The relationship between patient safety culture and adverse events: A questionnaire survey

Thev

XuLi-a Deb Schc Schd Sche De

International Journal of Nursing Studies xxx (2014) xxx–xxx

A R

Artic

Rece

Rece

Acce

Keyw

Adv

Pati

Pati

Chin

Nur

*

G Model

NS-2333; No. of Pages 9

Plq

002

http

e relationship between patient safety culture and adverseents: A questionnaire survey

e Wang a, Ke Liu b,*, Li-ming You b, Jia-gen Xiang c, Hua-gang Hu d,feng Zhang b, Jing Zheng b, Xiao-wen Zhu e

partment of Orthopaedics, The First Affiliated Hospital of Chongqing Medical University, China

ool of Nursing, Sun Yat-sen University, China

ool of Nursing, Guangzhou University of Chinese Medicine, China

ool of Nursing, Soochow University, China

partment of Nursing, School of Medicine, Jinan University, China

T I C L E I N F O

le history:

ived 17 April 2013

ived in revised form 10 December 2013

pted 18 December 2013

ords:

erse events

ent safety

ent safety culture

ese hospital

sing

A B S T R A C T

Background: Patient safety culture is an important factor in the effort to reduce adverse

events in the hospital and improve patient safety. A few studies have shown the

relationship between patient safety culture and adverse events, yet no such research has

been reported in China.

Objectives: This study aimed to describe nurses’ perception of patient safety culture and

frequencies of adverse events, and examine the relationship between them.

Design: This study was a descriptive, correlated study.

Setting and participants: We selected 28 inpatient units and emergency departments in 7

level-3 general hospitals from 5 districts in Guangzhou, China, and we surveyed 463

nurses.

Methods: The Hospital Survey on Patient Safety Culture was used to measure nurses’

perception of patient safety culture, and the frequencies of adverse events which

happened frequently in hospital were estimated by nurses. We used multiple logistic

regression models to examine the relationship between patient safety culture scores and

estimated frequencies of each type of adverse event.

Results: The Positive Response Rates of 12 dimensions of the Hospital Survey on Patient

Safety Culture varied from 23.6% to 89.7%. There were 47.8–75.6% nurses who estimated

that these adverse events had happened in the past year. After controlling for all nurse

related factors, a higher mean score of ‘‘Organizational Learning-Continuous Improve-

ment’’ was significantly related to lower the occurrence of pressure ulcers (OR = 0.249),

prolonged physical restraint (OR = 0.406), and complaints (OR = 0.369); a higher mean

score of ‘‘Frequency of Event Reporting’’ was significantly related to lower the occurrence

of medicine errors (OR = 0.699) and pressure ulcers (OR = 0.639).

Conclusions: The results confirmed the hypothesis that an improvement in patient safety

culture was related to a decrease in the occurrence of adverse events.

� 2013 Elsevier Ltd. All rights reserved.

Corresponding author at: School of Nursing, Sun Yat-sen University, 74 Zhongshan 2nd Road, Guangzhou 510089, China. Tel.: +86 20 87334850.

E-mail address: [email protected] (K. Liu).

Contents lists available at ScienceDirect

International Journal of Nursing Studies

journal homepage: www.elsevier.com/ijns

ease cite this article in press as: Wang, X., et al., The relationship between patient safety culture and adverse events: Auestionnaire survey. Int. J. Nurs. Stud. (2014), http://dx.doi.org/10.1016/j.ijnurstu.2013.12.007

0-7489/$ – see front matter � 2013 Elsevier Ltd. All rights reserved.

://dx.doi.org/10.1016/j.ijnurstu.2013.12.007

http://dx.doi.org/10.1016/j.ijnurstu.2013.12.007

mailto:[email protected]


http://www.sciencedirect.com/science/journal/00207489


X. Wang et al. / International Journal of Nursing Studies xxx (2014) xxx–xxx2

G Model


What is already known about the topic?

� Theoretically, patient safety culture (PSC) is regarded asan important factor in the prevention of adverse events(AEs) in healthcare settings, but the relationship lacksquantitative evidence.� Few studies have examined the relationship between

PSC and AEs.� Although some studies have investigated the rates of

some AEs and nurses’ perception of PSC, no study hadexamined the relationship between them in Chinesehospitals.

What this paper adds

� Nurses’ perception of PSC was not satisfactory and thenurse-estimated AEs were high in Chinese hospitals.� Improvement in nurses’ perception of PSC was related to

a decrease in the occurrence of patient AEs.

1. Introduction

1.1. Background

The Institute of Medicine (IOM) emphasized that it wasimportant for healthcare organizations to establish a safetyculture to ensure that patients were not inadvertentlyharmed by errors in the care which was supposed to healthem (Stefl, 2001). PSC in healthcare organizations,specifically hospitals, includes communication foundedon mutual trust, good information flow, shared perceptionof the importance of safety, organizational learning,commitment from management and leadership, and thepresence of a non-punitive approach to incident and errorreporting (Sanders and Cook, 2007). AEs have become aglobal problem; they are an important indicator of patientsafety (Baker et al., 2004). The IOM pointed out thatpreventable AEs happen not only due to individual factorssuch as inadequate skills or knowledge of nurses ordoctors; it claimed that system errors due to problems inareas such as management, work environment and staffingare a more important aspect of preventable AEs (Stefl,2001). It is therefore urgent to build a safety system, andbuilding a safety culture is the first step towards it(Maurette, 2002).

