Skip to main content

Table 5 Frequency and agreement between methods of determination of nursing home residence (n = 150, Weighted n = 520,260)

From: Validation of a method for identifying nursing home admissions using administrative claims

 

Algorithm 1*

Algorithm 2

Expert Review

Frequency (%) of NH stay

   

Unweighted

6 (4.0)

18 (12.0)

18 (12.0)

Weighted

3,320 (0.6)

5,473 (1.1)

5,473 (1.1)

Correspondence with expert review

   

   Agreement

   

Unweighted

92.0%

98.7%

-

Weighted

99.6%

99.9%

-

   Sensitivity

   

Unweighted

0.33

0.94

-

Weighted

0.61

0.97

-

   Specificity

   

Unweighted

1.00

0.99

-

Weighted

1.00

1.00

-

   Kappa (95% CI)

   

Unweighted

0.47 (0.22, 0.71)

0.93 (0.85, 1.00)

-

Weighted

0.75 (0.74, 0.76)

0.97 (0.96, 0.97)

-

  1. NH = nursing home; CI = confidence interval
  2. *Algorithm 1 defined a person-month as a nursing home stay if there was at least one day in the month where the nursing home RULE = "PROBABLE"; otherwise the person-month was considered not to be a nursing home stay.
  3. † Algorithm 2 defined a person-month as a nursing home stay if there was at least one day in the month where the nursing home RULE = "PROBABLE" or RULE = "POSSIBLE"; otherwise the person-month was considered not to be a nursing home stay.