Adherence to weekly oral bisphosphonate therapy: cost of wasted drugs and fractures.

UNLABELLED: In an observational cohort of patients treated with biphosphonates (BP), we observed that poor adherence to these drugs causes important expenditures in terms of avoidable fractures. Of particular interest are the amounts of money wasted by patients who did not take their BPs long enough to obtain a clinical benefit. INTRODUCTION: A large proportion of patients initiated with oral weekly BP therapy stop their treatment within the first year. The objective of this study was to estimate the impact of the poor adherence to BPs in terms of drug wasted and avoidable fractures. METHODS: The study was done on primary and secondary prevention cohorts from the Régie de l'assurance maladie du Québec (Québec). The concept of the "point of visual divergence" was used to determine the amount of wasted drug. The risk of fracture was estimated using Cox regression models. The hazard ratios of compliant patients (+80%) versus non compliant patients were used to estimate the number of fractures saved. RESULTS: The cost of wasted drugs was $25.87 per patient initiated in the primary prevention cohort and $30.52 in the secondary prevention cohort. If all patients had been compliant, 110 fractures would have been avoided in the primary prevention cohort and 19 fractures in the secondary prevention cohort. The cost of these avoidable fractures per patient initiated on BP therapy was $62.95 in primary prevention cohort and $330.84 in secondary prevention cohort. CONCLUSIONS: This study confirms that poor adherence to oral BPs leads to a significant waste of money and avoidable fractures.

Differences in persistence among different weekly oral bisphosphonate medications.

SUMMARY: We evaluated the differences in persistence with weekly oral bisphosphonate therapy according to the initial drug. Persistence to weekly oral preparations remains suboptimal, particularly in patients who receive generic alendronate. Alternative solutions are needed to improve the real life effectiveness of osteoporosis therapies. INTRODUCTION: Poor persistence is widespread with oral osteoporosis (OP) therapy. The objective of this study was to evaluate the persistence among OP patients started on weekly oral bisphosphonates (BP). METHODS: Patients newly initiated on branded risedronate, branded alendronate, or generic alendronate once weekly were selected from the Régie de l'Assurance Maladie du Québec databases. The cohort included patients with and without a previous OP fracture. The probability and the risk factors for early discontinuation were estimated using Cox regression models. RESULTS: The study cohort included 32,804 patients. After 1 year, a significant difference in persistence on oral BP therapy was found. The patients started on branded risedronate were 11% more likely to stop OP therapy than patients started on branded alendronate. Risk of discontinuation doubled in patients initiated with generic alendronate compared to patients started on branded alendronate. Male gender was associated with a 25% increase risk of early discontinuation. No statistical association was found between previous OP fracture and early discontinuation. CONCLUSION: This study provides further evidence of poor persistence to newly initiated oral weekly BP therapies, particularly for the patients started on generic alendronate.

Osteoporosis among patients with type 1 and type 2 diabetes.

Both diabetes and fractures are prevalent in adults. The relationship between diabetes and osteoporosis is complex and, although it has been investigated extensively, the subject remains controversial. While low bone mineral density (BMD) is consistently observed in type 1 diabetes, the relationship is less clear in type 2 diabetes, with some studies reporting modestly increased or unchanged BMD. Both type 1 and type 2 diabetes have been associated with a higher risk of fractures. Despite discrepancies between BMD and fracture rates, clinical trials uniformly support the fact that new bone formation and bone microarchitecture and, thus, bone quality, are altered in both types of diabetes. Although a causal association between diabetes and osteoporosis cannot be established on the basis of existing data, it is possible to conclude from many studies and from a better understanding of the physiopathology of diabetes that it can increase the risk of fractures through skeletal (decreased BMD and bone quality) and extraskeletal (increased risk of falls) factors. Even though osteoporosis screening or prophylactic treatment in all patients with type 1 and type 2 diabetes is not being recommended at present, such patient populations should be given general guidelines regarding calcium and vitamin D intakes, exercise and the avoidance of potential risk factors for osteoporosis. The extent of diagnostic and therapeutic interventions should be based on the individual's risk profile for fractures.

Assessment of Canadian provincial expenditures in depressed patients treated with venlafaxine XR versus SSRIs: the APEX Study.

