A risk model and nomogram for high-frequency hearing loss in noise-exposed workers

Occupational hearing loss is the second most common form of sensory-related hearing loss after age-related hearing loss and is a major occupation-related condition worldwide [1]. The World Health Organization (WHO) reported that 16% of adult hearing loss cases were caused by occupational noise exposure [2]. The WHO also indicated that the number of individuals with hearing loss could increase to 630 million by 2030 and may reach more than 900 million by 2050 [3].

As one of the largest global producers, China has many manufacturing, construction, and mining enterprises [4]. Noise pollution has become one of the key public hazards [5]. Due to the high cost of occupational noise services, mismanagement of occupational health and insufficient personal protective equipment for noise-exposed workers [6], the National Health and Family Planning Commission (NHFPC) of the People’s Republic of China (PRC) reported that the incidence of occupational ear, nose and throat diseases (OENTD) has increased in recent years. New cases of OENTD exceeded those of occupational poisoning to become the second most common occupational disease after occupational pneumoconiosis since 2015, and 95.90% of OENTD cases were noise-induced hearing loss (NIHL). From 2015 to 2019, the NHFPC reported that the numbers of new cases of OENTD were 1097 cases, 1276 cases, 1608 cases, 1528 cases, and 1623 cases, respectively [7‐11].

In 2009, the China Centers for Disease Control and Prevention (CDC) launched the National Key Occupational Diseases Survey (NKODS), a nationwide surveillance project of ten key occupational hazards, namely, coal dust (coal silica dust), silica dust, asbestos, benzene, lead, noise, brucella, welding fumes, carbon disulfide, and phosphorus compounds, to cope with the increasingly severe challenges of occupational diseases. Noise is one of the ten key occupational hazards in the NKODS. The NKODS-Sichuan is the provincial level surveillance project of the NKODS.

HFHL is a characteristic of occupational hearing loss that develops slowly, affecting the higher frequencies first and extending gradually to the lower frequencies [12]. Previous studies focused on the prevalence or factors of HFHL in different regions [13‐17], HFHL-related diseases [18‐20], and HFHL-related mental disorders [21‐23]. Regarding the risk model, the researchers paid more attention to the modelling of noise exposure risk in the workplace. However, few studies have provided risk models of HFHL or hearing loss. Lewkowski et al. [24] identified the risk of miners for developing occupational hearing loss and they found that miners who used planers, sanders, grinders, large machinery, and power hammers in work tended to suffer a high-level noise exposure. Pentti Kuronen et al. [25] constructed a risk model of NIHL in military pilots. It is well known that pilots have better health surveillance and management than other professions. Hong et al. [26] developed a prediction model using 379 South African adults, the predictors included the CD4 count, age, serum albumin level, and body mass index. This model can identify participants with drug resistant tuberculosis who are at high risk of developing minoglycoside-induced hearing loss. This research modelling the risk of disease-related hearing loss is not the same as the hearing loss induced by occupational noise. Therefore, this risk model is not appropriate for predicting HFHL in noise-exposed workers. Kuang et al. [27] predicted the individual risk of HFHL in 822 machinists. The predictors included age, sex male, limited earplug wearing, and high noise intensity. These results provide theoretical support for the prevention of HFHL among machinists. However, this study still has some limitations. On the one hand, the study was limited by its single-centre design, small sample size, and industry type whereby all the objects were mechanists. On the other hand, the study lacked an additional set of objects for external validation as well as a cut-off point to identify the high and low risk groups of workers. Furthermore, most studies have shown that the industry type and enterprise type are related to hearing loss [28]. The early prediction and diagnosis of HFHL in different industrial and enterprise types could provide evidence for the comprehensive prevention of HFHL.

Nomograms are pictorial representations of a complex mathematical formula and are reliable and convenient statistical predictive tools based on the indicators of a clinical event, enabling the calculation of an individual probability with a simple and visual graphical representation [29]. Nomograms are widely used to predict the probability of the higher risk of disease in clinical, which is usually used to predict survival in cancer patients, especially prostate cancer and breast cancer [30, 31]. Few studies on the development of a nomogram for HFHL have been published. As workers in high noise environments are at a high risk of hearing loss, especially HFHL, a risk model and nomogram may be valuable for management and prevention in workplaces with high levels of noise exposure. The approaches for assessing occupational noise exposure have been tested, such as questionnaire-based algorithms [32] and artificial intelligence [33]. However, those studies assessed noise exposure in the workplace or risk assessment of different task. Based on the current situation of HFHL, the development of a predictive tool using the NKODS dataset is a valuable effort for applied research on HFHL prevention. Besides, Chinese researchers have suggested that a predictive tool based on the NKODS dataset would be valuable in China to help regulators and managers effectively use the NKODS dataset and provide early warning for populations at risk for HFHL [34].

Considering the above-mentioned information, the objective of this study was to construct a risk model to identify the predictors of HFHL and develop a HFHL nomogram to calculate the individual risk for noise-exposed workers. The results of this study are expected to provide technical support and enhance application-oriented research on HFHL.

