Applied Nursing Research 61 (2021) 151476

Available online 7 July 2021
0897-1897/© 2021 Elsevier Inc. All rights reserved.

Resilience as a mediator between compassion fatigue, nurses’ work
outcomes, and quality of care during the COVID-19 pandemic

Leodoro J. Labrague a, *, Janet Alexis A. de los Santos b

a Fundamentals and Administration Department, College of Nursing, Sultan Qaboos University, Muscat, Oman
b College of Nursing, Visayas State University, Philippines

A R T I C L E I N F O

Keywords:
Compassion fatigue
Resilience
Turnover intention
Job satisfaction
Quality of care
Nursing
COVID-19 pandemic

A B S T R A C T

Background: Nurses in the frontline of the battle against COVID-19 are highly vulnerable to compassion fatigue
(CF), which may affect their mental health, work effectiveness, and patient safety outcomes. However, no studies
have investigated nurses’ CF in relation to job outcomes and care quality during the pandemic.
Aims: This study aims to examine the mediating role of resilience in the relationship between CF and frontline
nurses’ job outcomes (job satisfaction and turnover intention) and care quality.
Design: An online, cross-sectional survey containing five self-report scales was used to collect data from 270
frontline nurses in selected hospitals in the Philippines.
Results: Overall, 38.5% of frontline nurses experienced medium to high CF during the second wave of the
pandemic. Increased CF was associated with poorer nurse-reported quality of care (β = − 0.145, p = 0.019),
lower job satisfaction (β = − 0.317, p = 0.001), and higher organizational turnover intention (β = 0.301, p =
0.001). Moreover, resilience fully mediated the relationship between CF and quality of care (β = − 0.088, p =
0.169), and partially mediated the relationship between CF and job satisfaction (β = − 0.259, p = 0.001), and CF
fatigue and organizational turnover intention (β = 0.272, p = 0.001).
Conclusion: Frontline nurses are at risk of developing CF during the pandemic. Psychological resilience reduces
the negative impact of CF on frontline nurses’ job satisfaction, turnover intention, and the quality of care in their
assigned unit. Proactive measures to reduce CF should be prioritized by nursing administrators. Resilience-
promoting interventions could foster job satisfaction and retention in nurses and, hence, the quality of care
delivered in their units.

1. Introduction

The coronavirus disease, or COVID-19, originated in China in the
later part of 2019 and has created an unprecedented burden for
healthcare systems around the world. The pandemic has brought addi-
tional stress and psychological burden to healthcare workers, particu-
larly those involved in the direct care and management of coronavirus
patients, posing an increased risk for the development of compassion
fatigue. Defined as “a state of emotional exhaustion due to the dynamics
of a caring relationship with an individual or a group of individuals who
have suffered a sudden or severe loss” (Sabo, 2011), compassion fatigue
(CF) often occurs as a consequence of persistent exposure to patients’
suffering, stressful work conditions, and inadequate utilization of mea-
sures to promote self-care (Peters, 2018; Sorenson, Bolick, Wright, &
Hamilton, 2017). The global estimate of CF prevalence among nurses

during the pre-pandemic period ranged from 22% to 60% (Zhang et al.,
2018) and was found to be influenced by myriad factors including
personal, work-related, and psycho-social factors (Cavanagh et al.,
2020; Ortega-Campos et al., 2020). Age, marital status, education,
gender, salary, ethnicity, and overall health condition were identified as
personal factors associated with CF. Work-related factors included years
of work experience, job rank, job status, work climate, and presence of
violence in the workplace; psycho-social factors strongly linked to CF
included family support, coping skills, psychological distress, and pro-
fessional identity (O’Callaghan, Lam, Cant, & Moss, 2020; Xie et al.,
2021). Manifestations of CF include decreased energy, reduced ability to
empathize, increased hopelessness, and heightened emotional exhaus-
tion (Peters, 2018).

When left unmanaged, CF may have untoward repercussions on
nurses’ well-being and health, resulting in mental health concerns (e.g.,

* Corresponding author.
E-mail address: [email protected] (L.J. Labrague).

