Journal of Antimicrobial Chemotherapy Advance Access published November 11, 2013
J Antimicrob Chemother doi:10.1093/jac/dkt458
A modified method for measuring antibiotic use in healthcare settings: implications for antibiotic stewardship and benchmarking Mamoon A. Aldeyab1,2*, James C. McElnay1, Michael G. Scott2, William J. Lattyak3, Feras W. Darwish Elhajji4, Motasem A. Aldiab5, Fidelma A. Magee2, Geraldine Conlon1 and Mary P. Kearney6 1
*Corresponding author. Clinical and Practice Research Group, School of Pharmacy, Queen’s University Belfast, Belfast BT9 7BL, Northern Ireland, UK. Tel: +44-28-90972033; Fax: +44-28-90247794; E-mail: [email protected]
Received 22 June 2013; returned 8 September 2013; revised 13 September 2013; accepted 22 October 2013 Objectives: To determine whether adjusting the denominator of the common hospital antibiotic use measurement unit (defined daily doses/100 bed-days) by including age-adjusted comorbidity score (100 bed-days/ age-adjusted comorbidity score) would result in more accurate and meaningful assessment of hospital antibiotic use. Methods: The association between the monthly sum of age-adjusted comorbidity and monthly antibiotic use was measured using time-series analysis (January 2008 to June 2012). For the purposes of conducting internal benchmarking, two antibiotic usage datasets were constructed, i.e. 2004– 07 (first study period) and 2008– 11 (second study period). Monthly antibiotic use was normalized per 100 bed-days and per 100 bed-days/ age-adjusted comorbidity score. Results: Results showed that antibiotic use had significant positive relationships with the sum of age-adjusted comorbidity score (P¼ 0.0004). The results also showed that there was a negative relationship between antibiotic use and (i) alcohol-based hand rub use (P ¼ 0.0370) and (ii) clinical pharmacist activity (P ¼0.0031). Normalizing antibiotic use per 100 bed-days contributed to a comparative usage rate of 1.31, i.e. the average antibiotic use during the second period was 31% higher than during the first period. However, normalizing antibiotic use per 100 bed-days per age-adjusted comorbidity score resulted in a comparative usage rate of 0.98, i.e. the average antibiotic use was 2% lower in the second study period. Importantly, the latter comparative usage rate is independent of differences in patient density and case mix characteristics between the two studied populations. Conclusions: The proposed modified antibiotic measure provides an innovative approach to compare variations in antibiotic prescribing while taking account of patient case mix effects. Keywords: antibiotic measure, age-adjusted comorbidity index, time-series analysis, pharmacoepidemiology
Introduction Evidence for a relationship between antibiotic use/misuse and the development of antibiotic-resistant bacteria, at both the individual and the population level, has been documented.1 – 4 This has subsequently increased morbidity, mortality and incurred additional hospital costs.5,6 Antibiotic resistance is a multifactorial problem that requires a multifactorial solution, of which optimizing antibiotic use has been considered a key component.7 – 9 The latter has been formalized through the establishment of hospital antibiotic
stewardship programmes in a range of healthcare settings, which consist of a set of coordinated strategies to improve the use of antibiotic medications with the goal of enhancing patient health outcomes.7 – 10 The surveillance of hospital antibiotic use and resistance patterns is considered a key aspect of high-quality antibiotic stewardship implementation, evaluation and performance. To measure antibiotic use, the use of defined daily doses (DDDs) per 100 bed-days has been recommended by the WHO Collaborating Centre for Drug Statistics and Methodology.10,11 In healthcare settings, expressing antibiotic use as DDDs per 100 patient-days has
# The Author 2013. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]
1 of 10
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Clinical and Practice Research Group, School of Pharmacy, Queen’s University Belfast, Belfast BT9 7BL, Northern Ireland, UK; 2Pharmacy and Medicines Management Centre, Northern Health and Social Care Trust, Ballymena BT43 6DA, Northern Ireland, UK; 3Scientific Computing Associates Corp., River Forest, IL, USA; 4Faculty of Pharmacy, Applied Science Private University, PO Box 166, Amman 11931, Jordan; 5Faculty of Computing and Information Technology – North Jeddah Branch, King Abdulaziz University, PO Box 80221, Jeddah 21589, Saudi Arabia; 6 Microbiology Department, Northern Health and Social Care Trust, Ballymena, Northern Ireland BT43 6DA, UK
Aldeyab et al.
