Everyone from Newt Gingrich to the New York Times is extolling the glories of the electronic health record (EHR) and lamenting its slow adoption by physicians. Senator Edward M. Kennedy's Wired for Health Care Quality Act (S. 1693) is designed to speed adoption of information technology by the health care system. Presidential candidates cite EHRs as part of the solution for health care reform. Thomas Goetz wrote in the New York Times that “the problem isn't a lack of software.”¹ No, the problem is the lack of good software.

Health care administrators, physician leaders and policy makers speak as if health information technology (HIT) has arrived – as if it were a static entity ready to begin improving patient care, decreasing costs and generally leading us to the promised land, if only Luddite physicians would fork over the cash. Few speak about the huge gap between what HIT could do and what it currently does.

As a front-line primary care specialist, I don't want to return to paper, but I do need an EHR that works. I say put Google on the job. Medical care involves sifting and synthesizing vast amounts of information. Google gets this. Google understands the importance of seamless access to information, concise display and creative use of symbols.

Without losing touch with my main screen, using Google's “Desktop Gadgets” (http://desktop.google.com/plugins) I can search Wikipedia, scan e-mails, consult a dictionary, translate words, review the calendar, check the weather and keep a to-do list. These barrier-free tools hardly interrupt the flow of work on the center of the screen. Thought and action are closely linked and supported by a sophisticated platform.

The EHR data maze

While Google Desktop Gadgets make access to information as unfettered as possible, HIT systems often sequester individual tidbits of information at the ends of nonintuitive labyrinths, with needless hurdles along the way. Clinicians need clear access to priority information, not an obstacle course. As an EHR user, I have to keep a thought in mind until, five clicks and two screens later, I can find related information. Then, I have to park all of this developing thought for four more clicks, three screens and a slow download until I get to the screen where I can take action. Repeat this process for dozens of tasks per patient and you have an environment that facilitates error and inefficiency.

In one of the EHRs I use, the medication list is a blizzard of undifferentiated characters, a monotone of font with unnecessary clutter, in that each medication name is repeated four times in a single line. Dosing information is ambiguous. (Does “Coumadin 4 mg 2 tabs” mean a total dose of 4 mg or 8 mg?) Furthermore, there is a wide gap on the screen between the medication name and the dose, so I've found myself inaccurately associating the dose of one medication with the medication on the line above or below. Beyond these display challenges, there is no grouping of medications by organ system or problem, so it is easy to overlook one of the patient's four diabetes prescriptions scattered through a list of 13 other medications.

Laboratory results reporting is likewise designed for error. The lab values are lost within a display muddled by second-priority information. It is easier to see that the test was drawn in Central Standard Time, as it always is in Iowa in January, than to track the flow of results. The signal to noise ratio is low.

Orthostatic blood pressures are splintered into a series of identical boxes, so it is difficult to discern which value is the systolic and which the diastolic and which values are lying, sitting and standing measurements. On the anticoagulation flow sheet, warfarin dosages and INRs are reported in two columns, on alternating lines, so the physician doesn't readily know whether the INR is associated with the dosage on the line above or the one below. Bedside blood sugars are reported in a single row, rather than organized by time of day, making it impossible to tell, for instance, whether the glucose values are running high in the evening and low in the morning. Again, it is a system designed to facilitate errors.

HIT is supposed to make care safer and better. And I believe that it will. Eventually. But for now I've witnessed more serious errors with the EHR than in my previous 25 years as a physician. There may be fewer errors of the type that can easily be counted but are unlikely to matter, such as spelling errors or applying the wrong units to a medication dosage. The greater concern is the errors that are less easily measured. These are the errors in thinking and decision making: cases where the physician was so distracted by the convoluted order tree that she forgot an important order; cases where the fragmented, disordered thinking imposed by the EHR obscured the big picture of the patient's care.

When I work through a colleague's note composed of prestructured text that conveys almost no useful content, patient care is compromised. When I am distracted by the hunt for information, clicking from screen to screen, trying to hold onto information from one corner of cyberspace while searching in another, I have less attention to give to the patient, and again there is a systemic occasion for error. When I become the unit secretary, transcriptionist, pharmacist, receptionist and medical records clerk, the time and energy remaining for the doctoring tasks of medical decision making and relationship building are limited. Errors will occur.

Making EHRs make sense

Imagine how health care could be improved by incorporating the best in information-display and navigation. Imagine an EHR with creative utilization of hovering, expanding data sources, novel visual cues, minimized clicking and time savers. It should not take four clicks to order a test or change a medication dosage. One should be able to read a complex list of medications and readily establish a clear cognitive picture. Differences between admission medications and current medications should be symbolically represented. Lab values should be consolidated, clear and contextualized. The user should not have to reconstruct data fields with every use. It shouldn't take 45 minutes to complete a task that used to take 20. When we have EHRs that incorporate such lessons, the power of the EHR will begin to be realized.

So let's get Google on it, and bring in a Toyota mindset while we are at it. Toyota understands the need for lean, streamlined workflow. Toyota understands that even a minor unnecessary step that causes worker fatigue or extra time is a problem. Many steps in HIT are redundant. They cause worker fatigue and monopolize time.

Every click should be ruthlessly evaluated: if it doesn't add value, it should be eliminated. Every character of information on a display should also be precisely studied. Is it necessary? Is it high-priority information? If not, put it in a position of lower priority or get rid of it. Distracting data, cluttering information and redundant characters are all rampant in existing HIT. We need better visual management of information.

Policymakers have embraced pay for performance, citing the EHR as a tool that makes this feasible. Our clinic has used an EHR for more than three years, yet had we chosen to participate in the Centers for Medicare & Medicaid's Physician Quality Reporting Initiative (PQRI), physicians in our department would have been reduced to human workhorses, ferrying information from one repository to another, diverting precious clinical time away from patient care to match up lists of hundreds of ICD-9 and Current Procedural Terminology (CPT) codes with the multitude of new five-digit PQRI G codes, all by hand, by memory and on paper. Our EHR is not nimble enough to match up these electronic data fi elds automatically. Conceptually this is just the sort of work for which a computer is well suited. But the gap between the promise and the reality of HIT means that many physicians are manually re-digitalizing digital data.

Robert M. Kolodner, MD, the new White House HIT chief, proposes to increase the adoption of HIT by using pay-for-performance incentives. No one had to be coerced to use a cell phone, a digital camera or the Internet. The problem with HIT is not the purchaser; it is the product. Physicians don't need to be bullied into buying something that doesn't fi t their needs; we need a better product.

If the federal government wants to improve the quality of health care and decrease the cost, it might offer incentives to technology innovators to create a streamlined, user-friendly EHR to replace the clunky technology under which we currently labor. Drug salespeople have better tools to track my prescribing habits than I have for prescribing; the UPS delivery woman, the clerk at Target and the taxi driver all have technology better designed for efficiency than I have for use in patient care.

Physicians aren't resisting and resenting HIT because we are dinosaurs or because we are out of step with the modern world, but because the available technology does not sufficiently support us in the safe and efficient performance of our work. There has been an almost ideological fervor in the assumption that all things electronic are good; in an evidenced-based world, we have accepted current health information technology on faith.