To differentiate diabetic ketoacidosis (DKA) from hyperglycemic ketosis due to cannabis hyperemesis syndrome (HK-CHS) in adults with type 1 diabetes.
Of 295 adults with type 1 diabetes who were seen with DKA-related ICD-10 codes, 68 patients with 172 DKA events meeting the inclusion criteria were analyzed. Cannabis use was defined as a positive urine test result for cannabis. Linear mixed models were used to define HK-CHS (pH ≥7.4 with bicarbonate ≥15 mmol/L [mEq/L]), and sensitivity and specificity were calculated using the receiver operating characteristic (ROC) curve.
Cannabis users had significantly higher pH (7.42 ± 0.01 vs. 7.09 ± 0.02) and bicarbonate (19.2 ± 0.61 vs. 9.1 ± 0.71 mmol/L) (P < 0.0001) compared with nonusers. The area under the ROC curve for a positive cannabis urine test result predicting HK-CHS was 0.9892.
In patients who present with DKA and higher pH, especially pH ≥7.4, cannabis use should be considered in the differential diagnosis.
Cannabis-related emergency visits doubled after legalization of cannabis in Colorado, and 30% of patients with type 1 diabetes reported using cannabis in 2017–2018 (1,2). We previously reported a two- to threefold increased risk of emergency visits due to hyperglycemia and ketosis in cannabis users compared with nonusers with type 1 diabetes (2,3). It is believed that hyperglycemic ketosis in cannabis users with type 1 diabetes is related to cannabis hyperemesis syndrome (CHS), which is characterized by early morning nausea progressing to severe vomiting that causes ketosis, followed by hyperglycemia (2–4), as opposed to typical diabetic ketoacidosis (DKA), where hyperglycemia would precede ketoacidosis mainly due to insulin omission or other precipitating factors. A case series from Australia showed severe alkalosis instead of acidosis in cannabis users with type 1 diabetes who presented with symptoms of DKA (5). Here, we investigated the acid-base parameters, among adults with type 1 diabetes who presented to an emergency department (ED) with symptoms of DKA, to differentiate typical DKA (ketoacidosis) and atypical DKA (ketoalkalosis), which we defined as hyperglycemic ketosis due to cannabis hyperemesis syndrome (HK-CHS).
Research Design and Methods
This study was approved in exempt category by the Colorado Multi-Institutional Review Board (Aurora, CO). Electronic medical records of adults (age ≥18 years) with type 1 diabetes who are followed at the Barbara Davis Center for Diabetes and seen in the ED with a DKA diagnosis (based on ICD-10 codes) between January 2016 and January 2021 were retrieved. All patient medical records were manually reviewed by two physicians (H.K.A. and A.C.) for inclusion and exclusion criteria. We included events that met following criteria (based on American Diabetes Association diagnostic criteria): venous blood glucose >250 mg/dL, anion gap >10, ketones as determined by serum β-hydroxybutyrate (BHB) ≥0.6 mmol/L (6), and available urine toxicology screening test report at the admission. Only the first available laboratory parameters on the day of admission were included. Patients without a urine drug test or with a positive urine drug test result other than cannabis at the ED visit, pregnancy, measurable serum alcohol (if tested), and positive urine ketones without a serum BHB ≥0.6 mmol/L were excluded.
Adults with type 1 diabetes with a positive urine test result for cannabis were defined as cannabis users, and those with a negative test result were defined as noncannabis users. Mean levels of venous pH, serum bicarbonate (HCO3−), anion gap, and BHB were compared between cannabis users and nonusers with type 1 diabetes using linear mixed models. Based on initial analysis (Fig. 1A), we defined HK-CHS as a venous pH ≥7.4 with serum bicarbonate ≥15 mmol/L. We then examined what percentage of HK-CHS episodes occurred in patients with a urine test result positive for cannabis versus in patients with a negative urine test result. We calculated the area under the receiver operating characteristic (ROC) curve and the sensitivity and specificity of cannabis use predicting HK-CHS presentation by using logistic regression analysis. All statistical analyses were conducted using SAS 9.4 software (SAS Institute, Cary, NC). A P value of <0.05 was considered statistically significant.
Of 295 adults with type 1 diabetes who were seen in the ED with DKA-related ICD-10 codes, 68 patients with 172 events meeting inclusion criteria were analyzed (Supplementary Fig. 1). Demographic characteristics of overall cohort and by cannabis use are reported in Supplementary Table 1.
All patients had elevated BHB (mean ± SE) 15.3 ± 2.1 mmol/L among cannabis nonusers and 13.7 ± 2.4 mmol/L among cannabis users (P = NS). In mixed linear regression models adjusted for age and sex, least square mean ± SE for venous pH was 7.09 ± 0.02 and 7.42 ± 0.01 (P < 0.0001), serum anion gap was 23.90 ± 0.71 and 20.94 ± 0.60 mmol/L (P = 0.01), and serum bicarbonate was 9.1 ± 0.71 and 19.20 ± 0.61 mmol/L (P < 0.0001) in patients with cannabis nonuse and cannabis use, respectively.