There are several methods of collection of AEs data,such as reviews of medical or nursing records, directobservation, reporting systems, nurses’ estimates, patientinterviews and so on (Cina-Tschumi et al., 2009; Flynnet al., 2002; Olsen et al., 2007). Each method has itsstrengths and weaknesses. For example, a review ofmedical or nursing records and direct observation canprovide accurate information of antecedents and out-comes of AEs, but implementing it may cost more time andstaffing (Flynn et al., 2002). A reporting system is anaccepted useful method to collect information about AEs,but the high missed report rate is its weakness (Grenier-Sennelier et al., 2002). Many hospitals have a reportingsystem, but most of them do not work well in China (Daiet al., 2009). Nurses’ estimates can not only collect mass ofdata in a short amount of time with little money andstaffing, but also collect more accurate information

without fear of punishment (Stratton et al., 2004).Although this method of data collection may be subjectto respondent bias and recall bias (Manojlovich andDeCicco, 2007), Cina-Tschumi et al. (2009) showed thatnurses’ estimated ‘‘patient fall’’ frequencies over theperiod of one year were concordant with continuously andsystematically assessed data and more accurate than thelatter over one month. This method has showed useful-ness in other studies (Aiken et al., 2001; Sochalski, 2004).For these reasons, we have chosen to use nurses’ estimatesto collect data about the frequencies of AEs over a period ofone year.

It has been suggested that establishing a good safetyculture can help prevent an error chain from causing a realerror (Reason, 1995), but the relationship between PSC andAEs lacks quantitative evidence. Some researches (Hansenet al., 2011; Singer et al., 2009; Zohar et al., 2007) foundthat PSC was negatively correlated with some AEs(medicine error, acute myocardial infarction, heart failureor Patient Safety Indicators which monitored AEs). Anotherstudy (Ausserhofer et al., 2012) found that none of the AEsit looked at (medication administration error, pressureulcer, patient fall, urinary tract infection, bloodstreaminfection and pneumonia) was significantly correlatedwith PSC, while rationing of nursing care was significantlycorrelated with patient satisfaction, medicine errors,bloodstream infection and pneumonia; this indicated thatnurse-related organizational factors may have affectedAEs. Studies that examined the relationship between PSCand a single type of AE employed controls for nurse-relatedfactors such as staffing ratio and education (Hansen et al.,2011; Vogus and Sutcliffe, 2007). We deduced from thisthat for our examination of the relationship between PSCand AEs, we would need to take account of nurse-relatedfactors.

To our knowledge, a few researches have beenconducted to explore the relationship between PSC andAEs, yet none examined the relationship between PSC andAEs in China.

1.2. Objectives

The objectives of this study were to describe nurses’perception of PSC and their estimate of the frequencies ofAEs and to examine the relationship between PSC and AEsin Chinese hospitals.

2. Methods

2.1. Design

This study was a descriptive, correlated study.

2.2. Sample and setting

Eight out of Guangzhou’s twelve districts are describedas main urban areas due to high population density. Basedon stratified sampling, we sampled seven level-3 hospitals(the large, high-tech hospitals with 1000–2500 beds) infive main urban areas of Guangzhou, covering 1 in 3 ofGuangzhou’s total count of level-3 hospitals and 5 in

Please cite this article in press as: Wang, X., et al., The relationship between patient safety culture and adverse events: Aquestionnaire survey. Int. J. Nurs. Stud. (2014), http://dx.doi.org/10.1016/j.ijnurstu.2013.12.007


8 ounideptotaunicrithadTotnurque

2.3.

and

by

(Someuse(‘‘st(‘‘ncou(PR‘‘strcalctheacrof othisHSOandtest(Xia

freqnurFlyprefor

(SWor tweyeamo‘‘evesti

inccurwo

2.4.

anaperdatteststatcoe

X. Wang et al. / International Journal of Nursing Studies xxx (2014) xxx–xxx 3

G Model


Plq

f all its main urban areas. One medical unit, one surgicalt, one Intensive Care Unit (ICU) and one emergencyartment from each hospital were selected, 28 units inl. The study surveyed all of the nurses working in these

ts from October 2010 to January 2011. The inclusioneria of nurse were as follows: registered nurses who

worked at least one year in the current hospital.ally, 640 questionnaires were delivered, 539 (84.2%)ses returned the questionnaires and 463 (72.3%)stionnaires were valid.

Measuring instrument

The Hospital Survey on Patient Safety Culture (HSOPSC) AE questionnaires were used to collect data.

The HSOPSC was pilot tested, revised and then releasedthe Agency for Healthcare Research and Quality (AHRQ)rra and Nieva, 2004). The HSOPSC contains 42 items toasure 12 patient safety culture dimensions. The itemsd 5-point Likert response scales of agreementrongly disagree’’ to ‘‘strongly agree’’) or frequencyever’’ to ‘‘always’’), so the mean score of each dimensionld be calculated. In addition, a Positive Response RateR) could be calculated for each item from responses ofongly agree/agree’’ or ‘‘always/most of the time’’. Toulate the PRR of one dimension, first step is to compute

PRR for each item and then calculate the mean PRRoss all items of this dimension. Similarly, the mean PRRsverall HSOPSC can be calculated. If the PRR is over 75%,

dimension is defined as ‘‘patient safety strength’’. ThePSC was translated into Chinese by this research team

content validity was established, and the reliability showed a good internal consistency: Cronbach’s ang et al., 2012) was 0.853.

We investigated the following 7 AEs which occurreduently in hospitals and considered to be sensitive tosing care as the indicators to report (Aiken et al., 2001;

nn et al., 2002; Yang et al., 2010): medicine error (ME),ssure ulcer (PU), patient falls (PF), physical restraintsmore than 8 h (PR � 8 h), surgical wound infectionI), infusion or transfusion reaction (IR/TR) and patients

heir family complaints (PC/FC). The frequencies of AEsre scored as ‘‘never happen = 0’’, ‘‘several times ar = 1’’, ‘‘once a month or less = 2’’, ‘‘several times anth = 3’’, ‘‘once a week = 4’’, ‘‘several times a week = 5’’,eryday = 6’’ in the past year using a 7-level rating scalemated by nurses.Additionally, nurses’ demographic data were collected,luding gander, age, the highest education in nursing,rent work unit, hospital working years and hoursrked per week.

