Todd I. Smith, MD

The clinical learning model in medical education, specifically in the third and fourth years of medical school and in residency and fellowship training, is driven by direct patient‐care experiences and complemented by mentorship and supervision provided by experienced physicians.[1] Despite the emphasis on experiential learning in medical school and graduate training, the ability of educators to quantify the clinical experiences of learners has been limited. Case logs, often self‐reported, are frequently required during educational rotations to attempt to measure clinical experience.[2] Logs have been utilized to document diagnoses, demographics, disease severity, procedures, and chief complaints.[3, 4, 5, 6] Unfortunately, self‐reported logs are vulnerable to delayed updates, misreported data, and unreliable data validation.[7, 8] Automated data collection has been shown to be more reliable than self‐reported logs.[8, 9]

The enhanced data mining methods now available allow educators to appraise learners' exposures during patient‐care interactions beyond just the diagnosis or chief complaint (eg, how many electrocardiograms do our learners evaluate during a cardiology rotation, how often do our learners gain experience prescribing a specific class of antibiotics, how many of the patients seen by our learners are diabetic). For example, a learner's interaction with a patient during an inpatient admission for community‐acquired pneumonia, at minimum, would include assessing of past medical history, reviewing outpatient medications and allergies, evaluating tests completed (chest x‐ray, complete blood count, blood cultures), prescribing antibiotics, and monitoring comorbidities. The lack of knowledge regarding the frequency and context of these exposures is a key gap in our understanding of the clinical experience of inpatient trainees. Additionally, there are no data on clinical exposures specific to team‐based inpatient learning. When a rotation is team‐based, the educational experience is not limited to the learner's assigned patients, and this arrangement allows for educational exposures from patients who are not the learner's primary assignments through experiences gained during team rounds, cross‐coverage assessments, and informal discussions of patient care.

In this study, we quantify the clinical exposures of learners on an acting internship (AI) rotation in internal medicine by utilizing the Veterans Affairs (VA) electronic medical records (EMR) as collected through the VA Veterans Integrated Service Network 10 Clinical Data Warehouse (CDW). The AI or subinternship is a medical school clinical rotation typically completed in the fourth year, where the learning experience is expected to mirror a 1‐month rotation of a first‐year resident.[10] The AI has historically been defined as an experiential curriculum, during which students assume many of the responsibilities and activities that they will manage as graduate medical trainees.[10, 11] The exposures of AI learners include primary diagnoses encountered, problem lists evaluated at the time of admission, medications prescribed, laboratory tests ordered, and radiologic imaging evaluated. We additionally explored the exposures of the AI learner's team to assess the experiences available through team‐based care.

METHODS

This study was completed at the Louis Stokes Veterans Affairs Medical Center (LSVAMC) in Cleveland, Ohio, which is an academic affiliate of the Case Western Reserve University School of Medicine. The study was approved by the LSVAMC institutional review board.

At the LSVAMC, the AI rotation in internal medicine is a 4‐week inpatient rotation for fourth‐year medical students, in which the student is assigned to an inpatient medical team consisting of an attending physician, a senior resident, and a combination of first‐year residents and acting interns. Compared to a first‐year resident, the acting intern is assigned approximately half of the number of admissions. The teams rounds as a group at least once per day. Acting interns are permitted to place orders and write notes in the EMR; all orders require a cosignature by a resident or attending physician to be released.

We identified students who rotated through the LSVAMC for an AI in internal medicine rotation from July 2008 to November 2011 from rotation records. Using the CDW, we queried student names and their rotation dates and analyzed the results using a Structured Query Language Query Analyzer. Each student's patient encounters during the rotation were identified. A patient encounter was defined as a patient for whom the student wrote at least 1 note titled either Medicine Admission Note or Medicine Inpatient Progress Note, on any of the dates during their AI rotation. We then counted the total number of notes written by each student during their rotation. A patient identifier is associated with each note. The number of distinct patient identifiers was also tallied to establish the total number of patients seen during the rotation by the individual student as the primary caregiver.

We associated each patient encounter with an inpatient admission profile that included patient admission and discharge dates, International Classification of Diseases, 9th Revision (ICD‐9) diagnosis codes, and admitting specialty. Primary diagnosis codes were queried for each admission and were counted for individual students and in aggregate. We tallied both the individual student and aggregate patient medications prescribed during the dates of admission and ordered to a patient location consistent with an acute medical ward (therefore excluding orders placed if a patient was transferred to an intensive care unit). Similar queries were completed for laboratory and radiological testing.

