2012 Symposium on Human Factors and Ergonomics in Health Care

MITIGATING RISKS ASSOCIATED WITH ADMINISTERING MULTIPLE INTRAVENOUS INFUSIONS: METHODS FOR ORGANIZING AND ANALYZING PROACTIVE RISK DATA

Andrea Cassano-Piché, M.A.Sc, P.Eng

Human Factors Engineer, Health Technology Safety Research Team University Health Network

Mark Fan, M.H.Sc

Human Factors Analyst, Health Technology Safety Research Team University Health Network

Anthony C. Easty, Ph.D, P.Eng, CCE

Health Technology Safety Research Team Leader University Health Network

Administering multiple infusions to a single patient is a requisite but risk-prone task that takes place in many patient care areas. Identifying and prioritizing the associated risks proactively requires detailed knowledge about the context, including a wide range of medication administration tasks and processes, which results in a large, highly interconnected data set. This paper discusses the methods used to collect, organize and analyze risk data related to the administration of multiple infusions in 12 patient care areas across 10 different hospitals, as well as the patient safety risk themes and associated hospital-based recommendations identified through this research.

Introduction  A major challenge of collecting and analyzing proactive risk data using ethnography in the healthcare setting is the volume and complexity of the data that arises from the number of degrees of freedom in the system (Savage, 2000). Unlike other complex socio-technical safety-critical systems where the physical components and monitoring sensors of the system being controlled by expert users are fixed, and are defined primarily by physical principles (e.g., nuclear power plants), healthcare workers are often required dynamically to construct the physical systems they control and monitor as they are caring for patients. This increases the amount of context-specific data that must be captured and also increases the difficulty of separating out requisite from extraneous complexity. Administering multiple IV infusions to a single patient is a complex task with considerable patient safety risk (Cassano-Piché, Fan, Sabovitch, Masino, & Easty, 2012). It is, however, a requisite task for many nurses caring for patients receiving complex treatment and has not been systematically evaluated in terms of the types of associated risks and the effectiveness of potential mitigating strategies (Health Technology Safety Research Team, Institue for Safe Medication Practices Canada, 2010). It thus warrants a detailed proactive risk analysis to mitigate potential patient safety risks. This paper describes a research approach used to systematically collect and analyze a large, ethnographic field data set to support a proactive risk assessment of failure modes associated with administering multiple IV infusions to a single patient. It also presents the themes of issues identified,

and the corresponding recommendations to hospitals for mitigating some of the identified issues.

Methods  

Data  Collection  Twelve ethnographic field studies were conducted across 10 Ontario hospitals in a wide range of patient care settings where multiple infusions are frequently administered to a single patient. Units included critical care environments, an emergency department, an inpatient oncology setting, and an outpatient oncology clinic. Four of the 10 field studies were conducted in pediatric settings and the remaining were conducted in adult care environments. A mixed methods approach was followed at each field study, which consisted of semi-structured interviews and direct observations of nurses caring for patients receiving multiple IV infusions. Each field study began with a semi-structured group interview of unit clinical administrators (e.g., nurse and pharmacy managers), clinical educators, medication prescribers, and risk managers. The number and type of participants varied across sites based on availability. Each interview lasted 2-3 hours and focused on the unit structure, infusion technologies in use, medication administration policies and practices, training and education, and challenges and safety issues related to administering multiple infusions. The interviews were followed by 2.5 days of direct observations conducted by two or three human factors researchers with experience in the healthcare domain. A nurse consultant on the project also participated in three of the field studies to further support the human factors researchers in understanding the clinical context.

Cop

yrig

ht 2

012

Hum

an F

acto

rs a

nd E

rgon

omic

s S

ocie

ty. A

ll rig

hts

rese

rved

. 10.

1518

/HC

S-2

012.

9452

8940

1.00

6

35

2012 Symposium on Human Factors and Ergonomics in Health Care

Qualitative field study data were collected in the form of photographs (Figures 1-2) and written notes. Issues within each field study were identified and tracked in a spreadsheet.

