Effectiveness of educational programs in hearing health: a systematic review and meta-analysis

Abstract

Objective

To evaluate the effectiveness of hearing health education programs aimed at preventing noise-induced hearing loss (NIHL), of recreational and occupational origin, by means of a systematic review and meta-analysis.

Design

The search strategy was carried out in on five electronic databases, as well as referrals from experts. The risk of bias was judged, and the random effects meta-analysis was performed. The certainty of the evidence was assessed.

Study sample

Effectiveness studies that used educational intervention in hearing health and prevention of NIHL were included.

Results

42 studies were included. The Dangerous Decibels program was the only one that could be quantitatively analysed and showed improvement in the post-intervention period of up to one week [SMD = 0.60; CI95% = 0.38–0.82; I2 = 92.5%) and after eight weeks [SMD = 0.45; CI95% = 0.26–0.63; I2 = 81.6%) compared to the baseline. The certainty of evidence was judged as very low.

Conclusions

The Dangerous Decibels program is effective after eight weeks of intervention. The other programs cannot be quantified. They still present uncertainty about their effectiveness. The level of certainty is still low for this assessment.

Keywords:

• Hearing
• education
• effectiveness
• noise-induced hearing loss
• health education
• systematic review

Introduction

The World Report on Hearing released by the World Health Organisation (WHO) in 2021 highlighted environmental and occupational noise as an important risk factor for hearing. The report suggested educational interventions to promote hearing health and prevent Noise-induced hearing loss (NIHL) in exposed populations (WHO 2021). In the last few years, many educational programs aiming to promote hearing health among children, adolescents or adults exposed to noise have been implemented using different educational methods and strategies (Khan, Bielko, and McCullagh 2018; Tikka et al. 2017). However, there are data gaps regarding the effectiveness of these programs in reducing the negative effects of exposure to harmful noise and in behavioural and habit changes amongst people.

Effectiveness is defined as an improvement in health or behaviour because of an intervention in a group or community (Truman et al. 2000). For a health promotion intervention to be effective, it must incorporate a range of educational strategies at the individual level while enabling support at the collective level to reinforce and encourage positive health attitudes. In addition, there should be potential for participants to spread health promotion messages to families and social networks at the community level. Thus, effectiveness depends directly on how the target audience acts and responds to the intervention (Noar and Corcoran 2007).

A systematic review (Khan, Bielko, and McCullagh 2018) assessed educational programs developed in environmental and occupational hearing health. It stated that intervention studies aimed at self-care and promotion of healthy behaviours related to hearing health usually aim to collect information from sound, hearing, awareness of risk factors (including environmental/leisure and occupational noise) and prevention of NIHL (including the supply of hearing protectors, facilitating feedback and communication training). However, the focus of that review was the description of intervention methodologies, with limited evaluation of their real effectiveness and no effective comparison between results was measured. In addition, there was a language restriction on the articles included, thus increasing the risk of publication bias.

Two other reviews related to the prevention of occupational hearing loss in adults were published (Brennan-Jones et al. 2020; Tikka et al. 2017). These Cochrane reviews identified specific strategies that have demonstrated effectiveness in reducing noise in the workplace. A more comprehensive systematic review and meta- analysis on this topic is justified to assess the effectiveness of these programs.

Therefore, this systematic review and meta-analysis aim to evaluate the effectiveness of hearing health education programs aimed at preventing noise-induced hearing loss (NIHL), of recreational and occupational origin, a broad demographic range by means of a systematic review and meta-analysis.

Material and method

This systematic review and meta-analysis were carried out in compliance with the PRISMA 2020 standard. (Preferred Reporting Items for Systematic Reviews and Meta-Analysis Checklist) (Page et al. 2021).

Eligibility criteria

The following targeted question was developed to evaluate the proposed objective: “How effective are hearing health education programs aimed at preventing NIHL?”

To consider the eligibility of studies to be included/excluded in this review, the acronym “PICOS” was used:

• Population (P) – general population.

• Intervention (I) – educational programs in hearing health that aim to prevent NIHL.