The VA EMR keeps an active problem list on each patient, and items are associated with an ICD‐9 code. To assemble the active problems available for evaluation by the student on the day of a patient's admission, we queried all problem list items added prior to, but not discontinued before, the day of admission. We then tallied the results for every patient seen by each individual student and in aggregate.

To assess the team exposures for each AI student, we queried all discharge summaries cosigned by the student's attending during the dates of the student's rotation. We assumed the student's team members wrote these discharge summaries. After excluding the student's patients, the resultant list represented the team patient exposures for each student. This list was also queried for the number of patients seen, primary diagnoses, medications, problems, labs, and radiology. The number of team admissions counted included all patients who spent at least 1 day on the team while the student was rotating. All other team exposure counts completed included only patients who were both admitted and discharged within the dates of the student's rotation.

RESULTS

An AI rotation is 4 weeks in duration. Students competed a total of 128 rotations from July 30, 2008 through November 21, 2011. We included all rotations during this time period in the analysis. Tables 1, 2, 3, 4, 5 report results in 4 categories. The Student category tallies the total number of specific exposures (diagnoses, problems, medications, lab values, or radiology tests) for all patients primarily assigned to a student. The Team category tallies the total number of exposures for all patients assigned to other members of the student's inpatient team. The Primary % category identifies the percentage of students who had at least 1 assigned patient with the evaluated clinical exposure. The All Patients % category identifies the percentage of students who had at least 1 student‐assigned patient or at least 1 team‐assigned patient with the evaluated clinical exposure.

DISCUSSION

As medical educators, we assume that the clinical training years allow learners to develop essential skills through their varied clinical experiences. Through exposure to direct patient care, to medical decision‐making scenarios, and to senior physician management practices, trainees build the knowledge base for independent practice. To ensure there is sufficient clinical exposure, data on what trainees are encountering may prove beneficial.

In this novel study, we quantified what learners encounter during a 1‐month team‐based inpatient rotation at a large teaching hospital. We effectively measured a number of aspects of internal medicine inpatient training that have been difficult to quantify in the past. The ability to extract learner‐specific data is becoming increasingly available in academic teaching hospitals. For example, VA medical centers have available a daily updated national data warehouse. The other steps necessary for using learner‐specific data include an understanding of the local inpatient processhow tests are ordered, what note titles are used by traineesas well as someone able to build the queries necessary for data extraction. Once built, data extraction should be able to continue as an automated process and used in real time by medical educators.

Our method of data collection has limitations. The orders placed on a learner's primary patients may not have been placed by the learner. For example, orders may have been placed by an overnight resident cross‐covering the learner's patients. We assumed that learners evaluated the results of all tests (or medication changes) that occurred at any time during their rotation, including cross‐cover periods or days off. In addition, our method for evaluating team exposure underestimates the number of team patients calculated for each learner by limiting the query only to patients whose hospital stay was completed before the student left the inpatient service. It is also difficult to know the how many of the exposures are realized by the learner. Differences in learner attention, contrasts in rounding styles, and varying presentation methods will affect the number of exposures truly attained by the learner. Finally, not all clinical exposures can be evaluated through review of an EMR. Clinical experiences, such as care coordination, patient education, and family counseling, cannot be easily extracted.

Data mining EMRs can enhance clinical medical education. Although our data collection was completed retrospectively, we could easily provide learner‐specific data in real time to ward attendings, chief residents, and program directors. This information could direct the development of teaching tools and individualization of curricula. Perhaps, even more importantly, it would also allow educators to define curricular gaps. Whether these gaps are due to the particular patient demographics of a medical center, the practice patterns and strengths of a particular institution, or career interests of a trainee, these gaps may skew the patient‐care experiences encountered by individual trainees. We can use these data to identify differences in clinical experience and then develop opportunities for learnersclinical, didactic, or simulatedto address deficiencies and provide well‐rounded clinical experiences.

Further investigation to better understand the relationship between direct patient‐care experience and clinical skill acquisition is needed. This information could help guide the development of standards on the number of exposures we expect our learners to have with different diagnostic or treatment modalities prior to independent practice. Using learner data to better understand the clinical experiences of our medical trainees, we can hopefully develop more precise and focused curricula to ensure we produce competent graduates.

