Smart Patients Get Smart Care™

The World’s Leading Authority for Chronic Lymphocytic Leukemia Patients

Steroids: Short-Term Use and the Increased Risk of Serious Complications: Correlation versus Causation (Part 3)

In science and medicine, information is constantly changing and may become out-of-date as new data emerge. All articles and interviews are informational only, should never be considered medical advice, and should never be acted on without review with your health care team.


This is the last of a three-part series on the risks and benefits of steroids in CLL.

This section reviews a recent publication on three serious risks associated with short-term steroid use. It also discusses the limitations of this type of research.

While there are many well-known problems with long-term use of steroids and several well-recognized short-term risks (see part 2 on steroids here), this recent open access study from the British Medical Journal (BMJ) published in March, 2017 entitled  Short term use of oral corticosteroids and related harms among adults in the United States: population based cohort study demonstrates that even short term, low dosage use (less than 20 mg of prednisone daily for less than two weeks) was associated with higher rates of sepsis (life-threatening infections), venous thromboembolism (blood clots in the veins that break off and embolize or travel through the blood stream possibly causing dangerous blockages in the lungs and other organs), and fractures within 30 days of starting on steroids.

The study design was to review all prescriptions for oral steroids over three years and then examine the incidence of fractures, sepsis and thromboembolism recorded within the subsequent 30 days since the prescription.


The results were remarkable:

  • More than 1 in 5 patients were prescribed oral steroids in the three year period that was studied.
  • Those studied were privately insured US outpatients between the ages of 18-64.
  • The most common reasons for the prescriptions were to help with upper respiratory tract infections, allergies and spinal conditions.
  • In the 30 days following initiation of oral steroids there was a
    • >5-fold increase in the risk of sepsis.
    • > 3-fold increase in the risk of thromboembolism.
    • Almost 2 fold increase risk of fractures.
  • These risks, although somewhat less, were still there even when “lower doses” of less than 20 mg a day were used.

Scary numbers indeed, but let’s next cut the data in a different way.

The risk of the problems happening to any individual is called the absolute risk.

The absolute risk after taking the steroids was:

  • 5 per 10,000 for sepsis
  • 14 per 10,000 for thromboembolism
  • 51 per 10,000 for a fracture

In other words, the risk of all three of these serious complications combined was considerably less than 1%.

Correlation versus causation

The larger issue is related to the axiom that even the most causal students of statistics will quickly want to quote when critiquing this study. They would tell us:

Correlation does not imply causation.

Just because there is an increased incidence of a complication associated with the prescribing of a particular medication does not mean that the medication caused the problem. It may or it may not be the cause.

Doctors don’t generally prescribe oral steroids to healthy patients or even to sick patients that they expect to quickly and easily recover.

If your upper respiratory infection or allergies or spinal trouble requires a pulse of steroids for control, it is likely more serious and more disabling.

Sicker patients and those with severe spinal back pain needing steroids to reduce swelling and pain may be less mobile. Immobility is a known risk factor for blood clots.

Those with significant respiratory infections or allergies requiring steroids to decrease the inflammation that compromises the proper drainage of bacteria in the airways already are at higher risk for serious infections.

All these patients may be less stable, weaker, and more likely to fall leading to a fracture.

In other words, those needing steroids may already be at higher risk for these complications. The complications may be the result of the severity of their problem requiring the steroids, not the steroids themselves.

The only way to find out for certain if the steroids are the cause of the increased risks would be to prospectively check the outcomes of those who needed steroids and randomize that population to either get steroids or not – a study that is both unethical and impractical.

That said, much of the data that we rely on about smoking and lung cancer is correlative. So too are many of the studies on blood pressure and strokes, or obesity and diabetes. There are many more examples.

Clearly there is an important role for this kind of correlative research.


Steroids are life-saving medications in CLL and should be used when needed, but they should be used in as low a dose as possible for as short a time as possible.

This study raises the issue that even short-term use is risky, but the absolute risk is small and the research does not prove that the steroids themselves are the cause of the increased complications.

You might want to read the article here and decide for yourself. This publication reminds us of the importance of looking beyond the headlines found in the abstract and digging into the details of any research.

Brian Koffman, MD  5/9/17