Randomized Controlled Trial of Pharmacare's Nebulizer to Inhaler Conversion Policy

by Bruce Carleton, Malcolm Maclure | May 01, 2001

Key Implications for Decision Makers

  • The primary audience for this report is drug benefits program managers. The secondary audience is drug policy evaluators who work with those managers. A third audience is prescribing physicians whose collaboration may be key to future policy trials.

  • Reference drug pricing programs are like drug formularies. They are based on the idea that if the best-available scientific evidence suggests that individual drugs within a given therapeutic class work equally well for a certain condition, the program will pay for the drug(s) that costs the least.

  • British Columbia Pharmacare's reference drug pricing policy, when applied to nebulized medications, resulted in a policy of not paying for nebulized drugs unless physicians gave a clinical justification in advance. This encouraged patients to use inhalers, which are less expensive and equally or more effective in most patients. The policy saved money.

  • The policy did not cause cost shifting (saving on drugs but increasing the costs in other sectors of the health system such as medical services.)

  • The cost of rigorous evaluation can be small. Evaluation can be efficiently conducted, without disrupting policy rollout.

  • Credibility of a policy is increased when the policy is evaluated with a randomized control group, the gold standard for drug evaluations. Drug benefit program managers who wish to understand the intended and unexpected impacts of new money saving policies can use our method called "randomized optional temporary exemption (ROTE)". A group of doctors was given the option of a temporary exemption for their patients from the policy, thereby creating a random group of "control" patients to compare to the patients affected by the policy.

  • It is faster, cheaper, and easier to evaluate a policy by using randomized optional temporary exemption than it is to organize rigorous external retrospective evaluations supported by health research grants.

Executive Summary

Purpose

The gold standard for evaluating drug efficacy is the randomized controlled trial. Evidence shows nonrandomized studies tend to produce biased results. Therefore, a drug policy evaluation also should involve a randomized control group. But is that feasible?

The purpose of this project was to test the feasibility of doing a better impact evaluation than the usual uncontrolled before-after comparisons, which are so often confounded by background trends. This was done in 1999, when British Columbia's Pharmacare program introduced a change in coverage for drugs that was delayed for six months in a randomized control group. In the policy studied here - the Nebulizer-to-Inhaler Conversion Program - BC Pharmacare paid for the full price of inhalers and did not pay for wet home nebulizers unless a physician gave an exemption based on special clinical needs of the patient.

Target Audience

The primary audience for this report is drug benefits program managers. The secondary audience is drug policy evaluators who work with those managers. A third audience is prescribing physicians whose collaboration may be key to future policy trials.

Policy Implications

In terms of the policy topic studied, there is an implication that reference drug pricing can save money. Reference drug pricing programs are like drug formularies. They are based on the idea that if the best-available scientific evidence suggests that individual drugs within a given therapeutic class work equally well for a certain condition, the program will pay for the drug(s) that costs the least.

With respect to how policies should be tested, policies can be more evidence-based if they are tested with the method used in this study. We call it "randomized optional temporary exemption (ROTE)". In this case, a group of doctors were given the option for their patients to be temporarily exempt from the policy. This created a random group of control patients not affected by the policy, with which to compare patients affected by the policy.

The simplicity of the study design means the data analysis can be done by people with relatively little research training. Because the control group is created before the policy is implemented, the impact analysis can be done as soon as the data are available - much sooner than retrospective analyses that assemble comparison populations after the policy. With a randomized control group, the evaluation is clearer, less controversial (less dependent on the judgement of the analysts), and more respected by critics.

Main Findings

Pharmacare's Nebulizer-to-Inhaler Conversion policy, which evolved from a reference drug-pricing policy, saved money (although less than expected) without causing cost shifting. This is important because the pharmaceutical industry claimed that Pharmacare's Reference Drug Program did not save money and caused cost shifting. Estimated savings from the policy were $600,593 during the six-month period of the policy and $1,063,975 saved over the full year after the policy began.

Medical services for respiratory illnesses did not increase significantly as a result of the policy. Hospitalization rates were the same in both groups. Unexpectedly, use of metered dose inhalers did not increase as use of nebulized medications decreased.

Without a simultaneous matched control group, evaluation of the policy would not have been possible due to confounding influences. Delayed implementation of the policy in the control group was smooth (only one physician out of 600 withdrew himself and his patients from the control group). Before-after comparisons would have been inconclusive because of seasonal trends in asthma and British Columbian physicians' "Reduced Activity Days" immediately after the policy began. The randomized control group enabled control for background trends.

Methodology/Data Collection

Physician addresses were geographically paired and randomized to heads or tails. The chair of the external panel of expert advisors tossed the coin to determine which group was sent a letter offering a 6-month exemption from (delay of) the policy. Exempt physicians' ID numbers were coded into the central computer serving the province-wide pharmacy network, PharmaNet. Outcomes were measured using routinely collected Ministry of Health data on drug use, medical services, hospitalizations and mortality. Quality of life questionnaires were sent to patients at baseline and at six months.