In an intriguing analysis in 1997, Franklin et al explored the Framingham data set, charting coronary heart disease risk according to levels of systolic and diastolic blood pressure. This analysis showed that at any level of systolic blood pressure, the CHD risk actually fell with increasing diastolic blood pressure levels. Thus a person with a blood pressure of 170/70 has around 30% higher risk than a person of similar age and characteristics with a blood pressure of 170/90. This is explained by the importance of pulse pressure. At any level of systolic blood pressure, an increased pulse pressure produces a log-linear increase in CHD risk.