Statistical analysis

We used SPSS (Version 17.0) to perform the statisticallysis. Descriptive statistics, including frequencies,centages and tables were used for nurses’ demographica, the HSOPSC and AEs. We used a series of Chi-squares to confirm that the samples came from a sameistical mass. We calculated intra-class correlationfficient (ICC) both within unit and hospital and in the

same unit of different hospitals to examine the validity ofthis method of measuring AEs.

To examine the relationship between PSC and AEs, weused bivariate regression models for each AE and 12HSOPSC dimensions at first; then we performed multiplelogistic regression models with one type of AE as thedependent variable and 12 HSOPSC dimensions asindependent variables with control for all nurse relatedfactors. The standard of significance was set at a < 0.05.Considering that the main purpose of our research was toexplore whether PSC levels related to rates of AE’s, wechecked that no data was lost after collapsing the reports ofAEs into a binary none/some variable. In those models, wecombined 7 categories of frequencies of AEs into abinomial variable named ‘‘never happened = 0’’ (responded‘‘never happened’’) and ‘‘had happened = 1’’ (responses of‘‘several times a year’’, ‘‘once a month or less’’, ‘‘severaltimes a month’’, ‘‘once a week’’, ‘‘several times a week’’ or‘‘everyday’’).

Since the explanatory and response variables camefrom the same investigated nurses, we performed split halfanalysis (Van den Heede et al., 2013) to test our findings.Half of the random sample from all the participating nurseswas used to recreate HSOPSC dimensions as the explana-tory variables, whereas the other half of the sample wasused to recreate the AEs as the response variables. Weconducted the split half analysis five times using SPSS(Version 17.0). These results showed negligible differencefrom the total sample results which reinforced ourfindings. Therefore, it was used in both of bivariateregression models and multiple logistic regression models.

2.5. Ethical considerations

The research proposal was approved by EthicalCommittee of School of Nursing, Sun Yat-sen University.Informed consents were obtained from all sampledhospitals and nurses. The survey was strictly anonymousto ensure nurses’ privacy.

3. Results

3.1. Participants

There were 463 nurses in our sample and they camefrom a same statistical mass as confirmed by Chi-squaretests. Most of the nurses were female (95.1%). They werebetween 21 and 53 years old, and the mean age was30.67 � 5.92. More than half of them (57.0%) were in theirtwenties. More than half of nurses (54.1%) did not hold abaccalaureate degree. These nurses worked in medical units(17.7%), surgical units (18.4%), ICU (21.0%) and emergencydepartments (43.0%). There were 43.5% nurses working formore than 10 years in their current hospital. About 41%nurses worked over 40 h per week (see Table 1).

3.2. Patient safety culture

The PRRs and mean scores of 12 HSOPSC dimensionsand overall score were calculated (see Table 2). The PRRsranged from 23.6% to 89.7%, and mean scores ranged from




G Model


2.58 � 0.60 to 4.10 � 0.47. The overall PRR was 57.4%. ThePRR of ‘‘Organizational Learning-Continuous Improvement’’(PRR = 89.7%) was the highest followed by ‘‘Teamwork withinUnits’’ (PRR = 86.5%). The PRR of ‘‘Staffing’’ (PRR = 23.6%) wasthe lowest. The PRRs of ‘‘Non-punitive Response to Error’’,‘‘Communication Openness’’ and ‘‘Frequency of EventReporting’’ were all less than 50%.

3.3. Adverse events

To examine the validity of this method of measuringAEs, we calculated intra-class correlation coefficient (ICC)

both within unit (average measure ICC from 0.514 to 0.854,P-values <0.001) and hospital (average measure ICC from0.402 to 0.817, P-value ranged from <0.001 to 0.036) andin the same unit of different hospitals (average measureICC from 0.230 to 0.943, P-value ranged from <0.001 to0.047) which showed the concordance of reported AE ratesin the same unit.

The majority of nurses estimated that these AEsoccurred ‘‘several times a year’’, followed by ‘‘once amonth or less’’. A few nurses estimated that these AEsoccurred ‘‘once a week’’, ‘‘several times a week’’ and‘‘everyday’’. There were 3.9% nurses estimated thatPR � 8 h happened ‘‘several times a week’’ and 7.8% nursesestimated it happened ‘‘everyday’’ (see Table 3).

After combining 7 categories of frequencies of AEs into abinomial variable, the nurse-estimated AEs ‘‘had hap-pened’’ ranged from 47.8% (PF) to 75.6% (IR/TR) during thepast year, and 71.2% were for PC/FC, 67.3% for PU, 62.3% forPR � 8 h, 49.5% for SWI, and 49.1% for ME.

3.4. The relationship between PSC and AEs

As shown in Table 4, before controlling for nurse relatedfactors, some dimensions were significant predictors toAEs. For instance, ‘‘Organizational Learning-ContinuousImprovement’’, ‘‘Teamwork Within Units’’, ‘‘TeamworkAcross Hospital Units’’ and ‘‘Frequency of Event Reporting’’were significant predictors for ‘‘Surgical wound infection’’.However, after controlling for nurse related factors, thesedimensions were no longer significantly to it. On thecontrary, after controlling for nurse related factors, thesignificant relationship appeared between some other PSCdimensions and AEs.

Table 1

Demographic information of nurses.

