Anda di halaman 1dari 18

Journal of Safety Research

33 (2002) 259 276


www.elsevier.com/locate/jsr

Organizational safety:
Which management practices are most effective
in reducing employee injury rates?
Alison G. Vredenburgh*
Vredenburgh and Associates, Inc., PMB 353, 2588 El Camino Real, Suite F, Carlsbad, CA 92008, USA
Received 6 December 2000; received in revised form 5 August 2001; accepted 19 November 2001

Abstract
Problem: While several management practices have been cited as important components of safety
programs, how much does each incrementally contribute to injury reduction? This study examined
the degree to which six management practices frequently included in safety programs (management
commitment, rewards, communication and feedback, selection, training, and participation)
contributed to a safe work environment for hospital employees. Method: Participants were solicited
via telephone to participate in a research study concerning hospital risk management. Sixty-two
hospitals provided data concerning management practices and employee injuries. Results: Overall,
the management practices reliably predicted injury rates. A factor analysis performed on the
management practices scale resulted in the development of six factor scales. A multiple regression
performed on these factor scales found that proactive practices reliably predicted injury rates.
Remedial measures acted as a suppressor variable. Discussion: While most of the participating
hospitals implemented reactive practices (fixing problems once they have occurred), what
differentiated the hospitals with low injury rates was that they also employed proactive measures
to prevent accidents. Impact on Industry: The most effective step that hospitals can take is in the
front-end hiring and training of new personnel. They should also ensure that the risk management
position has a management-level classification. This study also demonstrated that training in itself is
not adequate. D 2002 National Safety Council and Elsevier Science Ltd. All rights reserved.
Keywords: Safety culture; Risk management; Injury reduction; Accidents; Hazards

Tel.: +1-760-434-4741; fax: +1-760-434-6029.


E-mail address: alisonv@nethere.com (A.G. Vredenburgh).

0022-4375/02/$ - see front matter D 2002 National Safety Council and Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 2 - 4 3 7 5 ( 0 2 ) 0 0 0 1 6 - 6

260

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

1. Introduction
On an average day, 17 US workers are killed and 16,000 are injured in work-related
accidents, resulting in a cost to industry of more than US$110 billion annually (Barr,
1998). This injury rate is increasing. Traditional safety efforts have focused on the
engineering aspects of safety; however, relatively few accidents (10%) are a consequence
of unsafe mechanical or physical conditions. While most on-the-job accidents and injuries
appear to result from employees unsafe acts, incidents typically are not caused by single
operator errors, but are end-events in a chain of interacting factors on several systems
levels (Wilpert, 1994). While many unsafe acts are committed, very few will penetrate an
organizations defenses to result in accident or injury (Reason, 1994).
It is becoming increasingly apparent that it is restrictive to discuss failures of large-scale
technological systems solely in terms of the technological aspects. Individuals, their
organizations, groups, and cultures are all-important factors in the design, construction,
operation, and monitoring of technological systems. Until recently, this issue has been
described in the related literature in terms of human error. While human error does
contribute to accidents, the behavioral causes of failure are often found to be far more
subtle when incidents are analyzed as part of a technological system (Pidgeon, 1991).
Many expectations are built into the current US health and safety legislation that
specifies the responsibilities of managers and employees with regard to safe working
practices. These suppositions are more likely to be fulfilled if a positive cultural attitude
toward safety exists. The costs of failure to comply with these expectations are increasing.
As workers become more educated, they are more likely to expect safer working
conditions; a more safety and environmentally conscious public is increasingly willing
to express its disapproval of companies that are perceived to behave carelessly. This public
reproach was evident during the American consumer boycott of Exxon gasoline following
the Valdez oil spill (Turner, 1991).
Researchers have found that safety performance is affected by an organizations
socially transmitted beliefs and attitudes toward safety (Ostrom, Wilhelmsen, & Kaplan,
1993). The concept of safety culture (Pidgeon, 1991) was developed as a result of the 1986
Chernobyl accident, which focused attention on the human and organizational elements
contributing to the unsafe operation of technological systems. Safety culture is an
organizations norms, beliefs, roles, attitudes, and practices concerned with minimizing
exposure of employees to workplace hazards (Turner, 1991). The goal of a safety culture is
to develop a norm in which employees are aware of the risks in their workplace and are
continually on the lookout for hazards (Ostrom et al., 1993). A safety culture motivates
and recognizes safe behavior by focusing on the attitudes and behaviors of the employees.
It is a processnot a program; it takes time to develop and requires a collective effort to
implement its many features (Barr, 1998).
Changing a companys culture is more difficult than issuing a new policy statement.
Traditional customs and practices constrain new thinking (Kletz, 1985). While many
authors on safety management attach great importance to a formal statement of a companys
safety policy, Kletz (1993) does not believe such a statement will impact a companys
accident record. He believes that the culture or common law of a company is more
influential, conveyed by such actions as a phone call from the head office immediately after

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

261

an incident, asking not if anyone was hurt, but when the plant would be back on line. In this
case, the cultural message is that production, not people or safety, is the priority.
Researchers have found a direct organizational culture performance link. According to
Siehl and Martin (1990), a strong organizational culture is one where espoused values
are consistent with behavior and where employees share the same view of the firm.
Conversely, a weak culture results when people at all levels of the hierarchy fail to share
the values espoused by management. The challenge facing organizations is to discover
how to displace existing cultural patterns where they lack an appropriate concern with
safety, and to replace them with new, self-perpetuating elements, which show a greater
degree of care. While there are many potential external influences that make it difficult to
define a strong safety culture across settings, there are many features that safety cultures
from successful organizations have in common. In order to cultivate a strong safety
culture, several measures can be taken.
Zohars (1980) study of safety climate used a factor analysis to identify climate
dimensions that could discriminate among factories based on their safety climate levels. A
few practitioners and experts (Cohen & Cleveland, 1983; Pidgeon, 1991; Turner, 1991)
described factors they believe to be prevalent in the safety culture of organizations that
have low injury rates. The variables described below are a compilation of the factors found
across several of these reports.
Six management practices have been consistently discussed in reports concerning
safety culture: (a) rewards, (b) training, (c) hiring, (d) communication/feedback, (e)
participation, and (f) management support. The objective of the current study was to
determine the extent to which these six variables predict employee injury rates.