Of 74 events among cannabis users, 72 (96%) had venous pH ≥7.4 and serum bicarbonate ≥15 mmol/L at the time of the presentation (Fig. 1A and Supplementary Table 2). Based on the findings, we defined hyperglycemic ketosis (venous blood glucose >250 mg/dL, anion gap >10, and serum BHB ≥0.6 mmol/L) as HK-CHS if venous pH was ≥7.4 and serum bicarbonate was ≥15 mmol/L at the time of the ED visit. When this venous pH and bicarbonate cutoff was used, the area under the ROC curve for positive cannabis urine test predicting HK-CHS was 98% (Fig. 1B), with sensitivity of 97% and specificity of 95%.
We conducted a sensitivity analysis using the BHB cutoff of ≥3.0 mmol/L. There were significant differences in pH and bicarbonate levels between cannabis users and nonusers. The area under the ROC curve for predicting HK-CHS was 98%, with sensitivity of 99% and specificity of 98% (Supplementary Table 3).
The diagnosis of DKA is based on a plasma glucose >250 mg/dL, increased anion gap, metabolic acidosis, positive serum ketones, and significant changes in arterial pH and serum bicarbonate with clinical symptoms (6). Parameters of arterial (in clinical practice venous) blood gas, such as pH and bicarbonate with anion gap, are used to classify DKA severity (6).
Physicians mostly rely on pH and bicarbonate to diagnose and classify DKA (6) and make floor versus intensive care unit admission decisions (7). Our study suggested that relying only on pH and bicarbonate can be misleading in cannabis users with type 1 diabetes. With increasing cannabis use and cannabis-related emergency admissions in adults with type 1 diabetes, it is imperative for clinicians to differentiate typical DKA from HK-CHS. Typical DKA is characterized by hyperglycemia, followed by acidosis, while HK-CHS is expected to have ketosis and alkalosis due to severe vomiting, followed by hyperglycemia. Cannabis causes significant gastric delay in animals and humans (8,9), and it is believed that progressive vomiting in cannabis users is likely the cause of metabolic alkalosis (2,3,10).
In patients with DKA, compound acid-base disturbance with a metabolic alkalosis component was described decades ago (11–13). Early reports of metabolic alkalosis with ketosis (aka diabetic ketoalkalosis) were described in patients with type 1 diabetes presenting with progressive vomiting and increased anion gap (11–13). In our study, cannabis users with type 1 diabetes mostly presented with metabolic alkalosis despite high anion gap ketosis. Our study confirmed the findings of a small case series from Australia that noted diabetic ketosis presenting in cannabis users has a different and conflicting acid-base profile compared with noncannabis users (5).
To our knowledge, this is the first study to analyze pH and bicarbonate cutoffs to differentiate between typical DKA and HK-CHS. Our study has many clinical and research implications. We suggest screening urine toxicology for cannabis in adults with type 1 diabetes who present to the ED with blood glucose ≥250 mg/dL, BHB ≥0.6 mmol/L, and have pH ≥7.4 (higher than expected) with bicarbonate of ≥15 mmol/L. This cutoff would predict 98% of HK-CHS events. Recognizing HK-CHS is the first step to counsel patients on cannabis cessation. Levels of pH and bicarbonate in cannabis user should be interpreted with caution to classify DKA severity and decision making for floor or intensive care unit admissions. We suggest prioritizing closing the anion gap and decreasing BHB with fluid replacement and intensive insulin therapy with clinical judgment in admission decisions in cannabis users with type 1 diabetes.
The large sample size, meticulous review of medical records by two physicians, and systematically defined HK-CHS cutoff are major strength of this study. However, the single-center, retrospective design, and limited patients with available urine toxicology results are limitations. Misclassification of DKA due to use of ICD-10 codes cannot be ruled out. Moreover, we could not verify the sequence of symptoms (DKA preceding vomiting or vomiting preceding DKA) from the medical records. Future prospective studies are required to confirm our findings.
In adults with type 1 diabetes presenting with hyperglycemic emergencies, HK-CHS should be considered in those with pH ≥7.4 and bicarbonate >15 mmol/L in presence of ketosis.
This article contains supplementary material online at https://doi.org/10.2337/figshare.17054168.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. H.K.A., G.L.K., A.M., and V.N.S. designed the study. H.K.A., A.C., and V.N.S. did the medical record review. J.S.-B. and G.L.K. did the statistical analysis. All authors contributed to drafting and confirming the manuscript. H.K.A. 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.