• Comparison (C) – non-participants in educational actions or “before and after” comparisons of the intervention.

• Outcome (O) – effectiveness of educational programs in hearing health.

• Study design (S) – randomised, quasi-randomised, or non-randomised clinical trials.

Inclusion criteria

Studies were included if their outcome was the evaluation of effectiveness and that used education intervention methods in hearing health and prevention of NIHL in children, adolescents, adults and the elderly. Randomised, quasi-randomised and non-randomised controlled clinical trials were included. Studies published in all languages were included with no restrictions on gender or time of publication.

Exclusion criteria

Studies were excluded based on the following criteria:

1. Studies that did not assess the effectiveness of hearing health education strategies;

2. Descriptive studies, reviews, book chapters, letters, editorials, expert opinions/personal opinions, conference and/or event summaries, case reports, or case series;

3. Studies referring to the development of protocol strategies in hearing health education (technical issues and/or pilot projects) and studies that presented only preliminary results of initial stages of development of educational strategies in hearing health education;

4. Studies with incomplete or unavailable data, after contact with authors.

Information sources and search strategy

Appropriate word combinations and truncations were selected and adapted for each of the following electronic databases: Embase, Latin American and Caribbean Health Sciences (LILACS), PubMed/Medline, Scopus and Web of Science and four gray literature databases (ASHAWIRE, Google Scholar, OpenGrey and ProQuest Dissertation and Thesis). Online Appendix 1 shows additional information on search strategies for all databases. In addition, a manual search of reference lists of selected articles was performed to obtain additional literature. Relevant studies on the subject were also requested from experts in the field. References were verified, and duplicate items were removed using the EndNote^® software (EndNote^® Basic X7 Thompson Reuters, New York, NY, USA). Searches were performed on 5 July 2020 and updated on 12 July 2022.

Selection process

The selection of articles was carried out in two phases. In phase 1, two reviewers (L.B. and L.M.A.G.) independently reviewed the titles and abstracts of all references. All articles that did not meet the inclusion criteria were excluded. In phase 2, the same reviewers independently read the selected articles in full. In case of disagreement, and when it was not resolved through discussion between the first and second reviewer, a third author (L.S.) was involved in the final decision.

To shield the reading of references and ensure independence and secrecy in both phases, the Rayyan website (http://rayyan.qcri.org) (Ouzzani et al. 2016) was used. The reviewers were shielded in all evaluations, and a member of the team (A.B.M.L.), who did not participate in the selection, acted as moderator.

Data collection process

Two reviewers (L.B. and L.M.A.G.) collected information from the studies included, and this information was discussed. The data collected consisted of study characteristics (author, year of publication, country, study design), population characteristics (sample size, age range), evaluation characteristics (composition of control and intervention groups, parameters of interest, method of evaluating outcomes, index used for evaluation), characteristics of results (results presented in relation to outcomes) and conclusions. Attempts were made to contact the authors and retrieve any unpublished data if the required data were incomplete. Three attempts were made to contact the first author, the corresponding author and the last author of the article. The time interval between attempts was one week.

Data Items

The interest outcome was the effectiveness of educational programs in hearing health that aim to prevent noise-induced hearing loss.

For studies in which the applied tool provided results through scores, mean values, standard deviation and sample size, the outcomes were extracted from the studies included in the synthesis of each group (control and experimental) or between the different times (pre- and post-intervention). In addition, the p values were also extracted for all comparisons.

Assessment of bias risk in the studies

The studies included were assessed as for methodological quality using the Joanna Briggs Institute Critical Appraisal Checklist (The Joanna Briggs Institute 2021). This tool covers 13 domains for randomised studies and nine domains for non-randomised studies. The judgement regarding the possible risk of bias in each of these domains was made by two independent reviewers (L.B. and L.M.A.G.), who used critical evaluation criteria to analyse all articles included, marking each criterion with “yes” or “no.” If not enough detail was reported in the study, the risk of bias was judged as “unclear,” and the authors of the original study were contacted for further information. The studies included were judged as “high risk,” “moderate risk,” and “low risk” when the domains with “yes” answers represented 0–49%, 50–69%, 70% or more, respectively, of the other domains (de Araujo et al. 2020). When necessary, disagreements were resolved through discussion with a third researcher (A.B.M.L.). Risk of bias graphs for all included studies were generated using the Robvis Tool. (https://mcguinlu.shinyapps.io/robvis/).