The clinical learning model in medical education, specifically in the third and fourth years of medical school and in residency and fellowship training, is driven by direct patient‐care experiences and complemented by mentorship and supervision provided by experienced physicians.[1] Despite the emphasis on experiential learning in medical school and graduate training, the ability of educators to quantify the clinical experiences of learners has been limited. Case logs, often self‐reported, are frequently required during educational rotations to attempt to measure clinical experience.[2] Logs have been utilized to document diagnoses, demographics, disease severity, procedures, and chief complaints.[3, 4, 5, 6] Unfortunately, self‐reported logs are vulnerable to delayed updates, misreported data, and unreliable data validation.[7, 8] Automated data collection has been shown to be more reliable than self‐reported logs.[8, 9]

The enhanced data mining methods now available allow educators to appraise learners' exposures during patient‐care interactions beyond just the diagnosis or chief complaint (eg, how many electrocardiograms do our learners evaluate during a cardiology rotation, how often do our learners gain experience prescribing a specific class of antibiotics, how many of the patients seen by our learners are diabetic). For example, a learner's interaction with a patient during an inpatient admission for community‐acquired pneumonia, at minimum, would include assessing of past medical history, reviewing outpatient medications and allergies, evaluating tests completed (chest x‐ray, complete blood count, blood cultures), prescribing antibiotics, and monitoring comorbidities. The lack of knowledge regarding the frequency and context of these exposures is a key gap in our understanding of the clinical experience of inpatient trainees. Additionally, there are no data on clinical exposures specific to team‐based inpatient learning. When a rotation is team‐based, the educational experience is not limited to the learner's assigned patients, and this arrangement allows for educational exposures from patients who are not the learner's primary assignments through experiences gained during team rounds, cross‐coverage assessments, and informal discussions of patient care.

In this study, we quantify the clinical exposures of learners on an acting internship (AI) rotation in internal medicine by utilizing the Veterans Affairs (VA) electronic medical records (EMR) as collected through the VA Veterans Integrated Service Network 10 Clinical Data Warehouse (CDW). The AI or subinternship is a medical school clinical rotation typically completed in the fourth year, where the learning experience is expected to mirror a 1‐month rotation of a first‐year resident.[10] The AI has historically been defined as an experiential curriculum, during which students assume many of the responsibilities and activities that they will manage as graduate medical trainees.[10, 11] The exposures of AI learners include primary diagnoses encountered, problem lists evaluated at the time of admission, medications prescribed, laboratory tests ordered, and radiologic imaging evaluated. We additionally explored the exposures of the AI learner's team to assess the experiences available through team‐based care.

METHODS

This study was completed at the Louis Stokes Veterans Affairs Medical Center (LSVAMC) in Cleveland, Ohio, which is an academic affiliate of the Case Western Reserve University School of Medicine. The study was approved by the LSVAMC institutional review board.

At the LSVAMC, the AI rotation in internal medicine is a 4‐week inpatient rotation for fourth‐year medical students, in which the student is assigned to an inpatient medical team consisting of an attending physician, a senior resident, and a combination of first‐year residents and acting interns. Compared to a first‐year resident, the acting intern is assigned approximately half of the number of admissions. The teams rounds as a group at least once per day. Acting interns are permitted to place orders and write notes in the EMR; all orders require a cosignature by a resident or attending physician to be released.

We identified students who rotated through the LSVAMC for an AI in internal medicine rotation from July 2008 to November 2011 from rotation records. Using the CDW, we queried student names and their rotation dates and analyzed the results using a Structured Query Language Query Analyzer. Each student's patient encounters during the rotation were identified. A patient encounter was defined as a patient for whom the student wrote at least 1 note titled either Medicine Admission Note or Medicine Inpatient Progress Note, on any of the dates during their AI rotation. We then counted the total number of notes written by each student during their rotation. A patient identifier is associated with each note. The number of distinct patient identifiers was also tallied to establish the total number of patients seen during the rotation by the individual student as the primary caregiver.

We associated each patient encounter with an inpatient admission profile that included patient admission and discharge dates, International Classification of Diseases, 9th Revision (ICD‐9) diagnosis codes, and admitting specialty. Primary diagnosis codes were queried for each admission and were counted for individual students and in aggregate. We tallied both the individual student and aggregate patient medications prescribed during the dates of admission and ordered to a patient location consistent with an acute medical ward (therefore excluding orders placed if a patient was transferred to an intensive care unit). Similar queries were completed for laboratory and radiological testing.