2. Six management practices studied


2.1. Worker participation
Worker participation (or employee involvement) is a behavioral-oriented technique that
involves individuals or groups in the upward communication flow and decision-making
process within the organization. The amount of participation can range from no
participation, where the supervisor makes all decisions, to full participation, where
everyone connected with, or affected by, the decision is involved.
Employees close to the work are recognized as often being the best qualified to make
suggestions about improvements. Participative managers will solicit opinions from other
individuals or groups before making final decisions, especially for those that affect the
employees. The empowerment of employees is both a management style and attitude.
Empowering workers provides them with authority, responsibility, and accountability for
required decisions and ensures that both employees and management are involved in setting
goals and objectives. It induces employees to do their best work as individuals and as a team,
while relieving the manager to plan, monitor, lead, and mentor (Cohen & Cleveland, 1983).
In the United States, employee involvement has tended to focus on greater personal
influence on the shop floor and on a greater role in the decision-making involving the
employees daily work experience (Cohen & Cleveland, 1983).

262

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

Safety committees have become a standard feature of workplace safety programs;


however, committees themselves do not necessarily mean effective employee involvement. Committees must be given real power to implement change. The members must be
in positions where they can have a positive impact on the committees work (such as
production and engineering supervisors), and must be well trained.
Participation has been found to be a key component in successful hospital injury
prevention programs. In a 1992 study done by the Department of Veterans Affairs Medical
Center (VAMC) in New Jersey, a program was implemented to reduce lost-time injury
cases. The intervention included the involvement of all levels of employees in every phase
of the safety program. This program dramatically reduced the lost-time injury cases within
1 year of implementation (Garrett & Perry, 1996).
2.2. Safety training
In order for employees to be active participants in a safety program, they must receive
occupational safety training. A well-designed and administered training program should
emphasize safe work practices and be derived from a true assessment of need. Training
should be followed with a program based on goal-setting and performance feedback
(Cohen & Jensen, 1984). Training program assessment should verify that the safe work
practices could be demonstrated to be effective and to endure beyond cessation of
performance feedback. According to Cohen and Jensen, there should be a redefinition
of group norms sustained through informal influences such as peer modeling of desired
behaviors, continued management support of the program, and a behavior sampling
procedure specifying performance-based criteria.
Safety training provides the means for making accidents more predictable. The basic
difference between safe employees and those who frequently get hurt is that safe employees
can recognize hazards and hazardous actions and understand the consequences. To improve
the quality of safety and health for all employees, organizations should institute a
systematic, comprehensive safety and health training program for new employees, provide
a mentor for these employees, and use a buddy system to help orient new employees in the
safety and health and quality systems. They should also institute a system of continual reeducation and retraining of employees in current safety and health issues (Roughton, 1993).
Several issues affect the perception of risk levels and should be understood when
training employees in occupational safety. People tend not to use the likelihood of injury in
their judgments of product safety; rather, the severity of injury plays the foremost role in
decisions to read warnings and act cautiously (Young, Brelsford, & Wogalter, 1990).
Vredenburgh and Cohen (1995) found that the level of perceived danger increased
compliance to warnings and instructions; therefore, it is critical that all employees are
trained to identify the hazards associated with their workplace.
2.3. Hiring practices
The development of a safety culture can be facilitated if recruitment criteria for new
personnel include the selection of people who are predisposed to displaying a safetyconscious attitude in their work. If an organization fosters a safety-conscious image, the

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

263

recruitment task will be influenced because those with compatible attitudes and expectations would be more likely to seek out this company, presumably in part due to a desire
for a safe work environment (Turner, 1991).
Eckhardt (1996) states that while interviewers cannot influence whether someone is an
inherent risk-taker, they can place such applicants in jobs with a corresponding level of
high-risk tasks. Recruiters can also select candidates with a lower propensity to take risks.
While it has been a long held belief, by different authorities, that many personal
characteristics such as gender, age, stature, and body weight are possible risk factors for
work-related injuries, none of these traits have been reasonably correlated with the
occurrence of injury-producing incidents or their severity. Ones medical history, medical
examination, or spine X-rays are sometimes able to reveal some clues of potential health
risks. However, medical advice or warning regarding a career choice might be disregarded
by an applicant who needs a job (Lin & Cohen, 1984b).
2.4. Reward system
People are motivated to behave in ways that lead to desired consequences; they will
modify their behavior to conform to a cultural norm if it is perceived that compliance will
lead to a desirable outcome. Culture is learned through a connection that is made between
behaviors and consequences. Thompson and Luthans (1990) state that since organizational
culture occurs in an environment where there are multiple reinforcements and reinforcing
agents, changing an organization involves identifying the various reinforcing agents in
order to determine their effects on the change process.
A correctly designed safety-incentive program reinforces the reporting of a hazard or an
unsafe act that leads to an injury while giving bonuses for fewer lost-time accidents. A
safety incentive program must be part of a campaign that runs parallel to safety education
and training. It must be directed at the prevention of accidents, not punishment after an
accident occurs (Peavey, 1995). Informational (feedback, self-recording), social (praise,
recognition), and tangible reinforcers (trading stamps, cash bonuses) have been used as
well as nonmonetary privileges (Komaki, Barwick, & Scott, 1978).
As with any policy, the effort to develop a strong safety culture is unlikely to be
effective if the organization is not reinforcing the desired behaviors (or is rewarding
inconsistent behaviors such as speed or production rates). A well-designed incentive
program offers recognition, which can help modify behavior. A key characteristic of a
successful incentive program is that it receives a high level of visibility within the
organization. Participants must be able to comprehend what the incentive program is
designed to accomplish and how their performance will be measured (Halloran, 1996).
Simply distributing prizes and money without pairing them with a clear, consistent set of
contingencies reduces the potential to achieve the desired outcome. It may even increase
the undesired behavior, more accidents (Swearington, 1996).
2.5. Management commitment
In one of the first investigations of safety climate, Zohar (1980) found that
managements commitment to safety is a major factor affecting the success of an

