Page 1 1 Reengineering Medicine: Development and The Role of IS - PDF document

Using IT to Improve Quality: Goals Past Results and Future Potential • Major gaps between evidence, practice • Costs high • Problems with errors David W. Bates, MD, MSc • Computerized decision support Medical Director of Clinical and Quality • Current Partners system Analysis, Partners Healthcare • Errors • Costs Chief, Division of General Medicine, • Guidelines Brigham and Women’s Hospital • Next 5 years at Partners IS • Conclusions Old Paradigm Leadership and IT • Authorities are infallible Leadership is the capacity to hold a shared vision • Heuristics work well of that we wish to create. • If in doubt, do it – Peter Senge • Clinical judgement and the “art of medicine” get you to the right answers The best way to predict the future is to invent it. • Community standards are correct – Peter Drucker David Eddy, Aetna Quality Forum 1999 New Paradigm The IOM Report • Report targets hospital errors: Mistakes killing • Authorities vary substantially thousands every year 11/30/99 • Heuristics don’t work • Medical errors kill 44,000-98,000 people per year • Clinical judgement is insufficient • “More people die from medical errors each year than from • Huge variation by community suicides, highway accidents, breast cancer, or AIDS” • “These stunningly high rates of medical errors - Therefore resulting in deaths, permanent disability, and • Need to begin to practice evidence-based unnecessary suffering - are simply unacceptable medicine in a system that promises to first ‘do no harm.’” David Eddy, Aetna Quality Forum 1999 William Richardson Page 1 1

Reengineering Medicine: Development and The Role of IS Implementation of POE • Could be changed by providing external aids • Physician involvement and leadership • Linking medical knowledge and patient-specific data • Decision to automate existing systems as is • Identifying options • Constant focus on speed • Without such tools, experts • Strong support from hospital administration • Make errors • Willingness to be flexible, modify system • Overlook available knowledge • Don’t sufficiently account for uniqueness • Patients could participate in decision-making Weed LL, Weed L, Federation Bulletin , 1994 Event monitor architecture Physician Coverage List Applications • Functions Applications Patient Applications page, email, (new data) (new data) • Identifies first and second-call physicians (new data) database write to file, • Manages physician rotation [real time • Handles evening coverage and signing out message] Inference engine Annun- • Facilitates delivery of computer-generated (decisions) ciators messages • Computer-page interface allows automated paging Rule Knowledge Coverage editor base List Pharmacy Computer System Field Handwriting example Test of Unsafe Orders Unsafe Order Not Detected Cephradine oral suspension IV 61% Vincristine 3 mg IV x 1 dose 62% (2-year-old) Colchicine 10 mg IV for one dose 66% (adult) Cisplatin 204 mg IV x 1 dose 63% Source: ISMP Medication Safety Alert! Feb 10, 1999 Page 2 2

Medication Error Frequency ADE Prevention Study: and Potential for Harm Key Results • 6.5 ADEs/100 admissions In 10,070 Orders • 28% preventable 530 Medication Errors 1.4 per admission • 3 potential ADEs for every preventable ADE 35 Potential ADEs • 62% of errors at ordering and transcription stages 5 Preventable ADEs • Systems analysis • No individual responsible for repeated errors • 1 in 100 medication errors results in an ADE • Systems should be designed to: • 7 in 100 represent potential ADEs • Make errors less likely • Catch those that do occur JAMA 1995;274:29-43 Improving the Quality of Drug Costs of ADEs Ordering w ith Order Entry • ADEs are expensive • Streamline, structure process • $2461 per ADE, $4555 per preventable ADE • Doses from menus • Annual BWH costs: • Decreased transcription • $5.6 million for all ADEs • Complete orders required • $2.8 million for preventable ADEs • These figures exclude costs of: • Give information at the time needed • Injuries to patients • Show relevant laboratories • Malpractice costs • Guidelines • Costs of admissions due to ADEs • Guided dose algorithms • Justifies investment in prevention efforts • Perform checks in background Drug-allergy Dose ceiling Drug-lab JAMA 1997;277:307-311 Drug-drug Drug-patient Chemotherapy Order: Allergy to Medication Patient Characteristics Page 3 3