The VA EMR keeps an active problem list on each patient, and items are associated with an ICD‐9 code. To assemble the active problems available for evaluation by the student on the day of a patient's admission, we queried all problem list items added prior to, but not discontinued before, the day of admission. We then tallied the results for every patient seen by each individual student and in aggregate.

To assess the team exposures for each AI student, we queried all discharge summaries cosigned by the student's attending during the dates of the student's rotation. We assumed the student's team members wrote these discharge summaries. After excluding the student's patients, the resultant list represented the team patient exposures for each student. This list was also queried for the number of patients seen, primary diagnoses, medications, problems, labs, and radiology. The number of team admissions counted included all patients who spent at least 1 day on the team while the student was rotating. All other team exposure counts completed included only patients who were both admitted and discharged within the dates of the student's rotation.

RESULTS

An AI rotation is 4 weeks in duration. Students competed a total of 128 rotations from July 30, 2008 through November 21, 2011. We included all rotations during this time period in the analysis. Tables 1, 2, 3, 4, 5 report results in 4 categories. The Student category tallies the total number of specific exposures (diagnoses, problems, medications, lab values, or radiology tests) for all patients primarily assigned to a student. The Team category tallies the total number of exposures for all patients assigned to other members of the student's inpatient team. The Primary % category identifies the percentage of students who had at least 1 assigned patient with the evaluated clinical exposure. The All Patients % category identifies the percentage of students who had at least 1 student‐assigned patient or at least 1 team‐assigned patient with the evaluated clinical exposure.

DISCUSSION

As medical educators, we assume that the clinical training years allow learners to develop essential skills through their varied clinical experiences. Through exposure to direct patient care, to medical decision‐making scenarios, and to senior physician management practices, trainees build the knowledge base for independent practice. To ensure there is sufficient clinical exposure, data on what trainees are encountering may prove beneficial.

In this novel study, we quantified what learners encounter during a 1‐month team‐based inpatient rotation at a large teaching hospital. We effectively measured a number of aspects of internal medicine inpatient training that have been difficult to quantify in the past. The ability to extract learner‐specific data is becoming increasingly available in academic teaching hospitals. For example, VA medical centers have available a daily updated national data warehouse. The other steps necessary for using learner‐specific data include an understanding of the local inpatient processhow tests are ordered, what note titles are used by traineesas well as someone able to build the queries necessary for data extraction. Once built, data extraction should be able to continue as an automated process and used in real time by medical educators.

Our method of data collection has limitations. The orders placed on a learner's primary patients may not have been placed by the learner. For example, orders may have been placed by an overnight resident cross‐covering the learner's patients. We assumed that learners evaluated the results of all tests (or medication changes) that occurred at any time during their rotation, including cross‐cover periods or days off. In addition, our method for evaluating team exposure underestimates the number of team patients calculated for each learner by limiting the query only to patients whose hospital stay was completed before the student left the inpatient service. It is also difficult to know the how many of the exposures are realized by the learner. Differences in learner attention, contrasts in rounding styles, and varying presentation methods will affect the number of exposures truly attained by the learner. Finally, not all clinical exposures can be evaluated through review of an EMR. Clinical experiences, such as care coordination, patient education, and family counseling, cannot be easily extracted.

Data mining EMRs can enhance clinical medical education. Although our data collection was completed retrospectively, we could easily provide learner‐specific data in real time to ward attendings, chief residents, and program directors. This information could direct the development of teaching tools and individualization of curricula. Perhaps, even more importantly, it would also allow educators to define curricular gaps. Whether these gaps are due to the particular patient demographics of a medical center, the practice patterns and strengths of a particular institution, or career interests of a trainee, these gaps may skew the patient‐care experiences encountered by individual trainees. We can use these data to identify differences in clinical experience and then develop opportunities for learnersclinical, didactic, or simulatedto address deficiencies and provide well‐rounded clinical experiences.

Further investigation to better understand the relationship between direct patient‐care experience and clinical skill acquisition is needed. This information could help guide the development of standards on the number of exposures we expect our learners to have with different diagnostic or treatment modalities prior to independent practice. Using learner data to better understand the clinical experiences of our medical trainees, we can hopefully develop more precise and focused curricula to ensure we produce competent graduates.

