Omboni S, Parati G, Zanchetti A, Mancia G
Cattedra di Medicina Interna, Ospedale S. Gerardo, Monza, Università di Milano, Italy.
J Hypertens. 1995 Oct;13(10):1105-12. doi: 10.1097/00004872-199510000-00005.
To address several methodological questions related to calculation of trough:peak ratio from 24 h ambulatory blood pressure (BP) recordings.
Data from patients with mild essential hypertension who were included in parallel group (n = 280) or cross-over studies (n = 39) were pooled. 24 h ambulatory BP recordings were available after 2- to 4-week washout from treatment and at the end of a 4- to 8-week period of treatment with calcium antagonists (n = 143), angiotensin converting enzyme inhibitors (n = 103) or placebo (73 patients from parallel group studies and 39 from a cross-over study). Each recording started between 0900 and 1000 h, immediately after the drug or placebo intake during the treatment phase. BP was measured at 15 min intervals during the day and at 15-20 min intervals during the night. Peak changes were calculated from systolic BP and diastolic BP 2-8 h after drug intake, and trough changes from readings taken during the last 4 h of the 24 h.
Peak changes induced by drug treatment were progressively reduced when data were averaged over 1, 2, 4 and 6 h. BP reproducibility showed a concomitant increase and the best compromise between correct estimate of peak changes and reproducibility was the average of the adjacent 2 h with the maximal BP fall. Peak and trough (average of last 2 h) changes showed a normal distribution, whereas trough:peak ratios showed non-normal distributions, large scatters and many individual values with no pharmacodynamic significance (namely, much above unity and below zero). Selecting responders to treatment reduced the dispersion and made the trough:peak ratio distribution normal. There was no correlation between trough:peak ratios and changes in BP variability (standard deviation of 24 h mean) induced by treatment. Placebo administration caused no trough but a modest peak fall. Peak changes during placebo also showed a wide scatter and a non-normal distribution, which makes correction with respect to average peak placebo data inappropriate in parallel-group studies. However, placebo correction may be performed for each subject in cross-over studies, leading to a reduction in peak changes and an increase in trough:peak ratio values.
When the trough:peak ratio is assessed from ambulatory BP, peak and trough changes should preferably be computed over a 2 h time window. To remove values with no pharmacodynamic significance, the analysis should preferably be conducted only in responders to treatment at peak. Although placebo is accompanied by some peak effect, placebo correction might be appropriate only for individual subjects in cross-over studies.
解决与从24小时动态血压（BP）记录计算谷峰比相关的几个方法学问题。
汇总纳入平行组研究（n = 280）或交叉研究（n = 39）的轻度原发性高血压患者的数据。在停用治疗2至4周后以及使用钙拮抗剂（n = 143）、血管紧张素转换酶抑制剂（n = 103）或安慰剂（平行组研究中的73例患者和交叉研究中的39例患者）进行4至8周治疗结束时，可获得24小时动态血压记录。每次记录在0900至1000时之间开始，在治疗阶段服用药物或安慰剂后立即进行。白天每隔15分钟测量一次血压，夜间每隔15至20分钟测量一次。峰变化根据服药后2至8小时的收缩压和舒张压计算，谷变化根据24小时最后4小时的读数计算。
当数据在1、2、4和6小时内平均时，药物治疗引起的峰变化逐渐减小。血压可重复性随之增加，在正确估计峰变化和可重复性之间的最佳折衷是相邻2小时内最大血压下降的平均值。峰和谷（最后2小时的平均值）变化呈正态分布，而谷峰比呈非正态分布，离散度大且有许多无药效学意义的个体值（即远高于1和低于0）。选择治疗反应者可减少离散度并使谷峰比分布呈正态。谷峰比与治疗引起的血压变异性变化（24小时平均值的标准差）之间无相关性。服用安慰剂未引起谷值下降，但引起适度的峰值下降。安慰剂期间的峰变化也显示出广泛的离散度和非正态分布，这使得在平行组研究中对平均安慰剂峰值数据进行校正不合适。然而，在交叉研究中可对每个受试者进行安慰剂校正，导致峰变化减小和谷峰比值增加。
当从动态血压评估谷峰比时，峰和谷变化最好在2小时时间窗内计算。为去除无药效学意义的值，分析最好仅在峰时的治疗反应者中进行。尽管安慰剂有一些峰值效应，但安慰剂校正可能仅适用于交叉研究中的个体受试者。
