A Check List for Similar Studies

An automated protocol for assigning early life exposures to longitudinal cohort studies

A Check List for Similar Studies

Based on what we've learned developing the ALGAE protocol, we have identified a number of issues we feel researchers may want to consider when they are setting up a similar kind of early life exposure study.

We hope the list we've provided can help you avoid pitfalls and obstacles that you may encounter when you are gathering, linking or cleaning the various kinds of data needed to support those studies. Please treat them as a kind of informal tick list of issues to consider and judge for yourselves how relevant they are for your own activities.

Defining Life Stages

Define the boundaries of temporally contiguous life stages.

ALGAE does not attach meaning to any one life stage, except the life stage in the early life analysis that starts with a birth date (typically called EL). However, the protocol assumes that study members occupy exactly one life stage for every day of their exposure time frame.

Review calculations for trimesters and conception date.

In the protocol, ALGAE uses the gestation age at birth in weeks to calculate conception date. The gestational age value can be assessed in at least one of two ways: foetal scans or the estimation of a pregnant mother's last menstrual period. Both of these ways of determining the gestation age at birth come with their own inaccuracies and this may in turn influence how researchers determine the conception date.

It is possible to alter the way ALGAE calculates conception date. Please review whether those ALGAE's way of calculating the conception date matches your needs.

Define the overall exposure time frame.

ALGAE relies on a collection of daily exposure values that are meant to accommodate all study members whose life stages overlap with those days. However you do this, it may find it useful to ensure your modelling effort covers what we call the overall exposure time frame. The boundaries of the time frame are calculated as:

  • start date: the earliest conception date of any study member in the cohort
  • end date: the latest last day of first year of life of any study member in the cohort

Once you have defined conception date and the boundaries of your life stages, you can determine this time frame.

If study members have a blank gestation age at birth value, what should the default be?

Be aware that ALGAE assumes it is by default 38 weeks.

Indicating Absences from Existing Residential Address History Data

Are there cohort variables that would indicate that study members spent significant time at addresses that weren't specified in residential address histories?

ALGAE assumes that study members will occupy some location that is specified in the residential address histories. However, study members may have occupied other locations for a period that is significant relative to the duration of the life stages of interest. This may mean their exposures may not always reflect what they would have received at the additional places they occupied.

Absences from residential address histories may be due to a number of circumstances:

  • vacations that could have been outside the study area
  • living with relatives who could live outside the study area
  • hospitalisation
  • homelessness
  • prison

Think about how you could use these variables together to determine a single Yes/No flag for the field absent_during_exp_period found in the table original_study_member_data table.

Are there cohort variables that would indicate whether a study member was at an address at conception that was different than their first recorded address?

ALGAE expects that it can place study members at exactly one residential address for each day of their exposure time frame, beginning at their conception dates. However, in many early life studies, the members may be recruited when their mothers are already pregnant with them. The first address registered for study members may be where their mothers are living when they enrol their children in the cohort. There then could be some uncertainty about whether the study member was at that address during conception. Consider looking for any variables that may indicate whether they had lived at some other addresses. Use them to create the Yes/No flag for the at_1st_addr_conception.

Creating Residential Address Histories

Have you considered whether situations where parents living at different locations have joint custody of child study members? Suppose the parents of a study member are living at different locations and the study member alternates between one location and another. In this scenario, a study member could be move between addresses in a way that isn't reflected in the address histories.

Do the addresses represent where study members live or where they receive mail?

Consider that for some study members, the addresses that are listed may be places where extended family members might pick up mail on behalf of study members.

Do address changes represent when study members actually started and stopped living at a location? Or do they represent when they were recorded in a database system?

Consider that a study member may have not yet or already moved by the time cohort administrators update that person's most current address.

How did cohort administrators receive updates about changes of address?

Consider the following ways that address changes could be conveyed to cohorts:

  • self-report through an online application
  • send a postal return
  • make a phone call
  • make an in-person visit
  • cohort administrators find out about address changes through digests from government databases

Each of these modes of communication may be associated with a different level of accuracy that the address period start and end dates could represent. For example, study members who made and in-person visit or a phone call might emphasise that the change of address is effective of a certain date. A postal return may suggest a lag time between when a letter is written and received. In some cases, cohort administrators may find out a child study member has moved because of updates in various government databases.