KEYWORDS:
state immunization coverage, communication

BACKGROUND:
CDC reports state-specific coverage results from the National Immunization Survey (NIS) as point estimates with confidence intervals. Often, these are translated into ranked lists used to compare coverage among states. Also, states need to compare their new point estimates with those from previous years. However, even though the NIS is one of the largest public health surveys, its state-level margin of error—reflected in the confidence intervals—often precludes confident claims that any state’s coverage is actually higher or lower than that of another state, or of the same state for a previous year.

OBJECTIVE:
Demonstrate the uncertainty of ranks by point estimates of coverage. Second, demonstrate new state-specific graphics available on-line from NIP to enable each state to (1) interpret state-specific NIS ranking results as precisely as the survey allows, and (2) communicate this information clearly to policy makers and media.

METHOD:
Using NIS data for 2001, simulation methods (Monte Carlo) were used to quantify the uncertainty in each state’s rank. Second, a graphical technique introduced by the U.S. Educational Testing Services depicts results of an analysis (constructing confidence intervals from the differences between state coverage estimates) that permits confident claims of which states have higher and lower coverage than a chosen reference state.

RESULT:
Ranks show considerable uncertainly. For example, the simulation method showed that the state ranked 10th by point estimate could have been as high as 4th to as low as 28th (1999 data). The state-by-state graphics are relatively easy to produce and facilitate meaningful comparisons of one state to the others.

CONCLUSION:
NIP will make state-by-state graphics available on line.

LEARNING OBJECTIVES:
Considerable uncertainty exists when ranking states by point estimates of coverage. Using a new graphic to interpret each year’s results and communicate to policy makers and media.