Access to off-label cancer drugs – or the use of drugs for cancer indications that are not approved by the FDA – is considered vital to the treatment of many cancer patients. However, recent literature on off-label use of oncology drugs has mostly focused on describing utilization patterns but pays very little attention to federal and state policies that may influence off-label use; furthermore, it has been noted that patients with ready access to off-label drugs through their insurance may have fewer incentives to enroll in the clinical trials that provide valuable evidence about drugs in new indications.

We hypothesized that health insurance mandates to provide coverage of off-label uses of cancer drugs cause higher utilization of those drugs for off-label indications. Higher off-label use of a particular drug in a given cancer site, in turn, slows enrollment and reduces the initiation and timely completion of clinical trials for those drugs in that cancer site.

The Specific Aims were:

1.     To estimate the effects on off-label drug utilization of two types of changes in health insurance coverage policies:

a)     state health insurance mandates that force private health insurance plans to cover off-label oncology drug use, and

b)    a 2008 rule change mandating that Medicare cover a larger number of off-label uses of cancer drugs than was previously the case.

2.     To analyze the effects of off-label use of particular drugs in selected cancer sites on clinical trial enrollment and on the initiation and timely completion of clinical trials for those drugs in those cancer sites.

3.     To conduct exploratory comparative effectiveness research on the effects of off-label drug use of particular drugs in selected cancer indications on patient survival and morbidity.

At this interim stage of the project, we have made substantial progress towards Aims 1 and 2.

Progress on Aim 1a:

We obtained from the Blue Cross Blue Shield Association a list of the existence and dates of enactment of legislation mandating that fully insured private health insurance plans provide coverage for off-label usage of cancer drugs. We then assembled a list of all approved infused anti-cancer medications, their dates of initial FDA approval and their initial and supplementary approval indications and approval dates. We also obtained the recommended on-label and off-label uses of all of these drugs in the two main compendia cited by state laws on off-label coverage: the American Medical Association Drug Evaluations, and the United States Pharmacopeia Drug Information for the Health Professional, for the period 1996-2007. Combined, these datasets allowed us to classify on-label, recommended off-label and non-recommended off-label uses of all approved anticancer drugs in each year from 1996-2007.

We obtained Truven Health Analytics Marketscan data on utilization of cancer drugs for a broad set of private health insurance plans nationwide from 1996-2007, covering the period when most of the changes in state laws took place. We selected 27 infused anti-cancer drugs approved prior to or during this time that both have high utilization in the Marketscan data and that according to the compendia have significant off-label use of drugs. To avoid complications in data analysis and interpretation of off-label use, we constructed a cohort of patients with continuous enrollment in their health plans for at least six months prior to first inpatient or outpatient services claim with a diagnosis of cancer, so that a consistent set of diagnosis and treatment information is available for them. Lastly, we merged in the data on state level policy changes and performed some initial analysis on variation in off-label utilization across states.

What remains to be done for Aim # 1a is to refine our measures of off-label utilization into compendia-recommended and non-recommended utilization, convert the diagnosis information into a summary comorbidity index and use this along with patient age and sex to construct case-mix standardized population rates of off-label utilization by state by year, and estimate difference-in-difference regressions models of the effect of state policy changes on off-label anti-cancer drug utilization.

Progress on Aim 1b:

We have taken the initial steps of purchasing the Clinical Pharmacology drug compendium for 2006-2010.

Progress on Aim 2:

The work performed for Aim #1a is setting the stage for the analysis of off-label use. We have also made progress in constructing a dataset of cancer clinical trials and their characteristics from 1970 to August 2012 by scraping the Physician Data Query database off XML pages hosted by a National Cancer Institute website and into a research-analytic dataset and cleaning this data. We also obtained data on all clinical trials registered by September 2012 on Clinicaltrials.gov from the Analysis of ClincalTrials.gov (AACT), already in research-analytic format.  These two datasets will be combined to obtain clinical trial initiation and completion dates and clinical trial enrolment data for subsets of cancer clinical trials to be used in our analysis.