Shi W et al / Acta Pharmacol Sin 2004 Mar; 25 (3): 357-365

Risk factors of adverse drug reaction from non-steroidal anti-inflammatory drugs in Shanghai patients with arthropathy¹

Wen SHI², Yong-ming WANG, Shao-li LI³, Min YAN³, Duan Li, Bin-yan CHEN, Neng-neng CHENG

Department of Pharmacology, School of Pharmacy, Fudan University, Shanghai 200032; ³The Center for Drug Reevaluation and The Center for Adverse Drug Reaction Monitoring, State Food and Drug Administration, Beijing 100061, China

¹ The project was supported by The Center for Drug Reevaluation and The Center for Adverse Drug Reaction Monitoring, State Food and Drug Administration, China. Grant No BK-PJ-0301.

² Correspondence to Wen SHI. Phn 86-21-5423-7465. Fax 86-21-5491-0623. E-mail vivianshi2001@yahoo.com.cn

Received 2003-05-06 Accepted 2003-09-12

KEY WORDS non-steroidal anti-inflammatory drugs; adverse drug reaction; risk factors; retrospective studies; quality of life

ABSTRACT

AIM: The study was to screen the possible risk factors of adverse drug reaction (ADR) induced by non-steroidal anti-inflammatory drugs (NSAIDs) in Shanghai patients with arthropathy. METHODS: The subjects were randomly selected from a database of outpatients with arthropathy from 9 main hospitals in Shanghai. A door to door retrospective epidemiological survey was used to collect demographic information about the patients, both individual and familial. This included data on their medical histories, lifestyle and dietary habits, history of smoking and alcohol consumption, history of drug therapy, quality of life (QOL) prior to NSAIDs intake, history of NSAIDs therapy and its ADR events, etc. Descriptive statistical methods and univariate analysis were also used to identify possible risk factors for ADRs induced by NSAIDs. RESULTS: Of the 1002 patients surveyed, the average length of NSAIDs intake was 2 years. ADR incidence from different NSAIDs was high, in a range from 46.7 %-66.2 %. In general, the candidate risk factors for ADRs were different for each NSAID. Each of the candidate risk factors were defined and studied in order to evaluate its role in the determination of ADRs from NSAIDs. "Family history of ADRs caused by NSAIDs" was found to be a significant risk factor for the four commonly used NSAIDs: meloxicam, diclofenac, nimesulide, and nabumetone. CONCLUSION: A retrospective epidemiological survey was useful in detecting the risk factors for ADRs caused by NSAIDs. The study found that different NSAIDs might have different risk factors and that there is no single risk factor universally applicable to all NSAIDs.

INTRODUCTION

In clinical practice, adverse drug reactions (ADRs) often follow the successful treatment of arthropathy by non-steroidal anti-inflammatory drugs (NSAIDs). To decrease or even avoid the high probability of ADR occurrence from NSAIDS whose incidence was above 30 % for long term intake^[1], many new types of NSAIDs like Cox-2 selective/specific inhibitors were developed. These drugs were clinically proven satisfactory for decreasing gastro-intestinal (GI) side effects, the most common ADR from NSAIDs^[2,3]. However, Cox-2 inhibitors are not safe in all respects, a feature shared with all traditional NSAIDs^[4]. Different NSAIDs have different typical ADRs and the severity of these can also differ according to the patient's specific circumstances^[1,5-10]. It should be possible, therefore, to develop some practical methods to minimize or even avoid ADRs by selecting the most suitable NSAID on an individual basis. The success in determining the risk factors for GI bleeding^[11] helps to support the hypothesis that it might also be possible to identify the risk factors for overall ADR occurrence and then select the most suitable NSAID for each particular patient. In this study, we tried to obtain some indicators of ADR risks through a broad scan of many variables from the demographic data, use of NSAIDs, lifestyle, quality of life (QOL), etc.

MATERIALS AND METHODS

Study sample The subjects met all the following enrollment criteria:

1) Being a patient with one of the following diseases: osteoarthritis, rheumatoid arthritis, ankylosing spondylitis, psoriatic arthritis, or some other non-malignant arthropathy, according to the American College of Rheumotology standards.

2) Being a Shanghai resident (of either gender).