OBJECTIVE: Empirical studies of antidepressant cost-effectiveness suggest that the use of venlafaxine may be no more costly than selective serotonin reuptake inhibitors (SSRIs) in the treatment of depression. The objectives of this study were to identify patients' characteristics and factors associated with the choice of antidepressant and to assess differences in persistence, healthcare utilization and direct medical costs associated with venlafaxine and SSRIs pharmacotherapy. RESEARCH DESIGN AND METHODS: We examined demographic and clinical characteristics of patients (n = 17 144) who received both a diagnosis of depression and a prescription for venlafaxine or an SSRI between 1996 and 2004 using the Quebec health administrative databases. Logistic regression models were used to identify factors independently associated with the choice of antidepressant. Persistence to treatment and overall direct medical costs during 12 months after initiation of therapy were assessed using Cox proportional hazard and GLM models, respectively. RESULTS: Age, sex, provider specialty, and prior 12-month healthcare utilization significantly influenced initial antidepressant choice. Fewer venlafaxine-treated patients discontinued their initial therapy relative to SSRIs' (persistence to initial treatment: 38.4% vs. 29.4% and 24.4% vs. 15.8% at 6 and 12 months, respectively; p < 0.0001), and they were less likely to require treatment switching. Overall 12-month direct medical costs for SSRI- and venlafaxine-treated patients were Can$2759 and Can$2604, respectively. Patients treated with SSRIs had significantly higher expenditures in a univariate analysis (cost ratio: 1.06 [95% CI: 1.02, 1.10]). However, after controlling for potential confounding factors such as patients' characteristics, prior healthcare utilization, and comorbid conditions in multivariate analyses, the overall expenditures were similar in both groups (cost ratio: 1.03 [95% CI: 0.99, 1.07]). CONCLUSIONS: Direct medical costs were generally similar among patients with depression treated with venlafaxine and SSRIs. In a 'real world' setting, the higher acquisition cost of venlafaxine is offset by savings due to fewer hospitalizations and fewer outpatient medical visits. Differences in treatment persistence may also, in part, explain the observed differences in average direct medical costs between venlafaxine and SSRIs.

3-Hydroxy-3-methylglutaryl coenzyme A reductase inhibitors and the risk of cancer: a nested case-control study.

BACKGROUND: During the past 15 years there has been an exponential increase in the number of prescriptions for lipid-lowering drugs. Uncertainties remain about the long-term impact of these medications on cancer, which is particularly bothersome given that the duration of these treatments may extend for several decades. OBJECTIVE: To explore the association between 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors and cancer incidence. METHODS: Using the administrative health databases of the Régie de l'Assurance-Maladie du Québec we performed a nested case-control study. We selected a cohort of 6721 beneficiaries of the health care plan of Quebec who were free of cancer for at least 1 year at cohort entry, 65 years and older, and treated with lipid-modifying agents. Cohort members were selected between 1988 and 1994 and were followed up for a median period of 2.7 years. From the cohort, 542 cases of first malignant neoplasm were identified, and 5420 controls were randomly selected. Users of HMG-CoA reductase inhibitors were compared with users of bile acid-binding resins as to their risk of cancer. Specific cancer sites were also considered. RESULTS: Users of HMG-CoA reductase inhibitors were found to be 28% less likely than users of bile acid-binding resins to be diagnosed as having any cancer (rate ratio, 0.72; 95% confidence interval, 0.57-0.92). All specific cancer sites under study were found to be not or inversely associated with the use of HMG-CoA reductase inhibitors. CONCLUSION: The results of our study provide some degree of reassurance about the safety of HMG-CoA reductase inhibitors.

Predictors of persistence of use of the novel antidiabetic agent acarbose.

BACKGROUND: A carbose is the first of a new class of antidiabetic agents, the alpha-glucosidase inhibitors. This study characterizes and identifies predictors of persistence of use of acarbose. METHODS: Medical, pharmaceutical, and demographic records were extracted for 2 cohorts of patients (social assistance recipients and seniors) from the databases of Quebec's provincial health plan. Patients were eligible for inclusion if they had received their first dispensation of acarbose between August 1, 1996, and December 31, 1997. The observation period included at least 1 year before the first dispensation and a minimum of 4 months after. RESULTS: New users of acarbose included 216 social assistance recipients and 677 seniors who were followed up for 82 914 and 270 041 person-days, respectively. Median persistence with acarbose treatment was 83 days (95% confidence interval, 75-105 days) for social assistance recipients and 105 days (95% confidence interval, 90-119 days) for seniors. In both cohorts, treatment by an endocrinologist vs another physician predicted longer treatment persistence. In the seniors cohort, additional determinants of (earlier) treatment discontinuation included a higher initial daily dose, previous treatment with insulin, and consultation with a gastroenterologist after treatment initiation. CONCLUSIONS: New users of acarbose showed low persistence in 2 cohorts of beneficiaries of Quebec's provincial health plan. Prescribing specialist was an important predictor of persistence in seniors and the socially assisted. The importance of 4 additional factors in seniors only led to hypotheses concerning population differences in treatment expectations and in the occurrence and tolerance of adverse effects.

