Abstract: Objective To analyze dialysis duration-related data registered by 109 HD (hemodialysis) units in Beijing in 2007 and 2008, understand distributing rules of dialysis duration, and provide new initial method and idea for HD quality evaluation. Methods Base on 5065 and 6506 cases whose dialysis duration were registered in these two years, the frequency distributing diagrams of prevalent HD patients’ dialysis duration were drawn and equations describing the characteristics of these diagrams were established. Then, correlation analysis were done among prevalent HD patients of the two years and died ones of 2008. Results Both of the frequency distributing diagrams of the two years had characteristics of exponential curves. Dialysis duration had good linear correlation with logarithm of proportion of corresponding cases’ frequency, and could be fitted by regression equations as lnCn=-1.54-0.30n and lnCn=-1.45-0.30n (n≥1). Through generalizing these equations, the unified equations as Cn=m×Ra-1 (n=0) or Cn=m×Rcn-1 (n≥1) could be used to describe distributions of different years (In which, m=[1/Ra+(1-Rcmax)/(1-Rc)]-1). The measured data of the two years matched very well with theoretical frequencies described by equations (χ2=5.63 and 11.61, P=0.58 and 0.11). If correlation analysis was done among data of prevalent HD patients in 2007 and 2008, and died ones in 2008, it could be found that correlation coefficients r of three groups were 0.99, 0.95 and 0.97 (P<0.01). Conclusion Distribution of dialysis duration can be described by equations. The coefficients of equations, such as Ra (acute decline rate, which is patients’ decline rate during the first year) and Rc (chronic decline rate, which is patients’ average decline rate during the years after the first year) prompt relative withdrawal speed of HD groups, so they might be new indicators to measure treatment quality of HD unit. Furthermore, distributions of dialysis duration have correlation between years of 2007 and 2008, or prevalent and died patients. So, a clue is provided for using recent data of prevalent HD patients to speculate corresponding data of prevalent and died HD ones of next year.

Key words: uremia, renal dialysis, time factors, epidemiology