Author, year, country | Study sample selection bias additional criteria for case– control studies | Bias in measurement of exposure | Bias in measurement of outcomes | Attrition bias | Analysis method reported and appropriate | Potential confounders adjusted for and listed | |
UK studies | |||||||
Bhoopalam et al 23, 1991, UK | Case definition adequate | YES From hospital records | HIGH Distance measured in straight line from home address | LOW Objective outcome (BBA) | LOW 3/134 (2.2%) BBA cases and 10/274 (3.6%) excluded from distance analysis | HIGH Descriptive analysis only | HIGH No adjustments |
Representative-ness of cases | YES BBA cases from 2 hospitals over 5 years | ||||||
Appropriate selection of controls | YES two controls from same hospital | ||||||
Definition of control appropriate | YES The outcome (BBA) could not have occurred | ||||||
Comparability of cases and controls | NO Significant differences in maternal age, parity, ethnicity and antenatal booking | ||||||
Other European Studies | |||||||
Pasquier et al 24, 2007, France | LOW Data from France Central-East malformation registry | LOW GIS software used to estimate distance between maternal residence and nearest maternity ward with neonatal surgical centre | LOW Objective outcomes (NM) | LOW 12 births (0.03%) missing survival data | LOW Method detailed and appropriate, univariate analysis & multiple logistic regression were reported | LOW Adjusted for parity, ethnicity and other characteristics | |
Blondel et al 5, 2011, France* | LOW Birth certificates | LOW GIS software used to estimate distance to hospital from home | LOW Objective outcome (BBA) | LOW 11% of births excluded | LOW Method detailed and appropriate, multilevel model was reported | LOW Adjusted for maternal age, occupation, parity and other characteristics | |
Pilkington et al 25, 2014, France | LOW French National Vital Statistics Registry | HIGH Distance calculated from centre of municipality not home address, using road networks provided by the French National Geography Institute | LOW Objective outcomes (SB, NM) | LOW 10% missing for type of pregnancy and 17% for maternal age | LOW Method detailed and appropriate, logistic regression analysis was reported | LOW Adjusted for maternal age, plurality, unemployment rate, single parent households | |
Fougner et al 26, 2000, Finland | Case definition adequate | NO Data from a questionnaire | HIGH Data for distance was self-reported in questionnaire | LOW Objective outcome (BBA) | HIGH Data from women who responded to questionnaire | HIGH Descriptive analysis only | HIGH No adjustment |
Representative-ness of cases | YES Cases from three hospitals in one county over 8 years | ||||||
Appropriate selection of controls | Yes Women who delivered their babies within 1 hour of arriving at hospital | ||||||
Definition of control appropriate | Yes Women with no BBA | ||||||
Comparability of cases and controls | Unclear | ||||||
Ovaskainen et al 27, 2015, Finland | Case definition adequate | YES Medical records | LOW Distance from women’s home calculated using web-based route planner | LOW Objective outcome (BBA) | High 13 out-of-hospital deliveries (19%) excluded as could not ascertain whether planned or not planned | LOW Method detailed and appropriate, logistic regression data were given | LOW Adjusted for single-mother, parity, and other characteristics |
Representative-ness of cases | YES Cases from one centre, but over 15 years | ||||||
Appropriate selection of controls | YES Births occurring immediately preceding and following case | ||||||
Definition of control appropriate | YES The outcome (BBA) could not have occurred | ||||||
Comparability of cases and controls | No Sig. differences for parity, partnership status, smoking, antenatal visits, labour duration and distance to delivery unit | ||||||
Canadian studies | |||||||
Lisonkova et al 28, 2011, Canada | LOW Population-based study | HIGH GIS used to calculate distance from postcode central point of residence to hospital; mostly using straight line distance | LOW Objective outcomes (SB; PM); NNU admission) | LOW 492 (1.7%) women excluded due to missing postcodes | LOW Method detailed and appropriate, multivariate regression analysis reported | LOW Adjusted for parity, single mother, low income, ethnicity and other characteristics |
*Blondel et al 5 2011 included in travel distance & OU closure.
BBA, born before arrival; GIS, Geographical Information System; NM, neonatal mortality; NNU, neonatal unit; PM, perinatal mortality; SB, still birth; sigs, significant.