Organizational Culture andIts Relation-
ship withHospital Performance in Public
Ping Zhou, Kate Bundorf, Ji Le Chang, Jin Xin Huang, and
Objective. To measure perceptions of organizational culture among employees of
public hospitals in China and to determine whether perceptions are associated with
Data Sources. Hospital, employee, and patient surveys from 87 Chinese public
hospitals conducted during 2009.
Study Design. Developed and administered a tool to assess organizational culture in
Chinese public hospitals. Used factor analysis to create measures of organizational cul-
ture. Analyzed the relationships between employee type and perceptions of culture
and between perceptions of culture and hospital performance using multivariate
Principal Findings. Employees perceived the culture of Chinese public hospitals as
stronger in internal rules and regulations, and weaker in empowerment. Hospitals in
which employees perceived that the culture emphasized cost control were more profit-
able and had higher rates of outpatient visits and bed days per physician per day but
also had lower levels of patient satisfaction. Hospitals with cultures perceived as
customer-focused hadlonger length ofstay butlower patient satisfaction.
Conclusions. Managers in Chinese public hospitals should consider whether the
culture of their organization will enable them to respond effectively to their changing
Key Words. Business and management, comparative health systems/international
health, hospitals, organization theory
In 2009, the Chinese government announced a major health care system
reform,with publichospitals being animportant target forreformefforts.Pub-
lic hospitals generate the bulk of their revenues from regulated fees charged
to patients and insurers. As the government sets fees for basic services very
low to make these services more accessible to patients, hospitals have strong
CALL FOR PAPERS FORTHEME ISSUE: GLOBAL HEALTH SYSTEMS
Health Services Research
incentives to over-provide more profitable high-tech services and pharmaceu-
ticals to remain financially viable. A key objective of the proposed reforms is
to changethe behaviorof publichospitals.
A key determinant of the effectiveness of the proposed reforms will be
how public hospitals respond to potentially dramatic changes in their external
environment. While major organizational changes without changes in organi-
zational culture often fail (Umiker 1999), little is known about the organiza-
tional culture of Chinese public hospitals. The main purposes of this study
were to determine how employees perceive organizational culture in China’s
public hospitals, to compare perceptions of hospital culture among different
types of employees, and to examine the association between employee per-
ceptionsof hospital culture and hospital performance.
While consensus does not exist on how to define organizational culture
(Cooke and Rousseau 1988; King and Byers 2007; Zhang, Li, and Pan 2009),
a commonly used definition is “the set of shared, taken-for-granted, implicit
assumptions that a group holds and that determine how it perceives, thinks
about, and reacts to its various environments” (Kreitner and Kinicki 2008).
Thus, the essence of culture is a core of basic assumptions. Behavioral norms
and values are a manifestation of these assumptions, and values and norms, in
turn, encourage activities that represent the expression of organizational
Organizational climate, in contrast, is defined as employees’ shared per-
ceptions regarding an organization’s policies, procedures, and practices,
which in turn serve as indicators of the types of behavior that are rewarded
and supported in work settings (Schneider, Gunnarson, and Niles-Jolly 1994;
Zohar and Luria 2010). Organizational culture is a broader concept than orga-
nizational climate, and organizational culture can be used to explain why an
organization focuses on certain priorities. While our study focuses on organi-
Address correspondence to Di Xue, Ph.D., M.P.H., M.D., Professor, Director, Department of
Hospital Management, School of Public Health, Fu Dan University, Shanghai, 200032, China;
e-mail: firstname.lastname@example.org. Ping Zhou, Ph.D., is with the Department of Hospital Administra-
tion, School of Public Health, Fu Dan University, Shanghai, China. Kate Bundorf, Ph.D., M.B.A.,
M.P.H., is with the School of Medicine, Stanford University, Stanford, CA. Ji Le Chang, M.D., is
with the Health Bureau of Gansu Province, Gansu Province, China. Jin Xin Huang, M.B.A., M.
D.,is withtheHealth DepartmentofHubeiProvince, HubeiProvince,China.
2140 HSR: Health Services Research 46:6, PartII (December 2011)
zational culture, we refer to some studies on organizational climate, particu-
larly in the contextof patientsafety, whichexaminerelatedissues.
