Skip to main content

Table 2 Leeds data: Agreement between admissions and cases with coding combinations with short admissions, readmission and elective admissions removed

From: ‘Caveat emptor’: the cautionary tale of endocarditis and the potential pitfalls of clinical coding data—an electronic health records study

Codes used Removal of admissions with:
< 3 day LOS Readmissions < 30 days Elective status True positive False positive True negative False negative Sensitivity Specificity PPV NPV
I33.0 primary only     433 97 1664 568 0.43 0.95 0.82 0.75
I33.0 primary only    431 67 1694 570 0.43 0.96 0.87 0.75
I33.0 primary only   426 65 1696 575 0.43 0.96 0.87 0.75
I33.0 primary only 406 57 1704 595 0.41 0.97 0.88 0.74
All primary codes     562 178 1583 439 0.56 0.90 0.76 0.78
All primary codes    559 114 1647 442 0.56 0.94 0.83 0.79
All primary codes   552 111 1650 449 0.55 0.94 0.83 0.79
All primary codes 529 96 1665 472 0.53 0.95 0.85 0.78
I33.0 in any position     549 162 1599 452 0.55 0.91 0.77 0.78
I33.0 in any position    547 125 1636 454 0.55 0.93 0.81 0.78
I33.0 in any position   540 117 1644 461 0.54 0.93 0.82 0.78
I33.0 in any position 514 102 1659 487 0.51 0.94 0.83 0.77
All primary and I33.0 secondary     669 242 1519 332 0.67 0.86 0.73 0.82
All primary and I33.0 secondary    666 172 1589 335 0.67 0.90 0.80 0.83
All primary and I33.0 secondary   658 163 1598 343 0.66 0.91 0.80 0.82
All primary and I33.0 secondary 629 141 1620 372 0.63 0.92 0.82 0.81
All codes except I38 secondary     697 299 1462 304 0.70 0.83 0.70 0.83
All codes except I38 secondary    694 209 1552 307 0.69 0.88 0.77 0.84
All codes except I38 secondary   686 196 1565 315 0.69 0.89 0.78 0.83
All codes except I38 secondary 653 168 1593 348 0.65 0.91 0.80 0.82
All codes     738 943 818 263 0.74 0.47 0.44 0.76
All codes    735 573 1188 266 0.73 0.68 0.56 0.82
All codes   726 538 1223 275 0.73 0.69 0.57 0.82
All codes 693 487 1274 308 0.69 0.72 0.59 0.81
  1. Note: Total number of clinically confirmed endocarditis cases = true positives + false negatives (1001); total number of endocarditis admissions = sum of all four columns (2762)