Nurse factors Total-n (%) Mean � SD Hospital 1 Hospital 2 Hospital 3 Hospital 4 Hospital 5 Hospital 6 Hospital 7 P-values

Gender – %

Female 425(95.1) 97.1 94.3 100 92.9 96.4 90.2 95.8 0.281

Male 22(4.9) 2.9 5.7 0 7.1 3.6 9.8 4.2

Age (in years) – % 30.67 � 5.92

21–25 102(23.4) 24.6 28.6 24.7 24.2 25.7 26.7 25.7 0.136

26–30 146(33.6) 35.5 32.8 31.1 22.6 30.0 22.4 24.3

31–35 100(23.0) 20.0 20.3 22.2 32.3 23.7 24.1 22.9

36–40 63(14.5) 12.3 11.3 16.2 14.5 13.7 21.1 20.0

41–53 24(5.5) 7.5 7.0 5.8 6.5 5.9 5.7 7.1

Highest education in nursing – %

Secondary & Advanced

diploma

248(54.1) 49.3 57.4 59.2 57.6 59.3 62.1 62.5 0.238

Baccalaureate & Master

degree

202(45.9) 50.7 42.6 40.8 42.4 40.7 37.9 37.5

Current work unit – %

Medical unit 82(17.7) 21.1 14.1 14.5 20.0 21.3 19.4 13.7 0.206

Surgical unit 85(18.4) 19.7 26.8 20.0 15.7 11.5 16.1 17.8

ICU 97(21.0) 11.3 18.3 27.3 22.9 26.2 38.7 6.8

Emergency

department

199(43.0) 47.9 40.8 38.2 41.4 41.0 25.8 61.6

Hospital working years – % 9.60 � 6.90

1–10 247(56.5) 58.7 56.5 52.0 49.7 53.9 54.3 52.9 0.224

11–20 161(36.8) 32.9 32.1 38.5 40.0 37.5 35.6 38.6

21–31 29(6.6) 8.4 11.5 9.6 10.3 8.6 10.1 8.6

Hours worked per week – % 38.14 � 3.58

30–40 264(58.7) 83.3 87.7 90.8 87.3 90.6 88.6 85.4 0.205

41–60 186(41.3) 16.2 12.3 9.2 12.7 9.4 11.4 14.6

Table 2

Descriptive statistics of the HSOPSC.

HSOPSC PRRa (%) Mean � SD

Organizational Learning-Continuous

Improvementb

89.7 4.10 � 0.47

Teamwork Within Unitsb 86.5 4.09 � 0.56

Supervisor/Manager Expectations &

Action Promoting Safety

73.8 3.81 � 0.52

Feedback and Communication

About Error

69.4 3.88 � 0.67

Hospital Handoffs and Transitions 68.1 3.37 � 0.58

Hospital Management Support for

Patient Safety

59.6 3.55 � 0.67

Overall Perceptions of Safety 53.0 3.35 � 0.53

Teamwork Across Hospital Units 50.1 3.38 � 0.58

Frequency of Event Reporting 44.2 3.40 � 0.87

Communication Openness 38.5 3.18 � 0.67

Non-punitive Response to Error 32.0 2.82 � 0.70

Staffing 23.6 2.58 � 0.60

Overall 57.4 3.46 � 0.34

a Positive Response Rate (PRR).b PRR > 75% was defined as patient safety strength.



tionwit‘‘Fetal

witPro‘‘Hoof 7we‘‘Or0.7incAbo

4. D

4.1.

ImpUnires(Ala201safecoudireheluniandThecouthe

wh(AlanurwotoopatandexppatMa(tot

Tab

Nur

AE

IR

PC

PU

PR

SW

M

PF

Infu

surg


G Model


Plq

In the multiple logistic regression models, ‘‘Organiza-al Learning-Continuous Improvement’’ correlated

h 3 out of 7 AEs, ‘‘Frequency of Event Reporting’’,edback and Communication About Error’’ and ‘‘Hospi-Management Support for Patient Safety’’ correlatedh 2 out of 7 AEs, ‘‘Supervisor Expectation & Actionsmoting Safety’’, ‘‘Non-punitive Response to Error’’ andspital Handoffs and Transitions’’ correlated with 1 out

AEs. The variance of OR is from 0.249 (the odds of PUre 24.9% as large for each unit increase in the score ofganizational Learning-Continuous Improvement’’) to39 (the odds of IR/TR were 73.9% as large for each unitrease in the score of ‘‘Feedback and Communicationut Error’’).

iscussion

Patient safety culture needs to be improved

In our findings, ‘‘Organizational Learning-Continuousrovement’’ (PRR = 89.7%) and ‘‘Teamwork Withints’’ (PRR = 86.5%) were patient safety strengths; these

ults were in line with other researchers’ findingshmadi, 2010; Chen and Li, 2010; El-Jardali et al.,0). From these findings, it can be seen that patientty strengths are almost identical across differentntries. We found, through interviews with the nursingctors of the hospital we surveyed, that all hospitals

d several training courses every month (at least one pert) to help nurses to improve professional knowledge

skills and to emphasized the importance of teamwork.y indicated a firm belief that through this effort theyld create an atmosphere of learning and cooperation in

organization and strengthen patient safety.The dimension with the lowest PRR was ‘‘Staffing’’,ich was also in line with other researchers’ findingshmadi, 2010; El-Jardali et al., 2010). About 70% ofses felt there was not ‘‘enough staff to handle therkload’’ and ‘‘work in crisis mode trying to do too much,

quickly’’. As previously reported (Aiken et al., 2002),ient failure-to-rescue (deaths following complications)

mortality were correlated with increases in nurses’erience of job dissatisfaction and burnout in highient–nurse ratio hospitals. A multi-centre study ininland China found when the nurse-to-patient ratio

total number of patients who stay on the unit) increased tothe 0.5 to <0.6 category, most patient outcomes weresignificantly improved (Zhu et al., 2012). Like otherresearches (Alahmadi, 2010; El-Jardali et al., 2010),‘‘Non-punitive Response to Error’’ was another dimensionwith low PRR (second-lowest in our study). Most surveyednurses felt ‘‘the person is being written up not a problemwhen an event is reported’’ and ‘‘their mistakes are heldagainst them’’, so they preferred being silent to admittingor reporting errors. The results indicated that a punitiveresponse to error may be a major barrier to reporting errorsupon identifications.

