THE battle against antibiotic resistance is being hampered by the grossly inadequate analytical techniques used by some medical researchers.
After reviewing research on antibiotic resistance from 1980 onwards, a team led by Peter Davey at the University of Dundee has concluded that many researchers fail to meet even the minimum standards required to make their research useful. 鈥淪eventy per cent of what is published is a waste of time,鈥� says Davey.
To test whether medical intervention or a new way of dealing with antibiotic resistance lowers the rate of infection or antibiotic resistance, researchers plot their results on a graph. But before they begin they need to collect a minimum of three previous data points, so that they have a baseline against which they can measure the effectiveness of their intervention.
Advertisement
But when Davey鈥檚 team reviewed the papers, they found that only 26 per cent met these minimum criteria. 鈥淔or those who don鈥檛 use proper methods we don鈥檛 really know what their interventions have accomplished,鈥� says Davey.
Other speakers highlighted further problems in the way epidemiologists analyse their data 鈥� assuming they collect enough data in the first place. While researchers track which bacterial infections resist treatment with certain antibiotics, their analysis is relatively simplistic and does not make the most of the information they can extract from it, says Dominique Monnet from Denmark鈥檚 Statens Serum Institute.
For example, many researchers plot the rate of antibiotic resistance against time in terms of years, which gives a steadily increasing line. But when researchers plot the same data in terms of months instead of years, the line spikes and dips like a roller coaster. While both plots accurately reflect reality, plotting monthly data is vastly more informative and highlights the seasonal fluctuations that affect both infection and resistance rates.
This type of graph, with data plotted at short time intervals over a long study period, is called a time series plot and is commonly used in economics to monitor stock prices. Monnet urged researchers to borrow further from economists, particularly a method called 鈥渢ime series analysis鈥�, which takes into account the fact that the antibiotic resistance rate one month will influence the rate in the next month, rather than treating it as an independent event 鈥� as traditional epidemiological methods would.
Using this technique, Monnet analysed several years data to predict what would happen to antibiotic resistance rates over the next six months. His predictions were a close match to reality.