264

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

organizations safety programs. This commitment can manifest itself through job training
programs, management participation in safety committees, consideration of safety in job
design, and review of the pace of work. For example, people working for a supervisor
that never mentions safety perceive that safety is unimportant; as a result, they will not
place a strong emphasis on safety (Hofmann & Stetzer, 1996). The degree to which
management values safety is expressed in its style and level of assumable risk. These
two factors are the most influential components of culture; however, safety professionals
have very little influence over these variables. When discussions of safety are conducted
in a false or insincere rhetoric, the phony statements are readily seen for what they are
(Turner, 1991).
In five plants recognized by the National Safety Council for no lost workdays, all of the
plants required advance approval by safety personnel for any changes in the design of the
work facilities. In four of the plants, the plant safety director had direct contact with the
plant manager on a daily basis (Cohen & Cleveland, 1983). The motivation to perform a
job in a safe manner is a function of both the individuals own concern with safety as well
as managements expressed concern for safety. Safety concerns must result in an
observable activity on the part of management; they must be demonstrated in their
behavior as well as their words (Hofmann, Jacobs, & Landy, 1995).
In the 1992 Veterans Hospital (VAMC) study, the guiding force behind the initiative to
reduce the number of injury cases was managements commitment, which began at the
very top management level, with the Medical Center Director. Without sincere support
from top hospital administrators, this project would not have achieved its level of success
(Garrett & Perry, 1996).
2.6. Communication and feedback
The role of feedback concerning employees performance is critical because behaviors
resulting in industrial accidents are not typically new occurrences. Their causes are deeply
rooted in past minor incidents, where damage was insignificant and workers and
bystanders were not injured (Kletz, 1993). Regular feedback on performance can be
communicated to employees through posted charts and a review of behavioral data in
safety meetings (Roughton, 1993).
The incubation model of disasters suggests that near-miss events will often differ from
actual incidents by the absence of the final trigger event and the intervention of chance.
Pidgeon (1991) states that organizations can interpret near-miss incidents as warning
signals. In some contexts, such as the aviation industry (Hall & Hecht, 1979), a high
premium is placed on the analysis and dissemination of incident data obtained on a nofault reporting basis. In the five National Safety Council award-winning plants, the
organizations had some form of employee hazard identification system in which they were
encouraged to report hazards to management (Cohen & Cleveland, 1983). In order to
encourage communication, it is important not to blame employees when accidents occur.
As managers have gained experience with the techniques used to improve quality, they
have learned the importance of improving the process of production. Many managers now
work to solve production problems upstream rather than inspecting for defects downstream (Roughton, 1993).

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

265

Consistent and forthright communication is an essential characteristic of any strong


organization. Good communication leads to trust, which is a fundamental element of
strength. In order for organizations to foster a climate where employees are alert to
hazards, they must have an appreciation of the employees and organizations tendency to
conceal and distort significant available information (Pidgeon, 1991). In order to influence
safety practices, feedback must be provided to the employees who are capable of using it.
It needs to be given to those working at the point in the process where their behavior can
effectively influence outcomes. People cannot behave in a safety-conscious manner unless
they have the authority to change their own actions to improve their work conditions. It is
illogical to ask employees to be careful if they do not have the power or discretion to avoid
hazards (Turner, 1991). Laws (1996, p. 26) writes, motivation is no big deal, you can
motivate a baboon. But if you dont back that motivation with tools, skills, training,
counseling, and leadership, then all you have is a highly frustrated, motivated ape that
cannot get the job done.

3. The hospital environment


In a study conducted by the National Institute for Occupational Safety and Health
(NIOSH), only 8% of the 3686 hospitals surveyed met all of NIOSHs basic components
of an effective occupational safety and health program for hospital employees (Lin &
Cohen, 1984a). Healthcare workers are at great risk for injury; nationally, the total lost
workday injury and illness incidence rates for hospitals (4.1) are greater than those for
private industry (3.6). Furthermore, workers in home health care (5.0) and nursing care
(8.8) are at greater risk than construction workers (4.9; Cal/OSHA, 1997).
Healthcare employees have to cope with psychological stress resulting from shift
rotations and the frequent need to work overtime. Of the 27 occupations ranked as the
highest stress, 15 were occupations found in hospitals (Lin & Cohen, 1984a). Since
hospitals never shut down, and rarely slow down, numerous hazards develop due to
fatigue from long hours, stress, rotating shift work, and changes in policy. One shift may
create a hazard (such as leaving a cart blocking an emergency exit) that is overlooked by
subsequent shifts (Johnson, 1997).
Hospital employees must face the daily traumas of life and death as well as the constant
interaction with a diverse patient population. Many people may expect that the occupational safety and health services would be conveniently accessible to hospital personnel;
however, there tends to be a lack of focus in this area shown by hospital administration,
which has received many complaints and criticisms from the affected healthcare workers
(Werdegar, 1980). A possible explanation of this deficiency is that since medical facilities
are to provide treatment and improve health, special facilities for employee care may
appear redundant (Lin & Cohen, 1984a).
The proper focus of attention to improve a hospitals quality of service, and employee
and patient safety is not on the personnel who participate in a flawed process, but the
processes that create the flaws. Hospitals that employ quality improvement methods have
been found to achieve new levels of efficiency, patient satisfaction, safety, clinical
effectiveness, and profitability (Berwick, Godfrey, & Roessner, 1990). However, due to

