Clinical studies on glucocorticoid should document adherence to risk prevention strategies: comment on the article by Madanchi et al

To the Editor:

I read with interest the report by Madanchi et al on the role of prednisone and 25‐hydroxyvitamin D (25[OH]D) and osteoporosis in systemic lupus erythematosus (SLE). ¹ The authors conclude that prednisone dose, even at 5 mg/day, and low mean 25(OH)D were found to be (potentially) modifiable predictors of osteoporosis. They also recommend avoiding prednisone use at any level in SLE to reduce the risk of bone loss and osteoporosis.

Several points can be raised. First of all, observational studies on the relationship between an adverse outcome and glucocorticoid treatment are strongly subject to confounding by indication, that is, where the indication (disease activity) is related both to the exposure (active disease more likely to be treated with prednisone) and the outcome (osteoporosis occurring due to active disease alone). ² The authors have tried to address this by looking at the residual effect of the factor prior disease activity (expressed somewhat crudely as a binary factor of mean SELENA‐SLEDAI [Systemic Lupus Erythematosus Disease Activity Index] > 2, yes or no) on bone mineral density (BMD) after correcting for prednisone use, but that is getting things in reverse. Rather, they should have looked at the residual effect of prednisone use (the exposure) on BMD after correcting for disease activity (the confounder). But even that analysis may be insufficient to fully correct for strong confounding. As disease activity and the choice and dose of glucocorticoids have a complicated relationship, disentangling the “pure effect” of glucocorticoids may require more complicated longitudinal modeling that accounts for the beneficial effect of current and prior glucocorticoid dose on disease activity.^{3, 4} For instance, when we compare two patients with the same level of disease activity, but only one is on prednisone, this patient would almost certainly have had a higher disease activity level if taken off the drug, something not accounted for by simply including disease activity in the model.

Second, I find no mention of recommended strategies to avoid osteoporosis and especially fractures, most importantly by protocolized bone density assessments in all patients treated with glucocorticoids for longer periods and application of antiresorptive agents in patients found to have a T score < –2.5. One is left wondering if and to what extent such strategies were applied in this cohort.

Third, recommending the avoidance of prednisone use at any level in SLE is overly simplistic and disregards the ongoing beneficial and even life‐saving effects of the drug. If it were so easy to avoid, why are so many patients on glucocorticoid therapy long term, even in this referral center? I think glucocorticoid‐avoidance strategies should be coupled with diligent action to avoid and treat adverse effects, such as bone loss, in patients in whom glucocorticoids cannot be stopped. ⁵

Supporting information

Disclosure form.