Does your data quality management include improvement processes?
If not, here are some tips on getting started
Data quality assessment and management is a lot easier when you have a database that contains information from multiple locations, such as information submitted by large health care networks.
Donna Fletcher, MPA, RHIA, data quality manager of Child Health Corporation of America (CHCA), has access to a database that includes more than 30 hospitals, including some of the largest nonprofit pediatric hospitals in the United States. This wealth of information from CHCA’s owner hospitals has enabled the organization to incorporate solid data quality assessment and management into its performance and clinical improvement processes.
Fletcher has shared guidelines on establishing such a process with other HIM professionals, including speaking about the subject at the 74th National Convention and Exhibit of the Chicago-based American Health Information Management Association (AHIMA), held Sept. 21-26 in San Francisco.
CHCA, based in Overland Park, KS, is a business alliance of 38 children’s hospitals nationwide, and the database used by Fletcher has detailed information on everything from the dosage of drugs prescribed to X-rays ordered.
When a particular hospital’s data appear to be out of the ordinary, Fletcher will investigate its practices and find out why the facility is doing things differently from its peers. Through her investigations and tracking of outcomes, she has developed strategies for improving data that are used for research and clinical purposes, as well as for reimbursement. Here are some of her recommendations:
• Draw upon any applicable database. HIM professionals working for smaller hospital systems may be able to use a state database for comparison purposes, Fletcher suggests.
First, they should make certain the state database has clinical data, diagnoses, and procedure codes, all of which might be found on billing documents. "These will allow you to make sure coding is consistent," Fletcher says.
Also, it’s a good idea to learn how conditions are defined in a particular region. In some places, physicians may describe a patient as having angina, whereas doctors in another region might describe the same condition as a heart attack, Fletcher says.
If the hospitals listed in a benchmarking database use different definitions from your own hospital, the coding will be different and not comparable, Fletcher says.
"So even if you have a patient with the same condition, you’ll have codes for one for an angina and codes for another for a heart attack," she says.
Since this sort of definition discrepancy could be true even within one hospital system, it’s wise to make certain there’s a process in place to ensure that all physicians define them the same way, Fletcher adds.
Coding roundtables address problems
• Ask physicians to explain coding issues. Coders often come across some description or definition that seems strange. When this occurs, it’s a good idea to have physicians meet with coders to explain why they are using unusual terms.
"We’d have coding roundtables where a physician would come in and tell us what was meant by a particular item," Fletcher says. "If there is any discrepancy in what the physician says, then we’d take it to the medical staff and ask for guidelines on coding that diagnosis or condition."
Those guidelines would be passed out to coders for future reference. They’re also distributed to everyone who is using the data, including clinical staff, she adds.
"The people who are collecting data and assigning codes need the information the most," Fletcher says. "However, anyone using the data needs to know the limitations and parameters of data before they draw conclusions."
• Be proactive in developing coding guidelines. Bring physicians together with HIM staff to discuss coding guidelines, Fletcher suggests.
"Have health information management folks ask physicians, How do you code XYZ?’" she says.
Or, when there’s some discrepancy in how clinicians and HIM staff believe a condition or diagnosis should be coded, the group will work on developing a consensus that can be put into guidelines.
Those guidelines are documented and made available to everyone through e-mail or posting them on a facility’s web site, Fletcher says.
"You can use the guidelines as a training tool for coders," she adds. "Any kind of reference documentation or coding clinic guidance also are included in the document, so people can follow up and check how we reached that consensus."
• Take a closer look at absent codes or codes used disproportionately by a particular physician or facility. Whenever you find this pattern, don’t automatically assume either overutilization or underutilization, as the codes might lead you to do. This kind of irregularity in code distribution often is an artifact of how facilities code various services.
Fletcher once discovered this kind of problem with the drug nitrous oxide, which is administered by respiratory therapy. She noticed that some hospitals appeared to be using the drug consistently, while others appeared to be not using it at all over a two-year period. That omission suggested to Fletcher that the drug really was being used, but that it was being coded in some odd place. After doing some digging, Fletcher learned that some hospitals coded the drug with respiratory therapy, while others coded it in pharmaceuticals.
"So we had our hospital come to a consensus on where it should be coded, and then we remapped or reassigned codes so that now everyone is coding it in the same area," Fletcher says. "They decided it would be coded in pharmacy because it’s administered hourly."
HIM directors should encourage their staff and others in the hospital system to use the coding data for quality purposes. "Our organization tries to develop ways for our members to use data so that when they find something inconsistent, they’ll bring the information to me," Fletcher says.
It’s when a hospital’s data are heavily used that data quality issues arise, Fletcher notes.
• Prioritize data elements, but work toward error-free documentation. "We have an advisory group that says these elements are so important that if the hospital doesn’t have them at 100%, it’s an issue," Fletcher says. "So each year I assess quality and completeness for those data elements, and our members then have information by data element."
Using these complete data, Fletcher is able to compare hospitals according to such details as admission hours.
By setting data collection priorities and reviewing them for errors, HIM professionals will find problems that previously were overlooked.
For instance, one hospital’s data included periods in the ZIP code numbers, Fletcher discovered during the course of a review.
Fletcher investigated why this punctuation mark was appearing erroneously in a data element, and then she asked the hospital to take corrective action to eliminate it.
The dot in the entry field wasn’t easy to discover, and the hospital itself had no idea why it was appearing there. "Sometimes you have to look at a printout of a list to see what looks weird," Fletcher says. "You have to do research for completeness, meaning there’s a value in every field and it’s valid.
"What I’m looking for is any one of the data elements that are not right for any reason," Fletcher says. "It’s important to have this level of precision now because of the number of databases that this information is fed into."
Even when HIM departments choose not to do benchmarking comparisons of coding data, it’s very likely that someone else is doing these comparisons, Fletcher notes.
"If the data are not consistent and accurate in your own facility, then that is a poor reflection on your hospital," Fletcher says. "If you’re singled out in any way, then that may be an issue for you, so the goal is 100% accuracy."
• Keep open communication lines with physicians. "Whenever you’re looking at an issue, look at the high-volume cases where you’re going to get the biggest bang for your bucks," Fletcher says. "Then find the population, obtain information, and routinely ask the physician head of the medical staff or someone else to look at it."
Show the medical expert the coding book and ask whether there is anything the hospital’s coders should do differently.
"Our physicians were very happy to do this, and we had no problem with asking for their help," Fletcher says.
Doctors were asked what needs to be documented to support certain diagnoses and procedure codes among the high-dollar cases, as well as for high-volume cases, she adds.
• Accept feedback from outside sources. Another strategy that has worked well for CHCA is an area-wide coding roundtable where coders from a CHCA hospital meet with coders from six or seven other hospitals in the area to discuss coding issues and provide coders with continuing education.
Also, HIM departments often will be alerted to data quality problems from the data warehouse vendor, who will send an error report back to the hospital, Fletcher says.
"I would prefer people to be more proactive, and if you are collecting data, then proactively review data to make sure it’s what you wanted," Fletcher says. "Most databases have an error threshold, so if you’re over the threshold it won’t accept the data, but if you’re under it, then there will be some errors that are not discovered."
But if a hospital first learns of a data problem from a vendor’s report, then the HIM staff should review the report and make quality improvements, she adds.