been used as an indicator for antibiotic selection density (the proportion of patients receiving antibiotics in the hospital) and the selection pressure. However, this measure does not provide information on the number of patients actually exposed to antibiotics, possibly indicating the need for additional measurement units that will allow meaningful understanding of hospital antibiotic use.12 Whereas there is a consensus regarding the role of antibiotic stewardship as a modality to improve patient care and healthcare outcomes,7 – 10 the most appropriate method for quantifying and reporting hospital antibiotic use is still debatable.12 – 14 Recently, an investigation with the objective of evaluating the effects of age-adjusted comorbidity and alcohol-based hand rub on monthly hospital antibiotic usage was carried out in the Antrim Area Hospital (Northern Ireland). The results showed that monthly antibiotic use was significantly related to the mean age-adjusted comorbidity index, a non-modifiable factor, indicating the possible utility of this variable in explaining hospital antibiotic use.15 The objective of this study was to determine whether adjusting the denominator of the common hospital antibiotic use measurement unit (DDDs/100 bed-days) by including age-adjusted comorbidity score (100 bed-days/age-adjusted comorbidity score) would result in more accurate and meaningful assessment of hospital antibiotic use. The utility of this method for better informed hospital antibiotic stewardship and benchmarking was explored.
usage datasets (two study periods), i.e. antibiotic usage for 2004 –07 and 2008– 11, were constructed. The determination of cut-off dates, at the end of period 1 and the start of period 2 in this second study phase, was based on the timing of an enhanced antibiotic stewardship programme, which was introduced in the study site hospital in January 2008. An overview of the study design and characteristics is provided in Figure S1 (available as Supplementary data at JAC Online).
Methods
Enhanced antibiotic stewardship
The study was carried out in the Antrim Area Hospital [Northern Health and Social Care Trust (NHSCT), Northern Ireland], a 426 bed district general teaching hospital serving a population of 420000. The hospital provides all acute, general medical and surgical services, supports a range of outpatient facilities and acts as a centre for the coordination of health service provision throughout a defined geographical area in Northern Ireland. The study was ecological in design (i.e. an observational study in which data are measured at the group level) and consisted of two phases. Firstly, a model aimed at measuring the association between the monthly sum of age-adjusted comorbidity score (a proxy measure for patient case mix) and monthly antibiotic use was developed. A proxy measure is an indirect measure or indicator that represents a certain outcome in the absence of the availability of a direct or exact measure. In developing this model, other predictors {i.e. proxy measures for infection control practices [alcohol-based hand rub] and clinical pharmacy activity [clinical pharmacist full-time equivalents (FTEs)]} were included. The model was built for the period January 2008 to June 2012. Secondly, and for the purpose of conducting internal benchmarking, two antibiotic
Monthly antibiotic usage data for hospitalized adult patients were obtained from the pharmacy computer systems retrospectively and were converted into DDDs. Monthly data for alcohol-based hand rub (measured in L), clinical pharmacist FTEs and occupied patient bed-days were determined retrospectively from the hospital information system. Monthly hospital adult admissions, with their corresponding age-adjusted comorbidity data, were obtained from the hospital episode statistics. Age-adjusted comorbidity scores were calculated by combining age with the Charlson comorbidity index.16 – 18 The Charlson score is a weighted index that includes 19 diagnoses, and each diagnosis is weighted according to a 1 year relative risk of mortality. The weighted index reflects the severity of each comorbid disease.16 – 18 With age-adjusted comorbidity, an additional score is added for age (i.e. one point is added for each decade over the age of 40 years). Following this, the summation of the monthly patient age-adjusted Charlson comorbidity score was calculated for inclusion in the analysis. The collection of data included the intensive care unit (one eight-bed unit). The paediatric department was excluded from the study.
Following a Clostridium difficile infection (CDI) outbreak, the Northern Health and Social Care Trust devised an antibiotic policy to minimize the use of high-risk antibiotics (second-generation cephalosporins, thirdgeneration cephalosporins, fluoroquinolones and clindamycin; January 2008), and classified other antibiotics as medium-risk (amoxicillin/clavulanic acid and macrolides) or low-risk antibiotics (all remaining antibiotics; September 2008).19,20 Clinical staff were encouraged to adhere to the hospital policy and their compliance with the hospital policy was observed and recorded using a standardized procedure form.21 The use of antimicrobials not included in the policy was monitored through exemption forms, which required authorization by consultant physicians. The exemption forms were assessed by the antimicrobial management team (AMT) as appropriate or inappropriate with a written explanation. Results were directly shared with the prescribing physicians.
