论文部分内容阅读
一、引言分析宫内及节育器避孕效果最简单的方法为指数方法,它假设因种种原因退出试验的终止率不随时间而改变,即为恒率。如:一个妇女自放器后在某个月脱落的机遇,不因月份变化而变化。当然,恒率的假设需要实际数据进行证实。为了避免这种恒率假设,Tietze 和potter 把生命表分析改变为按月计算。他们提出的方法便于估计宫内节育器使用一定时间后(一年或二年)各种终止原因的比例。为了比较不同宫内节育器的避孕效果,Pool 提出了对这些比例作大样本标准误估计计算。Tietze 提出了简化估计方法,且在实际应用中最为常用。Tietze-Potter-Pool 提出的生命表是蒋氏(蒋庆明)的精确生命表分析方法的近似。以上方法有关标准误是以大样本为基础的。这里的样本数是指各种原因的终止数,并非指参加试验的妇女数。因各种原因的终止数不足够大的话,可能产生明显的误差。长时间观察后,这种误差可能会消逝。另一个次要的问题是:这些方法在确定的时间区间内(通常为一个月)假设率不变,这需要由实际数据作进一步证实。
I. INTRODUCTION The simplest method of intrauterine and IUD contraception is the exponential method. It assumes that the termination rate due to various reasons does not change with time, that is, the constant rate. For example, a woman’s chance of falling off within a month after releasing the dispenser does not change according to the month. Of course, the assumption of constant rate needs to be confirmed by actual data. To avoid this constant assumption, Tietze and potter changed the life table analysis to monthly. Their method makes it easy to estimate the proportion of various causes of termination of a IUD after a certain period of time (one year or two years). In order to compare the contraceptive effectiveness of different IUDs, Pool proposed a large sample standard error estimate calculation for these ratios. Tietze proposed a simplified estimation method that is most commonly used in practice. The life table presented by Tietze-Potter-Pool is an approximation of Chiang’s (Chiang Hing-ming) exact life table analysis method. The standard error of the above method is based on large samples. The sample size here refers to the number of termination due to various reasons and does not refer to the number of women participating in the trial. Due to various reasons, the termination number is not large enough, then there may be significant errors. After a long period of observation, this error may disappear. Another secondary issue is that these methods do not change their hypothesis over a defined period of time (usually one month), which needs to be further substantiated by the actual data.