In this cross-sectional study, physicians completed an online survey designed to assess their perceptions of dose escalation. Before completing the survey, physicians provided electronic consent and completed online screening questions to determine whether they were eligible to participate. The online survey was designed to take approximately 20–30 min to complete, and physicians who completed the survey were reimbursed. Participants provided informed consent before completing the survey. All procedures and materials were approved by a central institutional review board (22131-01A, Ethical and Independent Review Services), which was conducted in accordance with the Declaration of Helsinki. All surveys were completed from July to September 2022.

Participants were primary care physicians and endocrinologists licensed to practice medicine in the USA. To be eligible for this study, physicians were required to have been in medical practice for ≥ 1 year, treated an average of ≥ 10 patients with T2D per month, and prescribed injectable T2D medication (i.e., insulin or glucagon-like peptide-1 receptor agonists) for ≥ 3 patients in the 6 months prior to survey completion. Physicians were excluded if they were practicing medicine in a state where the Sunshine Act prohibits participation (i.e., Vermont and Massachusetts). An estimated sample size of approximately 400 to 500 physicians (approximately 70% primary care physicians, 30% endocrinologists) was targeted, roughly equally distributed across four US regions (Northeast, Midwest, South, and West). As the planned analyses were descriptive without a key statistical comparison, no power analysis was conducted when determining the sample size target. The sample size target was determined on the basis of similar surveys published in the past as well as practical implications.

Participants were recruited through a healthcare provider (HCP) panel built over a period of approximately 12 years by sourcing HCP contact data through direct physician outreach, conferences, LinkedIn outreach, ZoomInfo, institution websites, and PubMed. HCP status is confirmed at the time of opting into the panel by verifying their National Provider Identifier and implementing other quality control metrics. Panelist data are periodically cross-checked against online databases to ensure profiling remains accurate as time passes. The recruitment strategy for most of the study was to conduct a spam-resistant continual email campaign, sending study invitations to verified primary care physicians and endocrinologists in the HCP panel. However, for the initial soft launch of 15 physicians, a controlled email campaign was used to avoid over-recruitment and pause data collection as needed to identify any potential issues with the survey prior to full launch. For the full launch, the continual email campaign was initiated, usually sending emails every 2 days during the study period. The email invitation briefly described the study and how much time it would take. A link was provided in the email invitation for the participant to click and be screened. If the screening questions determined that the participant was eligible, the participant would continue by completing the online physician survey. For this study, a total of > 30,000 email invitations were sent, and recruitment was discontinued when the sample size target was reached.

To inform development of the survey, four clinical experts (three physicians with experience treating T2D and one clinical researcher who designs trials of medication for T2D) were interviewed about dose escalation. They were asked about factors they consider when deciding whether to escalate the dose of T2D medication, advantages and disadvantages of dose escalation, and the importance of dose escalation relative to other attributes of medication used to treat T2D. The content of the online survey was based on input provided during these qualitative interviews with physicians and a clinical researcher.

The survey began with instructions for completion. Respondents were instructed to answer the questions while thinking about “the broad range of antihyperglycemic medications for type 2 diabetes, including oral treatment like metformin, empagliflozin, and oral semaglutide; basal and meal-time insulin; and non-insulin injectable medications like liraglutide, injectable semaglutide, and dulaglutide.” Then, a series of questions were administered to determine whether the physicians met study inclusion criteria. Physicians who met criteria continued by completing three additional background questions (see physician characteristics in Table 1). The next set of questions assessed the importance of dose escalation relative to other medication attributes. These items asked physicians to select from a list of medication attributes (presented in Tables 2 and 3) to indicate which attributes most commonly prevent patients from reaching treatment goals and which attributes were most commonly perceived as a barrier to prescribing medication.

The final series of questions was designed to provide insight into physicians’ decisions regarding dose escalation. For example, one question asked physicians to select from among a list of factors (presented in Table 4) they consider when deciding whether to escalate a dose of medication for T2D. Another item asked physicians to report the most common reasons for escalating a dose over the past 6 months, again by selecting from a list of potential reasons (presented in Table 5).

After each of the questions where physicians selected multiple responses from a list of options (i.e., Tables 2, 3, 4, 5), physicians were asked to rank their selections in order of importance. The exact wording of the key questions is presented in footnotes below the relevant tables.

After completing the draft survey, it was formatted for online completion and administered to the first 15 participants (i.e., 10 primary care physicians and 5 endocrinologists). Data collection was paused after these initial 15 participants so that results could be examined to ensure that the survey was functioning as intended. Results of the interim analysis led to two edits prior to continuing with data collection. First, the item assessing gender was moved to the screening section earlier in the survey so that gender could be considered as part of the screening criteria. Second, one question was deleted because it appeared to be potentially confusing. After making these two edits, the full data collection was allowed to proceed until the sample size target was met. Because no survey content was changed or added following the initial 15 completions, results from the pilot participants were included in the final dataset.

Descriptive statistics were used to summarize participants’ responses and to characterize the sample in terms of sociodemographic characteristics and clinical background. For continuous variables, the mean, median, standard deviation, and range were calculated. Categorical variables are reported as frequencies and percentages. Analyses were conducted with SAS software version 9.4 (SAS Institute, Cary, NC).