A report from the Institute of Medicine in 1999, To Err is

Transcription

1 EDUCATION AND TRAINING Effectiveness of a graduate medica education program for improving medica event reporting attitude and behavior Y M Coye, S Q Mercer, C L Murphy-Cuen, G W Schneider, L S Hynan... See end of artice for authors affiiations... Correspondence to: Dr Y M Coye, Interna Medicine, The University of Texas Southwestern Medica Center at Daas, 5323 Harry Hines Bouevard, Daas, Texas , USA; yvonne.coye@ utsouthwestern.edu Accepted for pubication 15 August Qua Saf Heath Care 2005;14: doi: /qshc Objectives: To evauate the effectiveness of an educationa program for improving medica event reporting attitude and behavior in the ambuatory care setting among graduate medica trainees. Design: One group pre- and post-test study. Setting: The University of Texas Southwestern Medica Center at Daas Famiy Medicine Residency Program. Participants: A famiy practice residents (n = 30). Intervention: Patient safety educationa program impemented through an introductory ecture and 6 monthy conferences, June to December 2002, invoving medica events that occurred in the ambuatory care setting. Outcome measures: Medica event reporting attitude and behavior at baseine and at 6 month foow up, and barriers to medica event reporting at the 6 month foow up. Resuts: Program attendance was significanty correated with medica event reporting attitude and behavior change (rho = 0.525, p = 0.003). The median change in medica event reporting attitude and behavior was zero and not statisticay significant (p = 0.566). Major barriers to medica event reporting were ack of time, extra paper work, and concern about career and persona reputation. Concusions: Attending the patient safety educationa program was key for promoting a positive medica event reporting attitude and behavior change among graduate trainees. Major barriers to medica event reporting were ack of time, extra paper work, and concern about career and persona reputation. Future research wi need to focus on reducing these barriers and to evauate the effectiveness of such a program over onger periods of time, since making a positive change in medica event reporting attitude and behavior must be made at the individua and organizationa eves. A report from the Institute of Medicine in 1999, To Err is Human, emphasizes that incorporating patient safety education into cinica training programs is a key mechanism for improving patient safety. 1 Furthermore, it is recommended that the initia exposure to patient safety shoud occur eary in undergraduate and graduate medica education programs and be ongoing through continuing medica education. 2 Mutipe activities can be directed towards understanding the causes of medica events and utimatey their potentia roe in promoting patient safety, which incude team training programs, information technoogy improvements, and medica event reporting programs. 3 In tandem with these activities, individua and organizationa attitudes must promote a supportive patient safety cuture. 1 4 This study focuses on one dimension of patient safety cuture namey, the attitude and behavior towards the reporting of medica events by graduate medica trainees. Past case studies have shown that medica events invoving graduate trainees (known in the US as interns, residents, and feows) have the potentia for causing harm to recipients of heath care. 5 Fortunatey, most medica events do not progress to cause patient harm due to the timey intervention by the graduate trainees themseves, other members of the heathcare team, or because the outcome of the medica event was benign in nature. Graduate medica education programs have historicay reied on the assessment of individua behaviors, through morbidity and mortaity conferences and facuty peer review, to anayze adverse medica outcomes incuding medica events. This approach has become outdated since current evidence indicates that a substantia proportion of medica events are attributabe to our processes of patient care rather than being the soe resut of poor individua cinica performance Athough there have been reports of incorporating medica event education into the medica curricuum, no educationa programs have been evauated for their effectiveness in improving the patient safety cuture of the heathcare system or its participants. 7 9 The purpose of this study was to evauate a patient safety educationa program for its effectiveness in improving attitude and behavior reated to medica event reporting in the ambuatory care setting among graduate medica trainees. The primary outcome for the study attitude and behavior reated to medica event reporting was assessed by a sef-administered questionnaire at baseine and at a 6 month foow up. A secondary outcome of the study was the barriers to medica event reporting which were assessed by a sef-administered questionnaire at the 6 month foow up. The educationa program of the study consisted of an introductory ecture foowed by six structured, facuty faciitated, monthy conferences. The introductory ecture covered the rationae for the study of medica events, particuary near misses. A near miss is a medica event that does not progress to an adverse outcome as the resut of a panned or unpanned identification in association with a change in the circumstances that ed to the medica event. 5 These types of medica events are frequent and readiy end themseves to group discussion due to the absence of high emotion that occurs with medica events that resut in adverse outcomes. 6 During the duration of the program, the residents had the option to report medica events using a paper based anonymous reporting system. METHODS Study design This one group pre- and post-test study 10 was conducted over 6 months to determine the effectiveness of an educationa

