Skip navigation

Risk stratification with NSAID and COXIB

 

Clinical bottom line

In this subgroup analysis of a large randomised trial, rofecoxib produced lower rates of upper gastrointestinal events at all risk levels, including patients with no common risk factors. The overall NNT was 41 to prevent an event at one year (compared, say, with NNTs for statins in secondary prevention of 10-30 over five years).

For many groups with higher levels of risk, NNts were 3-15 at one year. Even without common risk factors the NNT was 59 at one year. With no common risk factors, rofecoxib reduced the risk of an upper gastrointestinal event to that found in a general elderly population.


Reference

L Laine et al. Stratifying the risk of NSAID-related upper gastrointestinal clinical events: results of a double-blind outcomes study in patients with rheumatoid arthritis. Gastroenterology 2002 123: 1006-1012

Study

This is an analysis of the VIGOR study, in which rheumatoid arthritis patients of 50 years or older (or 40 years or older but receiving corticosteroid therapy) were randomised between rofecoxib 50 mg daily or naproxen 1000 mg daily, using matching placebos. The duration was for one year, and were seen at intervals over that period. Excluded were patients using aspirin or anticoagulants, proton pump inhibitors or full-strength histamine antagonists.

The primary endpoint was clinical upper GI events (perforation, obstruction, and major bleeding, and symptomatic ulcers identified on a clinically indicated workup). All these events were judged by a blinded, independent, endpoint committee working to prespecified criteria.

Results were analysed to investigate patient criteria that were linked to risk of clinical upper GI event.

Results

There were 8,076 patients followed for a median of nine months.There were 177 confirmed upper GI events, occurring at 2.1 per 100 patient years in the rofecoxib group and 4.5 per 100 patient years with naproxen. This gave a number needed to treat (NNT) of 41 for rofecoxib treatment to prevent one upper GI event occurring with rofecoxib that would have occurred with naproxen. All NNTs are given at one year.

Risk factors

Table 1 below gives the rates per 100 patient years for a range of risk factors, together with the one year NNT. Some of the most important are picked out specifically. In general those people with more risk from their age, medical history of advanced disease derive most benefit from rofecoxib, as seen by lower (better) NNTs.

Table 1: Risk factors for upper gastrointestinal clinical event

Rate per 100 patient years with:

Baseline characteristic

Rofecoxib

Naproxen

NNT at one year

Age <65

1.6

3.2

66

Age 65-74

3.3

7.4

25

Age>75

4.5

14.5

10

No prior GI event

1.7

3.7

51

Prior uncomplicated event

5.2

13.5

12

Prior complicated event

10

19

11

ACR class 1-3

2.0

4.3

43

ACR class 4

4.3

14.3

10

RA less than 25 years

2.0

4.1

48

RA more than 25 years

3.1

9.3

17

No steroid use

2.0

3.0

106

Steroid use

2.1

5.7

28

Baseline NSAID

1.9

3.9

49

No baseline NSAID

3.1

7.6

22

Baseline H2A

1.7

4.1

42

No baseline H2A

5.8

8.8

33

No history of GI symptoms

1.8

3.7

53

History of GI symptoms

3.6

8.8

19

Age

Older age is a known risk factor with NSAIDs. Figure 1 shows that the risk is much reduced with rofecoxib

Figure 1: Risk and NNTs with age


Prior gastrointestinal events

Prior history of gastrointestinal bleeding is a known risk factor with NSAIDs. Figure 2 shows that the risk is much reduced with rofecoxib.

Figure 2: Risk and prior gastrointestinal events


Prior symptoms

Prior history of gastrointestinal symptoms is a known risk factor with NSAIDs. Figure 3 shows that the risk is much reduced with rofecoxib.

Figure 3: Risk and prior gastrointestinal symptoms


Combinations of risk factors

For some combinations of baseline risk factors the risk with NSAID was very high, and the reduction in the rate with rofecoxib produced some very low and excellent NNts of between 3 and 11 (Table 2).

Table 2: Combinations of risk factors

Rate per 100 patient years with:

Baseline characteristic

Rofecoxib

Naproxen

NNT at one year

Steroid use, 65 or older

3.5

12.4

11

Prior event, 65 or older

16

29

8

Prior event, steroid use

5.8

21

7

Prior event, steroid use, 65 or older

10

41

3

No risk factors

The effect of rofecoxib was present even in patients with no risk factors. The rate per 100 patient years with naproxen was 1.9, and with rofecoxib was 0.2. The relative risk was 0.12 (0.02 to 0.96) and the NNT for rofecoxib to prevent one event at one year compared with naproxen was 59 (Figure 4).

Figure 4: No common risk factors


Comment

There are many interesting things in this fascinating paper, but we need to have a bit of caution because the number of events was not great, and in some subgroups the benefit of rofecoxib did not quite achieve statistical significance. That minor quibble apart, there are some stark take home messages:

The rate of events with rofecoxib in people with no common risk factors was the same (2 per 1,000 person years) as that found for people not taking any NSAID in an elderly community control .

In many patient groups the benefit of rofecoxib in terms of NNTs for one year were low, and at least equivalent to those we expect from statins in secondary prevention, where NNTs of 10-30 are found for the prevention of fatal and nonfatal strokes and heart attacks, but over five years.

Finally, the benefits are highest in higher risk groups in generally in accord with current (2003) guideance on their use in the UK.