Examination of Health Care Providers’ Use of Language in Diabetes Care: A Secondary Qualitative Data Analysis

We conducted a secondary qualitative analysis to explore the use of language in diabetes care among HCPs. The aim of the original qualitative study was to identify HCPs’ perceived barriers and supports for the management of diabetes in rural southeastern Ohio (31). We used focused ethnography to conduct in-depth face-to-face and telephone interviews with HCPs who cared for people with diabetes in the region. Focused ethnography is a qualitative approach designed to access beliefs and practices to improve our understanding of factors surrounding health and disease (32). Furthermore, HCPs shared beliefs and practices about diabetes, including the diagnosis of diabetes, self-care behaviors (e.g., blood glucose monitoring and physical activity), beliefs about diabetes (e.g., the need to lower or prevent hyperglycemia), and language used in diabetes management (e.g., insulin, glucose, lancets, and ketones).

We used maximum variation sampling, a form of purposive sampling (33), to recruit a wide range of participants from different medical backgrounds. Inclusion criteria included HCPs who were ≥18 years of age who could read and speak in English and practiced in southeastern Ohio, with at least 5% of their practice consisting of people with diabetes, to ensure experience with treating diabetes. Participants were recruited via emails, telephone calls, and word of mouth.

The Ohio University Office of Research Compliance approved the protocol (Institutional Review Board #16-E-342) and all recruitment procedures and materials. Written (during face-to-face interviews) or verbal (during telephone interviews) informed consent was obtained from all participants.

One trained qualitative researcher (E.A.B.) conducted all interviews, asking participants broad, open-ended questions about diabetes in the region, diabetes management, and barriers and supports to diabetes care. No questions about diabetes language were included in the interview guide; thus, the diabetes language used in the interviews may reflect how the HCPs discuss diabetes with their patients, colleagues, and community. The interview guide has been published elsewhere (31).

In-depth interviews were conducted either in-person at medical offices or university spaces or via telephone per participants’ preference. Interviews were audio-recorded and lasted 30–90 minutes each. Interviews were conducted in the spring and summer of 2017. Data were collected until saturation was reached (i.e., when no new information emerged from the interviews) (33).

All interviews were digitally audio-recorded and transcribed verbatim, with names and identifiers removed to protect participant confidentiality. The research team, consisting of two behavioral diabetes researchers and one medical student, performed quality checks of the transcribed files by listening to the audio-recordings to validate the transcriptions.

The multidisciplinary research team used content and thematic analyses to analyze the data using NVivo software (34). Specifically, members of the research team performed content analysis (35,36) by independently coding key words and phrases in the transcripts to identify language not recommended by the following organizations: Diabetes Australia (26), IDF (28), ADA and ADCES (29), and Diabetes Canada (30). The coders met to review the coded data; discrepancies were reviewed, discussed, and resolved through consensus (37). The Cohen κ coefficient for intercoder reliability was 0.981, indicating almost perfect agreement (38,39). Finally, the research team met to identify themes or abstract concepts linking substantial portions of the interviews together. The themes described participants’ use of diabetes language that occurred multiple times, both within and across transcripts.

We used analyst triangulation to support the credibility (validity) of the data by limiting selective perceptions and cognitive biases in the analysis (40). Next, we used thick descriptions and verbatim quotations to support the transferability (external validity) of the data (41). Finally, we conducted an inquiry audit to support dependability (reliability) of the data. Specifically, a researcher not involved with the study examined the research process and product to determine whether the findings and conclusions were supported by the data (42).

A total of 42 HCPs participated in in-depth interviews (23 in person and 19 via telephone). The average age of participants was 45.1 ± 12.0 years. Two-thirds of the participants self-identified as cisgender women (n = 28, 66.7%) and 15 (35.7%) identified as primary care physicians. On average, participants had 17.7 ± 11.5 years of clinical experience and an estimated diabetes caseload of 46.8 ± 29.6% (Table 1). Transcript identifiers are used with quotations indicating participant number and provider type. Three themes emerged and are discussed below.

