How Analytics Can Help with Chart Chasing

Introduction

Record retrieval and review, or “chart chasing “, can be a time-consuming and resource-intensive effort for both health plans and providers. Below we discuss reasons why this painful process is necessary, how and when it is done, and how to make it a more efficient process.

What is Chart Chasing?

“Chart chasing ” is a catch-all term for the various processes commercial and Medicare Advantage (MA) plans use to retrieve medical records from providers.

The logistics of chart chasing vary substantially depending on how the providers are related to the plan. For example, a closed-panel health maintenance organization (HMO) may have direct electronic access to patient medical records. This is particularly the case if all providers on the panel are employees of the HMO and utilize the same health information technology systems to capture and store medical records. Other types of health plans, such as preferred provider organizations (PPOs), may work with groups of contracted physicians and may also cover out-of-network services. In this case, the plan is less likely to have direct electronic medical record access. Requests may need to be made between the plan and provider by electronic messages (e.g., email or a web interface) and even fax. Phone calls may also be needed to confirm how the request should be processed or to get status updates on the request.

Once charts are collected, the needed data elements (e.g., supporting information for a diagnosis code) must be identified and captured from either the paper record or PDF/electronic record. Some health plans hire a vendor to assist in collecting the chart and/or extract the needed data elements. Having access to technology that can digitize the chart and identify different data elements, particularly supporting medical documentation for various diagnoses codes, can help increase efficiencies. Such technology generally involves machine learning and natural language processing, which can assist in converting unstructured text data into a more usable format. Once in a more usable electronic format, it should be possible to link different charts for the same enrollee, as well as to link charts to medical billing/encounter data.

Why is Chart-Chasing done?

There are different reasons that medical records need to be shared between the plan and provider. The term chart-chasing is usually reserved for situations when it is necessary for the plan to correct the risk score accuracy of its enrollees by closing gaps in reported diagnoses and diagnoses supported in medical record documentation. A prime example occurs when a diagnosis for a chronic condition mapped to a hierarchical condition category (HCC) in prior years is not reported in the current year. Some plans also use predictive modeling to identify conditions that may be applicable even if they have not been documented in past years. These may be conditions that have a high likelihood of being present based on an enrollee’s age and gender, as well as their drug utilization, lab results, or past medical procedures.

Some degree of chart chasing may also be necessary to confirm services rendered during a medical visit were covered under the enrollee’s benefit design. However, in these cases, providers may be more motivated to furnish charts and other information needed to support reimbursement of covered services.

When is chart-chasing done?

Chart-chasing is generally done on a retrospective basis. That is, a plan will retrospectively conduct predictive modeling and other data mining to identify enrollees with a high likelihood for medical conditions not reported on medical records in the current year. In addition, the plan may look for cases where conditions are on the billing or encounter data but were not supported by evidence in the medical records to date (or vice versa). Then, with this list of potentially under-documented conditions, health plans will pull all the medical records for the enrollee, review them, and engage the applicable providers for appropriate corrective action.

For plans with advanced analytics capabilities, including natural language processing and machine learning, and access to electronic medical records (EMRs), it may be possible to conduct concurrent coding or assist in the charting on a prospective basis. In the concurrent case, plans design workflows such that computer-assisted coding capabilities can follow provider EMR submissions on a near-real-time basis, ensuring coded information is accurately represented in the claims. This helps reduce unnecessary retrospective chart reviews and can minimize provider burden associated with retrieving charts. In addition, given the timeliness of the intervention, care management teams are more likely to have access to accurate clinical data when it matters the most to the patient.

In the prospective case, tools can be deployed within the EMR systems so that providers can be prompted with relevant information at the point-of-care to get the coding right the first time around. Such information could include a list of previously noted chronic conditions associated with the patient, as well as relevant lab tests, medications, or other services. This information would be particularly relevant during a patient’s annual wellness visit.

What are common problems faced when chart-chasing?

The immediate challenge plans face is access to consistent and complete data. For example, plans which do not have direct access to enrollees’ records experience poor retrieval rates due to something as minor as the lack of consistency between the medical provider’s contact information and the actual location of the medical records. This can occur for several reasons. For example, the rendering provider may be associated with multiple service locations, not all of which would have access to the required medical records.

Another issue is provider education regarding relevant documentation requirements. It is one thing to have documentation and another for it to meet the guidelines required for payment. For example, to support an HCC, clinical documentation in the patient’s health record must not only identify the presence of the condition, but also indicate the qualified provider’s assessment and/or plan for management of the condition. Documentation errors are also an issue. These can be as simple as coding transposition errors that may result in the wrong diagnosis being submitted on a claim or actual incorrect notation on a medical record. In either of these cases – incomplete information or erroneous information – the effort to retrieve required data remains incomplete and necessitates pulling more information or correcting what is found. Depending on scope and scale, this quickly consumes resources.

Some states have also passed laws to protect patient privacy, such as by requiring patient authorization for information related to mental health. This can make it challenging to get medical record documentation around related conditions. In addition, plans may have difficulties obtaining records related to out-of-network referrals to specialists with whom they have no contractual relationship.

Moreover, not only does chart-chasing require high labor and transaction costs in the direct retrieval activity, but there are also substantial indirect costs associated with adapting to ever-changing provider workflows and vendor relationships.

The good news is there are now solutions facilitating data access and retrieval, and they will continue to get better as adoption of standards and interoperability expand.

What are recommendations to improve the chart-chasing process?

For plans that must conduct a substantial amount of retrospective chart-review, first and foremost it is recommended to target the effort as much as possible. Targeting highest risk diagnoses, best patient records, or likely to be under-reported conditions are all examples of strategies plans can use to maximize use of resources and minimize exposure to the inevitable common problems described above. Likewise, plans can use analytics to identify providers most likely to underreport to concentrate training efforts. Targeting improves efficiency of the process, increases the return on investment, and reduces provider abrasion.

In addition, plans should aim to incorporate as many prospective coding capabilities as possible. Assisting providers with computer-assisted coding capabilities can help reduce errors and burden on providers, both during initial coding as well as by minimizing the need for future chart pulls and reviews. Such prompts could even be useful in creating proactive requests for relevant medical record documentation from out-of-network providers in the referral.