Successful NHMRC grants commencing in 2012 were announced in October. SSAI member Professor John Carlin and a team he leads have been granted $2.5M funding over five years for a Centre of Research Excellence named The Victorian Centre for Biostatistics (VICBiostat).
The following interview with John Carlin was conducted by Ian Gordon from the Statistical Society of Australia (SSAI) for the SSAI Newsletter.
"IG: First, congratulations: you must be very pleased with this outcome. How many person-days were required to prepare the application?
Thanks Ian, it was certainly very gratifying that the NHMRC saw fit to make this award, to a discipline that tends to be a bit of an invisible handmaiden to the researchers who frame the “real research” on the big issues of public health. As for the person-days, it’s a good question but I’m afraid the data collection was very disorganised so I can only make a very imprecise estimate... leaving aside the months if not years of pondering the general ideas writing the grant occupied me and my colleagues for the best part of a month and probably amounted to 1-2 weeks of person-time. I presume you are referring to recent research suggesting that the person-days required to prepare and assess grant applications might be better spent doing more research, with funds distributed by lottery. Having just won an award I couldn’t possibly agree with that suggestion of course.
IG: Tell us about the structure of the Centre and the groups involved.
The proposed Centre is a collaborative venture between biostatisticians at the Murdoch Children’s Research Institute (me and A/Prof Michael Coory), the Department of Epidemiology & Preventive Medicine at Monash University (Prof Andrew Forbes and A/Prof Rory Wolfe), and the Centre for Molecular Environmental, Genetic & Analytical Epidemiology (MEGA) at The University of Melbourne (A/Prof Lyle Gurrin and Dr Julie Simpson). We aim to create an academic centre for biostatistics research and training across our three locations. It is now widely recognised that the discipline of biostatistics is of critical importance to a broad range of health and medical research, as research in these areas focusses increasingly on studies of large cohorts of individuals followed over lengthy periods of time, data linkage with disease and healthcare registries, and the conduct of complex trials that aim to determine optimal treatment strategies. However, the discipline struggles to remain viable in Australia because of the absence of major centres of sufficient critical mass. Our proposal aims to address this structural problem by establishing a collaborative centre that will bring together many of the leaders of this field in Victoria.
IG: In broad terms, what do you hope the Centre will achieve? What will be different, in five years’ time, because of the Centre’s work?
Broadly speaking, the aim of the Centre is to foster the careers of a new generation of biostatistical leaders, by developing an integrated program of methodological and collaborative research that will be led largely by postdoctoral-level biostatisticians who will be supported by the grant. It is important that training in biostatistics include both methodological and collaborative work, and this typically extends the time that is needed to achieve a high level of proficiency and research output. The research program will be closely linked to training programs in biostatistical methods, including specialist PhD training as well as courses for epidemiologists and other health researchers. We expect the Centre will play an active role in the dissemination of sound statistical methods throughout the health research sector. Although the Centre is Victorian-based, building on existing strengths and collaborations, we hope its impact will be national, building on linkages with our colleagues in other states. In the end we would like to see the establishment of a critical mass of advanced trainees and enhanced collaborations on a sufficient scale and with sufficient visibility to underpin the ongoing development of the discipline. In our application we contrasted the situation in Australia with that in North America and the UK where substantial investment in biostatistics has been made over several decades by the national research funding agencies, resulting in the establishment of large academic departments, centres and national networks. To date in Australia, although many health researchers recognise the need there have been no vehicles for building the necessary high-level capacity.
IG: More specifically, I guess you have particular research themes that you and the team intend to pursue. Can you tell us what they are?
In our application we pointed to the substantial volume of academic research, published typically in journals of mathematical statistics, that develops biostatistical methods at a broad and often abstract level. We then outlined the concept of a “methodological and translational pipeline” which we see as guiding the work of the centre. This proposes that applied biostatistical research is required at multiple levels in order to enable the fruitful application of new statistical methods in health and medical research studies: (i) the methodology needs to be understood at a deep level in order for it to be tailored to suit practical application; (ii) it needs to assessed analytically or by numerical simulation in realistic practical scenarios; (iii) it needs to be translated into more accessible language and software to enable broad use by the practising applied biostatistician or data analyst working in government, industry or university departments; and (iv) awareness of the methodology needs to be promulgated widely to the broader research community in order to enable its acceptance and ultimate use in health/medical research studies that will inform appropriate health policy or practice. With this in mind, we proposed a program of research across eight areas within four broad themes:
- Analytical issues in cohort studies (missing data problems; longitudinal outcomes competing with loss to follow-up due to death; repeated exposure measures and causal questions).
- Modelling of multilevel structured (correlated) data (pharmacokinetic-pharmacodynamic modelling related to antimalarial treatment; regression models for binary outcomes from matched pairs)
- Development of new designs for intervention studies: cluster randomised crossover trials with application to intensive care research
- Treatment comparisons beyond head-to-head randomised trials (propensity score analyses in end-stage renal disease; methods for indirect comparisons).
No doubt the actual research will unfold in unexpected directions but we are very firm on the principle that the research should be both novel and useful. Being useful requires the work to be embedded within collaborative relationships with a range of epidemiologists and clinical researchers, while being novel is a little harder to define – following our “pipeline” metaphor it doesn’t always mean that new mathematical concepts need to be developed." top |
