Background and objective

Occupational parental exposures before conception or during pregnancy may be hazardous to fertility and optimal child development and a number of chemical and physical agents, and factors related to work organization (work strain, work schedule, shift work), have been suspected to interfere with normal reproduction process. InEurope, there are many pregnancy and birth cohort that have collected information on maternal occupation but few of them have enough statistical power to examine the effects of specific occupational exposures and adverse birth outcomes in detail, and therefore pooling of data across European cohorts may be particularly valuable.

Assessment of occupational exposures in European birth cohorts has been summarized (ENRIECO WP2 protocol: "Methods and Approaches of Evaluating Occupational Exposures in European Birth Cohorts" - online available www.enrieco.org). Assessment across the cohorts is very diverse and ranges from simple occupational title, to questionnaires on specific exposures, application of job-exposure matrices or biological markers of exposure.

The aim of this case study is to evaluate the risk of adverse birth outcomes (reductions in birth weight and gestational age) for specific "at risk" maternal occupations using combined data from European birth cohorts and evaluate the heterogeneity between countries in such effects.

Methods

Selection criteria for cohorts inclusion

1. Birth cohorts having recorded maternal occupations held at any time during pregnancy (including collected at birth) and having already coded this information into occupational codes

2. Birth cohorts that started the enrolment after 1990

Exposure variables

From participating birth cohorts it is expected that they have collected information on maternal occupation(s) held between the period starting one month before pregnancy until birth; however, only but one occupation during pregnancy will be selected for the pooled analysis. If cohorts have asked occupation history once during pregnancy, they will have to provide the corresponding code of this occupation. In cases that cohorts have asked occupational history more than once during pregnancy, they will be asked to provide only one occupational code and specify the criterion of selection (the longest job, first interview, etc.).

We ask each cohort to indicate whether the mother worked during pregnancy or not. If yes, cohorts will provide the corresponding occupational code and specify if the job was held during each specific period: 1 month before pregnancy, 1st, 2nd or 3rd trimester and the date of end of work during pregnancy (maternity leave).

*Occupational code:

A list of final occupations for analysis (grouped in different occupational sectors) will be defined when data from cohorts has been received, and the prevalence of each occupation is known. The ENRIECO case study already identified a list of occupational sectors of a-priori of interest: 1) Health; 2) Day-care; 3) Cleaning; 4) Agriculture and gardening; 5) Electricity, electronic, optic workers; 6) Lab work and chemical industry; 7) Food industry; 8) Printing and painting; 9) Hairdressers and cosmeticians.

These specific occupations will be defined according to ISCO 88 code. We have chosen this occupational code because it has many national equivalents and appears to be the most frequently reported in the participating cohorts. Cohorts will send the corresponding occupational codes to CREAL (Maribel Casas: mcasas@creal.cat):

- cohorts that have already coded occupational information into ISCO 88: send ISCO 88 codes to CREAL;

- cohorts that have coded the occupational information into other codes that can be translated to ISCO 88 (case of ALSPAC): send the corresponding codes to CREAL and its staff will be responsible to translate these codes to ISCO 88. CREAL staff will also check that the definition of each occupational sector is equivalent to the definition of ISCO 88.

*Physical load

Based on the occupational codes for job titles, a specific job-exposure matrix will be developed for physical load by Lex Burdorf. This JEM will focus on heavy lifting, awkward back postures, and forceful movements. The development of the JEM will follow two different approaches. First, available information on self-reported aspects of physical load within the participating cohorts will be collected and, permitting sufficient comparability of questions, collated into a JEM ("internal JEM"). This procedure will also allow us to evaluate within-cohort variance in self-reported physical load. Second, a JEM currently under construction will be applied. This JEM is based on detailed information from the Health 2000 Study and the Finnish National Work and Health Surveys (FNWHS), which cover over 300 occupations.

Outcomes variables

The study will focus on reductions in birth weight and gestational age as the main outcome variables. Birth weight will be analysed as a continuous outcome expressed in grams and as low birth weight, defined as birth weight below2500 grams. Gestational age will be analysed as a continuous outcome expressed in weeks and as pre-term birth, defined as gestational age of less than 37 weeks.

Confounding variables

A list of potential confounders has been listed in section 8: Scientific outline.

Data analysis

A centralized approach for combined analyses will be followed, where possible. Each cohort will prepare the necessary information and variables dataset and send it to CREAL, where relevant variables will be harmonized. All the mothers included in each cohort, even if they have not been working during pregnancy (and 1 month before pregnancy), will be included in the dataset and send to CREAL.

The population samples will be limited to:

- live born infants

- AND singleton infants (in case of twin births, both twins - and their mother - should be excluded)

- AND with known values for birth weight, gestational age

Missing values do not need to be replaced by the birth cohort data managers. If necessary, a global approach to handle missing values will be developed centrally (CREAL).

Analyses will start with descriptive tables of all variables in each cohort. On the basis of this, a detailed analysis plan will be prepared and agreed by all cohorts before further analytical models will be applied (based on criteria such as the number of subjects in different occupations). This analysis plan will include the selection official occupations and occupational groups for analysis, the choice of reference/control group, the choice of confounding variables, and the analytical approach to be used (pooled and/or meta-analysis). Heterogeneity between cohorts will be explored using appropriate statistical models.