Responses Rate and Demographic Characteristics
One hundred internal medicine resident physicians participated in the survey, of which 74 were female. Seventy-five were either in Year 1 (R1) or Year 2 (R2) of their internal medicine residency programme, and 45 spent over 80% of their time in direct patient care. All participants worked for inpatient services. Furthermore, 99% reported no experience with medico-legal proceedings (Table 1).
Table 1
Table 1: Socio-demographics of Medical Interns (N = 100)
Baseline characteristics
|
Frequency (%)
|
Sex:
Male
Female
|
26(26.0)
74(74.0)
|
Training level-residency programme
R1
R2
R3
R4
|
36(36.0)
39(39.0)
15(15.0)
10(10.0)
|
Time spent in direct patient care
41–60%
61–80%
> 80%
|
18(18.0)
37(37.0)
45(45.0)
|
Types of clinical practice
Inpatients (All or Mostly)
All Inpatients
Mostly Inpatients
Mostly Outpatients
All Outpatients
|
100(100.0)
0
0
0
0
|
Experience with medico-legal proceedings
Yes
No
|
1(1.0)
99(99.0)
|
R1: first year in the residency training medical programme
R2: the second year in the residency training medical programme
R3: the third year in the residency training medical programme
Reporting Actual Errors
Among the 100 participants, who primarily worked in inpatient care, 38% acknowledged self-reporting a minor error that had prolonged treatment or caused discomfort. In contrast, 13% acknowledged reporting a significant error that had caused disability or death (Table 2). The study found that 12% and 38% of residents had never disclosed a major or minor error, respectively. Considering all acknowledged errors together (reported and non-reported), 35.6% of the participants acknowledged having made at least one minor or significant error: 72% of the residents in Year 1 (R1), 94% in Year 2 (R2), 3% in Year 3 (R3), and 100% in Year 4 (R4). Approximately 75% acknowledged a minor error, and 25% acknowledged a significant error.
Table 2
Table 2: Actual Experiences with Reporting or Non-Reporting of Errors (N = 100)
Type of Errors
|
Residents Answering Yes (%)
|
Minor Errors
Have you ever made a mistake that prolonged treatment or caused discomfort and reported the mistake to your institution?
R1
R2
R3
R4
Have you ever made a mistake that prolonged treatment or caused discomfort and not reported the mistake to your institution?
R1
R2
R3
R4
|
38(38.0)
12(31.6)
16(42.1)
9(23.7)
1(2.6)
38(38.0)
12(31.6)
13(34.2)
7(18.4)
6(15.8)
|
Major Errors
Have you ever made a mistake that caused disability or death and reported the mistake to your institution?
R1
R2
R3
R4
Have you ever made a mistake that caused disability or death and not reported the mistake to your institution??
R1
R2
R3
R4
|
13(13.0)
2(15.4)
5(38.5)
6(46.2)
0(0.0)
12(12.0)
0(0.0)
3(25.0)
5(41.7)
4(33.3)
|
Reporting Hypothetical Errors
The likelihood of reporting a hypothetical error depended on the outcome of the error (Table 2). In total, 63% of the participants indicated that they would report a hypothetical error if it resulted in no harm to a patient, 84% if it resulted in minor harm, and 99% if it resulted in major harm with full recovery (Table 3).
Table 3
Table 3: Responses to Hypothetical Errors Vignette (N = 100)
How likely is it that you would report an error?
|
Outcome 1
No Harm (%)
|
Outcome 2
Minor Harm (%)
|
Outcome 3
Major Harm (%)
|
Very likely
|
20(20.0)
|
35(35.0)
|
56(56.0)
|
Likely
|
43(43.0)
|
49(49.0)
|
43(43.0)
|
Not sure
|
22(22.0)
|
16(16.0)
|
1(1.0)
|
Unlikely
|
0(0.0)
|
0(0.0)
|
0(0.0)
|
Very unlikely
|
15(15.0)
|
0(0.0)
|
0(0.0)
|
Knowledge and Attitudes Regarding Error Reporting
Most residents believed that reporting errors improves the quality of care for future patients (97%). However, only 75% knew how to report errors to their institution, and only 53% knew what errors should be reported. Some respondents believed that it was difficult to be sure of the true causes of adverse events (38%), but almost all of them believed that reporting errors was worth their time. According to 44% of the respondents, they were more likely to report errors in anticipation of feedback. Notably, fourth-year resident physicians were less knowledgeable than other residents regarding the types of medical errors that should be reported in their institutions (Table 4).
