spections were conducted. Although the rate of reported cases has been increasing each year, the sensitivity of this method is extremely low because most cases of elder mistreatment are not reported to any social service authorities (Pillemer and Finkelhor  found only 7 percent of cases reported to authorities), and those incidents that are reported must be judged as valid to be considered substantiated. Again, the criteria by which a report is considered founded vary widely by center, as do the definitions of abuse. Ultimately, it is the judgment of individual caseworkers that determines whether or not a mistreatment event has occurred.
A notable strength of agency record review studies such as that conducted by the National Center on Elder Abuse is the highly detailed nature of the data regarding the abuse event. Specifically, the context of elder mistreatment, the perpetrator characteristics, demographic variables, and social structures are usually specified and documented somewhere in agency records. Moreover, there is a relatively strong level of confidence that indicated cases did, in fact, occur. Relative to epidemiological surveys that are conducted solely for data collection and analysis (as opposed to service delivery), information from agency records exists independent of research protocols and is therefore relatively inexpensive to transfer to the research realm.
By contrast, several significant weaknesses characterize agency record-based investigations. This method requires collecting data from a wide variety of agencies that may use dissimilar definitions of mistreatment. Even more problematic is that individual agencies vary widely in the resources directed to investigation of cases, training of caseworkers, and follow-up and substantiation of cases. Thus, even when standard definitions and criteria are used, the means by which agencies determine whether an event meets these criteria will differ. As such, sensitivity and reliability of findings will suffer. The utility of this approach for epidemiological researchers is further affected by the quality of agency record maintenance, accessibility to records, accessibility of the agency personnel, and overall quality of record keeping by an agency.
Overall, the agency record review methodology is indicated when the population of older adults suffers from cognitive impairment and cannot otherwise be interviewed. However, this method is less sensitive than in other methods applicable to cognitively impaired populations and should probably be used only to guide initial efforts insofar as gross approximations of elder mistreatment are needed.
The National Elder Abuse Incidence Study (NEAIS) sponsored by the Administration for Children and Families and the Administration on Aging