The Joint Commission Journal on Quality and Patient Safety 2020; 46:282–290

Health Care Provider Factors Associated with

Patient-Reported Adverse Events and Harm
Traber D. Giardina, PhD; Kathryn E. Royse, PhD; Arushi Khanna, MPH; Helen Haskell, MA; Julia Hallisy, DDS;
Frederick Southwick, MD; Hardeep Singh, MD

Background: Patients can provide valuable information missing from traditional sources of safety data, thus adding new
insights about factors that lead to preventable harm. In this study, researchers determined associations between patient-
reported contributory factors and patient-reported harms experienced after an adverse event (AE).
Methods: A secondary analysis was conducted of a national sample of patient-reported AEs (surgical, medication, di-
agnostic, and hospital-acquired infection) gathered through an online questionnaire between January 2010 and February
2016. Generalized logit multivariable regression was used to assess the association between patient-reported contributory
factors and patient-reported harms (grouped as nonphysical harm only, physical harm only, physical harm and emotional or
financial harm, and all three harms) and adjusted for patient and AE characteristics.
Results: One third of patients (32.6%) reported experiencing all three harms, 27.3% reported physical harms and one
additional harm, 25.5% reported physical harms only, and 14.7% reported nonphysical harms only. Patients reporting
all three harms were 2.5 times more likely to have filed a report with a responsible authority (95% confidence interval
[CI] = 1.23–5.01) and 3.3 times more likely to have also experienced a surgical complication (95% CI = 1.42–7.51). Odds
of reporting problems related to communication between clinician and patients/families or clinician-related behavioral issues
was 13% higher in those experiencing all three harm types (95% CI = 1.07–1.19).
Conclusion: Patients’ experiences are important to identify safety issues and reduce harm and should be included in patient
safety measurement and improvement activities. These findings underscore the need for policy and practice changes to
identify, address, and support harmed patients.

D espite heightened efforts to research and prevent ad-

verse events (AEs),1 they are common in medicine,2–4
with estimates suggesting that two to three AEs occur per
mation directly from patients. Even though there are exist-
ing models for collecting the data from patients,23 , 24 there is
little information on when and how these data, if collected,
100 primary care consultations.5 Although these AEs typ- are being used by health care systems.
ically include only physical harm, patients may experience Improving patient experience is an essential part of
emotional and financial harm as a result of an AE,6 , 7 which patient-centered care. As such, leveraging information from
may have long-term impact for patients and families.6 , 8 , 9 patients can fill in the gaps such as the circumstances lead-
Patients are uniquely positioned to provide a 360 degree ing up to an AE and the factors that contributed to the AE.
view of the AE and identify factors that contribute to the A systematic review of patients’ perceptions of AEs in the
AE and the harm experienced. outpatient setting found that patients were able to identify
Methods to identify AEs are still evolving, and many AEs technical medical aspects (for example, errors in diagnosis,
are either poorly documented in the medical record or not treatment, intervention, and medication) and quality issues
captured at all.10–15 The integration of patients’ perspectives that contribute to error (for example, patient-physician re-
may lead to better understanding of AEs; patients can iden- lationship, coordination, communication).20 This AE in-
tify preventable AEs and the factors that contribute to those formation may be captured via patient complaint data.
events,16 , 17 none of which are necessarily recorded in the In fact, a recent study by Gillespie and Reader tested a
medical record or existing reporting systems.10 , 11 , 15 , 18–22 tool to analyze patient complaints and identify “hot spots”
There is a growing interest in improving patient engage- of patient harm and near misses, highlighting the value
ment in safety, but there is no current nationally standard- of patient-identified issues and information contributing
ized method or requirement to collect this valuable infor- to those issues.23 Increasing our knowledge about factors
contributing to AEs and resultant harm from the patient
perspective could lead to new opportunities for feedback,
1553-7250/$-see front matter
© 2020 The Authors. Published by Else-
learning, and prevention, as well as opportunity for patient
vier Inc. on behalf of The Joint Commission. healing.
This is an open access article under the CC BY-NC-ND license.
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
Adverse events usually result from a multitude of con-
https://doi.org/10.1016/j.jcjq.2020.02.004 tributory factors that can lead to patient harm.25 Currently,
Volume 46, No. 5, May 2020 283