Figure 1. Multiple IV infusions connected to a single IV line via manifolds

Figure 2. IV tubing intertwined between the pump and the patient

Data  Analysis  The field study data set was large, and presented a challenge in terms of how to record the data such that the complex

cause-consequence relationships between issues both within and across field studies was preserved. The first attempt at managing the complexity of the data was to group the issues into task-based themes (Table 1). The themes provided a useful framework for dividing up the problem space so that individual researchers could explore each theme independently and concurrently, but did not make explicit how contributing factors across the system influenced problems in multiple themes. Table 1. Themes of issues identified from the field study data

Themes of Issues Secondary Infusions Line Set-up/Removal Line Identification Dead Volume Management IV Bolus Administration

To this end, the data were mapped onto a structural hierarchy representing the actions and decisions at all levels of the system (Figure 3). This type of structural hierarchy is a tool of Rasmussen’s 1997 Proactive Risk Management Framework (Rasmussen, 2007; Rasmussen & Svedung, 2000). Each element on the hierarchy (i.e., box) represented a potential contributing factor to a patient safety issue related to administering multiple IV infusions. The cause-consequence relationship between the factors was made explicit by a line connection creating chains of elements that sequentially led to outcomes that cause patient harm. Decisions and actions most directly connected to these outcomes were identified as primary issues. A risk analysis of the primary issues was conducted using a Healthcare Failure Modes and Effects Analysis (De Rosier, Stalhandske, Bagian, & Nudell, 2002). The HFMEA rating scheme is shown in Table 2. The cause-consequence chains associated with the critical issues were analyzed to identify potential mitigating strategies.

Results  More than 100 field study data elements were mapped onto the structural hierarchy. Factors with the potential to contribute to patient harm were identified across all levels of the system. Twenty-two primary issues were identified, of which 17 were identified as critical issues based on the HFMEA risk analysis. Critical issues were defined as primary issues with a hazard score great than 24. Mitigating strategies were identified for several of the critical issues, which included at least one mitigating strategy for each theme except dead volume. Mitigations focused on interventions that could be implemented at the hospital or hospital-unit level of the system. The mitigating strategies were communicated as nine recommendations to hospitals and are presented in Table 3 and described fully in Cassano-Piché, Fan, Sabovich et al, 2012.

36

2012 Symposium on Human Factors and Ergonomics in Health Care

Table 2. Healthcare Failure Modes and Effects Analysis Rating Scheme for Multiple IV Infusion Issues Rating Detectability

1 No checking process is required to help detect the error. 2 A checking process is required to help detect the error,

but this process is not well defined and relies on human vigilance.

3 The error is not detectable based on current standards for knowledge/experience.

4 The error cannot be detected by any reasonable human process.

Rating Likelihood 1 Remote = the error may happen within the next 5 years 2 Uncommon = the error is likely to happen within the next

2–5 years 3 Occasional = the error is likely to happen within the next

1–2 years 4 Frequent = the error is likely to happen within the next

year Rating Severity

1 Minor = Error results in no harm, or the potential harm is unknown.

2 Moderate = Patient is temporarily harmed by the error. 3 Severe = Patient is permanently harmed by the error. 4 Critical = The error causes the patient’s death.

Table 3. Recommendations to hospitals based on an analysis of the structural hierarchy

Recommendation 1 When initiating a secondary medication infusion (often

referred to as a “piggyback” infusion), nurses should verify that the secondary infusion is active, and that the primary infusion is not active, by viewing the activity in both drip chambers. Full drip chambers should be partially emptied to restore the visibility of drips.

2 Continuous high-alert medications (Institute for Safe Medication Practices Canada, 2005) should be administered as primary infusions. Continuous high-alert medications should not be administered as secondary infusions.

3 Secondary infusions should be attached to primary infusion sets that have a back check valve. If infusion sets without back check valves are also available on the unit, multiple strategies should be employed to ensure that the types of tubing available are easily differentiated, and that the likelihood of a mix-up is minimized.

4 Hospitals should work towards the use of gowns that have snaps, ties, or Velcro on the shoulders and sleeves.

5 If an “emergency medication line” that is controlled by an infusion pump is set up on a patient, it is strongly suggested that the associated primary IV tubing be labelled as the emergency medication line at the injection port closest to the patient. The label should be prominent, and visually distinct from all other labels in the environment.