Assessmentof Hospital Culture
Twoconceptualframeworks areoften usedtoassesshospitalculture:theDeni-
son model and Quinn and Rohrbaugh’s competing values framework (CVF).
The Denison framework is based on four cultural traits: mission, consistency,
adaptability, and involvement (Denison 1990). Mission refers to a long-term
directionfortheorganization; consistencyreferstothevaluesandsystems that
are the basis of a strong culture; adaptability refers to the ability to translate
the demands of the business environment into action; and involvement refers
to building human capability, ownership, and responsibility. Each of these
traits is characterized by three sub-dimensions. The Denison model has been
used to assess culture in a variety of industries (Hatchand Cunliffe2006).
Studies in the health services field often use Quinn and Rohrbaugh’s
CVF (Quinn and Rohrbaugh 1981). In the CVF, there are two sets of compet-
ing values. The first is centralization and control over organizational processes
versus decentralization and flexibility. The second is whether the organization
is oriented toward its own internal environment and processes or the external
environment and relationshipswithoutside entities (suchas regulators, suppli-
ers, competitors, partners,and customers).
In our study, we developed a tool for organizational culture assessment
(TOCA) drawing from both models. We used three dimensions (consistency,
adaptability, and involvement) mainly from Denison model; we used CVF to
form a fourth dimension of “orientation,” which reflects the extent to which
the organization focuses on external expectations of stakeholders. We also
added the elements of “internal regulations and rules” and “cost control” to
the dimension of consistency, to capture salient issues for Chinese public
Different Perceptions of Hospital Culture
A subcultureis asubset ofan organization’smemberswho identify themselves
as a distinct group within the organization and who routinely take action on
the basis of their unique collective understandings (Hatch and Cunliffe 2006).
Subcultures may form within a hospital among employees who have similar
interests, who share professional, gendered, occupational identities, or who
interact more due to shared territory or equipment. For example, in U.S.
Hospital CultureandHospital Performance2141
hospitals, different types of employees have different perceptions of organiza-
tional patient safety climate (Thomas, Sexton, and Helmreich 2003; Hart-
mann et al. 2008; Singer et al. 2009), with senior managers having more
positive perceptions than either frontline workers or supervisors (Singer et al.
2008). As managers, physicians, health technicians, nurses, and other employ-
ent environments, they may represent different subcultures within a hospital
with different perceptions of the organizational culture. On the basis of find-
ings of Singer et al. (2008), we propose that a key factor in determining per-
ceptions of organization culture is the extent to which employees interact with
patients. We hypothesize that managers’ perceptions of organizational culture
will differ from those of frontline workers who interact directly with patients.
Relationship betweenHospital Cultureand Hospital Performance
While both managers and academic researchers believe that organizational
culture can influence performance (Kreitner and Kinicki 2008), studies of the
correlation between organizational culture and organizational performance
do not produce consistent results (Damanpour 1992; Denison, Haaland, and
Goelzer 2004; Kreitner and Kinicki 2008). In the health care field, studies
have analyzed different indicators of performance, such as quality improve-
ment activities, patient-care quality and efficiency, effectiveness of provider
teams, health care provider job satisfaction, and patient satisfaction, making it
difficult to identify consistent relationships across studies (Coeling and Wilcox
1988; Platonova et al. 2006; Williams et al. Konrad 2007; Zazzali et al. 2007).
In addition, a vast majority of literature on the organizational culture of hospi-
tals examines the United States or other high-income countries. Little is
known about hospital organizational culture in countries with different socio-
economic and cultural environments (Helfrichet al. 2007).
We analyze the relationship between organizational culture and four
types of performance indicators, which encompass key concerns of policy
makers and the public regarding hospital behavior. The indicators include
resource use per patient (length of stay [LOS]), productivity in resource use
(outpatient visits per physician per day [OVPPPD], bed days per physician
per day [BDPPPD]), short-term profitability, patient satisfaction with medical
care, and employee satisfaction.
When examining the relationship between culture and performance,
we develop hypotheses based on a subset of the sub-dimensions of culture
within each of the dimensions we identify above (see Appendix SA2 for a
2142 HSR: Health Services Research 46:6, PartII (December 2011)
list of the dimensions and sub-dimensions). We develop hypotheses based
on sub-dimensions, rather than on dimensions, because the different
sub-dimensions may have different relationships with specific performance
measures. The sub-dimensions within a dimension, however, are highly
correlated by construction. Thus, in empirical models, we drop one sub-
dimension from each dimension we analyze, and our method for choosing
the dropped sub-dimensions is discussed in the data analysis section.
