June 15, 2024

Study design

We collected data via an online questionnaire between 1 February and 18 April 2022 in written text and videos in GSL translated by a certified SLI. The questionnaire was administered via SoSci Survey v3.2.55 hosted on Charité in-house servers in accordance with European data protection laws. We pretested the survey twice with a female deaf participant and a female SLI, and we made improvements to wording and questions together. Many questions of our questionnaire were adapted from Höcker et al. [10] who in turn developed their questions from qualitative interviews with DHH people in a participatory fashion. Participants were recruited via emails to DHH associations or community multipliers and via social media channels.

Measures and definitions

The questionnaire contained 49 items. The original German version and its English translation can be found in the supplementary material.

We collected sociodemographic data for age, gender (male, female, diverse), residential environment (rural, urban) and educational background (highest degree). Educational background was dichotomised into vocational training (0) and university degree [1]. This established classification follows the German education system, and it represents the two common paths of further education after leaving secondary school.

For the following measures, we used the percentage of maximum possible (POMP) score as a linear transformation using the formula “((xi—min)/(max—min)) * 10” where xi are the individual observed values, min is the scale’s minimum value and max is the scale’s maximum value [16]. The resulting POMP score ranges between 0 and 10 and is thus intuitively interpretable in descriptive statistics and regression analyses.

A communication score (CS) was calculated with four 4-point Likert-scale items that included literacy in reading and writing words, understanding, and speaking spoken language. Responses ranged between very badly [1] and very well [4]. Following a principal component analysis with one factor (Eigenvalue = 2.25, 56% variance explained), we calculated the mean across the four items and transformed it into a POMP score [16] with a range between low CS (0) and high CS [10]. The scale had an internal consistency of Cronbach’s α = 0.74.

Conviction to receive help in acute medical situations (HELP) was measured by asking “Do you feel you get adequate help in medical emergency situations? “ on a 4-point Likert scale ranging from strongly disagree [1] to strongly agree [4], transformed into a POMP score ranging from strongly disagree (0) to strongly agree [10]. Medical emergency situations were specified to include all situations in which someone needs unplanned medical help; this includes unplanned visits, e.g., to the doctor, dentist, or hospital. All participants were asked this item regardless of their experience in emergency medical situations.

Satisfaction during the latest doctor’s visit (SAT) was measured with asking “Regarding your latest unplanned doctor’s visit, how satisfied were you with your appointment in general?” on a 4-point Likert scale ranging from very dissatisfied [1] to very satisfied [4], transformed into a POMP score ranging from very dissatisfied (0) to very satisfied [10]. Data were collected only from participants with previous emergency care experience (n = 245).

Concerns during doctor visits (CDV) were assessed via seven items e.g., being misunderstood or being misdiagnosed, on a scale from not concerned at all (0) to very concerned (100). Principal component analysis revealed one factor (Eigenvalue = 4.26, 61% variance explained). We calculated the mean across the seven items and transformed it into a POMP score with a range between low CDV (0) and high CDV [10]. The scale had an internal consistency of Cronbach’s α = 0.89. All participants were asked this item regardless of their experience in emergency medical situations.

Miscommunication with medical staff (MisC) was assessed via three 4-point Likert scale items on a scale from totally disagree [1] to totally agree [4] focusing on unasked questions in fear of being misunderstood, wrong diagnoses because of miscommunication, and feelings of helplessness and dependence because of deafness. Principal component analysis revealed one factor (Eigenvalue = 1.82, 61% variance explained). We calculated the mean across the four items and transformed it into a POMP score with a range between low MisC (0) and high MisC [10]. The scale had an internal consistency of Cronbach’s α = 0.67. Data were collected only from participants with previous emergency care experience (n = 245).

Avoiding medical attention (AMA) was assessed with the dummy variable (yes/no) whether participants did not use medical help even if they had symptoms indicating a condition. Subsequently, participants were asked to specify reasons for avoiding medical attention, e.g., fear of being misunderstood or not knowing where to go, with yes/no options and multiple responses.

Inclusion criteria

We included DHH participants ≥ 18 years old living in Germany.

Exclusion criteria

Participants younger than 18 years old and those without hearing impairment were immediately directed to the end of the survey without collecting any further data. After data collection, we excluded participants if they reported previous participations, if they only responded to ≤ 25% of questions, or if implausible time to response ratios were recorded [17].

Sample size calculation

Following exchange with the Charité Institute for Biometry and Clinical Epidemiology, we set the minimum number of participants at N = 100 enabling linear regression analyses with five predictor variables.

Statistical analysis

Data were analysed using IBM SPSS Version 27. We calculated descriptive statistics (mean and standard deviation), t-tests, ANOVAs, and correlations where applicable. We calculated two multivariable linear regression analyses to investigate individual factors (CS, CDV, MisC, age, gender, and location) associated with convictions of receiving help (HELP) and satisfaction with the last consultation (SAT), and a logistic regression analysis to investigate factors related to avoidance of medical attention (AMA). In logistic regression, we used backward elimination, eliminating non-significant predictors (p > 0.05) with every step. We calculated odds ratios (OR) and corresponding ninety-five percent confidence intervals (95%CI) for each predictor’s association with the outcome.

link

Leave a Reply

Your email address will not be published. Required fields are marked *