Diet and probucol in lowering cholesterol concentrations. Additive effects on plasma cholesterol concentrations in patients with familial type ii hyperlipoproteinemia.

Probucol [4,4-(isopropylidendithio bis)(2,6-di-t-butylphenol)], as as an adjunct to diet, was evaluated for its effect on lowering the plasma cholesterol level in patients with familial hypercholesterolemia (type II). The trial had a double-blind, placebo-controlled, crossover design. About half of the 30 patients responded to a low-cholesterol modified-fat diet with a decrease in the plasma cholesterol level of approximately 13%. When probucol was added to the diet of the responders, their plasma cholesterol level was lowered a further 13%. Patients who did not respond to the diet did show reduced plasma cholesterol concentrations when receiving probucol plus the diet. Analysis of the cholesterol content of the various lipoprotein fractions showed that the low-density lipoproteins accounted for most of the total plasma cholesterol level decrease. There was, as expected, no effect on plasma triglyceride concentrations. Neither the 7-dehydrocholesterol nor the desmosterol level was increased in the plasma of patients treated with probucol for three months. Probucol is useful as an adjunct to diet in lowering plasma cholesterol levels in patients with familial hypercholesterolemia. The drug was well tolerated by all patients.

Effects of smoking and chronic hepatitis B on lidocaine and indocyanine green kinetics.

Kinetics of lidocaine (L) and indocyanine green (ICG), substances with a high hepatic extraction ratio, was studied in 9 normal subjects (4 smokers and 5 nonsmokers) and in 6 patients with chronic type B hepatitis without portal hypertension. L metabolism was studied in each subject after intravenous and oral administration. The data were used to calculate L systemic and oral clearances, L systemic bioavailability, and apparent hepatic blood flow. In smokers, L systemic bioavailability was decreased secondary to a marked increase in oral clearance, reflecting induction of drug-metabolizing activity. In patients with chronic hepatitis, L systemic and oral clearances were higher than in the normal. These findings indicate that hepatic handling of drugs with a high hepatic extraction ratio, such as L, might be enhanced in patients with chronic type B hepatitis. L disposition approach was validated in 5 patients by comparing results with those using the ICG clearance and extraction method at the time of hepatic vein catheterization. The L systemic bioavailability after oral administration is a reflection of first-pass clearance by the liver and might be a useful kinetic method for evaluating overall ability of the liver to remove drugs with high hepatic extraction ratios.

Tixocortol pivalate, a corticosteroid with no systemic glucocorticoid effect after oral, intrarectal, and intranasal application.

Tixocortol pivalate is a corticosteroid with topical anti-inflammatory activity equal to that of hydrocortisone. It was evaluated in a group of 18 normal subjects to determine whether it exerted any systemic glucocorticoid activity after single oral or intrarectal doses and after short-term dosing by the intranasal route. Effects of tixocortol pivalate were compared to those of oral dexamethasone and intrarectal betamethasone 21-phosphate. By the three routes, tixocortol pivalate does not induce any changes in plasma cortisol, leukocyte counts (neutrophils, lymphocytes, monocytes, eosinophils), blood glucose, or 24-hr urinary excretion of sodium and potassium, whereas there were changes after dexamethasone and betamethasone. Tixocortol pivalate, however, increased urinary free cortisol-like substances. It is concluded that tixocortol pivalate given for short periods by nonparenteral routes does not induce a measurable systemic glucocorticoid effect.

Do statins cause cancer? A meta-analysis of large randomized clinical trials.