Finally, our performance measures encompass only a subset of possible
hospital performance indicators. Thus, we identify hypotheses only for the
subset of the dimensions of culture we assess for which we have strong a
priori hypotheses regarding their effects on the available performance mea-
sures. We hypothesize that the following relationships exist between specific
aspects of culture and these four types of indicators of organizational perfor-
mance (also see Table 1).
Orientation. A hospital with a culture emphasizing social responsibility will
put the interests of society ahead of those of individual hospitals or patients.
Public hospitals in China have relatively high occupancy rates (90.0 percent
in average in 2010) and relatively long LOS (10.7 days in average in 2010)
(Chinese Ministry of Health 2011). And the perception exists that capacity
constraints prevent many people who need treatment from receiving it.
Thus, the notion of social responsibility in this context refers to reducing
LOS for individual patients to provide access for more patients. While the
possibility exists that this may not be in the social interest due to negative
effects of shorter stays on quality of care, because LOS is unusually long in
China relative to other countries, we believe that this type of unintended
effect is unlikely.
We propose that hospitals compete based on profitability, which is dri-
ven by the volume of profitable services they provide as they are paid by fee-
for-service. Thus, we hypothesize that, hospitals with cultures emphasizing
competition will use resources more productively, resulting in more outpa-
tient visitsand BDPPPD,and willbe more profitable.
Consistency. We hypothesize that a culture emphasizing cooperation among
employees will be associated with a greater employee satisfaction. A strong
culture of internal rules and regulations, in contrast, will be associated with
lower levels of employee satisfaction. Theoretically, a consistency culture will
Hospital Culture andHospital Performance2143
enable an organization to make consistent efforts to reach its goals. The
TOCA allows us to measure the extent to which the culture is consistent with
respect to the goal of cost containment, but not other goals. We hypothesize
that a hospital with a culture of cost containment will have shorter LOS, more
OVPPPD, and more BDPPPD as cost containment goals create pressure to
use resources more efficiently. This, in turn, will lead to a greater short-term
profitability but lowerpatient satisfaction.
Involvement. Involvement cultures emphasize the development of organiza-
tional manpower. Consistent with other research demonstrating a positive
Hypotheses of the Relationship between Hospital Culture and
LOS OVPPPDBDPPPD ROIOE
Note. Cell entry indicates the direction of change in the performance measure associated with an
increase in the strength of the culture measure. The table includes a subset of the sub-dimensions
measured in the TOCA. We did not develop hypotheses for the sub-dimensions of sustainable
earity in empirical models. See the data analysis section for a discussion. We did not have any
hypotheses for the relationship between organizational learning and the available performance
measures.Thus,thesesub-dimensionsarenot includedin Table 1.
BDPPPD, bed days per physician per day; ESOHDR5Y, employee satisfaction; LOS, length of
stay; OVPPPD, outpatient visits per physician per day; ROIOE, ratio of operational income over
2144HSR: Health Services Research 46:6, PartII (December 2011)
association between involvement cultures and employee satisfaction and
greater efficiency in the delivery of medical care (Platonova et al. 2006), we
hypothesize that employee satisfaction will be greater in public hospitals with
culturesemphasizing capability developmentand empowerment.
Adaptability. Organizations with a culture of adaptability can make timely
adjustments to strategic objectives in response to changes in the external envi-
ronment (Zhang, Li, and Pan 2009). While public hospitals with more adapt-
able cultures will have better performance as a result, the effect on indicators
of performance depends on the hospital objectives. As we do not observe the
objectives of hospitals, our hypotheses are limited to specific aspects of adapt-
ability.Wehypothesize that organizationswith a culture of customer focus will
have higher levels of patient satisfaction, as well as longer LOS and fewer
OVPPPD,as employees place a greater focus on patientcare.
