1092.full

The Relationship of Cognitive, Personality, and Academic
Measures to Anesthesiology Resident Clinical Performance
David L. Reich, MD*, Suzan Uysal, PhD*, Carol A. Bodian, DrPH‡, Suzanne Gabriele, MA*,
Mary Hibbard, PhD†, Wayne Gordon, PhD†, Martin Sliwinki, PhD†, and Richard D. Kayne, MD*
Departments of *Anesthesiology, †Rehabilitation Medicine, and ‡Biomathematical Sciences, The Mount Sinai School of
Medicine, New York, New York

Cognitive skills (including vigilance), personality fac- these were subjected to a multivariate test (Mantel-
tors, and standardized academic test performance may Haenszel). Cognitive variables predicting poor clinical
be associated with clinical competence in anesthesiol- performance were difficulty performing a rapid mental
ogy to varying degrees. Sixty-seven anesthesiology res- arithmetic test requiring divided attention and com-
idents in training at one center between 1993 and 1995 mission errors during complex visual target detection.
were administered the modified Vigil (For Thought, Personality variables predicting poor clinical perfor-
Ltd., Nashua, NH), the Paced Auditory Serial Addition mance were introversion and flexibility. A predictive
Test, the California Personality Inventory, the State- academic variable was poor anesthesia knowledge as
Trait Anxiety Inventory, and five standardized aca- measured by using two different tests during the first
demic performance tests. The clinical performance of month of training. There were varying levels of inde-
anesthesiology residents was rated on a quarterly basis pendence among these variables. Implications: Early
by a clinical competence committee. A growth curve academic test performance and certain cognitive and
model indicated that there was significant variability in personality tests were associated with the clinical per-
clinical competence at the start of residency and a statis- formance of anesthesiology residents. The predictive
tically significant improvement over time, and that the value of these findings should be confirmed in a pro-
relative ranking of the residents remained stable over spective, multicenter study.
the course of training. Of 46 potential variables, 7 were
associated (P Ͻ 0.10) with poor clinical performance; (Anesth Analg 1999;88:1092–1100)

S
election and evaluation of resident candidates in sustain focused attention (vigilance), divide attention,
many disciplines of medicine has traditionally and process information rapidly may also be associated
been based on academic criteria, such as United with clinical competence. Vigilance (as assessed by mea-
States Medical Licensing Examination scores and medi- suring the reaction times to simulated abnormal intra-
cal school grades. Although academic test scores can operative physiological data) is of significance in anes-
serve as predictors of academic achievement, they may thesia competency, particularly with regard to the
be less useful in predicting actual clinical performance avoidance of critical incidents (6).
(1,2). Several studies have reported that specific person- We hypothesized that cognitive, personality, and
ality attributes are associated with clinical performance academic variables would be predictors of clinical per-
and have suggested that personality inventories may be formance during anesthesiology residency.
useful in selecting candidates for residency training pro-
grams (3–5). Another personality variable, trait-anxiety
(the tendency to respond to a wide range of situations as Methods
dangerous or threatening), and the cognitive abilities to
The study protocol was approved by the institutional
review board of the Mount Sinai School of Medicine.
This work was supported by an educational research grant from All residents in the anesthesiology residency program
the Foundation for Anesthesiology Education and Research.
Presented in part at the annual meeting of the American Society at The Mount Sinai School of Medicine between 1993
of Anesthesiologists, San Francisco, CA, October 1994. and 1995 were asked to participate in the study; only
Accepted for publication February 5, 1999. the data from residents who gave written, informed
Address correspondence and reprint requests to David L. Reich, MD, consent to participate were used. Of the 113 residents,
The Mount Sinai School of Medicine, One Gustave L. Levy Pl., Box
1010, New York, NY 10029. Address e-mail to david reich@smtplink 67 (59%) consented to participate in the study. Over
.mssm.edu. the 3 yr during which the study was conducted, 40