information about AEs relies on low-yield clinician and staff creditation, disciplinary, or other regulatory authority [such
incident reporting, morbidity and mortality conferences, as state medical boards or The Joint Commission]), type of
autopsy, root cause analyses, and electronic triggers (a spe- adverse event (surgical events, hospital-acquired infection,
cific set of clinical clues used to flag records of patients at diagnostic error), and type of harm. Respondents were also
higher risk of harm so that they can be reviewed for possi- asked to indicate which of 13 contributory factors were in-
ble safety events26 ). Error reports from patients, often left volved (for example, “Patient was not given the informa-
out of data collection, may provide new insights into safety- tion needed to make an informed decision,” “Health care
related problems and types of contributory factors that are personnel did not listen to patient or family,” “Health care
most harmful and thus potential targets of interventions. To personnel did not seem concerned about the patient” [see
advance knowledge in this area, we analyzed a large sample Appendix 1 for full list]) and how they affected outcomes.
of error reports gathered by patient advocates to determine
the association between patient-reported contributory fac- Contributory Factor Selection
tors and patient-reported harm.
We were interested in determining how perceived contrib-
utory factors related to health care personnel (HCP) com-
METHODS munication, HCP behavior, HCP medical knowledge, and
HCP monitoring and timely response to patients were as-
We conducted a secondary analysis of self-reported harms
sociated with the four harm outcome groups. Therefore, we
captured on a national online questionnaire. This voluntary
selected 13 of the 19 items from the scale in the EPC ques-
questionnaire was developed by two patient advocates and
tionnaire that listed “Contributing Factors to the Adverse
posted on the Empowered Patient Coalition (EPC) website
Event,” and we excluded all questions related to sanitation
from January 2010 to present.27 , 28 The survey was created
and to the patient’s discharge from the health care facility.
based on the categorization of adverse events by the Office
Each item on the scale ranged from “no response or did not
of Inspector General.29 Additional questions were added to
occur” or “NA, occurred but not a serious problem,” “se-
assess harm and were based on the advocates’ personal expe-
rious problem in patient’s care,” to “major factor affecting
riences and those of fellow patients and families. This study
patient outcome,” and we assessed 0–3 points for each re-
was approved by the Institutional Review Boards of the Bay-
sponse. Patients were instructed to “Choose as many as ap-
lor College of Medicine and the Michael E. DeBakey VA
ply” of the 13 items from the scale of contributing factors
[US Department of Veterans Affairs] Medical Center.
(see Appendix 1 for the 13 contributory factors). We mea-
Eligibility Criteria sured scale internal consistency and reliability using Cron-
bach’s alpha.30
We selected patients who self-reported any harm as an out-
come from an adverse event between January 2010 and
January 2016 and who also answered at least 1 of the 13 Data Analysis
potential questions about contributory factors related to Exploratory Factor Analysis of Contributory Fac-
their self-reported harm (see Appendix 1, available in on- tors. We conducted an exploratory factor analysis (EFA)
line article). Harms were classified as physical (brain dam- on the contributory factors to identify the underlying rela-
age, chronic pain, need for additional surgery, readmission tionships and their association with the reported four harm
to the hospital within 30 days, loss of bowel or bladder groupings.31 , 32 We used EFA because it is appropriate for
control, disfigurement, short-term loss of function [< 3 the interval measurement structure for the 13 questions.
months], long-term loss of function [> 3 months], and Item-item correlations for the 13 questions were assessed
permanent loss of function or disability), emotional (post- to investigate multicollinearity, which occurs when two or
traumatic stress or emotional trauma), and financial loss. more items in a factor analysis are redundant, and underly-
Based on questionnaire structure, participants were able to ing concepts (for example, communication). Resulting fac-
select multiple harms; therefore, participants were separated tor groupings were rotated using varimax rotation orthog-
into four outcome harm groups for analysis: (1) nonphys- onally, which allowed assessment of each contributory fac-
ical harm only (emotional and/or financial), (2) physical tor independent of each other. Because respondents were
harm only, (3) physical harm and emotional or financial instructed to skip questions that did not apply to them,
harm, and (4) all three harms (emotional, financial, and we included all adverse event respondents in the analysis
physical harm) to ensure that the categories were mutually if they answered at least 1 of the 13 questions. Any skipped
exclusive and exhaustive. questions among the remaining 12 items were weighted
The questionnaire collected patients’ demographic data as 0 = “No response or Factor did not occur or NA.” We
(age, location, gender, year of event, size of community), discarded items with lower factors loadings (< 0.40), and
whether the patient officially filed a report of the AE to a re- remaining contributory factors were summed together ac-
sponsible authority (for example, health care delivery orga- cording to resulting high factor groupings (≥ 0.40).31 We
nization where the AE occurred, or any type of licensing, ac- dropped all eigenvalues under 1.00, which is the general-
284 Traber D. Giardina, PhD, et al. Health Care Provider Factors

ized cutoff,31 , 32 and scree plots were used to visually con- stepwise logistic regression model with “nonphysical injury”
firm the number of contributory factors to retain in the as the reference. We used SAS 9.4 (SAS Institute Inc., Cary,
EFA. For the purposes of adjustment, we then compared North Carolina) to conduct all statistical analysis.
the resultant contributory factor groupings, along with pa-
tient factors (gender, community size, and participant age at
event), and adverse event–related factors (time since adverse RESULTS
event, diagnostic error, adverse medication event, hospital- Of 449 patient self-reports, 351 (78.2%) mentioned phys-
acquired infection, and surgical complication, and whether ical, emotional, or financial harms and responded to at
they filed a report of the AE to a responsible authority or least one question about contributory factors for the ad-
not) for significant differences (p < 0.05) across the four verse event. One third (32.6%) indicated experiencing all
harm groups. Categorical variables were first assessed using three harms, 27.3% reported physical harms and one ad-
the chi-square test and continuous variables using the Stu- ditional harm, 25.5% reported physical harms only, and
dent’s t-test. Variables with p values < 0.10 in univariate 14.7% reported only nonphysical harms. Most respondents
analysis were entered into a generalized logit multivariable (69.5%) were female and experienced the adverse event in-

Table 1. Characteristics of Patients and Patient-Reported Adverse Events Across Categories of Self-Reported
Harms (N = 334)∗
Patient Characteristics

