Data integrity ensures reliability of outcomes

Here are tips to correcting collection mistakes

When a Kentucky home care agency received its first report based on data collected with the Outcome and Assessment Information Set (OASIS), it looked as though the agency needed to improve its dyspnea care.

"We followed steps to identify the proper standard of care and to determine how far away we were from that standard," says Marly Auerbach, RN, BSN, MPH, director of quality management for Lifeline Home Health Care in Somerset, KY. The freestanding agency serves the rural southeastern and southwestern portions of Kentucky.

The agency is part of the OASIS project, officially called the National Medicare Quality Assurance and Improvement Demonstration, conducted by the Center for Health Services and Policy Research (CHSPR) in Denver.

The agency decided to hold a mini-refresher course for staff and develop a standard of care for each visit during a short period of time, Auerbach says. But it didn’t work because the problem wasn’t only in patient care. It was also in the data collection. It turned out the nurses didn’t understand what dyspnea was and were identifying too many patients as having dyspnea.

"There was a lot of inaccuracy in reporting which patients needed this level of intervention, so we had an overinflated count," Auerbach says. "And consequently, those people weren’t getting better because they didn’t need to."

For example, patients who had temporary shortness of breath were being diagnosed with dyspnea. These brief episodes might have occurred after exertion. "But it’s not the same thing as a cardiac patient who seeks to get dressed in the morning and is huffing and puffing," Auerbach says.

The agency held meetings, asked staff for input, and again emphasized education. Auerbach says they simplified the agency’s teaching tool for dyspnea, making it one page of very practical instructions. (See dyspnea teaching tool, p. 176.) The changes appear to be working, although the agency will not know until the May OASIS report whether dyspnea outcomes have improved, she says.

SunPlus Home Health in San Diego, which also is part of the OASIS demonstration project, has had a similar data collection problem. Quality managers found that nurses sometimes wrote down the wrong diagnosis for patients who were readmitted to the hospital, says Estelle Wolf, RN, former director of professional services for the freestanding agency that has about 6,500 visits a month.

For example, if a patient had a primary diagnosis for wound care, but had returned to the hospital for treatment of congestive heart failure, the nurse might write that the patient was admitted to the hospital for wound care. This would skew the wound care outcomes numbers on the OASIS report, Wolf says. (See story on how SunPlus solved data collection problems, p. 174.)

Home care agencies often have problems with collecting accurate OASIS data, says Peter W. Shaughnessy, PhD, professor and director of CHSPR.

"One of the biggest issues is data accuracy, and that’s going to be an issue nationally," Shaughnessy says. "If data are not of sufficient integrity, then all this reporting and so on will not help agencies out one bit."

Shaughnessy was the principal investigator with the center’s work in developing OASIS and the Outcome-Based Quality Improvement (OBQI) process. The Baltimore-based Health Care Financing Administration (HCFA) and the Robert Wood Johnson Foundation in Princeton, NJ, funded OASIS and OBQI.

Audit data internally

As Lifeline Home Health Care and SunPlus Home Health quality managers discovered, sometimes you will not realize you have a data collection problem until after you receive an outcomes report. "It’s very important to establish a program of internally auditing data and encouraging staff to make sure the information is accurate and that they don’t leave items off and so on," Shaughnessy says.

One way to ensure quality data is to follow a set process in handling OASIS outcome reports. Shaughnessy and CHSPR researchers Kathryn Crisler, MS, RN, and Karin Conway, MBA, RN, spoke about how to handle OASIS data and outcome reports at the recent Washington, DC-based National Association of Home Care (NAHC) conference held in Atlanta.

The research center has had experience in handling OASIS collection and analysis for the past few years. Shaughnessy and Crisler told quality managers who attended the NAHC conference there are some basic steps they should take to improve the entire data collection and reporting process. Here are their suggestions:

1. Commit to making the most of OASIS collection.

Quality managers should make sure their employees understand what OBQI is, what outcomes are, and how they can be used, Shaughnessy says."One of my concerns with the national program is that agencies will say, I’ll do just what I have to do and no more’ to meet the letter of the Medicare mandate," he says. Agencies that take this attitude will have difficulty making quality improvement projects out of their reports, he adds.

Shaughnessy says he empathizes with the burdens carried in recent years by the home care industry. And some agencies look at OASIS with the same amount of disdain they view other government impediments thrown at them, he adds. "It’s hard for them to separate the utility of this from the burden that everything else represents," Shaughnessy says.

Agencies must recognize the importance of a standardized data collection system and outcomes reports if they expect to receive any benefit from this investment. "Some agencies will know the value of this and be able to deal with it better than others," Shaughnessy adds.