4.2. Adverse events should be prevented

Our findings showed that nurse-reported occurrence ofAEs was high. Notably, PR � 8 h was estimated ashappening as frequently as ‘‘several times a week’’ and‘‘everyday’’. There were 27.1% nurses estimated it hap-pened ‘‘once a month or less’’ to ‘‘everyday’’ (scored 2–6)which was higher than a national survey (16.7%) (Zhu et al.,2012), and 62.3% nurses estimated that PR ‘‘had hap-pened’’, which was higher than other reports (Hofso andCoyer, 2007; Ljunggren et al., 1997). It may be the case thatthere are few standard guidelines for physical restrainingin general hospital in China, thus nurses use physicalrestrain mainly according to their personal experience. Inour findings, 71.2% nurses estimated that PC/FC ‘‘hadhappened’’, which is higher than those estimated by nursesin the USA (49.1%), Canada (43.4%) and Germany (32.6%)(Aiken et al., 2001). This indicates that the relationshipbetween healthcare providers and patients in China wasnot as harmonious as it could be. Therefore, nurses shouldimprove communication skills and hospital managersshould build a multichannel patient complaint system tohandle the PC/FC as soon as possible to reduce avoidabledisputes (Hsieh, 2009).

4.3. Improvement of patient safety culture has the potential

to decrease adverse events

To our knowledge, previous researches investigatedalmost a whole country and they analyzed the statistic athospital level (Ausserhofer et al., 2012; Mardon et al.,2010; Singer et al., 2009). We could not conduct that

le 3

se-estimated adverse events in the past year (n = 463).

Never happened n (%) Had happened n (%)

Several times a year Once a month or less Several times a month Once a week Several times a week Everyday

/TR 106 (24.4) 268 (61.8) 47 (10.8) 10 (2.3) 1 (0.2) 0 2 (0.5)

/FC 126 (28.8) 254 (58.1) 37 (8.5) 11 (2.5) 4 (0.9) 2 (0.5) 3 (0.7)

142 (32.7) 256 (59.0) 28 (6.5) 4 (0.8) 2 (0.5) 2 (0.5) 0

� 8 h 164 (37.7) 153 (35.2) 43 (9.9) 22 (5.0) 2 (0.5) 17 (3.9) 34 (7.8)

I 218 (50.5) 180 (41.7) 24 (5.6) 8 (1.8) 0 1 (0.2) 1 (0.2)

E 220 (50.9) 193 (44.7) 11 (2.5) 3 (0.7) 3 (0.7) 2 (0.5) 0

228 (52.2) 199 (45.5) 9 (2.1) 0 0 0 1 (0.2)

sion or transfusion reaction (IR/TR), patients or their family complaints (PC/FC), pressure ulcers (PU), physical restraints more than 8 h (PR � 8 h),

ical wound infection (SWI), medicine error (ME), patient falls (PF).

lysis with 7 hospitals, therefore we analyzed at
al number of nurses on all shifts on the unit divided by ana


Table 4

Bivariate and multiple logistic regression results of the relationship between PSC and AEs.

Unadjusted (bivariate) models Adjusted (multiple) models

r/OR P-values r/OR P-values

Medicine error

Supervisor Expectation & Actions Promoting Safety �0.057/0.945 0.761 0.488/1.630 0.109

Organizational Learning-Continuous Improvement �0.310/0.733 0.139 �0.578/0.561 0.117

Teamwork Within Units �0.155/0.856 0.379 0.070/1.073 0.802

Communication Openness �0.175/0.769 0.239 0.319/1.376 0.136

Feedback and Communication About Error �0.145/0.865 0.324 0.152/1.164 0.492

Non-punitive Response to Error �0.010/0.810 0.943 �0.124/0.883 0.512

Staffing �0.033/0.965 0.839 0.172/1.188 0.441

Hospital Management Support for Patient Safety �0.422/0.656 0.004* �0.673/0.510 0.006*

Teamwork Across Hospital Units �0.115/0.891 0.488 0.050/1.051 0.853

Hospital Handoffs and Transitions �0.217/0.805 0.198 �0.042/0.959 0.866

Overall Perceptions of Safety �0.255/0.775 0.165 �0.365/0.694 0.170

Frequency of Event Reporting �0.333/0.717 0.004* �0.358/0.699 0.021*

Pressure ulcer

Supervisor Expectation & Actions Promoting Safety 0.106/1.112 0.592 0.338/1.403 0.300

Organizational Learning-Continuous Improvement �0.243/0.784 0.274 �1.224/0.249 0.002*


Communication Openness �0.108/0.898 0.496 �0.162/0.850 0.493

Feedback and Communication About Error 0.068/1.070 0.665 �0.672/0.413 0.037*

Non-punitive Response to Error �0.174/0.841 0.238 �0.405/0.667 0.045*

Staffing 0.034/1.035 0.843 0.080/1.084 0.736

Hospital Management Support for Patient Safety �0.048/0.953 0.754 0.034/1.034 0.894

Teamwork Across Hospital Units 0.112/1.118 0.525 0.221/1.248 0.436

Hospital Handoffs and Transitions 0.050/1.051 0.779 �0.024/0.976 0.925

Overall Perceptions of Safety 0.272/1.313 0.167 0.269/1.309 0.333

Frequency of Event Reporting �0.285/0.752 0.021* �0.448/0.639 0.006*

Patient falls


Organizational Learning-Continuous Improvement �0.205/0.815 0.323 �0.260/0.771 0.484


Communication Openness 0.142/1.153 0.336 0.239/1.270 0.270

Feedback and Communication About Error 0.013/1.013 0.930 0.068/1.071 0.764

Non-punitive Response to Error 0.190/1.209 0.167 0.196/1.217 0.301

Staffing �0.068/0.934 0.674 �0.133/0.875 0.559

Hospital Management Support for Patient Safety �0.013/0.987 0.925 �0.124/0.883 0.609