3) The patient began treatment with NSAIDs some time between 1996 January 1st and 2000 December 31st. Also included were patients with a history of NSAID treatment who renewed such treatment any time between 1996 January 1st and 2000 December 31st, with an interval of more than 6 months between the two NSAIDs applications.

4) All patients should have taken NSAIDs continuously or intermittently for at least 1 year.

5) If the patient was on NSAIDs intermittently, the combined therapy time should exceed more than half of the total observational period.

Concomitant diseases did not exclude patients from enrollment, but subjects with no detailed address information, no compliance with the investigation, who were older than 85 years, had memory disorders or personality disorders such as paranoia, deliria, schizophrenia, etc, were excluded.

Data collection and preparation The collection of survey data was limited to the period from 1996 January 1st to 2002 January 31st. Data were collected using the investigator-administrated Case Report Form (CRF) through door-to-door visits. Letters and phone calls were used to clarify any concerns about the recorded information on CRFs. Interviewers were all trained and examined to improve their level of understanding of the CRF questions.

Further more, they were all trained to be qualified in terms of good interview techniques and field survey bias control. Twenty percent of all completed CRFs were randomly selected for re-interview by other interviewers to double-check information reliability.

Data on CRFs were then keyed into the Access database and the data were double-checked by computer. If any abnormal deviation was identified, the suspected data was corrected by the study coordi-nator, data administrator and clinical specialists. In such cases, proper records, including reasons for the corrections, were made and signed. The database was locked when it was completed.

The retrospective epidemiological survey strictly followed ICH GCP guidelines. Each of the subjects was given a full explanation of what would be involved in the study and fully understood what the study was about. All subjects signed the consent form before they were enrolled in the study. The Ethic Committees of the nine involved hospitals approved the study protocol before the study started.

Data collected includes patient's demographic information, history of NSAIDs taken and associated ADRs, concomitant drug therapy, family and personal anamnesis, smoking and alcohol consumption history, lifestyle, dietary habits, mood, previous drug therapy, and QOL prior to NSAIDs therapy.

All recorded adverse events (AEs) appearing during the period of NSAIDs intake were comprehensively evaluated by the ADR evaluation group composed, of the researchers and professional rheumatologists. Cases were excluded only if the relationship analysis indicated that the AE was impossible or unrelated to NSAIDs, or after withdrawal of the suspected NSAIDs the AE symptoms increased in severity.

Statistical analysis SAS (Version 8.0) was used for all statistical analysis. Descriptive statistics of the measurement data were expressed as mean±SD and were used to interpret the overall use of NSAIDs. the R×C Chi-Square test was used to evaluate the enumeration data, such as gender, education, dosage details, etc. Wilcoxon's test was used to evaluate rank data, such as smoking, drinking alcohol, tea and coffee and other lifestyle traits, etc. as well as measurement data such as age, weight, blood pressure, medical payments, diseases course, etc. as many of the measurement data were not normal distributed. A logistic regression model was used to determine the risk factors when all the variables were considered as a whole. A discriminatory function analysis was adopted for the determination of the discriminatory equation.

RESULTS

A total of 1002 patients were enrolled in the study, 230 (22.95 %) male, and 772 (77.05 %) female, with a mean age of 53.7±13.2 (9-84 a). The course of disease was 2517.4±2244.5 d. Among all the patients, 847 (84.5 %) were diagnosed with rheumatoid arthritis, 44 (4.4 %) with osteoarthritis, 57 (5.7 %) with psoriatic arthritis, 16 (1.6 %) with prolapsed lumbar intervertebral disc, and another 38 with other forms of arthopathies. Patients 980 (97.8 %) had at least one concomitant drug therapy during the observation period. All the patients had at least a one-year history of NSAIDs intake.