Table 4
Table 4: Knowledge Regarding Error Reporting (N = 100)
Knowledge Variables
|
Residents Answering Yes (%)
|
Total
|
R1
|
R2
|
R3
|
R4
|
P value
|
1. Reporting medical errors to one’s institution improves the quality of care for future patients
|
97
|
94.4
|
100.0
|
100.0
|
90.0
|
0.245
|
2. I know how to report medical errors in my institution
|
75.0
|
72.2
|
76.9
|
73.3
|
80.0
|
0.944
|
3. I know what kinds of medical errors should be reported in my institution
|
53.0
|
52.8
|
61.5
|
60.0
|
10.0
|
0.031
|
4. It is hard to be certain about the true causes of adverse events in the practice of clinical medicine
|
38.0
|
41.7
|
28.2
|
40.0
|
60.0
|
0.275
|
5. I would more likely report errors in my institution if I knew I would receive feedback afterwards
|
44.0
|
30.6
|
48.7
|
60.0
|
50.0
|
0.195
|
6. Disclosing errors to my institution is worth my time because my actions can change the system of care
|
99.0
|
100.0
|
100.0
|
93.3
|
100.0
|
0.126
|
Most resident physicians felt obligated to tell their patients the facts necessary for them to understand what happened. Most of the respondents (80%) felt an obligation to make it clear that a mistake had occurred. Most physicians acknowledged that they would disclose errors even if it could harm their reputation or increase incidences of malpractice. However, third-year residents were less likely to report errors as mistakes if it came at a high personal cost. Many participants reported telling their patients about errors that they had made, although only 80% of them believed that disclosing errors would help alleviate their guilt; some believed that disclosing errors had nothing to do with building trust in the treating doctor. Most participants (92%) reported that they would disclose an error based on their judgement of whether the information would help or harm patients. The majority said that when they make a medical mistake, they would be their own worst critic. Simultaneously, third-year residents were less likely to agree with this statement; 80% believed that if there were no malpractice risks, it would be much easier to disclose the errors to the patients (Table 5).
Table 5
Table 5: Attitude Regarding Error Reporting (N = 100)
Attitude Variables
|
Residents Answering Yes (%)
|
Total
|
R1
|
R2
|
R3
|
R4
|
P value
|
1. Whenever I commit a mistake, I feel obliged to tell my patient the necessary facts to enable them to understand what exactly happened.
|
97.0
|
97.2
|
97.4
|
93.3
|
100.0
|
0.793
|
2. When a mistake occurs, I feel obliged to clarify that whatever happened was a medical mistake (preventable adverse event).
|
80.0
|
77.8
|
94.9
|
53.3
|
70.0
|
0.005
|
3. Disclosing medical errors is the right thing to do, even if it comes at a significant personal cost (e.g., harms my reputation or increases the risk of malpractice).
|
93.0
|
100.0
|
94.9
|
73.3
|
90.0
|
0.007
|
4. It is important for me to inform my patients about the errors committed by me because that is how I would want to be treated if I were a patient.
|
98.0
|
97.2
|
97.4
|
100.0
|
100.0
|
0.877
|
5. If I made a medical mistake, disclosing the mistake to my patient would help alleviate my feelings of guilt.
|
80.0
|
86.1
|
82.1
|
73.3
|
60.0
|
0.277
|
6. Informing my patient about a medical error I have made in their care strengthens my patient’s trust in me as a physician.
|
36.0
|
36.1
|
28.2
|
53.3
|
40.0
|
0.383
|
7. My decision to disclose an error to my patient depends on whether I think the information will help or harm them.
|
92.0
|
91.7
|
92.3
|
93.3
|
90.0
|
0.992
|
8. When I make a medical mistake, I am my own worst critic.
|
83.0
|
83.3
|
87.2
|
93.3
|
50.0
|
0.025
|
9. If there are no malpractice risks related to disclosing medical mistakes to patients, it would be much easier to talk to my patients about mistakes when they occur.
|
80.0
|
86.1
|
84.6
|
66.7
|
60.0
|
0.137
|
Possible Consequences of Error Reporting
As shown in Table 6, Most residents were concerned about the negative patient or family reaction (98%), their professional discipline (97%), and malpractice litigation (95%) The more experienced residents (R4) were more concerned than others (Table 6).
