Title: Improvement of Parenteral Antibiotic use in a University Hospital in Colombia
1Improvement of Parenteral Antibiotic use in a
University Hospital in Colombia
Pérez A, Dennis RJ, Rodriguez B, Castro AY,
Delgado V, Lozano JM
- Clinical Epidemiology and Biostatistics Unit,
Pontificia Universidad Javeriana, Bogotá, Colombia
2ABSTRACT
- Problem Statement In Colombia, there has been no
incentive in the past for continuing quality
assessment, by ongoing monitoring, of antibiotic
prescription practices. - Objectives To evaluate the effect of an
intervention to improve antibiotic prescribing
practices in a University- based hospital. - Design Quasi-experimental before/after study
with a planned intervention interrupted time
series analysis. - Setting Tertiary care hospital caring for
private and institutional patients. - Study Population Hospitalized patient
prescription census from 10 clinical services,
including Gynecology and Obstetrics, Surgery,
Medicine, and Pediatrics. A total of 2716
prescriptions were collected between June, 1997
and April, 2000. - Intervention A structured antibiotic order form
implemented between two data collection phases
between week 82 and week 102. All hospitalized
and prescribed patients completed the form since
week 82.Physicians in charge of grup prescription
in each service completed the forms. The Hospital
designed the form with the help of the research
tem. We also implemented an educational campaign
with conferences for physicians and posters for
all the clinical services, and blood pressure
cuffs for anaesthesiologists. - Outcome Measures Hospital weekly rate of
incorrect prescriptions of (A) aminoglycosides in
dose interval less than 24 hours (gentamicin,
amikacin, streptomycin and netilmicin) (B)
cephradine and cephalothin in dose interval
greater than 6 hours (C) ceftazidime and
cefotaxime in dose interval greater than 8 hours
and (D) any antibiotic prescribed one hour before
or after incision in surgery. - Results Interrupted time series intervention
analysis was conducted for three antibiotic
groups of the hospitals weekly rate of incorrect
prescriptions. Pre-intervention Auto-Regressive
Integrated Moving Average (ARIMA) models were
identified, estimated and diagnosed for the four
time series (A,B,C,D). Time series (A) was an
ARIMA (0,1,2) with corresponding estimates and
standard error (SE) as theta10.36 (SE0.102) and
theta20.49 (SE0.101), respectively. Time series
(B) was an ARIMA (0,1,1) with corresponding
estimate 0.82 and SE0.07. Time series (C) was
an ARIMA (0,0,1) with corresponding
estimate-0.72 and SE0.08. Time series (D) was
an ARIMA (0,1,1). These models were used in the
post-intervention series to test for pre-post
series level differences. An abrupt constant
change was significant in A, C and D time series,
indicating a 47, 7.3 and 20 reduction on
incorrect prescriptions after intervention. - Conclusions High rates of incorrect prescription
were reduced after the intervention. This
intervention, consisting of both an educational
campaign and introduction of a structured
prescription form with built-in deterrents of
selection of inappropriate dosing intervals, can
be implemented in a teaching hospital in Latin
America. Such an intervention leads to measurable
decreases in the proportion of incorrectly
prescribed antibiotics.
3BACKGROUND
- Uncontrolled use of antibiotics abuse and
potential unwarranted events and costs - Need for
- Ongoing monitoring of antibiotic prescription
practices - Implementation of interventions to improve
inappropriate behavior - Pharmacy and Infection Control Committees
identified a critical area as - Use of expensive antibiotics
4BACKGROUND
- Pharmacy and Infection Control Committees
identified critical areas - Use of expensive IV antibiotics
- Implementation of an adverse drug reaction
surveillance program - Use of sedatives and hypnotics
- Drug modification as a function of renal
condition - Adequate pharmacological prevention of UGI
bleeding and thromboembolism
5RESEARCH OBJECTIVES
- To assess the appropriateness of the observed
antibiotic prescription patterns. - To implement a hospital wide intervention aimed
to improve inappropriate practices. - To assess the potential cost/savings profile of
the intervention from the payer point of view.
