Patterns of recovery from pain after cesarean delivery

Abstract

We know very little about the change in pain in the first 2 months after surgery. To address this gap, we studied 530 women scheduled for elective cesarean delivery who completed daily pain diaries for 2 months after surgery through text messaging. Over 82% of subjects missed fewer than 10 diary entries and were included in the analysis. Completers were more likely to be Caucasian, nonsmokers, and with fewer previous pregnancies than noncompleters. Daily worst pain intensity ratings for the previous 24 hours were fit to a log(time) function and allowed to change to a different function up to 3 times according to a Bayesian criterion. All women had at least one change point, occurring 22 ± 9 days postoperatively, and 81% of women had only one change, most commonly to a linear function at 0 pain. Approximately 9% of women were predicted to have pain 2 months after surgery, similar to previous observations. Cluster analysis revealed 6 trajectories of recovery from pain. Predictors of cluster membership included severity of acute pain, perceived stress, surgical factors, and smoking status. These data demonstrate feasibility but considerable challenges to this approach to data acquisition. The form of the initial process of recovery from pain is common to all women, with divergence of patterns at 2 to 4 weeks after cesarean delivery. The change-point model accurately predicts recovery from pain; its parameters can be used to assess predictors of speed of recovery; and it may be useful for future observational, forecasting, and interventional trials.

Figure 1.. Subject disposition

Number of subjects are in parentheses. The two datasets discussed in this study are the complete dataset used for the primary analysis and the missing embedded dataset used for the sensitivity analysis.

Figure 2.. Missing data

Proportion of subjects in the complete dataset (blue) and missing embedded dataset (red) with each of the possible number of daily diary data points missing.

Figure 3.. Pattern of recovery from pain in the study population

A) Heat map of worst daily pain over the 2 month study period for the 435 subjects in the complete dataset. For depiction of typical recovery, the black line is smoothed from the mean value for all subjects on that day. B) Proportion of subjects changing from one function of recovery to another as a function of postoperative day. All subjects in the complete dataset (n=435) had at least one change point, shown in red. Fewer had at least 2 change points, shown in green, and only 20 subjects had a 3^rd change point, shown in blue.

Figure 4.. Cluster analysis of group patterns of recovery in the complete dataset

Modeled worst daily pain over the two month period assigned to 6 clusters, depicted in a rainbow pattern from most rapid recovery in purple to most slowest recovery in red. Number of members in each class in parentheses.

Figure 5.. Individual data within clusters

Modeled data from all individuals within each of the 6 clusters, depicted in a rainbow pattern from most rapid recovery in purple to most slowest recovery in red. Thin lines are individual subjects and thick line is the modeled pattern for that cluster.