The first year of ICD-10 came and went without many complaints. The anxiously anticipated coder productivity issues were not as troublesome as many people expected, while claims denials and revenue disruptions were relatively minimal.

But October 1, 2016 marked another new era in the classification system: The Centers for Medicare and Medicaid Services year-long grace period, in which it accepted claims as long as they were submitted in the right ICD-10 family, came to a close and now CMS is demanding greater specificity — as are commercial insurers.

"The surprise was it went smoother than we anticipated," said Carol Paret, who led the ICD-10 transition at Memorial Hermann in Texas. "At the six- to-eight-week mark, we wondered, when is the shoe going to drop?"

Now, experts worry that many of the ICD-10 fears that were calmed through these various grace periods could have pushed back negative effects.

As providers must now code with the correct specificity, they still face issues in staffing enough qualified coders and getting claims out the door on a timely basis.

Memorial Hermann runs its own coding school, training and hiring its own qualified coders, which helps keep vacancies to a minimum.

[Are EHRs getting better? Readers rank vendors higher than last year in new survey]

"We did lower our coding productivity standards by 20 percent and are still down," said Paret, who is senior vice president and president of the system's Community Benefits Program. "We're still struggling a bit with documentation at the specificity we need."

Paret added that the system is making sure that claims are thoroughly checked before filing, but that still may not be enough.

"Denials as a whole are a challenge. We're expecting more when the grace period is done," she said.

Paret projects the new level of difficulty to remain through the next 18 to 24 months.

Luckily, Memorial Hermann did a lot of work upfront with payers before last year's rollout. They looked at claims paid under ICD-9, recoded them for ICD-10, and sent them to the insurers to see areas in which provider and payer agreed and disagreed.

Getting feedback from insurers as to the specific codes that will be accepted remains a challenge for most providers, said Andrew Boyd, MD, an assistant professor of Biomedical and Health Information Sciences at the University of Illinois in Chicago and an ICD-10 industry leader.

"I have not seen the insurance companies come out and say these are the codes we will reimburse on," Boyd said. "Historically they don't disclose what they reimburse for. This is not something that has been a transparent process."

This is because insurers want providers to submit claims based on the accuracy of a patient's condition, not based on which codes get paid, he said.

"If we give four codes, magically everyone will submit four codes," Boyd said, suggesting that to be common thinking among insurers. "There's a lot of money at stake."

The provider meanwhile, is looking at what is reimbursed and what is not, and the level of detail needed to get reimbursed by different insurers.

"If two codes equally describe an illness — and one gets paid and one doesn't — they'll obviously choose the one that gets paid," Boyd said.

Now that the grace period has ended, this process will have to be learned all over again, he said.

What Boyd sees as the real issue for ICD-10 is not the increase in the number of codes but that providers are not mining the data for analytics.

"Talk about information- driven healthcare," he said. "How often are systems looking at the data? Most people are not trying to understand the numbers, but to make the numbers go down."

[Special report: Precision medicine: Analytics, data science, EHRs in the new age]

Manny Pena, CEO and founder of H.I.M. ON CALL, said he also sees that if providers are getting paid, they don't want to know information beyond that, at least right now.

H.I.M. ON CALL held a national coding contest that saw a 62 percent accuracy rate for medical/surgical cases under ICD-10.

From 500 participants and 2,000 cases coded, Pena said, overall they're seeing 83 percent accuracy and 64 percent productivity in ICD-10, compared to 95 percent under ICD-9. 

"We were very surprised. Clearly we did not expect these overall low scores. Experienced people also had low scores. We knew it wasn't going to be in the '90s," said Pena, a former senior director of the health information management department for New York Presbyterian Hospital's Columbia-Presbyterian Medical Center. "Even with our own people we had to go through multiple passes. It's not like the old days of ICD-9, now with so many codes and conditions, you have to go through multiple passes."

This article originally appeared on Healthcare Finance. 

Twitter: @HealthITNews

Like Healthcare IT News on Facebook and LinkedIn