COXIBs in the real world

Clinical bottom line

Cyclo-oxygenase-2 inhibitors (COXIBs) are as effective in real worl settings as they were predicted to be from randomised trials and meta-analyses of randomised trials. They are safer than NSAIDs and cause many fewer gastrointestinal bleeds.

Reference

M Mamdani et al. Observational study of upper gastrointestinal haemorrhage in elderly patients given selective cyclo-oxygenase-2 inhibitors or conventional non-steroidal anti-inflammatory drugs. BMJ 2002 325: 624-629.

Study

This was a population based retrospective study using databases with 1.3 million people aged 66 years or older in Ontario. It covered a period of year from April 2000, the date on which two COXIBs, celecoxib and rofecoxib became available for prescription (for osteoarthritis, and rheumatoid arthritis for celecoxib). There is free access to prescriptions to all elderly people in Ontario.

People taking rofecoxib, celecoxib, non-selective NSAID and a random sample of 100,000 people were chosen (a general elderly population, not matched by age, or sex). The outcome chosen was admission to hospital for upper gastrointestinal haemorrhage from comprehensive linked databases.

Users were defined as those who were given at least two successive prescriptions and who received enough drug for at least 30 days of observation.

Results

Out of 1.3 million older people, 365,000 (28%) had a prescription for an NSAID during the period. The mean age was about 75 years, and they had six prescription drugs in the previous year. The numbers of patients in each group, and some of their characteristics, are shown in Table 1:

Table 1: Included patients (characteristics reported as percent of total for that treatment)

Characteristic	Controls	NSAID	Diclofenac + misoprostol	Rofecoxib	Celecoxib
Number	100,000	5,391	5,087	14,583	18,908
Hospital admission in last year	12	19	18	20	19
Use of GPA within 180 days	17	25	25	42	41
Use of opioid analgesics within 180 days	11	26	26	31	30
Prior GI procedure	18	20	21	32	31
Prior upper GI haemorrhage	1	1	1	3	3
Aspirin use	12	19	19	18	18
Anticoagulants	7	5	5	10	10
PPI	6	8	8	22	20

The groups were not the same.

Compared with controls, users of NSAIDs (including diclofenac plus misoprostol) had more hospital admissions, a history of use of gastroprotective agents (GPA), use of opioid analgesics and used more aspirin.
Compared with controls, users of COXIBs had more hospital admissions, had a greater history of use of GPA, used more opioid analgesics, had a greater history of prior GI procedures, had a greater history of GI haemorrhage, used more aspirin, anticoagulants and proton pump inhibitors.
Compared with NSAID users, users of COXIBs had a greater history of prior GPA use, prior GI procedures and use of aspirin and proton pump inhibitors.

Despite these differences, pointing to a higher risk with COXIBs than a general elderly population, the rate at which hospital admission for upper gastrointestinal haemorrhage occurred was barely different between users of COXIBs and controls (Figure 1). The rate for NSAID users, and users of diclofenac plus misoprostol, was about four times greater than control.

Figure 1: Relative rate ratios of upper gastrointestinal haemorrhage

Comment

This is a detailed and interesting paper, telling us that what the randomised trials and meta-analyses of randomised trials promised, namely a dramatic reduction if GI bleeds with COXIBs compared with NSAIDs, actually occurs in practice. This is despite the fact that we have a phenomenon here known as "channelling", whereby those patients who are sicker are more likely to get the more effective medicine. Here the patients given COXIBs were at greatest risk of bleeding, followed by NSAID users, followed by controls. So this might be regarded as a tougher test than a randomised trial, in which randomisation ensures that risks will be equally distributed between treatment groups.

What we have to remember is that the actual number of bleeds was small (Table 2), and we therefore need to avoid making too much of these data. What is comforting is that the four-fold increased risk with NSAID is what is found in many other studies, and that the background risk of bleeds in untreated patients is again about the same as is seen in other studies. So there are good reasons for putting some credence in the results.

Table 2: Bleeds and rates

Characteristic	Controls	NSAID	Diclofenac + misoprostol	Rofecoxib	Celecoxib
Number	100,000	5,391	5,087	14,583	18,908
Number of bleeds	82	17	13	43	32
Number per 1000 patient years	2.2	13	10	7.3	3.6
NNH		403	592

The paper also gives us numbers needed to harm for a follow up of 295 days, which for NSAIDs is 403. Since a reasonable estimate is that the death rate with upper gastrointestinal haemorrhage is about 1 in 10, this would generate a number needed to kill of about 4,000 for NSAIDs, again within reasonable agreement with other estimates.

Trying to draw conclusions from a comparison of the COXIBs would not be sensible. One of the main reasons, apart from the small numbers of events, is that of dose. The paper describes 200 mg celecoxib and 25 mg as the upper end of the prescribing range. This may have been the case in Canada at that time, but we may not be comparing like with like.

The final comment is about diclofenac plus misoprostol. Though doses of either or both were not given, there was not much evidence here for any special benefit.