Synthesis method and summary effect

To measure the effect size between the pre- and post-intervention moments, a meta-analysis of random effects, weighted by the inverse variance method, was performed. As data from a same educational intervention were presented on different scales, the standardised difference between means (SMD) was used, measuring the observed effect size (G of Hedges) between the two moments. SMD values of 0.2–0.5, 0.5–0.8 and >0.8 were considered as small, moderate and large effect sizes, respectively (Andrade 2020). Heterogeneity was evaluated by the inconsistency index (I²), Cochran’s Q test and variance values (Tau²), estimated using the restricted maximum likelihood method.

To enable a broader analysis and interpretation of the global effect size for these comparisons, an analysis of the p values was performed using the Albatross plot (Harrison et al. 2017).

All analyses and graphs were performed and created using the Stata software, version 16.0 (Stata Corp LLC, College Station, USA). The significance level adopted for all analyses was 5%, and the respective 95% confidence intervals (95% CI) were calculated.

Reporting bias

The existence of publication bias was assessed using the funnel plot and the Egger test, considering a significance level of 5%.

Assessing the certainty of evidence

The certainty of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE). The certainty of the evidence generated was judged as high, moderate, low or very low, considering five domains: risk of bias, inconsistency, imprecision, indirect evidence and publication bias (Guyatt et al. 2008).

Results

Selection of studies

The search strategy carried out in the scientific databases resulted in 1853 articles. Excluding 197 duplicate articles, 1656 articles were selected for title and abstract reading. Of these articles, 42 were selected for full reading (phase 2), of which 13 were excluded (Online Appendix 2). Two articles were added after the search was updated (October 2022). Another seven articles were included from the search in the gray literature and search for manual references and four works were indicated by experts, thus resulting in 42 articles included for the synthesis (Figure 1).

Figure 1. Prisma 2020 flow diagram for new systematic reviews which included searches of databases, registers and other sources.

Characteristics of studies

Among the 42 articles included, twenty-nine articles were published in English, twelve were published in Portuguese and one in Spanish. The following countries were their sources: Australia, Belgium, Brazil, Canada, Cuba, United States, New Zealand, Sweden and Turkey. The publication year varied between 1990 and 2021, with a greater number of Brazilian articles.

Sample sizes varied between 14 and 5013 participants, aged between five and 65 years old, divided into 26 studies in the adult age group, 16 studies in children and adolescents’ group and one at the community level including different age groups, totalling 42 studies. As for gender, there was a predominance of males in the samples of the 42 studies included.

In terms of study design, nine were classified as randomised (Cavallari et al. 2021; Gomes et al. 2023; Hong et al. 2006; Knobloch and Broste 1998; Lusk et al. 2004; Martin et al. 2013; Neufeld et al. 2011; Rocha et al. 2011; Seixas et al. 2011) and 33 as non-randomised (Alvarenga et al. 2008; Araújo et al. 2013; Blasca et al. 2013; Bramati et al. 2022; Bramatti, Morata, and Marques 2008; Bulunuz et al. 2017; Elander and Hellström 1995; Ewigman et al. 1990; Gates and Jones 2007; Gilles and Paul 2014; Gonçalves et al. 2009; Gondim 2022; Griest, Folmer, and Martin 2007; Heupa et al. 2011; Incekar et al. 2019; Knobel and Lima 2014; Lacerda et al. 2013, 2015; Lopes 2021; Moreira and Gonçalves 2014; Neitzel et al. 2008; O’Brien, Driscoll, and Ackermann 2015; Piccino 2019; Reddy et al. 2017; Rodríguez Aldana et al. 2020; Samelli et al. 2015; Saunders et al. 2015; Taborda et al. 2021; Takada et al. 2020; Taljaard, Leishman, and Eikelboom 2013; Trabeau et al. 2008; Voaklander et al. 2009; Welch et al. 2016).

