Australasian Society of Clinical Immunology and Allergy Adverse Reactions to NSAIDs David Henry Faculty of Medicine and Health Sciences The University of Newcastle NSW Australia
Aims of this presentation To show how epidemiological methods can be used to investigate adverse drug effects To review the main ‘predictable’ adverse effects of NSAIDs on the GI tract kidneys and cardiovascular system To discuss how such data can be used in health services research to influence practice and policy
Background Non steroidal anti-inflammatory drugs (NSAIDs) are widely used for management of a range of painful disorders NSAIDs are the commonest reported cause of serious adverse reactions to drugs NSAIDs can cause serious ADRs in any body system, but those involving the upper GIT the CVS and the kidneys are the most common The outcomes of these ADRs are hemorrhage, ulcer perforation, renal failure, heart failure
Questions that need to be addressed How frequently are these drugs used? What are the levels of risk of development of serious GIH, functional renal impairment, congestive heart failure? –To what extent are these risks dependent on factors such as drug dose, half-life, COX selectivity?
Questions that need to be addressed (cont) What are the associated public heath burdens? What are the practice and policy implications?
How commonly are these drugs used? Prescription survey : In million prescriptions were written for NSAIDs in Australia Prevalence survey: In 1987/88 23% of women and 20% of men aged >65 years in Newcastle had taken NSAIDs in the previous 4 days (not including aspirin). This figure dropped to around 15% by 1994
What are the levels of risk for development of serious gastrointestinal hemorrhage and ulcer perforation, and what factors modify this risk?
Principal Methodology: Case-control Study Major advantages over traditional case series Involves definition and careful selection of –subjects with the disease (cases) –subjects without the disease (controls) Unbiased measurement of prior use of the drugs of interest
Principal Methodology: Case-control Study The Odds Ratio (adjusted for confounders ) provides an estimate of the Relative Risk of the outcome in exposed compared with non-exposed individuals The Odds Ratio is an accurate estimate of the Relative Risk when the condition under study is uncommon
Relative Risk Background Risk 1/1000 Relative Risk 4.0 Risk in exposed group 4/1000 Excess risk 3/1000
Major gastrointestinal complications Case-control study to determine the risk of development of GI complications Case-control study to determine the risk of dying of GI complications Collaborative meta-analysis to determine the variation in risk with individual compounds
Major gastrointestinal complications (cont) Information on 644 cases of GIH or ulcer perforation and 1268 community and hospital controls (matched for age and sex) Overall Relative Risk 3.0 (2.3, 3.5) –Dose response relationship obvious –Apparent differences in risk between individual drugs
Death from major GI complications Information on 80 cases who died of peptic ulcer complications and 160 controls who survived hospitalisation with peptic ulcer complications –Controls were matched on age, sex, site and nature of complication and analyses were adjusted for co-morbidity Adjusted RR of dying of peptic ulcer complications with use of NSAIDs or aspirin were 1.1 (0.6, 2.1) and 1.2 (0.5, 1.9) respectively
Collaborative Meta-analysis Investigators from six countries pooled data from 12 epidemiological studies with data from 10,000 cases and 30,000 controls The main aim was to investigate the range of risks with individual agents and the extent to which this could be explained by dosage Results were combined by ranking as different combinations of drugs had been included in different studies
Estimated relative risks of major gastrointestinal complications with individual drugs AzaPiroxKetoIndoNapAspSulinDiflDiclIbup Estimated relative risk
Ibup n=11 Dicl n=9 Difl n=2 Fenop n=2 Asp n=6 Sulin n=5 Nap n=12 Indo n=12 Pirox n=11 Ketop n=7 Tolm n=2 Aza n= Rank from low risk to high risk: avge of 12 top scores Estimated relative risks of major gastrointestinal complications with individual drugs Summary ranking method Estimated relative risks of major gastrointestinal complications with individual drugs Summary ranking method
Overall RR around 3.0. Higher for perforation Case-fatality rate unaffected by NSAIDs 34% of elderly bleeds due to NSAIDs Significant differences between individual drugs: ibuprofen and diclofenac have the lowest risks and piroxicam and ketoprofen the highest Major gastrointestinal complications : Conclusions
Cardiac and Renal Complications Vasodilator PGs are important in maintaining renal blood flow, and (possibly) modulating systemic vascular resistance in times of stress NSAIDs have been shown to reduce GFR and increase systemic vascular resistance in selected individuals There have been very few epidemiological studies and these have been confined to end- stage renal failure
Cardiac and Renal Complications Case-control study of 365 cases of CHF and 658 matched controls Case-control study of 110 cases of functional renal impairment (FRI) and 189 matched controls The estimated RR for hospitalisation with first episode of CHF with NSAIDs was 2.8 (1.5, 5.1). Etiologic fraction 19% The estimated RR for hospitalisation with evidence of FRI with NSAIDs was 1.5 (0.80, 2.9).
