Defending Computerized Physician Order Entry From Its Supporters

The American Journal of Managed Care, July 2006, Volume 12, Issue 7

Journal

Journal

The article by Palen et al in this issue of theis an important contribution to the literatureon healthcare information technology(IT).1 First, it is one of the few studies of a decision supportsystem (DSS) that focuses on alerts for laboratorymonitoring. Second, it reports negative findings; thealerts do not generate greater physician adherence toappropriate laboratory monitoring. Third, it is an especiallygood test of a part of a DSS: (1) physiciansreceived elaborate 1-on-1 training on the laboratorymonitoring alerts, their meaning, the triggering factors,and the timing; (2) the alerts focused on a limited set ofspecific medications for which the monitoring was relevant;(3) the selected monitoring alerts were carefullychosen from respected sources; and (4) local physiciansreviewed and approved the monitoring guidelines(alerts), thus presumably enhancing buy-in. Last, thiswas willing to publish a study of negative findings,an accomplishment in itself.

Equally important, the findings by Palen et al—andtheir comment that their results raise questions aboutcomputerized physician order entry (CPOE)—highlighta persistent error in the research literature on the technology,the inability or refusal to distinguish betweenCPOE and DSS.2,3 This error impedes appreciation ofthe strengths and weaknesses of both technologies, andcontributes to the cacophony in error reduction claimsamong vendors, in-house IT staff, researchers, and evaluators.4-8 Note that clarification of the scope of CPOEand DSS is not a mere academic distinction. The waywe define CPOE and its related advantages and problemsis entwined in the political and commercial evangelizingfor the technology as a redeeming force forhealthcare's dual cardinal sins of high costs and medicationerrors.2-8 The need, therefore, for a cleardelineation will not only benefit researchers, but willalso help the national effort at implementing healthcareIT.2-4,6-9

The Scope of CPOE

DSS is not an integral part of CPOE.2,3 Less than 10%of hospitals have CPOE,8,9 and only a very small proportionof those hospitals also have DSS.2 Moreover, somehospitals have DSS without CPOE systems. The vast literatureclaiming that CPOE systems reduce the incidenceof medication errors is often based on combiningthe effects of CPOE with the benefits of DSS.2-5 Theseclaims are undoubtedly not intentionally deceptive, butrather are the result of definitional sloppiness aided bythe hopefulness of vendors and some researchers. Someclaims for the effectiveness of CPOE systems includenot only the benefits of DSS but also of electronic healthrecords and electronic medication administrationrecording systems, further inflating the apparent utilityof CPOE.

On the other side of the equation, when analysis ofCPOE systems incorporates the problems of DSS, theresults often detract from the apparent value of CPOE.Some CPOE systems are freighted with the difficulties ofDSS, such as frustration with annoying warnings (alarmfatigue), conflicting criteria for warnings, physician irritationat being reminded of obvious steps, and integrationof DSS plus CPOE with other IT systems (ie,pharmacy and nursing systems).2 Rather than reflectingthe advantages of CPOE, such expansive definitionsmight result in an inappropriately low estimate of thesystem's efficacy. Indeed, understanding the often difficultmarriage of DSS with CPOE is essential to our evaluationof the contribution of Palen et al. Their findings,contrary to what the authors suggest, should not beviewed as a problem with CPOE. Palen and colleaguesstudied an aspect of a DSS, not of CPOE. While thewarnings about appropriate laboratory tests failed toproduce improvements in physicians' monitoring, thewarnings, to our knowledge, did not cause the physiciansto reduce the quality of their original medicationorders—and writing appropriate medication orders isthe primary function of CPOE.

What Palen and colleagues illustrate is how difficult itis to produce an effective DSS,2,3,6,10-12 even in a trial thathad remarkable resources to train each physician, selectthe medications and laboratory tests, etc.

Perverse Logic

Those who see CPOE as a panacea for medicationprescribing errors might be reluctant to accept mydefense of the technology because they tend to see DSS(along with electronic health records and medicaladministration recording systems) as a part of CPOE,rather than as extraordinarily promising but sometimestroubled siblings. Conflating the technologies leads us tocompare apples to oranges, inaccurately comparing differentarrays of systems across hospitals.

I suggested above that defining the scope of CPOE isnot just an academic exercise. That's because CPOE hasbecome a part of federal healthcare policy, hospital andinvestor strategies, and hardball market economics. Inan effort to promote CPOE as the panacea for healthcare'smajor problems, its supporters embrace an imperialdefinition of CPOE's scope and reported utility.3

CPOE is a promising technology in itself, which canbe integrated with other technologies to improve itsvalue. Its supporters only force the system to absorb theproblems of other technologies if they insist CPOE systemsare those other technologies—a claim I suggest isconfusing and unhelpful.

Accordingly, I do not accept the suggestion by Palenand colleagues that their valuable findings oblige us toquestion aspects of CPOE. Their findings do raise criticalquestions about DSS, and that's a serious contributionin itself.

Am J Manag Care.

1. Palen TE, Raebel M, Lyons E, Magid DM. Evaluation of laboratory monitoringalerts within a computerized physician order entry system for medication orders.2006;12:389-395.

JAMA.

2. Garg A, Adhikari N, McDonald H, et al. Effects of computerized clinical decisionsupport systems on practitioner performance and patient outcomes: a systematicreview. 2005;293:1223-1238.

JAMA.

3. Wears R, Berg M. Computer technology and clinical work: still waiting for Godot.2005;293:1261-1263.

J Biomed Inform.

4. Bates D. Computerized physician order entry and medication errors: finding abalance. 2005;38:259-261.

JAMA.

5. Koppel R, Metlay J, Cohen A, et al. Role of computer physician order entry systemsin facilitating medication errors. 2005;293:1197-1203.

N Engl J

Med.

6. Bates D, Gawande A. Improving safety with information technology. 2003;348:2526-2534.

Health Aff.

7. Poon EG, Blumenthal D, Jaggi T, Honour M, Bates D, Kaushal R. Overcomingbarriers to adopting and implementing computerized physician order entry systems.2004;23:184-190.

J Am Med Inform Assoc.

8. Ash JS, Gorman PN, Lavelle M, et al. A cross-site qualitative study of physicianorder entry. 2003;10:188-200.

CPOE Digest 2006.

9. KLAS Enterprises. Orem, Utah: KLAS Enterprises; 2006.

10. Silverstein S. Sociotechnologic issues in clinical computing: common examplesof healthcare IT failure 2006. Available at: http://www.ischool.drexel.edu/faculty/ssilverstein/medinfo.htm. Accessed April 28, 2006.

IEEE Trans Sys Man and Cybernet—Part A: Sys

Humans.

11. Nemeth C, Cook R, Woods D. The messy details: insights from the study oftechnical work in healthcare. 2004;34:689-692.

J Am Med Inform Assoc.

12. van der Sijs H, Aarts J, Vulto A, Berg M. Overriding of drug safety alerts incomputerized physician order entry. 2006;13:138-147.