Nonphysical Physical Harm Physical Harm & Physical, Emotional P Value

Harm n = 49 Only n = 85 Emotional or & Financial Harm
(14.7) (25.5) Financial Harm n = 109 (32.6)
n = 91 (27.3)
Sex† 0.40
Female 34 (70.8) 50 (61.7) 65 (72.2) 83 (77.6)
Male 14 (29.2) 31 (38.3) 25 (27.8) 24 (22.4)
Age AE occurred Mean (SD)† 51.3 (14.8) 53.4 (13.5) 49.8 (13.5) 48.6 (10.4) 0.01‡
Years between event and report§ Mean 2.8 (4.2) 6.5 (9.4) 6.8 (7.9) 6.4 (6.7) 0.02‡
(SD)†
Event location 0.78
Not a hospital 14 (29.2) 22 (26.8) 24 (27.0) 35 (32.7)
Hospital 34 (70.8) 60 (73.2) 65 (73.0) 72 (67.3)
Community size at event location† 0.26
≤ 500,000 population 22 (47.8) 39 (49.4) 50 (56.8) 66 (61.7)
> 500,000 population 24 (52.2) 40 (50.6) 38 (43.2) 41 (38.3)
Adverse Event Characteristics
Presence of diagnostic error‡ 0.01†
No 32 (65.3) 44 (51.8) 45 (49.5) 34 (31.2)
Yes 17 (34.7) 41 (48.2) 46 (50.5) 75 (68.8)
Presence of adverse medication event .022557‡
No 20 (40.8) 55 (64.7) 46 (50.5) 50 (45.9)
Yes 29 (59.2) 30 (35.3) 45 (49.5) 59 (54.1)
Presence of hospital-acquired infection 0.01‡
No 37 (75.5) 40 (47.1) 51 (56.0) 55 (50.5)
Yes 12 (24.5) 45 (52.9) 40 (44.0) 54 (49.5)
Presence of surgical complication < 0.0001‡
No 32 (65.3) 28 (32.9) 30 (33.0) 23 (21.1)
Yes 17 (34.7) 57 (67.1) 61 (67.0) 86 (78.9)
Patient reporting to appropriate 0.01‡
authority
No 26 (53.1) 48 (56.5) 41 (45.1) 34 (31.5)
Yes 23 (46.9) 37 (43.5) 50 (54.9) 74 (68.5)
∗ For categorical variables, the number of participants with the characteristic, and the percentage of persons in the outcome harm group
with the characteristic, are presented.
† Missing data.
‡ Significant at 0.05.
§ The category of diagnostic error included misdiagnosis, laboratory or pathology error, delay in diagnosis or treatment, proper tests

not ordered, and test results being lost or misplaced.

AE, adverse event; SD, standard deviation.
Volume 46, No. 5, May 2020 285

Table 2. Unstandardized Factor Loadings and Item Descriptions of Contributory Factors (Questions 1–13)
(N = 334)∗
Factor 1 Factor 2 Factor 3 Mean SD
Contributory Factors
Communication and Health Care Personnel Behavior
Health care personnel did not seem concerned about the patient. 0.78 0.15 0.14 1.62 1.32
Health care personnel did not listen to patient or family. 0.72 0.10 0.17 1.82 1.33
Health care personnel did not seem familiar with the patient’s case. 0.60 0.25 0.19 1.16 1.31
Health care personnel did not communicate important information to patient. 0.56 0.11 0.20 1.67 1.36
Health Care Personnel Fatigue
Health care personnel seemed overtired or fatigued. 0.09 0.83 0.19 0.57 1.03
Health care personnel seemed overworked, rushed, or behind schedule. 0.27 0.68 0.15 0.95 1.23
Timeliness of Health Care Personnel Response
Nurse did not respond quickly to the call button. 0.19 0.18 0.78 0.57 1.00
Doctor was slow to arrive. 0.39 0.22 0.52 0.72 1.15
∗ Items appear in order of importance. Factor loadings are unstandardized. Factor loadings in boldface type represent the factor onto
which the item loaded the strongest (Factor 1 = Communication and Health Care Personnel [HCP] Behavior; Factor 2 = HCP Fatigue;
Factor 3 = Timeliness of HCP Response). Rotated factor with loadings < 0.4 were considered unimportant and removed, while factor
loadings > 0.50 were used as a first step for grouping. Grouped factors were listed in order of how much variation they explain.

