Occupational noise exposure, noise annoyance, hearing-related symptoms, and emotional exhaustion - a participatory-based intervention study in preschool and obstetrics care

Abstract A participatory-based intervention was performed in Sweden, aimed at improving the sound environment in one preschool (n = 20) and one obstetric ward (n = 50), with two controls each (n = 28, n = 66). Measured sound levels, and surveys of noise annoyance, hearing-related symptoms and emotional exhaustion were collected before, and three and nine months after the interventions, comparing intervention and control groups over time. The results of this first implementation in a limited number of workplaces showed significantly worsening of hyperacusis, sound-induced auditory fatigue, emotional exhaustion and increased sound levels in the preschool, and worsening of noise annoyance in both intervention groups. Increased risk awareness, limited implementation support and lack of psychosocial interventions may explain the worsening in outcomes, as might the worse baseline in the intervention groups. The complexity of the demands in human-service workplaces calls for further intervention studies.


Introduction
Noise exposure is one of the major causes of occupational disease. 1 The European Work Authorities highlights a lack of risk assessments in female-dominated sectors. 2 In human-service settings, such as education and health care, traditional interventions such as attenuating or removing noise sources are inadequate, as the noise arise mainly from intense speech communication, children playing or medical alarms. 3][6] In a Swedish obstetrics ward, 46% of 61 workshifts exceeded 80 dB LAeq (7-10 h) , and 5% exceeded 85 dB LAeq (7-10 h) , with several instances above 115 dB LAFmax. 7Exceeding these levels pose a risk of hearing damage.
A Swedish study found increased risk ratios (RR) for sound-induced auditory fatigue (RR 2.4, 95% confidence interval CI 2.2-2.5),hyperacusis (RR 2.3, 95% CI 2.1-2.5),difficulty perceiving speech (RR 1.9, 95% CI 1.7-2.0),tinnitus (RR 1.7, 95% CI 1.5-1.9)and hearing loss (RR 1.6, 95% CI 1.5-1.8) in female preschool teachers (n ¼ 4718), compared to randomly selected women (n ¼ 4122). 8Increased hazard ratio (HR) of hyperacusis (HR 3.4, 95% CI 3.0-3.7)has also been found in Swedish preschool teachers compared to women working in low noise exposure (<75 dB LAeq (8h) ). 9 One study on obstetrics personnel in Sweden found increased odds ratio (OR) for CONTACT Sofie  sound-induced auditory fatigue (OR 1.04, 95% CI 1.00-1.07)and tinnitus (OR 1.04, 95% CI 1.00-1.09) in relation occupational noise. 7wedish and Danish studies have shown that stress-related disorders (e.g.burnout and emotional exhaustion) are common in preschool personnel and midwives. 10,11The noise-stress model outline how noise cause physiological arousalby activating the sympathetic nervous system and the endocrine system. 12Interestingly, studies have indicated an association between stressand hearing, 13 such as increased hyperacusis during an acute stress test among women with emotional burnout. 14High working demands is pervasive in human-service work, 15 andmay thus also be important to reduce to prevent hearing disorder.Consistent with the noisestress model, noise annoyance may modify or mediate the risk of hearing-related symptoms. 7,16otably, 70% of preschool teachers and 49% of obstetrics personnel in Sweden report noise annoyance at work, 7,17 compared to 27% in the general female population. 17here is a lack of noise intervention studies in communication-intense human-service settings.While studies in the obstetrics care are lacking, a few studies in preschools in Sweden and Germany have found small or non-significant effects from acoustic or organizational measures on outcomes including noise annoyance and burnout. 6,18-21Small or absent effects may result from a lack of "measure-to-challenge correspondence" (the interventions do not target the existing problems). 22Compared to top-down implementation (e.g.from management), which can lack relevance and motivation among employees, employee-driven change is often focused on actual problems and realistic solutions, 23 and may achieve a sustainable change by integration with the ordinary systematic work environment work. 24his emphasizes the possibilities of noise intervention studies using a participatory-based approach.
This study aims to assess the effects on hearing-related outcomes, noise annoyance and emotional exhaustion of context-specific interventions developed using a participatory-based approach in one preschool and one obstetrics ward in Sweden.The hypothesis was an improved sound environment, and a decreased reporting of hearing-related outcomes, noise annoyance and emotional exhaustion.

