Introduction
Material and methods
Search strategy
Study inclusion
Data extraction and quality assessment
Results
Study characteristics
Authors, country, year [ref]
|
Sample population
|
Data years
|
Type of administrative database
|
Study size (n)
|
ICD version
|
Diagnostic coding field position
|
Reference/gold standard
|
Sn
|
Sp
|
PPV
|
NPV
|
---|---|---|---|---|---|---|---|---|---|---|---|
Cevasco et al., USA, 2011 [19] | General surgical | 2003 to 2007 | Population-based, inpatient Veterans Affairs hospital | 112 | ICD-9-CM | Secondary | Medical chart review | – | – | 53% | – |
General surgical | 2005 to 2007 | Population-based, inpatient community hospital | 164 | ICD-9-CM | Secondary | Medical chart review | – | – | 41% | – | |
Gedeborg et al., Sweden, 2007 [20] | ICU-specific | 1994 to 1999 | Population-based, inpatient | 4,181 | ICD-9b
| Principal, secondary | ICU database | 45.7% | 97.5% | 45.9% | 97.5% |
ICU-specific | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-10b
| Principal, secondary | ICU database | 52.5% | 92.6% | 28.0% | 97.3% | |
ICU-specific and DI | 1994 to 1999 | Population-based, inpatient | 4,181 | ICD-9b
| Principal, secondary | ICU database | 17.2% | 99.4% | 56.1% | 96.3% | |
ICU-specific and DI | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-10b
| Principal, secondary | ICU database | 20.1% | 98.4% | 40.9% | 95.7% | |
ICU-specific | 1994 to 1999 | Population-based, inpatient | 45 | ICD-9b ICD-10b
| Principal, secondary | Sepsis clinical trial patients | 42.2% | 95.5% | 7.4% | 99.5% | |
ICU-specific | 1994 to 1999 | Inpatient intensivist-coded ICU database | 45 | ICD-9b ICD-10b
| Principal, secondary | Sepsis clinical trial patients | 51.5% | 92.6% | 5.6% | 99.6% | |
ICU-specific | 1994 to 1999 | Population-based, inpatient | 4,181 | ICD-9c
| Principal, secondary | ICU database | 43.0% | 98.0% | 49.7% | 97.4% | |
ICU-specific | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-10c
| Principal, secondary | ICU database | 43.0% | 95.6% | – | – | |
ICU-specific | 1994 to 1999 | Population-based, inpatient | 4,181 | ICD-9b
| Principal | ICU database | 31.7% | 99.2% | 63.4% | 97.0% | |
ICU-specific | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-10b
| Principal | ICU database | 21.8% | 97.9% | 36.4% | 95.8% | |
ICU-specific: CAS | 1994 to 1999 | Population-based, inpatient | 4,181 | ICD-9b
| Principal | ICU database | 51.1% | 99.4% | 66.7% | 98.9% | |
ICU-specific CAS | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-10b
| Principal | ICU database | 31.8% | 99.0% | 41.5% | 98.3% | |
ICU-specific CAP and DI | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-9b
| Principal | ICU database | 19.1% | 99.8% | 64.3% | 98.2% | |
ICU-specific CAS and DI | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-10b
| Principal | ICU database | 17.6% | 99.4% | 42.8% | 97.9% | |
ICU-specific CAS | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-9c
| Principal | ICU database | 47.9% | 99.5% | 70.3% | 98.8% | |
ICU-specific CAS | 1994 to 1999 | Population-based, inpatient | 3,434 | ICD-10c
| Principal | ICU database | 27.1% | 99.0% | 39.7% | 98.2% | |
ICU-specific CAS | 1994 to 1999 | Population-based, inpatient | 45 | ICD-9c ICD-10c
| Principal | Sepsis clinical trial patients | 46.9% | 97.4% | 9.9% | 99.7% | |
ICU-specific CAS | 1994 to 1999 | Inpatient intensivist-coded ICU database | 45 | ICD-9c ICD-10c
| Principal | Sepsis clinical trial patients | 31.2% | 98.5% | 10.9% | 99.6% | |
Grijalva et al., USA, 2008 [21] | Rheumatoid arthritis | 1995 to 2004 | Inpatient database | 45 | ICD-9-CM | Principal, secondary | Medical chart review | – | – | 80% | – |
Ibrahim et al., Australia, 2012 [22] | General ICU | 2000 to 2006 | Inpatient database | 1,645 | ICD-10-AM | Principal | ICU database | 44.1% | 98.9% | 88.2% | 90.6% |