266

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

the health services traditional orientation toward sick care rather than health maintenance
and hazard prevention and because of the concern for the cost of safety and health
management, the implementation of an employee safety and health program has rarely
been considered a top priority by hospital administration. Hospital employees typically do
not participate in hazard management; thus, a great deal of resources are often required to
train employees in communication and problem-solving skills and in quality/measurement
techniques (NIOSH, 1983). Quality assessment often lacks the necessary evaluation tools.
It also is missing a general theory regarding the source of hazards in the complex processes
of health care. To the extent that quality measurement tools have been developed, they
tend to unveil the symptoms, not their underlying causes (Berwick et al., 1990).
Several common types of injuries to hospital employees have been recognized and
identified: strains and sprains, needle punctures, communicable diseases, toxic and
hazardous substances, dermatitis (caused by handling cleansers, medicines, antiseptics,
and solvents), and thermal burns (primarily in food service, laundry, and sterilizing areas).
Back sprains and strains are the most common injuries to hospital workers; 46% of nurses,
aides, orderlies, and attendants report back injuries, as opposed to 26% in private industry
occupations (Cal/OSHA, 1997). These injury data must be collected and reported to the
Joint Commission on Accreditation of Healthcare Organizations (JCAHO) and the
Occupational Safety and Health Administration (OSHA).
The hospital environment was selected for the current study because injury data were
thought to be readily available to use as a criterion measure and because it represents a
growing and dynamic influence within the US service industries. It was also selected
because incidence rates among various sectors of health services are at least one-third
above the average service industry rates (NIOSH, 1983).
While few people would dispute the importance of workplace safety, a review of the
literature reveals a paucity of well-controlled studies demonstrating the efficacy of
workplace safety programs. While well-controlled studies are reported in simulated work
settings, in-house safety programs are notable for their lack of systematic assessment. The
accounts reported in trade journals are primarily anecdotal.

4. Method
4.1. Participants
Participants were risk managers from 62 hospitals located in several states in the United
States. They were recruited from professional organizations for hospital risk managers as
well as direct mail and phone solicitation to hospitals. Participation was voluntary. Public,
private, and investor-owned hospitals were solicited. All participating hospitals were
medical/surgical; none were neuropsychiatric or nursing homes. Most respondents were
managers (55 or 89%); seven (11%) were not in management.
Rousseau (1990) states that research concerning culture frequently focuses upon key
informants who are identified as those possessing special or more complete knowledge
than others in the organization. Hospital risk managers were presumed to have more
complete information about their institutions risk management programs, and thus, were

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

267

solicited as participants. While the 62 participants performed the risk management


function, 26 had the title of risk manager, 3 were in human resources, 10 were in
administration, 8 were called safety officers, 5 were nurses, and 8 did not fit into the above
categories. Seventy-four percent of the respondents had worked at their participating
hospital for 5 or more years.
Respondents were asked if their facility had undergone a change in ownership during
the 3-year period of study. Of the 62 respondents, 57 (92%) reported no change, 3 (5%)
reported a change in owner with no impact on the responses to survey items and 2 (3%)
reported a change that may have impacted survey items.
Participating hospitals ranged in size from 55 to 6000 employees (full-time equivalent),
with a median size of 390 employees. Most of the hospitals, 40 (65%) were public, 21
were private (34%), and one provided no information.
4.2. Level of analysis
Climate researchers have often debated issues concerning the level of analysis and
aggregation procedures. Climate surveys, which collect data at the individual level,
provide limited information about the actual activities of an organization. Culture research
calls for in-depth case studies that may require multiple organizations as the units of
analysis to compare performance. In order to change an organization, data concerning an
organization, as well as how that organization compares to others are required (Reichers &
Schneider, 1990). The goal of the present study is to determine what factors, across
organizations, predict injury rates; thus, the unit of study is at the organizational level.
When studying organizational culture, the focal unit (whole organization, department, or
work group) must be specified. When the level of analysis is not identified, there is a risk
of ambiguity that Rousseau (1990) feels plagued much of the early climate research. In the
current study, participating risk managers were informed, Please keep your responses
general, to your hospital as a whole.
4.3. Instruments
4.3.1. Organizational factors data
Perception surveys have been used to effectively identify improvements in and
deterioration of safety system elements (Ostrom et al., 1993). Participating hospitals were
evaluated for the six predictor variables via an instrument designed for this study. Items
were adapted from Ostrom and his colleagues study and pilot-tested with risk managers
from three hospitals. Items that were unclear to these risk managers were reworded until
there was agreement in the interpretation of the questions.
Participants were given the instructions, Considering only the hospital where you
work, please respond to the following questions by circling the appropriate number
following each question. Extent scales 1 = no extent to 5 = a great extent followed each
item. Three items were developed to assess each of the six management practices. A
sample item used to assess training was To what extent do you believe that the safety
training provided to personnel is adequate to enable them to assess hazards in their
work areas? A sample item to evaluate level of participation was To what extent do

268

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

safety committees or teams have the power to implement change? See Table 1 for all
items.
Additional items solicited the frequency of the safety meetings, the frequency that top
administrators met with safety committee members, and the type of rewards used. There
was also an item to determine whether the hospital had changed ownership (such as