hospital (69.2%), although gender and setting did not dif- HCP did not seem concerned about the patient, (2) HCP
fer among harm groups (p > 0.05). Age at event (p = 0.01), did not listen to the patient or family, (3) HCP did not
years since event (p = 0.02), presence of a diagnostic error seem familiar with the patient’s case, and (4) HCP did not
(p = 0.01), presence of a medication error (p = 0.01), pres- communicate important information to the patient. Two of
ence of a hospital-acquired infection (p = 0.01), presence of the 13 items about patients’ perceptions of the providers’
a surgical complication (p < 0.0001), and whether the pa- physical/mental state loaded onto Factor 2 (HCP Fatigue)
tient filed a report with a responsible authority (p = 0.01) and included (1) HCP seemed overtired or fatigued and
were significantly different among the four harm groups (2) HCP seemed overworked, rushed, or behind schedule.
(Table 1). Factor 3 focused on timeliness of HCP response to patients:
nurse did not respond quickly to the call button, and doctor
Contributory Factors and Harm was slow to arrive. Scale reliability for the items in all three
None of the 13 contributory factor items had correlations > factors was relatively high (α = 0.87). All three factors were
0.90 to suggest multicollinearity; however, there were sev- used in the subsequent univariate logistic regression with
eral correlations > 0.60 which suggested that an EFA was harm as the outcome.
appropriate. We identified three interpretable groupings of In univariate analysis, 9 of the 10 adverse event char-
the original 13 contributory factors (Factor 1: Communica- acteristics and harm outcomes met the inclusion thresh-
tion and HCP Behavior; Factor 2: HCP Fatigue; and Factor old (p = 0.10) for assessment in the multivariable step-
3: Timeliness of HCP Response) (Table 2). Examination of wise model (Table 3). After adjustment, only presence of
the scree plot of eigenvalues indicated that factors 3 and 4 hospital-acquired infection, presence of surgical complica-
appeared to be identical (see Appendix 2); factor 4 (con- tion, patient filed a report of his or her AE to a responsible
tributory factor items 1 and 3) was dropped because it ex- authority, and communication and HCP behavior (Factor
plained very little of the variance. Only Factor 1 met the 1) were significantly (p < 0.05) related to harm outcomes.
Kaiser criterion of an eigenvalue above 1.00 suggesting that The presence of a diagnostic error was close to statistical sig-
it was the only one that explained an above average amount nificance in multivariable analysis (p = 0.06), most likely
of the variance from the 13 variables. The eigenvalue for because it was highly documented by those with physical
Factor 2 was 0.85 and cumulatively explained 91.6% of harm (data not shown). Compared to patients who indi-
the variance, so it was retained for analysis. Factor 3 re- cated experiencing only physical harm, patients who expe-
sulted in a modest improvement of 8.41% of explained vari- rienced all three harms were 2.5 times more likely to have
ance. The resulting extracted factor solution average com- officially filed a report of their AE to a responsible authority
munality estimates, which reflect each item’s variance ex- (adjusted odds ratio [OR] = 2.49, 95% confidence inter-
plained by each model factor, were moderately improved val [CI] = 1.23–5.01, p = 0.01) and 3.3 times more likely
from 0.45 to 0.50. Four of the 13 contributory factor items to have also experienced a surgical complication (adjusted
loaded onto Factor 1 (Communication and HCP Behavior) OR = 3.27, 95% CI = 1.43–7.51, p = 0.01). The odds of
and included interpersonal and communication issues: (1) problems related to communication and HCP behavior
286 Traber D. Giardina, PhD, et al. Health Care Provider Factors

Table 3. Univariate and Multivariable Logistic Regression for Patient and Adverse Event Characteristics and Harm
Outcomes (N = 334) [Reference group: physical harm only]
Univariate Multivariable

OR 95% 95% CI P Value OR 95% 95% CI P Value

CI low high CI low high
Age at occurrence of event 0.04∗ 0.37
All 3 harms 0.97 0.94 0.99 0.01∗
Nonphysical harm only 0.99 0.96 1.01 0.32
Physical & 1 other harm 0.98 0.95 1.00 0.04∗
Years between event and report 0.01∗ 0.10
All 3 harms 1.00 0.96 1.04 0.92 1.00 0.95 1.04 0.87
Nonphysical harm only 0.85 0.77 0.94 0.01∗ 0.98 0.81 0.98 0.02∗
Physical & 1 other harm 1.00 0.97 1.04 0.93 1.00 0.96 1.05 0.88
Presence of diagnostic error 0.01∗ 0.06
All 3 harms 2.71 1.52 4.83 0.01∗ 1.29 0.62 2.70 0.49
Nonphysical harm only 0.65 0.32 1.32 0.24 0.44 0.18 1.11 0.08
Physical & 1 other harm 1.06 0.60 1.88 0.85 0.61 0.30 1.26 0.18
Presence of a hospital-acquired 0.01∗ 0.05
infection
All 3 harms 0.85 0.49 1.48 0.56 1.22 0.61 2.44 0.58
Nonphysical harm only 0.28 0.13 0.59 0.01∗ 0.33 0.13 0.82 0.02∗
Physical & 1 other harm 0.67 0.38 1.19 0.17 0.84 0.42 1.67 0.61
Presence of a surgical complication < 0.0001∗ < 0.0001∗
All 3 harms 1.98 1.05 3.72 0.03∗ 3.27 1.43 7.51 0.01∗
Nonphysical harm only 0.26 0.13 0.53 0.01∗ 0.21 0.08 0.55 0.01∗
Physical & 1 other harm 1.02 0.56 1.86 0.95 1.19 0.55 2.59 0.66
Presence of an adverse medication 0.01∗ 0.37
event
All 3 harms 2.25 1.27 3.98 0.01∗
Nonphysical harm only 2.86 1.41 5.80 0.01∗
Physical & 1 other harm 1.84 1.03 3.31 0.04∗
Patient reporting† 0.01∗ 0.02∗
All 3 harms 2.74 1.54 4.87 0.01∗ 2.49 1.23 5.01 0.01∗
Nonphysical harm only 1.10 0.55 2.18 0.80 0.72 0.31 1.70 0.45
Physical & 1 other harm 1.57 0.88 2.79 0.12 1.43 0.72 2.84 0.31
Communication and HCP behavior < 0.0001∗ < 0.0001∗
All 3 harms 1.13 1.09 1.18 < 0.0001∗ 1.13 1.07 1.19 < 0.0001∗
Nonphysical harm only 1.03 0.98 1.08 0.25 0.99 0.92 1.06 0.69
Physical & 1 other harm 1.06 1.02 1.11 0.01∗ 1.08 1.02 1.14 0.01∗
HCP fatigue 0.77 –
All 3 harms 1.06 0.92 1.13 0.40
Nonphysical harm only 0.98 0.82 1.17 0.84
Physical & 1 other harm 1.02 0.88 1.18 0.82
Timeliness of HCP response < 0.0001∗ 0.20
All 3 harms 1.31 1.16 1.47 < 0.0001∗ 1.04 0.81 1.35 0.75
Nonphysical harm only 1.07 0.92 1.24 0.3732 1.01 0.75 1.37 0.94
Physical & 1 other harm 1.17 1.04 1.31 0.0110∗ 0.82 0.63 1.06 0.13
∗ Significant at 0.05.
† Patient filed a report of his or her AE to a responsible authority.
OR, odds ratio; CI, confidence interval; HCP, health care personnel.

