A systematic search was performed on Medline, Embase, PsycINFO, the Cochrane Central Register of Controlled Trials; CINAHL) (Cumulative Index to Nursing and Allied Health Literature), Scopus, NIOSHTIC-2 (National Institute for Occupational Safety and Health), ClinicalTrials.gov, and World Health Organization (WHO) International Clinical Trials Registry Platform. The search strategy employed was consistent with that reported in a previous Cochrane review (Morata et al. 2024). There were no restrictions on language, publication year, or publication status. Citations were imported into EndNote, where duplicates were removed. The databases were last searched on February 5, 2025.

The inclusion criteria were (1) UBA studies involving fit-tested workers who used HPDs and were exposed to noise levels of more than 80 dBA; (2) studies that reported noise attenuation levels provided by various type of HPDs used by workers; (3) studies involving fit testing procedures and different formats of instructions for workers on how to properly select and fit HPDs. The primary outcome was noise attenuation measured either as a PAR and/or PAR pass rate. PAR pass rate is the percentage of workers who passed an established level of sufficient attenuation from their HPDs, determined based on their individual noise exposure levels. The secondary outcome was the HPD use rate, defined as the percentage of workers who used HPD during their work shift.

Eligible studies were selected in two stages. First, the 17 UBA studies identified in the previous Cochrane review (Morata et al. 2024) were automatically included. The last search of the Cochrane review was February 29, 2024. Second, search results from February 29, 2024 to February 5, 2025 were imported to the web-based software Covidence. Two review authors (WG, TM) independently screened the title and abstracts of all studies identified from the new search. The same two authors then conducted full-text screening. Any discrepancies between reviewers were resolved by discussion or by consulting a third review author (AS).

One review author extracted study characteristics using the same data collection form from the previous Cochrane Review (Morata et al. 2024) and entered data in a Microsoft Excel (Microsoft^® Excel^® for Microsoft 365 MSO, Version 2402) spreadsheet to perform a meta-analysis using a random-effects model (Neyeloff et al. 2012). The second review author spot-checked data entry for accuracy against the study report. For dichotomous outcomes, risk ratios (RRs) were calculated for PAR pass rate and the rate of HPD use. For continuous outcomes, mean differences (MDs) were used for PAR. The sample mean and standard deviation (SD) were estimated as described by Van et al. (2014) when studies reported only sample size, median, range and/or interquartile range.

All training formats reported in the included studies were categorized into two levels: extensive and simple. Extensive training was defined as any format training that includes individualized instruction, while simple training was defined as any training format without an individualized instruction component. Initially, all workers underwent pre-intervention fit testing, and the first PAR was recorded as pre-intervention PAR. Subsequently, some workers took a post-intervention retest after receiving an extensive and/or simple instruction either immediately (same day), in the short-term (up to 6 months), or in the long-term (beyond 6 months). This retested PAR was recorded as the post-intervention PAR. Later, a follow-up fit test was conducted among those who had completed the post-intervention test to obtain follow-up PAR values; this follow-up fit test could include or exclude refresher training. The refresher training was also categorized as either extensive or simple instruction.

To assess the effectiveness of HPD fit testing with extensive instruction, the post-intervention PARs were compared with the pre-intervention PARs. This comparison allows for an evaluation of how long the effectiveness of extensive instructions is maintained.

To determine whether refresher instruction was necessary to maintain the effectiveness of extensive instruction, the follow-up PARs were compared with the post-intervention PAR values obtained immediately after extensive instruction. This comparison also allows evaluation of which type of refresher instruction (extensive, simple, or none) would be most effective and how often it would need to be provided to maintain the effectiveness of extensive instruction.

In addition, the rate of HPD use among workers was compared between the pre-and post-intervention fit testing, as well as between follow-up and the post-intervention fit testing. This comparison aims to examine whether HPD fit testing accompanied by instruction enhances adherence to HPD use.

The data were pooled from studies deemed sufficiently homogenous in terms of intervention, participants, and settings. A meta-analysis was performed with random-effects model, if more than one study provided useable data in any single comparison. Statistical significance alpha levels were set at 0.05.

