Search

No standards exist for reporting the accuracy and completeness of pregnancy histories created by linking the birth and fetal death certificates for all the deliveries occurring to one woman. To link certificates, analysts use deterministic and/or probabilistic approaches. Errors in linkage occur randomly and non-randomly. Any type of error can cause incorrect estimation of the magnitude of relationships. Methods for assessing linkage correctness are proposed. Analysts can detect errors in linkage by comparing the linkage results with the pregnancy history reported by the mother. The analyst interviews a random sample of women, ascertaining the dates and outcomes (stillbirth or live birth) for their births and, if the linkage used certificates from one state, the state where they occurred. For each woman, he/she then assesses the accuracy and completeness of this linkage by comparing it with her reported pregnancy history. An alternative approach is to chronologically sequence each woman's births. The parity for the most recent birth shows the number of babies born to the same woman and should equal the number of births that the analyst linked. In the absence of maternally reported pregnancy histories, an analyst can use data on the certificates to assess linkage correctness. Although this will show whether the correct number of births have been linked, it provides no information concerning the accuracy of linkages. It may be, however, the most universally applicable way of reporting the completeness of the linkage of deliveries.

Data from the Jamaican Perinatal Mortality Survey had been used to create a statistical model using logistic regression. From this a simple additive scoring system to predict perinatal death was devised and tested on the 2 cohort months of the study. The score had a theoretical range of 0-28 points, with the higher the score, the greater the likelihood of a perinatal death. For a cut-point of 7, sensitivity was 43% and specificity 84%. A cut-point of 8 resulted in 27% sensitivity and 94% specificity. Higher cut-points resulted in much reduced sensitivity but enhanced specificity (e.g. cut-point 10: 11% sensitivity, 99% specificity). However, it is likely that these estimates are optimistically high, and to achieve unbiased estimates of sensitivity and specificity the score needs to be tested on a sample of the population from which it was not derived before implementation takes place. Meanwhile, the cut-off level for implementation will depend on appropriate resources available.