Table 1
Management practices survey items
Management practice

Survey item

Rewards

To what extent do you think that work-related injuries are due to a


lack of rewards for reporting hazards?
To what extent are employees rewarded for reporting a safety hazard
(e.g., thanked, have employee recognized in hospital newsletter,
receive cash or other awards)?
To what extent are employees punished for reporting a safety hazard
(e.g., they are ignored or told to keep it quiet)?*
To what extent do you believe that the safety training provided to personnel
is adequate to enable them to assess hazards in their work areas?
To what extent does the training program perform assessments following
instruction to verify that the safe work practices are being carried out in
the work areas?
To what extent do you think that work-related injuries are due to a lack
of training?*
To what extent do you think that work-related injuries are due to a
lack of management support in correcting employee safety hazards?
To what extent do supervisors in your hospital enforce safe
working procedures?
To what extent do the administrators of your hospital demonstrate
that safety is important to them (e.g., take immediate action to
eliminate safety hazards, list safety issues high on the agenda of
management meetings)?
To what extent does your hospital use a hazard reporting system
where employees can communicate hazard information before
incidents occur?
To what extent are near-miss incidents analyzed as warning signals
that must be studied and corrected?
To what extent do you think that work-related injuries are due to a lack
of feedback to employees about their unsafe behavior?*
To what extent are employees hired based on a good safety record in
their previous positions?
To what extent does management seek information about job candidates
prior safety performance in selecting or transferring employees?
To what extent do you think that work-related injuries are due to a lack
of hiring people who are safety conscious?*
To what extent do employees participate in identifying safety problems?
To what extent does management solicit opinions from employees before
making final decisions?
To what extent do safety committees or teams have the power to
implement change?

Training

Management commitment

Communication
and feedback

Selection

Participation

* Reverse-scored.

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

269

from public to private). If they responded yes, they were then asked if their responses
would be different if they pertained to the hospital under previous ownership. There was
little demographic data requested to protect anonymity of the hospitals. Internal
consistency reliability of the overall management practices scale was calculated at .86
(coefficient a).
4.3.2. Hospital injury data
Participating hospitals provided data for the criterion variable via an instrument
designed specifically for this study. This form provided respondents with a space to
document the number of injuries for 15 injury categories. Injury categories were
generated based on injury data (OSHA 200 forms) provided as examples from two
facilities. Participants were asked to Please fill in the number of injuries of each type
that occurred to hospital employees during the 3 previous years, in the shaded area
below. These categories included sprains, strains, and fractures, communicable or
infectious diseases, needle punctures, and fractured/crushed fingers or hands.
Participants were asked to indicate the number of injuries (frequency) of each type that
occurred to hospital employees during the 3-year period (1994 1996). There was also a
statement in the instructions, If you cannot easily provide the information in the format
as it is requested on this form, you may send copies of your hospitals OSHA 200 forms
covering this 3-year period. Remember to remove your hospitals name from the top of
the forms to insure confidentiality. If the data was sent on the OSHA 200 forms, the
researcher tallied and converted the data onto the survey form and destroyed the OSHA
forms.
In order to maintain confidentiality, few demographic questions were included. The
three demographic items requested the type of facility (public, private, investor-owned),
the hospitals full-time equivalent (FTE) employment (to control for facility size differences), and whether this included per diem and contractor employees.
4.3.3. Severity data
In order to assign weights to calculate the composite criterion, severity data were
collected. Since severity is very difficult to measure from existing data due to a variety of
regulatory and organizational constraints, perceived severity ratings by an expert panel
were used. The instrument (with the same 15 injury categories) was pilot-tested with two
physicians. Some changes were then made to the instructions to increase clarity. The 14
experts who completed this questionnaire were all physicians who work in hospitals as
part of their job. Because physicians are used as experts to testify in court as to the extent
of damages resulting from an injury, they were selected to rank the injuries concerning
their severity.
Participants were instructed, For my doctoral dissertation, I need to have rankings
assigned to each of the risk factors listed below in order that I may assign weights for the
statistical analysis. Considering your work in the hospital(s) where you practice, and based
on your experience, please rank the hazards from 1 (not severe) to 15 (extremely severe).
Please take into consideration such factors as days off work, permanent or long-term
inability to perform current job duties, medical expenses, and whether the hazard is lifethreatening (not probability of occurrence).

270

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

4.4. Procedure
Participants were solicited via telephone to participate in a research study concerning
hospital risk management. During this discussion, the experimenter explained the system
to protect anonymity. Research materials, a confidentiality agreement, and a pre-addressed
return envelope were mailed to volunteers. Responses were assigned random numbers;
thus, it was impossible to link data to its hospital.
Participants received follow-up calls or faxes to remind them to return the surveys.
Since the researcher did not know who responded, all original contacts were recontacted.
There was a statement on the written reminder that thanked them if they had already
returned the survey.
4.4.1. Response rate
Out of the 194 phone calls (only calls that reached the risk manager were counted), 125
agreed to participate (64%) and were mailed surveys. Of these 125 surveys, 74 were
returned (59%); however, only 62 had criteria information and were usable for the
regression analyses (50%).

5. Results
The central question addressed in this study concerned the degree to which six
management practices predicted hospital employee injury rates. To evaluate this issue,
several steps were required.
5.1. Expert rankings
In order to compare hospitals employee injury rates, it was necessary to determine both
the frequency and severity of the injuries. Severity data were collected as expert rankings,
which were converted into an interval scale and used to weight the frequency data. An
estimate of reliability of the responses by the 14 expert raters was .82. Two of the raters
were found to have contributed the most to the variance in the ratings. By removing both,
reliability increased to .87. The rankings of the remaining 12 raters were then converted
into an interval scale using Thurstones discriminate model.
5.2. Computing the criterion
There was a large range in the frequencies of the different injury types. Table 2 presents
the mean (adjusted for number of employees to control for hospital size) and relative
frequencies of the 15 injury types. The mean frequency is the annual average for each
injury type based on 3-years injury data. The first step in computing the criterion measure
was to weight the frequencies of each injury type with the severity factor developed from
the physicians expert rankings. The total number of (weighted) injuries per hospital
ranged from 19 to 1780 injuries, with an average of 262 injuries and a standard deviation
of 327. The total number of (weighted) injuries was divided by the number of full-time