Most research has supported that long-term noise exposure in workers is likely to cause HFHL [35]. In this study, we identified the risk factors for HFHL and also developed a HFHL nomogram for noise-exposed workers.

In our study, 10.06% (3232/32121) of the workers had HFHL. The prevalence of HFHL in Sichuan is higher than that in China. The prevalence is higher than that in Ordos (5.5%), Wuhan (7.3%), Jining (4.6%), and Tianjin (8.3%) [36‐39], and lower than that in Zhoushan (22.02%) and Sanmenxia (12.0%) [40, 41]. Occupational hearing loss has historically been difficult to compare because the criteria for HFHL among workers vary from country to country [42].

In our risk model, the results indicated that males (OR = 3.25) experience more effects after exposure to occupational noise than females. The reason may be due to males usually having greater exposure to noise at work than females due to differences in occupational categories and lifetime work history. Several animal and human studies have demonstrated that women may be protected against hearing loss because of oestrogen and its signalling pathways [43, 44]. Age (OR = 1.09) and NED (OR = 1.15) have a negative effect on HFHL, and research has showed that age and NED are important factors for changes in the hearing threshold. Our study strongly supports these results [45‐47].

Interestingly, our risk model showed that age and NED had independent effects on HFHL when the BHFTA was adjusted for age. There could be two explanations for this result: first, the study sample is predominantly comprised of male workers, and studies have shown that male workers are generally more vulnerable to occupational hearing loss than female workers [48]. There was a greater proportion of females was used to control for age, which may have accounted for some of the effects. Second, in the risk model, the variable age may represent occupational hearing loss due to other exposures and not the influence of age on HFHL over time, given that the exposure duration is related to age. Exposures to other factors such as ototoxic chemicals in the workplace may not be detected by occupational environmental monitoring. Chemicals alone or combined with noise have recently become a concern as a cause of occupational hearing loss [49]. In the NKODS-Sichuan, noise-exposed workers were usually asked whether they were exposed to toxic chemicals in the workplace, but in general, workers often do not know or remember which chemicals they have encountered. In addition, toxic chemicals may have greater effects with longer exposure durations or in older workers. This may also explain why age and NED had independent effects on HFHL when age was adjusted for in this study. It is difficult to detect toxic chemical exposure in such nationwide screenings for NIHL among noise-exposed workers. However, toxic chemical effects on occupational hearing loss still should be considered in the future [50].

The risk model showed that the mining industry (OR = 2.63) has a strong negative effect on HFHL, with a higher positive rate of HFHL than the other industry types in both the validation cohort and training cohort. Our results suggest that mining workers should have more hearing protective measures and receive more attention from managers. In Poland, according to a report by the Central Statistical Office, the number of workers who exceeded the noise level (85 dB) was approximately 200 thousand, with the highest numbers in industries related to mining, metal and metal product production, textiles and wood production [51]. The mining of minerals necessitates the use of heavy energy-intensive types of machinery and equipment, leading miners to be exposed to high noise levels [52]. Based on the risk model and nomogram, workers who are employed in POE are more likely to experience HFHL than those in other enterprise types. Private enterprises are mostly small-sized enterprises, which are limited by less investment in occupational disease prevention, a lack of occupational health management, and no personal protective equipment. Moreover, the level of noise exposure for those in POE is more serious than those in SOE and FOE, and POE should be regarded as a key enterprise that is in need of prevention and management strategies for HFHL. In addition, the supervision department should pay more attention to the use of hearing protection by noise-exposed workers in POE [53, 54].

We also developed a HFHL nomogram based on the risk model. In the training cohort, the final risk model had good discrimination, as demonstrated by an AUC of the ROC curve of 0.713 (95% CI: 0.704–0.722), and the model continued to have excellent discriminatory ability in the validation cohort (AUC = 0.714, 95% CI: 0.695–0.733). Our nomogram was successfully subjected to independent external validation, which revealed good calibration and better discrimination in the validation cohort than in the training cohort. The results indicate that the risk model and nomogram for HFHL can be an effective tool to assess noise-exposed workers’ risk for HFHL. Moreover, due to its easy-to-use and the visualization features, noise-exposed workers can use the tool on a daily basis. Furthermore, the maximum Youden’s index revealed that noise-exposed workers have a risk probability of HFHL greater than 9.6%. Administrative staff should pay more attention to providing hearing protection for or transferring the position of noise-exposed workers. Regarding further research on the HFHL nomogram, researchers can focus on: (1) comparing other assessment tools with the HFHL nomogram in different occupation-specific populations and (2) the risk probability of HFHL could be considered a new variable in further studies on occupational hearing loss.

This study still has two limitations. (1) Generally, the direct predictor of HL is noise cumulative exposure (NCE), but the NKODS did not collect the NCE levels of the participants. If data on area noise monitoring instead of the NCE levels were included in the model, differences in the results may have been observed. Therefore, we used the NED to represent the level of noise exposure. (2) Exposure to toxic drugs and chemicals was self-reported by the workers in the NKODS. Workers may be unaware of toxic drugs that they have taken or toxic chemicals that may be present in their work environment, which may have affected our results.