The majority of HCPs referred to people with diabetes as “diabetic.” Thirty of the 42 participants (71.4%) used the label “diabetic” in their interviews a total of 167 times, with some participants using this term more than 20 times (Table 2). For example:

In the context of the interview questions, participants were asked to describe barriers to diabetes care and reasons for not achieving diabetes targets. Participants’ responses attributed unmet targets to not performing self-care behaviors. These participants did not use person-centered language but rather used labeling language, as shown in the following examples.

Several providers used language that labeled people with diabetes as people who did not follow medical recommendations. Sixteen of the 42 participants (38.1%) referred to people with diabetes as “noncompliant” when discussing self-care recommendations, as shown in the following quotations.

Several participating HCPs used judgmental language to describe people not engaging in self-care or not achieving A1C targets. For example, 25 participants (59.5%) referred to A1C levels with good or bad value judgments and/or placed an emphasis on achieving “control” (Table 2). For example:

In addition, participants used language with negative connotations when discussing self-care behaviors such as blood glucose monitoring. Fourteen participants (33.3%) used the term “testing” for blood glucose monitoring, which implies a pass/fail result. These participants used “test” and “testing” instead of monitoring, as shown in the following examples.

Eight participants (18.6%) also used “regimen” when discussing self-care behaviors, which is a term that may feel discouraging and/or limiting to people with diabetes. For example:

Seven participants (16.7%) used “adherence” when referring to “regimens” and self-care behaviors. Like compliance, the word “adherence” suggests that a person with diabetes is taking orders from someone else rather than making their own choices for self-care, as shown in the following example.

Only one participant did not use any of the nonrecommended language during the interviews. Two participants, without prompting, noted that language could be stigmatizing and harmful to people with diabetes. These two HCPs stated that they did not like the use of the term “noncompliant.” One participant mentioned how using the term “noncompliant” in the medical chart could negatively affect future clinical care for a person with diabetes.

The other participant commented that this term was frequently heard from other HCPs in rural southeastern Ohio and that its use in a clinical setting was not appropriate.

In this secondary qualitative analysis of HCPs’ use of language in diabetes care, participants frequently used stigmatizing language when referring to people with diabetes. Participants spoke with labeling language and used language with negative connotations. Nearly three-fourths of participants used the label “diabetic” when referring to people with diabetes, and more than one-third used the label “noncompliant” when discussing self-care recommendations. In addition, participants emphasized achieving glycemic “control” and placed good or bad value judgments on A1C levels. Participants also used the terms “testing,” “regimen,” and “adherence,” which may be perceived as discouraging and limiting to people with diabetes. Importantly, one participant did not use any nonrecommended language during the interview, and two participants recognized the negative impact language could have on people with diabetes. These findings offer a real-world glimpse of how HCPs talk about people with diabetes, given that the original aim of the study was not focused on the use of language.

Our findings align with the conclusions of prior qualitative research exploring the experiences of people with diabetes regarding nonrecommended language. Qualitative research by Speight et al. (26) in 2012 and Dickinson (27) in 2018 revealed that people with diabetes felt judged, stigmatized, blamed, and shamed by the language used by HCPs. Specifically, participants in these studies stated that certain terms made them feel as if their diabetes was being over-simplified or presumed to be easily managed (27). Participants also noted that the use of “noncompliant” implied that someone else was responsible for their health care decisions, which in turn removed autonomy from their own diabetes-related decisions (27). In our study, the participating HCPs used the term “noncompliant” to describe people with diabetes who were not following their medical recommendations, echoing the sentiments of the participants in the Dickinson study (27). Moreover, Speight et al. (26) found that people with diabetes want a more active role in their diabetes care and that being labeled as “noncompliant” left them feeling upset and unheard. In these two studies, participants requested that HCPs stop using harmful labels such as “diabetic” and “noncompliant” as well as judgmental terms such as “control” and “should” (27). They said they would feel more supported, comfortable, and understood by HCPs if these terms were no longer used (27). Finally, Dickinson (27) highlighted the need to address language broadly, arguing that new terms with the same meaning and intention could be used in place of current terms. Thus, widespread education to call attention to stigmatizing and harmful language is needed to improve care for people with diabetes.