(Factor 1) as contributing to adverse event were 13% higher 67.2% and 78.6% lower in these groups than those report-
in those experiencing all three harms and 8% higher in pa- ing physical harm, respectively.
tients with more than physical harm (physical plus one)
compared to patients who experienced physical harm alone
(adjusted OR = 1.13, 95% CI = 1.07–1.19, p < 0.0001; DISCUSSION
adjusted OR = 1.08, 95% CI = 1.02–1.14, p = 0.01, re- We analyzed a large sample of patient-reported adverse
spectively). After adjustment, those reporting only financial events gathered by patient advocates to determine the as-
and/or emotional harm were significantly less likely to also sociation between self-reported contributory factors and
report experiencing surgical complications (p = 0.01) and harms experienced by patients. Our analysis of 334 adverse
hospital-acquired infections (p = 0.02), as these odds were events found that more than half of the patients in our sam-
Volume 46, No. 5, May 2020 287

ple experienced multiple harms—emotional, physical, and for Healthcare Research and Quality (AHRQ) and RAND
financial, or some combination. Patients who experienced Corporation Health Care Safety Hotline,41 , 42 while less
all three harms were more likely to have experienced a sur- successful than expected, identified important challenges:
gical AE, filed a report of the AE to a responsible authority, The causes of safety events are complex and may compli-
and experienced problems related to communication and cate structured data collection; the system must meet the
HCP behavior. needs of its users; and the system must respect legal and
AEs can have a significant impact on patients’ regulatory requirements about confidentiality and data pro-
lives,6 , 33 , 34 and our analysis shows that following an tection. These challenges will need to be addressed to gather
adverse event, patients may experience multiple harms. reports that offer more complete data from patients and
These patients likely were exasperated by communication family/caregivers about their experiences from before, dur-
and HCP behavior issues. In our previous qualitative work ing, and after the clinician-patient encounter. This will en-
analyzing a subset of patient- and family-reported diagnos- able health care organizations to develop and use systematic
tic error narratives from this data set, respondents revealed methods to collect patient reports of adverse events and pro-
clinician behavioral and interpersonal factors contributed vide opportunities to measure patient-identified factors that
to diagnostic errors.35 Patients can experience a myriad contribute to adverse events. In addition, health care orga-
of emotions following an AE—guilt, fear, humiliation, nizations will need to develop a safety culture that encour-
helplessness—that can have a lasting impact, such as loss of ages patient reporting and values gathering patients’ per-
trust and health care avoidance,6 , 9 , 20 and financial harm.36 spectives on these issues.
Though there have been patient safety initiatives to en- Patients report communication breakdowns and behav-
courage patients to speak up about patient safety issues, ioral issues as a major contributing factor leading to harm.
only about half feel comfortable speaking to their care team Even after adjustment for all other factors, patients with
about a mistake.37 Patients’ emotional harm may be exacer- multiple types of harms were more likely to report prob-
bated if a patient feels that he or she should have spoken up lems related to communication and HCP behavior as con-
but did not or did speak up and was ignored. Open com- tributing factors. Growing evidence suggests that patients
munication on the part of the clinician about the AE may and their caregivers recognize miscommunication and un-
reduce emotional harm following the event, though it is professional clinician behaviors as threats to their safety.43
not necessarily the norm in practice.36 Nonphysical harms, Communication issues continue to be a frequent contribut-
though less visible and harder to measure, are gaining ing factor to adverse events in the literature.20 , 40 , 44 A recent
attention.6 Recently, Bell et.al. proposed a research agenda systematic review of patients’ perspectives of adverse events
to address emotional harm and identified four research found that patients highlight the role of communication
priorities: patient-centered taxonomy of harm and healing, problems between clinicians and patients (for example, use
epidemiology of emotional harm, awareness-raising, and of medical jargon without explanation, lack of explanation
best practices development and dissemination.8 Identifying of test results) and deficits in the patient-clinician relation-
these types of harms and their contributory factors can ship (for example, lack of respect, lack of time) in adverse
help stimulate policy initiatives to implement programs events.20 These unprofessional behaviors, or disrespect, can
to appropriately address them after an event as well as have far-reaching consequences for patients such as loss of
interventions to prevent them in the first place. trust and negative health outcomes.45 In another study,
We found patients not only can identify safety issues23 , 38 patients identified lack of communication between clini-
but also factors that directly contribute to AEs7 , 16 and pa- cians and a general lack of patient-centered communication
tient harm, which may not be otherwise uncovered. Certain toward patients as problematic (for example, patients are
HCP communication and behavioral issues may be particu- aware they are not receiving the best care if their care team is
larly hard to for health systems to identify outside of patient not effectively communicating issues/symptoms, care plans,
reports. Patient reporting of events provides valuable data or test results, among the team members or to the patients
that can contribute to organizational learning.39 , 40 Existing in a meaningful way).43
research has shown that patients are able to identify con- Based on our results, there appears to be an increased
tributing factors, such as knowledge issues and communi- risk of multiple harms among patients who experienced
cation and safety issues, and want to understand the causes a surgical AE. The prospect of surgical procedures may
of their AE.16 These data can be collected in a standard- cause patients to experience anxiety and stress, leaving them
ized way23 ; however, there are no current US national stan- “psychologically vulnerable,”46 which may exacerbate emo-
dardized efforts to collect AE–related information directly tional harm. Financial harm may include need for addi-
from patients. Next steps in patient safety improvement tional care and/or additional surgical procedures.36 It can
should prioritize developing and testing methods to engage also result when a patient is disabled as a result of the AE and
patients in error reporting as well as leveraging and test- is no longer able to work. Loss of employment also affects
ing existing taxonomies for collecting patient data. Previ- health care insurance benefits, which can have devastating
ous attempts to solicit patient reporting of AEs, the Agency impacts on harmed patients.
288 Traber D. Giardina, PhD, et al. Health Care Provider Factors