©1999 by the International Anesthesia Research Society
1092 Anesth Analg 1999;88:1092–1100 0003-2999/99

ANESTH ANALG ECONOMICS AND HEALTH SYSTEMS RESEARCH REICH ET AL. 1093
1999;88:1092–1100 PREDICTIVE EVALUATION OF RESIDENTS

two-letter sequences embedded in visual noise. Stim-
uli were presented for 85 ms, separated by an 850-ms
interstimulus interval. To better simulate the divided
attention demands present in operating rooms (ORs),
this task was modified according to the methodology
of Solberg and Mateer (unpublished data) to include
an auditory signal detection task consisting of an au-
dio tape recording of ventilator disconnect alarms in-
terspersed among normal OR background noise.
Thus, both auditory and visual signal detection tasks
were presented simultaneously. Errors of omission
and commission were recorded for both visual and
auditory attention tasks. Sustained and divided atten-
tion were also measured using the Paced Auditory
Serial Addition Test (PASAT) (8), in which subjects
Figure 1. The computerized daily evaluation system display of the add pairs of random digits presented via an audio
criteria for evaluating an anesthesiology resident in the domain of tape, such that each digit is added to the one imme-
essential character attributes. diately preceding it. The test consists of four series of
50 digits, and the rate of digit presentation increases
Table 1. Demographic Characteristics of Anesthesiology
across series. In this study, the first three rates of
Resident Study Sample
presentation were administered. In addition to the
Gender ability to sustain and divide attention, the PASAT
M 70.1 requires rapid information processing.
F 29.9 Residents’ academic performance was recorded for
Race
Caucasian 74.6 the following five examinations:
Non-Caucasian 25.4
Medical school
1. Anesthesia Knowledge Test 1 Pretest (Metrics
American 85.1 Associates, Inc., Chelmsford, MA) (first week of
International 14.9 training)
2. Anesthesia Knowledge Test 1 Posttest (Metrics
Values are expressed as percentages.
Associates) (first month of training)
3. Anesthesia Knowledge Test 6 (Metrics Associ-
first-year residents entered. During the first year of the ates) (after 6 mo of training)
study, 13 second-year and 14 third-year residents also 4. ABA-ASA In-Training Examination (Joint Coun-
entered the study. All test results were coded to pre- cil on In-Training Examinations, Park Ridge, IL)
serve participant confidentiality. Demographic char- (first month of training)
acteristics (age, race, gender, and United States versus 5. ABA-ASA In-Training Examination (Joint Coun-
international medical graduate status) were recorded. cil on In-Training Examinations) (after 13 mo of
Psychological tests were administered during the training)
first 6 mo of residency training using four published The members of the clinical competence committee
instruments with demonstrated validity and reliabil- and the residents were not blinded to the results of
ity. Both residents and clinical competence commit- these examinations.
tee members were blinded to the psychological test Clinical performance for anesthesiology residents
results. was evaluated using the five criteria defined by the
Personality was assessed using the California Per- American Board of Anesthesiology: essential character
sonality Inventory (CPI), which consists of 462 items attributes, acquired character skills, clinical skills,
comprising 23 personality scales. Trait-anxiety was judgment, and knowledge (9). All residents were rated
measured using a 20-item scale of the State-Trait Anx- on a scale of 1–5 of increasing competence (Fig. 1). At
iety Inventory (7). The Mount Sinai School of Medicine, the daily perfor-
The Vigil (For Thought, Ltd., Nashua, NH), a con- mance of anesthesiology residents was rated by at-
tinuous performance test, was used to measure sus- tending anesthesiologists using a computerized eval-
tained focused attention. It is a visual signal detection uation system. Using the same criteria, the clinical
task that requires subjects to observe a computer competence of anesthesiology residents rotating at af-
screen for an extended period of time (approximately filiate hospitals was assessed manually on a monthly
8 min) and to respond rapidly to simple, single-letter basis by attending physicians who had worked with
targets or to complex targets consisting of specific those residents. Both the computerized and manual

