As the use of administrative databases to study acne becomes more common, accurate identification of acne patients in these datasets is essential to avoid potential misclassification bias.1–3 While a single-center Canadian study has evaluated the validity of using a single International Classification of Disease (ICD)-9 code for acne, there is a need to understand the optimal approaches to identify patients with acne using newer ICD-10 codes.4
To evaluate several strategies to identify patients with acne using ICD-10 codes, we conducted a retrospective, multi-center study among 300 patients, 12–40 years-old who were evaluated in the outpatient dermatology clinics at the University of Pennsylvania Health System and NYU Langone Health between January 1, 2016 and June 30, 2019 and had at least one ICD-10 code for acne (L70.0, L70.8, L70.9) were included. A control cohort was compromised of 300 randomly selected patients from the same clinical population who did not have an ICD-10 diagnosis code for acne.
A focused chart review was performed on each patient record to determine if there was documentation of clinical acne, which was defined as a description in the physical exam of comedones, papules, pustules or cysts or a discussion of acne in the clinical encounter note. It was also recorded whether each patient had a subsequent visit with another ICD-10 code for acne or if they were prescribed an acne medication (topical retinoid, topical antibiotic, oral antibiotic, spironolactone or isotretinoin) within 6 months of the first code for acne. A focused chart review in the control group was conducted to evaluate for any documentation consistent with acne.
The sensitivity, specificity, negative predictive value (NPV) and positive predictive value (PPV) were compared for three classification strategies: 1) one ICD-10 code for acne, 2) at least two ICD-10 codes for acne within 6 months, and 3) one ICD-10 code and a prescription for an acne medication within 6 months.
Using one ICD-10 code for acne, the PPV, NPV, sensitivity, and specificity were 0.990, 0.977, 0.977, and 0.990 respectively. Using two ICD-10 codes for acne, the PPV, NPV, sensitivity, and specificity were 1.000, 0.751, 0.615, and 1.000 respectively. Using one ICD-10 codes and a prescription for an acne medication, the PPV, NPV, sensitivity, and specificity were 0.993, 0.977, 0976, and 0.993 respectively (Table 1 and 2). However, the estimated sensitivity of using one code, two codes, and one code and a medication prescription decreased to 0.812, 0.670, and 0.804, respectively, if the frequency of encounters for acne was 15%, which is approximately the frequency of encounters for acne in the community (Supplemental Table 1).5
Table 1.
Subject Characteristics
| Acne | Control | |
|---|---|---|
| Age, years, mean (SD) | 23.9 (6.8) | 28.8 (6.8) |
| Female, % | 69.6 | 62.0 |
| Acne treatments, % | ||
| Topical retinoid | 80.7 | |
| Topical antibiotic | 33.7 | |
| Oral antibiotic | 25.0 | |
| Spironolactone | 13.7 | |
| Isotretinoin | 7.3 | |
| Non-acne diagnoses coded as acne, n | ||
| Folliculitis | 2 | |
| Perioral dermatitis | 1 |
Table 2.
Characteristics of Each Classification Approach
| Classification Approach | PPV (95% CI) | NPV (95% CI) | Sensitivity (95% CI) | Specificity (95% CI) |
|---|---|---|---|---|
| One ICD-10 code for acne | 0.990 (0.979–1.000) | 0.977 (0.960–0.994) | 0.977 (0.960–0.994) | 0.990 (0.978–1.000) |
| Two ICD-10 codes for acne | 1.000 (1.000–1.000) | 0.751 (0.708–0.794) | 0.615 (0.555–0.675) | 1.000 (1.000–1.000) |
| One ICD-10 code and medication prescription | 0.993 (0.984–1.000) | 0.977 (0.960–0.994) | 0.976 (0.959–0.994) | 0.993 (0.984–1.000) |
PPV: positive predictive value; NPV: negative predictive value
When using administrative databases, the ideal method for accurate case identification is both sensitive and specific. Our analysis suggests that using one ICD-10 code for acne or one ICD-10 code and an acne prescription may be preferred over requiring two codes for acne for acne case identification in large administrative databases. Both approaches yield a PPV and NPV >0.97. Future studies should examine whether these findings generalize to acne patients seen outside of dermatology.
Supplementary Material
Funding/Support:
Barbieri is supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health under award number T32-AR-007465 and receives partial salary support through a Pfizer Fellowship grant to the Trustees of the University of Pennsylvania.
The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interests Disclosures: The authors have no other conflicts of interest to disclose.
Institutional Review Board Approval: This study was deemed exempt the University of Pennsylvania Institutional Review Board and the NYU Langone Health Institutional Review Board.
References
- 1.Noe MH, Mostaghimi A. The Challenges of Big Data in Dermatology. J Am Acad Dermatol. Published online May 9, 2018. [DOI] [PubMed] [Google Scholar]
- 2.Barbieri JS, James WD, Margolis DJ. Trends in prescribing behavior of systemic agents used in the treatment of acne among dermatologists and nondermatologists: A retrospective analysis, 2004–2013. J Am Acad Dermatol. 2017;77(3):456–463.e4. [DOI] [PubMed] [Google Scholar]
- 3.Barbieri JS, Choi JK, Mitra N, Margolis DJ. Frequency of Treatment Switching for Spironolactone Compared to Oral Tetracycline-Class Antibiotics for Women With Acne: A Retrospective Cohort Study 2010–2016. J Drugs Dermatol JDD. 2018;17(6):632–638. [PubMed] [Google Scholar]
- 4.Ejaz A, Malaiyandi V, Kim WB, Rogalska T, Alhusayen R. Validating the diagnostic code for acne in a tertiary care dermatology centre. Eur J Dermatol EJD. 2015;25(5):469–471. [DOI] [PubMed] [Google Scholar]
- 5.Wilmer EN, Gustafson CJ, Ahn CS, Davis SA, Feldman SR, Huang WW. Most common dermatologic conditions encountered by dermatologists and nondermatologists. Cutis. 2014;94(6):285–292. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.