The primary data sources are surveys of 93 public hospitals, their employees,
and their patients in Shanghai, Hubei Province, and Gansu Province con-
ducted between June and October of 2009. The selection of regions and the
sampling of hospitals within regions were designed to capture varying levels
of socioeconomic status within China. We first selected three provinces repre-
senting high, middle, and low levels of socioeconomic status. We then selected
three districts or prefecture-level cities representing high, middle, and low lev-
els of socioeconomic status within each province. Finally, we randomly
selected three to four tertiary hospitals, three to four secondary hospitals, and
three to four community hospitals in each district or city. In Shanghai, nine
tertiary general hospitals were selected from the region as a whole (because
tertiary general hospitals are distributed very unequally among the districts).
In the hospital survey, we collected measures of hospital performance that are
routinely reported to the government, including LOS, outpatient visits per
year, bed days per year, number of physicians in the hospital, annual hospital
operational income, and annual hospital operational expense.
Employee and patient surveys were administered in each hospital using
paper-based questionnaires. For the employee survey,10 percent of managers
(at least 10 managers) and 10 percent of physicians, nurses, and health
Hospital Culture andHospital Performance 2145
technicians (at least 30 in each group) were randomly selected to receive a sur-
vey in the secondary-level and tertiary general hospitals, and 50 percent of the
managers, 10 physicians, 5 nurses, and 5 health technicians were randomly
selected to receive a survey in community hospitals. If this algorithm resulted
in fewer than 20 people surveyed in a community hospital, then all employees
in the community hospital were selected for the survey. In this study, “man-
ager” refers to employees with management responsibilities at top and middle
levels, includingphysician-managers, nurse-managers, andtechnician-manag-
ers. Frontline workers are employees without management responsibilities
who interact directly with patients.
In their survey, employees evaluated 80 statements regarding the
organization’s culture. The rating scale was 1 (fully disagree), 2 (essentially
disagree), 3 (partially disagree), 4 (partially agree), 5 (essentially agree), and
6 (fully agree). When the data were analyzed, the rating scores of the state-
ments that were phrased negatively were reversed so that a higher score
represents a view that the culture is stronger along a particular dimension.
The employee survey also included questions about employee characteris-
tics and satisfaction with the overall hospital development in the most
recent 5 years. The rating scale for the satisfaction question was 1 (very dis-
satisfied), 2 (dissatisfied), 3 (relatively dissatisfied), 4 (relatively satisfied), 5
(satisfied), and 6 (very satisfied). All responses to the employee survey were
For the patient survey, 50 patients treated in the outpatient setting and
50 patients admitted to each hospital were randomly selected to receive an
anonymous questionnaire. The scale for the question for overall satisfaction
with medical care provided in the hospital was the same as that for employee
Organizational Culture. We used the TOCA to develop measures of orga-
nizational culture. The TOCA included 80 items, grouped into four
dimensions, including orientation, consistency, involvement, and adapt-
ability, and 13 sub-dimensions (see Appendix SA2 for a sample question,
translated from Mandarin, from each sub-dimension). We consulted with
experts of hospital management in developing questions and adjusted
some questions based on the results of pilot tests. Using factor analyses,
we developed measures of organizational culture from the items on the
TOCA. The scores were calculated according to the framework of the
2146 HSR: Health Services Research 46:6, PartII (December 2011)
TOCA and were weighted according to the loadings of the first eigenvec-
tor on the dimensions of organizational culture in principal component
We conducted item analysis (item correlation and Cronbach’s alpha),
exploratory factor analysis (principal factor analysis with rotate = promax),
and confirmatory factor analysis (structural equation model) to test the reli-
ability and validity of the TOCA (Hoyle 1995; Byrne 2001; Grembowski
2001; Arbuckle 2003; Cole, Ciesla, and Steiger 2007). In confirmatory factor
analysis, we used modification indices (MIs) to modify the model and used the
fitness indices to select the best model from alternative models. Based on these
analyses, four dimensions based on 73 items were ultimately included in the
TOCA (see Appendix SA3). Orientation (F1) included the sub-dimensions of
social responsibility (F11), sense of competition (F12), and sustainable devel-
opment (F13); consistency (F2) included the sub-dimensions of core values
(F21), internal regulations and rules (F22), cooperation (F23), and cost control
(F24); involvement (F3) included sub-dimensions of capability development
(F31), team orientation (F32), and empowerment (F33); and adaptability (F4)
included the sub-dimensions of creating change (F41), organizational learning
(F42), and customerfocus (F43). The item scoreswere correlated withthe total
score (correlation coefficients ranged from 0.43 to 0.81) and were also corre-
lated with the related dimension score (the correlation coefficients ranged
from 0.59to 0.85).
