hearing impairment (patient), is hearing aid fitting (intervention) beneficial compared to no intervention (comparison) for improving self-reported hearing and communication difficulties in everyday life (outcome)?” Other PICO questions would be generated if he were particularly interested in one of the other intervention options or if he were completely unsure about which option to choose and wanted to know about the evidence for all options.
EBP Step 2: Access the Information
The second step in the EBP process is to assemble and evaluate the best evidence. This is accomplished by searching relevant databases. A good starting point is the Cochrane Database of Systematic Reviews (http://www.cochranelibrary.com), a resource for systematic reviews in health care and health policy. Cochrane Reviews uses a defined protocol to identify and select studies that meet specific quality criteria, ensuring the resulting review provides the best available evidence. Research evidence about audiologic interventions can also be found in citation search engines. Wong and Hickson (2012b) suggest the use of Scopus (http://www.scopus.com), PsychINFO (http://www.apa.org/psychinfo), the Cumulative Index to Nursing and Allied Health Literature (CINAHL) (http://www.cinahl.com), and PubMed (www.pubmed.gov). It is possible to search a number of databases at once by using systems such as Web of Science or Medline. Most of these databases require a subscription, but if you do not have access, then PubMed and Google Scholar are popular freely available search engines.
A good starting point for search terms to use in the databases is to take the keywords from the PICO question. For example, search terms to start with based on the PICO question above would be: mild hearing impairment, hearing aids, benefit, and outcomes. Alternative terms, such as mild hearing loss, should be included in the search, and limits can be placed on the search to refine the results obtained. It is also important to specify criteria such as the language of publication (e.g., English only), age of participants (e.g., adults only), the year of publication (e.g., from 2010 to present), and type of publication (e.g., systematic review and peer-reviewed journal article). A PubMed search using these suggested search terms and limits was conducted and identified 99 references that could provide evidence for the example PICO question. Some of the relevant systematic reviews and peer-reviewed journal articles identified in this search are discussed later in the chapter. There is a useful tutorial at the PubMed website to learn how to search in the database (http://www.nlm.nih.gov/bsd/disted/pubmedtutorial/). Databases such as Embase (http://www.embase.com) allow searches using PICO question formulation. It is also recommended to discuss search approaches initially with a librarian, as this can save valuable time in the long run.
EBP Step 3: Appraise the Articles Found
The next step involves determining the relevance of the articles you find to your clinical question, and then if they are relevant, critically reviewing them. Rating the quality of a particular piece of research is central to EBP. Table 27–1 shows a hierarchy of research evidence specific to the provision of hearing aids developed by Cox (2005).
Levels of Evidence
Level 1 evidence is available from systematic reviews, which are comprehensive evaluations of the literature that try to limit bias by systematically identifying, appraising, and summarizing research on a particular topic. Primary data sources are evaluated, and criteria for inclusion in such a review are predetermined and explicitly stated in the review. Such synthesis of research is important because single studies can and do provide conflicting evidence, and a systematic review can help establish why this occurs. Systematic reviews are also extremely useful for clinicians, as they provide a summary of the evidence on a particular clinical question. For example, the question proposed earlier was, “For patients with acquired mild sensorineural hearing impairment (patient), is hearing aid fitting (intervention) beneficial compared to no intervention (comparison) for improving self-reported hearing and communication difficulties in everyday life (outcome)?” In relation to this PICO question, a PubMed search identified the two highly relevant reviews: a Cochrane Review by Ferguson et al. (2017) and another recent systematic review, by Johnson, Danhauer, Ellis, and Jilla (2016). Findings of both reviews are discussed later in this chapter.
Systematic reviews aside, there are various types of experimental studies mentioned in Table 27–1 which may involve an intervention (randomized controlled trials and nonrandomized intervention studies) or be noninterventional or observational, such as cohort studies, case-control studies, and cross-sectional surveys. It is important to understand the differences between them. Many textbooks discuss these types of studies, and Cox (2005) provides an excellent review relevant to audiology. The first and most rigorous, from a scientific perspective, is the randomized controlled trial (RCT). In an RCT, participants who meet stated inclusion criteria are randomly allocated to groups receiving different forms of treatment or to a no-treatment or placebo treatment group. Humes et al. (2017) provide an example relevant to this chapter’s PICO question in an RCT where participants were randomly allocated to receive hearing aids with amplification programmed by an audiologist; hearing aids with amplification programmed by the participant themselves; or hearing aids programmed to provide no amplification. In an RCT, participants and the researchers who assess them should be blind to group membership, that is, not aware of which treatment a participant is receiving. This step is necessary to reduce any potential bias. If both participants and researchers are unaware of the treatment, the study will be described as a double-blind RCT. If the researcher knows the nature of the treatment but the participant does not, this is referred to as a single-blind RCT. Occasionally the term triple-blind is used. This refers to the fact that the participant, the researcher, and the person doing the final analysis of the data are all blind to the different types of treatment. Randomization of individuals to groups should mean that the different groups will be equal on all variables except the type of treatment, although this equality needs to be checked statistically. RCTs vary in quality depending on how careful the researchers have been to follow the guidelines for such studies (e.g., http://www.consort-statement.org).
