Limitations of Databases

To the Editor.

The study by Marbella et al¹ uses administrative health data to examine patterns of rehospitalization and hospital charges by newborn hospital length of stay. As federal legislation mandating reimbursed minimum length of stay after delivery recently became effective, this study could not be more timely. Although we agree in the main with both the methods and the conclusions drawn by Marbella et al, we wish to raise some additional issues for the benefit of others who may conduct research in this arena.

First, researchers should pay close attention to strengths and limitations of administrative and clinical health data.^2–4 Points for consideration include: 1) the attribution of individual patients to diagnosis and procedure codes, and diagnosis-related groups (DRGs) in hospital discharge databases; 2) quantifying length of stays in relation to maternal admission for labor and delivery and neonatal admission after delivery; and 3) the curious relationship between hospital charges, which are reported in administrative health databases, and costs, which are not.

Second, careful …