Study design, setting and selection of participants
The study was designed as a non-randomized non-blinded intervention aimed at improving the sound environment in female-dominated humanservice communication-intense settings (preschools and obstetrics care) by implementing context-specific interventions developed using a participatory approach.
Figure 1 shows the timeline, the study phases and intervention activities.A purposive selection was used to ensure that the intervention workplaces had the necessary characteristics of female-dominated human-service communication-intense settings, and that they had resources to engage in the participatory approach.All workplaces were located in the V€ astra G€ otaland County, Sweden.One intervention preschool was selected via the head of schools in one municipality.The preschool principal had the resources and motivation to participate.One intervention obstetrics ward was selected after having participated in a previous study, which informed the researchers of the workplace characteristics.They also had resources to renovate the department as part of the intervention.Two control workplaces were recruited for each intervention workplace, closely matching size and building characteristics.All preschools were in suburban medium sized municipalities.The number of children ranged from around 50 in one control preschool to around 80-90 in the intervention preschool and the other control.In the obstetrics, the low-risk labor wards were included.The intervention ward was located at a university hospital in a larger city, with approximately 4000 births per year.The controls were among the next largest wards in in smaller neighboring cities, with around 3000 births per year each.One larger obstetrics ward and two preschools initially declined participation.One control preschool was excluded due to plans of a stress-reducing intervention.
All employees were invited to participate.Managers were not considered participants but were involved in the intervention process.
The study was approved by the Regional Ethical Review Board, Gothenburg Sweden, No. 659-18.Participants received written information and signed a written informed consent form.

The intervention processes
Figure 1 shows the intervention process, initiating with a reference group meeting in May 2017 including researchers, representatives from the intervention workplaces, and key stakeholders, explaining the study aim, and giving presentations on noise, stress and organizational interventions.Then, separate qualitative interviews, reported elsewhere, 3 were conducted to assess the needs, opportunities, and ideas for interventions.These results fed into the intervention via a second reference group meeting held in May 2018, including the intervention workplaces, where selection and priority ofinterventions was initiated.Then, baseline measurements were collected, after which implementation of interventions begun.To support the implementation process, a workshop (1.5-2 h) was held at each intervention workplace, including results from the qualitative interviews. 27n accordance with the participatory method, the workshops encouraged active participation from the personnel. 25,26Two additional meetings were held with managers and one union representative at the obstetrics intervention to support the implementation.In September 2019, a third reference group meeting was held focussing on following through with the implementation, and included evaluation of the process and descriptive data on the implementation so far.Adaptations of the interventions after this point were slight.After the final data collection, follow-up qualitative interviews were performed, reported elsewhere. 25inally, results were reported back to the workplaces.

The intervention content
Table 1 details the intervention content, which included changes to the physical environment, and organizational and psychosocial changes.All employees who were not on leave of absence were considered to have received theseworkplace level interventions.

Questionnaire surveys
Survey data was collected at three timepoints: baseline (t0), and two follow-ups approximately 3-4 months (t1), and 9 months (t2) after baseline.The first follow-up was performed after the workshop, during the implementation process, as per the participatory approach. 28ll personnel received a questionnaire and two reminders attheir home address or at work.Including items on hearing loss, tinnitus, difficulty perceiving speech at work, sound sensitivity (hyperacusis) and sound-induced auditory fatigue, previously diagnostically validated in the obstetrics care (except for difficulty perceiving speech at work previously assessed as a combination of work and leisure time), 29 and in Think Aloud interviews on preschool teachers. 8Noise annoyance at work was assessed using an item adapted for the workplace, based on the ISO/TS 15666:2003. 30Emotional exhaustion was assessed using the five items from the emotional exhaustion subscale of the Maslach Burnout Inventory general survey (MBI-GS), 31 in Swedish, 32 with a modified 6-point response scale,excluding the alternative "A few times each week".For MBI-GS, internal missing were replaced by the mean if three or more items were reported.All response alternatives and the categorization defining the analyzed variables (binary, ordinal or scale) are detailed in Table 2.The survey also included items measuring potential confounders, for example age, noise at work, manager support, general health, as detailed in supplement A.

