AHA Data Collection Methods 

AHA Annual Survey of Hospitals

The AHA conducts a voluntary census (survey) of hospitals in the United States and Associated Areas. Hospitals report information on their organizational structure, service lines, utilization, finances, insurance and payment models, and staffing for a specific time-period. AHA Annual Survey data are a primary reference for government agencies (for example, Centers for Disease Control and Prevention, U.S. Census Bureau, Centers for Medicare and Medicaid Services) and journals and sites (for example, Health Services Research, JAMA, U.S. News and World Report). Historically, the AHA Annual Survey has achieved a response rate at or above 75%. We use a rigorous evaluation method to validate survey responses. All hospitals are encouraged to participate in the Survey. Participating hospitals are not required to be members of the American Hospital Association.

AHA Annual Survey Information Technology Supplement

The AHA partners with the Office of the National Coordinator (ONC) of the United States Department of Health and Human Services (HHS) to survey hospitals on the implementation of the Promoting Interoperability initiative. Data are collected on computerized system capabilities, patient engagement, health information exchange, interoperability, and EHR systems and IT vendors. The data are cited in ONC Data Briefs, Health Affairs, and The New England Journal of Medicine as well as other journals. Historically, the AHA Annual IT Supplement Survey has achieved a response rate near 60%. Data responses are evaluated for validity. All hospitals are encouraged to participate in the Survey. Participating hospitals are not required to be members of the American Hospital Association. .   

AHA Estimation Process 

Each year, AHA’s Data Center administers the AHA’s Annual Survey. The purpose of the survey is to collect utilization, financial, and personnel information on each of the nations’ hospitals. The survey’s overall response rate exceeds 75% each year, which is considerably high for a voluntary survey given its length and the size of the universe (over 6,000 hospitals). For hospitals that do not respond at all, or not respond fully to the survey, a statistical methodology is run against their records to estimate missing values.   

The following describes the two major approaches used for estimation:   

  • For nine key variables ‐ total admissions; total births; total inpatient days; total expenses; total full‐time employees; total surgical operations; total outpatient visits; total part‐time employees; and total revenue – estimates are generated from regression models.
  • For each variable, the previous year’s data (base year) along with estimation status, percentage change in state median, MSA size and bed size are used as the independent variables, while the current year’s value is the dependent value. The regression model generates a coefficient for each independent variable, which later will be used in estimating the current year’s value. In other words, the current year’s missing value is “predicted” by multiplying the base year data with the corresponding coefficients derived from the regression model.

Alternatively, components of the key variables and all other variables eligible for estimation are generated from a matrix of estimators:   

  • An estimator is a ratio of two variables: numerators are the variable to estimate; denominators are an indicator variable such as beds, bassinets or a total variable in which the numerator is an additive component. In the first step, the universe of hospitals is stratified into 52 stratum or cells. The stratification is based on the intersection of the following variables, control, service, bed size, and length of stay.
  • Estimators are calculated using all reported data in each stratum. Each estimator takes on a different value for each stratum (thus the matrix). Using payroll expense as an example, matrix estimation occurs as follows. For hospitals in the same stratum, payroll expense as a percentage of total expense is computed based on reported data only. For a non‐responding hospital in this same stratum, the estimated payroll expense value is derived by applying the above mentioned percentage to the non‐respondent’s total expense value regardless of whether that value is reported or estimated.

Please note that missing facilities and services collected through Section C of the survey are never estimated when missing. Beds, control, service, length of stay also are not estimated but are present in all cases. Values for these variables come from the AHA master facility inventory system. For an institution to be listed in the inventory file there must be a value for those variables. The compilation and management of the inventory is independent of the Annual Survey process