General ICU | 2000 to 2006 | Inpatient database | 45 | ICD-10-AM | Principal | ICU database | 16.5% | 99.8% | 93.9% | 86.8% | |
Iwashyna et al., USA, 2014 [12] | General | 2009 to 2010 | Population-based, inpatient | 111 | ICD-9-CM Angus | All | Medical chart review | 50.3% | 96.3% | 70.7% | 91.5% |
General | 2009 to 2010 | Population-based, inpatient | 111 | ICD-9-CM Explicit | All | Medical chart review | 9.3% | 100% | 100% | 86.0% | |
General | 2009 to 2010 | Population-based, inpatient | 111 | ICD-9-CM Martin | All | Medical chart review | 16.8% | 99.8% | 97.6% | 87.0% | |
Lawson et al., USA, 2012 [23] | General surgical | 2005 to 2008 | Population-based claims data | 13,410 | ICD-9-CM | All | ACS-NSQIP inpatient surgical database | 46.3% | 94.0% | – | – |
Madsen et al., Denmark, 1998 [24] | General | 1994 | Population-based, inpatient | 471 | ICD-10, Danish version | Unknown | Bacteraemia database | 5.9% | – | 21.7% | – |
Ollendorf et al., USA, 2002 [25] | Severe sepsis clinical trial patients | No dates given | Population-based, inpatient claims | 122 | ICD-9-CM | All | Severe sepsis clinical trial patients | – | – | 75.4% | – |
Quan et al., Canada, 2013 [26] | General surgical | 2007 to 2008 | Population-based, inpatient | 117 | ICD-10 | Secondary | Medical chart review | – | – | 9.8% | – |
General surgical | 2007 to 2008 | Population-based, inpatient | 34 | ICD-10 | Secondary | Medical chart review | – | – | 12.5% | – | |
Ramanathan et al., USA, 2014 [27] | Surgical patients | 2012 to 2013 | Surgical inpatient | 243 | ICD-9-CM | All | Medical chart review | 82.3% | 78.3% | 91.1% | 62.1% |
Schneeweiss et al., USA, 2007 [28] | General | 2001 to 2004 | Population-based, inpatient | 158 | ICD-9-CM | Principal | Medical chart review | – | – | 91% | – |
Whittaker et al., USA, 2013 [29] | ED admitted inpatients | 2005 to 2009 | Population-based, inpatient | 1,735 | ICD-9 (severe) | All | Medical chart review | 20.5% | – | – | – |
ED admitted inpatients | 2005 to 2009 | Population-based, inpatient | 1,735 | ICD-9 (severe) | All | Medical chart review | 47.2% (Angus) | – | – | – | |
ED admitted inpatients | 2005 to 2009 | Population-based, inpatient | 321 | ICD-9 (shock) | All | Medical chart review | 49.5% | – | – | – | |
ED admitted inpatients | 2005 to 2009 | Population-based, inpatient | 321 | ICD-9 (shock) | All | Medical chart review | 42.4% | – | – | – | |
ED admitted inpatients | 2005 to 2009 | Population-based, inpatient | 321 | ICD-9 (shock) | All | Medical chart review | 75.1% (Angus) | – | – | – |
Performance characteristics
Author
|
ICD version
|
ICD codes used
|
---|---|---|
Cevasco et al., USA, 2011 [19] | ICD-9-CM | 0380, 0381, 03810, 03811, 03812, 03819, 0382, 0383, 78552, 78559, 9980, 99591, 99592, 03840, 03841, 03842, 03843, 03844, 03849, 0388, 0389 |
Gedeborg et al., Sweden, 2007 [20] | ICD-9 |
Sepsis, wide criteria: 020–023, 027A, 032, 037, 040A, 041, 060, 061, 065, 071, 074C, 078G, 078H, 112X, 118, 590, 790H, 790 W |
ICD-10 |
Sepsis, wide criteria: A19–A36, A44.0, A49, A54.8, A69.2, A75–A79, B00.7, B00.9, B01.8, B01.9, B02.7–B02.9, B05.8, B05.9, B34.9, B38–B64, R50, T79.3, T81.3–T81.6, T83.6, T83.8, T84.5–T84.7, T85.7, T88.0, Y95 | |
ICD-9 |
Sepsis, narrow criteria: 036C–036E, 036X, 038, 084, 112 F, 117D, 286G, 999D | |
ICD-10 |
Sepsis, narrow criteria: A02.1, A04.0–A04.3, A39–A41, A42.7, A48, A90–A99, B37.7, B38.7, B39.3, B40.7, B41.7, B42.7, B44.7, B45.7, B46.4, B95–B99, D65, T80.2 | |
Grijalva et al., USA, 2008 [21] | ICD-9-CM | 003.1, 036.2, 785.52, 790.7, 038.x |
Ibrahim et al., Australia, 2012 [22] | ICD-10-AM |
Sepsis: A40.0, A40.1, A40.2, A40.3, A40.8, A40.9, A41.0, A41.1, A41.2, A41.3, A41.4, A41.5, A41.52, A41.58, A41.8, A41.9 |