The clinical learning model in medical education, specifically in the third and fourth years of medical school and in residency and fellowship training, is driven by direct patient‐care experiences and complemented by mentorship and supervision provided by experienced physicians.[1] Despite the emphasis on experiential learning in medical school and graduate training, the ability of educators to quantify the clinical experiences of learners has been limited. Case logs, often self‐reported, are frequently required during educational rotations to attempt to measure clinical experience.[2] Logs have been utilized to document diagnoses, demographics, disease severity, procedures, and chief complaints.[3, 4, 5, 6] Unfortunately, self‐reported logs are vulnerable to delayed updates, misreported data, and unreliable data validation.[7, 8] Automated data collection has been shown to be more reliable than self‐reported logs.[8, 9]

The enhanced data mining methods now available allow educators to appraise learners' exposures during patient‐care interactions beyond just the diagnosis or chief complaint (eg, how many electrocardiograms do our learners evaluate during a cardiology rotation, how often do our learners gain experience prescribing a specific class of antibiotics, how many of the patients seen by our learners are diabetic). For example, a learner's interaction with a patient during an inpatient admission for community‐acquired pneumonia, at minimum, would include assessing of past medical history, reviewing outpatient medications and allergies, evaluating tests completed (chest x‐ray, complete blood count, blood cultures), prescribing antibiotics, and monitoring comorbidities. The lack of knowledge regarding the frequency and context of these exposures is a key gap in our understanding of the clinical experience of inpatient trainees. Additionally, there are no data on clinical exposures specific to team‐based inpatient learning. When a rotation is team‐based, the educational experience is not limited to the learner's assigned patients, and this arrangement allows for educational exposures from patients who are not the learner's primary assignments through experiences gained during team rounds, cross‐coverage assessments, and informal discussions of patient care.

In this study, we quantify the clinical exposures of learners on an acting internship (AI) rotation in internal medicine by utilizing the Veterans Affairs (VA) electronic medical records (EMR) as collected through the VA Veterans Integrated Service Network 10 Clinical Data Warehouse (CDW). The AI or subinternship is a medical school clinical rotation typically completed in the fourth year, where the learning experience is expected to mirror a 1‐month rotation of a first‐year resident.[10] The AI has historically been defined as an experiential curriculum, during which students assume many of the responsibilities and activities that they will manage as graduate medical trainees.[10, 11] The exposures of AI learners include primary diagnoses encountered, problem lists evaluated at the time of admission, medications prescribed, laboratory tests ordered, and radiologic imaging evaluated. We additionally explored the exposures of the AI learner's team to assess the experiences available through team‐based care.

METHODS

This study was completed at the Louis Stokes Veterans Affairs Medical Center (LSVAMC) in Cleveland, Ohio, which is an academic affiliate of the Case Western Reserve University School of Medicine. The study was approved by the LSVAMC institutional review board.

At the LSVAMC, the AI rotation in internal medicine is a 4‐week inpatient rotation for fourth‐year medical students, in which the student is assigned to an inpatient medical team consisting of an attending physician, a senior resident, and a combination of first‐year residents and acting interns. Compared to a first‐year resident, the acting intern is assigned approximately half of the number of admissions. The teams rounds as a group at least once per day. Acting interns are permitted to place orders and write notes in the EMR; all orders require a cosignature by a resident or attending physician to be released.

We identified students who rotated through the LSVAMC for an AI in internal medicine rotation from July 2008 to November 2011 from rotation records. Using the CDW, we queried student names and their rotation dates and analyzed the results using a Structured Query Language Query Analyzer. Each student's patient encounters during the rotation were identified. A patient encounter was defined as a patient for whom the student wrote at least 1 note titled either Medicine Admission Note or Medicine Inpatient Progress Note, on any of the dates during their AI rotation. We then counted the total number of notes written by each student during their rotation. A patient identifier is associated with each note. The number of distinct patient identifiers was also tallied to establish the total number of patients seen during the rotation by the individual student as the primary caregiver.

We associated each patient encounter with an inpatient admission profile that included patient admission and discharge dates, International Classification of Diseases, 9th Revision (ICD‐9) diagnosis codes, and admitting specialty. Primary diagnosis codes were queried for each admission and were counted for individual students and in aggregate. We tallied both the individual student and aggregate patient medications prescribed during the dates of admission and ordered to a patient location consistent with an acute medical ward (therefore excluding orders placed if a patient was transferred to an intensive care unit). Similar queries were completed for laboratory and radiological testing.

