Fathers are often missing from administrative health data. A new approach could help fill the gap.
Administrative health data that link people to their parents can help researchers answer a wide range of questions about health across generations. They can be used to study how health conditions and social circumstances are shared across families, how parents’ experiences and exposures may affect their children, and which factors may help predict health outcomes.
But while children can often be linked to their mothers, links to their fathers are much more likely to be missing.
In a new study published in the International Journal of Population Data Science (IJPDS), researchers used administrative health data from Manitoba, Canada, to develop and test a way to address missing father-child links.
The study included nearly 140,000 people, about one-quarter of whom had no father recorded in the administrative data. People without paternal linkages were more likely to have younger mothers, live in lower-income areas, and be born more recently.
This means that simply excluding people without a recorded father from research could affect who is represented and potentially influence study findings.
The researchers developed a “candidate father” approach that used information already available in the data, including residential postal codes and the ages of potential parents, to identify plausible fathers.
They first tested the method using historical data in which the recorded fathers were known. After artificially removing paternal links, the researchers were able to identify at least one plausible candidate father for nearly 80% of people.
They then examined whether using candidate fathers produced findings similar to those from recorded fathers. They tested the approach using two examples: the relationship between paternal and offspring mental disorders, and between paternal and offspring ischemic heart disease.
Estimates using candidate fathers were generally close to those based on the recorded fathers, particularly when levels of missing paternal linkage were low.
The researchers also compared the candidate father approach with other ways of handling missing paternal information, including excluding people without a recorded father. Each method has strengths and limitations, and performance differed according to the health outcome being studied and how much paternal information was missing.
The findings provide researchers with another option for including fathers in population-based studies when father-child links are incomplete. They also show that how missing paternal information is handled can matter, and that researchers should consider whether their findings are consistent across different methods.
Further research could refine the candidate father approach using additional sources of family information and test how well it works in other populations and data systems.
“Fathers are important to many questions about health across generations, but they can be difficult to include in research using administrative data because father-child links are often incomplete,” says lead author Amani Hamad
“Our study provides another approach to handle this missing information and shows the importance of considering how different approaches may affect study findings.”
Click here to read the full article
Amani F. Hamad, PhD, Canada Research Chair in Population Data Science and Data Curation (Tier 2), Assistant Professor, College of Community and Global Health, Rady Faculty of Health Sciences, University of Manitoba, Canada