Introduction
Short courses of oral corticosteroids are commonly prescribed for inflammatory, allergic and respiratory conditions. In people with type 2 diabetes, clinical attention often focuses on hyperglycaemia, while other potential complications of brief treatment may receive less attention.
Yao and colleagues used Taiwan’s National Health Insurance Research Database to examine serious adverse events after oral corticosteroid bursts in adults with type 2 diabetes.
How was the study conducted?
The population-based study included 36,048 adults with type 2 diabetes who received an oral corticosteroid burst between 2008 and 2022. A burst was defined as continuous treatment for 14 days or less; the median course lasted 3 days.
A self-controlled case series design compared event rates after corticosteroid initiation with baseline rates in the same individuals. This helps control for stable individual characteristics, although changes in illness severity and other time-varying factors may still influence the results.
| Study feature | Details |
|---|---|
| Population | 36,048 adults with type 2 diabetes |
| Mean age | 61.6 years |
| Exposure | Oral corticosteroid course lasting ≤14 days |
| Median course duration | 3 days |
| Assessment periods | Days 6–30 and 31–90 |
| Outcomes | Gastrointestinal bleeding, pneumonia, sepsis, heart failure and fracture |
What happened during the first 30 days?
After correction for multiple comparisons, three serious outcomes were significantly more frequent during days 6–30:
| Outcome | Adjusted incidence rate ratio | 95% confidence interval |
|---|---|---|
| Pneumonia | 2.06 | 1.65–2.57 |
| Heart failure | 1.87 | 1.27–2.77 |
| Gastrointestinal bleeding | 1.71 | 1.52–1.94 |
Pneumonia had the largest point estimate, with an observed event rate approximately twice the baseline rate. Heart failure and gastrointestinal bleeding were also more frequent.
These relative measures do not show how many additional patients experienced an event. An incidence rate ratio of 2.06 does not mean that 2.06% of patients developed pneumonia. Absolute excess risks and numbers needed to harm were not reported.
Sepsis and fracture did not remain statistically significant during the early period after correction for multiple comparisons. Syncope, used as a negative-control outcome, was not associated with corticosteroid exposure.
Did the associations persist beyond 30 days?
During days 31–90, gastrointestinal bleeding remained more frequent, with an incidence rate ratio of 1.26 (95% confidence interval 1.14–1.38). Fracture was also associated with corticosteroid bursts during this period, with an incidence rate ratio of 1.42 (95% confidence interval 1.21–1.68).
The early pneumonia association did not persist into the later period. Heart failure was also not included among the statistically supported 31–90-day findings after correction for multiple comparisons.
The timing therefore differed by outcome: pneumonia and heart failure signals were concentrated within the first month, gastrointestinal bleeding persisted longer, and fracture appeared during the later period.
How should the findings influence prescribing?
The study does not show that short corticosteroid courses should be avoided whenever a patient has type 2 diabetes. Corticosteroids may remain appropriate when the indication is clear and the expected benefit is substantial.
The findings support a more deliberate assessment before prescribing:
| Clinical question | Practical consideration |
|---|---|
| Is systemic corticosteroid treatment clearly indicated? | Confirm that it is likely to provide meaningful benefit. |
| Is an appropriate alternative available? | Consider local, inhaled, topical or non-steroidal treatment where supported. |
| Is the patient particularly vulnerable? | Consider previous gastrointestinal bleeding, infection risk, heart failure, frailty, fracture risk and interacting medicines. |
| Is the planned course appropriate? | Avoid extending treatment without reassessing the indication and response. |
| Is follow-up required? | Consider early review in patients with substantial comorbidity or worsening symptoms. |
Patients should receive advice appropriate to their individual risk. Fever, worsening respiratory symptoms, new oedema or breathlessness, evidence of gastrointestinal bleeding, marked hyperglycaemia or a suspected fracture should prompt clinical reassessment.
What are the limitations?
This was an observational study. The conditions that prompted corticosteroid treatment may themselves have contributed to subsequent complications. Although the investigators adjusted for several acute illnesses and medicines, residual time-varying confounding cannot be excluded.
Prescription data could not confirm whether patients took the medication. Lifestyle information was unavailable, and some outcome misclassification was possible.
The study was conducted in Taiwan, so differences in population characteristics and prescribing practices may affect its generalisability. The lack of absolute event rates also limits direct estimation of the benefit–harm balance for an individual patient.
Clinical takeaway
Brief oral corticosteroid treatment should not automatically be considered low risk in adults with type 2 diabetes. Gastrointestinal bleeding, pneumonia and heart failure were more frequent during the first 30 days after initiation in this nationwide observational study. The findings support careful patient selection, a clear indication and proportionate follow-up, while recognising that the study demonstrates association rather than causation.
Full reference
Yao TC, Hsu CC, Sheu WHH, et al. Oral Corticosteroid Bursts and Risk of Serious Adverse Events in Adults With Type 2 Diabetes. JAMA Network Open. 2026;9(9):e2635604. Published September 23, 2026. doi:10.1001/jamanetworkopen.2026.35604