Sound level measurements
Sound levels were measured each timepoint during one work week (five days in preschools and seven days in the obstetrics).The SVANTEK SV 104 (IEC 61252 − 2002, Class 2, IEC 61672-1 -2013, range 43 to 141 dBA Peak) dosimeters were used, except at baseline in the obstetrics intervention where the Larson Davis 705þ (IEC 61252 − 1993, Class 2, range 40 to 143 dBA Peak) was used.Sound levels were logged every 30 s. Personnel wore the dosimeter on the right shoulder close to the ear and wrote a log including start and stop times and work activities.
Equivalent daily A-weighted decibel level (dB LAeq), minimum two hours, and the highest Aweighted fast maximum decibel level (dB LAFmax) per measurement, were calculated from eight dosimeters per day in preschool and from ten dosimeters in obstetrics care.Work shifts in obstetrics (day, evening, and night) were similar and thus combined.Calculations were done in Python, after data managing in excel to exclude loggings when the dosimeter was not reported to be worn orwhen the personnel left the workplace (e.g., being outside, or at a nearby ward) was excluded.Coffee and lunch breaks were included to reflect the individual dose.

Statistical analysis
Baseline characteristics and prevalence are presented as valid percentages for categorical variables, and as mean and standard deviation (SD) for continuous variables.Repeatedly measured outcomes were analyzed using the generalized estimating equation (GEE), including participants who responded at baseline and at least one of the two follow-ups.Each workplace setting was analyzed separately.Response categories from the survey were merged if there were few responses, and assessed as either binary or ordinal, as detailed in Table 2.Each outcome was analyzed in separate models, controlling for its baseline value.Potential confounders were selectedusinga theoretical approach, then assessed statistically (forward stepwise) and kept if beta changed �15%.A timepoint variable (follow-up 1 or follow-up 2) was analyzed to assess change over time, and included if beta changed �15%.Dosimeter data was analyzed for each setting separately by comparing the group average equivalent (dB LAeq) and maximum (dB LAFmax) sound levels between the intervention and the control group at each timepoint using t-test with Bonferroni correction, and over timepoints within each group using One-way ANOVA with post hoc pairwise comparisons using Bonferroni.
All tests were two-sided, with p-value <0.05 considered statistically significant.Analyses were performed using IBM SPSS Statistics 28.0 and Stata/ SE 17.0.

Results
Figure 2 details the number of participants responding to the survey at each timepoint.The response rate at baseline in preschool was 96% in the intervention group and 82% in the control group, and in obstetrics (including hourly employees) 49% in the intervention group and 73% in the control group.The response rates decreased over time.The preschool intervention group was slightly older (mean 49 years) than the controls (mean 38 years).In the obstetrics the mean age was 50 years in both groups.
Table 2 shows that baseline outcomes were slightly more prevalent and with more frequently occurring symptoms in both intervention groups compared to their control.The results from the GEE analysis (Table 3) did not show any significant improvement from the intervention (a negative b would indicate an improvement in the outcome, whereas a positive b indicates a worsening in the outcome).Hyperacusis changed in a direction toward improvement in the obstetrics setting, but the model was not significant.A similar trend was seen for difficulty perceiving speech in obstetrics and tinnitus in preschool.In contrast, a significant worsening was found for noise annoyance in both settings, although descriptive data for the intervention group in the obstetrics care showed less worsening than the control group (Figure 3).Hyperacusis, sound-induced auditory fatigue, and emotional exhaustion also showed a significant worsening in preschool.There was no significant effect of time.
Figure 4 shows that equivalent sound levels was slightly higher in the preschools compared to the obstetric wards, and that in the preschool setting the mean equivalent levels increased significantly in the intervention group from baseline to follow-up 2 with a medium effect size of the overall model (ANOVA p ¼ 0.005, post hoc t0-t2 p ¼ 0.004).As shown in Table 4, there were no other significant differences.
Figure 5 illustrates that a larger proportion reported improvements in sound environment than in psychosocial work environment in both intervention groups.Notably, at the final followup a third of the preschool group and a third of the obstetrics group perceived a worsening of sound environment and psychosocial environment respectively.

