Jacques LeLorier

Discrepancies between meta-analyses and subsequent large randomized, controlled trials.

BACKGROUND Meta-analyses are now widely used to provide evidence to support clinical strategies. However, large randomized, controlled trials are considered the gold standard in evaluating the efficacy of clinical interventions. METHODS We compared the results of large randomized, controlled trials (involving 1000 patients or more) that were published in four journals (the New England Journal of Medicine, the Lancet, the Annals of Internal Medicine, and the Journal of the American Medical Association) with the results of meta-analyses published earlier on the same topics. Regarding the principal and secondary outcomes, we judged whether the findings of the randomized trials agreed with those of the corresponding meta-analyses, and we determined whether the study results were positive (indicating that treatment improved the outcome) or negative (indicating that the outcome with treatment was the same or worse than without it) at the conventional level of statistical significance (P<0.05). RESULTS We identified 12 large randomized, controlled trials and 19 meta-analyses addressing the same questions. For a total of 40 primary and secondary outcomes, agreement between the meta-analyses and the large clinical trials was only fair (kappa= 0.35; 95 percent confidence interval, 0.06 to 0.64). The positive predictive value of the meta-analyses was 68 percent, and the negative predictive value 67 percent. However, the difference in point estimates between the randomized trials and the meta-analyses was statistically significant for only 5 of the 40 comparisons (12 percent). Furthermore, in each case of disagreement a statistically significant effect of treatment was found by one method, whereas no statistically significant effect was found by the other. CONCLUSIONS The outcomes of the 12 large randomized, controlled trials that we studied were not predicted accurately 35 percent of the time by the meta-analyses published previously on the same topics.

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.

Risk of ocular hypertension or open-angle glaucoma in elderly patients on oral glucocorticoids

BACKGROUND Ocular hypertension and open-angle glaucoma are well-known side-effects of treatment with topical ophthalmic glucocorticoids. There is uncertainty about the risk of these disorders with oral glucocorticoid therapy. METHODS Data from the Quebec universal health insurance programme for the elderly were used to identify 9793 patients with a new diagnosis of ocular hypertension or open-angle glaucoma, or on newly prescribed treatment for these disorders (cases). 38,325 controls were randomly selected from ophthalmology patients seen in the same month and year as the case (index date). Current use of oral glucocorticoids was defined as that within 14 days of the index date. All glucocorticoid doses were converted to the equivalent amount of hydrocortisone. The case-control analysis was done by conditional logistic regression and adjusted for age, sex, systemic hypertension, diabetes mellitus, ophthalmic glucocorticoids, glucocorticoid injections, and variables related to general health. FINDINGS The mean ages of cases and controls were similar (74.9 [SD 6.3] vs 74.7 [6.4]). The adjusted odds ratio of ocular hypertension or open-angle glaucoma for current users of oral glucocorticoids compared with non-users was 1.41 (95% CI 1.22-1.63). There was a dose-related increase in the adjusted odds ratios for current users: 1.26 (1.01-1.56) for less than 40 mg per day of hydrocortisone, 1.37 (1.06-1.76) for patients on 40-79 mg per day, and 1.88 (1.40-2.53) for patients on 80 mg or more per day. The odds ratios also increased with the duration of treatment over the first 11 months of exposure. INTERPRETATION The use of oral glucocorticoids increases the risk of ocular hypertension or open-angle glaucoma in elderly patients. In patients in this age-group who need long-term treatment with high doses of oral glucocorticoids, monitoring of intraocular pressure may be justified.

The Use of the Bootstrap Statistical Method for the Pharmacoeconomic Cost Analysis of Skewed Data

SummaryIn 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.

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 individuals risk profile for fractures.

Statins and Cancer Risk

OBJECTIVE Despite numerous randomized clinical trials and observational epidemiologic studies, evidence on the potential effectiveness of statins for prevention of cancer remains controversial. The objective of this study was to investigate the relation between lipophilic statin use and cancer occurrence. METHODS We conducted a retrospective observational study based on a medical administrative database in the province of Quebec, Canada (1998-2004). Patients aged 45 years or more and discharged from the hospital alive after admission for acute myocardial infarction were included. High- and low-dose statin use were defined as a filled prescription, within 3 days after index discharge, at or above (below) the statin-specific target dose, for any of the 4 lipophilic statin medications: atorvastatin, simvastatin, lovastatin, or fluvastatin. Statin non-use was defined as non-use of any statins while simultaneously using major non-statin cardiac medications. A total of 30,076 patients, including high-dose statin users (n=6015), low-dose statin users (n=5323), and non-users (n=18,738), were followed for up to 7 years. Multivariable Cox regression analyses were performed to estimate associations between statin dose category and the incidence of admission to hospital with a diagnosis of any type of cancer. RESULTS The crude incidence rates of hospital admission with the diagnosis of any type of cancer were 13.9, 17.2, and 26.0 per 1000 person-years in statin high-dose users, low-dose users, and non-users, respectively. The estimated adjusted hazard ratios were 0.75 (95% confidence interval [CI], 0.60-0.95) for statin use at high dose and 0.89 (95% CI, 0.75-1.07) for statin use at low dose. No significant time-dependence of the effect of statins at either dose was detected. CONCLUSION The use of lipophilic statins at sufficiently high dose might be associated with a clinically important reduction in the incidence of cancer.