Antibiotic measure calculations Antibiotic use was normalized per 100 bed-days and per 100 age-adjusted comorbidity score separately. A modified antibiotic usage measuring unit was devised to take into account both patient bed-days and patient
Table 1. Estimates of time-series analysis model for monthly antibiotic use (R 2 ¼0.58) Term Sum of age-adjusted comorbidity scores Alcohol-based hand rub (L/100 bed-days) Clinical pharmacist FTEs (100 bed-days) Moving average a
Lag timea
Coefficient (SEM)b
T ratio
P value
0 2 3 5
0.927000 (0.244047) 20.085489 (0.039737) 20.487303 (0.155512) 20.625627 (0.124044)
3.798450 22.151364 23.133538 25.043594
0.0004 0.0370 0.0031 ,0.0001
Represents the delay necessary to observe the effect (in months). Indicates the size and direction of the effect.
b
2 of 10
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Setting and study design
Hospital data
JAC
A modified method for measuring antibiotic use
Clinical pharmacist FTEs
May-12
Mar-12
Jan-12
Nov-11
Sep-11
Jul-11
May-11
Mar-11
Jan-11
Nov-10
Antibiotic use
0.6
180 160
0.5
140
0.4
120 100
0.3
80
0.2
60 40
0.1 Mar-12
May-12
Jan-12
Sep-11
Nov-11
Jul-11
May-11
Mar-11
Jan-11
Sep-10
Nov-10
Jul-10
Mar-10
May-10
Jan-10
Sep-09
Nov-09
Jul-09
May-09
Jan-09
Mar-09
Nov-08
Sep-08
Jul-08
May-08
Mar-08
20 Jan-08
0
Antibiotic use, DDDs/100 bed-days
May-12
Jan-12
Mar-12
Nov-11
Jul-11
Sep-11
May-11
Jan-11
Mar-11
Nov-10
Sep-10
Jul-10 Jul-10
May-10
Jan-10
Mar-10
Nov-09
Jul-09
Sep-09
May-09
Jan-09
Mar-09
Nov-08
Sep-08
Jul-08
May-08
(c)
180 160 140 120 100 80 60 40 20 0
Antibiotic use
Sep-10
Mar-10
May-10
Jan-10
Sep-09
Nov-09
Jul-09
Mar-09
May-09
Jan-09
Sep-08
Nov-08
Jul-08
May-08
Mar-08
Jan-08 Jan-08
Alcohol-based hand rub 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
180 160 140 120 100 80 60 40 20 0
Antibiotic use, DDDs/100 bed-days
Antibiotic use
9000 8000 7000 6000 5000 4000 3000 2000 1000 0
Mar-08
Alcohol-based hand rub, L/100 bed-days
Time-series analysis methods, including the use of linear transfer function techniques, were applied to study the relationship of age-adjusted comorbidity index, alcohol-based hand rub and clinical pharmacy activity to
Sum of age-adjusted comorbidity scores
(b)
Clinical pharmacist FTEs/100 bed-days
Statistical analysis
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Sum of age-adjusted comorbidity scores
(a)
antibiotics involved in the restriction policy (i.e. second-generation cephalosporins, third-generation cephalosporins, fluoroquinolones, clindamycin, penicillins with extended spectrum, and combinations of penicillins including b-lactamase inhibitors) was considered. Antibiotic comparative usage rate was also calculated for the latter and presented separately.
0
Antibiotic use, DDDs/100 bed-days
comorbidity score, i.e. DDDs/100 bed-days/age-adjusted comorbidity {[number of antibiotic DDDs in a specific month/(patient days×sum of age-adjusted comorbidity score in that month)]×100}. Antibiotic comparative usage rate was calculated by dividing the average monthly antibiotic use during the second study period by the average monthly antibiotic use during the first study period. The latter was carried out considering all antibiotics, i.e. total antibiotics for systemic use (J01). However, to address possible effects with regard to changes in antibiotic volumes due to the restriction policy (i.e. an increase in amoxicillin and amoxicillin/clavulanic acid agents, which may lead to an apparent increase in antibiotic use regardless of increased patient activity), another analysis that excluded all
Figure 1. Monthly antibiotic use versus (a) sum of age-adjusted comorbidity score, (b) use of alcohol-based hand rub and (c) clinical pharmacist FTEs, January 2008 to June 2012, Antrim Area Hospital.
3 of 10
Aldeyab et al.
explain the use of antibiotics (see the Statistical analysis section in the Supplementary data available at JAC Online).22 – 28 The series were checked for stationarity (i.e. having a constant mean and variance) using the augmented Dickey– Fuller test for unit roots.26 All variables were logarithmically transformed and first-order differencing was applied. The coefficient of determination (R 2), which corresponds to the percentage of the variance of the observed time-series explained by the model, was determined. Significance tests for parameter estimates were used to eliminate the unnecessary terms in the model. All analyses were performed using EViews 6 software (QMS, Irvine, CA, USA) and the SCA Statistical System software (River Forest, IL, USA). All results are reported based on the EViews software estimation summaries.