2 384 Coye, Mercer, Murphy-Cuen, et a program for improving attitude and behaviour of graduate trainees to medica event reporting in the ambuatory care setting. The program was integrated into The University of Texas Southwestern Medica Center at Daas (UT Southwestern) Famiy Medicine Program ongitudina curricuum. The UT Southwestern institutiona review board approved the study protoco. Study site The UT Southwestern Famiy Medicine Residency Program is housed in two cinic sites. One is ocated on the UT Southwestern campus and the other is a community based cinic ocated in a Daas suburb. The campus cinic is jointy operated by UT Southwestern and Parkand Heath and Hospita System (Parkand), a pubic hospita system. The patient popuation using this cinic is primariy referred from Parkand s hospita faciity for foow up care. This cinic serves an ethnicay diverse patient popuation comprised of 60% Hispanics, 20% African-Americans, 18% white, and 2% other, of which more than 85% ive at or beow 100% poverty. Approximatey 50% of this popuation has heath insurance coverage, primariy Medicare and Medicaid. This cinic has over patient visits per year and is served by 18 residents (six in each of three years), two registered nurses, and four medica assistants. The community based cinic serves a patient popuation that, for the most part, has managed care or commercia heath insurance coverage. The ethnic distribution within the cinic is 55% white, 25% Hispanic, 16% African-American, and 4% other. This cinic has 9000 patient visits per year and is served by 12 residents, two registered nurses and 1.5 medica assistants. Study participants US famiy practice postgraduate residencies are of 3 years duration. This study incuded a 30 graduate trainees in the UT Southwestern Famiy Medicine Residency Program, with 10 from each of the three postgraduate residency years. Patient safety educationa program We deveoped the program using the human error in medicine teaching approach of Gosbee and Stahhut. 11 The program consisted of six 1 hour conferences on patient safety and near misses hed monthy from Juy to December 2002, preceded by a 1 hour introductory ecture on medica error given by two of the study investigators (Coye and Mercer) in Juy The content of the program s ecture incuded a brief overview of the impact of medica error on patient safety and had five educationa objectives: (1) to define medica error; (2) to define medica event; (3) to describe the conditions that promote medica events; (4) to describe the process (root cause anaysis) used to identify causes of medica events; and (5) to state the purpose of a medica event reporting system. We used a variant of root cause anaysis known as modeing to identify the causes of the medica events in this project during the conferences on near miss and patient safety. Modeing uses the coective experiences of the participants to describe how a particuar type of medica event can occur and how it might be prevented in the future. 12 The introductory ecture used a video dramatization of a medica event taken from a current teevision show ER, couped with a study investigator ed arge group discussion of the causa anaysis for this medica event. Each of the program s subsequent six monthy conferences had an educationa theme and featured the case discussion of one ambuatory care event and the associated anaysis. The case study was derived from actua medica events that had occurred within one of the study s cinic sites during the 6 months before the initiation of the study. The description and circumstances reated to these medica events were obtained through facuty interviews. A written case package that identified the moderators, the theme for the conference, and a case presentation divided into two sections ( What happened? and Discovery and recovery ) was maied to the study participants 1 week before the conference. Tabe 1 provides a summary of the cases presented at the six conferences. The format of the conferences consisted of the foowing, in order of occurrence: (1) opening remarks (5 minutes) by the facuty moderators on the educationa theme for the case to be discussed; (2) reading of the case aoud by a moderator; (3) sma group discussions (15 minutes) by residents faciitated by a famiy practice facuty member at which time the group coectivey competed a structured worksheet to identify the type of medica event, potentia causes, and its prevention strategies; (4) arge group discussion (30 minutes) on the anaysis of the medica event that incuded interventions for prevention, if possibe; and (5) case summary (10 minutes) by the facuty moderators suppemented by written materias