Cholecystitis: K81.0, K83.0 | ||
Peritonitis: K65.9 | ||
Pneumonia: J13, J15.9, J18.0, J18.8, J18.9, J85.2 | ||
Perforation: K22.3, K27.5, K63.1 | ||
Lawson et al., USA, 2012 [23] | ICD-9-CM | 038, 78552, 99591, 99592, 9980, 99859, 99931 |
Madsen et al., USA, 1998 [24] | ICD-10, Danish version | A42.7, A41.3, A54.8, P36, P36.5, 36.4, P36.8, P36.2, P36.1, A02.1, A40.0, A40.2, A41.9, A40.8, O08.0, O85.9, A41.1, A41.2, A40.9, O75.3, A41.4, A41.5, P36.0, P36.3, P36.9, A41.0, A40.1, A40.3, A28.2, A41.8 |
Ollendorf et al., USA, 2002 [25] | ICD-9-CM | 038.3, 022.3, 790.7, 038.42, 038.49, 038.40, 038.41, 054.5, 036.2, 038.2, 038.43, 003.1, 038.8, 038.9, 020.2, 038.44, 038.1, 038.0 |
Schneeweiss et al., USA, 2007 [28] | ICD-9-CM |
Bacteremia: 038.-, 790.7 |
Quan et al., Canada, 2013 [26] | ICD-10-CA | A40.0, A40.1, A40.2, A40.3, A40.8, A40.9, A41.0, A41.1, A41.2, A41.3, A41.4, A41.5, A41.8, A41.9, R57.8, T81.1 |
Iwashyna et al., USA, 2014 [12] | ICD-9-CM | Angus positive: |
Severe sepsis: 995.92; Septic shock: 785.52; | ||
OR codes used to identify infection: 001, 002, 003, 004, 005, 008, 009, 010, 011, 012, 013, 014, 015, 016, 017, 018, 021, 022, 023, 024, 025, 026, 027, 030, 031, 032, 033, 034, 035, 036, 037, 038, 039, 040, 041, 090, 091, 092, 093, 094, 095, 096, 097, 098, 100, 101, 102, 103, 104, 110, 111, 112, 114, 115, 116, 117, 118, 320, 322, 324, 325, 420, 421, 451, 461, 462, 463, 464, 465, 481, 482, 485, 486, 491.21, 494, 510, 513, 540, 541, 542, 52.01, 562.03, 562.11, 562.13, 566, 567, 569.5, 569.83, 572.0, 572.1, 575.0, 590, 597, 599.0, 601, 614, 615, 616, 681, 682, 683, 686, 711.0, 730, 790.7, 996.6, 998.5, 999.3; | ||
AND acute organ dysfunction codes: 785.5, 458, 96.7, 343.3, 293, 348.1, 287.4, 287.5, 286.9, 286.6, 570, 573.4, 584 | ||
ICD-9-CM |
Explicit code positive: 995.92, 785.52 | |
ICD-9-CM |
Martin positive: 038, 020.0, 112.5, 112.81; AND acute organ dysfunction codes: 785.5, 458, 96.7, 343.3, 293, 348.1, 287.4, 287.5, 286.9, 286.6, 570, 573.4, 584 OR 995.92 OR 785.52 | |
Ramanathan et al., USA, 2014 [27] | ICD-9-CM | 995.91, 995.92, 785.52 |
Whittaker et al., USA, 2013 [29] | ICD-9 | 995.92, 785.52, Angus coding method (see Iwashyna et al. [12]) |
1. Identify article as study of assessing diagnostic accuracy | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
2. Identify article as study of administrative data | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
3. State disease identification & validation as goals of study | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
Methods: participants in validation cohort
| ||||||||||||
4. Age | 1 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 1 | 1 | 1 | 1 |
5. Disease | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
6. Severity | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
7. Location/jurisdiction | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 |
8. Describe recruitment procedure of validation cohort | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 |
9. Inclusion criteria | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 |
10. Exclusion criteria | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 0 | 1 |
11. Describe patient sampling (random, consecutive, all, etc.) | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 |
12. Describe data collection | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 |
13. Who identified patients and did selection adhere to patient recruitment criteria | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 |
14. Who collected data | 1 | 0 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 |
15. A priori data collection form | 1 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 |