Contents lists available at ScienceDirect

Applied Nursing Research

journal homepage: www.elsevier.com/locate/apnr

https://doi.org/10.1016/j.apnr.2021.151476
Received 14 May 2021; Received in revised form 1 July 2021; Accepted 3 July 2021

Applied Nursing Research 61 (2021) 151476

2

burnout, depression, stress, and anxiety) leading to work impairment
(Pérez-García et al., 2021), job dissatisfaction (Kim et al., 2017), and
eventual nurse turnover (Wells-English, Giese, & Price, 2019). These
conditions could ultimately influence the way nurses provide care,
resulting in increased incidence of adverse events, patient errors and
missed nursing care, and poorer health service delivery (Alharbi, Jack-
son, & Usher, 2020a, 2020b, 2020c; Peters, 2018). Hence, many
healthcare organizations have implemented various efforts to address
this issue and to assist healthcare workers (HCWs), including nurses, to
effectively deal with the stress that accompanies the pandemic.

Nurses on the frontline against COVID-19 encounter numerous
stressors in addition to stressors found in usual circumstances, including
increased patient workloads, lack of reliable personal protective
equipment (PPEs), additional coronavirus protocols, inadequate prepa-
ration related to care of infected patients, and poor working conditions
to effectively carry out their duty (Arnetz, Goetz, Arnetz, & Arble, 2020).
These conditions, along with heightened fear of being infected or un-
knowingly infecting their family and friends (Khattak et al., 2021;
Labrague & de los Santos, 2021a, 2021b), as well as infection control
measures (e.g., social distancing, lockdown), could negatively affect
nurses’ ability to be compassionate and increase their risk of developing
CF (Alharbi et al., 2020a, 2020b, 2020c). Available research has shown
that as many as 70% of frontline nurses suffered from moderate to severe
CF during the pandemic, with 97% of nurses experiencing at least one
symptom of CF (Erkin, Konakçı, & Duran, 2021). A higher rate of CF was
mostly observed in nurses assigned to critical care units, emergency care
units, and units designated for treating and managing patients with
COVID-19 (Ruiz-Fernández et al., 2020), although a significant pro-
portion of nurses from other hospital units were also seen to suffer from
CF related to caring for non-COVID-19 patients during the pandemic
(Alharbi et al., 2020a).

CF during the pandemic may affect the mental health of frontline
nurses, leading to work impairment and reduced capacity to provide
quality of care to their patients (Labrague, 2021; Pérez-García et al.,
2021). With the continuing battle against COVID-19, along with the
increasing number of individuals being infected, the emergence of new
coronavirus variants, and seemingly no end in sight for the pandemic,
frontline nurses will continue to suffer from CF unless measures are
initiated. Considering the costs associated with CF, it is imperative that
proactive measures to adequately manage CF be prioritized by hospital
and nursing administrators. Harnessing nurses’ personal resources,
including psychological resilience – defined as a person’s ability to
rebound from adversities or stress-provoking events – could be an
important measure to address CF in this group of nurses.

Resilience has been identified as an essential protective factor
against the mental and psychological health effects of traumatic events
and adversities including calamities, disaster emergency situations, and
outbreaks of infections (Labrague, 2021; Pollock et al., 2020). Pre-
pandemic studies consistently associated psychological resilience with
sustained health (e.g., mental, psychological, and emotional health) and
positive job outcomes (e.g., increased work performance, job engage-
ment, and retention) in nurses despite stressful situations (Cooper et al.,
2020; Cooper et al., 2021; Hart et al., 2014). Moreover, during disease
outbreaks, including the previous outbreaks of Ebola, SARS, H1NI, and
MERS-COV, such personal resources safeguarded nurses’ mental and
psychological well-being and allowed them to continuously carry out
their professional nursing role (Maiorano, Vagni, Giostra, & Pajardi,
2020; Polloch et al., 2020). During the height of the coronavirus crisis,
evidence showed that nurses who had adequate levels of resilience
better handled the negative impact of the pandemic, resulting in
reduced anxiety, post-traumatic stress, emotional exhaustion, and
depression (Yörük & Güler, 2021; Zhang et al., 2021). It is therefore not
surprising that many healthcare institutions have invested heavily in
programs and interventions to harness resilience in healthcare workers
serving in the forefront of the pandemic, and hence to improve their
overall health and work performance (Adimando, 2018; Dreher,