1094 ECONOMICS AND HEALTH SYSTEMS RESEARCH REICH ET AL. ANESTH ANALG
PREDICTIVE EVALUATION OF RESIDENTS 1999;88:1092–1100

Table 2. Significance Levels for Associations Among Selected Predictor Variables
PASAT3 Vigil-Co CPI-V1 CPI-V2 CPI-FX AKT-1 ITE-1

PASAT3 1.00
Vigil-Co 0.05 1.00
CPI-V1 0.09 0.82 1.00
CPI-V2 0.90 0.26 0.30 1.00
CPI-FX 0.82 0.20 0.71 Ͻ0.01 1.00
AKT-1 0.80 0.91 0.19 0.17 0.68 1.00
ITE-1 0.93 0.57 0.59 0.44 0.61 Ͻ0.01 1.00

PASAT3 ϭ third trial of the Paced Auditory Serial Addition Test, Vigil-Co ϭ commission errors during complex visual target detection on the Vigil (For
Thought Ltd., Nashua, NH), CPI-V1 ϭ California Personality Inventory Introversion/Extroversion Scale, CPI-V2 ϭ California Personality Inventory Norm-
Favoring/Norm-Doubting Scale, CPI-Fx ϭ California Personality Inventory Flexibility Scale, AKT-1 ϭ Anesthesia Knowledge Test 1 (administered during the
first week of training), ITE-1 ϭ ABA-ASA In-Training Examination (administered during the first month of training).

Table 3. Independent Contribution of PASAT3 to Predict evaluation systems provided attending anesthesiolo-
Poor Clinical Outcome gists the opportunity to make written comments. All
evaluations over 3-mo intervals were presented to the
Low commission High commission
errors errors
15–20 members of the clinical competence committee.
(Vigil-Co Յ5) (Vigil-Co Ͼ5) P ϭ 0.05 At quarterly meetings, the committee members used
PASAT3 Յ30 42.9 75.0
these daily and monthly evaluations to arrive at con-
PASAT3 30–40 10.0 60.0 sensus evaluation scores for each resident in the five
PASAT3 Ͼ40 14.3 0.0 domains of clinical competence.
Clinical competence data from the subset of subjects
Extroversion Introversion
(CPI-V1 Յ50) (CPI-V1 Ͼ50) P ϭ 0.04
entering the study as first-year residents were ana-
lyzed using growth curve methods (10) to determine
PASAT3 Յ30 33.3 80.0
the pattern of changes in clinical competence scores
PASAT3 30–40 18.2 50.0
PASAT3 Ͼ40 10.5 25.0
over the course of the residency. The demographic
variables were evaluated for all subjects by using ei-
Norm-doubting Norm-favoring ther ␹2 or Wilcoxon analyses to determine whether
(CPI-V2 Յ50) (CPI-V2 Ͼ50) P ϭ 0.01
these variables predicted status at baseline or change
PASAT3 Յ30 66.7 40.0 over time.
PASAT3 30–40 40.0 20.0 Residents were classified as having poor clinical
PASAT3 Ͼ40 16.7 9.1
performance if they left the program for poor perfor-
Inflexible Flexible mance or were in the lowest 25th percentile for Ͼ50%
(CPI-Fx Յ50) (CPI-Fx Ͼ50) P ϭ 0.01 of their evaluations. All other residents were classified
PASAT3 Յ30 25.0 71.4 as having good clinical performance. The cognitive,
PASAT3 30–40 20.0 40.0 personality, and academic tests produced 46 different
PASAT3 Ͼ40 0 21.4 scores for each resident. Each of the scores was di-
vided into quartiles, and the outcomes were examined
Poor academic Good academic
knowledge knowledge
for suggestions of appropriate categories for the sta-
(AKT-1 Յ50) (AKT-1 Ͼ50) P ϭ 0.02 tistical analyses. On this basis, scores were grouped
PASAT3 Յ30 44.4 100.0
into two or three categories, and ␹2 tests of association
PASAT3 30–40 40.0 0.0 or of trend were used to identify which of these meas-
PASAT3 Ͼ40 17.6 0.0 ures showed evidence of association with poor clinical
performance. A value of P Ͻ 0.10 was the criterion
Poor academic Good academic
knowledge knowledge
used for this screening process.
(ITE-1 Յ15) (ITE-1 Ͼ15) P ϭ 0.01