PURPOSE: Although the short-term safety and tolerability of statins has been well established, their potential carcinogenicity in the long term is still debated. The goal of this study was to determine whether long-term treatment with statins is associated with an increased risk of fatal and nonfatal cancers. METHODS: We searched the Medline database between January 1966 and December 1999 for randomized, controlled trials of human subjects in which monotherapy with a statin was compared with placebo. No language restrictions were applied. Only trials with a minimum treatment duration of 4 years and a minimum of 1,000 subjects were included. Studies that did not provide information on fatal or nonfatal cancers were excluded. Data on fatal and nonfatal cancers and all-cause mortality were extracted by a single nonblinded reviewer. Overall crude estimates of risk difference were computed by summing the numerators and denominators of trial-specific risk estimates. RESULTS: Five trials met the inclusion criteria. The estimated differences in absolute risk between treatment and placebo were as follows (negative risks indicate that treatment was safer than placebo): all nonfatal cancers, 0.0% (95% confidence interval [CI]: -0.8% to 0.8%); all fatal cancers, -0.1% (95% CI: -0.7% to 0.4%); all fatal and nonfatal cancers combined, -0.1% (95% CI: -1.0% to 0.7%); and all-cause mortality, -1.5% (95% CI: 2.8% to 0.2%). CONCLUSION: This study demonstrates no association between statin use over a 5-year period and the risk of fatal and nonfatal cancers. This conclusion is limited by the relatively short follow-up of the studies analyzed. Similar analyses of data from studies with longer follow-up periods would be valuable.

Shanghai trial of nifedipine in the elderly (STONE).

OBJECTIVE: To assess the effectiveness of nifedipine treatment in elderly hypertensives. METHODS: A single-blind trial was conducted under the direction of the Shanghai Institute of Hypertension in 1632 subjects aged 60-79 years alternatively allocated to either nifedipine or placebo after a 4-week placebo run-in period between 1987 and 1990 with mean follow-up of 30 months. Clinical events and risk modification were analysed in collaboration with the University of Montreal. Seventy-four patients with severe hypertension were reallocated to active nifedipine treatment after placebo run-in. RESULTS: Cox's proportional hazards model accounting for covariates demonstrated a highly significant decrease in the probability of events: 'original treatment assignment' analysis indicated that 77 events occurred in the placebo and 32 in the nifedipine group. Similar significances were achieved with 'actual treatment' or 'changes excluded' (excluding reallocated subjects) analyses. A significant reduction in relative risk was observed for strokes and severe arrhythmia with an overall decrease from 1.0 to 0.41 (95% confidence interval 0.27-0.61). CONCLUSION: Nifedipine treatment diminished the number of severe clinical outcomes in elderly hypertensives significantly.

Selection of controls in database case-control studies: glucocorticoids and the risk of glaucoma.

In case-control studies conducted using computerized databases, controls are often selected as a random sample from the base population. This representative choice of controls is intended to guard against selection bias. We show, using data from a database case-control study, that such a definition of controls may also lead to selection bias under two conditions: (1) if the target disease has a prolonged asymptomatic clinical course with its detection depending on a specific physical examination and (2) if exposed patients have a higher likelihood of having the disease detected than unexposed patients. The extent of the bias that could result from the use of randomly selected controls was investigated in the context of a case-control study of the risk of ocular hypertension or glaucoma associated with the use of glucocorticoids, conducted using the Quebec universal health insurance computerized databases. This article also illustrates that a computerized database can be useful to empirically explore opportunities for bias.

Impact of a cost-sharing drug insurance plan on drug utilization among older people.

OBJECTIVES: In August 1996, the Régie de l'assurance-maladie du QuEbec (RAMQ), the government body responsible for medical insurance in the Canadian province of Quebec, introduced a cost-sharing drug insurance plan. Before this plan, individuals age 65 years and older had to pay Canadian (CDN)$2 per prescription, with the remaining cost paid by the RAMQ. With the new plan, beneficiaries may have to pay an amount between CDN$200 and CDN$925 per year, depending on their income. Concerned that this financial constraint imposed on older people might have an impact on the use of medications, we investigated whether the consumption of four classes of medications, antihypertensive agents, anticoagulants, nitrates, and benzodiazepines, was affected by the drug plan implementation. DESIGN: Time series models with pre/post comparison group. SETTING: Administrative computerized databases of the RAMQ. PARTICIPANTS: Random sample of Quebec residents age 65 years and older registered in the provincial drug plan between August 1992 and June 1997: 54,771 users of nitrates, 133,146 users of antihypertensive agents, 45,534 users of anti-coagulants, and 26,165 users of benzodiazepines. MEASUREMENTS: We modeled the monthly consumption of the medications under study between August 1992 and June 1996. Monthly drug consumptions predicted from the models were compared with those observed for the 13 months (August 1996 to August 1997) following the implementation of the new drug plan using 95% confidence intervals. The number of prescriptions dispensed served as an indicator for drug consumption. RESULTS: During the study period we observed a nonstatistically significant decrease in number of prescriptions of 5.1% for nitrates, 1.1% for antihypertensive agents, and 0.8% for benzodiazepines, and a nonstatistically significant increase of 1.6% for anticoagulants. CONCLUSION: Residents of Quebec age 65 years and older were not found to have reduced significantly their consumption of nitrates, antihypertensive agents, anticoagulants, and benzodiazepines during the 13 months that followed the implementation of a cost-sharing drug insurance plan.

