Pharmacoepidemiology and rheumatic disorders

Active-comparator design and new-user design in observational studies

Over the past decade, an increasing number of observational studies have examined the effectiveness or safety of treatments for rheumatoid arthritis. Unlike randomized controlled trials (RCTs), however, observational studies of drug effects have methodological limitations such as confounding by indication. Active-comparator designs and new-user designs can help mitigate such biases in observational studies and improve the validity of their findings by making them more closely approximate RCTs. In an active-comparator study, the drug of interest is compared with another agent commonly used for the same indication, rather than with no treatment (a 'non-user' group). This principle helps to ensure that treatment groups have similar treatment indications, attenuating both measured and unmeasured differences in patient characteristics. The new-user study includes a cohort of patients from the time of treatment initiation, enabling assessment of patients' pretreatment characteristics and capture of all events occurring during follow-up. These two principles should be considered when designing or reviewing observational studies of drug effects.

Tocilizumab in rheumatoid arthritis: a case study of safety evaluations of a large postmarketing data set from multiple data sources

OBJECTIVES

To evaluate the magnitude of serious adverse events (SAEs) observed in postmarketing reports of tocilizumab (TCZ) for rheumatoid arthritis (RA) in relation to SAEs observed in TCZ clinical trials and external epidemiology data.

METHODS

A total of 64,000 patient-years (PY) of TCZ exposure was needed to determine, with 90% power, whether rates of SAEs of interest (eg, death, hepatic, gastrointestinal, and cardiovascular) were ≥50% higher (agreed with the Food and Drug Administration) than expected. Reporting rates were calculated for spontaneously reported SAEs, open-label or unblinded postmarketing clinical trials (phase 3b/4), and a Japanese postmarketing surveillance program in the global postmarketing safety database. Event rates were calculated for the registrational placebo-controlled trials and long-term extension data. External comparators for anti-tumor necrosis factor (aTNF)-treated RA patients were derived from a US-based health care insurance claims database or published literature.

RESULTS

The global postmarketing safety database provided 65,099 PY of TCZ exposure; the aTNF external comparator population provided 53,360 PY. Spontaneous reporting rates per 100 PY (95% confidence interval) were 8.3 (8.1, 8.5) SAEs, 0.39 (0.34, 0.44) deaths, 0.06 (0.04, 0.08) serious hepatic events, 0.15 (0.12, 0.18) serious gastrointestinal events, 0.09 (0.07, 0.12) serious myocardial infarctions, 0.15 (0.12, 0.18) serious strokes, and 0.07 (0.05, 0.09) cardiac deaths in the global postmarketing safety database. These were of similar magnitude to corresponding rates from registrational clinical trials, the aTNF external comparator population, and published literature.

CONCLUSIONS

SAE rates observed among postmarketing TCZ users were similar to those of various comparison populations. Predetermined design of studies to compare postmarketing AEs using multiple data sources is a useful strategy that can be applied to other medications.

A systematic review on the application of pharmacoepidemiology in assessing prescription drug-related adverse events in pediatrics

OBJECTIVES

To conduct a systematic review of recent case-control and cohort studies in assessing adverse drug events (ADEs) among pediatric patients aged between 0 and 18 years; to establish strengths and limitations of pharmacoepidemiology when applied in evaluating pediatric drug safety; and to identify areas of pediatric drug safety that may be suitable for future pharmacoepidemiological investigations.

METHODS

A PubMed search was conducted using a list of keywords representing drugs, adverse drug events, case control and cohort studies, and pediatric population. Selection criteria were used to extract relevant studies published from 1/1 2000 to 7/1 2006.

RESULTS

Twenty-seven studies met our criteria. Of them, 12 used a prospective cohort design. The others were either case-control (seven) or retrospective cohort (eight) studies, based on the analysis of existing databases. More than half of the studies included more than 500 subjects. Inclusion and exclusion criteria for participants in these studies were generally not very stringent. The subjects closely resembled patients in the real-world settings. The length of study follow-up ranged from 1 day to 40 years. In eight studies, the length of follow-up surpassed 5 years. Potential confounding factors were taken into consideration in all of the studies. But the database-based studies generally failed to control for some important clinical variables such as disease severity. Misclassification of drug exposures also occurred in some of these studies.

CONCLUSIONS

Despite some limitations, pharmacoepidemiology proves to be useful for assessing ADEs in pediatrics. With appropriate study design, this methodology can bolster our understanding about the safety of pediatric drug use. Several areas of pediatric drug safety may especially be suitable for future pharmacoepidemiological investigations. These areas include the safety of polypharmacy, long-term drug effects, and off-label drug use.