The authors aimed to use the Risk of Bias in Non-randomized Studies of Interventions (ROBINS-I) tool to assess the risk of bias in the eligible UBA studies (Sterne et al. 2024). If the evidence allowed, the five considerations (risk of bias, inconsistency, imprecision, indirectness, and publication bias) of the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) method was employed to assess the certainty of a body of evidence as it related to the studies that contributed data for the pre-specified outcomes.

The latest search yielded 284 references in total. During the title and abstract screening, 277 were not eligible for inclusion. The remaining seven articles underwent full-text review, and one was excluded because of incomplete data. This process resulted in 6 recent UBA studies being added to 17 UBA studies (reported in 18 articles) identified in the 2024 Cochrane review (Morata et al. 2024). A total of 23 UBA studies with completed datasets were included in this review (Figure 1, Appendix Table 1). No RCT studies were found. Workers from 14 studies (Appendix Table 1) who did not achieve the desired PAR during the initial fit tests and/or with incorrect fit skill received an intervention followed by follow-up fit tests. In contrast, workers from nine studies (Appendix Table 1) underwent initial and follow-up fit tests regardless of the PAR levels they achieved at their initial tests.

Of 5575 workers who were recruited in the 23 UBAs, a total of 4629 workers were eligible and participated in the initial fit test to obtain their pre-intervention PARs. Of these, 4184 received extensive instructions and were fit tested to estimate their post-intervention PARs. Among them, 3374 were retested on the same day when they received extensive instructions, 456 were tested within 6 months after receiving extensive instructions, and 351 were tested more than 6 months later. In addition, 1275 took a third test to estimate their follow-up PARs, either within 6 months or beyond six months after their second test.

Seventeen studies were conducted in manufacturing facilities with 11 located in China, two in the United States, with one each from Canada, Taiwan, Korea, and Japan. In addition, one study took place at the U.S. Department of Defense, two were in the U.S. Marine Corps, and one was at an energy distribution company in Canada. Furthermore, two studies were performed at a public university in Brazil (Appendix Table 1). All workers reported being exposed to noise at work. Fourteen studies documented noise exposure levels ranging from 80 to 107 dBA, while some peak levels from rifles reached up to 160 dB SPL. Only five studies provided detailed descriptions of noise measurement methods and results (Gong et al. 2021; J. Liu et al. 2024; Murphy et al. 2016; Shu et al. 2025; Zhong et al. 2023). The ages of participants, reported by 16 studies, ranged from 18 to 69 years old (Appendix Table 1). Three studies included only male participants, one study primarily included female workers, and 17 studies reported a predominance of male participants (Appendix Table 1).

Eight studies used the real-ear attenuation at threshold (REAT, ISO 2018) method to estimate PAR values, while the remaining 15 studies used the microphone in the real-ear (MIRE, ANSI 2018) method (Appendix Table 1). Only one study measured PAR levels for canal caps, two studies tested earmuffs, and the remaining studies tested different types of earplugs including foam, pre-molded, and push-to-fit earplugs (Appendix Table 1).

Three types of training format were defined: extensive, simple, and no instruction. To evaluate the effectiveness of intervention, 14 studies compared no instruction with extensive instruction (Appendix Table 1); three studies compared extensive instruction with simple instruction — specifically, HPD manufacturers’ package instruction in two studies (Assunção et al. 2019; Takada et al. 2020) and group instruction in one study (Tsukada and Sakakibara 2008). In addition, one study reported the intervention effect of a short video instruction (simple), expert-fit instruction (extensive), or integrated video training and expert-fit instruction (extensive), compared with no instruction (Federman et al. 2021).

Twenty-one studies (Appendix Table 1) measured the immediate effects of fit testing combined with instruction. Eight studies assessed short-term effects, while four studies measured long-term effects (Appendix Table 1). Eight studies evaluated the effect of follow up fit testing with refresher training weekly (Martin et al. 2019), biannually (Assunção et al. 2019), or over a period within 6 months (Gong et al. 2019; Liu et al. 2020; Tsuakda and Sakakibara 2008; Lü et al. 2022) or in 12 months or more (Fu et al. 2024; Liu et al. 2020; Murphy et al. 2016). In addition, Kim et al. provided annual individualized instruction to earplug users for four consecutive years to evaluate the effect of annual training (Kim et al. 2019). Only one study examined whether fit testing with instructions contributed to the use of hearing protection (Tsukada and Sakakibara 2008).

