The value of the number-needed-to-treat method in antiepileptic drug trials

Fenfluramine responder analyses and numbers needed to treat: Translating epilepsy trial data into clinical practice

OBJECTIVE

Clinical trials typically report antiepileptic drug efficacy by evaluating reduction in monthly convulsive seizure frequency (MCSF) through group response (active versus placebo). Although useful for regulatory purposes, population statistics do not easily translate into clinical practice, where treatment decisions are made on an individual-patient basis. Responder analyses help bridge this gap by showing proportions of patients who achieved various MCSF improvement levels. Deriving numbers needed to treat (NNTs) to achieve clinically desirable response levels can further inform individual decision-making. We calculated the NNT with fenfluramine to achieve "clinically meaningful" (≥50%) or "profound" (≥75%) MCSF reductions in patients with Dravet syndrome (DS).

METHODS

NNT to achieve ≥50% or ≥75% MCSF reduction was assessed using longitudinal data from two phase 3 studies for adjunctive fenfluramine in DS patients aged 2-18 years. NNT was calculated: 1/((Experimental-Responder Rate)-(Control-Responder Rate)).

RESULTS

In Study 1, NNTs to achieve ≥50% and ≥75% MCSF reduction were 1.8 and 2.1 at 0.7 mg/kg/day fenfluramine. In Study 2, these NNTs were 2.0 and 3.1, respectively. These results were seen as early as Weeks 6-7 and were sustained through Weeks 14-15.

INTERPRETATION

For every two to three patients with DS treated with fenfluramine in these trials, one patient achieved ≥50% or ≥75% MCSF reduction, respectively, compared with placebo (large treatment effect size; Cohen's d≈0.8). Responder analyses and NNTs can aid in clinical decision-making by offering clinically important information that is complementary to the population mean data on the chance of an individual patient achieving meaningful levels of MCSF improvement. (NCT02682927/NCT02826863, NCT02926898).

A perspective on cannabinoids for treating epilepsy: Do they really change the landscape?

With the licensing of cannabidiol for drug resistant seizures in Dravet and Lennox Gastaut syndromes in the United states in 2018, interest in the potential for cannabis-based-medicinal products to meet currently unmet needs for people with epilepsy continues to grow. This review summarizes current knowledge and discusses the implications for future research and practice. Both cannabidiol and tetrahydrocannabinol, the main components, have been extensively studied in animal models, with multimodal mechanisms of action proposed. Only pure cannabidiol formulations have been rigorously evaluated in controlled trials thus far, with modest but significant improvements in motor seizures. Adverse effects include diarrhoea, somnolence and reduced appetite, with mostly acceptable tolerability, but a not insignificant (up to 1 in 23) risk of serious adverse events. Recognized drug interactions include with valproate (increased risk of hepatotoxicity) and clobazam (contributing to somnolence, increased secretions, probably chest infections, and potentially efficacy). Whilst there is public (and producer) interest in products also containing tetrahydrocannabinol, clinicians have justifiable concerns about exposing a group already vulnerable to mental health and neurobehavioural comorbidities to the associated additional risks in these domains. Artisanal preparations, with often inconsistent/unknown constituents are frequently used but not recommended. A gulf exists between the actual evidence, including a lack of comparative studies and public beliefs, fuelled by media and anecdote. Continued education of the public, policymakers, researchers and healthcare providers about what is and isn't yet known, together with on-going good quality research is essential to mitigate against future potential risks, particularly in relation to vulnerable populations. This article is part of the special issue entitled 'New Epilepsy Therapies for the 21st Century - From Antiseizure Drugs to Prevention, Modification and Cure of Epilepsy'.

Designing noninferiority tuberculosis treatment trials: Identifying practical advantages for drug regimens with acceptable effectiveness

In this Collection Review for the Novel Treatments for Tuberculosis Collection, Piero Olliaro and Michael Vaillant discuss the considerations when choosing a non-inferiority margin that is meaningful from statistical, ethical, clinical, and health standpoint.

The number needed to treat in pairwise and network meta-analysis and its graphical representation

OBJECTIVE

The objective of this study was to present ways to graphically represent a number needed to treat (NNT) in (network) meta-analysis (NMA).

STUDY DESIGN AND SETTING

A barrier to using NNT in NMA when an odds ratio (OR) or risk ratio (RR) is used is the determination of a single control event rate (CER). We discuss approaches to calculate a CER, and illustrate six graphical methods for NNT from NMA. We illustrate the graphical approaches using an NMA of cognitive enhancers for Alzheimer's dementia.