Hospital Handoffs and Transitions 0.127/1.135 0.448 0.216/1.241 0.393


Frequency of Event Reporting �0.030/0.971 0.791 �0.208/0.812 0.188

Physical restraints for more than 8 h

Supervisor Expectation & Actions Promoting Safety 0.244/1.276 0.204 0.344/1.410 0.276


Teamwork Within Units 0.047/1.048 0.794 0.317/1.373 0.273

Communication Openness �0.379/0.684 0.015* �0.613/0.542 0.010*


Non-punitive Response to Error 0.005/1.005 0.970 �0.127/0.881 0.520

Staffing �0.181/0.834 0.277 �0.132/0.876 0.577

Hospital Management Support for Patient Safety �0.011/0.989 0.939 �0.149/0.861 0.544



Overall Perceptions of Safety 0.086/1.090 0.645 0.181/1.199 0.503

Frequency of Event Reporting �0.199/0.820 0.090 �0.282/0.754 0.069

Surgical wound infection


Organizational Learning-Continuous Improvement �0.636/0.529 0.003* �0.611/0.543 0.102

Teamwork Within Units �0.455/0.634 0.012* �0.279/0.757 0.340




Staffing �0.174/0.840 0.282 0.037/1.038 0.868

Hospital Management Support for Patient Safety �0.272/0.762 0.061 0.061/1.063 0.798

Teamwork Across Hospital Units �0.428/0.625 0.012* �0.198/0.820 0.465

Hospital Handoffs and Transitions �0.693/0.500 <0.001** �0.741/0.477 0.004*


Frequency of Event Reporting �0.262/0.770 0.022* �0.246/0.782 0.112

Infusion or transfusion reaction

Supervisor Expectation & Actions Promoting Safety �0.412/0.663 0.078 �0.023/0.977 0.951

Organizational Learning-Continuous Improvement �0.194/0.824 0.426 0.067/0.609 0.880


G Model




indsamvalcondiffanawhThebel

afteshiwemusignwaconIt iorgshi

momeoutevecoutheandnurIt istogthestra(Kim

Tab

Pa

Adju

year

*

**


G Model


Plq

ividual level. We have tested and confirmed that theple came from the same statistical mass and the

idity of nurse-estimated method which indicated thecordance of reported AEs rates in the same unit oferent hospitals. In addition, we have done a split halflysis both in bivariate and multiple logistic regressionsich were not different from the total sample analysis.se analysis methods reinforced our results therefore we

ieved it could be used in further researches.There was a marked difference in the result before andr controlling nurse-related factors, some relation-

ps vanished while others appeared. Additionally, therere a few positive correlation coefficients in theltiple logistic models, though none of them wereificant. We thought the most likely reason for these

s that some potential related factors were nottrolled, such as the unit or hospital related factors.

ndicated that further study need to explore potentialanizational factors when examination of the relation-p between PSC and AEs.According to our findings in the multiple logisticdels, ‘‘Organizational Learning-Continuous Improve-nt’’, which got the highest PRR (89.7%), could predict 3

of 7 AEs. Because organizational learning was heldry month in surveyed hospitals, nurses felt learningld prevent AEs indeed. Mardon et al. (2010) also foundre was an inverse relationship between this dimension

rates of AEs. A good learning climate could also reducesing related medication errors (Chang and Mark, 2011).

better for nurses to learn specific knowledge and skillsether before and after AEs happened, they can discuss antecedents and outcomes of AEs, and developtegies to prevent and solve problems more successfully

et al., 2007).

In our study, ‘‘Frequency of Event Reporting’’ was ininverse proportion to the rate of AEs (ME and PU). While47.8–75.6% nurses estimated AEs ‘‘had happened’’, morethan half of surveyed nurses did not report the AEs inhospital report system. Possibly because of ‘‘blameculture’’ and the attitude of ‘‘none of my business’’, nursesmay have hesitated to report errors which were not foundto avoid receiving punishment, and errors which werenothing with themselves to avoid getting into trouble. Thatis why ‘‘near-miss’’ errors as well as real errors were notfound or discussed and consequently resulted in the sameerrors occurred again. Event reporting, an essentialcomponent for achieving a learning culture, only happensin a non-punitive and just environment, where people canreport events without being blamed (Smits et al., 2008). Aresearch (Kantelhardt et al., 2011) conducted in aneurosurgical department in Germany developed coun-ter-strategies using reporting system data, reported a 12%reduction in medication-related AEs in only 5 months.Where previous researches have shown that effectivereporting systems have a relationship with reduced ratesof AEs (Kantelhardt et al., 2011), our findings indicated thatnurse’s attitude towards reporting is a key to increase thereporting rate. Therefore, we suggested that hospitalmanagers can implement a non-punitive reporting systemand reward reporting nurses to increase the reporting rate(Peng and Wang, 2012).

4.4. Limitations

First, because we collected the data in level-3hospitals, sampling bias remained possible and it wasnot representative of the whole nurse population inGuangzhou. Nevertheless, our study surveyed one

le 4 (Continued )

Unadjusted (bivariate) models Adjusted (multiple) models

r/OR P-values r/OR P-values

Teamwork Within Units �0.128/0.880 0.538 �0.051/0.950 0.883


Feedback and Communication About Error 0.208/1.232 0.218 0.553/0.739 0.041*


Staffing �0.148/0.863 0.434 0.086/1.090 0.965

Hospital Management Support for Patient Safety �0.359/0.698 0.040* �0.686/0.504 0.027*

Teamwork Across Hospital Units �0.388/0.678 0.049* �0.222/0.801 0.500

Hospital Handoffs and Transitions �0.423/0.655 0.034* �0.640/0.527 0.034*


Frequency of Event Reporting 0.066/1.069 0.614 0.257/1.293 0.166

tients or their family complaints

Supervisor Expectation & Actions Promoting Safety �0.086/0.918 0.681 0.716/0.646 0.029*


Teamwork Within Units �0.235/0.791 0.237 �0.124/0.884 0.684

Communication Openness 0.065/1.067 0.691 0.218/1.243 0.358

Feedback and Communication About Error 0.055/1.056 0.732 0.317/1.373 0.183


Staffing �0.332/0.717 0.064 �0.189/0.828 0.429

Hospital Management Support for Patient Safety �0.345/0.708 0.035* �0.481/0.618 0.065

Teamwork Across Hospital Units �0.191/0.826 0.296 0.072/1.074 0.803



Frequency of Event Reporting �0.300/0.741 0.018* �0.237/0.789 0.147

sted models were controlled nurses’ demographic factors, including gander, age, the highest education in nursing, current work unit, hospital working

s and hours worked per week. Nurse-estimated adverse events (0 = never happened, 1 = had happened).

