In hospitalized patients, tramadol use was shown to be associated with hypoglycemia, according to a case-control study recently published in the Journal of Diabetes & Metabolic Disorders.1
In the United States, tramadol is among the top 25 most frequently prescribed drugs.2 Recent epidemiological surveys of diabetic and nondiabetic outpatients taking therapeutic doses of tramadol indicate that the drug is associated with an increased risk for symptomatic hypoglycemia and a related increased risk for hospitalization.3 “However, the overall magnitude of hypoglycemic risk associated with tramadol remains unclear and no recommendations for patient monitoring are currently available,” note the investigators in the current study1
In a retrospective analysis of hospitalized patients conducted at the University of Colorado Hospital in Aurora, computerized inpatient prescription data collected from July 2013 to June 2015 were searched to identify patients >18 years of age who had received ≥1 dose of oral tramadol for acute or chronic pain management and had ≥2 blood glucose measurements performed within 5 days of the first recorded in-hospital tramadol administration. Hypoglycemia was defined as having ≥1 recorded blood glucose concentration ≤70 mg/dL. A contemporary comparator group of patients receiving ≥1 dose of oral immediate-release oxycodone was similarly identified, with data collected during the month of December 2015.
The presence/absence of a diabetes mellitus diagnosis in the patients’ history/current problem list was used to aggregate patients into 2 sets: patients with no documentation (non-diabetic-set), and patients with type 1 (T1DM-set), or type 2 (T2DM-set) diabetes mellitus (DM) . The medical records of patients in the tramadol group non-diabetic-set were further examined for factors that may contribute to the development of hypoglycemia. The sub-group of patients with potential contributing factors was excluded from intragroup comparisons of hypoglycemia occurrence rates.
In the tramadol group (n=3588), 2927 patients (81.6%) had inclusionary blood glucose data available (n=2207 [75.4%] non-diabetic; n=47 [1.6%] with T1DM; n=673 [23.0%] with T2DM)). A total of 1155 patients were in the oxycodone group, with 967 patients (83.7%) with inclusionary blood glucose data available (n=716 [74.0%] non-diabetic patients; n=36 [3.7%] with T1DM; n=215 [22.2%] with T2DM).
Intragroup comparisons of blood glucose concentrations showed no association between a DM diagnosis and blood glucose in either group. Intergroup comparisons showed lower blood glucose in the tramadol vs oxycodone group in the non-diabetic-set (mean nadir blood glucose, 91.9±16.4 vs 98.1±19.9 mg/dL; P <.001), but no appreciable differences between treatment groups in the T1DM-set (P =.381), or the T2DM-set (P =.122).
Intragroup comparisons indicated that in the tramadol group the occurrence of hypoglycemia was lower in non-diabetic patients vs patients with T1DM or T2DM (n=103 [4.7%] vs n=22 [46.8%] for T1DM or n=113 [16.8%] for T2DM; P <.001 for both). In non-diabetic patients, causality between hypoglycemia and tramadol was deemed possible in 1.2% of patients, and probable in 3.5% of patients. In non-diabetic patients, inter-group comparisons showed higher hypoglycemia occurrence in the tramadol vs oxycodone group/group (3.5% vs 1.1%; P <.002); no significant differences between treatment groups were found in patients with T1DM or T2DM. In patients with no diabetes, tramadol-associated hypoglycemia was observed in younger patients (52.0±17.19 vs 57.3±18.61 years; P =.027) and predominantly female patients (74.0% vs 59.8%; P =.012) compared with patients who received tramadol but did not experience hypoglycemia.
“Blood glucose monitoring should be performed in patients with diabetes,” and it should be “strongly considered for hospitalized patients who do not have diabetes when tramadol therapy is initiated,” concluded the researchers.
Limitations and Disclosures
- The study was performed in a single institution, used an uncontrolled design, and suffered from relatively small sample sizes
- Diagnostic categorization depended on information obtained from patients’ snapshot histories/problem lists
- Accounting of symptoms was subjective, and the need for hypoglycemic episode treatment intervention was not evaluated