PASAT3 Յ30 57.1 50.0
PASAT3 30–40 25.0 25.0 Results
PASAT3 Ͼ40 23.1 0.0
The correlations among the five clinical competence
Values are presented as the percentage of subjects meeting the given domains (essential character attributes, acquired char-
criteria.
PASAT3 ϭ third trial of the Paced Auditory Serial Addition Test, Vigil- acter skills, knowledge, judgment, and clinical skills)
Co ϭ commission errors during complex visual target detection on the Vigil were statistically significant at every 3-mo assessment.
(For Thought Ltd., Nashua, NH), CPI-V1 ϭ California Personality Inventory
Introversion/Extroversion Scale, CPI-V2 ϭ California Personality Inventory For example, at the 3-mo assessment, the range of
Norm-Favoring/Norm-Doubting Scale, CPI-Fx ϭ California Personality In- correlation coefficients among all pairs of clinical com-
ventory Flexibility Scale, AKT-1 ϭ Anesthesia Knowledge Test 1 (adminis-
tered during the first week of training), ITE-1 ϭ ABA-ASA In-Training
petence variables was 0.51– 0.94 (all P Ͻ 0.001). This
Examination (administered during the first month of training). indicates that very little additional information would


Table 4. Independent Contribution of Vigil-Co to Predict Poor Clinical Outcome
Poor mental arithmetic Moderate mental arithmetic Good mental arithmetic
(PASAT3 Յ30) (PASAT3 30–40) (PASAT3 Ͼ40) P ϭ 0.08