Results Analysis of data showed that monthly antibiotic use had significant positive relationships with the sum of age-adjusted comorbidity score (concomitant effect; coefficient, 0.927; P¼ 0.0004; Table 1). This means that a level change in the sum of age-adjusted comorbidity score by 1% would cause a level change in antibiotic use by 0.927% (direct relationship). The results also showed that there was a negative relationship between antibiotic use and (i) alcohol-based hand rub use (2 months delay; coefficient, 20.085; P¼ 0.0370; Table 1) and (ii) clinical pharmacist FTEs (3 months delay; coefficient, 20.487303; P ¼ 0.0031; Table 1). This means that a level change in alcohol-based hand rub use and clinical pharmacist FTEs by 1% would cause a level change in antibiotic use by 0.085% and 0.487% (inverse relationship), respectively. Based on examination of the residuals of the final model, adjustment was made for two outliers (i.e. February 2009 and March 2011) by inclusion of binary dummy variables (i.e. a value of 1 if an outlier is present; 0 otherwise.) in the model. For the final model, R 2 was 0.58, i.e. 58% of the variation in the monthly antibiotic use in the study site hospital was explained by the model. Graphs for antibiotic use versus sum of age-adjusted comorbidity score, alcohol-based hand rub and clinical pharmacist FTEs are shown in Figure 1.
Modified measure and benchmarking Age-adjusted comorbidity scores were calculated for 167576 admissions. Distributions of comorbidity and age-adjusted comorbidity score are shown in Table 2. Data showed a strong association between the monthly sum of age-adjusted comorbidity score and the monthly number of admissions (Pearson’s correlation, 0.92, P,0.0001). Over the two study periods (2004 –07 and 2008 –11), the average observed monthly antibiotic use was 8062.7 and 11407.6 DDDs, respectively, with an increasing trend in antibiotic use being observed (P,0.0001; Tables 3 and 4). Similar increasing trends were observed for both study periods when antibiotic use was normalized per 100 bed-days (P,0.0001; Table 4). Characteristics and trends in monthly antibiotic use are shown in Tables 3 and 4. The average monthly sums of age-adjusted comorbidity scores for both study periods were 4194.3 and 5061.5, respectively; an increasing trend was observed (P,0.0001; Table 4). When antibiotic use was normalized per age-adjusted comorbidity and per bed-days/age-adjusted comorbidity, a significant increasing trend was observed for the period 2004–07 (P,0.0001 and
4 of 10
2004– 07 (n¼76479) frequency (%)
number of patients
frequency (%)
Comorbidity score 0 51 174 1 12 874 2 7092 3 2400 ≥4 2939
66.9 16.8 9.3 3.1 3.8
56393 17235 9513 3832 4124
61.9 18.9 10.4 4.2 4.5
Age-adjusted comorbidity score 0 27 653 1 7194 2 6368 3 6705 4 7867 5 8169 6 5921 7 3079 8 1616 ≥9 1907
36.2 9.4 8.3 8.8 10.3 10.7 7.7 4.0 2.1 2.5
28306 8898 7514 8111 9463 10637 8117 4785 2490 2776
31.1 9.8 8.2 8.9 10.4 11.7 8.9 5.3 2.7 3.0
Term
Score
number of patients
2008 –11 (n¼91 097)
P,0.0027, respectively). However, antibiotic use during the period 2008–11 for both DDDs/age-adjusted comorbidity score and DDDs/100 bed-days/age-adjusted comorbidity score remained stable (P,0.1311 and P,0.6857, respectively; Table 4). Plots for monthly antibiotic use (DDDs) versus use of antibiotics normalized for different denominators (i.e. bed-days, age-adjusted comorbidity score and bed-days/age-adjusted comorbidity score) are presented in Figure 2. Graphs for antibiotic use DDDs/100 bed-days versus (i) antibiotic use DDDs/age-adjusted comorbidity scores and (ii) DDDs/100 bed-days/age-adjusted comorbidity score are shown in Figure 3. The average observed monthly antibiotic use during the second study period was higher (115.8 DDDs/100 bed-days, 2008 –11) than during the first study period (88.7 DDDs/100 bed-days; Table 3). This has contributed to a comparative usage rate of 1.31, i.e. after considering patient density, antibiotic use during the second period of the study was 31% higher than that during the first study period (Table 5). Normalizing antibiotic use per age-adjusted comorbidity score resulted in a comparative usage rate of 1.06, i.e. after taking into account the effect of patient case mix, antibiotic use during the second period of the study was 6% higher than that during the first study period. Considering a denominator that took into account both patient density and patient case mix characteristics led to a comparative usage rate of 0.98 (Table 5), i.e. the average antibiotic use was 2% lower in the second study period. Importantly, the latter comparative usage rate is independent of differences in patient density and case mix characteristics between the two studied populations. The exclusion of specific antibiotic agents showed similar patterns of decreasing overall antibiotic use when considering the studied adjustment units (Table 5).