distributed to the participants. Study variabes Six predictor variabes were chosen for this study: age, sex, US versus non-us medica schoo training, type of graduate training degree, postgraduate year, and frequency of conference attendance. We incuded US versus non-us medica schoo training as a predictor variabe since differences in cuture and attitude have been noted in individuas receiving training in different regions of the word. 13 Simiary, we incuded the type of graduate degree as the investigators wanted to examine the differences in medica event reporting behavior between individuas with an MD degree and those with a DO degree, as there are differences in their undergraduate medica education curricua ( org/om.htm, accessed 4 Apri 2005). The main outcome variabe for the study was change in attitude and behaviour to medica event reporting. A secondary outcome variabe category incuded barriers to medica event reporting. Medica event reporting attitude and behavior questionnaire The 5 item questionnaire was based on the work of Gosbee and Stahhut 11 reated to medica event reporting attitude and an adaptation of the Prochaska and DiCemente s stages of change mode for measuring behavior change. 14 Prochaska and DiCemente deveoped this mode to measure behavior change in smoking cessation research, which has since been extended to cessation of other probem behaviors as we as the acquisition of new behaviors (such as mammography screening and exercise). 14 Research shows that this mode predicts behavior change reated to smoking cessation. Appications of this mode measure behavior change categoricay. The categorica measures for the Prochaska and DiCemente mode incude pre-contempation, contempation, preparation, and action. Pre-contempation incudes individuas not currenty reporting medica events and not panning on doing so in the next 6 months. Individuas cassified as contempators are not currenty reporting medica events but are considering doing so in the next 6 months. Those in the preparation stage are not currenty reporting medica events but are panning to start reporting within the next 30 days and have attempted to report a medica event within the past year. Action stage participants have begun reporting events within the ast 6 months. The medica event reporting attitude and behavior questionnaire (tabe 2) incuded five eves of possibe attitude and behavior reated to the reporting of medica events

3 Medica education and medica event reporting 385 Tabe 1 misses Summary of cases presented at the six conferences on patient safety and near Conference theme Case synopsis Medica event Introduction Imperfect information Verba communication Information hand off Physician sips Residency training and human factors errors Patient in ER receives another patient s type specific bood (video dramatization) (negative attitude, pre-contempation, contempation, preparation, and action). Barriers to medica event reporting survey Since it is we known that there is substantia underreporting of medica events due to a number of barriers Bood not taken off the infuser from previous case Nursing strike in progress Many temporary staff in the emergency department Charge nurse in the emergency department Trauma suite assumes someone ese checked the identity of the bood hanging on the infuser Charge nurse continuay interrupted during the case Patient prescribed hypertension Medica record unavaiabe medication over the teephone from home after cinic hours No independent confirmation of high bood pressure to which they had previousy had an adverse reaction Patient did not reca or remind physician of the history of the adverse reaction drug reactions Physician did not ask patient about drug reaction history Patient received injection of Depo-testosterone instead of Depo-Provera The resident did not communicate with either the patient or outside physician regarding the abnorma resuts of the prenata birth defect screening test Physician correcty wrote intended dosages for the diabetes medication in the patient s medica record, but wrote the corresponding prescription for a higher dosage than intended Foow up on an after hours abnorma potassium test (panic vaue) not appropriate Tabe 2 Medica event reporting attitude and behavior questionnaire Questions (1) Do you think that it is important for physicians in training to report medica events? (2) Are you seriousy considering reporting medica events within the next 6 months? (3) Are you panning to start reporting medica events in the next 30 days? (4) Have you reported medica events during the ast 6 months? (5) When did you start reporting medica events? Possibe responses Yes/No Yes/No Yes/No Yes/No (If no, skip question 5) Enter month and year Nurse received verba rather than printed orders Medications had simiar names Inexperienced temporary nurse No supervision of temporary nurse No forma protoco at cinic for the communication of abnorma resuts to the patient or outside physician No forma protoco for the cinic attending physicians to review resuts of tests ordered by the residents Resident did not bring test resut to the attention of the cinic attending physician The higher dosage was commony prescribed but wrong, and the ower dose, athough