Provider surveys cover a universe of transactions—rather than people—it is typically not possible to reliably infer prevalence of conditions among the human population from these sources because they do not reflect a standard and well defined population.
Put another way, provider surveys are limited to the people who present in medical settings—they cannot cover those who do not receive care.
Provider surveys depend on the cooperation of providers—physicians, hospitals and free-standing medical facilities. They can suffer biased coverage if large provider organizations decide not to participate.
Provider surveys typically involve the coding and classification of procedures and diagnoses in terms of a standard set of categories. these often reflect providers’ practices in classifying cases—particularly if their compensation is influenced by coding practices.