Delivering clinically appropriate, data-driven results is the underlying principle of all projects undertaken by the Statistics Team at Costello Medical. We are passionate about our multidisciplinary project teams who provide a unique combination of rigorous clinical understanding and comprehensive statistical knowledge, enabling us to deliver the most meaningful answers to our clients’ questions, and seamlessly incorporate them into many different deliverables.
Our team successfully delivers a diverse range of projects and particularly enjoys supporting clients on challenging analyses that require creative approaches to succeed. We appreciate the importance of staying at the forefront of the rapidly evolving field of medical statistics, and our team are trained in a variety of advanced statistical methods.
Our dedicated team routinely delivers projects in the following areas:
- Working with aggregate and patient-level data in comparative research: meta-analysis, network meta-analysis (NMA) and indirect treatment comparison (ITC), including the use of matching-adjusted indirect comparison (MAIC) and simulated treatment comparison (STC) techniques
- Analysis of randomised and observational clinical trial individual patient data (IPD) and real-world evidence: exploratory data analysis, multivariate linear, non-linear and logistic regression, multicollinearity analyses and estimation of transition probabilities
- Undertaking survival analyses: Cox regression, piecewise analyses, proportional hazards testing and IPD simulation
What Makes Statistics at Costello Medical Different?
Combination of clinical and statistical knowledge. Our extensive experience of working with clinical studies allows us to critically evaluate the research that feeds into our statistical analysis. The scientific skills possessed by the multidisciplinary Statistics Team also ensure that the results and clinical implications of the statistical analyses are interpreted and communicated correctly and persuasively for the purposes of the project.
Technical expertise. We have worked on a wide variety of projects and are committed to expanding our technical expertise; members of our Statistics Team have attended formal training given by leaders in data analysis techniques and obtained higher degrees and certificates in statistics. Our team also undertakes novel research and has expertise in a variety of statistical software, such as SAS, R, Stata and WinBUGS.
Reasonable budget. Due to the careful scoping of projects and the clinical knowledge of our multidisciplinary Statistics Team, we identify the most appropriate methods to perform effective and efficient statistical analysis from the project outset. We therefore believe that we can offer a high quality statistical analysis service that will fit your budget, and we set out our fees in a clear and transparent way.
Cross-company collaboration. Costello Medical’s Statistics Team works closely with other specialised teams within the company, meaning that any subsequent stages of a project (such as HTA submissions, value materials or publications) benefit from the involvement of dedicated specialists as well as the supervision and oversight of the original project manager and statistician throughout.
To request a quote, or for more information on Costello Medical’s diverse statistics services, please contact Lucy Eddowes.
In order to perform a NMA to inform reimbursement discussions across Europe, we first completed a systematic review, which identified more than 30 studies for assessment for inclusion in the evidence synthesis. We followed a detailed feasibility assessment process to decide which studies should be included and to devise 12 different networks containing up to 18 studies in each. Each step was recorded and justified carefully, which was particularly important in a NMA that will be critiqued by reimbursement agencies. We analysed binary and continuous variables, and the analyses were conducted in a Bayesian framework.
Due to the class-based nature of the treatments included in the NMA, in addition to the standard NMA we also conducted hierarchical NMA, which groups treatments by drug class to allow more accurate modelling of the treatment-level effects and produces pairwise comparisons between drug classes. We also investigated the variation in study outcomes (heterogeneity) and inconsistency across trials both statistically (by fitting models such as unrelated mean effects models) and through qualitative assessment of the baseline characteristics and study designs.
We approached this project in a systematic and meticulous way, and ensured that each stage of data extraction and model coding was checked by a second Analyst. The final result was a robust assimilation of a great volume of data. From this, we provided clinically valid interpretation of the results to support our client’s reimbursement applications across Europe.
We performed a full statistical analysis of a prospective, observational study that enrolled over 800 patients, supporting the study from protocol development to publication. The main objectives were to explore the factors associated with the eligibility and willingness of the patient population to receive currently available treatments, in order to help position a new product in the market.
We developed the study protocol, designed the surveys and distributed the surveys to over 10 study centres. During the study, we monitored the whole data collection process and reviewed and assessed the data for quality and adherence to the protocol. Upon receipt of the survey responses, we carefully performed data entry and cleaning. The results were summarised with descriptive statistics and multivariate analysis using binary logistic regression was conducted to explore which demographic and clinical characteristics significantly affected patient eligibility and willingness to receive treatment. Further analyses are ongoing to identify confounders and assess multicollinearity, and to investigate the impact of potential drug-drug interactions.
Capitalising on our expertise in developing publications, this study and the analysis has been presented at a conference as a poster and is being written into a manuscript suitable for submission to an international peer-reviewed journal.
We undertook statistical analysis of a small, single-arm observational study that enrolled patients with a rare genetic disorder. The aim was to determine if treatment with a specific pharmacological agent slowed disease progression based on the rate of change in several clinical rating scales.
Our feasibility assessment highlighted a number of potential issues, such as limited data availability and the potential insensitivity of some clinical rating scale subcomponents to disease progression. A variety of different statistical methods were therefore used to explore whether the drug was effective at slowing disease progression; standard linear regression was used to investigate whether the duration of treatment had an effect on the clinical rating scores, and an exploratory analysis used hierarchical clustering methods to determine if there was a noticeable difference between the rates of clinical score change before and after treatment. The results of the analysis indicated that treatment with the drug does slow disease progression, adding to the body of observational evidence in support of offering this currently unlicensed treatment to patients. These results have been accepted as a poster presentation at a rare disease statistical analysis methodology conference, and are being prepared for publication in a peer-reviewed journal.