Background
Historically, professional organizations have emphasized the central role of glycemic control (e.g., hemoglobin A1c (HbA1c) < 7%) in diabetes management in order to reduce the risk of long-term complications of the disease [
1]. However, a growing recognition of the burden of severe hypoglycemia has begun to shift diabetes care toward a greater emphasis on prevention of adverse drug events (ADEs) associated with diabetes treatment [
1]. Emergency department (ED) visits for severe hypoglycemia are now more common than those for hyperglycemic crisis [
2] and are associated with significant healthcare costs [
3]. Alarmingly, the reported rates of severe hypoglycemia have not fallen for the past 20 years [
4] and current surveillance continues to underestimate its overall burden [
5]. Severe hypoglycemia, defined as an event requiring the assistance of another person [
6], is associated with significant morbidity and greater risk of death [
3,
6‐
10], especially for older adults, as it may exacerbate or may be exacerbated by their many comorbidities. In older patients, diabetes agents are implicated in 1 in 5 ED visits for ADEs [
11] and at least one episode of hospitalization for hypoglycemia is associated with subsequent re-hospitalization and mortality [
3]. For older adults with diabetes, a history of severe hypoglycemic episodes is also associated with a greater risk of dementia [
12], fall-related fractures [
13], and decreased quality of life [
14]. Therefore, prevention of hypoglycemic events should be considered a high priority in older individuals.
In order to develop successful interventions to reduce hypoglycemic events, it is important to understand the underlying causes of severe hypoglycemia. Although multiple predisposing factors for hypoglycemia have been reported, such as lack of access to food [
15‐
21], social vulnerability [
17,
21], lack of support or knowledge [
15], and treatment errors [
3,
9], precipitating factors for hypoglycemia are often under-reported in the literature [
15]. In a national study examining insulin-related hypoglycemia and errors leading to ED visits, precipitating factors for insulin-related hypoglycemia were documented in only 20.8% of cases [
3], indicating that the majority of precipitating factors are still left to be identified. Furthermore, professional organizations and the US Department of Health and Human Services have called for data to better understand and thus prevent hypoglycemia [
1,
22].
One step toward understanding the underlying causes for hospitalizations of hypoglycemia is to elicit the perspectives of patients on the reasons for their hypoglycemia. Accordingly, using qualitative methods, we conducted interviews with older patients with diabetes admitted to the hospital for severe hypoglycemia to investigate this complex phenomenon and to explore its cause(s).
Methods
Overall study design
We performed a qualitative analysis using in-depth interviews with older patients with diabetes admitted to the hospital with hypoglycemia. We chose qualitative methods because they are well-suited for investigating complex phenomena with multiple potential causes [
23].
Participant recruitment
Between October 2014 and February 2016, we recruited patients from Yale-New Haven Hospital (a large, tertiary referral teaching hospital located in an urban setting on the east coast) and its affiliate, Saint Raphael’s Hospital. We used electronic medical records to identify patients with diabetes, who were 65 years or older, and who were admitted with a primary or secondary diagnosis of diabetes-related hypoglycemia defined based on ICD-9 and ICD-10 codes. Following a careful review of the medical records to confirm the patient was actually admitted for hypoglycemia, potential participants were screened for delirium and dementia with validated tools (CAM, Confusion Assessment Method, and the Mini-Cog, respectively) to rule out significant cognitive impairment. Patients were considered eligible if they did not have delirium or significant cognitive impairment, had the capacity to participate in the study as assessed directly by the research staff, did not have severe hearing impairment, and were not admitted from a nursing home. All participants that passed the screening process and agreed to participate provided written informed consent. The study was approved by the institutional review board at Yale University.
Data collection
Data were collected from in-person one-on-one in-depth open-ended interviews to explore participants’ experiences with diabetes management and hypoglycemia. Interviews were conducted during the admission so as to limit recall bias. Interviews were conducted using a standard qualitative interview guide, but allowing participants to direct the course of discussion as much as possible. The interviews began with a broad “grand tour” question [
24], “Tell me about the reasons you were admitted to the hospital this time.” The interview guides explored the reason for admission, challenges experienced dealing with diabetes, medications and adherence, home life, financial issues, transportation, depression/cognitive function, functional disability, contact with the provider, and how the admission could have been potentially avoided. Supplementary Fig. 1 in Additional file
1 contains the interview guide that was used.
In addition to the established questions, the interviewer used prompts and probes to clarify concepts, elicit detail, and extend the narrative. All interviews were professionally recorded and transcribed. Additional data were collected from the review of the hospital electronic medical record, including comorbid conditions and medications used prior to admission.
Data analysis
We analyzed the data using the constant comparative method [
25], where coding and analysis took place simultaneously in order to systematically extract major themes that were then used to develop a theory [
26]. Verbatim interview transcripts were coded using an iterative, inductive process by an interdisciplinary research team consisting of a medical student (W.P.), an organizational psychologist (D.B.), and an endocrinologist specializing in hypoglycemia (K.L). A codebook was developed to capture the emerging themes and was updated whenever new themes emerged or new connections between themes were discovered.
Each of the 3 coders individually coded the transcripts and then as a team, discussed data interpretation and thematic emergence, negotiated a consensus on the major themes, and identified potential sources of individual bias. Albeit rare, disagreements between coders were thoroughly discussed and clarified using exemplar quotes and concrete textual examples. Ultimately any disagreements led to a constructive re-evaluation of the codebook and major themes.
Several techniques were used to ensure the scientific rigor of our study: consistent use of an interview guide, professional audiotaping and transcription, the use of researchers with diverse professional backgrounds, an initial calibration period to ensure inter-rater reliability, and an audit trail to document all analytic decisions.
