Background
It has been suggested by previous research findings that surgery is an effective treatment for sports-related injuries. This study aimed to explore the improvement of range of movement of the arm after surgery at one month and 12 months, and examine the relationships between movement of arm at baseline, and gender, the severity of the injury, or the dominant side.
Methods
Range of arm movement (degree), age (years), gender (male=1 or female=0), dominant side (dominant or not), and severity of the injury, was collected from 96 patients from hospital outpatients clinics in four Brisbane hospitals.
Study
To identify whether there was an overall improvement in the range of movement over time, a One Way ANOVA test was conducted. This was the most appropriate analytical test since the dependent variable (range of movement) is a categorical variable, hence ANOVA would enable easy comparison of means for the three ranges of movement and the fact that ANOVA is a parametric test. In addition, the variability of range of movement was assumed to be homogenous as confirmed by a non-significant Levene’s test, F(2, 285) =.610, p>.05. Using ANOVA would enable post-hoc tests to be conducted to clearly identify which groups differed (Field, 2009). In this case, it was necessary to identify at what time (baseline, 1-month post-surgery and 12 months post-surgery) the range of movement differed.
The mean range of arm movement during baseline was 44.33 degrees with a standard deviation of 16.15. During 1 month post-surgery, the mean range of arm movement was 75.75 degrees with a standard deviation of 16.51. The mean range of movement at 12 months post-surgery was 1.23E2 with a standard deviation of 16.52.
There was homogeneity of variances in arm movement between the three-time periods since the Levene’s statistic was not significant (p =.544 and this is greater than.05).
From the ANOVA table (Table 2), the F ratio is significant, F(2, 285) = 556.846, p<.05, indicating that there were significant differences in the range of arm movement during baseline, 1-month post-surgery and at 12 months post-surgery. Since this does not necessarily indicate that the means in arm movement were significantly different from each other, it was necessary to conduct post hoc tests. In this case, Tukey HSD test was used as a post hoc test to determine between which time periods the significant differences in arm movement occurred. There were significant differences (p =.001 in all the time periods) in the range of arm movement in all three time periods. In other words, there was a significant difference in range of arm movement during baseline, 1 month after surgery and 12 months after surgery. In summary, the ANOVA and post-hoc tests indicate that there was an improvement in range of arm movement (in degrees) over time.
In an effort to identify whether range of arm movement at baseline depended on patient’s age, a linear regression analysis was conducted. Since the range of movement is a continuous variable and it is the dependent variable in this case, and age is a continuous and independent variable, linear regression was the most appropriate analysis. In this case, there was only one predictor variable (age) and it is a continuous variable which predicts a single continuous outcome variable (range of arm movement).
The mean age of all the 94 arm injury patients was 23.22 years with a standard deviation of 4.32. On the other hand, the mean range of arm movement was 44.77 degrees with a standard deviation of 16.02. The Pearson correlation coefficient between range of arm movement at baseline and age of the patient is shown in Table 3 as being non-significant, rp =.096, single tailed p>.05. This indicates that range of arm movement at baseline did not depend on the patient’s age.
The value of R square was.009 and this can be converted to a percentage to 0.9 percent. This implies that 0.9 percent variance in arm movement was explained by the patient’s age considering other variables to be constant. This is an indication that age did not affect the range of arm movement significantly at baseline.
The regression equation derived from Table 4 is y =.356x + 36.505. From this equation, the y-intercept value (constant) is 36.505 and this is a positive value implying that the intercept occurs above the x-axis. Table 4 also displays the unstandardised regression coefficient between range of arm movement at baseline and the patient’s age as B=.356. From this output, it is evident that as age changed by 1 unit, the range of arm movement changed by.356 units. The 95% CI for B ranges from -.409 to 1.120 and for the intercept ranges from 18.444 to 54.566 indicating that from this sample, the intercept of the population was 95 percent likely to be within the above ranges. Table 4 indicates that the standardized Beta was 0.096 implying that 0.096 degrees of arm movement at baseline was dependent on the patient’s age when all other factors are held constant. This was a very small variation in arm movement as a factor of patient’s age.
To determine whether severity of arm injury depended on gender and dominant side, a logistic regression was conducted. It is important to note that the severity of injury and dominant side are categorical variables thus logistic regression was a suitable analysis in this case. Logistic regression is the most ideal analysis where the outcome variable is categorical (Field, 2009), and in this case severity of injury was a categorical variable and the outcome was based on categorical predictor variables (gender and dominant side). In this case, there was need to determine which variable (between gender and dominant side) predicted injury (severe or not severe) In addition, the fact that there were two independent variables (gender and dominant side) affecting the dependent variable (severity of injury) called for a logistic regression analysis as a viable way of determining the effect of the two independent variables on severity of injury.
In 50 patients, arm injury was on the dominant side whereas 50 patients had arm injury on the non-dominant side. There were 50 female patients and 46 male patients. Of the 96 patients in this study, 83.3 percent (80) did not have severe arm injury with the rest, 13.7 percent (16), having severe arm injury.
A logistic regression analysis was conducted to determine whether severity of arm injury depended on the patient’s gender as well as the dominant side of injury. The influence of gender on the severity of arm injury at baseline was significant, B = 2.357, p <.05, indicating that the patient’s gender was a key determinant of the severity of the injury. The odds ratio in Table 5 was 10.56 implying that female patients were 10.56 times more likely to have severe arm injury at baseline compared to male patients. The 95% confidence interval for this odds ratio case ranged from 2.24 to 49.80.
Looking at the influence of dominant side of injury on severity of injury, it is evident that dominant side of injury was not a significant determinant of severity of injury at baseline (B = -.064, p >.05). The odds ratio in this case was.938 implying that patients who had arm injury on the dominant side were.938 times more likely to have severe arm injury at baseline compared to patients whose injury was on the non-dominant side. The 95% confidence interval for this odds ratio ranges from.297 to 2.966. In summary, it is evident from the logistic regression that severity of injury at baseline depended on patient’s gender but not on the dominant side.
Discussion
Looking at the association between range of movement and sports-related injury, it is evident that range of arm movement improved significantly from baseline to 1 month after surgery and 12 months post-surgery. The significant improvement of range of movement over time confirms other research findings that surgery is an effective treatment for sports-related injury. It is evident that patient’s age caused a non–significant effect on the range of movement at baseline in sports-related arm injury. Severity of arm injury at baseline depended on gender of the patient with females experiencing more severe injuries compared to male patients. On the contrary, the dominant side of injury is not a significant determinant of severity of injury at baseline. It can therefore be concluded that surgery is effective in treating sports-related arm injury and the severity of arm injury at baseline is a factor of patient’s gender, with age causing no significant contribution on arm movement during baseline treatment period.
References
Field, A. (2009). Discovering statistics using SPSS (3rd Ed.). Thousand Oaks, CA: Sage Publications.
Polit, D. F. (2010). Statistics and data analysis for nursing research (Second Edition). Upper Saddle River, NJ: Pearson Education.
Appendix
Table 2: ANOVA for Range of Arm Movement (Degrees)
Table 3: Pearson Correlation for Range of Arm Movement (Degrees) and Age of the Patient
Table 4: Linear Regression Coefficients for Range of Arm Movement and Age of the Patient
Table 5: Logistic Regression Output for Gender and Dominant Side