The effect of massage and shaking on infants with colic in a clinical trial concerning the misspecification

Background: Statistical models are used to investigate the relationship between variables in statistical studies. Considering the variety of statistical models, finding the most suitable model is a complex work. This study aimed to compare different models in the treatment of infants' colic and the misspecification of specificity. Methods: This randomized clinical trial was conducted on 100 infants with colic in the pediatric clinic of Amir Kabir Hospital in Arak, the intervention and control groups were randomly divided into two groups. The collection and analysis of the data was performed in 2016. After teaching massage to mothers of the intervention group, they were asked to perform massage on infants three times a day during the week. In the control group, mothers can relieve the symptoms of colic by shaking the infant. Parents recorded the number and severity of crying daily in the checklist. Finally, by using different models, R software, SAS, and goodness of fit, the best model was introduced. Results: In the massage group, the mean crying intensity of infants with colic decreased from 5.01 units on the first day to 2.47 units on the seventh day. On the other hand, the difference in mean sleep time changed from 1.81 hours in favor of the shaking group on the first day to 1.26 hours in favor of the massage group on the seventh day. Also, the severity of crying in the infants of the massage group was significantly higher than the impulse group. Regarding the grace of marginal models, the first-order self-return correlation structure was the best grace and for some variables, the model had random effects with a gamma distribution for the random component. Conclusion: Massage can reduce infants' colic. Statistically, in the case of a nonlinear model, the variance of estimates is more than estimated to be influenced by the misspecification of the correlation structure.