Exposure to allopurinol and the risk of cataract extraction in elderly patients.

OBJECTIVE: To determine whether exposure to allopurinol is associated with an increased risk of cataract extraction in elderly patients. METHODS: We conducted a case-control study using data from the Quebec universal health insurance program for all elderly patients. The 3677 cases were patients with a cataract extraction between 1992 and 1994. The 21,868 controls were randomly selected among patients not diagnosed with cataract and matched to cases on the date of the extraction. We determined the odds ratio of cataract extraction according to the cumulative dose and duration of allopurinol use relative to nonusers, using conditional logistic regression analysis. The analysis was adjusted for the effects of age, sex, diabetes mellitus, hypertension, glaucoma, and ophthalmic and oral corticosteroid exposure. RESULTS: A cumulative dose of allopurinol of more than 400 g or a duration of use of longer than 3 years were associated with an increased risk of cataract extraction, with odds ratios of 1.82 (95% confidence interval [CI], 1.18-2.80) and 1.53 (95% CI, 1.12-2.08), respectively. No increase in risk was observed for lower cumulative doses or shorter exposure periods. CONCLUSION: Long-term administration of allopurinol increases the risk of cataract extraction in elderly patients.

Estrogen replacement therapy: determinants of persistence with treatment.

OBJECTIVE: To evaluate the persistence rate for estrogen therapy and to identify its determinants. METHODS: From the Quebec health insurance database we chose a cohort of 4527 women 35 years and older who received social assistance and were new users of estrogen between January 1989 and December 1997. Incident use was defined by the absence of any dispensed prescription of estrogen in the 3 years before the index date (date of first dispensed prescription). We estimated the cumulative persistence rate of treatment by Kaplan-Meier failure time analysis and identified its determinants with the Cox proportional hazards model. RESULTS: From the initial cohort, 3395 women (75%) renewed their first dispensed prescription and 905 (20%) continued treatment after 4 years. The determinants measured at the index date and significantly associated with a better persistence rate (relative risk [RR]) were younger than 60 years (RR 1.15, 95% confidence interval [CI] 1.01, 1.30), low dosage (RR 1.49, 95% CI 1.32, 1.70), continuous progestin combination (RR 1.40, 95% CI 1.27, 1.54), and a gynecologist as the first prescribing physician (RR 1.15, 95% CI 1.03, 1.21). Also, coronary heart disease or at least one risk factor for it in the year before the index date was associated with a better persistence rate for estrogen replacement therapy (RR 1.15, 95% CI 1.05, 1.22). CONCLUSIONS: The persistence rate for estrogen therapy is poor, implying that few women take it long enough to benefit from it.

Savings in direct medical costs from the use of tobramycin solution for inhalation in patients with cystic fibrosis.

OBJECTIVE: Two identical 24-week, double-blind, placebo-controlled trials of tobramycin solution for inhalation (TOBI [PathoGenesis Corporation, Seattle, Washington]) in cystic fibrosis patients with chronic Pseudomonas aeruginosa infection were conducted in the United States. The aim of the present study was to extrapolate the US trial data to a Canadian setting, using Canadian costs to estimate the savings in direct medical costs that might result from use of a similar 24-week TOBI regimen versus usual care in 2 Canadian provinces. BACKGROUND: Cystic fibrosis is a genetic disease in which persistent respiratory infection, usually due to P. aeruginosa infection, is the major cause of morbidity and mortality. METHODS: The US trials demonstrated that TOBI produced significant improvements in pulmonary function test results, reduced sputum levels of P. aeruginosa, and resulted in a 26% reduction in the probability of hospitalization (95% CI, 2%-43% vs placebo in the clinical trials). Individual patient data from the US trials were used to calculate the mean number of days in hospital as well as the mean number of days of home intravenous or oral antibiotic therapy. To adjust for Canadian pricing, pertinent economic data were obtained from Statistics Canada and the Ontario and Quebec health ministries. Demographic and baseline data were obtained from health surveys conducted by the Canadian Cystic Fibrosis Foundation. RESULTS: Economic analysis showed that the use of TOBI for 24 weeks would result in estimated mean per-patient savings in direct medical costs (in Canadian dollars) of $4055 in Ontario and $4916 in Quebec, which would substantially offset the Canadian acquisition price of $8602 for the same 24-week period. CONCLUSIONS: Assuming that the percentage of reduction in hospital days observed in the US trials would also occur in the Canadian clinical setting, use of TOBI would reduce the use of health care services, particularly hospital days, and lead to substantial savings in direct medical costs that would offset its acquisition price. Whether this reduction actually occurs after TOBI enters the Canadian market is a subject for future investigation.