Hughes, Handley, & Tavakoli, 2019).
With regards to CF, a handful of studies showed the efficacy of

resilience in reducing fatigue associated with caring for patients
(Alharbi et al., 2020a, 2020b, 2020c; Maiorano et al., 2020). Psycho-
logical resilience, along with increased work engagement and positive
coping, was attributed to a significant reduction in CF levels among
frontline nurses during the early months of the pandemic (Cao & Chen,
2020; Cho & Jung, 2014; Maiorano et al., 2020). Based on this evidence,
psychological resilience may also influence the effects of CF on nurses’
work outcomes and patient safety outcomes. However, despite these
assumptions, the mechanism underlying this relationship remains un-
explored. Understanding this mechanism is critically important in the
formulation of interventions aimed at supporting the mental health of
frontline nurses in order to ensure the delivery of quality, safe, and
compassionate care to their patients.

1.1. Theoretical framework

This study was anchored on the Compassion Fatigue Model (CFM;
Coetzee & Laschinger, 2018), which conceptualizes lack of resources,
inadequate positive feedback, and nurses’ reactions to personal distress
as drivers of compassion fatigue (Coetzee & Laschinger, 2018). These
resources include personal (resilience, coping, self-esteem), organiza-
tional (hospital resources, staffing), and energy resources (knowledge,
energy, time). According to this model, nurses who are poorly resourced
(e.g., lack of resilience, inadequate hospital resources) and lacked
empathic focus on the patient are more likely to experience patient’s
needs as a threat to their resources, resulting in compassion fatigue.
Conversely, nurses who are well-resourced and have patient-oriented
focus tend not to perceive the patient’s needs as a threat, thereby
reducing the risk for compassion fatigue (Coetzee & Laschinger, 2018).
The current pandemic poses a significant threat to nurses’ resources,
which could lead to compassion fatigue and subsequently affect their job
productivity and the overall delivery of nursing care.

1.2. Hypothesize model

Based on the review of the literature, the following hypotheses were
formulated. First, we hypothesized that CF may negatively influence job
satisfaction (Hypothesis 1) and quality of care (Hypothesis 2), and
positively influence organizational turnover intention (Hypothesis 3).
Moreover, we hypothesized that positive coping may mediate the rela-
tionship between compassion fatigue CF and job satisfaction (Hypoth-
esis 4), organizational turnover intention (Hypothesis 5), and quality of
care (Hypothesis 6). In other words, resilience may reduce the negative
effects of CF, resulting in heightened job satisfaction, reduced organi-
zational turnover intention, and increased perceptions of care quality
(Fig. 1).

2. Methods

2.1. Design, samples, and settings

A cross-sectional study using an online survey was conducted in
selected hospitals in the Philippines to collect data from nurses who
were caring for coronavirus patients. To be eligible for the study, par-
ticipants were required to be registered and licensed nurses, working in
the hospital in the previous six months and directly involved in the care
of suspected or infected coronavirus patients. The sample size was
calculated using an online calculator (Soper, 2021) for multiple
regression. Considering a small effect size, statistical power of 80%,
alpha of 0.05, and four initial predictors in the multiple regression, the
required sample size was 242. The link to the survey was sent to 300
frontline nurses in the region, of which 270 responded.

L.J. Labrague and J.A.A. de los Santos

Applied Nursing Research 61 (2021) 151476

3

2.2. Instrumentation

The 13-item short version of the Compassion Fatigue Scale (CFS) was
used to assess CF in frontline nurses in terms of their burnout and sec-
ondary trauma. Nurses responded to CFS items using a 10-point Likert
scale ranging from 1 (never) to 10 (very frequent). The CFS mean score
was calculated by adding all items; a higher mean score reflects a higher
degree of CF. The CFS mean scores were divided into three classifica-
tions: low CF (1.00–2.33), medium CF (2.34–3.66) and high CF
(3.67–5.00) (Barnett & Flores, 2016). The criterion and predictive val-
idity were confirmed in previous studies (Barnett & Flores, 2016).
Moreover, previous research reported the internal consistency of the
scale within the acceptable level, ranging from 0.85 to 0.91 (Adams,
Figley, & Boscarino, 2008).

Frontline nurses’ capacity to bounce back from stressful situations
brought about by the pandemic was examined using the 4-item Brief
Resilient Coping Skills (Smith et al., 2008). Nurses rated items using a
five-point Likert scale ranging from 0 (does not describe me at all) to 5
(describes me very well), with a higher composite score representing
higher resilience. The criterion validity was established in previous
research, while the scales’ internal consistency was found to be accept-
able (Labrague & de los Santos, 2020).