Discussion
The main result was the significant worsening after the intervention in some outcomes, instead of the hypothesized improvement.In the obstetrics, only noise annoyance worsened significantly.In the preschool, worsening of hyperacusis, sound-induced auditory fatigue, noise annoyance and emotional exhaustion, and increased equivalent sound levels was found.The worsening is partly explained by a larger improvement in the control groups.If starting with worse values makes it harder to improve, the effect could also be partly explained by worse baseline values in the intervention groups.Control workplaces were selected to match the intervention workplaces and analyses were adjusted for potential confounding differences, but differences may still have influenced.The significantly increased   sound levels in preschool supports the hypothesis of a causal relationship with health outcomes, but the worsening may also be a result of the intervention raising awareness of the health risks of noise exposure.
A lack of resources and support for implementation may have decreased the intervention effect.This is supported by a concurrent qualitative evaluation, 27 where personnel expressed positive attitudes toward the changes but also described hindering factors.An earlier study in the same population highlighted that improvements were much wanted, 3 which may have increased the expectations of improvement.Blinded or doubleblinded occupational interventions are nearly impossible, 33 and high expectations have been suggested to explain worsening of outcomes in participatory interventions, 26 as have employee disappointment due to poor implementation. 12hile there are studies aimed at improving the sound environment in hospital settings such as intensive care, 34 we have found no studies from the obstetrics care.Previous interventions in preschool have shown divergent results, but we still expected improvement particularly in noise annoyance, 18 and sound-induced auditory fatigue, 35 and decreased sound levels. 6,19,20The sound levels were comparable to previous studies both in preschools and obstetrics care, 7,18 and may be high enough to pose a risk to hearing.Previous interventions did not assess hearing-related symptoms, but one assessed whether hearing protection  devices decreased personnel burnout. 21Similar to our study, they found a non-significant increase in burnout, while another preschool study found significant increased stress after interventions aimed at improving the sound environment. 20t has previously been suggested that organizational needs will be prioritized over needs of a research intervention study. 36We argue that high work demands may impact particularly participatory-based interventions where personnel and managers are responsible for implementation.Fewer intervention measures were directed at the psychosocial work environment, despite of, or due to, the increasingly demanding work environment with cutbacks, lay-offs, and increased workload. 3,27his may have overshadowed improvements from the sound environment interventions, and it could explain the increased emotional exhaustion.The personnel have described that noise causes stress but also that stressful working conditions made the noise more difficult to handle, 3 which potentially explain the increased noise in the preschool.We hypothesize that stress and noise have a synergistic effect on health, as reported for environmental noise and job strain, 37 partly through noise annoyance causing a stress reaction, 38 as described in the noise-stress model. 12

Strengths and limitations
A limited sample size may cause type-II errors.Confidence intervals were wide and estimates uncertain, but descriptive data of individual change generally supported the statistical analysis.The limited sample motivated analysis of both follow-ups, although implementation was ongoing at the first follow-up, and motivated including individuals responding at only one follow-up.This may mean that the full or final intervention effect was not analyzed, though no significant time effect was found.The limited number of intervention sites and the purposive selection could introduce bias and hinder generalization.Hence, further studies could confirm or counter the results from this first trial using these participatory-developed interventions.
There were fundamental similarities in the interventions in the two settings, matching the similar work environments.A strength of the participatory methodis context-specific interventions, 26 and high measure-to-challenge correspondence. 22he approach is however limited in estimating effects of specific measures.Randomized controlled trials (RCTs) are still considered gold standard, but are difficult to implement in occupational settings and has limitations in real-world effectiveness. 33he participatory method requires consideration of the personnel's influence over and involvement in the intervention process. 28nvolvement was mainly representative via the reference group.Representative involvement has however not been shown to result in less influence or participation. 28The workplaces had limited influence over the process.It is common, 28 and it may have influenced the results since process participation is associated with the perception of change and thus with the intervention outcome. 39nother potential weakness is that organizational influence was not a goal in itself.It has been suggested important for sustainable change. 28he accompanying qualitative evaluation reported elsewhere, 27 is an important strength of the overall project as it provides a nuanced and multifaceted analysis of the personnel's perceptions of the intervention process and outcome.It showed that the overall experience was positive. 27ence, we suggest further studies using the participatory approach to develop innovative, contextspecific, and comprehensive interventions, to meet the complex issues in these communication-intense settings.These should consider and analyze the process and degree of implementation, personnel participation and expectations, and the potentially hindering effect of concurrent work-related factors such as stress and high workload.Although difficult, follow-up studies should evaluated the interventions in larger samples, and ideally at more intervention sites with randomized assignment to the intervention group.