6 When setting up multiple IV infusions at the same time (e.g., a new patient requires many ordered infusions immediately, routine line changes), infusions should be set up one at a time, as completely as possible, before setting up the next infusion. Set-up tasks required for each infusion vary and may include:

• labelling (e.g., IV tubing, pump); • spiking and hanging the IV bag; • connecting the IV tubing to the pump; • programming the IV pump;

Organisation

Government

ProfessionalPractice

andTechnologyRegulators

Management

Staffand

PatientActivities

Equipmentand

surroundings

LabelsMed lines not

labeled (tubing)

LabelsMed lines

not labeled (pump)

LabelsDrug libraries do not display drug

name prominently or indicate access

information

LabelsExternal labels

applied to pumps to provide drug

name and access info

LabelsExternal pump labels not removed prior to

setting up a new infusion on the pump

LabelsDrug tubing labels wrapped around IV tubing leave part of

the drug name obscured

LabelsLabel placed on incorrect

drug tubing line

LabelsGeneric and brand name

used inconsistently across pump bag and drug tubing labels for a single patient (all 3 labels written and

applied by the same RN) see SRH

Labels-Illegible

handwriting on pump/drug tubing

labels-Poor visibility of

labels

LabelsInconsistent use of

colour to identify CVC lines across vendors

LabelsInconsistent use of terminology for drug abbreviations and

access locations on labels (between RNs

on same unit)

LabelsNo labels designed specifically for med line, drug tubing of

pump labeling in use

LabelsPatient access information on the pump label is inaccurate

LabelsPump/Drug

tubing labels fall off

LabelsNo drug

tubing labels applied to

lines

DRequired

medications not delivered to the patient

EDelay/

interruption in administration

of critical medications

AInfusion delivered at

incorrect rate

BIncompatible medications

running together

CVessicant medication

in PIV

PumpSingle interface used to program multiple

pump channels/modules

PumpSoftware/User

interfaceDesign issues

PumpHighly potent medications administered

concurrently via a pump with poor flow

rate accuracy

PumpDrug library/dose

calculator not used to program a

primary infusion

PumpPrimary infusion

programming error (wrong values

entered)

Line DiagnosisTransporting/ambulating

patients tangles the

lines

Line DiagnosisTubing

intertwined like spaghetti

Line DiagnosisNo standardized approach to line

setups (arrangement, components used, etc)

Line DiagnosisNo policies or best practice outlined for how to exchange IV

setup information during handover

Line DiagnosisLine setup information

not formally communicated at

handover

Line DiagnosisLine tracing

process prone to error

Line DiagnosisSlower running line

attached upstream of faster running line(s) and no chaser/driver

attached

Line Diagnosis

Line identified incorrectly

Line DiagnosisIV Bags do not physically align with the pumps

they are attached to

Line DiagnosisMedication connected

to/disconnect from incorrect line

Line DiagnosisIncorrect line

removed

Line DiagnosisSettings changed on

the wrong pump/channel

Dead VolumeSignificant change in rate of one or more of the fluids connected to other rate-critical medications

Line change required

Dead VolumeCVP line flushed

Dead VolumeMedications in dead volume of new tubing not in correct proportions for

undermined length of time

Dead VolumeOne or more rate-critical

medications (fluids) connected with other (fluids) medications into a single piece of tubing

Dead VolumeRate-critical medication

administered via CVP line

Inadequate training on managing multiple IV infusions

High risk medication delivered

Patient is receiving multiple

concurrent IV infusions

Lack of guidance given to nursing education programs

and hospitals about multiple IV infusion concepts required in

the curriculum/training program

IV medications not D/C when

other administration routes become

available

RNs work at more than one organization

SecondarySecondary tubing

connected to the incorrect port along the tubing (ie

downstream of pump instead of upstream)

SecondarySecondary clamp not unclamped

SecondarySecondary IV

bag connected to incorrect primary line

SecondaryLack of experience

setting up secondary infusions

SecondarySecondary infusions perceived as low risk

SecondaryLength of time to program via

drug library

SecondarySecondary infusion not programmed via

the drug library

SecondaryNo drug library available for secondary infusions (SMH/

Colleague)SecondaryNo bolus

appropriate profiles in the drug library

SecondaryBag height difference

not great enough

SecondaryTraditional pump used

SecondarySecondary infusion programming error

SecondaryPump software design prevents the use of the secondary mode if the

primary line is programmed using a drug library/drug

dose calculator(Graseby, Baxter...?)