Limitations been validated. Despite these limitations, our study evalu-

Our study has several limitations. Purely nonphysical harms ates the association between patient-identified factors that
were experienced by 49 patients (14.7%), but because of contribute to AEs and harm and highlights that patients
data limitations we could not conduct additional qualitative who experience an AE are likely to be experiencing multi-
inquiry or analysis. However, upon reviewing open-ended ple harms.
responses, some of the respondents may not have catego-
rized their harm as physical because they felt it was tem-
porary or minor. For example, some respondents described CONCLUSION
dealing with a postsurgical infection and/or pain but did
Our study highlights the multiple harms that patients may
not select one of the physical harm categories. This sug-
experience following an AE and patient-identified factors
gests that respondents may not have felt that their experi-
that contribute to those harms. Patients’ experiences are es-
ence and/or pain rose to the level of the physical harm op-
sential to identify adverse events and the factors that con-
tions listed in the questionnaire. Although details in these
tribute to those events and should be part of national efforts
cases were not always available, some of these respondents
in patient safety measurement and improvement activities.
selected “no injury” or “no serious injury” despite includ-
The findings underscore the need for new policy and prac-
ing a comment describing excruciating pain or an infec-
tice initiatives to identify, address, and support harmed pa-
tion that required multiple rounds of antibiotics. Finally, for
tients. More systematic collection of patient event reports
some patients who selected emotional and financial harm
will help advance the science of reducing preventable harm
only, there was no detailed information about the AE, and
in health care.
we cannot speculate on these patients’ experiences. Future
questionnaires should account for short-term physical harm
Funding. This project is funded by an Agency for Healthcare Research
and include more nuanced harm categories to represent a and Quality Mentored Career Development Award (K01- HS025474) and
broad range of experiences. partially funded by the Houston VA Health Services Research and De-
Although our sample was national, it does not reflect velopment Service and the Center for Innovations in Quality, Effective-
ness and Safety (CIN 13–413). Dr. Singh is supported by the VA Health
all types of patients’ experiences of adverse events. Expe- Services Research and Development Service (IIR 17-127), the Presiden-
riences are also not representative of all patients who ex- tial Early Career Award for Scientists and Engineers (USA14-274), the VA
perience health care. Voluntary patient surveys are inher- National Center for Patient Safety, The Gordon and Betty Moore Founda-
tion, and a CanTest Research Collaborative funded by a Cancer Research
ently biased because respondents are a self-reporting, self- UK Population Research Catalyst award (C8640/A23385). These funding
selected population willing to share their experience via an sources had no role in the design and conduct of the study; collection,
management, analysis, and interpretation of the data; and preparation,
online questionnaire. Sociodemographic information (for review, or approval of the manuscript. There are no conflicts of interest for
example, race/ethnicity, education) was not collected. We any authors. The views expressed in this article are those of the authors
cannot account for recall bias—the majority of reports oc- and do not necessarily reflect the position or policy of the US Department
of Veterans Affairs or the US government.
curred two years or more after the AE. Although the sur- Conflicts of Interest. All authors report no conflicts of interest.
vey is publicly available online, the survey was initially ad-
vertised through patient safety advocate networks and may
have resulted in our sample having a more comprehensive Traber D. Giardina, PhD, is Investigator, Health Policy, Quality
knowledge of patient safety issues, being more empowered, & Informatics Program, Center for Innovations in Quality, Effec-
and experiencing more severe AEs than the general patient tiveness and Safety (IQuESt), Michael E. DeBakey Veterans Affairs
(VA) Medical Center, Houston, and Assistant Professor, Depart-
population. A larger study with a more generalizable source ment of Medicine, Baylor College of Medicine (BCM), Houston.
population will be needed to confirm the results. This anal- Kathryn E. Royse, PhD, is Epidemiologist, IQuESt, and Instructor,
Department of Medicine, BCM. Arushi Khanna, MPH, is Project
ysis does not shed light on frequency of events or inform any Coordinator, IQuESt. Helen Haskell, MA, is Founder and President,
epidemiologic assessment, and we did not have access to the Mothers Against Medical Error, Columbia, South Carolina. Julia Hal-
medical records or clinicians involved. The outcome vari- lisy, DDS, is Founder, Empowered Patient Coalition, San Francisco.
Frederick Southwick, MD, is Professor, Department of Medicine,
able of harm was broken into four categories based on the University of Florida. Hardeep Singh, MD, is Chief, Health Policy, Quality
data. Participants were able to select any category of harm, & Informatics Program, IQuESt, Michael E. DeBakey VA Medical Center,
Houston, and Professor, Department of Medicine, BCM. Please address
resulting in significant overlap among the categories. To en- correspondence to Traber D. Giardina, [email protected].
sure that the harm outcome categories were exhaustive and
exclusive, we categorized them based on the data rather than
how they would make sense conceptually. It could be that
patients who experienced physical and emotional harm are SUPPLEMENTARY MATERIALS
very different from those who experienced physical and fi-
Supplementary material associated with this article can be
nancial harm. We do not have detailed data on emotional
found, in the online version, at doi:10.1016/j.jcjq.2020.02.
harm. Types of emotional harm, timing of this harm, and
004.
duration requires further study. Finally, the survey has not
Volume 46, No. 5, May 2020 289

