2004S-0170 - Medicare Prescription Drug, Improvement, and Modernization Act of 2003, Section 1013: Suggest Priority Topics for Research
FDA Comment Number : EC40
Submitter : Mr. Dennis Cotter Date & Time: 06/07/2004 07:06:06
Organization : MTPPI
Health Professional
Category :
Issue Areas/Comments
GENERAL
GENERAL
First, I would like to thank CMS for bringing these questions related to 1013 to the public for input and thank the symposium for the opportunity to comment. I certainly believe that there is an important role to be played by CMS in collecting and promulgating information about the effectiveness of pharmaceuticals and that such efforts can help make the health care system more efficient overall as well as helping individual patients receive the best possible care at the lowest possible cost.

Analyzing effectiveness in the real world, as opposed to efficacy in trials, is difficult, however. The need to capture information that reflects the breadth of experience of actual patients requires large populations and makes controlled clinical trials prohibitively expensive. In such populations, there are many factors that can cloud the connection between patient outcomes and the treatments, specifically pharmaceuticals in real word usage and make it difficult to infer causality.

A decision to issue a prescription for a drug for a specific patient is made in a context of that patient's history, state of health, and preferences. Each of these aspects can influence outcomes from treatment in obvious and in subtle ways, but one of the most critical issues is that patients who have more serious problems are more likely to be treated and treated more aggressively.

Failing to properly control this treatment-by-indication or state-of-health bias will lead to understating the effectiveness of any pharmaceutical regimen being analyzed based on real-world observations, or even the possibility of inferring a detrimental effect to a drug that is beneficial.

One possible data source for capturing data on drug use among large populations is administrative or claims data resulting from the process of billing insurance carriers for medical care. Such data sets enable the analysis of patient histories over extended periods by accumulating claims for services, which including diagnoses and procedures. However, such claims rarely include clinical information in needed to assess the need for treatment or the severity of disease.

We have been actively engaged in research using data from one of the rare cases where such clinical data is collected as part of the claims process. - the case of epoetin to treat anemia. In this case, Medicare requires providers to report the hematocrit, or red blood cell level of patients in order to be reimbursed for their anemia treatment.

Since anemia is the primary driver of epoetin dosing, we hope to apply advanced statistical techniques to correct for the tendency of patients with the most severe anemia to receive the most aggressive treatment. Our preliminary findings have been submitted to CMS in response to their request related to epoetin policy.

We feel that this example can be used as a pattern for the possibilities of collecting limited clinical data via the claims process to be used for
measuring effectiveness.

This suggestion may only be usefully applied in cases of treatments for chronic diseases when those treatments are heavily focused on improving measured clinical parameters such as blood pressure, cholesterol levels or glucose tolerance among others.

In addition, using the claims process to collect such information will only be useful when the prescription is itself billed to the insurance provider and is accompanied by an encounter with a provider that is also billed. If the test yielding the physiological data to be analyzed can be required for payment, and the connection between that measurement and the prescription can be reliably made.

Collecting even limited clinical information in these circumstances could greatly enhance the ability of researchers to develop evidence of effectiveness based on administrative data that include large populations.