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Time-series analysis
Table 2. Distribution of comorbidity and age-adjusted comorbidity score in Antrim Area Hospital, 2004– 11
Average monthly antibiotic use, 2004 – 07
DDDs/100 bed-days
111.1 2.1 280.36
1.22 0.22 3.09
0.0268 0.0005 0.0676
0.0003 0.000005 0.0007
1.38 0.25 3.50
600.1 1.8 1465.46
5.98 0.22 14.61
0.1019 0.0003 0.2479
0.0010 0.000003 0.0025
193.33
2.11
0.0457
0.0005
2.39
205.77
2.09
0.0364
0.0004
1.80
371.09
4.07
0.0887
0.0010
4.61
567.3
5.71
0.0993
0.0010
4.93
3101.36
33.96
0.7373
0.0081
38.46
3994.11
40.99
0.7316
0.0075
35.39
52.25
0.58
0.0126
0.0001
0.66
20.62
0.21
0.0038
0.00004
0.18
295.64
3.26
0.0722
0.0008
3.69
14.8
0.15
0.0028
0.00003
0.13
63.53
0.7
0.0154
0.0002
0.79
43.45
0.44
0.0078
0.0001
0.38
1.66 77.46 187.35
0.02 0.83 2.06
0.0004 0.0176 0.0450
0.000004 0.0002 0.0005
0.02 0.94 2.33
93.02 248.61 334.18
0.95 2.51 3.38
0.0167 0.0437 0.0583
0.0002 0.0004 0.0006
0.82 2.17 2.92
24.71
0.27
0.0060
0.0001
0.31
53.68
0.53
0.0090
0.0001
0.46
1322.75 65.4 0.63 94.54 939.57 163.3 106.83
14.4 0.71 0.01 1.04 10.31 1.78 1.17
0.3108 0.0154 0.0002 0.0226 0.2239 0.0384 0.0257
0.0034 0.0002 0.000002 0.0003 0.0025 0.0004 0.0003
16.31 0.80 0.01 1.18 11.68 2.02 1.33
1909.12 72.3 1.04 253.09 197.25 518.2 155.49
19.52 0.73 0.01 2.55 1.99 5.21 1.58
0.3466 0.0124 0.0002 0.0440 0.0348 0.0902 0.0279
0.0036 0.0001 0.000002 0.0004 0.0004 0.0009 0.0003
536.6
5.89
0.1282
0.0014
6.67
471.63
4.69
0.0829
0.0008
4.05
0.4
0.0087
0.0001
0.45
138.18
1.38
0.0232
0.0002
1.19
0.59 88.7
0.0082 1.9179
0.00009 0.0211
0.67 100.00
58.32 11407.55
0.38 115.8
0.0102 2.0320
0.00010 0.0207
36.73 34.48 8062.67
DDDs/100 bed-days/ age-adjusted comorbidity score
% of J01 use (based on DDDs/100 bed-days)
DDDs
DDDs/100 bed-days
DDDs/ age-adjusted comorbidity score
DDDs/100 bed-days/ age-adjusted comorbidity score
% of J01 use (based on DDDs/100 bed-days) 5.16 0.19 12.62
16.86 0.63 0.01 2.20 1.72 4.50 1.36
0.33 100.00
JAC
5 of 10
Tetracyclines (J01A) Amphenicols (J01B) Penicillins with extended spectrum (J01CA) b-Lactamase-sensitive penicillins (J01CE) b-Lactamase-resistant penicillins (J01CF) Combinations of penicillins including b-lactamase inhibitors (J01CR) First-generation cephalosporins (J01DB) Second-generation cephalosporins (J01DC) Third-generation cephalosporins (J01DD) Monobactams (J01DF) Carbapenems (J01DH) Trimethoprim and derivatives (J01EA) Combination of sulphonamides and trimethoprim (J01EE) Macrolides (J01FA) Lincosamides (J01FF) Streptomycins (J01GA) Aminoglycosides (J01GB) Fluoroquinolones (J01MA) Glycopeptide (J01XA) Steroid antibacterials (J01XC) Imidazole derivatives (J01XD) Nitrofuran derivatives (J01XE) Other antibacterials (J01XX) Antibiotics for systemic use, total (J01)
DDDs
DDDs/ age-adjusted comorbidity score
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Antimicrobial class (ATC group)
Average monthly antibiotic use, 2008 –11
A modified method for measuring antibiotic use
Table 3. Characteristics of monthly antimicrobial use in Antrim Area Hospital, 2004 –11
Aldeyab et al.
