Children who experience maltreatment and other forms of adverse events are at an increased risk of early death, acute and chronic health outcomes, and a wide range of factors that impact functioning and well-being into adulthood [1], [2], [3]. Preventing child maltreatment, which includes all forms of child physical and sexual abuse as well as neglect and emotional injury, is a public health priority. To direct and inform prevention efforts, public health research approaches such as leveraging population-level linked administrative data are increasingly used to study and identify factors that predict child maltreatment [4].

Studies linking statewide birth cohorts with child welfare records have emerged as an effective approach to studying maltreatment at the population level [5], [6], [7], [8]. For example, Putnam-Hornstein and colleagues linked all California birth records with child welfare and other administrative records. They estimated statewide maltreatment incidence, identified risk factors, quantified relationships between maltreatment and injuries, assessed prenatal substance exposure and maltreatment, and described the impact of neighborhood affiliation and healthcare utilization [4], [6], [9], [10]. They also examined intergenerational effects of maltreatment and highlighted racial disparities in child welfare involvement [11], [12]. Similar though less extensive statewide linkages have been used to explore the association between birth anomalies and child maltreatment, identify children at risk of maltreatment-related mortality, and identify child maltreatment risk and protective factors [7], [8], [13], [14], [15].

While linkages of administrative records to create birth cohorts is useful, the approach has limitations. Information recorded in administrative birth records does not capture complex factors associated with maltreatment such as parental experiences with violence, housing or economic stability, mental/behavioral health of the caregiver, caregiver beliefs regarding health-related behaviors, health care access, parenting and social supports, or community factors. Integrating population representative survey data can mitigate these limitations.and be used to assess the relation of these factors to child maltreatment at the population-level [16], [17], [18], [19].

The State of Alaska developed and internally validated a novel birth population linkage approach [14], [20], [21] using a mixed-design method similar to that suggested by Bertolli in 1995 [22]. Specifically, the Alaska Longitudinal Child Abuse and Neglect Linkage project (ALCANLink) integrates survey responses from the Pregnancy Risk Assessment Monitoring System (PRAMS) survey with child welfare and other administrative data. PRAMS, a surveillance project supported by the Centers for Disease Control and Prevention (CDC) and implemented by state, territorial, or local health departments, collects standardized and jurisdiction-specific, population-based data on maternal attitudes and experiences before, during, and shortly after pregnancy and is weighted to represent the annual cohort of residents who delivered a live birth [23]. Alaska has used the rich epidemiologic information contained in the ALCANLink data to quantify relationships between events in the pre-birth and newborn periods, such as intimate partner violence and household challenges, and subsequent child maltreatment, Adverse Childhood Experience scores, and early educational outcomes [20], [24], [25], [26], [27], [28], [29].

Using PRAMS, which is weighted to the statewide birth population, as the base cohort may enable cross-jurisdiction analyses using questionnaire elements that are standard between jurisdictions. However, this method has not been replicated or externally validated in another jurisdiction. We had two objectives: 1) Apply the ALCANLink method in Oregon to demonstrate the reproducibility in a jurisdiction with different PRAMS sampling strata and coverage rates, child welfare policies, and state health department organizational structure; and 2) Assess the accuracy of the method in Oregon and Alaska by comparing incidence proportion estimates of alleged, investigated, and substantiated maltreatment before age nine years in PRAMS cohorts with those observed in the corresponding birth cohorts for 2009.