The VA EMR keeps an active problem list on each patient, and items are associated with an ICD‐9 code. To assemble the active problems available for evaluation by the student on the day of a patient's admission, we queried all problem list items added prior to, but not discontinued before, the day of admission. We then tallied the results for every patient seen by each individual student and in aggregate.

To assess the team exposures for each AI student, we queried all discharge summaries cosigned by the student's attending during the dates of the student's rotation. We assumed the student's team members wrote these discharge summaries. After excluding the student's patients, the resultant list represented the team patient exposures for each student. This list was also queried for the number of patients seen, primary diagnoses, medications, problems, labs, and radiology. The number of team admissions counted included all patients who spent at least 1 day on the team while the student was rotating. All other team exposure counts completed included only patients who were both admitted and discharged within the dates of the student's rotation.

RESULTS

An AI rotation is 4 weeks in duration. Students competed a total of 128 rotations from July 30, 2008 through November 21, 2011. We included all rotations during this time period in the analysis. Tables 1, 2, 3, 4, 5 report results in 4 categories. The Student category tallies the total number of specific exposures (diagnoses, problems, medications, lab values, or radiology tests) for all patients primarily assigned to a student. The Team category tallies the total number of exposures for all patients assigned to other members of the student's inpatient team. The Primary % category identifies the percentage of students who had at least 1 assigned patient with the evaluated clinical exposure. The All Patients % category identifies the percentage of students who had at least 1 student‐assigned patient or at least 1 team‐assigned patient with the evaluated clinical exposure.

DISCUSSION

As medical educators, we assume that the clinical training years allow learners to develop essential skills through their varied clinical experiences. Through exposure to direct patient care, to medical decision‐making scenarios, and to senior physician management practices, trainees build the knowledge base for independent practice. To ensure there is sufficient clinical exposure, data on what trainees are encountering may prove beneficial.

In this novel study, we quantified what learners encounter during a 1‐month team‐based inpatient rotation at a large teaching hospital. We effectively measured a number of aspects of internal medicine inpatient training that have been difficult to quantify in the past. The ability to extract learner‐specific data is becoming increasingly available in academic teaching hospitals. For example, VA medical centers have available a daily updated national data warehouse. The other steps necessary for using learner‐specific data include an understanding of the local inpatient processhow tests are ordered, what note titles are used by traineesas well as someone able to build the queries necessary for data extraction. Once built, data extraction should be able to continue as an automated process and used in real time by medical educators.

Our method of data collection has limitations. The orders placed on a learner's primary patients may not have been placed by the learner. For example, orders may have been placed by an overnight resident cross‐covering the learner's patients. We assumed that learners evaluated the results of all tests (or medication changes) that occurred at any time during their rotation, including cross‐cover periods or days off. In addition, our method for evaluating team exposure underestimates the number of team patients calculated for each learner by limiting the query only to patients whose hospital stay was completed before the student left the inpatient service. It is also difficult to know the how many of the exposures are realized by the learner. Differences in learner attention, contrasts in rounding styles, and varying presentation methods will affect the number of exposures truly attained by the learner. Finally, not all clinical exposures can be evaluated through review of an EMR. Clinical experiences, such as care coordination, patient education, and family counseling, cannot be easily extracted.

Data mining EMRs can enhance clinical medical education. Although our data collection was completed retrospectively, we could easily provide learner‐specific data in real time to ward attendings, chief residents, and program directors. This information could direct the development of teaching tools and individualization of curricula. Perhaps, even more importantly, it would also allow educators to define curricular gaps. Whether these gaps are due to the particular patient demographics of a medical center, the practice patterns and strengths of a particular institution, or career interests of a trainee, these gaps may skew the patient‐care experiences encountered by individual trainees. We can use these data to identify differences in clinical experience and then develop opportunities for learnersclinical, didactic, or simulatedto address deficiencies and provide well‐rounded clinical experiences.

Further investigation to better understand the relationship between direct patient‐care experience and clinical skill acquisition is needed. This information could help guide the development of standards on the number of exposures we expect our learners to have with different diagnostic or treatment modalities prior to independent practice. Using learner data to better understand the clinical experiences of our medical trainees, we can hopefully develop more precise and focused curricula to ensure we produce competent graduates.