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

271

Table 2
Averages and relative frequencies of the injury types (across all hospitals)
Injury type
Sprains, strains, and fractures
Needle punctures, blood exposure
Contusions
Lacerations/cuts
Cumulative trauma disorder (CTD)
Other (allergic reactions,
unknown causes)
Disease exposure
Burns
Abrasions
Eye injuries
Skin disease
Finger injuries
Toxic exposure
Mental stress
Human or animal bites
Total

Mean
frequency/year

Percent
of total

Mean frequency/year
(per 100 employees)

40.25
14.94
14.87
10.35
7.51
6.53

34
13
13
9
6
5

6.00
2.70
2.10
1.90
1.40
0.87

5.35
3.23
3.78
3.08
2.86
2.20
2.01
0.86
0.50
118.11

5
3
3
3
2
2
1
1
0
100

0.69
0.61
0.60
0.59
0.36
0.34
0.34
0.19
0.13
18.82

employees for each hospital, resulting in the number of (weighted) injuries per employee
for each participating hospital. These values ranged from 0.03 to 1.12, with a mean of 0.43
(weighted) injuries per employee and a standard deviation of 0.22.
Several of the responding risk managers included a note with their injury data
indicating that their hospitals did not record illness data. Some respondents noted that
their hospitals recorded exposure to illnesses, while others reported that they recorded only
actual infections. As a result of the variability in the recording and reporting of exposure to
diseases/illnesses, the incident types, communicable or infectious diseases and needle
punctures, blood exposure were not used in the calculation of the criterion variable; thus,
only the 13 injury types were used (see Table 2 for the 15 injury categories). The final
criterion variable (one observation per hospital) ranged from 0.02 to 0.90 injuries per
employee, with a mean of 0.30 injuries and a standard deviation of 0.17.
5.3. Predicting injuries
The principal analysis was a linear multiple regression that assessed the predictive
capacity of management practices (subscales) Participation, Management support, Training, Hiring practices, Communication/feedback, and Rewards of hospital employee injury
rates. The multiple correlation was .41, R2=.165, and adjusted R2=.137. A significant F
statistic [ F(2,59) = 5.84, P < .01] indicated a reliable linear relationship between the
management practice subscales and the criterion. The only management practice that
individually predicted injury rates was Hiring practices; the multiple correlation was .268,
R2=.07, and adjusted R2=.06. A reliable linear relationship [ F(1,60) = 4.66, P < .05] was
established.

272

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

Another variable that was found to account for the differences in injury rates was
whether the person who performed the risk management function was classified as a
manager. The multiple correlation was .274, R2=.075, and the adjusted R2=.06. A
significant linear relationship [ F(1,60) = 4.87, P < .05] was found.
The size of the hospital (number of employees) also predicted injury rates, with smaller
hospitals averaging more injuries per employee than the larger ones. The multiple
correlation was .380, R2=.145, and the adjusted R2=.130.
An exploratory factor analysis was conducted to verify that the management practices
items (predictors) loaded onto the expected subscales (the six management practices). Six
factors, with eigenvalues > 1, emerged in 10 iterations and were rotated using the Varimax
method. In total, the solutions accounted for 69% of the variance in the data: Factor 1
accounted for 29.7% of the variance; Factor 2 accounted for 10.8%; Factor 3 accounted for
8.0%; Factor 4 accounted for 7.2%; Factor 5 accounted for 6.6%; and Factor 6 accounted
for 6.1%. The items comprising the six factor scales did not correspond to the categories
derived from the practitioners reports (the management practices subscales). As a result of
the factor analysis, six factor scales were developed. Items were accepted for a factor scale
if they had a correlation with the factor (factor loading coefficient) greater than .70.
A multiple regression analysis was performed using the six factor scales identified in
the factor analysis to determine whether the factor scales predicted injury rates. The six
factor scales were entered (items equally weighted within the factor scales). Factors 1 and
2 made a significant contribution to the prediction of the variance in hospital employee
injury rates; the multiple correlation was .385, R2=.15, and adjusted R2=.12. A relationship
between Factors 1 and 2 and injury rates [ F(2,59) = 5.14, P < .01] was found.
Table 3 provides the items comprising Factors 1 and 2; the items comprising Factor 1
(near-miss incidents are analyzed as warning signals, and supervisors enforce safe work
practices) were reactive. There was a safety violation that needed correction. Factor 2
contained proactive practices concerning the initial selection and training of employees.
Because Factor 1 was positively related to injury rates (i.e., the more hospitals performed
these desirable actions, the higher their employee injury rates), Factor 1 was acting as a
suppressor variable. Factor 1 controlled for a portion of the error variance of Factor 2,
Table 3
Best predictors of employee injury rates items comprising factors 1 and 2
Factor

Correlations
with criterion

.227

.192

Critical questions
Predict injury rates
To what extent are employees hired based on a good safety record in their
previous positions?
To what extent does management seek information about job candidates prior
safety performance in selecting or transferring employees?
To what extent does the training program perform assessments following instruction
to verify that the safe work practices are being carried out in the work areas?
Suppressor
To what extent are near-miss incidents analyzed as warning signals that must be
studied and corrected?
To what extent do supervisors in your hospital enforce safe working procedures?

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

273

Table 4
Model summary
Factor

R2

Adjusted R2

S.E.