Diclofenac, ibuprofen, indomethacin, nimesulide, meloxicam, and nabumetone were found to be the top six orally taken NSAIDs among Shanghai patients with arthropathy. The incidences of ADR of these drugs were summarized in Tab 1. Diclofenac was particularly widely taken, with 926 (92.4 %) patients taking this NSAID among those in the survey. We also found that patients typically switched NSAIDs, with 834 (83.2 %) patients found to have switched NSAID (accounted by chemical names of drugs) at least once, for different reasons. Among these 834 patients, 301 (30.04 %) switched 2 NSAIDs, 230 (22.95 %) switched 3, 163 (16.27 %) switched 4, 110 (10.98 %) switched 5 and 30 (2.99 %) switched 6 or more. The main reasons for withdrawal and switching of NSAIDs were: in 248 (24.75 %) cases prescription changes, in 224 (22.36 %) cases ADR intolerance, 205 (20.46 %) thought the drugs were ineffective, 38 (3.79 %) thought they had recovered or their health condition had improved and 58 (5.79 %) were for other unknown reasons.

Tab 1. The use of NSAIDs and incidence of ADR in Shanghai participants with arthropathy.

No	Generic name of NSAIDs	Case	No ADR/%*	ADR/%	Average dosage	Average dosage
					per day/mg	duration/aD
1	Diclofenac	926	532(57.5)	394 (42.5)	99.2±77.0	2.00±0.38
2	Ibuprofen	447	255 (57.0)	144 (32.2)	740.7±352.1	2.01±0.37
3	Indomethacin	436	224 (51.4)	212 (48.6)	240.1±127.4	2.00±0.31
4	Nimesulide	266	147 (55.2)	119 (44.7)	198.5±67.7	2.05±0.35
5	Meloxicam	248	149 (60.1)	99 (39.9)	14.7±11.5	2.30±0.74
6	Nabumetone	145	96 (66.2)	49 (33.8)	988.2±421.9	1.98±0.14
7	Diclofenac+misoprostol	97	50 (56.7)	47 (48.5)	144.2±139.9	1.99±0.14
8	Sulindac	69	37 (53.6)	32 (46.4)	353.2±124.3	1.99±0.12
9	Acemetacin	53	28 (52.8)	25 (47.2)	107.3±50.5	2.13±0.52
10	Piroxicam	45	21 (46.7)	24 (53.3)	26.7±12.2	1.98±0.15
11	Oxaprozin	25	25 (56.0)	11 (44.0)	432.0±160.0	2.08±0.49
12	Acetylsalicylic acid	17	10 (58.8)	7 (41.2)	567.7±179.4	2.00±0.00

ADR data for which the number of cases was less than 10 were omitted.
* R×C Chi-square test, ^DWilcoxon's test.

There were 504 (50.3 %) patients who suffered from ADRs from NSAIDs. Among these, 366 (36.5 %) patients suffered one ADR, 101 (10.1 %) patients suffered 2 ADRs and 37 (3.7 %) suffered 3 ADRs. Tab 2 shows the composition of different ADRs caused by NSAIDs and the number of patients who received therapy for the ADRs.

Tab 2. ADR events induced by NSAIDs.

ADR data for which the number of cases was less than 10 were omitted.

The risk factors of the ADRs varied widely among different NSAID subgroups. In total, 35 variables were found significant for the top 6 NSAIDs. No single variable was found to be universally applicable as a risk factor for all of the NSAIDs. Among the 35 variables, the "family history of ADR caused by NSAIDs" was the sole significant risk factor for the following commonly used NSAIDs: meloxicam, diclofenac, nimesulide and nabumetone. Other variables include "Compared to one year ago, how would you rate your health in general now?" (related to ADRs in diclofenac, ibuprofen and indomethacin), "Daily alcohol consumption" (related to ADRs in meloxicam, diclofenac and indomethacin), "Course of the disease" (related to ADRs in meloxicam, diclofenac and nabumetone), "Degree of personal care" (related to ADRs in diclofenac, nabumetone and nimesulide) and "Concomitant drug therapy" (related to ADRs in ibuprofen, nimesulide and indomethacin). The remaining 7 variables were significant in two of the six NSAIDs subgroups and the last 22 variables were significant for one out of the six NSAIDs subgroups (Tab 3.1-3.6. *R×C Chi-square test, the statistics is c². ^DWilcoxon's test, the statistics is Z)

Tab 3.1. Univariate analysis for scan of the risk factors of ADR caused by meloxicam. * R×C Chi-square test, ^DWilcoxon's test.

Tab 3.2. Univariate analysis for scan of the risk factors of ADR caused by diclofenac.

Tab 3.3. Univariate analysis for scan of the risk factors of ADR caused by ibuprofen.