correct, was an exception for the usuay prescribed doses Training program cuture assumes that residents know how to hande this type of situation No cear guideines for residents on how to respond to after hours panic vaues Patient reported feeing we inherent in medica cuture, we deveoped a survey to identify these barriers based on the reported experience of others who have impemented medica event reporting sytems These barriers were categorized into eight categories, each as a singe item. A ninth category, designated as other, was incuded to enabe respondents to record barriers not incuded on the ist. At the 6 month foow up the study participants were asked to compete the 9 item survey Tabe 3 (n = 30) Characteristics of study participants Characteristic No (%) Age (years) Range Median age 30.5 Sex Mae 10 (33.3%) Femae 20 (66.7%) US medica schoo training 23 (76.7%) Graduate training degree MD 23 (76.7%) DO 7 (23.3%)

4 386 Coye, Mercer, Murphy-Cuen, et a anonymousy, identifying a of the factors that prevented them from reporting medica events. Data coection Data for the demographic variabes were obtained by review of the UT Southwestern Famiy Medicine Residency Program records. A study investigator (Murphy-Cuen) recorded attendance at each conference. The outcome variabes, medica event reporting attitude and behavior change, and barriers to medica event reporting were derived from sefadministered questionnaire data. The medica event reporting attitude and behavior questionnaire was sef-administered and competed by the program s participants before the introductory ecture in June 2002 and after the ast conference on near misses and patient safety in December The data for these variabes were entered into a Microsoft Exce spreadsheet. Statistica anaysis Using the scoring protoco deveoped for the stages of change mode, the answers to the medica event reporting attitude and behavior questionnaires (tabe 2) were used to cacuate the pre-program and post-program scores for attitude and behavior reated to medica event reporting. 14 No response to question 1 in the medica event reporting attitude and behavior questionnaire (tabe 2) indicated a negative attitude to medica event reporting. Positive responses for one or more of the first four questions of the medica event reporting attitude and behavior questionnaire (tabe 2) determined the five behaviora eves for medica event reporting as foows: (1) pre-contempation (question 1), (2) contempation (questions 1 and 2), (3) preparation (questions 1, 2, and 3), and (4) action (questions 1 and 4). If assigned to the action category, question 5 specified the amount of time for medica event reporting. There were five eves to the scoring protoco that corresponded to the attitude and behaviora eves for medica event reporting. At each of these eves, beginning with the negative attitude category and foowed by the four behaviora eve categories (pre-contempation, contempation, preparation, and action), scores were assigned in increments of 0.2 ranging from 0 (negative attitude) to 0.8 (action). Changes in the attitude and behavior eve categories were cacuated by subtracting the post-program scores from the pre-program scores. Study participants characteristics for the continuous measures are presented as medians and range, and for the dichotomous measures as numbers (frequency of event) and proportions. The Wicoxon signed rank test was used to compare the changes in rank for the pre-program and Number of study participants Negative attitude Baseine Foowup Pre-contempation Contempation Preparation Attitude and behaviour eve Action Figure 1 Change in attitude and behaviour to medica event reporting from baseine to foow up at 6 months. Change score (2) (6) (6) (6) post-program medica event reporting attitude and behavior scores. Spearman rank order correations (rho) were used to describe the association between the characteristics of the study participants and the change in medica event reporting attitude and behavior score. SPSS Version 11.5 was used to anayze the data. The statistica tests used were two taied. Statistica significance was set at a p eve of RESULTS Characteristics of study participants Thirty graduate trainees participated in the study, representing 100% of the residency program. The median age of the participants was 30.5 years, 33.3% were mae, 76.7% received their medica training in the US, and 76.7% had MD degrees (tabe 3). For the three resident groups, age, sex, conference attendance, US versus foreign medica schoo training, graduate training degree, and postgraduate year for residency training were found to be simiar. Change in attitude and behavior to medica event reporting The response rate for the medica event reporting attitude and behavior questionnaire was 100% (n = 30). The median change in attitude and behaviour to medica event reporting was zero, and was not significanty different 6 months after impementation of the patient safety educationa program compared with baseine (p = 0.566, Wicoxon sign rank test). Figure 1 shows the attitude and behaviour to medica event reporting at baseine and 6 months after impementation of the program. On average, 65% of the study participants attended each of the patient safety