A review of methods used in assessing non-serious adverse drug events in observational studies among type 2 diabetes mellitus patients.

Clinical drug trials are often conducted in selective patient populations, with relatively small numbers of patients, and a short duration of follow-up. Observational studies are therefore important for collecting additional information on adverse drug events (ADEs). Currently, there is no guidance regarding the methodology for measuring ADEs in such studies. Our aim was to evaluate whether the methodology used to assess non-serious ADEs in observational studies is adequate for detecting these ADEs, and for addressing limitations from clinical trials in patients with type 2 diabetes mellitus. We systematically searched MEDLINE and EMBASE for observational studies reporting non-serious ADEs (1999-2008). Methods to assess ADEs were classified as: 1) medical record review; 2) surveillance by health care professionals (HCP); 3) patient survey; 4) administrative data; 5) laboratory/clinical values; 6) not specified. We compared the range of ADEs identified, number and selection of patients included, and duration of follow-up. Out of 10,125 publications, 68 studies met our inclusion criteria. The most common methods were based on laboratory/clinical values (n = 25) and medical record review (n = 18). Solicited surveillance by HCP (n = 17) revealed the largest diversity of ADEs. Patient surveys (n = 15) focused mostly on hypoglycaemia and gastrointestinal ADEs, laboratory values based studies on hepatic and metabolic ADEs, and administrative database studies (n = 5) on cardiovascular ADEs. Four studies presented ADEs that were identified with the use of more than one method. The patient population was restricted to a lower risk population in 19% of the studies. Less than one third of the studies exceeded pre-approval regulatory requirements for sample size and duration of follow-up. We conclude that the current assessment of ADEs is hampered by the choice of methods. Many observational studies rely on methods that are inadequate for identifying all possible ADEs. Patient-reported outcomes and combinations of methods are underutilized. Furthermore, while observational studies often include unselective patient populations, many do not adequately address other limitations of pre-approval trials. This implies that these studies will not provide sufficient information about ADEs to clinicians and patients. Better protocols are needed on how to assess adverse drug events not only in clinical trials but also in observational studies.

Comparing adverse event rates of oral blood glucose-lowering drugs reported by patients and healthcare providers: a post-hoc analysis of observational studies published between 1999 and 2011.

BACKGROUND: Non-serious symptomatic adverse drug events (ADEs) affect the real benefit-risk ratio of a drug. Currently, such ADEs are quantified in different ways, often using reports from a healthcare provider or patients, resulting in large variations in estimated rates. Several studies showed that patients report bothersome or symptomatic ADEs more frequently than providers, but no comparisons to an external reference or gold standard have been made. OBJECTIVE: We conducted a literature review to assess the agreement and concurrent validity of healthcare provider- and patient-oriented methods for quantifying symptomatic ADEs of oral blood glucose-lowering drugs in patients with type 2 diabetes mellitus. METHODS: We systematically searched MEDLINE and EMBASE databases for observational studies reporting on rates of ADEs in patients treated for type 2 diabetes that were published between 1999 and 2011. We included nine observational studies reporting absolute rates of symptomatic ADEs in patients receiving monotherapy. We calculated 95% confidence intervals and assessed agreement between rates observed with different methods. We assessed concurrent validity using the range noted in the Summary of Product Characteristics (SPC) as the gold standard. RESULTS: A comparison of rates reported by patients and providers was only possible using three studies of metformin that assessed mainly gastrointestinal (GI) ADEs. Provider-oriented methods by means of medical record review gave lower rates for abdominal pain (0.6-3.7%), dyspepsia (1.3-2.8%) and constipation (0.6-1.0%) than a patient questionnaire method (8.5%, 11.9% and 20.7%, respectively). For diarrhoea, the patient-reported rate (5.2%) was in agreement with the provider-based rates (1.6-7.6%). The majority of the rates reported by providers and patients were not corresponding with the ranges in the SPC. For GI ADEs the rates were all lower, whereas for lactic acidosis and hypoglycaemia the rates were higher. CONCLUSION: Although it has repeatedly been proposed that patients' reports on safety should be incorporated with providers' reports, especially for symptomatic ADEs, the number of observational studies using patient-oriented methods for assessing ADEs other than hypoglycaemia are limited. Provider-based measurement tended to underestimate symptomatic ADEs. Patient-oriented methods seemed to give ADE rates that were closer to the rates reported in the SPC.