Felix Reed-Tsochas explains why his research on complex networks can have an impact far beyond the business sphere
Mention ‘networks’ in a business school context and most people tend to think of swapping business cards and making career-furthering personal connections.
These sorts of interactions do create a social network that, when analysed, can yield interesting and useful information about marketing and engagement activities, career and work trends, and so on – all very relevant to the business world. But it is merely the tip of the iceberg in the study of complex networks, which has relatively recently and rapidly become a major topic of interdisciplinary research.
Complex networks are everywhere: there are biological and ecological networks, social and economic networks, transport networks, supply networks, and computer and communication networks. They are characterised by multiple interactions that are neither totally regular nor totally random. And by studying them we can improve our understanding of, for example, why traffic becomes congested; where faults in factory assembly lines may occur; or ways in which networks of financial firms operate – and sometimes fail.
Those are fairly obvious business-focused cases. But increasingly we are seeing how the sort of complex network analysis we do in a business school can intersect with completely different areas of operation and research. By working at these intersections we can shed sometimes startling new light on a range of human problems.
A recent example of this is a research project led by one of my doctoral students, National Institutes of Health Oxford-Cambridge Scholars Program Fellow Jeff Lienert, who is co-supervised by Dr Laura Koehly at the Social Behavioral Research Branch of the National Human Genome Research Institute (NHGRI) in the USA. Together with Dr Christopher Marcum, also from the NHGRI, we investigated how cancer patients undergoing chemotherapy can be affected by their social interaction with other patients during treatment.
Hospitals are prime examples of complex social systems that we can represent and analyse as networks. Although we might think of a normal outpatient appointment, for example, as a simple meeting between patient and clinician, each visit also comprises a number of other formal and informal interactions, such as with the receptionist, other patients in the waiting room, and possibly other health professionals or technicians. We can use a range of innovative statistical analyses of such interactions to reveal patterns and meaningful insights.
Our study was based on data provided by Jon Finney at the Nuffield Department of Medicine, University of Oxford, that covered the administrative data and electronic medical records between 2000 and 2009 of 4691 cancer patients undergoing chemotherapy. Chemotherapy wards are generally open, with treatment chairs or beds; and, because the treatment must be delivered according to a strict timetable, patients often overlap with the same group of people when they are in the ward. So while they probably do not know each other at the start of their treatment, over the course of several months they may become more familiar with each other, whether through direct interaction or observation.
We used the data to construct a network of patients and their ‘co-presence’ in the chemotherapy ward, which we then compared with their five-year survival rate. The five-year survival rate is the percentage of people who live at least five years after chemotherapy treatment is completed. For example, a five-year survival rate of 70 percent means that an estimated 70 out of 100 people are still alive five years after chemotherapy.
Our results suggest that co-presence matters. If patients consistently received treatment with the same set of other patients, this improved their five year survival rates, compared to circumstances where their treatment was delivered in the presence of a frequently changing set of other patients. Additionally, patients were a little more likely to survive for five years or more after chemotherapy if they repeatedly shared the same space during chemotherapy with other patients who also survived for five years or more. However, patients were a little more likely to die in less than five years after chemotherapy when they repeatedly shared the same space during chemotherapy with those who died in less than five years. When patients were around those during chemotherapy who died in less than five years following chemotherapy, they had an estimated 72 percent chance of dying within five years following their chemotherapy. The best outcome was when patients interacted with someone who survived for five years or longer: their chance of dying within five years was reduced to an estimated 68 percent.
As with many other types of research, the enormous sensitivity surrounding this project lies in the difference between broad trends and individual practice, as well as in questions of direct applicability. To people who have not directly encountered cancer, either in themselves or people close to them, these findings are interesting – exciting, even, as they suggest other avenues of research that could further improve cancer survival rates overall.
To anyone with experience of cancer, though – particularly if they have lost someone to the disease – it immediately raises painful and personal questions. How do you know which groups are going to have the better survival rate? Can you choose who you get to have chemotherapy with? And, most agonisingly, could a different chemotherapy schedule have helped a loved one to live longer? It is one of the reasons that researchers in these sorts of areas have to be so careful about how they communicate their findings. In the media’s typically punchy style, a small rise in risk, for example, only too easily becomes interpreted as direct causation and we all stop eating burnt toast.
These implications of our research afford a rare opportunity for business school scholars to translate research into practice through careful collaboration with hospital administration. Now we know that there is a significant social influence exerted by patients who are co-present in the chemotherapy ward, further research can be directed towards understanding why. If the observed social influence operates via changes to stress, for example, then reducing patient stress in the ward without changing patient scheduling may have a positive effect on all patients. Or, given that our analysis was based on co-presence only, with no investigation of the quality or type of interactions the patients experienced, it may be possible to experiment with encouraging more direct social support while in the chemotherapy ward. Even without further research, our findings call into question recent moves in some hospitals to replace communal treatment spaces with private chemotherapy rooms.
This is what we mean at Oxford Saïd when we talk about the role of business in tackling world-scale problems. It is not that individual businesses should be expected to address challenges such as climate change or healthcare on their own, or that for-profit organisations and a competitive market are the best possible models for all sectors. But there are skills and resources that have been developed by or for business that can be applied to great effect in other sectors – as, indeed, business can learn from other sectors in its turn. Openness and collaboration are the key.