educationa program conferences (range 55 76%), with none of the study participants having attended significanty more or ess of these sessions than the others. Of the study participant characteristics, ony the number of patient safety educationa program conferences attended by the study participants (median 4, range 2 6) were significanty correated with the change in the medica event reporting attitude and behavior score (rho = 0.525, p = 0.003) at the 6 month foow up (fig 2). Barriers to medica event reporting A of the medica events reported into the paper based event reporting program were from the famiy medicine program cinics. Cinic staff initiated 73% of the medica event reports, the remaining reports were initiated by the graduate trainees. The response rate for barriers to the medica event reporting (3) Number of conferences attended Figure 2 Change in attitude and behavior to medica event reporting in reation to number of conferences attended. The number of study participants for each of the medica event reporting attitude and behavior scores and the number of conferences attended (1 6) are incuded in parentheses if the number is greater than 1. (2)

5 Medica education and medica event reporting 387 Tabe 4 Barriers to positive medica event reporting change by study participants (n = 22) Barriers to medica event reporting* Lack of time due to other cinic duties 13 Too much paperwork invoved in the reporting 6 process Reporting interrupts the work process 6 Career and persona reputation may be at stake 7 Institution and/or residency training program is not 3 ikey to make changes based on the medica event reporting resuts Facuty do not encourage residents to report medica 3 events Timey and high quaity feedback on medica event 1 reports for the purpose of resident training is not adequate Resident medica event reporting does not contribute 1 to my training as a resident No of study participants *The other category for barriers to medica event reporting incuded six responses indicating that there were no medica events noted during the study period to report. survey was 73.3% (n = 22). The most frequenty cited barrier was a ack of time to report due to other cinic duties (13 responses). An opportunity for respondents to suggest a barrier not incuded in the instrument was provided. Major barriers to medica event reporting were ack of time, extra paper work, and concern about career and persona reputation (tabe 4). DISCUSSION Athough the medica event educationa program reated to the ambuatory care practice setting impemented in our study was not associated with a positive change in attitude and behaviour to medica event reporting at the 6 month foow up compared with baseine (fig 1), it was noted that participation in this program was key for promoting a positive change in medica event reporting attitude and behavior among graduate trainees (fig 2). Major barriers to medica event reporting were ack of time, extra paper work, and concern about career and persona reputation. The patient safety program impemented in this study to promote a positive attitude and behaviour to medica event reporting among graduate trainees in the ambuatory care practice setting used methods for promoting this change that are consistent with recommendations for teaching medica students and residents about error in heath care. 11 These methods incuded an introductory ecture foowed by a video dramatization of a bood transfusion event taken from an episode of the currenty running teevision show ER, and medica event cases derived from ambuatory care that were reevant to the earner s training experience. These medica event cases were discussed using an interactive and probem based format that incuded a discussion of countermeasures to reduce the probabiity of their recurrence (tabe 1). In addition, based on the study of socia change, 16 it is ikey that a key factor for promoting attendance in this program is the active participation of facuty and other educationa roe modes for patient safety invoved in its curricuum. In addition, because of the competitive environment in which medica graduates are trained 17 and the negative impact that a medica error can have on a physician s emotiona state, 18 we conducted a survey of the study participants to identify the barriers to anonymous medica event reporting at the 6 month foow up. In this survey, 59% of the survey respondents indicated that the most important barrier to medica event reporting was the ack of time due to cinica duties (tabe 4); 27% of the respondents indicated that the other important barriers were the excess paper work invoved in reporting a medica event and the risk of jeopardizing one s career and persona reputation due to reporting these events. Furthermore, 27% of the program participants reported that they did not recognize any medica events to report, indicating that they may have had a ack of awareness as to what constituted a medica event. Two of the program participants reported that they were unfamiiar with the process for reporting medica events, which coud have been because they did not attend the introductory ecture to the program. These findings are consistent with past research, which indicates that the major factors