A crossover study is another type of RCT and has some advantages because participants can act as their own controls and all participants receive the treatment in the end. In this type of study, participants are first randomly allocated to different groups as they are for an RCT and then one group begins a particular type of treatment and the other groups begin other treatments. At a certain point in time, those who have completed one form of treatment begin the other, so that all participants complete all treatments. A crossover study design can also be used to trial different hearing aids or different hearing aid features, one example being Cox, Johnson, and Xu (2016).
The main aim of the Cox, Johnson, and Xu’s (2016) study was to test the effectiveness of different hearing aids for adults with mild to moderate hearing impairment. Forty-five participants with an average age of 70 years, of which 26 had never worn amplification, trialed four pairs of hearing aids in a single-blind, crossover design. The hearing aids differed in features (basic versus advanced) and were from two different manufacturers. The group results showed that hearing-related quality of life was at least moderately better with the hearing aid compared with not using any amplification. The participants also reported the hearing aids were effective in speech clarity, listening fatigue, and sound comfort. However, data analysis did not show significant outcome differences between the four different hearing aids, and the participants’ subjective preferences were in line with this, as there was no clear trend preferring advanced features over basic features, or one manufacturer over another.
A lower level of evidence is available from cohort studies. These are common in hearing aid studies and involve the observation of the effects of a particular intervention in a single cohort or group of participants. Results for the experimental cohort are compared to a group who did not receive the intervention. The major difference between a cohort study and an RCT is that random allocation to groups does not occur, so there is the potential for bias and for results to be influenced by variables other than the type of intervention. One example of a cohort study is that of Collins, Souza, O’Neill, and Yueh (2007). The researchers retrospectively investigated hearing aid outcomes from 74 participants who attended either individual or group hearing aid fitting and follow-up visits. Collins et al. (2007) found that there was no statistical difference between the participants who attended group appointments in terms of a reduction in self-reported hearing difficulties (as measured by the Hearing Handicap Inventory for the Elderly [HHIE]; Ventry & Weinstein, 1982) when compared to the participants who were provided individual appointments. Additionally, the participants who attended group visits reported greater hearing aid satisfaction and hours of hearing aid use.
In case-control studies, participants with particular outcomes are grouped and studied retrospectively to investigate factors that influence outcomes. One group (cases) could be people who achieve high levels of benefit from hearing aids and another group (controls) could be people who do not achieve benefit. Thus, the outcome is measured after the intervention and then the groups are compared on different variables to identify those that are predictive of group membership. There are numerous examples of studies of this kind in the hearing aid literature (Helvik, Wennberg, Jacobsen, & Hallberg, 2008; Hickson et al., 2014; Kelly-Campbell, Thomas, & McMillan, 2015). Roup and Noe (2009) described such a study relevant to hearing aid outcomes for mild hearing impairment, in which they compared hearing aid outcomes for participants with two different types of pure-tone audiogram configurations. The 53 participants in this study had an average mild hearing impairment and were grouped according to their hearing threshold at 2000 Hz (at or below 30 dB hearing level [HL] versus at or above 40 dB HL). All were fitted bilaterally with hearing aids in the 4 years prior to the study. Perceived hearing aid benefit was measured using the Abbreviated Profile of Hearing Aid Benefit (APHAB) (Cox & Alexander, 1995), and hearing aid satisfaction was measured with the Satisfaction With Amplification in Daily Life (SADL) (Cox & Alexander, 1999) questionnaires. The study found that both groups reported benefit from, and satisfaction with, hearing aids and that there was no significant difference in these outcome measures between the two audiogram configuration groups. Relevant to the PICO question, this study suggests hearing aids can be beneficial for adults with mild hearing impairment. As Roup and Noe (2009) point out, the two groups in this study design were selected based on degree and configuration of hearing loss, and other factors were not controlled. There was, for example, a significant difference in age between the two groups and, had hearing aid benefit or satisfaction differed between the two groups, it would have been difficult to ascertain if those differences were due to hearing loss configuration or age. This is an inherent limitation in nonrandomized studies such as case-control studies.
Cross-sectional survey studies involve analyzing data, often collected through the use of questionnaires or demographic surveys at one point in time, to determine the relationship or association between various factors in a population or group. Cross-sectional study designs are common in prevalence studies but cannot, by design, infer any causation between factors. One example of a cross-sectional study pertinent to this chapter’s PICO question is that of Mick and Pichora-Fuller (2016), who investigated the association between unaddressed hearing loss and poor physical and mental health in older adults. The study found that adults between 60 and 69 years who had not addressed their hearing loss by, for example, wearing hearing aids, had a higher incidence of lower cognitive ability and social isolation compared to those who did not have a hearing loss.