In terms of the participants of the included studies, there was 25 studies with a population of workers (Alvarenga et al. 2008; Araújo et al. 2013; Bramati et al. 2022; Bramatti, Morata, and Marques 2008; Cavallari et al. 2021; Elander and Hellström 1995; Ewigman et al. 1990; Gates and Jones 2007; Gonçalves et al. 2009; Heupa et al. 2011; Hong et al. 2006; Incekar et al. 2019; Lusk et al. 2004; Moreira and Gonçalves 2014; Neitzel et al. 2008; O’Brien, Driscoll, and Ackermann 2015; Reddy et al. 2017; Rocha et al. 2011; Samelli et al. 2015; Saunders et al. 2015; Seixas et al. 2011; Takada et al. 2020; Trabeau et al. 2008; Voaklander et al. 2009; Welch et al. 2016).

Sixteen studies with school children (children and adolescents) (Blasca et al. 2013; Bulunuz et al. 2017; Gilles and Paul 2014; Gomes et al. 2023; Gondim 2022; Griest, Folmer, and Martin 2007; Knobel and Lima 2014; Knobloch and Broste 1998; Lacerda et al. 2013, 2015; Lopes 2021; Martin et al. 2013; Neufeld et al. 2011; Piccino 2019; Taborda et al. 2021; Taljaard, Leishman, and Eikelboom 2013) and one study included the general community (Rodríguez Aldana et al. 2020).

The educational strategies used were all interactive, but with a variable content and different methods in their structures and applications. Eighteen studies used different educational strategies – using participative methodologies (Moreira and Gonçalves 2014), inductive-deductive methods (Rodríguez Aldana et al. 2020), with measurement of environmental noise (Bulunuz et al. 2017; Elander and Hellström 1995; Gates and Jones 2007; Incekar et al. 2019; Knobloch and Broste 1998; Lacerda et al. 2015; Rodríguez Aldana et al. 2020), carrying out audiological tests (in order to draw more attention to the problem of noise and its repercussions on hearing) (Lacerda et al. 2015), through tests with objective data (Knobloch and Broste 1998), workshops (Gonçalves et al. 2009; Lacerda et al. 2013; Moreira and Gonçalves 2014), community integration (Rodríguez Aldana et al. 2020), management (Rodríguez Aldana et al. 2020), training based on protocols (Araújo et al. 2013; Bramati et al. 2022; Gonçalves et al. 2009; Rocha et al. 2011; Samelli et al. 2015; Seixas et al. 2011; Takada et al. 2020; Trabeau et al. 2008), reminders with brochures sent via postal service (Gates and Jones 2007), in addition to standardised education and hearing conservation programs (Ewigman et al. 1990; Incekar et al. 2019; Knobloch and Broste 1998; Lacerda et al. 2015; Neitzel et al. 2008; O’Brien, Driscoll, and Ackermann 2015), theatre, conversation circles, games and music (Lacerda et al. 2013, 2015).

Five studies were lectures with the use of graphic materials, printed resources, figures and texts in conversation forms, handouts adapted from contents of the World Health Organisation and flip charts (Alvarenga et al. 2008; Blasca et al. 2013; Knobloch and Broste 1998; Neitzel et al. 2008; Rocha et al. 2011). Another six studies used audiovisual material and multimedia interaction (interactive CDs, tele-education, campaigns in various telecommunication media - television, radio, internet – twitter, Facebook, websites), applications and computer programs (Araújo et al. 2013; Gilles and Paul 2014; Lusk et al. 2004; Martin et al. 2013; Saunders et al. 2015) and visual resources and animations (Araújo et al. 2013; Bulunuz et al. 2017; Elander and Hellström 1995; Gilles and Paul 2014; Hong et al. 2006; Lusk et al. 2004; Saunders et al. 2015). Three studies used interviews (individuals and focus groups) (Bulunuz et al. 2017; O’Brien, Driscoll, and Ackermann 2015; Saunders et al. 2015) and ten studies used the Dangerous Decibels^® (DD) program as an intervention strategy (Bramati et al. 2022; Gomes et al. 2023; Gondim 2022; Griest, Folmer, and Martin 2007; Knobel and Lima 2014; Lopes 2021; Martin et al. 2013; Piccino 2019; Reddy et al. 2017; Welch et al. 2016).