Regarding the assessment of risk of bias and quality of evidence, it is important to note that all the UBA studies identified in this review reported only one pre-intervention outcome measurement and one post-intervention outcome measurement for each participant. While the ROBINS-I tool is particularly relevant for UBA studies, it is not appropriate to those that rely on single pre- and/or post-intervention outcome measurements for each participant. Such studies are typically judged to be at serious or critical risk of bias because it is impossible to determine whether pre-post changes are due to the intervention rather than other factors (Sterne et al. 2024). Furthermore, UBA studies are observational in nature and focus solely on subjects who underwent the intervention. As a result, they usually warrant downgrading from low to very low quality according to the GRADE handbook (Schünemann et al. 2013). Therefore, all included studies were judged to be at serious risk of bias, which resulted in very low quality of evidence and should be considered with caution.

This mixed method review of 23 UBA studies revealed that providing extensive instructions immediately followed by fit testing achieved the greatest increase in PAR values and PAR pass rates for earplugs, compared with the pre-intervention PARs from fit testing without instruction. The increase in mean PAR (MD 12.88 dB; 95% CI: 10.03 to 15.66) is greater than that reported in the Cochrane review (MD 8.34 dB; 95% CI: 7.32 to 9.36). One possible explanation for this difference is that 14 UBA studies only provided intervention for workers who failed to achieve the desired PAR and/or with inappropriate use of HPDs, indicating that the pre-intervention PAR values for these workers were relatively lower than those in the randomized controlled trials (RCTs) analyzed in the Cochrane review (Morata et al. 2024). However, the increase in PAR levels is consistent regardless of the difference in the study design and rigor.

When there was a time interval of one week to 17 months between the extensive training and the post-intervention fit tests, a longer time interval between the instructions and the fit test was associated with lower attenuation. An exception was noted in the Park and Casali (1991) study, where workers received extensive instructions before a 3-week experimental period and were fit tested once a week. Their 3-week post-intervention mean PARs for earplugs improved by 12 dB compared with their pre-intervention PARs (Park and Casali 1991). This increase was even greater than the immediate effects observed in the randomized controlled trial (RCT) study (Federman et al. 2021). Beyond the different study designs, a possible explanation is that frequent fit testing (once a week) may enhance the effectiveness of extensive instruction.

For earmuff, one study (Liu and Yang 2018) observed improvements in mean post-intervention PAR values at immediate, 6-month, and 12-month test after participants received extensive instruction. The results, along with those from earplugs, consistently show that an interval of 6 months or more between instructions and fit testing significantly reduces the increase in PAR values from baseline. This finding indicates that providing extensive instruction immediately followed by a fit test leads to the greatest improvement in PAR values or pass rates. Conversely, longer intervals between instructions and fit testing (whether 6 or 12 months) are associated with lower increase in attenuation. This suggests that a follow-up fit test is necessary after workers have demonstrated their ability to receive effective protection following extensive training.

One study reported a change in the rate of earplug use, showing an increase from 4% to 66% over two months following extensive training (Tsukada and Sakakibara 2008). In another study, workers indicated that they increased their earplug usage after initial fit tests, however, the rate of HPD use was not reported (Lü et al. 2022). A 2024 study found that 138 out of 192 workers removed their earplugs during their initial visit for conversations or equipment checks; however, this number decreased to only 94 workers citing the same reasons after 17 months of their fit testing and instructions (Fu et al. 2024). These findings suggest that measuring the rate of HPD use may not comprehensively assess the impact of HPD fit testing since workers are generally required to wear HPDs when exposed to hazardous noise. Neitzel and Seixas (2005) noted that construction workers who reported “always” using HPDs in high noise environments actually wore them only one-third of the time when their noise exposure exceeded 85 dBA. Addressing factors contributing to non-use—such as communication, comfort, and convenience—could enhance the consistency of HPD usage. Alternative metrics like HPD wearing duration, frequency of removal, or comfort levels may serve as better indicators for assessing consistent HPD use.

