Skip to main content

Table 1 Assessing the accuracy of NHS numbers in routine data.

From: The SAIL databank: linking multiple health and social care datasets

Data Source

Type of Record Linkage

Result of comparing the NHS number allocated by the record linkage process with the original submitted NHS number

  

Same

Different

Not found

% Agreement

% Disagreement

% Linked

  

Allocated NHS number equals the submitted NHS number

Allocated NHS number differs to the submitted NHS number

An NHS number is not found by the record linkage process

Of the records that were allocated an NHS number, the percentage that were allocated an NHS number equal to the NHS number submitted

Of the records that were allocated an NHS number, the percentage that were allocated an NHS number different to the NHS number submitted

Of the records that were processed, the percentage that were allocated an NHS Number

  

a

b

c

= a/(a+b)

= b/(a+b)

= (a+b)/(a+b+c)

Primary Care Practice Clinical Systems (GP) (n = 229,117)

DRL

223,344

40

5,733

99.982%

0.018%

97.498%

 

PRL – 99% cut off

227,778

51

1,288

99.978%

0.022%

99.438%

 

PRL – 95% cut off

228,288

55

774

99.976%

0.024%

99.662%

 

PRL – 90% cut off

228,479

56

582

99.976%

0.025%

99.746%

 

PRL – 50% cut off

228,699

61

357

99.973%

0.027%

99.844%

Secondary Care Hospital Admissions (PEDW) (n = 264,868)

DRL

216,062

323

48,483

99.851%

0.149%

81.695%

 

PRL – 99% cut off

244,692

410

19,766

99.833%

0.167%

92.537%

 

PRL – 95% cut off

247,865

439

16,564

99.823%

0.177%

93.746%

 

PRL – 90% cut off

249,024

453

15,391

99.818%

0.182%

94.189%

 

PRL – 50% cut off

250,155

465

14,248

99.815%

0.186%

94.621%

  1. This shows the level of agreement between NHS numbers supplied in the General Practice (GP) dataset (n = 229,117) and the Patient Episode Database for Wales (PEDW) dataset (n = 264,868) with those allocated by the matching process using by DRL and PRL. The NHS Administrative Register (NHSAR) was used as the reference.