P < 0.05.

P < 0.001.




G Model


third of level-3 hospitals in Guangzhou and thesehospitals cover five out of eight urban districts of thiscity.

Second, we used the nurses’ estimates to collect thefrequencies of AEs which might cause bias more or less.Some Chinese studies reported that about 50% AEs werenot reported through administration channel (Liu et al.,2008; Xiang et al., 2012). Furthermore, Barbara et al. foundthat nurses’ estimates of AEs over 1 year were reliable;therefore our design chose nurses to estimate thefrequencies of AEs over last 1 year as well. Also, we testedthe validity of this measure which indicated the con-cordance of reported AE rates both within unit and hospitaland in the same unit of different hospitals. We also suggestthat further studies could use various methods to collectdata at the same time.

Third, there was a possibility that unmeasured vari-ables could confound our results. We believe that we hadaccounted for key factors of nurses as covariates. Someresearchers found that other potential factors may beinvolved, such as nurse–patient ratio, which may indicatethe need to introduce controls for other factors in futurestudies.

Forth, this was a cross-sectional study which could notidentify the effects of PSC on AEs. However, the multiplelogistic regression models could predict that higher PSCscores would be related to lower incidence rates of someAEs.

5. Conclusion

We examined the relationship between PSC and AEs inlevel-3 hospitals in Guangzhou, China, and analyzed therelationship at individual level. Our study found thatnurses’ perceptions of PSC were not satisfactory, and thatproportions of nurse-estimated AEs were high. Building anon-punitive environment and developing nurse’s initia-tive to report AEs voluntarily was necessary. This studyreinforced the findings of previous studies which identi-fied that PSC was a predictor for AEs, and an improvementof PSC was related to a decrease in the occurrence ofAEs. Further study is needed to determine the generalityof these results to a large scope of hospitals, and toidentify interventions which would improve PSC so as toreduce AEs.

Conflict of interest: None of the authors has a conflict ofinterest with respect to the authorship and or publicationof this article.

Funding: This research was supported by the China Medical

Board (grant 10-021), Li-ming You, principal investigator (PI).

Ethical approval: The research project was approved by Ethics

Committee of School of Nursing, Sun Yat-sen University.

Acknowledgement

Our sincere appreciation is extended to all participatinghospitals and nurses.

References

Aiken, L.H., Clarke, S.P., Sloane, D.M., Sochalski, J., Silber, J.H., 2002.Hospital nurse staffing and patient mortality, nurse burnout, andjob dissatisfaction. JAMA 288 (16) 1987–1993.

Aiken, L.H., Clarke, S.P., Sloane, D.M., Sochalski, J.A., Busse, R., Clarke, H.,Giovannetti, P., Hunt, J., Rafferty, A.M., Shamian, J., 2001. Nurses’reports on hospital care in five countries. Health Aff. (Millwood) 20(3) 43–53.

Alahmadi, H.A., 2010. Assessment of patient safety culture in SaudiArabian hospitals. Qual. Saf. Health Care 19 (5) e17.

Ausserhofer, D., Schubert, M., Desmedt, M., Blegen, M.A., De Geest, S.,Schwendimann, R., 2012. The association of patient safety climate andnurse-related organizational factors with selected patient outcomes:a cross-sectional survey. Int. J. Nurs. Stud. 50 (2) 240–252.

Baker, G.R., Norton, P.G., Flintoft, V., Blais, R., Brown, A., Cox, J., Etchells, E.,Ghali, W.A., Hebert, P., Majumdar, S.R., O’Beirne, M., Palacios-Der-flingher, L., Reid, R.J., Sheps, S., Tamblyn, R., 2004. The CanadianAdverse Events Study: the incidence of adverse events among hospi-tal patients in Canada. CMAJ 170 (11) 1678–1686.

Chang, Y., Mark, B., 2011. Effects of learning climate and registered nursestaffing on medication errors. Nurs. Res. 60 (1) 32–39.

Chen, I.C., Li, H.H., 2010. Measuring patient safety culture in Taiwan usingthe Hospital Survey on Patient Safety Culture (HSOPSC). BMC HealthServ. Res. 10, 152.

Cina-Tschumi, B., Schubert, M., Kressig, R.W., De Geest, S., Schwendimann,R., 2009. Frequencies of falls in Swiss hospitals: concordance betweennurses’ estimates and fall incident reports. Int. J. Nurs. Stud. 46 (2)164–171.

Dai, H.S., Shi, Y., Mao, Y.F., 2009. Research status quo and thinking ofnursing adverse events reporting system. Nurs. Res. 23 (23) 2137–2138.

El-Jardali, F., Jaafar, M., Dimassi, H., Jamal, D., Hamdan, R., 2010. Thecurrent state of patient safety culture in Lebanese hospitals: a study atbaseline. Int. J. Qual. Health Care 22 (5) 386–395.

Flynn, E.A., Barker, K.N., Pepper, G.A., Bates, D.W., Mikeal, R.L., 2002.Comparison of methods for detecting medication errors in 36 hospi-tals and skilled-nursing facilities. Am. J. Health Syst. Pharm. 59 (5)436–446.