RESULTS

The NNT calculation using a relative effect measure, such as OR and RR, requires a CER value, but different CERs, including mean CER across studies, pooled CER in meta-analysis, and expert opinion-based CER may result in different NNTs. An NNT from NMA can be presented in a bar plot, Cates plot, or forest plot for a single outcome, and a bubble plot, scatterplot, or rank-heat plot for ≥2 outcomes. Each plot is associated with different properties and can serve different needs.

CONCLUSION

Caution is needed in NNT interpretation, as considerations such as selection of effect size and CER, and CER assumption across multiple comparisons, may impact NNT and decision-making. The proposed graphs are helpful to interpret NNTs calculated from (network) meta-analyses.

Iatrogenic ulnar nerve injury after the surgical treatment of displaced supracondylar fractures of the humerus: number needed to harm, a systematic review

BACKGROUND

Supracondylar fractures of the humerus are common pediatric elbow injuries. Most displaced or angulated fractures are treated by closed reduction and percutaneous pinning, with either a crossed pin or lateral pin configuration. The purpose of this study was to conduct a systematic review to determine if there is an increased risk of iatrogenic nerve injury associated with the crossed pin configuration.

METHODS

Relevant articles were identified by searching electronic databases and hand searching-related journal and conference proceedings. Within each trial, the risk of iatrogenic ulnar nerve injury was calculated for each pinning technique. For studies comparing crossed versus lateral pinning, the resulting trial-based differences in risk estimates were pooled using a random effects meta-analysis. A number needed to harm was determined using the pooled risk difference.

RESULTS

Thirty-two trials consisting of 2639 patients were used in the pooled analysis. The pooled risk difference of iatrogenic ulnar nerve injury is 0.035 (95% confidence interval, 0.014-0.056), with a higher incidence of injury in the crossed pinning group. The weighed number needed to harm for the crossed pinning is 28 (95% confidence interval, 17-71).

CONCLUSIONS

The results of this review suggest that there is an iatrogenic ulnar nerve injury for every 28 patients treated with the crossed pinning compared with the lateral pinning. Further research is necessary to ensure that the optimal pinning technique is chosen to treat these factors.

LEVEL OF EVIDENCE

Level III.

The Application of Number Needed to Treat (NNT) to Clinical Problems in Radiotherapy

The goals of this report are: 1) to review the number needed to treat (NNT) concept, which, although well established in many sectors of medicine, is still relatively new to the radiotherapy community; 2) to discuss several clinical radiotherapy examples illustrating the inherent advantages of the NNT approach; and 3) to discuss potential future roles of the NNT concept within radiotherapy.

Approaching clinical problems in prostate cancer radiotherapy using the number needed to treat (NNT) technique

The goals of this article are to review the application of the number needed to treat (NNT) concept to selected clinical problems in prostate cancer radiotherapy. Particular emphasis will be placed on (1) comparison of radiotherapy with other treatment options for early-stage disease, (2) the role of hormone therapy in addition to radiotherapy over a spectrum of disease presentation, and (3) systematic comparison of adjuvant versus salvage radiotherapy in the post-prostatectomy setting. Limitations of NNT calculations based on non-randomized comparisons also are discussed.

