Etiology
Short-term use of oral corticosteroids was
linked to increased risk for sepsis, VTE,
and fractures
Clinical impact ratings: 多多多多多多夞 多多多多多夞夞
Question
In adults, is short-term use of oral corticosteroids associated
with adverse events (sepsis, venous thromboembolism [VTE], or
fracture)?
Methods
Design: Population-based cohort study with linkage of a nation-
wide health care insurer's health records and outpatient phar-
macy service between 2012 to 2014 comparing all oral cortico-
steroid users with nonusers; and a self-controlled case series
comparing within-patient treatment and posttreatment periods
with previous untreated periods.
Setting: USA.
Patients: 1 548 945 adults 18 to 64 years of age (327 452 corti-
costeroid users, mean age 45 y, 51% women; 1 221 493 nonus-
ers, mean age 44 y, 44% women) who were enrolled with the
insurer for ≥ 1 year before study start and continuously between
2012 to 2014. Exclusion criteria were use of oral corticosteroids
during 2011, ≥ 30 days of cumulative oral corticosteroid use
during the study period, exclusive use of nonoral corticosteroids
or oral budesonide, solid organ or bone marrow transplanta-
tion, cancer, or adverse events of interest (sepsis, VTE, or frac-
ture) during 2011.
Risk factors: Oral corticosteroid use for < 30 days.
Outcomes: Corticosteroid use; adverse events (first recorded
diagnosis for each of sepsis, VTE, or fracture) identified through
International Classification of Diseases, Ninth Revision, Clinical
Modification, diagnostic codes; in the self-controlled case se-
ries, first adverse events 5 to 30 days and 31 to 90 days after
prescription fill-date were compared with events before the fill
date.
Main result
21% of patients received ≥ 1 prescription for oral corticoste-
roids for a median of 6 days. Overall incidence rates for sepsis,
VTE, and fracture were 1.8, 4.6, and 21.4/1000 person-years at
risk, respectively, in the corticosteroid user group; and 1.0, 2.4,
and 14.3/1000 person-years at risk, respectively, in the nonuser
Association between short-term use of oral corticosteroids
and adverse events in adults*
Outcomes N Incidence rate ratio (95% CI)
At 5 to 30 d† At 31 to 90 d†
Sepsis‡ 1556 5.30 (3.80 to 7.41) 2.91 (2.05 to 4.14)
Venous thromboembolism§ 4343 3.33 (2.78 to 3.99) 1.44 (1.19 to 1.74)
Fracture|| 20 090 1.87 (1.69 to 2.07) 1.40 (1.29 to 1.53)
*Abbreviations defined in Glossary. Reference period was 5 to 180 d before prescrip-
tion date.
†Days after prescription fill date.
‡Admission to hospital for sepsis; adjusted for use of antibiotics, 5-HT3–receptor
antagonists, antidepressants, antiinflammatory agents, antimuscarinics, opiate ago-
nists, and phenothiazine.
§Inpatient or outpatient claims; adjusted for use of antibiotics, androgens, anxiolytics,
antiinflammatory agents, azoles, calcium-channel blockers, coumarin, diuretics, opi-
ate agonists, and platelet aggregation inhibitors.
||Inpatient or outpatient claims; adjusted for use of antiinflammatory agents, cycloox-
ygenase 2 inhibitors, and opiate agonists.
doi:10.7326/ACPJC-2017-167-4-020
姝 2017 American College of Physicians JC20
Downloaded From: https://annals.org/ by a Columbia University User on 08/15/2017
Waljee AK, Rogers MA, Lin P, et al. Short term use of oral corticoste-
roids and related harms among adults in the United States: popu-
lation based cohort study. BMJ. 2017;357:j1415.
group. The results of the self-controlled case series are in the
Table.
Conclusion
In adults, short-term use of oral corticosteroids was associated
with increased risk for sepsis, venous thromboembolism, and
fractures.
Source of funding: No external funding.
For correspondence: Dr. A.K. Waljee, University of Michigan,
Ann Arbor, MI, USA. E-mail awaljee@med.umich.edu. 
Commentary
The analysis of the large database by Waljee and colleagues
found increased risk for sepsis, fracture, and thromboembolic
events 5 to 90 days after filling a prescription for a short course
of oral steroids. Steroids directly affect white blood cell counts
and bone remodeling as early as the first dose (1). The causal
link between steroids and VTE is less clear.
This retrospective cohort study design allowed for the analysis
of > 300 000 steroid users, but it comes with the potential for
confounding. It is reasonable that sicker people are both at
“risk” for receiving steroids for their illness and at risk for sepsis,
fracture, and VTE from the underlying condition. The study was
designed to minimize confounding; patients were included if
they received ≤ 30 days of steroids during 2012 to 2014 but had
not received steroids in 2011, and they were compared with
themselves rather than a control group. Any important unmea-
sured factors would presumably (but not certainly) be present in
patients before and after they received steroids.
More than 300 000 patients received steroids, and 1556 were
admitted for sepsis. The rate of admission for sepsis for any
given patient was 5 times higher in the 30 days after a steroid
prescription than in the 180 days before. Sepsis isn't common,
but the increased risk is clinically important.
The most common indications for a short course of steroids in-
cluded upper respiratory infections, allergies, bronchitis, and
musculoskeletal conditions. Steroids are effective for many of
these conditions. In practice, they have been considered a bet-
ter option than opioids or antibiotics. The results of this study
should not stop the prescription of steroids, but should prompt
clinicians to consider safer alternatives and use steroids only
when treatment benefits outweigh the risks.
Kate Rowland, MD, MS
Rush University
Chicago, Illinois, USA
Reference
1. Kauh E, Mixson L, Malice M-P, et al. Prednisone affects inflammation, glu-
cose tolerance, and bone turnover within hours of treatment in healthy indi-
viduals. Eur J Endocrinol 2012;166:459-67.
ACP Journal Club Annals of Internal Medicine 15 Aug 2017