Discussion
In this study of older adults with diabetes admitted for severe hypoglycemia, participants did not report the previously described reasons for hypoglycemia, including lack of access to food [
15‐
21], or lack of supports for diabetes management from patients’ families [
15] as well as our hypothesized reasons, such as lack of support from the healthcare system and patients’ lack of knowledge about how to treat hypoglycemia. Although treatment errors are thought to be major contributors to hypoglycemia among older persons [
3,
9], only one participant described a medication mix-up.
Instead, we found evidence for a pervasive emphasis on the avoidance of hyperglycemia by both patients and the medical profession. In our study, older patients who had long-standing diabetes, good knowledge of hypoglycemia treatment, and several instances of recent hypoglycemia, still reported surprise at their hypoglycemic events. Many reported they were “doing everything right,” and we suggest that “doing everything right” is deeply entrenched with the prevention of hyperglycemia, rather than hypoglycemia. This was also evidenced by patients’ preoccupation with the avoidance of foods with high-glycemic indices that would cause hyperglycemia, even after they acknowledged that perhaps not eating enough food is what caused hypoglycemia. This emphasis on hyperglycemia was also echoed in patient experiences with the healthcare system; patients detailed their clinicians’ views on bringing down their HbA1c and a “top number” to avoid, while explaining that they rarely reported low blood sugars to their healthcare team and instead dealt with hypoglycemia themselves. This attention focused on glycemic control may have contributed to surprise at the occurrence of hypoglycemia, or neglect of preceding milder hypoglycemic events.
Although efforts are underway to shift from the prioritization of reaching pre-specified glycemic targets to a more balanced approach [
27], our results suggest that the healthcare system’s previous emphasis on avoidance of hyperglycemia has been internalized by many older patients with diabetes. Participants in our study still prioritized the prevention of hyperglycemia over hypoglycemia. The emphasis on avoidance of hyperglycemia is further entrenched by existing quality measures targeting hyperglycemia, with none that specifically measure hypoglycemic events [
28]. As a result, clinicians may be unaware of their patients’ hypoglycemic events [
29], and our findings suggest that patients may be reluctant to report low blood sugars to their clinicians because they deem them as less likely to require medical intervention than hyperglycemic events. To correct this unbalanced approach and address this gap in communication, we propose that the healthcare system makes hypoglycemia a “medical” priority; improve communication strategies about the risks of hypoglycemia; include severe hypoglycemia as a quality measure for clinicians; and dedicate as much time, focus, and monitoring to low blood sugars as is typically dedicated to high blood sugars in older patients at risk for severe hypoglycemia.
Another major theme in our study was the tension in self-management of diabetes: between patients’ desire to adhere to prescribed treatment plans and their need to adjust treatment based on their experience and changes in their diet or activity. Some patients reported that they needed to eat just to compensate for the insulin they were taking. Many participants may not have felt empowered to share concerns about their treatment with the healthcare team or to make changes to their regimen when they felt something was amiss. This disconnect highlights a tension around where the control of self-management resides: the patient or the clinician. Perhaps with good intentions and supported by current recommendations to simplify complex regimens [
6,
30,
31], physicians recommend treatment plans that are less flexible, but simpler to implement. However, the lack of flexibility and patient control may in fact predispose patients to severe hypoglycemia. In contrast, structured education programs studied in type 1 diabetes, like DAFNE, have allowed patients to retain more control and flexibility in their self-management [
32,
33], while also improving glycemic control [
32,
34,
35] and reducing severe hypoglycemia [
4,
35,
36]. Due to a lack of existing literature [
4], we propose that further research should expand upon the role of structured education programs on the management of type 2 diabetes and severe hypoglycemia, while also addressing the prevalence of self-management challenges in older adults who experience severe hypoglycemia.
It is also possible that clinicians make assumptions about patients’ capacity to self-manage their disease. In previous qualitative studies, healthcare professionals cited concerns regarding the suitability of self-management for patients they deemed not capable [
37,
38]. With our predominantly female, African-American, urban population, we cannot discount the influence of gender, race, and socioeconomic status in clinicians’ assessments of patients’ capacity to self-manage. There is a paucity of literature focused explicitly on bias and self-management, but previous studies have reported differences in physicians’ perceptions of black vs. white patients, with black patients more likely to be seen as a risk for noncompliance [
39] and experiencing shared-decision making less often than white patients [
40]. Therefore, in order to develop safe and effective management plans, we propose that clinicians work closely with patients to create plans that “fit” – plans that match patients’ willingness and capacity for self-management, while tempering assumptions regarding patients’ capability to self-manage.
Limitations
Our study has several limitations. Participants were recruited after hospital admission for severe hypoglycemia; therefore, our findings were unable to capture the hypoglycemic events that were managed in the outpatient setting or outside of the healthcare system. Our sample consisted of predominantly African-American, female participants from New Haven, so our results may not be representative of all older adults with diabetes who have experienced severe hypoglycemia. Furthermore, we did not collect data on all of the potential sociodemographic factors that could have influenced our findings. Although certain themes likely transcend differences in socioeconomic status, such as the idea that the prevention of hyperglycemia is more important than the prevention of hypoglycemia, the lack of food insecurity and medication mix-ups could potentially be explained by a patient population of higher socioeconomic status. However, our findings did achieve our aim of richly capturing the experiences of older patients with severe hypoglycemia admitted to Yale Hospital and its affiliate, St. Raphael’s. The last limitation is inherent to the methodology of qualitative work; the intent of our study, was not to prove or refute hypotheses but to generate them. We propose that further quantitative research is needed to assess the prevalence of the themes that emerged in our study and to evaluate the efficacy of potential interventions.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.