The standard approach to sampling in all 40 US states in which PRAMS is in operation involves use of state birth certificate files to generate stratified, random samples of women with recent live births . However, in Ireland, access to the Irish birth certificate file is restricted under the Data Protection Act 1988 . Consequently, Cork University Maternity Hospital (CUMH), a large urban obstetric unit in the south of Ireland delivering almost 9,000 babies per year  or approximately 12%  of all Irish births was chosen to test the PRAMS methodology.
Using a sampling frame of live births recorded 3 – 5 months before the study start date at CUMH, 124 women were randomly sampled using a random start and constant sampling fraction. The sample size for the pilot was based on methods recommended by Thabane et al., 2010  using a confidence interval approach to estimate sample sizes for pilot studies. To estimate a projected response of 65% [based on CDC minimum weighted response rates] with a lower confidence limit of 60% and upper confidence limit of 75% we sampled 124 women to the study. Women were included by manually counting and extracting a record from each of two delivery books, one recording caesarean sections and the other recording both spontaneous and instrumental vaginal births. In line with the PRAMS protocol mothers of stillbirths, neonatal deaths and triplets or more were excluded . This was done using the APGAR score recorded on the delivery record and cross-checking with hospital records for the period. Sampling was done proportionally in a ratio of 1:4 to represent the underlying proportion of caesarean deliveries in the population of approximately 25% . Available information including name, address, maternal age, infant sex, gestational age at delivery, delivery method, APGAR score and birth weight were extracted from delivery records of selected women.
PRAMS protocol and study materials
Letters and information sheets were prepared explaining the purpose of the study and its importance to improving maternal and child health in Ireland. The PRAMS survey covers a range of topics including exposures during pregnancy such as alcohol and smoking, care received, and socio-demographic information .
For this study, questions were carefully modified for content, language and overall layout. Content which pertained specifically to the United States such as Medicaid status and enrolment in the Women, Infants and Children Supplemental Nutrition Programme (WIC) were removed and replaced with medical card and private health insurance questions relevant to Ireland. Language changes were applied to conform to commonly used terms in Ireland such as; “antenatal” rather than “prenatal”; “contraception” rather than “birth control” and “health care professional” rather than “health care worker”. Demographic information including race/ethnicity, marital status, nationality, educational attainment and health insurance status provided by the birth certificate file in the United States were added to the questionnaire, as these were not available from the CUMH delivery record. Validated questions from other longitudinal or cross sectional studies were used to supplement the PRAMS questionnaire or replace questions where necessary including questions on pregnancy history from the Avon Longitudinal Study of Parents and Children (ALSPAC) , questions on complications during pregnancy from GUI  and Growing up in Australia  and questions on sexually transmitted infections from The Irish Survey of Sexual Health and Relationships (ISSHR) . These questions were chosen over some PRAMS Phase 6 questions for comparability.
Recent research suggests that the ascertainment of alcohol exposure during pregnancy in studies focused on documenting patterns of alcohol consumption during gestation could be optimized by examining more carefully, the dose, pattern and timing of exposure . Moreover, the most recently available estimates in the Irish general population suggest that up to 77% of women regularly drink alcohol in Ireland compared to an EU average of 68% while over 42% of all female drinkers are classified as having harmful drinking patterns . As a result, we developed questions on maternal alcohol consumption specific to these needs for PRAMS in Ireland which took into account the dose, pattern and timing of alcohol exposure during pregnancy based on work by O’Leary et al., 2010 .
Diet around the time of pregnancy has a substantial impact on maternal and infant health but remains one of the key determinants of health that has never been addressed in PRAMS in the United States. PRAMS does not collect data on diet potentially due to feasibility issues and the impact on response. Furthermore, little is known about effective approaches by which dietary data can be collected in PRAMS. Thus, we chose to randomize participants to receive a validated semi-quantitative Food Frequency Questionnaire [FFQ] to collect information on maternal diet in order to assess its impact on response rates. Random allocation was achieved using a random number generator in Microsoft Excel. For all 124 women, a random number between 0 and 1 was generated. The 50% of women with the highest numbers were selected to receive the validated FFQ. This FFQ was previously adapted from the European Prospective Investigation of Cancer (EPIC) study , validated in the Irish general population  and used in the Survey of Lifestyle Attitudes and Nutrition (SLAN) 1998, 2002 and 2007  in the Irish general population.
The layout and design features of the questionnaire were also adjusted to incorporate the most recent evidence from a Cochrane Systematic Review on improving response rates to postal questionnaires . This involved changing PRAMS questions to a horizontal rather than vertical orientation which has been shown to increase response rates .
This research protocol and all study materials administered within this study received ethical approval from the Cork Research Ethics Committee of the Cork Teaching Hospitals (CREC).