Table 6
Table 6: Possible Consequences of Error Reporting
Variables
|
Residents Answering Yes (%)
|
Total
|
R1
|
R2
|
R3
|
R4
|
P value
|
In general, when thinking about disclosing medical mistakes, I am concerned about the following possible consequences.
|
1. Negative patient/family reaction
|
98.0
|
100.0
|
97.4
|
93.3
|
100.0
|
0.446
|
2. Malpractice litigation
|
95.0
|
94.4
|
94.9
|
93.3
|
100.0
|
0.888
|
3. Professional discipline
|
97.0
|
97.2
|
94.9
|
100.0
|
100.0
|
0.709
|
4. Loss of reputation amongst colleagues
|
73.0
|
58.3
|
82.1
|
80.0
|
80.0
|
0.104
|
5. Blame from colleagues
|
62.0
|
50.0
|
69.2
|
60.0
|
80.0
|
0.215
|
6. Negative publicity
|
77.0
|
75.0
|
79.5
|
73.3
|
80.0
|
0.944
|
Demographic Factors and Actual Errors
Bivariate analysis was used to assess the dependent variable(reporting and not reporting of errors , and the independents variables; sex, training level, direct patient care, and type of practice)(Table 7) .
Table 7
Table 7: Reporting Actual Errors According to Demographic Variable (bivariate analysis)
|
Minor Errors Reported
|
Major Errors Reported
|
No
|
Yes
|
No
|
Yes
|
Sex
|
Male
|
15 (57.7%)
|
11 (42.3%)
|
24 (92.3%)
|
2 (7.7%)
|
Female
|
47 (63.5%)
|
27 (36.5%)
|
63 (85.1%)
|
11 (14.9%)
|
Training level
|
R1
|
24 (66.7%)
|
12 (33.3%)
|
34 (94.4%)
|
2 (5.6%)
|
R2
|
23 (59.0%)
|
16 (41.0%)
|
34 (87.2%)
|
5 (12.8%)
|
R3
|
6 (40.0%)
|
9 (60.0%)
|
9 (60.0%)
|
6 (40.0%)
|
R4
|
9 (90.0%)
|
1 (10.0%)
|
10 (100.0%)
|
0 (0.0%)
|
Time spent in direct patient care
|
41-60%
|
13 (72.2%)
|
5 (27.8%)
|
17 (94.4%)
|
1 (5.6%)
|
61-80%
|
24 (64.9%)
|
13 (35.1%)
|
31 (83.8%)
|
6 (16.2%)
|
More than 80%
|
25 (55.6%)
|
20 (44.4%)
|
39 (86.7%)
|
6 (13.3%)
|
Type of practice
|
All inpatient
|
8 (47.1%)
|
9 (52.9%)
|
16 (94.1%)
|
1 (5.9%)
|
Mostly inpatient
|
54 (65.1%)
|
29 (34.9%)
|
71 (85.5%)
|
12 (14.5%)
|
Direct logistic regression was performed to determine the impact of several demographic factors on reporting actual errors. The model contained four independent variables (sex, training level, direct patient care, and type of practice)and dependent variable (reporting actual minor errors). Moreover, the complete model containing all the predictors was statistically significant (p < 0.05), indicating that it could distinguish between who is reporting and not reporting actual minor errors; the model correctly classified 72.2% of the cases. As depicted in the table 8, only two independent variables made unique and statistically significant contributions to the model (training level and type of practice). The strongest predictor was training level, with an odds ratio of 1.879, which indicated that the internal medicine residents in R3 were more likely to make actual minor errors than those in R2 or R1. In addition, the type of practice had a significant impact on disclosing minor medical errors, with an odds ratio (OR) of 0.244. When the type of practice was for all inpatients, the residents who spent all their time in inpatient care services were more likely to report actual minor errors than others. There was no significant effect of sex or direct patient care (p > 0.5) (Table 8).
Table 8
Table 8: Association Between Demographic Factors and Reporting of Actual Minor Errors Using Binary Logistic Regression
|
B
|
S.E.
|
Wald
|
df
|
Sig.
|
Exp(B)
|
Constant
|
.851
|
1.034
|
.676
|
1
|
.411
|
2.341
|
Sex
|
-.377
|
.334
|
1.278
|
1
|
.258
|
.686
|
Training level
|
.631
|
.194
|
10.530
|
1
|
.001**
|
1.879
|
Time spent in direct patient care
|
.216
|
.176
|
1.514
|
1
|
.218
|
1.241
|
Type of practice
|
-1.409
|
.412
|
11.718
|
1
|
.001**
|
.244
|
Dependent variable: Reporting actual minor errors. **p < 0.01, *p < 0.05, OR: odds ratio
The forest plot in Figure 1 depicts the binary logistic regressions for reporting actual errors as the dependent variable; it was used to graphically represent the ORs and their corresponding confidence intervals (CI; 95%). The diamonds reflect the ORs, while the whiskers represent the confidence limits (95% CIs). The plot illustrates the likelihood of reporting actual errors. The CI indicates the level of uncertainty around the measure of effect, which is expressed as an OR.