Table 4. General characteristics and trends in antibiotic use adjusted per patient days and per age-adjusted comorbidity score, Antrim Area Hospital, 2004– 11 Trend (2004– 07)
Trend (2008–11)
Term
coefficient
P value
coefficient
P value
Admissions Patient bed-days Sum of age-adjusted comorbidity scores Total antibiotic DDDs Antibiotic use (DDDs/100 bed-days) Antibiotic use (DDDs/age-adjusted comorbidity score) Antibiotic use (DDDs/100 bed-days/age-adjusted comorbidity score)
9.296407 21.80346 23.60427 104.7636 0.924069 1.391882 0.000104
,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 0.0027
5.977366 24.77079 44.18167 112.3938 0.834901 0.388966 20.000011
,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 0.1311 0.6857
The process of developing policies and identifying targets for quality improvement in healthcare systems depends largely on the adequate collection and interpretation of a range of quality indicators. Different sources of data, with several advantages and limitations, are available for quantifying hospital antibiotic use, including pharmacy purchases, patient prescription profiles and pharmacy deliveries to wards. The latter data have been widely used for surveillance since they are easily accessible and readily available. The use of DDDs as a unit of measurement provides a reflection of the quantitative volumes of antibiotic usage and helps to enable comparisons between different institutions and countries. However, concerns have been expressed over which denominator should be used in order to express hospital antibiotic usage.12 – 14 More importantly, it has been shown that antibiotic use varies widely between different hospitals,29,30 with several studies suggesting excessive use, thus making it difficult to conduct proper benchmarking across hospitals.3,31,32 High antibiotic use in a particular setting is often justified by clinicians as relating to differences in patient morbidities.33 – 35 The measurement of patient comorbidity may therefore play a major role in explaining variations in hospital antibiotic use and subsequently facilitating the conduct of more realistic benchmarking. The results of the present research showed that monthly antibiotic use was significantly related to the age-adjusted comorbidity score. The importance of measuring comorbidity lies in its ability to predict morbidity, mortality, cost, hospitalization and the burden of the disease.16 – 18 As such, this factor can provide risk-adjustment criteria for patient case mix purposes and can have a vital role in explaining trends and variation in hospitals’ antibiotic use.15 Interestingly, changing age-adjusted comorbidity score had a much higher impact within the model when compared with the two other significant variables (Table 1). The latter observation, together with the fact that the role of patients’ age-adjusted comorbidity as a proxy measure for the disease burden has been proven,16 – 18 suggested the possible utility of this non-modifiable variable as an appropriate denominator for the calculation and presentation of hospital antibiotic use. The findings of the study also showed that an increase in the use of alcohol-based hand rub was associated with a decrease in antibiotic use. Antibiotics administered to patients in hospital are used to treat community- and healthcare-acquired infections.
6 of 10
The use of alcohol-based hand rub was shown to be an effective method to reduce the spread and acquisition of healthcareacquired infections in hospitals; the latter are associated with significant increases in length of hospital stay and antibiotic treatment.24,36,37 It is interesting to note that variations in hospital pathogens’ prevalence might be followed by similar variations in pathogens’ prevalence in the surrounding community. MacKenzie et al.38 showed that the monthly prevalence of methicillinresistant Staphylococcus aureus (MRSA) in the community was strongly related to the monthly MRSA prevalence observed 1 month previously in the hospital, suggesting that the reason for the increase in MRSA prevalence in the community was a hospital MRSA outbreak. Thus, increasing use of alcohol-based hand rub, which is a proxy measure for adequately practised infection control, might be leading to a subsequent reduction in hospital antibiotic use. However, further prospective research is needed to assess the impact of enhancing different infection control practices (e.g. alcohol-based hand rub, surveillance activity, education, etc.) on antibiotic use in healthcare settings. In addition, the number of clinical pharmacist FTEs was used as a proxy measure for clinical pharmacy activity. The results showed that an increase in the clinical pharmacy services provided across the hospital was associated with a decrease in antibiotic use. This is in line with other local research that showed that medicine appropriateness is improved via improved medication management by the clinical pharmacy team.39 In Antrim Area Hospital, a major CDI outbreak occurred in 2008.19 The CDI outbreak was ended following the implementation of an action plan that involved improving communication, infection control practices, environmental hygiene and surveillance, and enhanced antibiotic stewardship.19 The latter