Implementation of CG-CAHPS

The Alliances have generally adopted one of two models when implementing standardized measurement using CG-CAHPS across a community: a decentralized (also known as “leveraged”) model or a centralized model.

  • Decentralized (or “leveraged”) model: Under this approach, individual medical practices contract with their existing survey vendors to sample their patients and implement the survey. The sample frame comes from the encounter records at each practice. This approach is also referred to as the “leveraged” model, since it builds on surveying activities already taking place among the participating practices in the community.
  • Centralized model: Under this approach, a party external to the medical practices contracts with a single survey vendor to administer the survey using one sample frame, drawn from either pooled health plan enrollment files or a compilation of medical records from the practices. The costs of this effort are covered by one or more external organization (e.g., a health plan, a state agency) or distributed across the participating practices.

The table below lays out the major advantages and disadvantages of these two models.[1]

Advantages and Disadvantages of the Two Common Implementation Models


Leveraged Model

Centralized Model



  • The sample frame can include all patients.
  • Participating practices can integrate CG-CAHPS core items into their current surveys.
  • Costs can be built into existing survey budgets.
  • The direct participation of medical practices in the survey effort may foster greater use of survey results to improve performance.
  • The external party can ensure consistency in sampling and administration across practices.
  • The model creates the potential for cost savings through economies of scale.
  • The efficiencies associated with a combined effort facilitate participation by smaller practices that cannot afford to survey patients on their own.



  • Significant coordination is needed to ensure consistency in sampling and administering the survey.
  • The costs of surveying may be prohibitive for smaller and independent practices.
  • The sample frame may be restricted to only certain patients, depending on the data source (e.g., only those covered by certain health plans).
  • Disengaging the survey process from the participating practices may limit their use of data for quality improvement.
  • The source(s) of funding for the effort may not be as stable.


There is no “right” model for a given community. The best approach (or mix of approaches) depends on a variety of market-specific factors, including current surveying activities, the influence of health plans and other payers in the market, the structure of the local delivery system (e.g., whether it is dominated by large integrated systems, small independent practices, or a mix of the two), and available expertise within the community (e.g., Alliance staff). Both approaches have been used successfully, as described in the sections below.

[1] Further information about these two models is in Leveraging Existing Patient Survey Efforts: A Decision Guide, available from the Robert Wood Johnson Foundation at