R2 change

2
1

.227
.385

.051
.148

.036
.120

0.165
0.158

.051
.097

F
.366
.341

3.253
5.143**

** P < .01.

which was negatively related to criterion (see Table 4 for the model summary). Thus,
while all of the participating hospitals, to some extent employed the reactive measures
comprising Factor 1, the primary difference in performance was that the participants with
lower injury rates also performed the proactive measures (Factor 2), while the hospitals
with high injury rates relied solely on putting out fires or fixing hazards after
problems had occurred.
Post-hoc tests were performed to determine whether there were possible alternate
explanations for the results. Since the type of hospital was categorical (public, private,
investor-owned), a nonparametric (independent sample Mann Whitney) test was performed; the type of hospital made no difference in the employee injury rates. A Mann
Whitney test was also performed to determine if the job title of the person (risk manager,
safety officer, nurse, etc.) who performed the risk-management function had any impact on
injury rates. No difference was found.

6. Discussion
The most important finding of this study is that when organizations take proactive
measures to protect their employees, the company derives a financial benefit in reduced
lost time and workers compensation expenses. While previous research has typically
discussed management practices as general goals, the current study systematically
examined the specific elements of these practices that predict employee injury rates.
Consistent with Eckhardt (1996) and Turner (1991), the current study found that the
consideration of safety performance in the selection of employees was found to be a
significant predictor of injury rates. The results from this study may help establish a bona
fide occupational requirement for requesting these data (to avoid discrimination claims).
Furthermore, since safety behavior is often tied to quality of performance, it is probable
that an added benefit of this approach may be an improvement in productivity.
When reviewing the results of this study, one may be inclined to believe that the items
comprising Factor 1 actually caused the poor injury performance. However, with further
analysis, it becomes apparent that Factor 1 is acting as a suppressor variable. It would be
incorrect to infer that analyzing near miss incidents and enforcing safety practices
increased injury rates; in fact, these measures were employed by both high and low
performing hospitals. Therefore, it is not recommended that hospitals discontinue the
practices of Factor 1, if they are in effect; however, resources and focus should be
channeled toward the proactive measures of Factor 2.
The most effective step that hospitals can take is in the front-end hiring and training of
new personnel. They should also ensure that the risk management position has a

274

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

management-level classification. This study also demonstrated that training in itself is not
adequate. Organizations must verify that the safe practices taught in the classes are being
implemented in the work areas. The results of this study can be used to determine which
factors to emphasize when performing an organizational development change in safety
culture. While it is not recommended that hospitals discontinue the reactive practices that
are in effect, resources and focus should be channeled toward more proactive measures.
Due to the high turnover of hospital personnel, selection of new employees is an ongoing
process; therefore, there is ample opportunity to consider safety records when selecting
new employees (Cal/OSHA, 1997).
A few hospital risk managers wrote that they were unable to obtain information about
job candidates prior injury records from other facilities. One reliable and valid approach
to solicit this type of information is through the behavioral based interview (Thornton &
Byham, 1982). To use this approach, the interviewer must be trained in the concepts and
techniques of behavioral interviews. An example of a question that may be used to assess
an employees safety record is Please describe the types of accidents or near misses you
have had in your current or previous jobs. Another example is Please provide an
example of when you had to call a co-workers attention to a possible violation of a safety
regulation. The applicant should describe the situation, the action he/she took, and the
result (Huck, personal communication, July 1998).
The injury rates at smaller hospitals were found to be higher than the larger ones. These
findings may result from these institutions having a less comprehensive safety program,
causing them to take a more reactive approach to injury prevention.
The data collected in this study to measure management practices reflected the
perceptions (and potential biases) of the risk managers. It is not possible to determine a
true level of these characteristics. The number of responses used in the factor analysis
poses another limitation; the 12 responses that had missing criterion data were included in
this factor analysis to raise the sample size to 74. This number is somewhat lower than the
rule of thumb of five per item (which would require 90 responses), due to the difficulty
in recruiting hospitals. However, according to Tabachnick and Fidell (1989, p. 603), If
there are strong, reliable correlations and a few, distinct factors, a sample size of 50 may
even be adequate, as long as there are notably more cases than factors. Distinct, strong
correlations were found in this study. This issue was further mitigated by using a high
factor-loading cutoff score (.70).
Since this was the first study of this type, a replication and extension of this work is
recommended. A recommended follow-up study is to select two comparable hospitals,
preferably within the same system, with the same (or similar) scores on the four
management practices comprising Factors 1 and 2. A preliminary assessment could be
conducted to establish a baseline measure of the performance of these practices. An
intervention can then be developed to maximize performance on the three proactive items
identified in this study as the best predictors of low injury rates (Factor 2). This
intervention would emphasize a front-end approach where new personnel are screened
and selected based on their past safety records. The approach could include behavior-based
assessment. In addition, after these new (and existing) personnel are trained in appropriate
safe work practices, an assessment will be performed to verify that the safe behaviors have
been implemented in the work areas.

