Article

Almost daily, the media reports another health-related discovery. A study in a leading medical journal finds an association between certain genetic variants and increased risk of heart disease.  Another suggests rethinking how bone density testing is used to diagnose and treat osteoporosis.  A clinical trial shows no evidence that a new anti-platelet therapy reduces the incidence of death or serious health outcomes relative to an existing treatment in coronary artery disease patients. Three major studies suggest that exposure to air pollution may be implicated in stroke or cognitive decline. 

Behind all of these findings? Biostatistics – the science of development of statistical theory and methods for application to data-driven challenges in the health and biological sciences. The classical tools of biostatistics – for design and analysis of randomized clinical trials, the gold standard approach for comparing treatments; for design and analysis of laboratory studies; for analyzing studies that follow large cohorts of patients over time; for “meta-analysis,” combining results of several studies in a principled way – have for decades provided the basis for learning from data while taking account of the inherent uncertainty.  

And for decades, collaborations between biostatisticians and domain scientists have produced scores of now ubiquitous techniques. Type “Kaplan-Meier” into Google Scholar, and you’ll find over 200,000 hits in the medical literature. Paul Meier, an iconic clinical trials biostatistician, and his collaborator proposed a method for analyzing data in the form of a “time-to-an-event” – like death – that is “censored” (unobserved) for some subjects because they have not experienced the event by the end of the study or “drop out” and disappear, leaving their event times unrecorded. This one biostatistical advance, and extensions of it developed since, have aided countless biomedical breakthroughs. The landmark clinical trial assessing the benefit of AZT for reducing the risk of transmission of HIV from infected pregnant women to their infants was stopped six months early using these methods, altering the landscape of HIV prevention. 

But what of the future? “Big Data” are here, and the challenges they pose for biomedical and biological science could not be more complex or exciting.  For biostatisticians, they bring unprecedented opportunities to collaborate with the scientists generating the data to develop innovative new theory and methods to tackle problems never envisioned by the biostatisticians of yesterday.  A list of emerging trends is as long as the emerging scientific challenges; here’s just a sample.

Personalized medicine

If there is one overarching theme defining this new era, it is personalized medicine. Fulfilling the promise of “the right treatment for the right patient” will require combing vast data on huge numbers of patients to determine the (probably very small) subsets that inform credible predictions of disease risk or are critical for developing targeted treatments and selecting those leading to the best health outcomes.  

Deciphering messages embedded in massive, high throughput genomics and other 'omics data will be key, and biostatisticians will need to engage with genomic scientists and be sufficiently nimble to react with novel methods to avalanches of new types of data arising from rapid innovations in technology.  Biostatisticians will be developing better techniques to tease out rare genetic variants that may be implicated in disease from whole-genome sequence data and to integrate multiple types of data – single nucleotide polymorphisms, gene expression, epigenetic factors such as DNA methylation that are heritable but not found in sequence information – to isolate regions of the genome that differ between diseased and normal subjects, all while protecting against the threat of false discoveries that could lead to blind alleys and scientific missteps.    And putting the puzzle together will require methods to uncover networks and pathways of disease based on integration of genomic, proteomic, “exposiomic,” clinical, and phenotypic data.   Biostatisticians will also be devising new approaches to discovering and evaluating the reliability and reproducibility of biomarkers – biological features or measures that reflect presence or progression of disease or effects of treatment – from these data to be used for diagnosis and to guide treatment in practice. 

Indeed, development of principled methods for determining treatment over the entire course of a patient’s disease using his or her evolving information may be within reach.  Treatment of chronic diseases or disorders such as cancer or depression involves a series of decisions at milestones in the disease process. Biostatistical researchers are adapting approaches used in artificial intelligence to estimate, using complex patient data, sequences of “rules” for selecting treatments across all decision milestones that produce the best health outcomes.  These methods could someday be used with detailed information from electronic medical records and even mobile devices, as promoted in the emerging area of mHealth. 

Redesigning clinical trials

The quest for new drugs and tailored treatment strategies is inspiring a rethinking of the way clinical trials are conceived and conducted. With the vast majority of therapeutic agents in the development pipeline failing to win regulatory approval and mounting evidence of differential benefits and risks of treatments for subgroups of patients with certain profiles, the bedrock practice of including broad swaths of patients in trials comparing new experimental agents to the standard of care, may someday be supplemented or replaced. Studies focusing on subgroups, initially or after an interim period in which evidence is gathered to support such focus, are one possibility; adaptive trials, in which the trial design is modified as evidence accrues, are another.  Biostatisticians are pursuing these and other design and analysis innovations, which could bring promising new treatments to the patients who can benefit from them faster.