involved the implementation, led by the AMT, of prospective auditing of antimicrobial use with direct interaction and feedback to the prescriber, formulary restriction and a preauthorization strategy.19,21 With such high awareness of the problem of antibiotic use and resistance in this hospital, the statistically significant increasing usage trend and the observed 31% increase in average antibiotic use (DDDs/100 bed-days) during the implementation of enhanced antibiotic stewardship (second study period; Tables 4 and 5) contradicted the natural expected outcomes. Nevertheless, adjusting antibiotic usage DDDs per age-adjusted comorbidity score resulted in only a 6% antibiotic use increase during the enhanced antibiotic stewardship period, highlighting the value of considering patient
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Discussion
JAC
A modified method for measuring antibiotic use
Antibiotic use, DDDs/100 bed-days
Sep-11
May-11
Jan-11
Sep-10
Jan-10
May-10
Sep-09
Jan-09
May-09
Sep-08
May-08
Jan-08
Sep-07
May-07
Jan-07
Sep-06
Jan-06
May-06
May-05
Jan-05
Sep-04
May-04
18 000 16 000 14 000 12 000 10 000 8000 6000 4000 2000 0
2.5 2 1.5 1 0.5
Antibiotic use, DDDs/ age-adjusted comorbidity
3
Sep-11
May-11
Jan-11
Sep-10
Jan-10
May-10
Sep-09
May-09
Jan-09
Sep-08
May-08
Jan-08
Sep-07
May-07
Antibiotic use, DDDs/100 bed-days/age-adjusted comorbidity 0.035
18 000 16 000 14 000 12 000 10 000 8000 6000 4000 2000 0
0.03 0.025 0.02 0.015 0.01 0.005 Sep-11
May-11
Jan-11
Sep-10
May-10
Jan-10
Sep-09
May-09
Jan-09
Sep-08
May-08
Jan-08
Sep-07
May-07
Jan-07
Sep-06
Jan-06
May-06
Sep-05
May-05
Jan-05
Sep-04
May-04
Jan-04
0
Antibiotic use, DDDs/100 bed-days/ age-adjusted comorbidity
Antibiotic use, DDDs
Jan-07
Sep-06
Jan-06
May-06
Sep-05
Jan-05
May-05
Sep-04
May-04
0
(c)
Antibiotic use, DDDs
180 160 140 120 100 80 60 40 20 0
Antibiotic use, DDDs/age-adjusted comorbidity
Antibiotic use, DDDs
Jan-04
Antibiotic use, DDDs
(b)
Antibiotic use, DDDs/100 bed-days
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Jan-04
Antibiotic use, DDDs
18 000 16 000 14 000 12 000 10 000 8000 6000 4000 2000 0
Sep-05
Antibiotic use, DDDs
(a)
Figure 2. Monthly hospital antibiotic use adjusted for different denominators, January 2004 to December 2011, Antrim Area Hospital. January 2008 represents the introduction of the enhanced antibiotic stewardship.
case mix when expressing hospital antibiotic use. Such findings were consistent with the non-significant antibiotic use trend, which was observed during the second study period (DDDs/ age-adjusted comorbidity; Table 4). However, taking into account a more robust unit of measurement (i.e. DDDs/100 bed-days/ age-adjusted comorbidity score) has shown a reduction in the average antibiotic use by 2% (during the second study period). This reduction in antibiotic use, together with the significant reduction in high-risk antibiotics (described elsewhere),19,20 demonstrates
the success of this antibiotic stewardship in the study site hospital and confirms its value in the management of hospital antibiotic use. The study has several strengths. Firstly, the employed statistical regression technique has taken cognizance of the problem of dependency existing between consecutive observations and for identifying lag effects between predictors and the outcome series, thus measuring significant relationships accurately.40 Secondly, the study used robust risk-adjustment criteria for the patient case
7 of 10
Aldeyab et al.
3
160
2.5
140 120
2
100
1.5
80 60
1
40
0.5
20
Antibiotic use, DDDs/ age-adjusted comorbidity
180
0 Sep-11
May-11
Jan-11
Sep-10
May-10
Jan-10
Sep-09
May-09
Jan-09
Sep-08
May-08
Jan-08
Sep-07
May-07
Antibiotic use, DDDs/100 bed-days/age-adjusted comorbidity
180
0.035
160
0.03
140
0.025
120 100
0.02
80
0.015
60
0.01
40
0.005
20
0
0
Antibiotic use, DDDs/100 bed-days/ age-adjusted comorbidity
Jan-07
Sep-06
May-06
Jan-06
Sep-05
May-05
Jan-05
Sep-04
May-04
Antibiotic use, DDDs/100 bed-days
Figure 3. Monthly hospital antibiotic use (DDDs/100 bed-days) versus (a) antibiotic use (DDDs/100 age-adjusted comorbidity score) and (b) antibiotic use (DDDs/100 bed-days/age-adjusted comorbidity score), January 2004 to December 2011, Antrim Area Hospital. January 2008 represents the introduction of the enhanced antibiotic stewardship. Table 5. Comparative antibiotic usage rates adjusted per patient days and per age adjusted comorbidity scores, Antrim Area Hospital, 2004–11 Average use Term
2004–07
2008 –11a
Comparative rate
Estimates including all antibiotics antibiotic use (DDDs/100 bed-days) antibiotic use (DDDs/age-adjusted comorbidity) antibiotic use (DDDs/100 bed-days/age-adjusted comorbidity)
88.7 1.918 0.0211
115.8 2.032 0.0207
1.31 1.06 0.98
Estimates excluding specific antibioticsb antibiotic use (DDDs/100 bed-days) antibiotic use (DDDs/age-adjusted comorbidity) antibiotic use (DDDs/100 bed-days/age-adjusted comorbidity)
36.66 0.786 0.0086
56.90 0.995 0.0101
1.55 1.27 1.17
a
This period represents the introduction of the enhanced antibiotic stewardship. Antibiotics excluded are second-generation cephalosporins, third-generation cephalosporins, fluoroquinolones, clindamycin, penicillins with extended spectrum, and combinations of penicillins including b-lactamase inhibitors.