Risk of bias in studies

Of the eight randomised interventional studies included, six had a moderate risk of bias (Figure 2). The other 33 non-randomised interventional studies included all had a low risk of bias (Figure 3).

Figure 2. Risk of bias assessed by the Joanna Briggs Institute critical appraisal tool for randomised interventional studies.

Figure 3. Risk of bias assessed by the Joanna Briggs Institute critical appraisal tool for non-randomised interventional studies.

Results of individual studies

• Children and youth population

Sixteen studies included school children (children and adolescents) (Blasca et al. 2013; Bulunuz et al. 2017; Gilles and Paul 2014; Gomes et al. 2023; Gondim 2022; Griest, Folmer, and Martin 2007; Knobel and Lima 2014; Knobloch and Broste 1998; Lacerda et al. 2013, 2015; Lopes 2021; Martin et al. 2013; Neufeld et al. 2011; Piccino 2019; Taborda et al. 2021; Taljaard, Leishman, and Eikelboom 2013).

With the use of DD intervention, eight studies were identified among the 16 selected. DD is a program that uses various playful educational strategies to inform children, young people and adults about the effects of loud sounds on hearing and thus promote hearing health. Authors confirmed that the DD educational program is effective in improving students’ knowledge and attitudes towards noise and the use of hearing protection. (Gomes et al. 2023; Gondim 2022; Griest, Folmer, and Martin 2007; Knobel and Lima 2014; Lopes 2021; Martin et al. 2013; Piccino 2019; Welch et al. 2016)

Another nine studies used different strategies showed the effectiveness of interventions (Blasca et al. 2013; Bulunuz et al. 2017; Gilles and Paul 2014; Knobloch and Broste 1998; Lacerda et al. 2013, 2015; Neufeld et al. 2011; Taborda et al. 2021; Taljaard, Leishman, and Eikelboom 2013).

• Adult population

Twenty-five studies included studies with workers (Alvarenga et al. 2008; Araújo et al. 2013; Bramati et al. 2022; Bramatti, Morata, and Marques 2008; Cavallari et al. 2021; Elander and Hellström 1995; Ewigman et al. 1990; Gates and Jones 2007; Gonçalves et al. 2009; Heupa et al. 2011; Hong et al. 2006; Incekar et al. 2019; Lusk et al. 2004; Moreira and Gonçalves 2014; Neitzel et al. 2008; O’Brien, Driscoll, and Ackermann 2015; Reddy et al. 2017; Rocha et al. 2011; Samelli et al. 2015; Saunders et al. 2015; Seixas et al. 2011; Takada et al. 2020; Trabeau et al. 2008; Voaklander et al. 2009; Welch et al. 2016).

Among them, two used the Dangerous Decibel program for workers (Bramati et al. 2022; Reddy et al. 2017), especially the Ecological Model of Health Promotion with a Multilevel Approach to Community Health. The developed program offers a simple, interactive, theory-based intervention that is well accepted and effective in promoting positive auditory health behaviours in the workplace.

The other studies used other approaches, showing the effectiveness of training (Alvarenga et al. 2008; Bramatti, Morata, and Marques 2008; Bulunuz et al. 2017; Cavallari et al. 2021; Elander and Hellström 1995; Ewigman et al. 1990; Gates and Jones 2007; Gilles and Paul 2014; Gonçalves et al. 2009; Heupa et al. 2011; Hong et al. 2006; Incekar et al. 2019; Lusk et al. 2004; Moreira and Gonçalves 2014; O’Brien, Driscoll, and Ackermann 2015; Rocha et al. 2011; Rodríguez Aldana et al. 2020; Samelli et al. 2015; Saunders et al. 2015; Seixas et al. 2011; Takada et al. 2020; Trabeau et al. 2008; Voaklander et al. 2009).

