Skip to main content

Table 3 Effect size of association, differences in prevalence and description of strength of association

From: Causal assessment of smoking and tooth loss: A systematic review of observational studies

   

Association

Prevalence of tooth loss (%)

 

Type of exposure

First author, year

Sex

Effect size

(95% CI)

Type

Current smoker

Non-smoker

Difference

Strength of association

Current smokers

Randolph, 2001

M

F

1.69 (1.31, 2.20)

OR

50

57

41

46

9

11

Moderate

Moderate

 

Klein, 2004

M, F

4.04 (2.52, 6.49)

OR

92.3

79.5

12.8

Moderate

 

Musacchio, 2007

M, F

4.01 (2.59, 6.20)

OR

48.1**

42.3**

5.8**

Weak

 

Ojima, 2007

M

F

2.21 (1.40, 3.50)

1.70 (1.13, 2.55)

OR

39.3

43.2

21.8

29.4

17.5

13.8

Moderate

Moderate

 

Hanioka, 2007

M

F

2.24 (1.28, 3.94)

2.74 (1.46, 5.16)

OR

36.9

38.9

28.5

38.6

8.4

0.3

Moderate

Weak

 

Mundt, 2007

M, F

2.3 (1.6, 3.4)*

OR

21.1**

8.4**

12.7**

Moderate

 

Krall, 2006

M

2.1 (1.5, 3.1)

HR

   

Weak

 

Dietrich, 2007

M

2.3 (2.1, 2.5)*

HR

   

Moderate

Former smokers

Randolph, 2001

M

F

1.26 (1.04, 1.54)

OR

45

55

41

46

4

9

Weak***

Moderate

 

Klein, 2004

M, F

1.57 (1.25, 1.98)

OR

85.8

79.5

6.3

Weak***

 

Musacchio, 2007

M, F

3.42 (2.42, 4.82)

OR

45.5**

42.3**

3.2**

Weak

 

Ojima, 2007

M

F

1.25 (0.55, 2.86)

0.52 (0.23, 1.18)

OR

26.3

20.0

21.8

29.4

4.5

-9.4

NS***

 

Hanioka, 2007

M

F

1.55 (0.88, 2.74)

1.17 (0.44, 3.09)

OR

38.6

34.3

28.5

38.6

10.1

-4.3

NS***

 

Mundt, 2007

M, F

1.7 (1.0, 3.1)*

OR

12.1**

8.4**

3.7**

NS***

 

Krall, 2006

M

1.3 (0.9, 1.7)

HR

   

NS***

 

Dietrich, 2007

M

1.2 (1.2, 1.3)*

HR

   

Weak***

  1. The evidence of weak to moderate association between smoking and tooth loss was consistent in high-quality studies, and the effect size was consistently smaller for former smokers than for current smokers.
  2. CI, confidence interval; OR, odds ratio; HR, hazard ratio; *extracted from the category that included the median value of the group as a representative because the effect size for all current or all former smokers was not reported; **calculated by reviewer based on date in the table in the original literature; ***lower rank than current smokers