REFERENCES 21. Khan A, et al. Families as partners in hospital error

1. Institute of Medicine. To Err Is Human: Building a Safer and adverse event surveillance. JAMA Pediatr. 2017 Apr
Health System. Washington, DC: National Academy Press, 1;171:372–381.
2000. 22. Walton MM, et al. Patients’ reports of adverse events: a data
2. Panagioti M, et al. Prevalence, severity, and nature of pre- linkage study of Australian adults aged 45 years and over.
ventable patient harm across medical care settings: system- BMJ Qual Saf. 2017;26:743–750.
atic review and meta-analysis. BMJ. 2019 Jul 17;366:I4185. 23. Gillespie A, Reader TW. Patient-centered insights: using
3. de Vries EN, et al. The incidence and nature of in-hospital health care complaints to reveal hot spots and blind spots
adverse events: a systematic review. Qual Saf Health Care. in quality and safety. Milbank Q. 2018;96:530–567.
2008;17:216–223. 24. Reader TW, Gillespie A, Roberts J. Patient complaints in
4. Classen DC, et al. ‘Global trigger tool’ shows that adverse healthcare systems: a systematic review and coding taxon-
events in hospitals may be ten times greater than previously omy. BMJ Qual Saf. 2014;23:678–689.
measured. Health Aff (Millwood). 2011;30:581–589. 25. Giardina TD, et al. Root cause analysis reports help
5. Panesar SS, et al. How safe is primary care? A systematic identify common factors in delayed diagnosis and treat-
review. BMJ Qual Saf. 2016;25:544–553. ment of outpatients. Health Aff (Millwood). 2013;32:1368–
6. Ottosen MJ, et al. Long-term impacts faced by patients and 1375.
families after harmful healthcare events. J Patient Saf. Epub. 26. Szekendi MK, et al. Active surveillance using electronic trig-
2018 Jan 17. gers to detect adverse events in hospitalized patients. Qual
7. Van Vorst RF, et al. Rural community members’ perceptions Saf Health Care. 2006;15:184–190.
of harm from medical mistakes: a High Plains Research Net- 27. Southwick FS, Cranley NM, Hallisy JA. A patient-initiated
work (HPRN) study. J Am Board Fam Med. 2007;20:135–143. voluntary online survey of adverse medical events: the per-
8. Bell SK, et al. A multi-stakeholder consensus-driven re- spective of 696 injured patients and families. BMJ Qual Saf.
search agenda for better understanding and supporting the 2015;24:620–629.
emotional impact of harmful events on patients and fami- 28. The Empowered Patient Coalition. Report a Medical
lies. Jt Comm J Qual Patient Saf. 2018;44:424–435. Event. Accessed Feb 25, 2020. https://empoweredpatient
9. Delbanco T, Bell SK. Guilty, afraid, and alone—strug- coalition.org/report- a- medical- event/.
gling with medical error. N Engl J Med. 2007 Oct 29. March 2010 report: https://oig.hhs.gov/oei/reports/
25;357:1682–1683. oei- 06- 08- 00221.pdf.
10. Kinnunen UM, Sarantob K. It is time for self-incident-report- 30. Bland JM, Altman DG. Cronbach’s alpha. BMJ. 1997 Feb
ing for patients and their families in every health care or- 22;314:572.
ganization: a literature review. Stud Health Technol Inform. 31. Floyd FJ, Widaman KF. Factor analysis in the development
2013;192:92–96. and refinement of clinical assessment instruments. Psychol
11. Zhu J, et al. Can we rely on patients’ reports of adverse Assess. 1995;7:286–299.
events? Med Care. 2011;49:948–955. 32. Fabrigar LR, et al. Evaluating the use of exploratory fac-
12. Dovey SM, et al. A preliminary taxonomy of medical errors tor analysis in psychological research. Psychol Methods.
in family practice. Qual Saf Health Care. 2002;11:233–238. 1999;4:272–299.
13. Fernald DH, et al. Event reporting to a primary care patient 33. Mazor KM, et al. Parents’ perceptions of medical errors. J
safety reporting system: a report from the ASIPS collabora- Patient Saf. 2010;6:102–107.
tive. Ann Fam Med. 2004;2:327–332. 34. Institute for Healthcare Improvement; NORC at the Uni-