The Job Satisfaction Scale was used to assess nurses’ satisfaction with
their present job during the height of the coronavirus crisis (Taunton
et al., 2004). Nurses responded to 7-items using a five-point Likert scale
ranging from 1 (strongly disagree) to 5 (strongly agree), with a higher
composite score representing a higher level of job satisfaction. The

concurrent and criterion validity of this scale were found to be excellent,
with acceptable reliability (0.91) (Taunton et al., 2004).

Nurses’ intention to leave their current organization was examined
by a single-item measure. Using a five-point Likert scale ranging from
0 (strongly disagree) to 5 (strongly agree), nurses responded to the item,
“Given the current situation, I am more likely to leave my profession”.
Research during the pandemic has established the validity and internal
consistency of the scale, with a value of 0.86 (Labrague & de los Santos,
2020).

A single question to measure nurses’ judgment and appraisal of the
quality of care provided in their assigned unit was used. Nurses
appraised the overall nursing care provided in their unit for the previous
days using a four-point Likert scale ranging from 0 (poor) to 4 (excellent),
with a higher score indicating a higher quality of care delivered. This
single-item measure was found to have excellent criterion and predictive
validity based on previous research, while the scales’ test-retest reli-
ability was found to be within the acceptable range (r = 0.79) (Labrague
& de los Santos, 2021b).

The internal consistencies of the multi-item scales were 0.91 for the
CFS, 0.85 for the Brief Resilient Coping Skills, and 0.87 for the Job
Satisfaction Scale. The test-retest reliabilities of the single-item mea-
sures were 0.88 for Organizational Turnover Intention, and 0.86 for
Quality of Care.

2.3. Data collection and ethical review

Ethical clearance for the study was provided by the Institutional and

Compassion Fatigue Quality of Care

Resilience

Compassion Fatigue Job Contentment

Resilience

Compassion Fatigue
Professional Turnover

Intention

Resilience

–

–

+

Fig. 1. Hypothesized models.

L.J. Labrague and J.A.A. de los Santos

Applied Nursing Research 61 (2021) 151476

4

Review Taskforce of (omitted for review purposes). Due to the ongoing
restrictions regarding in-person collection of research data, a link to the
online questionnaire was sent to nurses’ official email addresses. The
questionnaire was created using Survey Monkey and Google Forms. The
link provided a brief description of the study and contained a section to
obtain informed consent, whereby nurses indicated whether they agreed
or declined to participate in the research. Nurses who agreed to
participate in the research were asked to click an ‘I wish to participate’
button, while those who declined to participate were asked to click an ‘I
don’t wish to participate’ button. To encourage a high response rate, we
sent weekly reminders to nurses’ email addresses asking them to com-
plete the survey. Nurses were reminded of the survey by sending weekly
reminders to their email address or social media accounts. To maintain
nurses’ confidentiality, no personal identifying information was
collected. Data collection took place from 1 November 2020 to 1
December 2020.

2.4. Data analysis

Data were analyzed using the Statistical Package for Social Sciences
(SPSS) version 23. Prior to conducting analyses, data were inspected for
missing values. Three questionnaires with missing values were removed
from the data analysis. Descriptive statistics were calculated, including
means, standard deviations, frequencies and percentages. To identify
relationships and significant differences in key study variables, t-tests,
Pearson’s r correlations, and ANOVA were used. The mediating effect of
resilience was assessed using the 3-step approach described by Baron
and Kenny (1986). First, the effects of the independent variable (CF) on
the mediator (resilience) were examined. Next, the direct influence of
the independent variable (CF) on the outcome variables (nurses’ job
satisfaction, organizational turnover intention, and quality of care) was
calculated. Finally, the direct influence of the mediator (resilience) on
the outcome variables (job satisfaction, organizational turnover inten-
tion, and quality of care) was examined. The significance of the medi-
ation model was examined using the Sobel Test.

