Statistical Modeling for Studying the Impact of ICD-10 on Health Fraud Detection

Introduction

The implementation of the electronic health record (EHR) has allowed healthcare organizations to collect and externally report/provide a greater amount of data to the public sector. The EHR is a systematic collection of electronic health information about an individual patient or population, and can include a range of data, including demographics, medical history, medication and allergies, immunization status, laboratory test results, radiology images, vital signs, personal statistics, and billing information. Fraud detection research in the field of healthcare management has primarily concentrated on medical billing data, which is derived from the EHR. After a patient’s medical record is updated by a physician or staff member, diagnosis and procedure codes are assigned by a medical coder. The appropriate medical codes and necessary data from the EHR are incorporated into an ANSI 837 file, which is submitted to the payer directly or via a clearinghouse. An insurance company is usually the recipient of this claim file, so the majority of appropriate fraud detection data comes from health insurance agencies.

Healthcare fraud literature originating from countries outside of the United States has used a variety of sources to acquire medical claims data. The National Health Insurance Administration (NHIA) in Taiwan has provided data to multiple research groups (Chan & Lan, 2001; Hwang, Wei & Yang, 2003; Liou, Tang & Chen, 2008; Wei, Hwang & Yang, 2000; Yang & Hwang, 2006), and an analogous NHIA program in South Korea, the National Health Insurance (NHI) system, has also contributed to studies (Shin, Park, Lee & Jhee, 2012). Two major Australian governmental health departments, the Health Insurance Commission (HIC) and Medicare Australia, have been reported as the data sources in numerous pertinent research projects (He, Graco & Yao, 1999; Hubick 1992; Shan, Jeacocke, Murray & Sutinen, 2008; Shan, Murray & Sutinen, 2009; Tang, Mendis, Murray et al., 2011; Williams, 1999). Healthcare claims data from private insurance companies located in Turkey and Chile has also been used by several researchers (Kirlidog & Asuk, 2012; Kumar, Ghani & Mei, 2010; Ortega, Figueroa & Ruz, 2006).

In the United States, various agencies within the Department of Health and Human Services (HHS) have been involved with research exploring the detection and prevention of healthcare fraud. The Centers for Medicare and Medicaid Services (CMS) and its predecessor, the Health Care Financing Administration (HCFA), both supplied members of academia with Medicare and Medicaid data (GAO 1996; Liu & Vasarhelyi, 2013; Musal, 2010; Shapiro, 2002). Researchers in the United States have also worked with private insurance companies and hospitals, using data collected by these organizations (Major & Riedinger, 2002; Ngufor & Wojtusiak, 2013).

Today, when an individual is seen or treated by a healthcare professional, a series of alphanumeric codes are still used to describe the medical diagnoses and services provided. This designated classification structure, the ninth iteration of ICD (International Classification of Diseases), implements the use of coding for healthcare management, public health and medical informatics, and insurance purposes. ICD-9 has been the coding standard in the healthcare industry since October 1^st, 1984. The primary purpose of ICD-9 is to translate written information from a patient’s clinical statement regarding diagnoses and inpatient procedures into a series of universally understandable designations.