Vigil-Co Յ5 42.9 10.0 14.3
Vigil-Co Ͼ5 75.0 60.0 0.0

(CPI-V1 Յ50) (CPI-V1 Ͼ50) P ϭ 0.01

Vigil-Co Յ5 10.7 40.0
Vigil-Co Ͼ5 42.9 100.0

Norm-doubting Norm-favoring
(CPI-V2 Յ50) (CPI-V2 Ͼ50) P ϭ 0.00

Vigil-Co Յ5 31.6 5.3
Vigil-Co Ͼ5 66.7 57.1

Inflexible Flexible
(CPI-Fx Յ50) (CPI-Fx Ͼ50) P ϭ 0.03

Vigil-Co Ͼ5 66.7 57.1

Poor academic knowledge Good academic knowledge
(AKT-1 Յ50) (AKT-1 Ͼ50) P ϭ 0.02

Vigil-Co Ͼ5 50.0 66.7

(ITE-1 Յ15) (ITE-1 Ͼ15) P ϭ 0.02

Vigil-Co Ͼ5 50.0 60.0


be obtained by analyzing these domains separately. Of the 67 participating residents, 3 (4%) left the
Therefore, the remainder of the results were derived program due to poor clinical performance, and 4 left
from analyses of the mean of the five domain scores. the program for other reasons. Of the 67 participating
A growth curve model was fit to the clinical com- residents who completed the residency program, 18
petence rating scores to model status at baseline and (27%) were classified as having poor clinical perfor-
improvement over the course of residency. This anal- mance based on their clinical competence ratings,
ysis was limited to the subset of subjects who entered yielding a total of 21 residents (31%) who met the
the study as first-year residents. The results from the criteria for poor clinical performance.
model indicate that the mean rating at the start of The demographic characteristics of the study sam-
residency (intercept) was 3.4 and that there was a ple are presented in Table 1. The sample was pre-
slight but statistically significant improvement over dominantly male, Caucasian, and American medical
time (slope 0.017/month; P Ͻ 0.01). Analysis of the school graduates. The median age was 28 yr. Age,
random effects showed significant individual differ- gender, race, and graduate status did not predict poor
ences in the intercept (z ϭ 4.41, P Ͻ 0.01), which clinical performance (P Ͼ 0.40).
indicates that there was significant variability among Of the 46 cognitive, personality, and academic vari-
the first-year residents at the start of residency. The ables, 7 met our criterion (P Ͻ 0.10) for a univariate
residents’ scores improved at an equivalent rate over predictor variable. The predictive cognitive variables
the course of the residency, as indicated by a random were low score on the third and most challenging trial
effect for slopes that was not statistically significant of the PASAT (PASAT3) and high number of commis-
(z ϭ 0.91, P ϭ 0.36). Thus, the relative ranking of the sion errors during complex visual target detection on
residents as reflected in the clinical competence rat- the Vigil (Vigil-Co). The personality variables that
ings was stable over time. Clinical competence ratings predicted poor clinical outcome were high CPI
at baseline also did not predict rate of change (z Ͻ 1). Introversion/Extroversion Scale score (CPI-V1), low


Table 5. Independent Contribution of CPI-V1 to Predict Poor Clinical Outcome

CPI-V1 Յ50 33.3 18.2 10.5
CPI-V1 Ͼ50 80.0 50.0 25.0

Low commission errors High commission errors
(Vigil-Co Յ5) (Vigil-Co Ͼ5) P ϭ 0.01

CPI-V1 Յ50 10.7 42.9
CPI-V1 Ͼ50 40.0 100.0

(CPI-V2 Յ50) (CPI-V2 Ͼ50) P ϭ 0.01

CPI-V1 Յ50 18.7 21.4
CPI-V1 Ͼ50 80.0 30.0

Inflexible Flexible

CPI-V1 Յ50 13.6 27.3
CPI-V1 Ͼ50 27.3 88.9

(AKT-1 Յ50) (AKT-1 Ͼ50) P ϭ 0.01

CPI-V1 Յ50 23.1 15.4
CPI-V1 Ͼ50 73.3 0.0

(ITE-1 Յ15) (ITE-1 Ͼ15) P ϭ 0.01

CPI-V1 Յ50 30.0 9.1
CPI-V1 Ͼ50 63.6 44.4


CPI Norm-Favoring/Norm-Doubting Scale score results demonstrate that the PASAT3 contributes ad-
(CPI-V2), and high CPI Flexibility Scale score (CPI-Fx). ditional information after controlling for the influence
Academic knowledge variables that predicted poor of each of the six other predictor variables. The pro-
performance were low scores on the Anesthesia portion of residents who ultimately demonstrated
Knowledge Test 1 administered during the first week poor clinical performance decreased as the PASAT3
of training (AKT-1), and the ABA-ASA In-Training score increased. This trend was observed for residents
Examination administered during the first month of with low and with high Vigil-Co scores, although the
training (ITE-1). overall level of poor performance differed in the two
Several of these variables were associated with each subgroups. Table 4 shows the comparable analysis of
other, and this association reached statistical signifi- the independent contribution of Vigil-Co. Five of six
cance in three instances (Table 2). Based on this and Mantel-Haenszel tests are statistically significant. This
the relatively small sample size, we believed that these analysis demonstrates that both low scores on PA-
data did not lend themselves to picking one “best” SAT3 and high rates of commission errors on the
combination of independent predictors, as in a multi- Vigil-Co (score Ͼ5) are independent predictors of
variate regression analysis. Instead, these data are pre- poor clinical performance.
sented in the form of contingency tables that allowed The independent contribution of high CPI-V1
tests of the independence of association among each of scores in predicting poor performance is shown
the seven univariate predictor variables. The Mantel- in Table 5. The statistically significant Mantel-
Haenszel test was used to test for statistical signifi- Haenszel tests demonstrate that CPI-V1 contributes
cance. These data are presented in Tables 3–9. additional information after controlling for the in-
The independent contribution of PASAT3 to pre- fluence of the six other predictor variables. In con-
dicting poor performance is demonstrated in Table 3, trast, the CPI-V2 was not a strong independent pre-
in which the statistically significant Mantel-Haenszel dictor in this sample; none of the Mantel-Haenszel