Table 1 (Supplemental Online Material) summarises the individual characteristics of the included studies.

Summary of results

Seven articles were included in the quantitative synthesis, all of which used the DD program as an educational strategy. Data for performing the meta-analysis were obtained by contacting the studies’ authors (Bramati et al. 2022; Gomes et al. 2023; Gondim 2022; Lopes 2021; Piccino 2019; Reddy et al. 2017; Welch et al. 2016). Griest, Folmer, and Martin (2007), Martin et al. (2013) and Knobel and Lima (2014) were not included because the data was not sent (Figure 4).

Figure 4. Forest plot of the meta‐analysis evaluating the scores comparing the pre-intervention moments and scores after up to one week of intervention of DD program.

When evaluating the scores comparing the pre-intervention moments and scores after up to one week of intervention, there was a moderate effect size between the two moments [SMD = 0.60; CI95% = 0.38–0.82; I² = 92.5%). The largest effect size occurred in the knowledge domain [SMD = 1.12; CI95% = 0.67–1.57; I² = 91.17%]. There was a large effect size for this comparison (Figure 4). Of the DD program domains in which studies reported p values for comparison pre-intervention and after one week, 80% of them showed statistical significance. Most domains that did not show statistical significance in this comparison were associated with a smaller sample size (Figure 5).

Figure 5. Forest plot of the meta‐analysis evaluating the scores comparing the pre-intervention moments and scores after a period of more than eight weeks post-intervention of DD program.

When the same domains were evaluated over a period of more than eight weeks post-intervention, the overall effect has diminished, but there is still a difference compared to the reference situation. [SMD = 0.45; CI95% = 0.26–0.63; I² = 81.6%) (Bramati et al. 2022; Gomes et al. 2023; Gondim 2022; Lopes 2021; Piccino 2019; Reddy et al. 2017; Welch et al. 2016). Likewise, the largest effect size occurred in the knowledge domain [SMD = 0.93; CI95% = 0.65–1.22; I² = 70.75%]. When considering the p values the studies reported for the DD domains in a period of eight weeks after intervention, studies with a small sample size showed statistical significance, especially as for the knowledge domain (Gomes et al. 2023; Gondim 2022; Lopes 2021; Piccino 2019; Reddy et al. 2017; Welch et al. 2016).

Reporting bias assessment

There was no significant publication bias through funnel plot analysis and Egger’s test (p > 0.05) (Online Appendix 3).

Certainty of evidence

The certainty of evidence was judged as very low for all analyses. The decrease in the certainty of evidence was because of the design of the studies included in the analysis (lack of randomisation in most of the studies included) and the high heterogeneity existing in both analyses (Online Appendix 4).

Discussion

The aim of this systematic review and meta-analysis was to evaluate the effectiveness of educational programs in hearing health aimed at the general population. Thus, 42 studies included in this systematic review met the eligibility criteria, out of a total of 1853 retrieved articles. The DD program was the only that could be analysed quantitatively, showing a positive impact in a period longer than eight weeks post-intervention. The greatest positive impact has been in the field of knowledge.

When considering the 42 included studies (Figure 1), all were effective regardless of the educational strategies used. There was a greater number of studies focused on the worker’s hearing health in the adult population (Table 1, Supplemental Online Material). In this respect, studies published between 1990 and 2010 show a predominance and greater concentration, among the selected studies, in the occupational field and among adults/workers. The greater production of articles analysing children and adolescents/school children was in the last 10 years (from 2010 to 2021). This follows the trend of the WHO reports, guidelines and recommendations (2021). The initial focus is on actions to reduce occupational noise, followed by a growing concern with environmental and leisure noise in schools and in the population of children and teenagers. The wide variety of educational strategies used, as well as the way of analysing the effectiveness of interventions, make methodological comparisons between studies (Table 1, Supplemental Online Material).

