Serious Infections in Children with Lupus: Uncommon But Important

This administrative database study has some important caveats for primary care providers, in spite of the limitations in study design.

Source: Hiraki LT, Feldman CH, Marty FM, et al. Serious infection rates among children with systemic lupus erythematosus enrolled in Medicaid [published online ahead of print February 19, 2017]. Arthritis Care Res. doi:10.1002/acr.23219. See AAP Grand Rounds commentary by Dr. Gloria Higgins (subscription required).

First, let's get the limitations out of the way. I like to call administrative database studies the "grain of salt" study design.* Administrative databases have information based on billing codes and the like, with little to no clinical detail available. Researchers can find out diagnoses, medications administered, duration of hospitalizations, and similar details, but such databases generally do not contain lab results nor any nuances of care that would be available from a true chart review. This study of US children with systemic lupus erythematosus (SLE) is taken from a Medicaid administrative database and has some good pearls for frontline pediatric providers.

The researchers looked at the years 2000 - 2006 for Medicaid-enrolled children 4 - 17 years of age, and found only 3500 children with that diagnosis. If you spread that over the entire country (actually in this study, 47 states plus DC were included), the average pediatric primary care practice would likely have only a few SLE patients. Infection rates were in the ballpark of 10 per 100 person-years, so a pediatrician in a medium-sized practice has a reasonable chance of never seeing a child with SLE and serious infection. Still, those same providers need to be able to recognize such situations.

What were the infections? Here, we again need to deal with the grain of salt problem. The investigators excluded systemic candidiasis diagnoses, because of prior data suggesting that discharge diagnosis codes were not accurate for that entity. Furthermore, they didn't count central nervous infections, because of the difficulty of distinguishing infectious causes from lupus encephalitis. Not only does that mean the 10 per 100 person-years number may be too low, it also cuts out a chunk of very important diagnoses from consideration. With that caveat, bacterial infections (mainly pneumonia, bacteremia, and cellulitis) accounted for the vast majority of infections, with herpes zoster the most common viral diagnosis. Fortunately, all of those diagnoses should be familiar to pediatric primary care providers. It didn't appear that the authors attempted to determine how many of the bacteremic episodes were associated with indwelling venous catheters; I suspect that would be tough to do with this database.

So, I'd take this information to mean that front line providers, when evaluating a child with SLE and infection concern, should be careful to perform a thorough evaluation of breath sounds and skin in particular, plus have a low threshold for obtaining blood culture as part of the initial evaluation. It would be interesting to see a chart review study to determine the yield of blood culture in this setting.

*In a little over 6 months, I've gone from a spoonful of sugar to a grain of salt. Of course, I had to do my own evidence search to find the origin of the term "grain of salt," and it seems it is open to debate. Clearly, Pliny the Elder deserves some credit since he published it, as part of a recipe for a poison antidote. Regardless of its origins, my point is the same: studies utilizing administrative databases have serious limitations. Don't rely on the precision of the findings, that's for future studies to define. However, clinicians can still find some savory bites among the grains.