How do we study factors such as drug dose, half-life and COX selectivity in epidemiological studies of NSAID toxicity?
Effects of Dose on Relative Risk of GI Complications with Individual NSAIDs: meta-analysis LowHighLowHighLowHigh RR, 95% CI IbuprofenNaproxenIndomethacin
Effects of Dose on Relative Risk of CHF
The Relationship between GI Risk and Drug Half-life Relative risks were ordered by the summary ranking procedure (meta- analysis) Published values for T1/2 were ranked Non parametric correlation: –Kendall’s Tau 0.504, P = 0.023
The Relationship between Half-life and risk of CHF Relationship between T1/2 and RR for first admission with CHF (multivariate): –<4 hrs 2.0 (0.56, 7.4) – hrs 4.3 (0.66, 27.4) –> 12 hrs 10.6 (1.1, 103.7) –P (for trend) = 0.016
The Relationship between Half-life and risk of Renal Impairment Relationship between T1/2 and RR for functional renal impairment (multivariate): –<4 hrs 1.3 (0.48, 3.6) – hrs 2.8 (0.66, 12.0) –> 12 hrs 3.8 (0.68, 21.0) –P (for trend) = 0.012
COX selectivity Cyclo-oxygenase exists in two isoforms, COX- 1 and COX-2 COX-1 is responsible for the ‘constitutional’ effects of PGs COX-2 is responsible for the inflammatory effects
COX selectivity Drugs were ranked by COX-1, COX-2 activities and COX-1/COX-2 ratio Relative risks were ordered by the summary ranking procedure (meta- analysis) Significant correlation with COX-1 activity, but not with the other variables
Public Health Issues GI complications: admissions in elderly subjects annually, and deaths Approximately 1500 admissions and 100 deaths are attributable to use of NSAIDs Cardiac Failure: 50,000+ admissions annually, with 5000 deaths in hospital and many more in the next year Approximately 8000 admissions and up to 800 deaths could be attributable to use of NSAIDs
What is happening to use of NSAIDs in Australia ?
Recent Trends in Use of NSAIDs McManus et al MJA 1996; 164: Overall use of NSAIDs fell from 50.1 DDD/1000/day in 1990 to 34.6 DDD/1000/day in 1994 (31% fall) In subjects over 60 years prescription use fell by 44%
Recent Trends in Use of NSAIDs McManus et al MJA 1996; 164: The fall in the use of ‘low risk’ drugs (ibuprofen, diclofenac and diflunisal) was greater than the fall in ‘high risk’ drugs (piroxicam and ketoprofen)
Selection of NSAIDs by GPs Analysis of NSAID prescribing by 2 GP Divisions with similar overall rates (per 100 MC services) Feedback of data may result in an overall reduction in prescribing and a shift to the left
Conclusions The high use of NSAIDs constitutes a problem of public health dimensions The cardiovascular effects are little recognised and in community terms may be more important than the GI effects There are significant differences in risk between the drugs. Both dose and drug half-life seem important COX-selectivity potentially is an important factor but assessment with current data difficult due to confounding