Among the studies that used the same methodology, ten studies that used the Dangerous Decibels^® (DD) program stood out, eight of which were applied to children and teenagers (Gomes et al. 2023; Gondim 2022; Griest, Folmer, and Martin 2007; Knobel and Lima 2014; Lopes 2021; Martin et al. 2013; Piccino 2019; Welch et al. 2016) and two to workers (Bramati et al. 2022; Reddy et al. 2017). All were effective; however, in some of the evaluated questions (risk habits - exposure to noise, attitudes, trends and barriers to the use of hearing protectors and behaviour towards noise), the results were not significant in the three- and six-month follow-ups after interventions (Figures 4, 5, and 6).

Figure 6. Albatross plot evaluating the p values for the comparison between pre-intervention and after one week, for the domains of the DD program.

The Dangerous Decibels^® (DD) Program is a public health program that exists since 1999 and is designed to reduce the incidence and prevalence of tinnitus and NIHL in children and young people. The educational program is based on the theory of cognitive behaviour with interactive strategies that could favour the increase of knowledge and the establishment or change in habits and behaviours. Its effectiveness has already been validated in different contexts.

The basic information, such as “What do we hear?”, “How do we hear?”, “How do our ears get damaged?”, “What is it like to have damaged ears?”, “How can I protect myself?” and the main messages of the program, such as “What are the sources of dangerous sounds?”, “What are the consequences of being exposed to dangerous sounds?”, “How do I protect myself from dangerous sounds?”, remained the same regardless of whether the population was schoolchildren or workers. Some examples used during the presentation changed according to the target population.

In the present study, the DD program proved effective in modifying knowledge, attitudes and behaviours related to noise and hearing loss when compared before the intervention and after one week. However, the improvements in intended attitudes and behaviours decreased after three months (Figures 4, 5, and 6). This shows the need for systematic interventions throughout the school year and beyond the 45-min classroom program. Some other suggestions for the continuity of the program in the classroom are adding DD to the Pedagogical Plan of schools, use of other strategies (such as the dummy Jolene, to evaluate the sound pressure level of headphones in games, among other resources), inclusion of topics related to noise and participation in scientific, physical, acoustic and environmental fairs, music festivals and/or other events foreseen in the school calendars, seeking to articulate actions in a multi- and transdisciplinary way and in dialogue with the realities and the "operating mode" of each school, municipality and country. The best prospects for the results of educational interventions when they are multimodal and continuous.

Still, there were some limitations in this SR, such as lack of detailing of interventions performed, wide variety of statistical analyses, many studies did not describe the statistical tests clearly, lack of a control group and/or paired groups, absence of some information such as age and its variations, in addition to the means and standard deviations of the results. It is suggested that instruments be adapted to analyse risks of bias specifically aimed at intervention studies of educational programs given that the instruments focus more on analysis of drug therapies and therapeutic procedures, making it difficult to access responses to effectiveness assessments for educational interventions.

The results corroborate, in part, with a systematic review published in 2018 (Khan, Bielko, and McCullagh 2018). Where the authors verified that the educational programs developed in hearing health had their real effectiveness evaluation limited by the difficulty of comparison of results due to the diversity of methodologies and intervention strategies used. Some limitations should be considered in studies designed with pre- and post-educational intervention assessments, such as the probability of existence of repetition test bias and the Hawthorne effect. This open spaces for changing responses in the post-test due to the memorisation of errors existing in the pre-test or to the modification of behaviour by simply observing the subject (Ho et al. 2018).

The present study also corroborates the findings of the Cochraine reviews (Brennan-Jones et al. 2020; Tikka et al. 2017), which revealed a lack of conclusive evidence in the studies, but all of which demonstrated efficacy. It is suggested to conduct future studies with longer follow-up periods and greater methodological detail.

Conclusion

The Dangerous Decibels program is effective after eight weeks of intervention. The other programs cannot be quantified. They still present uncertainty about their effectiveness. The level of certainty is still low for this assessment.

Register

The protocol for this systematic review was registered at the PROSPERO website (International prospective register of systematic review - Centre for Reviews and Dissemination University of York) - CRD42020202075.

Disclosure statement

No potential conflict of interest was reported by the author(s).