Table 6. Independent Contribution of CPI-V2 to Predict Poor Clinical Outcome

CPI-V2 Յ50 66.7 40.0 16.7
CPI-V2 Ͼ50 40.0 20.0 9.1


CPI-V2 Յ50 31.6 66.7
CPI-V2 Ͼ50 5.3 57.1

(CPI-V1 Յ50) (CPI-V1 Ͼ50) P ϭ 0.20

CPI-V2 Յ50 18.8 80.0
CPI-V2 Ͼ50 21.4 30.0

Inflexible Flexible

CPI-V2 Յ50 28.6 47.4
CPI-V2 Ͼ50 15.4 41.7

(AKT-1 Յ50) (AKT-1 Ͼ50) P ϭ 0.33

CPI-V2 Յ50 52.4 0.0
CPI-V2 Ͼ50 30.0 18.2

(ITE-1 Յ15) (ITE-1 Ͼ15) P ϭ 0.12

CPI-V2 Յ50 54.6 28.6
CPI-V2 Ͼ50 35.0 11.8


tests reached statistical significance (Table 6). High (Vigil-Co) and poor performance on a test of rapid
CPI-Fx scores seem to convey moderate indepen- mental arithmetic (PASAT3) were both independent
dent prediction of poor performance, which reached predictors of poor clinical performance. The excessive
or approached statistical significance in all six errors of commission indicate that the subjects dem-
Mantel-Haenszel tests (Table 7). onstrated impulsive behavior in a test situation that
The AKT-1 and ITE-1 test results are presented in required vigilance. Specifically, the impulsive subjects
Tables 8 and 9. These tests were independent of per- had trouble discriminating targets from background
sonality variables, but not of the cognitive tests, as noise. This test is analogous to the OR environment, in
reflected by the Mantel-Haenszel tests. which an impulsive resident would be expected to
have difficulty discriminating pertinent clinical abnor-
malities from artifactual or distracting data. The rapid
Discussion mental arithmetic ability could be described as the
Psychological factors are associated with clinical com- ability to “think on one’s feet.” Slow speed of mental
petence in residency training programs(3–5). In this processing and difficulty ignoring irrelevant informa-
study, we examined cognitive, personality, and aca- tion are undesirable characteristics in an anesthesiol-
demic knowledge test scores in anesthesiology resi- ogist that may be detected by this test.
dents and investigated their relationship to clinical In the personality domain, the independent predic-
competence over the course of training. Several meas- tors of poor performance were introversion and high
ures were independent predictors of poor clinical per- flexibility. Introverted individuals, as defined by the
formance, as defined by poor clinical competence rat- CPI-V1, are described as reticent, shy, reserved, and
ings or discontinuation from the residency program reluctant to initiate or take decisive social action (as
for poor clinical performance. opposed to outgoing, confident, talkative individuals
In the cognitive domain, excessive errors of commis- with social poise and presence). Highly flexible indi-
sion on a test of sustained and divided attention viduals (as defined by the CPI-Fx) are described as


