Political and Public Policy Implications of missing data and unlikely values: The example of California's Birth Certificate Data

When conducting the core public health function of surveillance, the quality of data used can have a significant effect on conclusion reached and the public policies that are implemented. An examination of birth certificate data from California counties from 1992 to 2003 indicates that there is considerable variation among local jurisdictions with regard to the amount of both unlikely and missing data. Birth certificates are more likely to be incomplete for infants who subsequently die and for sub-populations that are at risk for poor birth outcomes. Decisions to exclude poor quality records when calculating health indicator rates means that cases with poor outcomes may be eliminated, and lead to incorrectly estimating progress toward Healthy People 2010 objectives. Analyzing the impact of unlikely and missing data from California birth certificates from 1992 to 2003 demonstrates that in 1992, the county-level rate range for improbable gestational age was 0.0% to 13% with a median of 3.2%. In 2003, the range was 0.0% to 17.7% with a median of 5.3%. In 1992, most counties with data problems were rural. In 2003, more counties have data quality problems and most are among the more populous counties. Additional analyses conducted utilizing Kotelchuck's algorithm to correct for unlikely birthweight and gestational age reveal that without using this corrections, preterm birth rates for Blacks and Asians are significantly underestimated while rates for Whites and Hispanics are not. Thus, before concluding that population-based rates are changing, it is important to evaluate and understand the impact of data quality.