Conclusions
This study could not show significant improvement from the intervention.Contrary, a significant worsening in hyperacusis, sound-induced auditory fatigue, emotional exhaustion and noise annoyance, and increased equivalent sound levels was found in preschool, and worsening noise annoyance was found in obstetrics.The results may be explained by larger improvements in controls, and if starting with worse values makes it harder to improve, the effect could also be partly explained by worse baseline values in the intervention groups.Raised awareness of risks in the intervention group may initially inflated the outcomes, but the increase in sound levels and concurrent worsening in outcomes also supports a causal relationship.A lack of psychosocial interventions may explain the worsening of emotional exhaustion.The complexity of the demands in communication-intense human-service workplaces calls for further studies on sustainable interventions.

Figure 1 .
Figure 1.Timeline of the study phases, the intervention activities, and the data collection.The control workplaces participated only in the data collection (dark grey), while the intervention workplaces participated in all the intervention activities and the data collection.

Figure 2 .
Figure 2. Participant flow diagram detailing number of personnel enrolled, lost to follow-up, responding at each timepoint, and the final study sample included in the analysis.

Figure 3 .
Figure 3. Descriptive data of the individual change in the outcomes as valid percentages reporting improvement (green), worsening (orange) or remaining unchanged (grey) from baseline to follow up (merging results from t1 and t2) within the intervention group (I) and control group (C) respectively, in the obstetrics care setting (left) and in the preschool setting (right).

Figure 4 .
Figure 4. Boxplots of daily A-weighted equivalent sound levels (dB LAeq) from dosimeter measurements in the obstetrics care (left) and in the preschools (right).

Figure 5 .
Figure 5. Descriptive responses of perceived changes in the work environment within the intervention groups, showing improvement (green), worsening (orange) or remaining unchanged (grey) from baseline (t0) to the first follow-up (t1), and between the two follow-ups (t1 and t2) in the obstetrics care setting (left) and in the preschool setting (right).

Table 1 .
The selected and prioritized context-specific intervention contents implemented in the obstetrics care and the preschool setting.Awareness of sound and noise, when for example buying new toys.Awareness was also increased for the kitchen staff.� New group composition at lunch.� Fewer children in hallway and dividing the children into small groups indoors.� Own voice use, not shouting, staff making a good example for children.� Discussing work responsibilities and how to support each other.� Workplace meeting discussions and reflections about sound environment and work environment.

Table 2 .
Baseline values for the outcome variables with categories or scale as analyzed, shown as valid column percentages (% � ), or mean and standard deviation (SD), within the obstetrics setting and the preschool setting, separating the intervention workplace from the controls.

Table 3 .
Generalized estimating equation (GEE) analyzing the effect of intervention by comparing outcomes at follow-up in the intervention versus the control group in the obstetrics and the preschool setting respectively (control group as reference).Definitions of outcome variable categories for binary outcomes and levels of ordinal outcomes are given in Table 2, second column from the left.a Crude models includes the baseline outcomes.Adjusted models in obstetrics were adjusted for: Age; Noise exposure at work; Percentage of full-time work; Noise annoyance; Disturbing, distracting, attention demanding sounds; Influence on resources; Time for reflecting about work; and Support from manager, in addition to baseline outcome.Adjusted models in preschool were adjusted for: Age; Noise exposure at work; Percentage of full-time work; and Time for reflecting about work, in addition to baseline outcome.b Crude models includes the baseline outcome.Adjusted model in obstetrics were adjusted for: Age and Noise exposure, in addition to baseline outcome.Adjusted model in preschool were adjusted for: Age; Noise exposure at work; Hearing protection at work; Disturbing, distracting, attention demanding sounds; Influence on resources; and Putting oneself aside, in addition to baseline outcome.c Crude models includes the baseline outcome.Adjusted model in obstetrics were adjusted for: Noise annoyance; General health; Division of household work and Recovery after work, in addition to baseline outcome.Crude model was the final model in preschool.� Maximum likelihood estimates not available (NA) due to quasi-complete separation (separation means that the predictor, in this case intervention group, is associated with only one outcome value). §

Table 4 .
Descriptive data and analysis of average sound levels within each group over timepoints and between control and intervention group at each timepoint.