contributing to under-reporting of medica events are understanding what constitutes a medica event, how to use the medica event reporting system, and whether the reporting is anonymous or mandatory The resuts of this survey highight two major areas that need to be addressed to increase medica event reporting by graduate trainees: (1) the medica event reporting process needs to be streamined to save time and eiminate unnecessary paper work; and (2) some of the survey respondents indicated that they did not note any medica events to report, as has been reported previousy with graduate trainees, 9 emphasizing that graduate medica training programs need to educate their trainees as to what constitutes a medica event and provide exampes of medica events throughout their training experience. Facuty need to be roe modes for their graduate trainees by reporting medica events themseves, encouraging their trainees to report medica events, and to provide emotiona counseing and support as needed to those touched by medica events or participating in the medica event discosure and anaysis process The concern about one s career and/or reputation being at stake due to discosure of medica events is to be expected, but shoud be ess threatening if the medica events are reported anonymousy. A potentia imitation of the study was that the adapted Prochaska and DiCemente stages of change mode 14 used in our study to assess the acquisition of medica event reporting behavior was not specificay vaidated for this purpose. However, it is important to note that Wiey and coeagues showed, in a recent study, 21 that the predictive vaidity of the stages of change mode 14 was supported by a significant association between the stages of change for medication adherence and eectronicay monitored medication taking behavior (p,0.03). Another recent study by Donovan and coeagues 22 determined that the test-retest reiabiity of the stages of change mode, 14 as appied to acquiring more exercise activity, was aso moderatey encouraging (k = 0.52). The most important imitation of the study is that a onger foow up period is needed to determine whether this study s educationa program is effective for promoting a positive change in attitude and behaviour to medica event reporting since such a change must be made at both the individua and organizationa eves. 23 It woud aso be worthwhie to test the effectiveness of this program in a variety of academic settings to further assess its effectiveness for promoting a positive change in medica event attitude and behavior. Future studies wi need to evauate the effectiveness of educationa programs for improving patient safety cuture over periods of time onger than 6 months and, if possibe, to incude a contro group in the study design. To date, appications of the Prochaska and DeCemente stages of change mode 14 have been successfu in changing heath behaviors within a 6 month period of time for exampe, cigarette smoking. Greater success for changing heath behaviors may be primariy reated to the individua having more contro over the associated environmenta infuences. Changing behavior reated to medica event

6 388 Coye, Mercer, Murphy-Cuen, et a Key messages N Past studies have shown that medica events invoving graduate medica trainees have the potentia for causing harm to heathcare recipients. N It is beieved that incorporating patient safety education into graduate medica training programs offers the opportunity to improve patient safety. N The study indicated that attending a patient safety educationa program was key for promoting a positive change in the attitude and behaviour to graduate medica trainees to medica event reporting at the 6 month foow up. N The study indicated that major barriers to medica event reporting were ack of time, extra paper work, and concerns about career and persona reputation N Facuty need to be roe modes for their graduate medica trainees by reporting medica events themseves, encouraging their trainees to report medica events, and providing emotiona counseing and support to those trainees touched by medica events or participating in the medica event discosure anaysis process. N Future research wi need to focus on reducing barriers to medica event reporting and to evauate the effectiveness of patient safety educationa programs over a onger period of time, since making a positive change in medica event reporting attitude and behavior must be made at both the individua and organizationa eves reporting woud probaby invove factors within the cuture of the organization or profession over which the individua may have itte contro. It is therefore ikey that successfu adoption of medica event reporting for graduate trainees wi require changes within the organization that promote patient safety. Westrum 23 indicated that a major step in moving an organization towards a generative safety cuture, or one in which the focus is on the members of an organization as a whoe moving forward to improve patient outcomes based on medica error information, was to engage a members of the organization in detecting and reporting medica events. Successfuy creating a generative patient safety cuture for a graduate medica training program is ikey to require integrating patient safety activities such as medica error discosure and participation in the anaysis of medica events and documenting competence for these activities. The key to creating this type of environment is to integrate patient safety earning activities throughout the residency curricuum, which wi require the residency program facuty to know how to use medica errors as educationa toos since much of medica education is apprenticeship. 