Aircraft (1906–1945) by Reijer Stolk

Dropot: Maltreatment research's arch enemy

Longitudinal cohorts are among the most powerful tools we have for understanding what happens to children exposed to maltreatment. Following children over time allows researchers to move beyond simple questions such as “Are maltreated children more likely to develop mental health problems?” and investigate much more clinically meaningful questions: Which children are most vulnerable? Which children recover? What factors protect children from adverse outcomes?

But there is a major challenge hidden within almost every longitudinal study of maltreatment: the children who are hardest to follow are often the children we most need to understand.

The scientific term is attrition: participants dropping out of a study over time. In any long-term cohort, some families move away, become difficult to contact, lose interest, or simply decide that participation is no longer possible. This is normal. However, in maltreatment research, dropout is not random.

One of the strongest predictors of whether someone remains in a longitudinal study is often the very thing researchers are trying to understand: exposure to maltreatment itself.

This creates a fundamental problem. Imagine a cohort that starts with 1,000 children. At the beginning of the study, researchers identify children exposed to neglect, abuse, and other forms of adversity. Ten years later, they invite everyone back to assess mental health, education, relationships, and functioning.

The children who return may not be representative of the children who started.

Those who remain may be the families with greater stability, more resources, better access to care, and stronger relationships with institutions. Meanwhile, children who experienced the most severe adversity may be more likely to disappear from the research process.

This creates a paradox: the group we most want to study may become the group we know the least about.

Why does maltreatment increase dropout?

There are several possible reasons. Families affected by maltreatment often face multiple challenges at the same time. Housing instability, financial stress, parental mental health problems, substance use, involvement with child protection services, and disrupted relationships can all make participation in research more difficult.

For some families, contact with researchers may also evoke difficult emotions. Discussing traumatic experiences, answering detailed questionnaires about family life, or repeatedly revisiting painful memories can feel burdensome. Trust in institutions may also be affected by previous experiences with healthcare, schools, or social services.

For adolescents who experienced maltreatment, practical factors become increasingly important. They may move frequently, leave unstable home environments, experience mental health problems, or become difficult to reach during transitions into adulthood.

These processes are not simply inconveniences for researchers. They can change the conclusions of studies.

Suppose maltreated children who are doing relatively well are more likely to remain in a cohort, while those experiencing the greatest difficulties are more likely to leave. Researchers may conclude that the long-term effects of maltreatment are smaller than they really are.

Alternatively, if only the most severely affected participants remain engaged because they receive intensive support or services, studies may overestimate the average impact of maltreatment.

The key issue is that dropout can introduce bias into our understanding of what maltreatment does over the life course.

Traditional statistical methods can help, but they cannot fully solve the problem. Researchers can use techniques such as weighting participants according to their probability of remaining in the study, multiple imputation of missing data, or sensitivity analyses exploring different dropout scenarios. These approaches are valuable, but they rely on assumptions about why people leave and what their outcomes would have been if they had stayed.

The solution therefore cannot only be statistical. We need innovation in how we design and maintain longitudinal studies.

This may mean building stronger relationships with families from the beginning, involving communities in research design, reducing the burden of participation, and offering flexible ways to contribute. Digital assessments, remote participation, linkage with routinely collected health and social data, and creative approaches to maintaining contact may all help.

Most importantly, researchers need to recognise that retention is not merely a logistical challenge. It is part of the science.

A cohort that successfully follows children exposed to maltreatment over many years provides invaluable insight into resilience, vulnerability, and opportunities for prevention. But achieving this requires understanding that dropout is not a random event. It is often a reflection of the very social and psychological difficulties researchers aim to study.

The future of maltreatment research depends not only on collecting more data, but on finding better ways to ensure that the voices of the most vulnerable children are not lost along the way.