Skip to main content

Table 2 Geometric mean of risk ratios and a summary analysis for different cohort size, outcome incidence and exposure prevalence of initiators of celecoxib or NSAIDs (ibuprofen or diclofenac) in a cohort 18–65 years old between 1 July 2003 and 30 September 2004 in the MarketScan database

From: Effects of aggregation of drug and diagnostic codes on the performance of the high-dimensional propensity score algorithm: an empirical example

Cohorts and variable selection methods

Median of exposed subjects

Median of exposed outcomes

Median of unexposed subjects

Median of unexposed outcomes

Geometric mean of RR*

25th-75th percentile of RR of samples†

Mean variable coverage %‡

 

n (%)

n (%)

n (%)

n (%)

 

Full cohort §

7197 (38)

46 (0.64)

11632 (62)

71 (0.61)

   

Unadjusted

    

1.05

  

Basic covariates

    

0.98

  

Basic and extended covariates

    

0.95

  

Basic and hd-PS covariates

    

0.92

 

100

Basic, extended and hd-PS covariates

    

0.94

 

100

Condition 1: 50% size sample

3594 (38)

23 (0.64)

5821 (62)

36 (0.62)

   

Unadjusted

    

1.02

(0.89-1.20)

 

Basic covariates

    

0.96

(0.84-1.11)

 

Basic and extended covariates

    

0.92

(0.80-1.09)

 

Basic and hd-PS covariates

    

0.88

(0.74-1.07)

65

Basic, extended and hd-PS covariates

    

0.89

(0.74-1.11)

65

Condition 2: 20% size sample

1441 (38)

0 (0.66)

2325 (62)

14 (0.60)

   

Unadjusted

    

1.10

(0.89-1.37)

 

Basic covariates

    

1.03

(0.82-1.29)

 

Basic and extended covariates

    

0.99

(0.79-1.24)

 

Basic and hd-PS covariates

    

0.94

(0.71-1.21)

41

Basic, extended and hd-PS covariates

    

0.95

(0.70-1.25)

41

Condition 3: 50% outcome incidence sample

7220 (38)

23 (0.32)

11667 (62)

36 (0.31)

   

Unadjusted

    

1.02

(0.89-1.19)

 

Basic covariates

    

0.96

(0.84-1.13)

 

Basic and extended covariates

    

0.93

(0.81-1.09)

 

Basic and hd-PS covariates

    

0.90

(0.78-1.08)

65

Basic, extended and hd-PS covariates

    

0.91

(0.78-1.08)

65

Condition 4: 20% outcome incidence sample

7233 (38)

10 (0.14)

11689 (62)

14 (0.12)

   

Unadjusted

    

1.00

(0.81-1.37)

 

Basic covariates

    

0.94

(0.73-1.25)

 

Basic and extended covariates

    

0.91

(0.69-1.19)

 

Basic and hd-PS covariates

    

0.85

(0.69-1.17)

42

Basic, extended and hd-PS covariates

    

0.86

(0.70-1.14)

42

Condition 5: 50% exposure prevalence sample

3599 (19)

22 (0.61)

15230 (81)

95 (0.62)

   

Unadjusted

    

1.02

(0.93-1.13)

 

Basic covariates

    

0.94

(0.86-1.05)

 

Basic and extended covariates

    

0.91

(0.83-1.02)

 

Basic and hd-PS covariates

    

0.88

(0.79-0.98)

81

Basic, extended and hd-PS covariates

    

0.88

(0.79-1.00)

81

Condition 6: 20% exposure prevalence sample

1440 (8)

9 (0.63)

17389 (96)

108 (0.62)

   

Unadjusted

    

0.97

(0.77-1.24)

 

Basic covariates

    

0.89

(0.72-1.15)

 

Basic and extended covariates

    

0.86

(0.70-1.08)

 

Basic and hd-PS covariates

    

0.89

(0.73-1.13)

73

Basic, extended and hd-PS covariates

    

0.89

(0.72-1.14)

73

  1. Abbreviations: basic covariates included continuous age, gender and calendar year; extended covariates included covariates adjusted for in published studies; hd-PS, high dimensional propensity score.
  2. *Geometric mean of the risk ratio observed in 100 samples at this sampling rate.
  3. †25th-75th percentiles of the risk ratio observed in 100 samples at this sampling rate.
  4. ‡Mean percentage of hd-PS variables in the full cohort also identified in samples.
  5. §For the full cohort, all values are the number, not median.
Back to article page