SecondaryNo bolus

feature on the pump

SecondaryBolus of primary infusion given via secondary mode

SetupMultiple IV infusions connected

at the same time

SetupIV bag-pump mismatch (pump

programmed before bag connected)

SetupIV bag-programming mismatch (pump programmed after bag

connected) SetupPatient

transfers from one unit to another (or from EMS)

SetupPump label-bag

mismatch (label applied after bag attached + pump programmed)

SetupNurses try to space the IV bags

apart to make the bag labels more visible

SetupColleague pumps tubing needs

to be moved up over time

SetupTubing too short to reach between bag and pumps when

multiple pumps are stacked vertically

SetupSecondary hangers used to lower primary bags when no secondary infusion is running

Setup3-way stopcock product design

issue

SetupUse of recalled 3-way large bore stopcock to

connect multiple IV lines

SetupLimited

central IV access

SetupConnecting multiple stopcocks in series

results in leaks

SetupIVs need to be

rearranged prior to patients going for CT imaging (ie free up a

PIV line)

SetupBatch process of steps during setup of multiple IV infusions

Patient Harm

!"#

!"#$

%

$%

$%

$%

$%

$%

$%

$%

$%

$%

$%

!"#

$%

!"#

!"#

!"#

$%

$%

!"#

!"#

!"#

$%

SecondaryNo backcheck valve

on primary tubing

Line DiagnosisIV hooks arranged

circularly on IV poles

SetupBackflow can

occur through Y connectors for 2 primary lines if

pressure differential too high

Line DiagnosisIncubator and

crib sides impede the continuous

tracing of lines

Line DiagnosisOnly single

channel pumps used, requiring

stacking of pumps

LabelsPre-printed, colour-coded drug name stickers are only

available for some drugs

LabelsColour-code scheme for drug tubing labels

has many similar colours for different

drugs

LabelsSyringe labels

applied lengthwise are obscured by the syringe clip

Dead VolumeNo standard means of

communicating that a fluid should not be bolused

because it is connected to other medications with sensitive infusion rates $

%

Line DiagnosisMultiple drugs

have same opaque colour

(propofol, lipids, breaskmilk, NG

feeds)

Line DiagnosisSecondary medication injected into incorrect

line (tubing or buretrol) (pediatrics)

Dead VolumeNo standard process for

managing the dead volume of a 2 or 3-way y-style connector if a drug is discontinued and removed

FBolus of

incorrect fluid/medication

administered

Dead VolumeMedication unhooked

from 2 or 3-way y-style connector and

there residual volume in the line is not

identifyable

SecondarySome hangers not

long enough for longer IV bags

SecondarySecondary

medication not in drug library

SetupRN height/

reach

A

!"#

SecondaryUnnecessary port on primary tubing

just below the pump

SetupSequential

intermittent infusions hung in advance of being connected to

the pump

SetupWrong IV bag connected to

the pump

SetupPumps need

to be returned to home unit

SetupLack of

standard drug concentrations between units

SetupDrug library

not configured for the unit patient is

transferring to

SetupDifferent

pumps used on different

units

SetupConcentration

of drugs is different

between the units

SetupDate of IV line (based on policy)

SetupIV tubing/

connectors not compatible

with new unit

1

3

45

6

7 8

9

10

11

12

13

14

15

16 17

18

19

21 22

23

24

25

26

27

28

29

30

31

32

34

35

36

37

38

39

4041

42 43

44

45

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

73

74

75

76

77

78

79

80

81

83

84

85

86 87 88 89 90 91

92

93

94

96 97

98

99

100

101

102

103

104

105

SecondaryBack flow of

secondary infusion into primary infusion

113

106

Line DiagnosisLine tracing

prcess lengthy

108

GAdministration

of a discontinued medication

98

E

!"#

!"#

9

51

106

A

107

111

Drug library not configured optimally

110

65

Procurement selection not

optimal for clinical needs

105

86

61

59

112

18

22

43

49 112

114

Hospital Budget constraints

Lack of usability standards by

device regulators (Canada)

61

109

Dead VolumeIV connector design

results in dead volume

114

35

11

1214 23 18 68

99

Figure 3. Structural Hierarchy of Factors Contributing the Multiple IV Infusion Administration Risks

37

2012 Symposium on Human Factors and Ergonomics in Health Care

• connecting the IV tubing to the appropriate location (e.g., patient access, manifold); and

• starting the pump (unless a secondary infusion must be set up prior to starting the pump, or other infusions need to be connected to a multi-port connector before flushing).

Minor modifications to this recommendation are required for routine line changes.