Table 7. Independent Contribution of CPI-Fx to Predict Poor Clinical Outcome

CPI-Fx Յ50 25.0 20.0 0.0
CPI-Fx Ͼ50 71.4 40.0 21.4


CPI-Fx Յ50 5.0 66.7
CPI-Fx Ͼ50 33.3 57.1

(CPI-V1 Յ50) (CPI-V1 Ͼ50) P ϭ 0.01

CPI-Fx Յ50 13.6 27.3
CPI-Fx Ͼ50 27.3 88.9

(CPI-V2 Յ50) (CPI-V2 Ͼ50) P ϭ 0.06

CPI-Fx Յ50 28.6 15.4
CPI-Fx Ͼ50 47.4 41.7

(AKT-1 Յ50) (AKT-1 Ͼ50) P ϭ 0.04

CPI-Fx Յ50 27.8 0.0
CPI-Fx Ͼ50 52.2 25.5

(ITE-1 Յ15) (ITE-1 Ͼ15) P ϭ 0.04

CPI-Fx Յ50 33.3 5.9
CPI-Fx Ͼ50 50.0 35.7


those who like change and variety, are easily bored by implies that the relative ranking of the residents by the
routine, and may be impatient and erratic (as opposed faculty tended not to change over the course of the
to stubborn, rigid, steady-paced individuals who like residency and that the faculty perceived that nearly all
an organized life). The potential disadvantages of cer- residents improved at nearly equal rates over time.
tain of these personality characteristics in an anesthe- We also observed a statistically significant correlation
siologist are self-evident. In contrast to our expecta- among the scores in the five domains of clinical com-
tions, anxiety was not a predictor of poor clinical petence. This implies that the faculty usually made
performance. summary judgments of the residents’ performances
In the academic domain, only the tests administered and did not differentiate, to any significant degree,
within the first month of initiation of anesthesiology among performances in the different domains. Never-
residency (AKT-1 and ITE-1) were mildly predictive of theless, the clinical competence committee did dis-
poor clinical performance. They were independent of criminate among the five evaluation domains for cer-
the personality variables, but not of the cognitive vari- tain residents. Another limitation of the evaluation
ables. The predictive value of the academic tests may process is that the evaluators were aware of previous
reflect the individual’s level of preparedness for anes- assessments of the residents’ performances.
thesiology training. Specifically, those who had stud- This is the first study to examine the relationship
ied anesthesiology-related facts in advance and who between the cognitive abilities (sustained and divided
were generally good test-takers performed better in attention, rapid information processing) and clinical
the residency. It may also reflect a difference in the performance of anesthesiology residents. Previous
motivation of the individuals to learn anesthesiology- studies have investigated the relationship between
specific facts. personality attributes (CPI) and clinical performance
The pattern of the clinical competence ratings was of anesthesiology residents (4,5) and have identified
notable in that residents’ scores began at different independence, empathy, socialization, well-being, and
levels but tended to improve equally over time. This achievement via conformance as attributes that have