Language affects views, perspectives, and practices (29). It can inflict harm and perpetuate stigma, whether intentionally or unintentionally (29). For these reasons, the American Psychological Association (43) and the ADA and ADCES (29) advocate for the use of person-first language, which emphasizes the person rather than a disabling or chronic health condition. In the context of diabetes, person-first language acknowledges that a person has diabetes rather than diabetes being who they are (e.g., “a person with diabetes” vs. “diabetic”).

Conversely, disability culture and its scholars challenge the notion of person-first language and recommend the use of identity-first language (44). Identity-first language places the disabling or chronic condition first by recognizing that it is an integral part of a person’s identity. Identity-first language is considered empowering because people acknowledge that their disabling or chronic condition is essential to who they are as a person. Furthermore, disability scholars argue that person-first language endorses stigma by implying that the disabling or chronic condition is problematic (45). For example, a 2017 analysis by Gernsbacher (45) showed that person-first language in scholarly writing was used more frequently to refer to children with the most stigmatized disabilities (e.g., intellectual disability) compared with children with less stigmatized disabilities (e.g., deafness).

To address this bias, some experts recommend that HCPs refer to all people, regardless of the presence of a disabling or chronic condition, with person-first language (45). However, other experts recommend the use of identify-first and person-first language interchangeably to acknowledge both people with and without disabling and chronic conditions (44). They argue that the use of both identify-first and person-first language is more inclusive and addresses the issues raised by disability culture and scholars (44).

People with diabetes are diverse, and what applies to one person may not apply to another. For this reason, HCPs cannot predict what any single individual might prefer. Therefore, HCPs should ask each person about their preferred terminology.

Study limitations included homogeneity of the sample with regard to gender, race, and geographical region. Thus, our qualitative findings may not be transferable or generalizable to people not represented in the sample. Furthermore, data from one geographic area (i.e., southeastern Ohio) may not reflect the diabetes language used by HCPs in other regions of the country. Future research should examine HCPs’ language use in different geographical regions. Importantly, our findings are consistent with the qualitative findings from Speight et al. (26) and Dickinson (27).

Next, the qualitative interviews were conducted in the spring and summer of 2017, which is when the ADA and ADCES published their position statement (29). Thus, the participating HCPs may have been less familiar with their recommendations against using some terms. On the other hand, the use of language has been a topic of discussion for more than two decades in the diabetes literature (46,47), and Diabetes Australia (26) and the IDF (28) published their recommendations in 2012 and 2014, respectively.

An additional limitation to this secondary qualitative analysis was we did not ask participants specific questions about language use in diabetes. Participants responded to questions about diabetes management and barriers and supports to diabetes care from a previous qualitative study, which may explain the use of language with negative connotations. Despite the emphasis on barriers and diabetes management, participants had the opportunity to use terms other than “noncompliant,” “control,” or “test.” Another important consideration was that the participants in the study may have used different language when speaking to their patients with diabetes; future research exploring HCPs’ language use via qualitative interviews and video-based observation of patient-provider interactions may be useful methods to answer this question. Finally, as a secondary data analysis, we did not perform member checking, which involves asking participants to review the findings as a technique to improve the accuracy of the findings. To address this limitation, the qualitative coders included behavioral diabetes researchers and one member from the diabetes community.

In this study, the participating HCPs frequently used stigmatizing language when discussing people with diabetes and their outcomes. We know from prior research that people with diabetes report feeling judged, stigmatized, blamed, and shamed by the words used by HCPs and that those feelings are associated with psychological distress and negative health outcomes. Thus, HCPs must strive to use diabetes language that is free of bias and avoids perpetuating stigma, blame, and shame. Continuing medical education is the ideal platform to disseminate national and international diabetes language recommendations to promote rapid uptake of best practices. Increasing HCPs’ awareness and knowledge of recommended diabetes language has the potential to improve patient-provider communication, relationships, and health outcomes.

The authors thank the HCPs who participated in this study.

No potential conflicts of interest relevant to this article were reported.

All authors provided substantial contributions to secondary data conception, design, and analysis; contributed to the interpretation of data; drafted and revised the article critically for important intellectual content; and read and approved the final version of the manuscript for publication. E.A.B. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.