11 Further progress towards successfuy institutionaizing patient safety as a vaue in the cuture of medicine is to make it an integra part of the academic curricuum, beginning with the first year of medica schoo 6 and extending throughout residency training and ater on in the curricuum of continuing medica education. 24 ACKNOWLEDGEMENTS The authors thank Juia Hoy Reza for interviewing facuty in The University of Texas Southwestern Medica Center at Daas Famiy Medicine and Community Medicine Department to obtain medica event case materias for the study.... Authors affiiations Y M Coye, S Q Mercer, Interna Medicine, The University of Texas Southwestern Medica Center at Daas, TX, USA C L Murphy-Cuen, G W Schneider, Famiy Practice and Community Medicine, The University of Texas Southwestern Medica Center at Daas, TX, USA L S Hynan, Psychiatry, Center for Biostatistics and Cinica Sciences, The University of Texas Southwestern Medica Center at Daas, TX, USA This study was funded by the 2001 GaxoSmithKine Pharmacy Research Award. The contents of the artice refect the view of the authors, not the officia position or poicy of GaxoSmithKine. Competing interests: none decared. REFERENCES 1 Kohn LT, Corrigan JM, Donason MS. To err is human: buiding a safer heathcare system. Washington, DC: Nationa Academy Press, Joint Committee of the Group on Resident Affairs and Organization of Resident Representatives. Report on patient safety and graduate medica education, Association of American Medica Coeges, Sandars J, Esmai A. The frequency and nature of medica error in primary care: understanding the diversity across studies. Fam Pract 2003;20: Pizzi LT, Godfarb NI, Nash DB. Promoting a cuture of safety. In:Shojania KG, Duncan BW, McDonad DM, et a.making heath care safer:a critica anaysis of patient safety practices.evidence Report/Technoogy Assessment Number 43.Prepared by the University of Caifornia at San Francisco Stanford Evidence-Based Practice Center under Contract No AHRQ Pubication No.01-E058. Rockvie, MD: Agency for Heathcare Research and Quaity, Battes JB, Shea CE. A system of anayzing medica errors to improve GME curricua and programs. Acad Med 2001;76: Leape LL. Error in medicine. JAMA 1994;272: Meyer BA. A student teaching modue: physician errors. Fam Med 1989;21: Habach JL, Suivan L. Medica error education in a third year medicine cerkship. Patient Safety Initiative 2000: Spotighting Strategies, Sharing Soutions. 9 Sorkin R, Caves JL, Kane GC, et a. The near miss resident conference: understanding the barriers to confronting medica errors. Semin Med Pract 2002;5: Campbe DT, Staney JC. Experimenta and quasi-experimenta designs for research. Boston: Houghton Miffin Company, Gosbee J, Stahhut R. Teaching medica students and residents about error in heathcare. In:Examining errors in heath care:deveoping a prevention, education and research agenda.proceedings of the 1st mutidiscipinary eadership conference convened by AAAS, AMA, Annenberg Center for Heath Sciences at Eisenhower, JCAHO. Rancho Mirage, CA: October, van der Schaaf TW. Near miss reporting in the chemica process industry, PhD Dissertation. Eindhoven, The Netherands: Eindhoven University of Technoogy, Hemreich RL. On error management: essons from aviation. BMJ 2000;320: Prochaska JO, DiCemente CC. Stages of change in the modification of probem behaviors. In: Hersen M, Eiser RM, Mier PM, eds. Progress in behavior modification. Sycamore, IL: Sycamore Pubishing Company, Barach P, Sma SD. Reporting and preventing medica mishaps: essons from non-medica near miss reporting systems. BMJ 2000;320: Rogers EM. Diffusion of innovations. New York: The Free Press, Christiansen JF, Levinson W, Dunn PM. The heart of darkness: the impact of perceived mistakes on physicians. J Gen Intern Med 1992;7: Pinkus R. Learning to keep a cautious tongue: the reporting of mistakes in neurosurgery, 1890 to In: Zooth L, eds. Margin of error: the ethics of mistakes in the practice of medicine. Hagerstown, MD: University Pubishing Group, Kuze AJ, Woof SH, Enge JD, et a. Making the case for a quaitative study of medica errors in primary care. Quait Heath Res 2003;13: Wu AW, Fokman S, McPhee SJ, et a. Do house officers earn from their mistakes? JAMA 1991;265: Wiey C, Redding C, Stafford J, et a. Stages of change for adherence with medication regimens for chronic disease: deveopment and vaidation of a measure. Cin Ther 2000;22: Donovan RJ, Jones S, Homan CD, et a. Assessing the reiabiity of a sage of change scae. Heath Educ Res 1998;13: Westrum R. Cutures with requisite imagination. In: Wise JA, Hopkin VD, eds. Verification and vaidation of compex systems: human factors aspects. Berin: Spring Verag, Ekin PL, Gorman PN. Continuing medica education and patient safety: an agenda for ifeong earning. J Am Inform Assoc 2002;9(Supp):S