3. Results

The link to the online survey was sent to 300 frontline nurses and 270
completed the survey. The majority of the participants were female (n =
201, 74.5%), not married (n = 153, 56.8%), and held BSN degrees (n =
214, 79.2%). The vast majority held a full-time job role (n = 244,
90.3%), while more than half held staff nurse positions (n = 188,
69.5%). More than half (77.6%) reported the adequacy of the personal
protective equipment in their workplace as ‘sufficient’ to ‘very suffi-
cient’, while 52.5% (n = 141) reported the adequacy of the staff as
‘sufficient’ to ‘very sufficient’ (Table 1). Pearson’s r coefficients identi-
fied negative correlations between CF and nurses’ age (r = − 0.156, p =
0.012) and years of nursing experience (r = − 0.170, p = 0.006).

The composites scores of the scales were as follows: 2.212 (CFS),
1.93 (organizational turnover intention), 3.505 (job satisfaction), and
3.281 (quality of care). Low CF was reported by 61.4% (n = 159) of
frontline nurses, while 28.2% (n = 73) and 10.4% (n = 27) reported
medium and high compassion fatigue, respectively. Bivariate analysis
identified significant correlations between key variables in expected
directions. CF negatively and significantly correlated with quality of
care (r = − 0.145, p = 0.01), job satisfaction (r = − 0.317, p = 001), and
positively and significantly correlated with organizational turnover
intention (r = 0.301, p = 0.001) (Table 2).

Table 1
Nurse characteristics (n = 270).

Variables Category n % Mean SD Test statistic p value

Age 34.861 8.832 − 0.156 0.012
Years of experience in nursing 10.671 7.532 − 0.170 0.006
Years of experience in the organization 6.065 5.580 − 0.081 0.193
Gender Male 69 25.5 2.154 0.958 − 0.576 0.566

Female 201 74.5 2.233 0.988
Marital status Married 117 43.2 2.079 0.992 − 1.923 0.056

Unmarried 153 56.8 2.315 0.960
Education BSN 214 79.2 2.179 0.958 − 1.029 0.307

MSN 56 20.8 2.342 1.054
Job status Fulltime 244 90.3 2.212 0.956 − 0.038 0.970

Part time 26 9.7 2.222 1.199
Job role Staff nurse 188 69.5 2.215 0.932 0.047 0.963

Nurse manager 82 30.5 2.208 1.086
Facility size Small 94 34.7 2.386 1.077 2.497 0.084

Medium 77 28.6 2.052 0.845
Large 99 36.7 2.175 0.964

Attendance in COVID-19 related trainings Yes 144 53.3 2.193 0.985 − 0.352 0.725
No 126 46.7 2.236 0.976

Vaccination status Vaccinated 211 78.0 2.161 0.923
Not vaccinated 59 22.0 2.398 1.148

Personal protective adequacy 0 0.099 0.111
Very insufficient 11 4.2
Insufficient 49 18.1
Sufficient 120 44.4
Very sufficient 90 33.2

Staff adequacy 0 − 0.121 0.051
Very insufficient 29 10.8
Insufficient 99 36.7
Sufficient 109 40.5
Very sufficient 32 12.0

Table 2
Correlations between key study variables.

Variables Mean SD 1 2 3 4

1. Quality of care 3.282 0.654 1
2. Job satisfaction 3.506 0.954 0.367*** 1
3. Turnover

intention
1.931 1.065 − 0.128* − 0.0106 1

4. Compassion
fatigue

2.213 0.979 − 0.145* − 0.317** 0.301** 1

** p = 0.001.
* p = 0.01.

L.J. Labrague and J.A.A. de los Santos

Applied Nursing Research 61 (2021) 151476

5

Using multiple regression, the direct and indirect effects of CF on
frontline nurses’ job satisfaction, organizational turnover intention, and
quality of care were tested (Table 3). First, CF was negatively associated
with quality of care (β = − 0.145, p = 0.019) and job satisfaction (β =
− 0.317, p = 0.001), and positively associated with turnover intention (β
= 0.301, p = 0.001). Next, resilience was positively associated with
quality of care (β = 0.217, p = 0.001) and job satisfaction (β = 0.272, p
= 0.001), and negatively associated with turnover intention (β =
− 0.178, p = 0.004). Additionally, resilience fully mediated the rela-
tionship between CF and quality of care (β = − 0.088, p = 0.169).
Further, resilience partially mediated the relationship between (a) CF
and job satisfaction (β = − 0.259, p = 0.001), and (b) CF and organi-
zational turnover intention (β = 0.272, p = 0.001) (Table 3). The final
models are shown in Fig. 2. In other words, resilience reduced the effects
of CF on frontline nurses’ job satisfaction, organizational turnover
intention, and perceived quality of care.