14. Makeham MA, et al. An international taxonomy for errors versity of Chicago. Americans’ Experiences with Medical
in general practice: a pilot study. Med J Aust. 2002 Jul Errors and Views on Patient Safety: Final Report, 2017.
15;177:68–72. Accessed Feb 25, 2020 http://www.ihi.org/about/news/
15. Weingart SN. et al. What can hospitalized patients tell us Documents/IHI_NPSF_NORC_Patient_Safety_Survey_
about adverse events? Learning from patient-reported inci- 2017_Final_Report.pdf.
dents. J Gen Intern Med. 2005;20:830–836. 35. Giardina TD, et al. Learning from patients’ experiences re-
16. Etchegaray JM, et al. Patients as partners in learning from lated to diagnostic errors is essential for progress in patient
unexpected events. Health Serv Res. 2016;51:2600–2614. safety. Health Aff (Millwood). 2018;37:1821–1827.
17. Kok J, Leistikow I, Bal R. Patient and family engagement in 36. Agency for Healthcare Research and Quality, Patient Safety
incident investigations: exploring hospital manager and in- Net. Book/Report: The Financial and Human Cost of
cident investigators’ experiences and challenges. J Health Medicine Error and How Massachusetts Can Lead the Way
Serv Res Policy. 2018;23:252–261. on Patient Safety, Jun 19, 2019. Accessed Feb 25, 2020
18. Weissman JS, et al. Comparing patient-reported hospital https://psnet.ahrq.gov/issue/financial- and- human- cost-
adverse events with medical record review: do patients medical- error- and- how- massachusetts- can- lead- way-
know something that hospitals do not? Ann Intern Med. patient-safety.
2008 Jul 15;149:100–108. 37. Bell SK, et al. Speaking up about care concerns in the ICU:
19. King A, et al. Time to listen: a review of methods to so- patient and family experiences, attitudes and perceived
licit patient reports of adverse events. Qual Saf Health Care. barriers. BMJ Qual Saf. 2018;27:928–936.
2010;19:148–157. 38. Entwistle VA, et al. Speaking up about safety concerns: mul-
20. Lang S, Velasco Garrido M, Heintze C. Patients’ views of ad- ti-setting qualitative study of patients’ views and experi-
verse events in primary and ambulatory care: a systematic ences. Qual Saf Health Care. 2010;19:e33.
review to assess methods and the content of what patients 39. Armitage G, et al. Patient-reported safety incidents as a new
consider to be adverse events. BMC Fam Pract. 2016 Jan source of patient safety data: an exploratory comparative
27;17:6. study in an acute hospital in England. J Health Serv Res Pol-
icy. 2018;23:36–43.
290 Traber D. Giardina, PhD, et al. Health Care Provider Factors

40. O’Hara JK, et al. What can patients tell us about the text responses using the Patient Reported Experiences and
quality and safety of hospital care? Findings from a UK Outcomes of Safety in Primary Care (PREOS-PC) question-
multicentre survey study. BMJ Qual Saf. 2018;27:673– naire. Health Expect. 2017;20:961–972.
682. 44. Vermeir P, et al. Communication in healthcare: a narrative
41. Weingart SN, et al. Implementation and evaluation of a pro- review of the literature and practical recommendations. Int
totype consumer reporting system for patient safety events. J Clin Pract. 2015;69:1257–1267.
Int J Qual Health Care. 2017 Aug 1;29:521–526. 45. Sokol-Hessner L, et al. Development of a framework to de-
42. Schneider EC, et al. Developing and testing the Health Care scribe patient and family harm from disrespect and promote
Safety Hotline: a prototype consumer reporting system for improvements in quality and safety: a scoping review. Int J
patient safety events. Rand Health Q. 2017 Jun 19;6:1. Qual Health Care. Epub. 2018 Nov 14.
43. Ricci-Cabello I, et al. Identifying patient-centred recom- 46. Carr EC, et al. Patient experiences of anxiety, depression
mendations for improving patient safety in general prac- and acute pain after surgery: a longitudinal perspective. Int
tices in England: a qualitative content analysis of free– J Nurs Stud. 2005;42:521–530.