The P&T formulary committees in large managed care facilities have struggled to balance the theoretical improved safety of the COX-l-sparing medications against their significantly increased acquisition costs. Although attempts have been made to determine their overall cost-effectiveness on a global basis, no one seems to have examined this problem in a way that practitioners can use in a clinical setting.

In the interest of providing a solution to this complex situation, I have attempted to develop a logical and applicable algorithm for the use of these COX-l-sparing agents in a large managed care system—the federal health system—in which the need for effective cost containment resulting from an ever-shrinking formulary budgets is most urgent (Table 1).

Table 1    Recommendations for the Use of Nonselective NSAIDs or COX-1-Sparing Drugs

Risk Level First-Line Agent Alternative
1 Low risk(gi score 0-10) Formulary nonselective NSAID If the nonselective NSAID is not tolerated or if it is ineffective, patients may use a second nonselective NSAID. If this is not tolerated or is ineffective, patients may use meloxicam.
2 Moderate risk (gi score 11-15) Formulary nonselective NSAID If there is no response to the nonselective NSAID or if intolerance devel­ops, patients may use another COX-1-sparing agent, such as meloxicam, celecoxib, rofecoxib (Vioxx®),or valdecoxib (Bextra®).
3 High risk(gi score 16-20)   If there is no response to meloxicam or if intolerance develops, patients may use a different COX-1-sparing agent, such as celecoxib, rofecoxib, or valdecoxib.
4 Substantial high risk (gi score >20) COX-1-sparing agent (e.g., meloxicam, celecoxib, rofecoxib or valdecoxib) If there is no response or if intolerance develops, patients may use a different COX-1-sparing agent, such as meloxicam, canadian celecoxib, rofecoxib, or valdecoxib.

For an algorithm to be effective, it not only needs sound scientific and pharmacoeconomic backing but also must be easy to use. By studying the best data available on reducing the risk of GI events and on the current costs of the medications on the federal formulary, one can establish a model for use throughout a diverse network of clinics and hospitals within the federal system and as a potential model in other large managed care networks.

The problems with effective use of the COX-l-sparing medications are more pronounced in closely managed, large formularies (e.g., in the federal health system) than in the private sector. Although these agents do have the potential to decrease mortality and costly hospitalizations, their indis-criminant use would lead to tens of millions of dollars of unnecessary formulary costs in an atmosphere of steadily increasing budgetary constraints.


Generic NSAIDs cost pennies per day, and for younger individuals at low risk, the additional expense of COX-l-sparing agents might not be justified. In an ideal world, cost would not be a concern and COX-l-sparing drugs would be used for every patient who needed them; however, with the current emphasis on cost containment, physicians can use the gi score to stratify a patient’s risk of significant GI events.
canadian cialis online

Table 2 Gastrointestinal (GI) Risk Assessment Tool (GI SCORE)

This scoring tool, developed by Dr. G.Singh and colleagues at Stanford Univer-
sity, is based upon data from 566 hospitalizations for serious GI injury from
6,386 patients with rheumatoid arthritis or osteoarthritis who were monitored
prospectively.The authors used COX proportional hazard models to determine
risk factors.The GI SCORE is calculated from individual patient responses to six
questions. Each question is assigned a certain number of points. After the
points have been added up, a GI risk score from 1 to 4 is assigned (1 = lowest
risk, 4 = highest risk).
1. How old are you?
Age        Points                     Age        Points
<20 years       0                      56-60          10
21-25          1                      61-65         12
26-30         3                     66-70         13
31-35         4                     71-75         14
36-40         5                     76-80         16
41-45         6                     81-85         17
46-50         8                      85+18
51-55         9 Points:
2. How do you rate your current health status on the
following scale?
Very Poor (4 points)      Poor (3 points)      Fair (2 points)
Well (1 point)              Very Well (0 points)
3. Has a physician ever told you that you have rheumatoid
arthritis (not osteoarthritis or other forms of arthritis)?
No (0 points)                                Yes (2 points)
4. If you are taking prednisone or another corticosteroid,
for how many months have you taken it in the past year?
0 (0 points)            1-3 (1 point)              4-6 (3 points)
7-10 (4 points)      11-12 (5 points)
5. Have you ever been hospitalized for a stomach or
intestinal problem, such as bleeding or an ulcer?
(if the answer is yes, skip the next question).
No (0 points)                                Yes (2 points)
6. If no, have you ever had gastrointestinal side effects
(heartburn, stomach pain, nausea, vomiting) when taking
NSAID pain relievers?
No (0 points)                                Yes (2 points)
Total Points:

Developed by Singh and colleagues at Stanford University, the GI scoring tool includes six simple questions to calculate a person’s annual risk of GI bleeding as a result of NSAID therapy (Table 2). This instrument has been validated in more than 40,000 patients. The model clearly predicts the risk of GI bleeding in patients at all levels of risk, and the clinician can determine the risk of a significant GI event in less than 30 seconds with information that is easily obtained from a patient’s chart. An intake nurse can calculate the gi score. The instrument is already being used in Veterans Affairs (VA) medical centers, in the federal health system, and in many large managed care networks.

The next step in developing a simple algorithm for the use of COX-l-sparing agents is to calculate the costs of the drug against the costs of the ADEs. From the current data, GI bleeding episodes are not completely eliminated by the use of COX-l-sparing medications Several factors account for this finding, including:

  • the inherent COX-l activity of even the most selective agents.
  • other causes of GI bleeding, such as Helicobacter pylori-induced peptic ulcer disease.
  • the concomitant use of aspirin.

The gi score can be used to weigh the cost of medication against the probable risk reduction in any given population and at a given risk level.

According to the best clinical data, a reduction in significant GI events of approximately 50% can be expected with COX-l-selective inhibitors in place of standard NSAIDs. This figure is supported by the VIoxx Gastrointestinal Outcomes Research (VIGOR) trial, which examined the risk of GI events with rofecoxib (Vioxx®, Merck), and by several studies, such as the SUCCESS tria and a meta-analysis of drug meloxicam (Mobic canadian, Boehringer Ingelheim); all of these trials indicated a risk reduction of at least 50%.

If we were then to estimate the anticipated risk reduction with any given score, we could predict the number of GI bleeding episodes that would be prevented. By comparing the cost of medication per l00 patients, based on actual usage, versus the cost of a GI event, we would be able to determine, according to a particular gi score, whether the use of a COX-l-sparing agent would be cost-negative, cost-neutral, or cost-saving. If we determined, at a given score, that a drug was either cost-neutral or cost-saving, we would implement the use of that medication.

The cost of each medication was determined directly from the federal supply pricing schedule. The practice of “pill-splitting” was considered if it was deemed to be cost-saving; today this is a common procedure in federal, military, and VA hospitals. Clearly, if changes in the pricing schedule occurred, it is possible that the algorithm might change.