Billing Codes Could be Useful in Capturing DME in Patients with NPDR

July 27, 2020

In research presented at ASRS 2020, investigators found 56% of patients with PDR had a DME code, with a sensitivity of 85%.

Suzanne Michalak, MD

Billing codes are becoming increasingly used for non-financial purposes in clinical trials, including determining the prevalence of disease, investigating healthcare outcomes and allocating resources.

However, there are very few studies validating the accuracy of ICD-10 codes for non-billing purposes.

A team, led by Suzanne Michalak, MD, Duke University Eye Center, validated the accuracy of the International Classification of Diseases, Tenth Revision-Clinical Modification (ICD-10) against clinical charts for diabetic macular edema (DME) in findings presented at the American Society of Retina Specialists 2020 (ASRS 2020) Virtual Sessions.

The investigators selected a random subset of 100 encounters for diabetic retinopathy at an academic medical center between October 2018 and January 2019 for review. The researchers excluded patients with a diagnosis of neovascular age-related macular degeneration or retinal vein occlusion.

The team calculated the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPE) for diabetic macular edema and assessed various procedures including optical coherence tomography (OCT) and intravitreal injection on their impact on PPV and NPV.

The patient population included 50 individuals with non-proliferative diabetic retinopathy (NPDR) and 50 patients with proliferative diabetic retinopathy (PDR). The average age of the patients was 58 years old, ranging from 27-86. Overall, 62% of patients had an ICD-10 code for diabetic macular edema.

The DME code had an overall sensitivity of 93%, specificity of 61%, PPV of 65%, and NPV of 92%. Of the individuals with NPDR, 30 (60%) had a DME code with a sensitivity of 97%, specificity of 75%, PPV of 85%, and NPV of 94%.

On the other hand, of the patients with PDR, 28 (56%) had a DME code, with a sensitivity, specificity, PPV, and NPV of 85%, 54%, 39%, and 91%, respectively.

A total of 84 patients obtained a same-day OCT, but this did not improve the PPV of the DME code. An intravitreal injection was administered in 16% of patients with DME within 1 month of the visit, which improved the PPV of a diabetic macular edema code to 86%.

There were also 6 of 22 false positive DME codes resolved of patients with a history of DME.

“Billing codes are more sensitive and specific for capturing DME in NPDR than PDR,” the authors wrote. “Reasons include increased use of intravitreal anti-VEGF agents for PDR with associated DME codes and more prior encounters for PDR patients, increasing the chance of carrying forward a code. Further study on whether associated procedure and diagnostic codes increase the accuracy of ICD-10 codes for DME is needed.”

In research also presented during ASRS 2020, researchers compared the efficacy and adverse effects of triamcinolone acetonide and dexamethasone in a real-world setting to treat diabetic macular edema and determined which corticosteroid is more effective in the long-term management of DME.

To treat diabetic macular edema, clinicians often administer corticosteroids when laser photocoagulation and anti-VEGF agents become ineffective. Triamcinolone acetonide is an injection administered every 3 months, while dexamethasone is an implant that can last up to 6 months.

In the retrospective study, the investigators examined 40 eyes from 32 patients treated for DME at a tertiary care center. Of the eyes treated, 24 were treated triamcinolone 2-4 mg and 16 with dexamethasone 0.7 mg implant. Each patient was followed for up to 12 months.

The study, “Validation of ICD-10 Diagnostic Codes for Diabetic Macular Edema,” was published online by ASRS 2020.