7 Multiple 3-way stopcocks joined together in series to connect multiple IV infusions into a single line are prone to leaks, which may often be undetectable. Hospitals should provide multi-port or multi-lead connectors, and nurses should use these connectors to join multiple IV infusions into a single line, as required.

8 Hospitals should develop a policy to limit the practice of manually increasing the infusion rate to administer a medication bolus of a primary continuous infusion. If a medication bolus is to be administered using the primary continuous infusion pump/pump channel, then the nurse should program the bolus dose parameters (i.e., total amount of medication to be given over a defined duration) into the pump without changing any of the primary infusion parameters. Some examples may include the following:

• programming a bolus using a dedicated bolus feature in the pump

• programming a bolus using the pump’s secondary feature

9 Hospitals should ensure that their smart pump drug libraries include hard upper limits for as many high-alert medications as are appropriate for each clinical area, in order to prevent the administration of a bolus by manually increasing the primary flow rate.

Discussion  The human factors proactive risk analysis of the processes associated with administering multiple IV infusions revealed that multiple IV infusion administration is a complex, risk-prone process that requires a systems approach to risk mitigation. The structural hierarchy tool from Rasmussen’s risk management framework (Rasmussen & Svedung, 2000) proved useful for organizing complex healthcare field study data, such that the relationship between many factors that influence human performance across levels of the system was explicit. This has several important implications for how risk-mitigating strategies are developed.

1. It supports a systematic analysis of the effect of a potential risk-mitigating strategy on all other elements of the system by making the relationship between all the elements explicit. This helps to minimize unintended negative consequences associated with changes to the system.

2. It supports a focus on solutions at higher levels of the system (e.g., government, regulatory bodies) than are usually considered, because the structural hierarchy tool prompts an inclusion of contributing factors at these levels.

3. It facilitates the aggregation of information across field sites, making the analysis more generalizable across the healthcare domain. Individual healthcare

organizations can each identify their own unique sub-set of issues from the set of factors presented on the hierarchy to create their own individual structural hierarchy as a first step towards developing mitigating strategies most important to their organization.

The nine recommendations identified in this study focused on the lower levels of the hierarchy, where human factors recommendations are traditionally made, because these recommendations can be immediately implemented to minimize risks. Mitigating strategies aimed at higher levels of the system (e.g., changes to clinical education programs, new standards for infusion systems) and solution approaches that require further study prior to determining the best mitigation approach are currently being investigated in the next phase of the study.

References  1. Cassano-Piché, A., Fan, M., Sabovich, S., Masino, C., &

Easty, A. C. (In Press). Multiple Intravenous Infusions Phase 1b: Practice and training scan. Ont Health Technol Assess Ser [Internet] . Toronto, ON, Canada

2. De Rosier, J., Stalhandske, E., Bagian, J. P., & Nudell, T. (2002). Using Health Care Failure Mode and Effects Analysis: The VA National Centre for Patient Safety's prospective risk analysis sistem. Joint Commission Journal of Quality Improvement , 28 (5), 248-267.

3. Health Technology Safety Research Team, Institue for Safe Medication Practices Canada. (2010, 09 24). Multiple Intravenous Infusions Phase 1a: Situation Scan Summary Report. Retrieved 03 19, 2012, from Health Technology Safety Research Team Web site: http://www.ehealthinnovation.org/files/Multiple%20IV%20Infusions_Phase1a_SummaryReport.pdf.

4. Institute for Safe Medication Practices Canada. (2005, 02 1). Secondary Infusions Require "Primary" Attention. ISMP Canada Safety Bulletin , 5 (2), pp. 1-2. Available at: http://ismp-canada.org/download/safetyBulletins/ISMPCSB2005-02SecondaryInfusions.pdf.

5. Rasmussen, J. (2007). Risk management in a dynamic society: A modeling problem. Safety Science , 27, 183-213.

6. Rasmussen, J., & Svedung, I. (2000). Proactive risk management in a dynamic society. Karlstad, Sweden: Swedish Rescue Services Agency.

7. Savage, J. (2000). Ethnography and health care. BMJ , 321, 1400-1402.

Acknowledgements  This research was commissioned by the Ontario Health Technology Advisory Committee, funded by Health Quality Ontario and conducted in collaboration with the Institute for Safe Medication Practices (ISMP) Canada. The authors wish to thank the members of the Multiple IV Infusions Expert Panel, the AAMI Multiple Line Management Working Group and the health care staff who supported and participated in the field studies.

38