Vitamin E and the Healing of Bone Fracture: The Current State of Evidence

Table 4

Commentary of statistical analysis and reanalysis of results from past animal studies of the effect of vitamin E on bone fracture healing.

Study	Result	Reanalysed result (where necessary or possible) and final conclusion (focusing on the vitamin E group versus control group)	Effect size and power of the result via G*Power (vitamin E group versus control group)	Sample size for future replication of results via G*Power
	(A) Radiological-based grade of bone formation
		Reanalysis was unnecessary for conclusion	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=0.56; Power = 0.28	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 59
	(1) Day 20 analysis	Mann-Whitney 𝑈 test: 𝑃>0.05
		Final conclusion: there was no significant difference between the treatment group and control group in terms of radiological-based grade of bone formation
		Reanalysis was unnecessary for conclusion	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=1.31; Power = 0.89	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 12
	(2) Day 30 analysis	Mann-Whitney 𝑈 test: 𝑃<0.01
		Final conclusion: the treatment group has significantly higher radiological-based grade of bone formation compared to control group
		Reanalysis was unnecessary for conclusion	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=2.34; Power ≈ 1.00	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 5
	(3) Day 40 analysis	Mann-Whitney 𝑈 test: 𝑃<0.01
Kurklu et al. [15]		Final conclusion: the treatment group has significantly higher radiological-based grade of bone formation compared to control group
	(B) Ratio of osteoblastic activity
	(1) Day 5 analysis	Reanalysis was unnecessary for conclusion	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=0.47; Power = 0.21	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 83
		Mann-Whitney 𝑈 test: 𝑃>0.05
		Final conclusion: there was no significant difference between the treatment group and control group in terms of osteoblastic activity
	(2) Day 20 analysis	Reanalysis was unnecessary for conclusion	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=2.35; Power ≈ 1	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 5
		Mann-Whitney 𝑈 test: 𝑃<0.01
		Final conclusion: the treatment group has significantly higher osteoblastic activity compared to control group
	(C) Histological-based grade of bone formation
	(1) Day 40 analysis	Reanalysis was unnecessary for conclusion	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=2.67; Power ≈ 1	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 5
		Mann-Whitney 𝑈 test: 𝑃<0.01
		Final conclusion: the treatment group has significantly higher histological-based grade of bone formation compared to control group
	(A) Callous volume
		Not enough information from the article for reanalysis	Not enough information from the article for calculation	Not enough information from the article for calculation
	(1) Day 14 analysis	Unknown statistical test: 𝑃>0.05
		Final conclusion: There was no significant difference of callous volume between the ovariectomized treatment group and ovariectomized control group
	(B) Radiological-basedscore of callous staging
		Reanalysis was performed using the tabulated data reported for the chi-square test in the article
		Descriptive results:
		(1) Sham group (𝑛=8): median = 2 (mean = 2.12, s.d. = 0.64)
		(2) Ovariectomized control group (𝑛=8): median = 3 (mean = 3.00, s.d. = 0.76)
	(1) Day 14 analysis	(3) Ovariectomized treatment group (𝑛=8): median = 3 (mean = 3.12, s.d. = 0.84)	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=0.15; Power = 0.06	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 811
		Kruskal-Wallis test: 𝐻(df=2)=6.5, 𝑃<0.05
		Posthoc Mann-Whitney 𝑈 test focused on the comparison between ovariectomized treatment group and ovariectomized control group (Bonferroni’s correction was not necessary in this case): 𝑈=29, 𝑃>0.05
Shuid et al. [18]		Final conclusion: there was no significant difference between the treatment group and control group in terms of radiological-based callous staging
	(C) Radiological-based score of fracture healing staging
	(1) Day 14 analysis	Reanalysis was performed using the tabulated data reported for the chi-square test in the article
		Descriptive results
		(1) Sham group (𝑛=8): median = 3 (mean = 2.88, s.d. = 0.35)
		(2) Ovariectomized control group (𝑛=8): median = 2 (mean = 2.38, s.d. = 0.52)
		(3) Overaiectomized treatment group (𝑛=8): median = 3 (mean = 2.88, s.d. = 0.35)	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=1.13; Power = 0.49	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 16
		Kruskal-Wallis test: 𝐻(df=2)=6.2, 𝑃<0.05
		Post-hoc Mann-Whitney 𝑈 test focused on the comparison between ovariectomized treatment group and ovariectomized control group (Bonferroni’s correction was not necessary in this case): 𝑈=16, 𝑃<0.05
