A
practical approach for calculating reliable cost estimates from
observational data: application to cost analyses in maternal and child
health.

Salemi JL¹, Comins MM, Chandler K, Mogos MF, Salihu HM.

Author information

Abstract

BACKGROUND:

Comparative
effectiveness research (CER) and cost-effectiveness analysis are
valuable tools for informing health policy and clinical care decisions.
Despite the increased availability of rich observational databases with
economic measures, few researchers have the skills needed to conduct
valid and reliable cost analyses for CER.

OBJECTIVE:

The
objectives of this paper are to (i) describe a practical approach for
calculating cost estimates from hospital charges in discharge data using
publicly available hospital cost reports, and (ii) assess the impact of
using different methods for cost estimation in maternal and child
health (MCH) studies by conducting economic analyses on gestational
diabetes (GDM) and pre-pregnancy overweight/obesity.

METHODS:

In
Florida, we have constructed a clinically enhanced, longitudinal,
encounter-level MCH database covering over 2.3 million infants (and
their mothers) born alive from 1998 to 2009. Using this as a template,
we describe a detailed methodology to use publicly available data to
calculate hospital-wide and department-specific cost-to-charge ratios
(CCRs), link them to the master database, and convert reported hospital
charges to refined cost estimates. We then conduct an economic analysis
as a case study on women by GDM and pre-pregnancy body mass index (BMI)
status to compare the impact of using different methods on cost
estimation.

RESULTS:

Over 60 % of inpatient charges for
birth hospitalizations came from the nursery/labor/delivery units, which
have very different cost-to-charge markups (CCR = 0.70) than the
commonly substituted hospital average (CCR = 0.29). Using estimated
mean, per-person maternal hospitalization costs for women with GDM as an
example, unadjusted charges ($US14,696) grossly overestimated actual
cost, compared with hospital-wide ($US3,498) and department-level
($US4,986) CCR adjustments. However, the refined cost estimation method,
although more accurate, did not alter our conclusions that
infant/maternal hospitalization costs were significantly higher for
women with GDM than without, and for overweight/obese women than for
those in a normal BMI range.

CONCLUSIONS:

Cost estimates,
particularly among MCH-related services, vary considerably depending on
the adjustment method. Our refined approach will be valuable to
researchers interested in incorporating more valid estimates of cost
into databases with linked hospital discharge files.