Clinical consequences of generic substitution of lamotrigine for patients with epilepsy

CLINICAL CONSEQUENCES OF GENERIC SUBSTITUTION OF LAMOTRIGINE FOR PATIENTS WITH EPILEPSY To the Editor: LeLorier et al.1 studied the risks associated with patients switching to and from generic antiepileptic drugs (AEDs) in Quebec. The authors did not consider that such changes could be attributed to promotionally driven doctor and patient preferences. Industry representatives vigorously promote the idea that generics are less potent (“up to 20% less effective”) than their brand name equivalents despite Food and Drug Administration (FDA) assertions to the contrary.2 Study patients taking generics underwent dose escalations. The authors suggest that dose escalations were in response to increased side effects, but this is counterintuitive. More plausibly, anxietyinduced dose escalations contributed to side effects and, in turn, switch-backs. The unspoken hypothesis that switches to generic led to more seizures is unaddressed by the presented data, which blur psychiatric and neurologic indications for lamotrigine (LTG). LTG is used heavily in psychiatry and most recent growth in sales is driven by the psychiatric market.3 A single claim submitted with a code for epilepsy is considered sufficient evidence that LTG is being prescribed as an AED, but this is unlikely. The leading outpatient diagnostic code as well as four of five diagnostic codes for outpatient visits and two of three diagnostic codes for inpatient hospitalizations were not for epilepsy. By contrast, all comparator non-AED drugs in this study lent themselves to readily available objective efficacy assessment (blood pressure and lipid levels). There is no such equivalent for any LTG indication. FDA standards for generic bioequivalence are the same standards applied to branded medication for between-batch variability.2 Bioequivalence is complex.4 For example, the area under the curve and maximum concentration but not time to maximum (tmax) concentration are used by the FDA in determining bioequivalence. Manufacturer disclosures of bioequivalence data indicate that branded LTG tmax varies from half an hour to 6 hours for various formulations of the 100 mg tablets.5 Clinical relevance of FDA-permitted variance within a brand or between brand and generic medication is unclear. Millions of doses of generic medications have been dispensed with no well-documented instances of therapeutic failures for medications produced in accordance with existing FDA standards.4 It is clear that promotional activity influences prescribing behavior. Furthermore, under experimental conditions, expensive placebos are more effective.6 Not only patients but neurologists are anxious about generic medications encouraged in this regard by controversial American Academy of Neurology (AAN) policies on generic substitution of AEDs.7 The authors should compare switches to and from generics by specialty of the prescriber (neurologist, psychiatrist, primary care) and by first indication for which the medication was prescribed, stratifying by promotional dollars spent per indication and specialty.

Using Healthcare Claims Data for Outcomes Research and Pharmacoeconomic Analyses

AbstractHealthcare claims data are a practical complement to data from randomized 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 socio demographic 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.

Vascular Access–Related Infections: Definitions, Incidence Rates, and Risk Factors

Hemodialysis is associated with a high risk of morbidity and mortality, often caused by infections. Infections account for approximately 15% of all deaths in this patient population (the second leading cause after cardiovascular events) and for about one-fifth of admissions. Approximately one-fourth of infection-related admissions are caused by dialysis-associated peritonitis or vascular access infection that may lead to such significant complications as endocarditis or death. Published studies that assessed the determinants of hemodialysis-related vascular infections reported inconsistent findings. Variations in the definitions of infection among these studies despite the existence of standard guidelines proposed by at least 3 major work groups may explain, at least in part, these inconsistencies. A comprehensive in-depth review of those studies is needed to examine the inconsistencies in the published results. We first revised the existing vascular access-related infection definitions, then conducted a narrative review of the published literature that examined predictors of vascular access-related infections, highlighting the heterogeneity in methods and findings. Better understanding of the risk factors for vascular access-related infections may inform efficacious prevention strategies and lead to early detection of infections and improved patient care.

Gastrointestinal health care resource use and costs associated with nonsteroidal antiinflammatory drugs versus acetaminophen: Retrospective cohort study of an elderly population

OBJECTIVE To estimate gastrointestinal (GI) health care resource use and direct costs associated with prescription nonsteroidal antiinflammatory drugs (NSAIDs) in an elderly population. METHODS Using the Government of Quebecs health insurance database, we obtained the medical, pharmaceutical, and demographic records of 73,850 senior citizens who, between 1993 and 1997, had either an NSAID or an acetaminophen prescription dispensed. The date of their first dispensed prescription for an NSAID or acetaminophen was termed their index date. Patients who were not taking oral corticosteroids or anticoagulants at their index date, were not diagnosed with cancer at their index date, and were not hospitalized and did not have any GI events during the year prior to their index date were included in the study. Patients who had a dispensed NSAID prescription at their index date formed the NSAID cohort; the others formed the acetaminophen cohort. All patients were followed up for 2 years. The daily direct costs of GI events incurred during NSAID therapy by the NSAID cohort were compared with those incurred during a similar followup period by the acetaminophen cohort. The difference in these average daily costs was attributed to NSAID use. RESULTS The NSAID cohort included 5,268 senior citizens and the acetaminophen cohort 2,245. More GI adverse events were observed in the NSAID cohort (odds ratio 2.48, 95% confidence interval 2.06, 3.00). The average daily direct cost of GI events for a day of NSAID therapy attributed to the NSAIDs was