interestingly, the duration of occupational noise exposure was not found to be significantly associated with NIHL in the study. We speculate that this may be because participants in our cohort took their health seriously and strictly adhered to occupational health and safety (OHS) protocols, which could have substantially mitigated the risk of NIHL [30]. However, the literature presents conflicting findings. One study reported no significant association between the age at which workers began noise-exposed jobs and the incidence of hearing loss [31], whereas another identified exposure duration as a key risk factor [32]. Further studies are needed to verify these differences. Several factors could explain these discrepancies. First, variations in industrial activities—particularly the noise intensity levels across different occupations—may significantly influence the outcomes. Second, the study cohorts may originate from different regions or countries, with potentially substantial differences in genetic background, resulting in varied susceptibility to noise-induced hearing loss. Additionally, pre-existing unhealthy conditions such as smoking and alcoholism may increase the risk of NIHL by impairing the body's antioxidative capacity. These factors were not considered in the current study and warrant further investigation. This discrepancy may be due to the influence of individual genetic susceptibility. Some individuals may carry genetic variants that render them more sensitive to noise, resulting in severe hearing loss despite shorter exposure durations. Lifestyle factors, such as chronic smoking and alcohol consumption, may further compromise auditory function. Smoking weakens the body's antioxidant defence system and disrupts tissue metabolism [33], while alcohol damages neural structures [34]. These findings suggest that hearing loss results from the complex interaction between genetic predisposition and environmental/lifestyle factors. Thus, years of noise exposure alone may not independently predict NIHL risk. Education level was not identified as an independent predictor of NIHL, education primarily reflects intellectual and psychosocial characteristics rather than direct physical exposure, such as to occupational noise. Our finding is supported by previous studies reporting similar outcomes [35]. Marital status was also not an independent risk factor for NIHL, in agreement with findings from a large cohort study [36]. Both education and marital status are more closely associated with psychosocial factors rather than physiological vulnerability to noise exposure. Hypertension and diabetes mellitus were not identified as independent risk factors for NIHL in this study. Although epidemiological studies often report associations between these conditions and hearing loss, the underlying mechanisms may involve shared lifestyle-related risk factors. Unhealthy behaviours contributing to hypertension and diabetes—such as poor diet and physical inactivity—can also impair auditory health through oxidative stress, vascular dysfunction, and chronic inflammation [37]. Therefore, these conditions may serve as mediators rather than direct causal factors for NIHL. Nonetheless, this study has limitations. First, noise cumulative exposure (NCE) is a more direct and objective predictor of hearing damage, but its data were unavailable, potentially affecting model precision; future studies should include NCE measurements to improve accuracy. Second, the population was mainly petrochemical workers, limiting generalizability to other occupational groups. Finally, due to constraints in the raw data, noise exposure duration could not be converted into a continuous variable and was included only as a categorical measure. This may have weakened the gradient association between exposure dose and NIHL risk, potentially limiting the ability to fully elucidate the underlying dose–response relationship. The developed nomogram provides a practical and visual tool for individual risk assessment of noise-induced hearing loss among petrochemical workers. It can be conveniently applied in occupational health examinations to identify high-risk individuals, enabling early screening, targeted intervention, and personalized hearing protection. In addition, this model may be generalized to other noise-exposed industrial populations with similar demographic and occupational characteristics, such as workers in manufacturing, mining, and construction. However, caution should be taken when extending the model to populations with different noise exposure levels, genetic backgrounds, or lifestyle profiles. Further validation in external cohorts is warranted to enhance its generalizability and clinical utility. Conclusion In summary, the prevalence of hearing loss among noise-exposed petrochemical workers in this study was 21.71%. Gender, age, earphone use, tinnitus, and family history of hearing loss were identified as independent influencing factors for NIHL. The nomogram model constructed using these variables demonstrated reliable predictive performance and may serve as a practical and intuitive tool for assessing NIHL risk. Ethical statement All participants provided written informed consent. This study was approved by the Ethics Committee of the Affiliated Hospital of Gansu University of Chinese Medicine (Approval No. 52 [2021]). Author contributions Jingchun Fan proposed the research idea; Xie Chen, Zhuanli Dong and Leilei He collected the data; Shisan Bao, Jiajia Wu and Xiaomei Jiang analyzed the data; Tingting Ma wrote the article. Data availability statement All data generated or analyzed in this study are included in the article. Funding This work was supported by the Key Research and Development Program of Gansu Provincial Department of Science and Technology (Grant No 25YFFA044); the Gansu Provincial Department of Education's Project for Young Doctoral Teachers in Higher Education Institutions (Grant No 2025QB-064); the Open Project of Key Laboratory of Dunhuang Medicine, Ministry of Education (Grant No DHYX24-17); and the Key Talent Project of Health in Gansu Province (2026) (Grant No 2026SWJWRC008). Declaration of competing interest No potential conflict of interest was reported by the author(s).