Why DKA Detection Needs to Start Before the Emergency Department

In the first episode of Patient Care Online’s Clinical Forum series on diabetes monitoring, Eden Miller, DO, founder and CEO of Diabetes and Obesity Care, and a panel of clinicians emphasized that diabetic ketoacidosis can no longer be viewed as a complication confined to classic type 1 diabetes. The discussion centered on a practical message for primary care: clinicians need to recognize DKA risk earlier, normalize ketone conversations, and prepare for a future in which glucose and ketone data may be interpreted together.

The June 16, 2026, Patient Care Clinical Forum, “Bridging the Gap in Diabetes Monitoring: Early DKA Detection and Optimized Glucose Management Across the Care Continuum,” focused on DKA recognition gaps, current practice barriers, and the outlook for continuous ketone sensing. Miller, a family medicine specialist with training in diabetology and obesity medicine, opened the session by comparing the current state of ketone monitoring to earlier eras of glucose monitoring, when diabetes care moved from urine glucose testing to fingersticks and then to continuous glucose monitoring. She said the same “journey” is now beginning with ketone sensing and encouraged clinicians to start having more direct conversations with patients about ketones, DKA risk, and sick day management.

Why is DKA being missed?

A central theme of episode 1 was that DKA is rising and remains vulnerable to delayed recognition, particularly when clinicians rely on outdated assumptions about who is at risk. Participants cited cases in which DKA was missed in the emergency department, including one clinician-patient who reported being denied ketone testing for 12 hours because of age and BMI despite symptoms consistent with DKA.

Miller described that same case during the discussion: a patient with a BMI greater than 40 was presumed to have type 2 diabetes, placed on metformin, and later presented to the emergency department with symptoms including tachypnea and Kussmaul breathing. According to Miller, the patient said it took 12 hours of advocating before ketone testing was performed. Her conclusion was direct: preconceived ideas about who “looks like” a DKA patient can delay diagnosis.

For primary care physicians, the implication is that DKA risk assessment should not stop at diabetes type, age, or body habitus. Miller noted that DKA is still predominantly associated with type 1 diabetes or autoantibody-positive disease, but she also highlighted a growing group of ketosis-prone patients with type 2 diabetes. She estimated that about two-thirds of DKA occurs in individuals with type 1 diabetes or autoantibodies, while about one-third occurs in individuals with type 2 diabetes.

Why are patients with type 2 diabetes increasingly relevant?

The episode repeatedly returned to the idea that patients with type 2 diabetes are the fastest-growing DKA cohort. Increasing concern among clinicians about antibody-negative type 2 diabetes, particularly in patients receiving sodium-glucose cotransporter-2 inhibitors, following ketogenic diets, or becoming insulin-deficient over time.

Miller emphasized that SGLT2 inhibitors can “hide” the degree of glycemic deterioration because they offload glucose through the urine. She described this as “lower threshold DKA,” arguing that glucose values below traditional DKA expectations should not reassure clinicians when ketones are elevated. She also flagged a particularly high-risk combination: ketogenic diet, SGLT2 inhibitor use, and relative insulin deficiency.

That point was reinforced through several real-world examples. In one case, Miller described a patient with latent autoimmune diabetes in adults who was taking a GLP-1 receptor agonist, using a low-dose SGLT2 inhibitor, and following a ketogenic diet. When ketones were checked, the patient had moderate ketones despite not reporting symptoms. Another clinician described a patient in a weight loss clinic who appeared to have type 2 diabetes, was treated with a GLP-1/SGLT2 combination, was taken off insulin, and later presented to the emergency department with DKA.

What are the limits of current ketone testing?

Episode 1 also addressed the gap between what clinicians know they should monitor and what is feasible in practice. Urine strips remain the default for many participants, not because they are preferred clinically, but because of access constraints. Participants acknowledged urine testing’s lag, inconvenience, and lack of real-time actionability compared with blood-based measurement.

Miller explained that current ketone monitoring in the US generally relies on either fingerstick ketone meters, which measure beta-hydroxybutyrate, or urine testing, which measures acetoacetate. Urine testing, she noted, lags behind real-time physiology, while fingerstick testing can be accurate but is still not continuous or easily actionable.

The discussion also highlighted protocol gaps. When Miller asked whether participants had ketone monitoring protocols in their organizations, one participant responded bluntly that “no formal protocol exists” was the polite way to describe current practice. Miller said most participants did not have a standard protocol and called the lack of outpatient standardization “frustrating.” The panel echoed that finding, noting that outpatient ketone monitoring protocols are largely absent for patients on SGLT2 inhibitors, following ketogenic diets, or with prior DKA history.

Why do patients need ketone action plans?

One of the clearest clinical recommendations from the episode was that patients should leave the hospital or emergency department after DKA with a ketone action plan. Miller drew an analogy to an EpiPen after an allergic reaction: if patients are discharged with tools and instructions after anaphylaxis, she argued, patients hospitalized for DKA should not leave without a ketone plan, follow-up, and monitoring strategy.

The panel described discharge without a ketone plan as a systemic failure, noting that some patients leave DKA hospitalizations without actionable monitoring plans and may not even be formally told they had DKA. Miller later said patients should have a ketone plan even before dual glucose-ketone sensors are available, adding that clinicians should discuss ketones, sick day management, and emergency department expectations more consistently.

What could continuous ketone sensing change?

The final portion of the episode looked toward dual glucose-ketone sensing. Participants described the technology as a potential paradigm shift, not only for DKA prevention but also for diabetes phenotyping. The panel noted that clinicians compared the potential impact of dual glucose-ketone sensors to the shift from urine glucose monitoring to CGM.

Miller said a ketone level greater than 3 mmol/L represents an urgent threshold and described a likely alert structure in which 3 mmol/L would be a high ketone alert, with optional lower alerts at less urgent levels. That tiered approach appeared to address some concerns about alarm fatigue; the panel noted that participants found reassurance in a hard threshold at 3 mmol/L with optional lower alarms.

Clinicians also discussed a possible professional-use dual glucose-ketone sensor that could be applied situationally—for example, around surgery, steroid courses, or in a high-suspicion type 2 diabetes case—without requiring full insurance authorization. One participant said a professional-use device could allow clinicians to place a sensor temporarily for patients on SGLT2 inhibitors before surgery, after steroid exposure, or when clinical suspicion for ketosis is high.