The use of the bootstrap statistical method for the pharmacoeconomic cost analysis of skewed data.

In pharmacoeconomics, the comparison of the costs of 2 different drugs used for the same treatment is of great interest. The problem is especially challenging when the drugs are likely to produce costly adverse effects in a small number of patients, which is often the case. The data are then skewed and traditional statistical methods to analyse the difference in the mean costs produced by 2 treatments may be inappropriate. The bootstrap method is presented as an alternative approach. A pharmacoeconomic cost-analysis example is presented and used throughout this article.

Using healthcare claims data for outcomes research and pharmacoeconomic analyses.

Healthcare claims data are a practical complement to data from randomised controlled trials (RCTs) for evaluating health outcomes in non-experimental settings and for generalising results to a broader population. Claims data are a relatively inexpensive way to obtain useful information about patient demographics, as well as healthcare resources used for specific medical conditions and procedures from large numbers of patients over extended periods of time. With claims data, it is possible to identify patients who meet specific medical or sociodemographic criteria, estimate their costs, define episodes of medical care, and measure outcomes more globally than is possible with RCT data. Statistical methods exist to address some of the inherent issues with claims data due to their limited clinical detail. We also identify extensions of claims data to productivity issues, the use of centralised claims data such as in Canada, and the application of new statistical methods to outcomes research literature such as sample selection correction methods.

Gastrointestinal-related healthcare resource usage associated with a fixed combination of diclofenac and misoprostol versus other NSAIDs.

OBJECTIVE: To compare gastrointestinal (GI) healthcare resource use (HCRU) and associated costs in patients taking a fixed combination of diclofenac and misoprostol versus other nonsteroidal anti-inflammatory drugs (NSAIDs). METHODS: We analysed a sample (49,033 patients) of the Government of Quebec Health Insurance Agency database. Patients were included in the study if they did not have GI events during the year preceding the date of their first NSAID prescription dispensing (the index date). Patients were followed up for 2 years. A 3-stage model was used to determine the factors that influenced the direct medical costs of GI HCRU: (i) a logistic regression model (model 1) to estimate the risk of GI HCRU; (ii) a linear regression model (model 2) to estimate the direct costs of GI HCRU for those who had such events; (iii) multiplying the estimated risks from model 1 by the estimated costs from model 2 gave the estimated direct costs of GI HCRU for all patients. STUDY PERSPECTIVE: Provincial government of Quebec, Canada. RESULTS: 1,533 patients were prescribed diclofenac/misoprostol at the index date and 10,540 another NSAID. Comorbidity markers were not significantly different between the 2 groups. Of the diclofenac/misoprostol patients, 23 (1.5%) were hospitalised for GI problems compared with 194 (1.8%) of the NSAID group; 403 (26.3%) of diclofenac/misoprostol patients used gastroprotective agents compared with 2,849 (27.0%) of the NSAID patients; 118 (7.7%) of diclofenac/misoprostol patients had GI diagnostic tests compared with 682 (6.5%) of the NSAID patients. The average direct medical cost of GI HCRU was 310.52 Canadian dollars ($Can)/patient (1997 values) in the diclofenac/misoprostol group compared with $Can231.19/patient (1997 values) in the NSAID group. When adjusted for baseline factors, the ratio of the total direct medical cost of GI HCRU in the diclofenac/misoprostol group to that of the NSAID group was 1.15 (95% confidence interval: 0.89, 1.48). CONCLUSIONS: Our data showed no significant differences in GI HCRU among patients taking diclofenac/misoprostol compared with those taking NSAIDs.