It is important to monitor the effects of medicines during pregnancy, to ensure they are effective and safe for the mother and unborn child in real world settings as well as trials. Existing patient clinical care records represent an opportunity to answer important questions about medicines taken during pregnancy. To help investigate this, a register of pregnancies in the Clinical Practice Research Datalink, which includes anonymised information on the start of each pregnancy and its outcome (live birth, still birth or pregnancy loss), has been created. However, some women have more than one pregnancy in the Register which appear to overlap. At least one of these records must be an error, and it is unclear how to handle these records in research studies. If we do not know which, if either, are correct and at which timepoints a woman was truly pregnant, it makes studying the effects of medicines difficult. This study intends to investigate potential reasons why these apparently overlapping pregnancies may occur in the Register. This work follows on from a similar study looking at pregnancies with missing outcomes in the Register. The results of this study will provide additional information about these types of pregnancy to researchers using the data for their own studies Information from these studies will then potentially be used to improve the methods by which the Register is created. Improvements will make this valuable resource more useful enabling researchers to investigate important issues such as the safety and effectiveness of medicine during pregnancy.
The Pregnancy Register algorithm generates a list of all pregnancies determined in the Clinical Practice Research Datalink (CPRD). A record in the register represents a pregnancy episode and includes information on pregnancy start and outcome. However, there are approximately half a million pregnancies which overlap with another pregnancy in the Register. Scenarios have been identified based on the algorithms logic and how the data is structured which may explain this. The scenarios describe four problems; Both pregnancies are real but one episode is a historical pregnancy; Both pregnancies are historical; Both pregnancies are real but the gestation of the pregnancies applied by the algorithm is wrong; The pregnancies are really one pregnancy which has been identified as two by the algorithm. Descriptive analysis will use an algorithmic approach to query CPRD data and linked datasets to look for supporting evidence for each of these scenarios. Potential reasons for why overlapping pregnancies may have been generated by the algorithm will be tabulated. This work follows on from a previous study which assessed pregnancies without an outcome recorded in the Register (ISAC 17_285R_2) Evidence from these studies will then be used to improve the Pregnancy Register algorithm to reduce the occurrence of overlapping pregnancies and increase the usefulness of this resource.
Pregnancy records relating to the antenatal, perinatal and postnatal period, and the pregnancy outcome in order to further understand which of the overlapping pregnancies in the Pregnancy Register are correct.
Jennifer Campbell - Chief Investigator - CPRD
Jennifer Campbell - Corresponding Applicant - CPRD
Caroline Minassian - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Helen McDonald - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Krishnan Bhaskaran - Collaborator - London School of Hygiene & Tropical Medicine ( LSHTM )
Rachael Williams - Collaborator - CPRD
CPRD Mother-Baby Link;HES Admitted Patient Care;HES Diagnostic Imaging Dataset;HES Outpatient;Pregnancy Register