In this multi-site study, PCPs endorsed the importance of achieving glycemic goals and of initiating insulin to reach glycemic targets for those uncontrolled on oral agents. However, most believed that several factors would mitigate their ability to intensify therapy in general and to initiate insulin in particular. Most of the respondents identified patient factors as a prime contributor to decisions to not initiate insulin. The respondents reported that at least 10% of their patients would refuse starting or continuing insulin therapy due to fear of the injections, fear of insulin, or the inconvenience of self-management. For most providers, this perceived resistance or lack of self-management skills in their patients was a barrier to initiating insulin therapy, and prior non-adherence to oral agents would dissuade most PCPs from initiating insulin with some patients.
Provider decision-making about whether to initiate or intensify medication therapy is a complex process affected by multiple factors,6
and clinical inertia is considered a major reason for inadequate metabolic control in diabetics.13–15
In prior studies, rates of appropriate medication intensification for poorly controlled hemoglobin A1c have ranged from 46 to 66%.14,16,17
However, these studies did not investigate the relative contribution of PCPs’ attitudes to their clinical inertia.
Our results lend insights into how providers incorporate their beliefs about patients’ prior adherence, preferences, and treatment risk and burden into their decision-making. Indeed, while most providers aimed for a hemoglobin A1c target of <7.0%, most also would adjust this target or refrain from starting insulin, a modification that is consistent with established guidelines.1,8,11
Given recent studies highlighting the linkage between inadequate health literacy among insulin users and hypoglycemia risk, and the long-term and acute consequences of hypoglycemia events,9,10,18
providers’ concerns about initiating insulin therapy may be both patient-centered and medically appropriate. With the potential for physical, psychological, and financial burdens of intensive therapy and monitoring, some have questioned the wisdom of applying quality indicators to promote tight glycemic control among all type 2 diabetics, arguing for more conservative prescribing practices.19,20
However, it is unclear how accurate PCPs are in their assessment of patients’ preferences, concerns, and self-management capacity. A separate part of the TRIAD Insulin Starts Project assessed attitudes towards insulin among patients with poorly controlled diabetes who fail to fill their first insulin prescriptions and found similar barriers to insulin therapy reported from the patient perspective.3
Fear of injections occurred for 30% of patients who failed to start insulin, and 97% of PCPs recognized that this is a major reason for patients’ failure to initiate insulin therapy. Meanwhile, 55% of patients who did not fill their insulin prescription felt that medication risks and benefits were not adequately explained by their providers, with 51% reported difficulty learning about their condition because of problems understanding written information (i.e., inadequate health literacy). Among patients who did not fill their insulin prescriptions, significant proportions expressed moderate to extreme concerns about their ability to give themselves shots (42%) and potential negative impact on their jobs (33%) and their social lives (38%).3
This suggests that some of these potential barriers might be addressed by more patient education and counseling and more focus on the quality of providers’ communication regarding insulin initiation.
Providers’ assumptions about self-management capacity could also contribute to disparities in intensification in response to suboptimal glycemic control. In our study, PCPs considered limited health literacy and language issues to be barriers to self-management, which is itself a barrier to insulin initiation. A recent study did find that patients with limited health literacy were 30–40% more likely to experience hypoglycemia compared to those with adequate health literacy.9
However, a patient’s capacity for self-management is not a static factor, and health care systems interventions can enhance diabetes self-management among diverse patient populations.21,22
For example, a randomized controlled trial of automated telephone self-management support among a low income population with limited English proficiency and poorly controlled diabetes demonstrated significant improvements in self-management behavior, including increased participation in self-management behaviors and physical activity.22
Providers’ beliefs that some patients lack the willingness or capacity to take insulin may reflect a need for more resources to help providers engage patients in effective shared decision-making and enhance patients’ capacity for self-management. In one study of older patients with diabetes, the higher patient ratings of their providers’ provision of information and participatory decision-making style were significantly associated with self-management behaviors such as diet, exercise, glucose monitoring, and foot care, and patients’ ratings of their self-management were associated with glycemic control.23
Our study highlights the prominence of perceived patient non-adherence in dissuading providers from initiating insulin. Consistent with our findings, another study suggested that academic general internal medicine physicians emphasized patient adherence, along with patient fear of injections and patient desire to prolong non-insulin therapy as major insulin barriers.24
In one study examining pharmacy claims, 23% of poorly controlled hyperglycemic patients had evidence of poor adherence and lack of treatment intensification by providers.17
Still, 30% of poorly controlled, hyperglycemic patients had no treatment intensification despite a lack of evidence of poor medication adherence, suggesting that providers may not accurately assess non-adherence in their patients or that this does not explain clinical inertia entirely.17
Meanwhile, interventions focused on improving medication adherence have been complex and marginal in effectiveness.25
More strategies are needed to help providers assess and manage non-adherence and to determine whether maintaining, decreasing, or intensifying a regimen is the most effective or patient-centered recommendation.
This study is limited in that, although the providers were drawn from three different health systems caring for diverse patient populations, our findings may not be generalizable to providers in other systems caring for different populations. We did not assess formally validity or reliability of the interview instrument. Because we asked PCPs to reflect on patients in general, our results may be susceptible to recall bias by providers attempting to recall factors in past decision-making experiences with their patient panel. An interview conducted immediately following a specific decision regarding a specific patient – particularly if paired with an interview with the patient itself – might be less subject to recall bias, but might also yield different insights into the insulin initiation decision-making process. Finally, our limited sample size precludes us from determining whether PCP factors such as years in practice or clinic site are related to attitudes. the study methodology asked PCPs to reflect on their patients in general;