2. Learn how to collect clean data.

Make sure the entire OASIS data set is complete for each patient, and make sure the data set is embedded within your assessment instrument, Shaughnessy advises. For example, look at the OASIS item on grooming. If an agency already has a grooming item on its assessment form, then take that out and combine it with the OASIS item on grooming, he says.

"Once you have duplicity, you have declining interest on the part of the providers," Shaughnessy says. "And they already think OASIS is overwhelming." A well-designed assessment form with OASIS data should take no more time at the start of care than the current form takes, he adds.

Re-evaluate every item, every time

Another important tip is to not allow nurses and therapists to easily see what previous OASIS items were for a patient. This encourages complacency and might result in staff missing important changes in a patient’s condition.

One of the demonstration project agencies, for example, used a vendor who set up the option for agencies to carry forward all of the OASIS items from the first data point at the start of care to the first follow-up point, Shaughnessy says.

"They could simply replicate the data, and the idea behind that is if the patient’s status hasn’t changed, why do we have to manually enter this all again?" he says. "But this creates a strong incentive to the clinician to not change anything, and that shows up in outcomes reports, where all the patients appear stable and remain the same."

CHSPR researchers tell agencies not to allow clinicians to carry data from one time point to another. Instead, they should re-evaluate patients at every step of the way.

Another way to ensure data accuracy is to conduct spot checks to evaluate the data’s accuracy. Another clinician could visit another employee’s patient and collect the same OASIS data that the primary provider had collected. If the spot-check person finds any discrepancies, then he or she can discuss these with the clinician.

3. Understand risk adjustment.

When home care agencies first begin to receive outcomes reports based on OASIS data, they may find that these reports have not been adjusted for risk. These are called unadjusted outcomes reports, Shaughnessy says.

What that means is agency X may have a predisposition toward an improvement in ambulation, so you would expect agency X’s outcomes to be better in ambulation than a national reference sample of agencies. The unadjusted outcomes reports would reflect that agency X has done a better job in ambulation.

Risk adjustment levels the playing field. It eliminates differences between a particular agency’s outcomes and the national reference sample when these differences are based on a particular agency having patients who are at greater or lesser risk of poor or superior outcomes.

Which patient attributes influence outcomes?

If an agency simply wants to compare one year’s outcomes to a previous year’s outcomes, then risk adjustment might not be necessary. It’s more useful when one agency is compared with other agencies.

The first step in risk adjustment is to determine the relationship between a given outcome measure and those patient attributes that influence the outcome. For example, suppose agency X had a hospitalization rate of 23%, which was slightly above the national sample hospitalization rate of 22.5%. However, this agency also had a smaller percentage of women patients than the national sample, and the national sample listed a hospitalization rate for women at half that for men. When agency X’s hospitalization rate is adjusted to reflect its case-mix difference, its risk-adjusted hospitalization rate actually is considerably better than the national sample.

4. Investigate care delivery.

CHSPR suggests agencies take these steps in the process of care investigation:

• select target outcome;

• enumerate important care behaviors correlated with the target outcome;

• determine target care behaviors;

• state problems or strengths;

• specify best practices to adopt or reinforce;

• spread best practices across your agency.

The quality manager should ask and answer these questions:

What are the best practices I want to put in place? Crisler suggests agencies specify exactly what a clinician should do, exactly when it should be done, and exactly how it should be done.

When I see a patient who is at a level less than fully independent in ambulation, what do I want to investigate in terms of pain?

• How do I determine whether I need to implement a specific type of pain management program?

• What are the indicators for that at start of care or at other points, and what is that pain management approach?

• How do I make sure my clinicians do it?

• If I decide activity intolerance was really the underlying problem, what does that mean?

• What kind of graded exercise program do I want to make sure happens?

Use quality improvement techniques to identify best practices and to encourage clinicians to follow these best practices, Crisler says.

"You have a specific outcome on which you start to focus," she adds. "And then you look at the care that was actually delivered to get to that outcome to decide what’s your problem or what is your area of strength."


Marly Auerbach, RN, BSN, MPH,Director of Quality Management, Lifeline Home Health Care, P.O. Box 938, Somerset, KY 42502-0938. Telephone: (606) 679-3952. Fax: (606) 678-4424.

Peter Shaughnessy, PhD, Professor and Director; Kathryn Crisler, MS, RN, Senior Research Associate; Karin Conway, MBA, RN, Senior Researcher, Center for Health Services and Policy Research, University of Colorado, Suite 306, 1355 South Colorado Blvd., Denver, CO 80222. Telephone: (303) 756-8350. Fax: (303) 759-8196. Web site: www.oasis/obqi.

Estelle Wolf, RN, Former Director of Professional Services, SunPlus Home Health, San Diego. Telephone: (619) 592-0609. Fax: (619) 592-0619.