The analysis of TOCA’s structural equation model using a randomly
assigned calibration sample (n = 1,718) showed that the root-mean-square
error of approximation (RMSEA) = 0.053, the standardized root mean
square residual (SRMR) = 0.052, the normed fit index (NFI), the incremental
fit index (IFI), non-normed fit index (NNFI), and the comparative fit index
(CFI) were greater than 0.85, and that the goodness-of-fit index (GFI), the
adjusted goodness-of-fit (AGFI), and the parsimony goodness-of-fit (PGFI)
were 0.760, 0.745, and 0.714, respectively. The analysis of TOCA’s structural
equation model by using a randomly assigned validation sample (n = 1,719)
showed that RMSEA = 0.052 and SRMR = 0.051, that NFI, IFI, NNFI, CFI
were all greater than 0.85, and that GFI, AGFI, and PGFI were 0.769, 0.754,
and 0.722, respectively. The TOCA adequately satisfied standard tests of
goodness of fit (Janssen, Jonge, and Bakker 1999; Henderson, Donatelle, and
Acock2002; Hau,Wen, and Cheng 2004).
Assessments of within-group agreement are required to determine
whether aggregated individual-level scores can be used as indicators of
group-level constructs (Dunlap, Burke, and Smith-Crowe 2003). Four
Hospital Cultureand Hospital Performance2147
complementary measures, ICC(1), ICC(2), rwg(j), and the F-statistic from a
one-way analysis of variance (ANOVA), are frequently used to justify statistically
the aggregation (Zohar and Luria 2005; Vogus and Sutcliffe 2007). Intraclass
correlation coefficients (ICCand ICC) measure homogeneity within the
group (values of the former between 0.05 and 0.30, and values of the later
equal to or above 0.7 are acceptable). R measures the degree to which individ-
ual responses within a group are interchangeable (values of 0.7 or greater are
acceptable). A significant F-statistic resulting from a one-way ANOVA with
groupmembership asindependentvariabledemonstrates differencesbetween
the groups (Vogus and Sutcliffe 2007). Based on the results of these tests, the
measures of organizational culture constructed in this study were character-
ized by high homogeneity within and high heterogeneity between the hospi-
tals (see Table 2).1
Hospital Performance. Six indicators were used to measure hospital perfor-
mance, including LOS, OVPPPD, BDPPPD, ratio of operational income
over operational expenditure (ROIOE), patient satisfaction, and employee
Table 2: Test of Homogeneity of Culture within Hospitals
CultureICC(1) ICC(2)FValue(One-Way ANOVA)rwg(j)Index
***p < .001.
2148HSR: Health Services Research 46:6, PartII (December 2011)
satisfaction with overall hospital development in recent 5 years (ESO-
We calculated means of the factor scores for the dimensions and the sub-
dimensions of organizational culture both overall and by type of employee
(manager, physician, nurse, and others). Analysis of variance was used to ana-
groups of employees.
We estimated mixed linear models using restricted maximum likelihood
to analyze the fixed effect of job type on employee perception of organiza-
tional culture, controlling for other employee characteristics and for hospital
random effects. We restricted these models to the total score and the four
dimensions of organizational culture as little difference existed across the sub-
dimensions of aparticular dimension.
We estimated separate hospital-level multinomial logistic regressions for
each of the six indicators of hospital performance to analyze the relationship
between organizational cultureand hospital performance. The dependentvar-
iable for each model was a three-level indicator of relative performance (less
than the 25th percentile, greater than or equal to the 25th percentile and less
than the 75th percentile, and greater than or equal to the 75th percentile). The
independent variables for each model were nine sub-dimensions of organiza-
tional culture. We dropped four sub-dimensions due to the existence of multi-
collinearity among sub-dimensions. The sub-dimensions that had the highest
variance inflation factor (VIF) were dropped one by one until all the VIFs of
sub-dimensions <10 (using “PROC REG” with the option of “VIF” in SAS).