		Final conclusion: the treatment group had significantly higher radiological-based fracture healing staging than control group
	(A) Radiological-based evaluation of callous formation staging (unspecified)
Paskalev et al. [8]	(1) Day 1, week 2, month 1, month 2, month 3, and month 4 analysis	Not enough information from the article for reanalysis; No statistical test result reported; Final conclusion: it was stated that callous formation stage was better in the treatment group compared to the control	Not enough information from the article for calculation	Not enough information from the article for calculation
	(B) Radiological-based evaluation of bone remodeling staging (unspecified)
	(1) Day 1, week 2, month 1, month 2, month 3, and month 4 analysis	Not enough information from the article for reanalysis; No statistical test result reported; Final conclusion: it was stated that bone remodeling stage was better in the treatment group compared to the control	Not enough information from the article for calculation	Not enough information from the article for calculation
	(A) Radiological-based grade of bone formation
	(1) Day 60 analysis	Reanalysis was performed using the tabulated data reported for the chi-square test in the article	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=2.5; Power ≈ 1	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 5
		Descriptive results
		Control group (𝑛=10): median = 3 (mean = 3.10, s.d. = 0.32)
		Treatment group (𝑛=10): median = 4 (mean = 3.90, s.d. = 0.32)
		Mann-Whitney 𝑈 test: 𝑈=10.0, 𝑃<0.01
		Final conclusion: the treatment group had significantly higher radiological-based grade of bone formation than control group
Turk et al. [21]	(B) Histological-based grade of fracture healing
	(1) Day 60 analysis	Reanalysis was performed using the tabulated data reported for the chi-square test in the article	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=2.5; Power ≈ 1	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 5
		Descriptive results
		Control group (𝑛=10): median = 3 (mean = 3.10, s.d. = 0.57)
		Treatment group (𝑛=10): median = 4 (mean = 3.90, s.d. = 0.32)
		Mann-Whitney 𝑈 test: 𝑈=14.5, 𝑃<0.01
		Final conclusion: the treatment group had significantly higher histological-based grade of bone formation than control group
	(A) Histological-based grade of fracture healing
	(1) Day 21 (16 days after the last treatment) analysis	Reanalysis was performed using the tabulated data reported for the chi-square test in the article	If 𝛼=0.05, 2-tailed; distribution: minimum ARE; Effect size, 𝑑=1.49; Power = 0.82	If 𝛼=0.05, 2-tailed; power = 0.80; distribution: minimum ARE; Sample size for each group (equal number) = 10
		Descriptive results:
Durak et al. [23]		Control group (𝑛=10): median = 2 (mean = 2.40, s.d. = 0.52)
		Treatment group (𝑛=10): median = 3 (mean = 3.30, s.d. = 0.68)
		Mann-Whitney 𝑈 test: 𝑈=17, 𝑃<0.01
		Final conclusion: the treatment group had significantly higher histological-based grade of fracture healing than control group
	(A) Callous index
		Reanalysis was unnecessary for conclusion	Not enough information from the article for calculation (absence of standard deviation value)	Not enough information from the article for calculation (absence of standard deviation value)
	(1) Day 14 analysis	Posthoc Mann-Whitney 𝑈 test: 𝑃>0.05
		Final conclusion: there was no significant difference between the Vit. E only treatment group and control group in terms of callous index
		Reanalysis was unnecessary for conclusion	Not enough information from the article for calculation (absence of standard deviation value)	Not enough information from the article for calculation (absence of standard deviation value)
	(2) Day 21 analysis	Post-hoc Mann-Whitney 𝑈 test: 𝑃>0.05
		Final conclusion: there was no significant difference between the Vit. E only treatment group and control group in terms of callous index
Sarisözen et al. [24]	(B) Histological-based grade of fracture healing
		Reanalysis was unnecessary for conclusion	Not enough information from the article for calculation (absence of standard deviation value)	Not enough information from the article for calculation (absence of standard deviation value)
	(1) Day 14 analysis	Post-hoc Mann-Whitney 𝑈 test: 𝑃>0.05
		Final conclusion: there was no significant difference between the Vit. E only treatment group and control group in terms of histological-based grade of fracture healing
		Reanalysis was unnecessary for conclusion	Not enough information from the article for calculation (absence of standard deviation value)	Not enough information from the article for calculation (absence of standard deviation value)
	(2) Day 21 analysis	Post-hoc Mann-Whitney 𝑈 test: 𝑃>0.05
		Final conclusion: there was no significant difference between the Vit. E only treatment group and control group in terms of histological-based grade of fracture healing