6RESEARCH DESIGN
METHODS
- Quasi-experimental pre-post time series design
- Reasons for not using an RCT
- Permanent rotation of residents, interns and
nurses very high potential for contamination bias
within and between wards, which would attenuate
any perceived effects - Selection of one other hospital as control
unfeasible control of measurable confounders
7SETTING
- San Ignacios Hospital, Bogotá, Colombia. June,
1997 - Hospitalized patients
- Obstetrics-Gynecology, Surgery, Medicine,
Pediatrics, Intensive Care Unit, others wards.
EXPERT PANEL
- PI, infectologist, representatives from GO,
Pediatrics, Internal Medicine, Surgery and
Nursing - Identifying tracer conditions
- Developing expected norms regarding the
appropriate use of antibiotics in selected
conditions - Developing data collection forms
- Delineating intervention
8INTERVENTION
- 1. Implementation of a new antibiotic order form
- 80 in US hospitals, 79 in British hospitals
- 2. Join educational intervention by researchers
and infectologist (lectures and posters) - 3. Logo bandblood pressure cuffs Do not forget
the prophylactic antibiotic one hour before
surgical incision. - Started in January/1999
9OUTCOMES
Hospital weekly proportion of incorrect
prescriptions
- Incorrect prescription
- Dose interval lt 24 h
- Dose interval gt 6 h
- Dose interval gt 8 h
- Prescription gt 1 hour before and/or after
incision
- Condition
-
- Aminoglycosides
- Cephradine/Cephalothin
- Ceftazidime/Cefotaxime
- Prophylactic prescription in surgery
10SAMPLE SIZE
- ? 0.05, two sided test
- ? 0.10
- ARIMA (2,0,0)
- ?1 0.3
- ?2 0.2
- 20 months of observation before and after
intervention 80 weeks pre-post
Gottman JM (1981) Time series analysis. Cambridge
Univ. Press, 335-67
11HYPOTHESIS
Stationary series (discrete and equally spaced
intervals)
Auto-regression process
Moving average process
Estimated from time series
Random shocks
12ETHICAL ISSUE
- Informing staff about prescription pitfalls
outside the intervention period may produce
temporary changes in habits that may attenuate
results - Data collector will not make staff aware of
minor prescription errors - Data collector will make staff aware of major
prescription errors
STATISTICAL ANALYSIS
- Identification of pre-intervention ARIMA model
- Diagnosis checks over residuals
- Akaike Information Criterion
- No seasonal component expected
- SAS 6.12 TSO 51, Unix
13RESULTS
Antibiotic Order Form
14RESULTS 1.Aminoglucosides
Pre-Intervention ARIMA (0,1,2)
- Abrupt Constant Change was statistically
significant.
-0.477 SE0.064 plt0.001
15RESULTS 2.Cephradine/Cephalothin
Pre-Intervention ARIMA (0,1,1)
Neither abrupt constant nor temporary change
were statistically significant.
16RESULTS 3.Ceftazidime/Cefotaxime
Pre-Intervention ARIMA (0,0,1)
Abrupt Constant Change was statistically
significant.
-0.073 SE0.03 plt0.05
17RESULTS 4.Prophylactic P. in Surgery
Pre-Intervention ARIMA (0,1,1)
Abrupt Constant Change was statistically
significant.
-0.199 SE0.069 p0.004
18DISCUSSION
- This study confirms previous reports of
reductions in the proportion of incorrect
antibiotic prescriptions by use of an educational
campaign and a structured antibiotic order form. - We believe that our structured prescription form
improved the quality of the prescriptions by
increasing the awareness of physicians about
correct dose intervals which is consistent with
previous studies reported in the literature.
19DISCUSSION
- Reduction in incorrect
- Prescriptions
- 47 Aminoglycosides
- 7.3 Ceftazidime
- Cefotaxime
- 20 Prophylactic P. in
- surgery
- No enough reduction in Prophylactic Prescriptions
in Surgery. - RCT not feasible due to permanent rotation of
residents/nurses, etc. - Ethical Issue informing staff about prescription
pitfalls.
20ACKNOWLEDGMENTS
- This work was supported by INCLEN, INC (grant
1004-97-6501) and Pontificia Universidad
Javeriana (grant 12-24-01-31).