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

275

While the level of reactive practices (Factor 1) should be measured, the practices should
not be changed. This intervention would be implemented in one of the facilities, with the
second acting as a comparison site. A program evaluation could be performed with 3years archival data as the baseline measure. Data could then be collected at 6-month
intervals following the intervention for a period of 3 years. Any preliminary differences
between the two facilities would serve as covariates. This proposed study would determine
whether taking a proactive approach positively influences safety culture to reduce
employee injury rates.
Acknowledgements
Many thanks to my dissertation chair, Richard Sorenson, who provided guidance
throughout this research. This study was funded in part by Error Analysis.
References
Barr, J. (1998). Cultivating culture. Occupational Health and Safety, 67(1), 32.
Berwick, D. M., Godfrey, A. B., & Roessner, J. (1990). Curing health care. San Francisco, CA: Jossey Bass.
Cal/OSHA (1997). A back injury prevention guide for health care providers. Sacramento, CA: State of California
Industrial Relations.
Cohen, H. H., & Cleveland, R. J. (1983). Safety program practices in record-holding plants. Professional Safety,
3, 26 33.
Cohen, H. H., & Jensen, R. C. (1984). Measuring the effectiveness of an industrial lift truck safety training
program. Journal of Safety Research, 15, 125 135.
Eckhardt, R. (1996). Practitioners influence on safety culture. Professional Safety, 7, 23 25.
Garrett, R. B., & Perry, A. J. (1996). A safer way to move patients. Occupational Health and Safety, 65(9), 60 64.
Hall, D. W., & Hecht, A. W. (1979). Summary of the characteristics of the air safety system reporting database.
Ninth Quarterly Report NASA, TM 78608, 23 34.
Halloran, A. (1996). Incentives benefit safety programs. Occupational Health and Safety, 65(6), 60 61.
Hofmann, D. A., Jacobs, R., & Landy, F. (1995). High reliability process industries: individual micro, and
macroorganizational influences on safety performance. Journal of Safety Research, 26(3), 131 149.
Hofmann, D. A., & Stetzer, A. (1996). A cross-level investigation of factors influencing unsafe behaviors and
accidents. Personnel Psychology, 49, 307 339.
Johnson, L. (1997). Safety in ward 3. Occupational Health and Safety, 66(7), 37 41.
Kletz, T. A. (1985). An engineers view of human error. Warwickshire, England: Institution of Chemical Engineers.
Kletz, T. A. (1993). Organizations have no memory when it comes to safety: a thoughtful look at why plants dont
learn from the past. Hydrocarbon Processing, 6, 88 95.
Komaki, J., Barwick, K. D., & Scott, L. R. (1978). A behavioral approach to occupational safety: pinpointing and
reinforcing safe performance in a food manufacturing plant. Journal of Applied Psychology, 63(4), 434 445.
Laws, J. (1996). The power of incentives. Occupational Health and Safety, 65(1), 24 28.
Lin, L.-J., & Cohen, H. H. (1984a). Hospital employee safety and health: current problems and practices.
Professional Safety, 1, 32 35.
Lin, L.-J., & Cohen, H. H. (1984b). Hospital employee safety and health (Part II). Professional Safety, 2, 28 32.
National Institute for Occupational Safety and Health [NIOSH]. (1983). Development and evaluation of an
employee hazard reporting and management information system in a hospital (Report #210-81-3102). Cincinnati, OH: Author.
Ostrom, C., Wilhelmsen, O. C., & Kaplan, B. (1993). Assessing safety culture. Nuclear Safety, 65, 163 172.
Peavey, B. (1995). Dont reward the safety cover-up. Occupational Health and Safety, 64(3), 69 72.

276

A.G. Vredenburgh / Journal of Safety Research 33 (2002) 259276

Pidgeon, N. F. (1991). Safety culture and risk management in organizations. Journal of Cross-Cultural Psychology, 22(1), 129 140.
Reason, J. T. (1994). A systems approach to organizational error. Proceedings of the 12th Triennial Congress of
the International Ergonomics Association, vol. 1 (pp. 94 96).
Reichers, A. E., & Schneider, B. (1990). Climate and culture: an evolution of constructs. In B. Schneider (Ed.),
Organizational climate and culture ( pp. 5 39). San Francisco: Jossey-Bass.
Roughton, J. (1993). Integrating quality into safety and health management. Industrial Engineering, 7, 35 40.
Rousseau, D. M. (1990). Assessing organizational culture: the case for multiple methods. In B. Schneider (Ed.),
Organizational climate and culture ( pp. 153 192). San Francisco: Jossey-Bass.
Siehl, C., & Martin, J. (1990). Organizational culture: a key to financial performance? In B. Schneider (Ed.),
Organizational climate and culture ( pp. 241 281). San Francisco: Jossey-Bass.
Swearington, M. H. (1996). Do safety incentive programs really help? Occupational Health and Safety, 65(10),
164 166.
Tabachnick, B., & Fidell, L. (1989). Using multivariate statistics (p. 603). New York: Harper Collins.
Thompson, K. R., & Luthans, F. (1990). Organizational culture: a behavioral perspective. In B. Schneider (Ed.),
Organizational climate and culture ( pp. 319 344). San Francisco: Jossey-Bass.
Thornton, G. C., & Byham, W. C. (1982). Assessment centers and managerial performance. New York: Academic Press.
Turner, B. A. (1991). The development of a safety culture. Chemistry and Industry, 4, 241 243.
Vredenburgh, A. G., & Cohen, H. H. (1995). High-risk recreational activities: skiing and scubawhat predicts
compliance with warnings. International Journal of Industrial Ergonomics, 15, 123 128.
Werdegar, D. (1980). Employee health. Nursing Clinics of North America, 15(14), 769 787.
Wilpert, B. (1994). Industrial/organizational psychology and ergonomics toward more comprehensive work
sciences. Proceedings of the 12th Triennial Congress of the International Ergonomics Association, vol. 1
(pp. 37 40).
Young, S. L., Brelsford, J. W., & Wogalter, M. S. (1990). Judgments of hazard, risk and anger: do they differ?
Proceedings of the Human Factors Society 34th Annual Meeting, (pp. 503 507).
Zohar, D. (1980). Safety climate in industrial organizations: theoretical and applied implications. Journal of
Applied Psychology, 65, 96 102.
Alison Vredenburgh holds a PhD in Industrial Organizational Psychology and a MS in Systems Management.
She is currently a postdoctoral research fellow at the School of Medicine (Department of Anesthesiology) at the
University of California, San Diego, where she is researching medical error. She is President of Vredenburgh and
Associates Inc., a consulting firm specializing in human factors and safety. Her principle publications are in the
areas of human factors, ergonomics, and workplace management practices. She is active in the Human Factors
and Ergonomics Society, where she has held several leadership roles.

Anda mungkin juga menyukai