ACR Meeting Abstracts

ACR Meeting Abstracts

Abstract Number: 1689

Using Electronic Health Record Algorithms to Accurately Identify Patients with Systemic Sclerosis

Lia Jamian, Leslie Crofford and April Barnado, Medicine, Vanderbilt University Medical Center, Nashville, TN

Meeting: 2017 ACR/ARHP Annual Meeting

Date of first publication: September 18, 2017

Keywords: ,

Session Information

Date: Monday, November 6, 2017

Title: Systemic Sclerosis, Fibrosing Syndromes and Raynaud's – Clinical Aspects and Therapeutics Poster II

Session Type: ACR Poster Session B

Session Time: 9:00AM-11:00AM

Background/Purpose: Systemic sclerosis (SSc) is a rare, chronic, autoimmune disease with high morbidity and mortality. The electronic health record (EHR) represents a powerful tool to study rare disorders such as SSc. Currently, there are no validated algorithms to identify SSc patients in the EHR. We sought to develop algorithms that incorporated not only billing codes but also labs and keywords to identify SSc patients accurately.           Methods: We analyzed data from a de-identified version of Vanderbilt’s EHR called the Synthetic Derivative (SD) that contains over 2.8 million subjects with longitudinal clinical data.  Within the SD, we identified 1899 potential SSc patients with at least one count of the SSc ICD-9 (710.1) code. Of these potential subjects, we randomly selected 200 as a training set for chart review to identify true case status. A subject was defined as a case if diagnosed with SSc by a Vanderbilt or external rheumatologist, dermatologist, or pulmonologist. Potential subjects were then classified as either 1) cases, 2) not cases with alternative diagnoses noted, 3) unconfirmed if there was uncertainty in the diagnosis, or 4) missing if clinical documentation was missing. A priori, we selected the following potential algorithm components based on clinical knowledge and available data: SSc ICD-9 code, positive anti-nuclear antibody (ANA) (titer ≥ 1:80), and a keyword of Raynaud’s phenomenon (RP) in the clinic notes. Positive predictive values (PPVs) and sensitivity were calculated for combinations of the above algorithm components.           Results: Of the 200 subjects in the training set, 81 were true cases on chart review, 65 not cases, 17 unconfirmed, and 37 with missing clinical documentation. The PPV for using 1 count of the SSc ICD-9 code was 50%, 68% for ≥ 2 counts, 82% for ≥ 3 counts, and 90% for ≥ 4 counts. PPVs increased when a positive ANA or RP keyword was added to the ICD-9 code. The algorithms with the highest PPVs of 96% were 1) ≥ 4 counts of the SSc ICD-9 code and RP keyword, 2) ≥ 3 counts of the SSc ICD-9 code and ANA positive and RP keyword, and 3) ≥ 4 counts of the SSc ICD-9 code and ANA positive and RP keyword (Table 1).  The algorithm with the highest combined PPV of 81% and sensitivity of 95% was ≥ 1 count of the SSc ICD-9 code and RP keyword.     Conclusion: We have developed novel algorithms to identify SSc subjects in the EHR using not only billing codes but also labs and keywords with a PPV of 96%.  These algorithms will allow researchers to identify and study patients with SSc more efficiently and accurately in the EHR.  

Table 1.

Algorithm*

Positive Predictive Value

Sensitivity

Counts of the ICD-9 code (710.1)*

 

 

≥ 1

50%

N/A

≥ 2

68%

85%

≥ 3

82%

77%

≥ 4

90%

67%

 

 

 

ANA positive£ AND ≥ 1 count of 710.1

45 %

81%

ANA positive AND ≥ 2 counts

70%

81%

ANA positive AND ≥ 3 counts

88%

67%

ANA positive AND ≥ 4 counts

93%

58%

 

 

 

Raynaud’s phenomenon (RP) keyword AND ≥ 1 count of 710.1

81%

95%

RP keyword AND ≥ 2 counts

91%

80%

RP keyword AND ≥ 3 counts

94%

71%

RP keyword AND ≥ 4 counts

96%

67%

 

 

 

ANA positive AND RP keyword AND ≥ 1 count of 710.1

74%

78%

ANA positive AND RP AND ≥ 2 counts

94%

76%

ANA positive AND RP AND ≥ 3 counts

96%

64%

ANA positive AND RP AND ≥ 4 counts

96%

57%

 

 

 

ANA positive OR RP keyword AND ≥ 1 count of 710.1

49%

98%

ANA positive OR RP AND ≥ 2 counts

72%

90%

ANA positive OR RP AND ≥ 3 counts

89%

76%

ANA positive OR RP AND ≥ 4 counts

93%

67%

*All algorithms included at least one or more counts of the Systemic Sclerosis ICD-9 (710.1). £ANA positive defined as titer ≥ 1:80.
Disclosure: L. Jamian, None; L. Crofford, None; A. Barnado, None.

To cite this abstract in AMA style:

Jamian L, Crofford L, Barnado A. Using Electronic Health Record Algorithms to Accurately Identify Patients with Systemic Sclerosis [abstract]. Arthritis Rheumatol. 2017; 69 (suppl 10). https://acrabstracts.org/abstract/using-electronic-health-record-algorithms-to-accurately-identify-patients-with-systemic-sclerosis/. Accessed .

« Back to 2017 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/using-electronic-health-record-algorithms-to-accurately-identify-patients-with-systemic-sclerosis/