Table 8. Independent Contribution of AKT-1 to Predict Poor Clinical Outcome

AKT-1 Յ50 44.4 40 17.6
AKT-1 Ͼ50 100 0 0


AKT-1 Յ50 25.9 50
AKT-1 Ͼ50 0 66.7

(CPI-V1 Յ50) (CPI-V1 Ͼ50) P ϭ 0.08

AKT-1 Յ50 23.1 73.3
AKT-1 Ͼ50 15.4 0.0

(CPI-V2 Յ50) (CPI-V2 Ͼ50) P ϭ 0.06

AKT-1 Յ50 52.4 30.0
AKT-1 Ͼ50 0.0 18.18

Inflexible Flexible

AKT-1 Յ50 27.8 52.2
AKT-1 Ͼ50 0.0 25.0

(ITE-1 Յ15) (ITE-1 Ͼ15) P ϭ 0.21

AKT-1 Յ50 40.7 35.7
AKT-1 Ͼ50 100.0 6.7


statistically significant positive correlations with resi- may not relate to clinical practice after residency. The
dency performance. Although those authors reported study is further limited by the incomplete enrollment
a moderate predictive effect, none of the correlation of anesthesiology residents. The group that declined
coefficients were Ͼ0.30, which indicates that a mini- to participate in the study may have had demo-
mal amount of variance (r2 Յ 0.09) was attributable to graphic, cognitive, personality, or academic character-
each of these individual factors. In contrast, in the istics that differed from the study group—a selection
current study, we examined the independent associa- bias. The data from nonparticipating residents are not
tion of cognitive, personality, and academic measures available to make such comparisons. Furthermore,
with poor clinical performance. this was a single-center study. It is therefore unknown
There are several limitations of our methodology. whether the results of this study would be similar at
Although based on a standard system and consistent other institutions.
over time, the clinical competence ratings are subjec- This was a longitudinal study of predictors of anes-
tive. Because the performance of individual residents thesiology residency performance at one institution.
was ranked relative to the performance of the entire Many statistical tests were performed. Despite this and
group and was not based on an absolute standard of other limitations, the data suggest that some cognitive,
minimal competency, the high rate of poor perfor- personality, and academic measures are associated with
mance is a consequence of the operational definition clinical competence in a sample of anesthesiology resi-
used for this study. The clinical competence ratings dents. The ability of these measures to predict individual
cannot be validated against a more objective measure resident’s clinical performance should be confirmed in a
because none exists. The clinical competence ratings prospective, multicenter study.
are also limited in that they are a global evaluation of
clinical performance and do not allow discrimination We gratefully acknowledge the assistance of Dr. Michelle Marks
between OR and non-OR performance. Furthermore, with the original design and implementation of this study.
clinical competence ratings made during residency


Table 9. Independent Contribution of ITE-1 to Predict Poor Clinical Outcome

ITE-1 Յ15 57.1 25.0 23.1
ITE-1 Ͼ15 50.0 25.0 0


ITE-1 Յ15 29.4 50.0
ITE-1 Ͼ15 9.5 60.0

(CPI-V1 Յ50) (CPI-V1 Ͼ50) P ϭ 0.07

ITE-1 Յ15 30.0 63.6
ITE-1 Ͼ15 9.1 44.4

(CPI-V2 Յ50) (CPI-V2 Ͼ50) P ϭ 0.04

ITE-1 Յ15 54.6 35.0
ITE-1 Ͼ15 28.6 11.8

Inflexible Flexible

ITE-1 Յ15 33.3 50.0
ITE-1 Ͼ15 5.9 35.7

(AKT-1 Յ50) (AKT-1 Ͼ50) P ϭ 0.38

ITE-1 Յ15 40.7 100.0
ITE-1 Ͼ15 35.7 6.7


6. Loeb RG. A measure of intraoperative attention to monitor
References displays. Anesth Analg 1993;76:337– 41.
1. Price PB, Taylor CW, Richards JM, Jacobsen TL. Measurement 7. Spielberger CD. State-Trait Anxiety Inventory. Palo Alto, CA:
of physician performance. J Med Educ 1964;39:203–11. Consulting Psychologists Press, 1970.
2. Wingard JR, Williamson JW. Grades as predictors of physician’s 8. Gronwall D. Paced auditory serial addition task: a measure of
career performance: an evaluative literature review. J Med Educ recovery from concussion. Percept Motor Skills 1977;44:367–73.
1973;48:312–22. 9. American Board of Anesthesiology. Defining competence in
3. Keck JW, Arnold L, Willoughby L, Calkins V. Efficacy of anesthesiology. Hartford, CT: American Board of Anesthesiol-
cognitive/noncognitive measures in predicting resident- ogy, 1993.
physician performance. J Med Educ 1979;54:759 – 65.
10. Bryk AS, Raudenbush SW. Hierarchical linear models. New-
4. McDonald JS, Lingam RP, Gupta B, et al. Psychologic testing as
bury Park, CA: Sage Publications, 1992.
an aid to selection of residents in anesthesiology. Anesth Analg
1994;78:542–7.
5. Gough HG, Bradley P, McDonald JS. Performance of residents
in anesthesiology as related to measures of personality and
interests. Psychol Rep 1991;3:979 –94.

1092.full

Recommended

Recommended

More Related Content

Viewers also liked

Viewers also liked (12)

Similar to 1092.full

Similar to 1092.full (20)

Recently uploaded

Recently uploaded (20)

1092.full