In October 2015, physicians across the United States transitioned from the International Statistical Classification of Diseases and Related Health Problems, Ninth Revision to the tenth revision (ICD-10-CM, the US version of the World Health Organization [WHO] ICD-10). Although the ICD-10-CM was a new concept for physicians in the United States, the international variant has been available since the early to mid-1990s. One of the main differences between the ICD-10-CM and the ICD-10 is the sheer number of diagnostic codes included in the US classification system: the ICD-10 has approximately 14,400 codes, whereas the ICD-10-CM has more than 144,000. The ICD-10-CM is so granular that it has been described as needlessly specific, absurd, and unnecessarily detailed. Some argue that this seemingly excessive detail, in theory, allows for more accurate labeling of diagnoses required both for billing and observational analysis.1
However, the value and utility of these data are highly dependent upon how accurately and effectively the classification system is used.2 As many opponents of the ICD-10 point out, in clinical practice coding is often “inconsistent, inaccurate, and incomplete.”1 In 2015, as the United States was rolling out its cumbersome version of the ICD-10, the National Academy of Sciences published a report aimed at improving diagnosis in healthcare in response to the statistic that 1 in 20 adults in the United States experiences a diagnostic error every year.3 At the beginning of 2019, nothing had yet been done to improve the situation; in fact, it may have been exacerbated by the ICD-10-CM.
The ICD-10-CM classification also does not allow clinicians to express “clinical concern” when there is insufficient, incomplete, or inconclusive evidence to support a firm diagnosis.2 Moreover, because the codes are collected for billing purposes, some argue that their use in research is “intrinsically flawed.”1,2,4 For example, if a clinically relevant ICD-10 code is not useful for billing purposes, it is likely to be left out of the coding altogether. On the other hand, a useful billing code, despite being clinically irrelevant, may prompt coders to ask physicians to add the code and amend the clinical documentation.2 Along those lines, physicians are often interrupted and forced by the electronic medical record to select a code that may not accurately represent the medical issue just so they can move on with their work. These interruptions are further exacerbated by impediments to efficient workflow such as expired code warnings, lack of coverage warnings, specificity prompts, retrospective prompts, or an inability to find the right code.2
Given the unparalleled dependency on ICD-10 coding for reimbursement, it is no surprise that diagnostic errors occur and that diagnostic accuracy is lacking. In an article published in Health Affairs, Robert Berenson, a fellow at the Urban Institute in Washington, DC, and Hardeep Singh, MD, chief of the Health Policy, Quality, and Informatics Program, Center for Innovations in Quality, Quality, Effectiveness, and Safety, Michael E. DeBakey Veterans Affairs Medical Center, and professor of medicine at Baylor College of Medicine, Houston, Texas, argue that current models fail to reward physicians for “quality or value of care.” They point out that diagnostic errors are difficult to identify in real time and current fee-for-service models exacerbate the problem by tolerating, and in some cases, even encouraging diagnostic errors.3
For example, some clinicians might take advantage of bundled payments for diagnoses requiring more extensive care, and consequently better pay, in patients with less severe illness requiring less care. Maybe a more prevalent scenario is patients with elusive diagnoses that require a more thoughtful and often a multidisciplinary approach. Current models encourage higher turnover and do not reward clinicians for taking their time to think through the diagnosis, brush up on the latest literature, and consult with colleagues. In the current assembly line, patients might pick up inaccurate diagnoses required for billing the office visits, tests, referrals, and procedures that do not accurately reflect the final conclusions. More troublesome is that current models allow for redundant testing and sometimes even inappropriate testing and procedures.
In some cases, such as cancer diagnoses, misdiagnosis is common, and patients often benefit from a second opinion. In fact, up to one-third of cancer diagnoses are missed on initial presentation. However, some health plans might establish hurdles that prevent or at least make it difficult for patients to seek out second opinions. One possible solution is to support team-based centers that aim to improve the diagnosis of challenging conditions. In the United Kingdom and Denmark, for example, there is a push for multidisciplinary centers aimed at the rapid and simultaneous assessment of multiple cancers. These centers offer an accurate cancer diagnosis or an “all-clear” within 2 weeks. By limiting the conditions the center focuses on, generously reimbursing the multidisciplinary team approach, and concentrating experts into a center, they are able to improve accuracy and speed up the process of making a diagnosis. In comparison, under the current standard of care in the United States, it can take several weeks to months of seeing multiple physicians (at different institutions and who do not communicate well with each other) just to establish a diagnosis. Even then, the diagnosis may still be incorrect.
Berenson and Singh argue that a possible solution to inaccurate diagnoses is changing our current payment models. Recognizing this issue, the US Department of Health and Human Services recently proposed new rules for the upcoming 2019 Medicare Physician Fee Schedule that would reduce the current 4 levels of care for outpatient visits to a single level for new patients and a single level for follow-up visits.3 However, as the investigators pointed out, this strategy is likely to exacerbate the problem because now physicians will receive the same payment for a 5-minute encounter as they would for a 35-minute encounter. Where is the incentive in this proposed fee schedule for a clinician to take their time to get a diagnosis right? Instead, the proposed changes are likely to shorten the already brief patient encounters. One strategy is to move forward with payment models that promote accountability for diagnostic performance.3 An obvious model that comes to mind is that of Accountable Care Organizations (ACOs), in which providers share the cost savings when they reduce the total cost of care. While there are some performance metrics used, current ACO models do not necessarily or directly reward diagnostic accuracy, although they may indirectly do so in improved quality care metrics.
Finally, in condition-based payment models adopted by the Center for Medicare and Medicaid Innovation, there is concern that the lack of diagnostic accuracy predisposes these models to abuse.3 For example, in oncology, payment streams designed for patients with more severe forms of a qualifying diagnosis may be claimed by physicians for milder cases of that cancer for which the payment stream was not intended. The problem lies in that the payment is triggered by the initial claim for chemotherapy with little follow-up to verify the accuracy of the diagnosis and the level of severity.3 One possible solution is to build multiple opinions into the diagnostic process for conditions that are based more on expert opinion rather than objective confirmatory evidence.3
These solutions share a common theme: they all propose changes to how physicians are paid for the care they deliver. I have previously argued that it may be time to uncouple documentation from billing, which I believe would allow physicians to focus more on accurate documentation than on billing. Unfortunately, in our current healthcare model, which is so heavily reliant on documentation for billing, it is unclear how to uncouple the tasks and fairly compensate physicians for their time and efforts. It is worth considering whether a universal or single-payer healthcare system might more easily offer answers to improving diagnoses and reducing diagnostic errors. I point to the specialized cancer centers in the United Kingdom and Denmark as examples of how a single-payer system might improve not only diagnostic accuracy but care for more challenging conditions.
- Weiner, MG. POINT: Is International Statistical Classification of Diseases and Related Health Problems, 10th Revision Diagnosis Coding important in the era of big data? Yes. Chest. 2018;153(5):1093-1095.
- Liebovitz DM, Fahrenbach J. COUNTERPOINT: Is International Statistical Classification of Diseases and Related Health Problems, 10th Revision Diagnosis Coding important in the era of big data? No. Chest. 2018;153(5):1095-1098.
- Berenson R, Hardeep S. Payment innovations to improve diagnostic accuracy and reduce diagnostic error. Health Aff (Millwood). 2018;37(11):1828-1835.
- Manchikanti L, Kaye AD, Singh V, Boswell MV. The tragedy of the implementation of ICD-10- CM as ICD-10: Is the cart before the horse or is there a tragic paradox of misinformation and ignorance? Pain Physician. 2015;18(4):E485-E495.