High Chemotherapy Dose High Chemotherapy Dose: Warning Requires Attending Approval Impact of BWH Inpatient Serious Medication Error Provider Order Entry Rates Before and After OE • Nizatidine use, for all oral H2 blocker orders, increased 12 Events/1000 Patient-days from 12% to 81% 10 • The percent of doses over the suggested maximum Delta = -55% 8 decreased from 2% to .6% p < .01 • The percent of orders for ondansetron, with a 6 4 frequency of 3 times daily, increased from 6% to 75% Phase I Phase II • The percent of bed rest orders with a consequent 2 order of heparin increased from 24% to 54% 0 Serious Medication Errors Teich, Arch Int Med 2000 Bates et. al. JAMA 1998 Reducing Drug Costs with “Panic” Laboratory Study Order Entry • For markedly abnormal results (K, Na, glucose, Hct) • Types of useful suggestions • Allows consideration of other factors • Drug interchange • Direct interface with paging system • Lower dose • “Before” data • Different route (IV-PO switches) • Median time to rx 2.5 hours • Guidelines for use • For 25% > 5.3 hours • RCT results • Mean time to rx 11% shorter (p<.0003) • Mean time to resolution 29% shorter (p=.11) • 95% physicians pleased to be paged Kuperman, JAMIA 1999 Page 4 4

Selected Laboratory Effect of Changing Default Dosing Frequency for Ceftriaxone Interventions • Charge display RCT 60 • No statistically significant effect 50 • BUT $1.7 million lower lab charges in intervention group Orders/week 40 BID • Redundant labs QD 30 • 67% reminders followed 20 • Annual charge savings $31,000, vs. estimate of $376,000 10 • Only 44% tests performed had computer order 0 • Substantial improvement possible if loop closed with 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 laboratory “back end” Week Other Laboratory Evaluations Guidelines: Vancomycin RCT • Initiation, renewals both targeted • Antiepileptic drug levels • Vancomycin use was reduced by • Only 28% of BWH inpatient levels appropriate intervention • RCT of structured ordering showed improvement • Bigger effect on renewals than on initiation • Digoxin levels • Magnitude of overall decreases • Only 16% of BWH inpatient levels appropriate • Vancomycin-days/prescriber 37% lower • Potential charge savings $388,000 • Duration of therapy 17% lower • PSA levels • Much of use likely still inappropriate • 19% inappropriate (age, frequency issues) • Further decreases possible by targeting specific • Thyroid studies indications • Initial testing TSH alone in only 73% of patients Guideline for Expensive Agent Low Yield Critique Page 5 5

Chest Radiographs and Alternate Exam Structured Ordering Percent Acceptable History Assess/R/O Before 78% 35% After 99% 99% Impact of Computer OE on Order Entry and Critical Paths Physician Time • Order writing took twice as long on computer • Critical paths specify what should happen for a • Medical HOs 44 min/day, recovered half specific day • Surgical HOs 73 min/day, no recovery • Essentially sequences of order sets • Daily and one-time orders accounted for most of • In place for 25 diagnoses change, increasing 3-fold • Have decreased LOS, costs, improved • Sets of orders took half the time they did before order satisfaction entry • Require physicians to select • Interventions dx at admission • Allows prompting about path • Introduction of “Write 1” • Reorganization of screens to facilitate access to OE • Increases likelihood path will be selected Results of Critical Path HO Satisfaction with OE Evaluation • 82% of admission diagnoses coded Overall satisfaction with OE • Half the diagnoses have an order set • Physicians select 40% of time when offered OE improves productivity • Substantial variation by diagnosis OE improves patient care • Total knee 77% • Pregnancy 54% OE reduces errors • Deep venous thrombosis 14% Surgery 1 2 3 4 5 6 7 Medicine 1=never, 7=always Page 6 6