4. Discussion

Findings of this study supported the hypothesized model and pro-
vided further evidence of the validity of the Compassion Fatigue Model,
which identifies resilience as a personal resource, to reduce the negative
effects of compassion fatigue on frontline nurses’ job outcomes and care
quality.

Prior to the pandemic, the prevalence of CF in nurses ranged from
22% to 60% (Ortega-Campos et al., 2020; Xie et al., 2021), while during
the initial surge of the coronavirus crisis, the prevalence reached 70%
(Erkin et al., 2021). In the current study, 38.6% of frontline nurses
experienced medium to high CF. This rate was lower relative to earlier
studies involving physicians and nurses in Spain in which 60.5% of
HCWs experienced high levels of compassion fatigue (Ruiz-Fernández
et al., 2020). Meanwhile, in a study conducted in Turkey (Erkin et al.,
2021), about 70% of nurses reported experiencing high to severe CF. It is
worth noting that in the current study, data was collected in selected
cities of the country with comparably lower rates of COVID-19 cases,
while the studies in Spain and Turkey were conducted in cities with high
numbers of cases. This could possibly explain the lower prevalence of CF
in our sample. Despite the lower percentage of frontline nurses in the
present study experiencing CF in comparison to previous studies (Erkin
et al., 2021; Ruiz-Fernández et al., 2020), such a result represents an
issue that requires prompt action from nursing administrators. With the
number of COVID-19 cases increasing daily, along with the emergence
of new coronavirus variants, the number of nurses experiencing CF may
continue to grow. Unless herd immunity is achieved or the majority of
the population is vaccinated, frontline nurses likely will continue to
suffer significant levels of CF as a result of caring for patients and wit-
nessing patient suffering and death due to coronavirus disease. As such,
it is imperative that proactive resilience measures be in place to
adequately support and safeguard the mental and psychological well-

being of nurses.
Additionally, our study found that job satisfaction was moderate to

high while organizational turnover intention was low, suggesting that
despite CF, nurses were still able to find satisfaction in their jobs,
resulting in lower turnover intention. Such a result differs from earlier
studies in the local (Labrague & de los Santos, 2020) and international
(Irshad, Khattak, Hassan, Majeed, & Bashir, 2020; Zhang et al., 2021)
context, in which nurses reported job dissatisfaction and increased levels
of organizational and professional turnover intention due to increased
fear of the coronavirus disease. Our data collection took place during the
months when COVID-19 cases were under control and positive rates
were declining due to stringent implementation of infection-control
measures (e.g., stay-at-home orders, social distancing), which may
explain our finding of a higher job satisfaction level and a lower level of
organizational turnover intention. By contrast, previous studies (Irshad
et al., 2020; Labrague & de los Santos, 2020; Zhang et al., 2021) were
conducted during the earlier surge of the pandemic. At that time, less
information was available regarding effective containment and pre-
vention measures, and the best protocols to manage suspected and
infected patients were still unknown; hence, nurses were apprehensive
and fearful, resulting in higher job dissatisfaction and increased inten-
tion to leave their work.

Bivariate analyses found significant negative correlations between
CF and nurses’ age and years of work experience. As nurses’ age and
years of work experience increased, their risk of experiencing CF
decreased. This result supports findings from a vast number of studies
during the pre-pandemic period in which nurses’ age and years of
nursing experience were strongly linked to CF (Alharbi et al., 2020a,
2020b, 2020c; Sacco, Ciurzynski, Harvey, & Ingersoll, 2015; Xie et al.,
2021). For instance, an integrative review by Alharbi et al. (2019) found
that younger nurses experienced a higher level of CF than older nurses,
while cross-sectional studies in China, Korea, the US, and Saudi Arabia
identified age and work experience as important risk factors for CF
(Alharbi et al., 2020a, 2020b, 2020c; Sacco et al., 2015; Xie et al., 2021).
A few possible explanations for these findings are presented here. First,
younger nurses have less experience in dealing with different stressors
and may have less mature problem-solving and decision-making skills
(Saintsing, Gibson, & Pennington, 2011), which are vital when handling
patient issues and caring for patients with complex …