Grenier-Sennelier, C., Lombard, I., Jeny-Loeper, C., Maillet-Gouret, M.C.,Minvielle, E., 2002. Designing adverse event prevention programsusing quality management methods: the case of falls in hospital. Int. J.Qual. Health Care 14 (5) 419–426.

Hansen, L.O., Williams, M.V., Singer, S.J., 2011. Perceptions of hospitalsafety climate and incidence of readmission. Health Serv. Res. 46 (2)596–616.

Hofso, K., Coyer, F.M., 2007. Part 2. Chemical and physical restraints in themanagement of mechanically ventilated patients in the ICU: a patientperspective. Intensive Crit. Care Nurs. 23 (6) 316–322.

Hsieh, S.Y., 2009. Factors hampering the use of patient complaints toimprove quality: an exploratory study. Int. J. Nurs. Pract. 15 (6) 534–542.

Kantelhardt, P., Muller, M., Giese, A., Rohde, V., Kantelhardt, S.R., 2011.Implementation of a critical incident reporting system in a neuro-surgical department. Cent. Eur. Neurosurg. 72 (1) 15–21.

Kim, J., An, K., Kim, M.K., Yoon, S.H., 2007. Nurses’ perception of errorreporting and patient safety culture in Korea. West. J. Nurs. Res. 29 (7)827–844.

Liu, Y.L., Zhang, L., Xu, J., Wang, G.L., Zhao, G.H., Ji, X.N., Yao, S.L., 2008.Investigation on hospital safety culture among nurses. J. Nurs. Sci. 23(23) 45–47.

Ljunggren, G., Phillips, C.D., Sgadari, A., 1997. Comparisons of restraint usein nursing homes in eight countries. Age Ageing 26 (Suppl. 2) 43–47.

Manojlovich, M., DeCicco, B., 2007. Healthy work environments, nurse–physician communication, and patients’ outcomes. Am. J. Crit. Care 16(6) 536–543.

Mardon, R.E., Khanna, K., Sorra, J., Dyer, N., Famolaro, T., 2010. Exploringrelationships between hospital patient safety culture and adverseevents. J. Patient Saf. 6 (4) 226–232.

Maurette, P., 2002. To err is human: building a safer health system. Ann.Fr. Anesth. Reanim. 21 (6) 453–454.

Olsen, S., Neale, G., Schwab, K., Psaila, B., Patel, T., Chapman, E.J., Vincent,C., 2007. Hospital staff should use more than one method to detectadverse events and potential adverse events: incident reporting,pharmacist surveillance and local real-time record review may allhave a place. Qual. Saf. Health Care 16 (1) 40–44.

Peng, H., Wang, Y., 2012. Reporting system construction of medicaladverse event and patients potential safety hazard. Chinese Hosp.Manage. 32 (12) 53–54.


http://refhub.elsevier.com/S0020-7489(13)00385-4/sbref0005



































































Reas

SandSing

Smi

Soch

Sorr

Stefl

Stra


G Model


Plq

on, J., 1995. Understanding adverse events: human factors. Qual.Health Care 4 (2) 80–89.ers, J., Cook, G., 2007. ABC of Patient Safety. Blackwell, Oxford.

er, S., Lin, S., Falwell, A., Gaba, D., Baker, L., 2009. Relationship ofsafety climate and safety performance in hospitals. Health Serv. Res.44 (2 Pt 1) 399–421.ts, M., Christiaans-Dingelhoff, I., Wagner, C., Wal, G., Groenewegen,P.P., 2008. The psychometric properties of the ‘Hospital Survey onPatient Safety Culture’ in Dutch hospitals. BMC Health Serv. Res. 8,230.alski, J., 2004. Is more better?: the relationship between nurse

staffing and the quality of nursing care in hospitals. Med. Care 42(2 Suppl.) I67–I73.a, J., Nieva, V., 2004. Hospital Survey on Patient Safety Culture.Westat, Rockville, MD., M.E., 2001. To err is human: building a safer health system in 1999.Front. Health Serv. Manage. 18 (1) 1–2.tton, K.M., Blegen, M.A., Pepper, G., Vaughn, T., 2004. Reporting ofmedication errors by pediatric nurses. J. Pediatr. Nurs. 19 (6) 385–392.

Van den Heede, K., Florquin, M., Bruyneel, L., Aiken, L., Diya, L., Lesaffre, E.,Sermeus, W., 2013. Effective strategies for nurse retention in acutehospitals: a mixed method study. Int. J. Nurs. Stud. 50 (2) 185–194.

Vogus, T.J., Sutcliffe, K.M., 2007. The Safety Organizing Scale: develop-ment and validation of a behavioral measure of safety culture inhospital nursing units. Med. Care 45 (1) 46–54.

Xiang, J.G., Liu, K., You, L.M., Zhang, L.F., 2012. Nurses’ attitude towardspatient safety culture in 3A-level general hospitals in Guangzhou. J.Nurs. Admin. 12 (6) 392–394.

Yang, S., Wang, X., Shao, Y.Y., Liu, Y.S., Ying, B., Dong, J., 2010. Analysis ofthe nursing adverse events based on 335 cases from the reportingsystem. Chinese J. Nurs. 45 (2) 130–132.

Zhu, X.W., You, L.M., Zheng, J., Liu, K., Fang, J.B., Hou, S.X., Lu, M.M., Lv, A.L.,Ma, W.G., Wang, H.H., Wu, Z.J., Zhang, L.F., 2012. Nurse staffing levelsmake a difference on patient outcomes: a multisite study in Chinesehospitals. J. Nurs. Scholarship 44 (3) 266–273.

Zohar, D., Livne, Y., Tenne-Gazit, O., Admi, H., Donchin, Y., 2007. Health-care climate: a framework for measuring and improving patientsafety. Crit. Care Med. 35 (5) 1312–1317.






































The relationship between patient safety culture and adverse events: A questionnaire survey

Documents

Transcript of The relationship between patient safety culture and adverse events: A questionnaire survey