Figure (1)
Demographic Factors and Reporting of Actual Major Errors
As shown in Table 9, only training level statistically significantly impacts reporting actual major errors, with an OR of 3.565. This indicates that internal medicine residents at higher training levels are more likely to report major errors. No other independent variables (sex, direct patient care, and type of practice) had a significant effect on error reporting (p > 0.5).
Table 9
Table 9: Association Between the Demographic Factors and Reporting of Actual Major Errors Using Binary Logistic Regression
|
B
|
S.E.
|
Wald
|
df
|
Sig.
|
Exp(B)
|
Constant
|
-6.411
|
1.954
|
10.767
|
1
|
.001
|
.002
|
Sex
|
.081
|
.509
|
.025
|
1
|
.874
|
1.084
|
Training level
|
1.271
|
.279
|
20.820
|
1
|
<.001**
|
3.565
|
Direct patient care
|
.151
|
.261
|
.335
|
1
|
.563
|
1.163
|
Type of practice
|
.681
|
.664
|
1.054
|
1
|
.305
|
1.977
|
Dependent variable: Reporting actual minor errors. **p < 0.01, *p < 0.05, OR: odds ratio
The forest plot in Figure 2 depicts significant associations achieved for training level as one significant predictor for reporting major errors. Simultaneously, the 95% CI crossed 1 for other variables, implying no significant impact of other variables (gender, direct patient care and type of practice).
Figure 2
Variables Associated with reporting actual errors
The classification table (Table 10) shows the results of the direct logistic regression performed to determine the impact of demographic, knowledge, attitudes, concerned consequences and willingness to disclose errors (using a hypothetical error vignette) variables on reporting minor errors. It shows that the model could correctly predict whether a physician would disclose a minor error 67% of the time. This is a significant improvement over random guessing, which would have resulted in a 50% accuracy rate.
The variables that were most predictive of disclosure were the physician's belief that disclosing errors is the right thing to do, even if it comes at a personal cost. However, the association was negative. These findings suggest that physicians who believe that disclosing errors is important are more likely to not disclose errors to their patients.
Table 10
Table 10: Classification table reporting the overall percentage of correctly classified reporting minor errors
|
Observed
|
Predicted
|
|
Minor reported
|
Percentage correct
|
|
No
|
Yes
|
Step 1
|
Minor reported
|
no
|
61
|
1
|
98.4
|
yes
|
32
|
6
|
15.8
|
Overall percentage
|
|
|
67.0
|
a. The cut-off value is .500
|
Direct logistic regression was performed to determine the impact of demographic, knowledge, attitudes, concerned consequences and willingness to disclose errors (using a hypothetical error vignette) variables on reporting major errors. The classification table (Table 11) shows that 93% of physicians would disclose a major error vs. 53.8%.
The variables in the equation that were found to be factors associated with a physician's decision to report a major medical error are the physician’s belief that disclosing errors is the right thing to do, even if it comes at a personal cost; the physician's belief that informing their patients about errors strengthens the patient–physician relationship; severity of the error; physician's training and experience; physician’s fear of retaliation from colleagues or patient; and physician's knowledge of the institution's policies and procedures for handling medical errors. However, the only significant relation was the physician's belief that disclosing errors is the right thing to do, even if it comes at a personal cost, although the association was negative.
Table 11
Table 11: Classification table reporting the overall percentage of correctly classified reporting major errors
|
|
Major reported
|
Percentage correct
|
|
No
|
Yes
|
Step 1
|
Major reported
|
No
|
86
|
1
|
98.9
|
Yes
|
7
|
6
|
46.2
|
Overall percentage
|
|
|
92.0
|
Step 2
|
Major reported
|
No
|
86
|
1
|
98.9
|
Yes
|
6
|
7
|
53.8
|
Overall percentage
|
|
|
93.0
|
a. The cut-off value is .500
|
Figure (2)