The dropped sub-dimensions included sustainable development, core values,
team orientation, and creating change. Although we had no hypotheses for
the sub-dimension of organization learning, we included it in the model as a
control variable. These models also included controls for hospital type (ter-
tiary, secondary, and community) and location (province).
Characteristics of Surveyed Hospitals, Employees, and Patients
Eighty-seven hospitals (93.55 percent of 93 sampled hospitals) participated in
the survey. Twenty-nine (33.33 percent) were tertiary general hospitals, 28
Hospital Culture andHospital Performance2149
(32.19 percent) were secondary-level general hospitals, and 30 (34.48 percent)
were community hospitals. Hospitals from Shanghai, Gansu Province, and
Hubei Province accounted for 37.93, 29.89, and 32.18 percent, respectively, of
Atotal of3,437hospitalemployeesparticipated inthesurvey(75.69per-
cent respondent rate); 52.87 percent of employee respondents were from ter-
tiary general hospitals, 31.48 percent from secondary-level general hospitals,
and 15.65 percent from community hospitals. A total of 22.84 percent were
managers, 31.62 percent were physicians, 27.14 percent were nurses, and
18.33 percent were other types of employees. The average age was
35.99 years and 37.65 percent were male. In all, 15.33 percent had master
and/or Ph.D. degrees, and 42.03 percent had worked at the hospital for
15 years or more. A total of 3,245 employees of 87 hospitals had no missing
data for the questions on organizational culture.
A total of 8,276 patients from 87 hospitals participated in the patient sur-
vey with 35.33 percent from tertiary, 33.11 percent from secondary, and 31.56
percent from the community hospitals; 48.36 percent of the patients were
male and 49.43 percent received care in the outpatient setting. The response
rate for the patient survey was 95.13 percent.2The mean and standard devia-
tion of patient age were 47.32 and 19.39,respectively.
Employee Perceptions of Hospital Culture
Overall, employees perceived the organizational culture as strong along most
dimensions (mean of the total score was 4.75, corresponding to a response
between partially and essentially agree) (see Table 3). The orientation dimen-
sion had the highest mean score (5.03) and involvement had the lowest (4.54).
Among the sub-dimensions, internal regulations and rules received the high-
est mean score (mean = 5.25), while empowerment received the lowest
(mean = 4.27).
Differences existed among the different types of employees in their rat-
ings of each sub-dimension of organizational culture except organizational
learning (see Table 3). On each measure, managers gave the highest ratings.
The analyses using mixed linear models showed that job type was highly cor-
related with perceptions of organizational culture after controlling for other
employee characteristics and hospital-level random effects (see Table 4). Con-
sistent with the unadjusted results, in the multivariate models, managers gave
2150 HSR: Health Services Research 46:6, PartII (December 2011)
Factor Scores of Hospital Culture Overall and by Type of Employee
(No. = 3,245)
(No. = 727)
(No. = 993)
(No. = 855)
(No. = 578)
†If the groupsaremarkedwiththesame letter, theirfactor means donot differstatistically by usingtheStudent-Newman-Keuls multiplerangetest.
***p < .001,
**p < .01,
*p < .05.
Hospital CultureandHospital Performance2151
Analysis of the Relationship between Employee JobTypes and Perceptions of Organizational Culture
UN(1,1)‡subject = hospital
Z = 5.33***
Z = 5.37***
Z = 5.22***
Z = 5.12***
Z = 5.21***
***p < .001,
**p < .01,
*p < .05.
†A total of 2,907 employees in 85 hospitals were included in mixed linear model. The models include controls for the fixed effects of employees’
characteristics (education, gender, age, working year) and hospital characteristics (level and location); the manager group is used as comparison group
that isindicated in parentheses.
2152HSR: Health Services Research 46:6, PartII (December 2011)
Table 5: Download full-text
ture and Hospital Performance
Logistic Analyses of the Relationship between Organizational Cul-
Note.Models estimated usinghospital-level multinomiallogistic regressions.
***p < .001,
**p < .01,
*p < .05.
†Themodelsinclude controlsforhospitallevel andlocation.
BDPPPD, bed days per physician per day; ESOHDR5Y, employee satisfaction with hospital
development in recent 5 years; LOS, length of stay; OVPPPD, outpatient visits per physician per
day;ROIOE,ratio ofoperational incomeoveroperationalexpense.
Hospital Culture andHospital Performance2153