Case series or case reports are considered a very low level of evidence in that they do not provide information about the effectiveness of one kind of intervention over another. Such studies are quite common in audiologic rehabilitation for the description of the outcomes of hearing device fitting (Bishop et al., 2017; Távora-Vieira et al., 2013; Uriarte et al., 2005; Vestergaard, 2006) and their value is that they can provide a baseline for comparison of outcomes across studies. In recent times, the availability of “big data” has allowed for the description of outcomes for some very large data participant samples, which increases the value of such studies as baselines for expectations about outcomes. For example, in a recent case series study, Timmer, Hickson, and Launer (2017b) used data extracted from 8,489 bilateral hearing aid fittings. Participants were retrospectively allocated to groups based on hearing impairment averaged across 500, 1000, 2000, and 4000 Hz to investigate if objective data-logging measures of hearing aid use are different for adults with mild hearing impairment compared to adults with moderate hearing impairment. Participant age was significantly different between the hearing impairment groups, with participants with a mild hearing impairment (average thresholds between 25 and 40 dB HL) being 7.8 years younger than those with a moderate hearing impairment (average thresholds between 40 and 55 dB HL) and 11 years younger than those with a moderately severe hearing impairment (average thresholds between 56 and 70 dB HL). The study found average daily hearing aid use was 8.5 hours, with no statistically relevant differences in use between adults with mild or moderate hearing impairment. Hearing aid use was significantly lower for participants with mild hearing impairment compared to those with moderately severe hearing impairment, but only for the use of the left hearing aid. Age, gender, or hearing aid style were not related to objective hearing aid use and the hearing aids for this large group of participants were typically used in quiet for 83.3% of the wearing time and in noisy listening situations for 12.4% of the time.
The lowest level of evidence is expert opinion and sometimes a number of experts collaborate to produce a consensus statement or a guideline. The latter may or may not be evidence based, and it is important to scrutinize such reports carefully to determine if they are evidence based. An example of a guideline relevant to the PICO question that is the focus of this chapter is the National Institute for Health and Care Excellence (NICE) guideline on “Hearing loss in adults: Assessment and management” (https://www.nice.org.uk/guidance/ng98). The full guideline is some 280 pages, but the main messages are summarized as a series of recommendations based on the detailed evaluation of the best available evidence as analyzed by an expert panel. For example, in relation to offering hearing aids to adults, the recommendation is “Offer hearing aids to adults whose hearing loss affects their ability to communicate and hear, including awareness of warning sounds and the environment, and appreciation of music” (NICE, 2018, p. 16).
More detailed descriptions of levels of evidence, along with a number of other very useful tools, such as tips for finding the best evidence and critical appraisal worksheets, are available from the Center for Evidence-Based Medicine (http://www.cebm.net) and in Wong and Hickson (2012a). It is also important to point out that, although the level of evidence hierarchy is commonly used, it is not always the only way to rate the value of a particular piece of evidence. Sackett et al. (1996) make the point that “Evidence-based medicine is not restricted to randomized trials and meta-analyses. It involves tracking down the best external evidence with which to answer our clinical question” (p. 72). In the case of the question proposed initially in this chapter, “For patients with acquired mild sensorineural hearing impairment, is hearing aid fitting effective compared to no intervention for improving self-reported hearing and communication difficulties in everyday life?” an RCT will provide the most appropriate evidence to answer the question. If, however, the question were “For adults with mild hearing impairment, what are the barriers and facilitators to hearing aid uptake and success?” a qualitative research methodology would be most appropriate (Knudsen et al., 2012). Hannes and Bennett (2017) point out that qualitative research is a valid means of creating evidence about “the feasibility, appropriateness or meaningfulness of a certain intervention or activity” (p. 226). Generally in qualitative research, the PICO question is adapted not to have a comparison and the “I” in PICO refers more to “issue” or “interest” rather than “intervention.” An example of a qualitative study investigating hearing aid use and adoption can be found in Gallagher and Woodside (2018), who conducted face-to-face semistructured interviews with hearing aid owners and nonowners with varying degrees of hearing impairment including mild, and utilized thematic analysis to answer their research question. For example, they found that perceived complexity of hearing aid use, attitudes toward both hearing impairment and hearing aids, and inadequacy of audiology services influenced hearing aid adoption. Qualitative studies, like quantitative studies, vary in quality and should be critically appraised to determine the credibility of the evidence presented (Hannes & Bennett, 2017).
EBP Step 4: Apply the Information
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