Measuring the efficacy of antiepileptic drugs

Clinical trials of new antiepileptic drugs (AEDs) include regulatory studies aimed at demonstrating efficacy and reasonable safety, post-marketing open-open label studies and longer term outcome studies. Regulatory trials involve a carefully selected population of patients and are conducted under rigorously standardised conditions. Data from such studies cannot often be translated into clinical practice. Pragmatic post-marketing studies using flexible dosing schedules allow clinicians to better judge the utility of the new drug in a wider population of patients with epilepsy and decide the most appropriate dosing schedules. This paper discusses some of the issues surrounding the measurement of efficacy of new AEDs in both pre- and post-marketing phases of their development. All of the newer AEDs are initially used in patients with refractory partial seizures as adjunctive treatment. These trials are generally parallel-group studies although cross-over designs have been employed. The use of placebo-control is uncontroversial in this type of study. Efficacy endpoints are generally manipulations of seizure frequency on study drug compared to control. Global outcome measures and health related quality of life scores can also be used to measure efficacy. As the standard AEDs are associated with a high rate of seizure remission in patients who receive them as monotherapy, demonstration of superior efficacy of a new agent in a comparative trial will require large numbers of patients in a design that takes into account the natural history of treated epilepsy. Comparing investigational agents to a standard AED in an 'active-control' study with demonstration of equivalent efficacy would seem to be an acceptable way of assessing efficacy of new AEDs in this population. Some regulators, however, do not accept equivalence as proof of efficacy and insist on demonstration of superiority compared to a control. The use of placebo alone in the control group is ethically dubious. Several innovative study designs have, therefore, been used to satisfy regulatory requirements, while maintaining patient safety including withdrawal to monotherapy using high versus low dose comparators. Observational outcome studies provide the best opportunity of exploring the long-term utility of individual AEDs. Such studies largely follow standard clinical practice and need considerable time and resources. They can, however, yield valuable information about the effectiveness of AEDs in everyday clinical practice. Data from regulatory trials should be complemented by postmarketing studies and longer term studies of outcome to help clinicians decide the best way of utilising new AEDs and establishing their role in the therapeutic armamentarium.

What can we learn from clinical trials of anticonvulsant drugs in epilepsy?

Any physician who intends to utilize the available antiepileptic drugs (AEDs) judiciously, cannot do so without being well versed on their pharmacological properties and the large body of evidence that is continuously accumulating on their relative efficacy and tolerability in different types of epilepsy. While informal observations such as retrospective surveys and case reports can be useful under special circumstances, prospective randomized clinical studies represent by far the most important tool by which objective information can be obtained about the clinical value of existing drugs. Even randomized trials, however, can produce misleading conclusions because of inherent weaknesses or bias in study design, analysis, and interpretation. Common deficiencies identified in some of the most recent drug trials in epilepsy include 1) inclusion of inappropriately heterogeneous patient groups (for example, patients with partial and primarily generalized seizures); 2) low statistical power due to insufficient sample size (for trials designed to show therapeutic equivalence); 3) inappropriate titration rates or suboptimal dosages or dosing schedules (often favouring the sponsor's product over the comparator); 4) insufficient duration of treatment; and 5) utilization of endpoints of questionable clinical significance. In part, some of the above shortcomings can be ascribed to the fact that most clinical drug trials are designed to address regulatory needs rather than to provide the type of information required for rational prescribing. Physicians need to be alerted about the importance of these issues, and they should make every possible effort to interpret critically the medical literature on which they rely to guide and support their therapeutic decisions.

Monotherapy trials: endpoints

Monotherapy antiepileptic drug (AED) trials can optimally provide information concerning efficacy and tolerability of one drug compared with placebo, as well as to different doses or rates of administration. Commonly, a drug is compared with one or more other drugs. The outcome measures will be dictated by the questions being asked. In most comparative studies, the single overall result that best defines success or failure is time of continuation on drug as expressed in a life table. Discontinuation before planned completion of the study implies insufficient efficacy or unacceptable adverse effects. A statistically significant difference between treatments provides important support for recommending the drug or dose with the best outcome. The criteria for continuation/discontinuation are defined in the design based on the expected outcome. The outcome of primary importance is efficacy in prevention of seizures or a decrease in severity. Complete control for the duration of the study is the ultimate goal but in some populations may not be possible. The number of subjects entering remission gives further information about long term outcome. Time to first (nth) seizure provides similar evidence of efficacy. Seizure rates allow comparisons of subjects retained for different lengths of time in the trial. Differences in seizure severity may be of clinical importance and multiple efforts have been made to develop instruments to accurately measure this outcome. Adverse effects of the drugs are the second major outcome. These can be expressed as incidence and/or prevalence. The presence and frequency of side effects will depend on how the study is designed and whether these effects are specifically sought by the investigators. Serious systemic safety outcomes are monitored, but the relative infrequency of occurrence and number of subjects in the trials usually do not provide enough power to detect statistically significant differences except for rash. Tolerability is more easily documented but is difficult to access accurately in the absence of placebo controls. Frequency, severity and persistence are measurable. Specific unwanted types of drug effects can be specifically studied using detailed neuropsychological test batteries. Some information concerning pharmacokinetic properties may be obtained but are better assessed in other types of trials. A final important outcome is the effect of drug therapy on quality of life. Although a favorable finding in this outcome is most desirable, the measures used are much less precise than those for efficacy and adverse effects.