Evidence from two uncontrolled studies (Martin et al. 2019; Assunção et al. 2019) indicated that when the same subjects received extensive refresher instructions (either weekly or biannually), the mean post-training PAR improved by an average of 1.53 dB compared with the previous post-intervention PAR (95% CI: 0.99 to 2.06). In another study, annual fit testing combined with extensive instructions for 172 workers over 2 years and 30 workers over 4 years resulted in a mean PAR increase of 3.45 dB (95% CI: 2.01 to 4.88) at the 4th year test, compared with the post-intervention PAR values measured immediately after extensive training at the first year (Kim et al. 2019).

Extensive instructions can be time-consuming. Murphy et al. (2016) reported that training sessions lasted between 3 to 7 minutes, while Gong et al. (2021) noted that the duration of fit testing combined with extensive training varied from 5 to 30 minutes, depending on the number of times workers needed to be trained, refitted, and retested. In contrast, group training, video presentations, or written instructions typically require less time and fewer resources than individualized extensive instruction. One study provided a 10-minute hearing health education and HPD training for groups of 8 to 16 U.S. Marine Corps recruits (Kulinski et al. 2024). This group training combined a standardized script, visual inspection, and immediate feedback from an occupational audiology staff member who provided recruits with the necessary skill to properly insert their HPDs. The study authors found a significant immediate increase of 14.6 dB in post-intervention PAR, with a slight decrease at one-month follow-up, yet still above pre-intervention levels. The study indicates that implementing small-group training at large-scale in-processing events, where resource constraints may limit one-on-one intervention, could be considered a viable option for refresher instruction (Kulinski et al. 2024).

This mixed method review found that providing follow-up fit testing with simple refresher instruction at 6-month intervals allows workers to achieve the same level of protection as they did with initial extensive training. Therefore, this review suggests that simple instruction could serve as an alternative intervention to maintain the effectiveness of extensive instruction once workers are proficient in selecting and wearing hearing protection devices correctly. However, no studies have examined the long-term effects (beyond 6 months) of follow-up using refresher simple instruction.

In contrast, the follow-up fit test without refresher instruction resulted in a decrease in the mean PAR for both earplugs (Kim et al. 2019; Liu et al. 2020) and earmuffs (Liu and Yang 2018) over 6 months to 1 year. This finding confirms that periodic follow-up fit testing with some forms of refresher instruction is necessary after workers have demonstrated their ability to achieve effective protection following extensive training.

The findings from this mixed method review indicate that providing extensive training can immediately enhance the noise attenuation offered by HPDs, aligning with the results reported in the Cochrane review published in early 2024. However, this initial increase in effectiveness tends to diminish over time. Fortunately, implementing periodic follow-up fit testing, along with either extensive or simple refresher instruction, can help maintain—and in some cases even improve—the effectiveness of the initial training. In other words, periodic follow-up fit testing with refresher training is beneficial for workers who demonstrate their ability to achieve effective protection.

From a practical perspective, biannual, annual, or biennial training would be more feasible than weekly sessions. However, few high-quality studies have evaluated the effects of periodic refresher instruction or the maintained effects of such instructions. These findings align with a previous Cochrane review (Tikka et al. 2017), which noted a limited number of studies examining the effectiveness of refresher training. Further research is likely to have an important impact and could modify the conclusions reached in this review.

Future randomized controlled trials (RCTs) could examine the need for optimal time intervals between tests with extensive instruction and periodic testing combined with refresher instructions. This mixed methods review indicates that periodic fit testing of HPDs with extensive or simple refresher instruction could help maintain the achieved attenuation from HPDs. To support this finding, RCTs of high or moderate quality could offer greater certainty to enhance the implementation of the intervention in workplace settings. Future studies also could use hierarchical clusters of noise exposure levels in the data analysis to better evaluate the training effects across different time frame.

For an HPD fit-testing strategy to be widely accepted in industry, it is important to identify its key elements and develop an efficient implementation model. In addition, addressing factors that contribute to non-use of HPDs—such as communication, comfort, and convenience—could improve the consistency of hearing protector use. Other outcomes, such as HPD wearing time, frequency of removal, and comfort levels, may also serve as valuable indicators for assessing consistency in HPD use.

As noted in the Cochrane review (Morata et al. 2024), two variables have not yet been explicitly investigated: the effect of experience with HPD use and past training history. Therefore, controlling for these variables in future studies appears warranted.