16. Disease classification | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
17. Split sample (that is, revalidation using a separate cohort) | 0 | 0 | 0 | 0 | 0 | U | 0 | 0 | 0 | 0 | 0 | 0 |
Test methods
| ||||||||||||
18. Describe number, training and expertise of persons reading reference standard | 1 | 1 | 1 | 1 | 1 | 0 | 0 | 1 | 1 | 1 | 1 | 1 |
19. If more than one person reading reference standard, quote measure of consistency (for example, κ) | 1 | 0 | 1 | N/A | N/A | 0 | 0 | N/A | 0 | 0 | 0 | 0 |
20. Blinding of interpreters of reference standard to results of classification by administrative data (for example, chart abstractor blinded to how that chart was coded) | U | 1 | 1 | U | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 1 |
Statistical methods
| ||||||||||||
21. Describe methods of calculating diagnostic accuracy | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 0 | 1 |
Results: participants:
| ||||||||||||
22. Report when study done, start/end dates of enrolment | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 |
23. Describe number of people who satisfied inclusion/exclusion criteria | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 |
24. Study flow diagram | 0 | 0 | 0 | 1 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 0 |
Test results:
| ||||||||||||
25. Report distribution of disease severity | 1 | 1 | 1 | 1 | 0 | 1 | 0 | 0 | 1 | 1 | 1 | 1 |
26. Report cross-tabulation of index tests by results of reference standard | 1 | 1 | 1 | 1 | 0 | 1 | 0 | 0 | 1 | 0 | 0 | 0 |
27. Report at least four estimates of diagnostic accuracy | 0 | 1 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 1 | 1 | 0 |
Diagnostic accuracy measures reported
| ||||||||||||
28. Sensitivity | 0 | 1 | 0 | 1 | 1 | 1 | 0 | 0 | 0 | 1 | 1 | 1 |
29. Specificity | 0 | 1 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 1 | 1 | 0 |
30. PPV | 1 | 0 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 0 |
31. NPV | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 0 | 0 | 1 | 1 | 0 |
32. Likelihood ratios | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
33. κ | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
34. Area under the ROC curve/C-statistic | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
35. Accuracy/agreement | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
36. Other (specify) | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
37. Report accuracy for subgroups (for example, age, geography) | 0 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 1 | 0 | 0 | 1 |
38. If PPV/NPV reported, does the ratio of cases/controls of validation cohort approximate prevalence of condition in the population? | 1 | 1 | N/A | N/A | 1 | N/A | N/A | N/A | N/A | 0 | 0 | N/A |
39. Report 95% CI for each diagnostic measure | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 0 | 1 | 1 | 1 | 1 |
Discussion
| ||||||||||||
40. Discuss the applicability of the validation findings | 1 | 1 | 1 | 1 | 1 | 1 | 0 | 1 | 1 | 1 | 1 | 1 |
Total score | 27 | 25 | 27 | 30 | 28 | 24 | 10 | 22 | 28 | 29 | 24 | 26 |
Discussion
Conclusions
Key messages
-
Sepsis is undercoded in administrative data using ICD-9- and ICD-10-based case definitions.
-
There is high heterogeneity across studies for coding sepsis in administrative data, which is dependent on the ICD codes used, the population studied, the criteria used to define sepsis and the diagnostic coding position, to name a few.
-
To improve the capture of true sepsis cases in administrative data, strategies should be considered that include data linkage, improving physician documentation, implementing specialized coding procedures for ICU patients and the use of at least eight coding fields for diagnosis to capture complex conditions such as sepsis.