Home  |   Subscribe  |   Resources  |   Reprints  |   Writers' Guidelines

E-News Exclusive

Assessing Your ICD-10 Coder Performance —
Four Steps to Move From Predictions to Certainty

By Paul Strafer, RHIA, CCS

It’s the question on every coding manager’s mind: How can one truly know how coders will perform once ICD-10 goes live? Much is banking on this answer. Claims denial rates are predicted to rise 100% to 200% in the early stages of ICD-10. Coders are expected to be up to 50% less productive and diagnosis-related group (DRG) shifts are predicted. The good news is that the Centers for Medicare & Medicaid Services’ end-to-end testing results thus far in 2015 have been positive for ICD-10 coding accuracy—only 3% of claims were denied due to ICD-10 coding errors during the January 2015 testing period.

With only two months ahead, now is the time for HIM directors to move from these early coding predictions to today’s coding reality. This article provides four tips to efficiently and effectively assess coder performance in preparation for October 1.

Why Coder Assessments Are Important
During these lean times in which ICD-10 budgets have been stretched to the maximum, coding managers need efficient and effective ways to assess performance so they can make correct staffing and training adjustments. Coding assessments ensure every dollar spent on ICD-10 education is a dollar well spent. They also enable managers to make more informed decisions regarding workload distribution, aligning ICD-10 auditors, and identifying coders who can specialize in certain procedures or diagnoses.

Specific benefits of conducting ICD-10 coder accuracy and productivity assessments include the following:

• Identify specific coder knowledge gaps and areas of team weakness.

• Target areas for coder refresher training and physician education opportunities.

• Recognize individual coder strengths for case assignment and coder specialization.

Dual coding is considered to be the best starting point to achieve these goals. However, because it’s difficult to know exactly how coders will perform when placed in a live ICD-10 production environment, the crystal ball remains cloudy. There are technical and logistical challenges involved in accurately assessing coders and establishing real-world ICD-10 coder benchmarks.

Specific Challenges
• Difficulty predicting the effects of certain variables. These include—but aren’t limited to—software glitches (eg, encoder problems or computer-assisted coding errors), a rise in queries, and claims submission challenges. Although some of these variables can be thwarted by proper planning and testing, it’s just not possible to identify—and proactively remediate—everything that could go wrong.

• Difficulty comparing ICD-9 and ICD-10 productivity. Dual coding provides a glimpse into how long it will take coders to code in ICD-10, but it’s only a glimpse. This is because of the “read-time” factor for each case. If coders assign an ICD-9 code first, they’ve already technically read and reviewed the chart. This first review makes coding the same case a second time, in ICD-10, more efficient. Likewise, if coders code in ICD-10 first, they require less time to code the same chart a second time in ICD-9. Read-time for cases makes it difficult for managers to calculate a true “apples-to-apples” comparison.

• ICD-10 experts don’t exist. Since nobody is an expert, how can facilities know for sure whether the auditing they perform on ICD-10 coded cases is correct? This is a big challenge for hospitals trying to assess performance, provide meaningful feedback, and set up an effective ICD-10 coding compliance program.

Many organizations appoint their lead coding expert in ICD-9 as their primary coding expert in ICD-10. This individual likely has the research skills and auditing abilities to excel. One major disadvantage is that this individual must be removed from current production in order to review cases in ICD-10.

• A manual tracking process is time-consuming. It takes time to review each and every chart coded in ICD-10. Four to five days are typically required for an ICD-10 coding manager to score and analyze accuracy and productivity results from coder training when real cases and actual clinical documentation are used—vs check-the-box tests within existing ICD-10 training applications.

Managers also must subsequently provide one-on-one feedback to coders (ideally immediately after completion of the exercise), and track performance over time. The larger the department, the more difficult this task becomes.

Despite these challenges, conquering coder assessments and building a solid ICD-10 coding compliance program is essential.

Four Steps to Gain Clarity About Coder Performance
Here are four tips to gauge real-time production in ICD-10, and ensure accurate and effective coder assessments.

1. Use your facility’s own medical records. The most realistic picture of coder production in ICD-10 can only be gleaned when using your facility’s own documentation. This may include a combination of handwritten notes, electronic notes, and scanned information.

Use today’s documentation to pinpoint areas for additional queries or clarification in ICD-10. The documentation your coders see today is likely what they’ll see in ICD-10. Even as clinical documentation improvement efforts continue to focus on ICD-10 specificity, documentation improvement takes time, and significant changes aren’t likely to appear by October 1.

2. Know what data to track. When assessing coder performance, pay attention to the following:

• Coding accuracy—Did the coder code both the diagnoses and procedures correctly?

• DRG accuracy—Did the DRG shift as compared with ICD-9? If so, track the reason. Did the coder assign the principal diagnosis correctly? Did the coder omit certain complications and comorbidities (CCs) or major CCs, or report CCs or major CCs that weren’t clinically validated? Tracking the reason for the DRG shift identifies whether the coder is struggling with the application of a particular guideline and not necessarily ICD-10 code assignment, or if the shift is a legitimate result of ICD-10 guidelines vs ICD-9.

• Trends in coding errors by code category, subcategory, and individual code. Do multiple coders struggle with the same topic or code?

3. Simulate a live environment. Consider the following techniques:

• Set an ICD-10 productivity standard, and hold coders accountable to it. If managers use ICD-9 standards, they’ll quickly be disappointed. Assume that productivity will be cut in half. Let coders know that this is your assumption. Encourage them to meet—and exceed—this standard when practicing.

• Require coders to use the coding book. All coders should double-check the encoder when coding in ICD-10. Asking them to do this while practicing prepares them for this task.

• Go live with ICD-10 templates and queries now. This gets coders and physicians accustomed to using them.

4. Make the most of your training/tracking software. Some ICD-10 educational software includes the following time-saving features to maximize coding manager productivity during coder scoring and assessments:

• a comprehensive training library using real medical records;

• a consistent answer key across the coding team, with flexibility to adjust according to organization-specific coding guidelines;

• immediate feedback indicating whether the record was coded correctly (including why the answer is correct or incorrect); and

• the ability to track and trend data by individual or across the entire department, making education more cost-effective and targeted.

Coding integrity is important today, and it will be even more paramount in ICD-10. Accurate coding enhances accurate reimbursement and helps organizations reap the long-term benefits of ICD-10. If your coders aren’t prepared, your organization isn’t prepared.

Coder assessments are a critical part of ICD-10 success. What’s your strategy, and will it help you accomplish the most efficient and effective coder training over the next few months?

Paul Strafer, RHIA, CCS, is coding and education manager at H.I.M. On Call.