Invest or perish: Why you need a computer upgrade

Physicians need outcomes info to manage risk

As demand for performance measurement grows, so too does the complexity and cost of data collection. To meet this challenge, group practices are investing in a new computer-based infrastructure. We asked Richard E. Dixon, MD, medical director of the National Independent Practice Association Coalition in Oakland, CA, to explain why and how physician practices should upgrade their systems.

Dixon was a member of an information systems advisory group for the National Committee for Quality Assurance in Washington, DC. He also has been a leader of the California Cooperative Healthcare Reporting Initiative (CCHRI), a collaborative of purchasers, payers, and providers that created a uniform method of collecting performance data.

Health plans complain about incomplete and poor quality data they receive from physician practices when they gather Health Plan Employer Data and Information Set (HEDIS) information. What is the primary problem with the data?

Physicians and their staffs have little experience or training in accurate diagnosis and procedure coding. Moreover, information required for claims payment does not necessarily capture some of the most important process and outcome measures of care. Therefore, claims and encounter data are notoriously inaccurate. Those inaccuracies are not unique to physicians’ ambulatory encounters, however.

Second, physicians have not had a strong economic incentive to report complete encounter records. Payments are often tied to the most expensive or complex procedure, and secondary or unbundled associated procedures are rejected for payment, even though they may be clinically important.

Finally, capitation as a reimbursement mechanism has further reduced physicians’ incentives to document encounters since capitation uncouples payments from diseases treated or procedures performed.

Most physician practices have computerized billing and administrative databases. Can they use those to collect outcomes information, such as HEDIS data?

There are many billing systems in use, and I do not know the capabilities of most of them. In theory, a billing system should be able to help collect outcomes data, and some vendors claim that their systems can prepare HEDIS reports.

But HEDIS reporting is not simple. HEDIS has very precise criteria to define the populations to be studied (which patients are included in the denominator) and equally precise criteria to define those interventions that satisfy HEDIS criteria (which interventions are included in the numerator for HEDIS rate calculations).

From our CCHRI experience, even skilled programmers at health plans frequently interpret HEDIS criteria incorrectly as they code their plans’ collection efforts. Therefore, I would need to examine the logic of any computer program purporting to report HEDIS data before I accepted its results as reliable.

Why should a physicians invest in outcomes analysis software or improved information systems? How will they benefit?

I believe that physicians who don’t make that investment — who don’t have effective information systems — will be forced out of business. They may have three years to develop such systems in a few competitive and sophisticated markets or a decade in other areas of the country, but the time is near.

No matter whether a physician is in an independent practice association (IPA) or medical group, works for a staff-model HMO, or has an individual direct contract with an HMO, that physician is at risk in our prospectively budgeted health care system.

As a result, it is critically important that a physician understands and is able to quantify the illness burden of the physician’s patients, is able to make certain that those patients are getting the right level of care — not too little or too much — and ensure that screening, prevention, and education efforts are targeted toward those who can best benefit.

Sophisticated purchasers are not willing to tolerate our inability to measure the processes and clinical outcomes of the care we provide. They will, at some point in the near future, withhold business from those provider organizations — or HMOs — that don’t have the systems.

Let me note that the Health Care Financing Administration and many of the state Medicaid agencies are increasingly behaving like commercial purchasers, so I expect the public-sector purchasers to place the same pressures. For most physicians, there will be no place to hide from these expectations.

What advice would you give group practices about choosing software?

Unfortunately, this software market is still immature, is changing rapidly, and is highly fragmented. There are too many competing proprietary models and too few industrywide standards. It’s a fact of life that any system installed now will need to be replaced or significantly upgraded in the future.

Other businesses have learned obsolescence comes quickly and adapting to changes in information technology is a continuing cost of doing business. Medicine is no different.

Does that mean a physician should wait until standards are set, until some software Goliath can impose stability? No. In my opinion the looming competitive pressures are too great.

What first steps should practices take to improve their information systems?

If I were selecting systems, I would focus on four factors. First, I believe the industry is moving toward open-architecture, and industry-standard hardware and software products. I would try to avoid highly proprietary systems or ones that use arcane programming languages.

I would like to be confident that I can maintain my system if my vendor goes out of business, and I would like to increase the chance that when I switch to a new system in a few years (as I certainly will), I can transfer my data to the new system. Both of these options are more likely to be available with open-architecture systems.

Second, I would focus on my most important immediate needs. Probably the most important need is to ensure that I can adapt my practice style to the system’s style. Then I would consider the following needs:

• the ability to provide accurate encounter data (diagnoses and procedures);

• the need to have a population-based system where we can define groups of patients, e.g., diabetics, or patients with congestive heart failure, or women who are candidates for mammography;

• the need to cut office overhead and increase physician productivity (sharing information within our IPA or medical group, having call-center protocols, automating prescription refills, etc.);

• having a system that makes me less dependent on the paper clinical record.

Less dependent on that paper record, not totally independent of it. We would all like an electronic medical record that captures the richness of a physician’s encounter with a patient, but the market is not demanding that yet.

Third, I would try to find a system that will readily permit critical clinical information to be shared electronically. I would want to be able to capture laboratory reports easily because that would make me more efficient.

I would want to be able to see a minimum clinical data set for my patients and those of my call-partners. That minimum data set need not be extensive — it might be no more than a problem list, a list of current medications, and an allergy list. A system that provides those minimum data would be even more important if I had more than a single office location.

Finally, I’d like my system to be compatible with the information systems in my IPA or medical group.