b
mix (i.e. the Charlson index), which is considered as the gold standard to assess comorbid risk in clinical research and uses readily available hospital databases.16 – 18 Thirdly, analysis showed a
8 of 10
strong association between the monthly sum of age-adjusted comorbidity score and the number of monthly admissions. This is of importance since age-adjusted comorbidity score can reflect
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Jan-04
0
(b) Antibiotic use, DDDs/100 bed-days
Antibiotic use, DDDs/age-adjusted comorbidity
Antibiotic use, DDDs/100 bed-days
Jan-04 Apr-04 Jul-04 Oct-04 Jan-05 Apr-05 Jul-05 Oct-05 Jan-06 Apr-06 Jul-06 Oct-06 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10 Jan-11 Apr-11 Jul-11 Oct-11
Antibiotic use, DDDs/100 bed-days
(a)
JAC
A modified method for measuring antibiotic use
clinical diagnosis.17 In addition to expressing antibiotic use per 100 bed-days, hospitals should be encouraged to express their antibiotic use per 100 bed-days/age-adjusted comorbidity score in order to allow the calculation of a comparative usage rate, while taking into account the effect of patient case mix. Following the determination of the percentage difference in antibiotic use, based on the identified antibiotic comparative usage rate, antibiotic policy makers should investigate the reasons behind a high or low antibiotic usage rate, thus identifying targets for quality improvement in the former and presenting best antibiotic prescribing practice for adoption by others in the latter.
Funding The study was carried out as part of our routine work.
Transparency declarations None to declare.
Supplementary data Figure S1 and a Statistical analysis section are available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/).
References 1 Patrick DM, Hutchinson J. Antibiotic use and population ecology: how you can reduce your “resistance footprint”. CMAJ 2009; 180: 416–21. 2 Goossens H. Antibiotic consumption and link to resistance. Clin Microbiol Infect 2009; 15 Suppl 3: 12– 5. 3 Hecker MT, Aron DC, Patel NP et al. Unnecessary use of antimicrobials in hospitalized patients: current patterns of misuse with an emphasis on the antianaerobic spectrum of activity. Arch Intern Med 2003; 163: 972–8. 4 Gould IM. Antibiotic policies to control hospital-acquired infection. J Antimicrob Chemother 2008; 61: 763–5. 5 Wilcox MH, Dave J. The cost of hospital-acquired infection and the value of infection control. J Hosp Infect 2000; 45: 81 –4. 6 Gould IM. The clinical significance of methicillin-resistant Staphylococcus aureus. J Hosp Infect 2005; 61: 277– 82. 7 Fishman N. Antimicrobial stewardship. Am J Med 2006; 119 Suppl 1: S53– 61. 8 Griffith M, Postelnick M, Scheetz M. Antimicrobial stewardship programs: methods of operation and suggested outcomes. Expert Rev Anti Infect Ther 2012; 10: 63–73. 9 Dellit TH, Owens RC, McGowan JE Jr et al. Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis 2007; 44: 159– 77. 10 Davey P, Brown E, Fenelon L et al. Interventions to improve antibiotic prescribing practices for hospital inpatients. Cochrane Database Syst Rev 2005; issue 4: CD003543. 11 WHO Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC Classification and DDD Assignment, 2012. Oslo, 2011. http://www. whocc.no/atc_ddd_publications/guidelines/ (20 September 2012, date last accessed).
9 of 10
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
monthly changes in hospital admissions. Therefore, expressing antibiotic use in terms of age-adjusted comorbidity score may possibly account for variations in admission rates and thus provide a meaningful understanding and interpretation of hospital antibiotic use.12 Fourthly, antibiotic use and the age-adjusted comorbidity score were calculated for all hospitalized adult patients; thus, information and selection bias were unlikely. Finally, adjustment was made for the confounding effect of different healthcare systems and hospital policies since the study used an internal benchmarking approach in the same study site hospital. However, the study has the following limitations. Firstly, although it is possible to argue that the increase in antibiotic use during the second study period was an apparent increase due to the increase in amoxicillin and amoxicillin/clavulanic acid, the exclusion of these agents showed similar patterns of decreasing overall antibiotic use when considering the presented adjustment measures (Table 5). This indicates that the increase in antibiotic use was genuine and possibly related to different reasons. Secondly, it is possible that other potential drivers for increased antibiotic use could be related to hospital infection rates and duration of treatment. Research from the study site Trust showed that the intervention introduced in 2008 (i.e. restricting high-risk antibiotics) had a positive impact on reducing the incidence rates of specific hospital pathogens (i.e. CDI pathogens, extended-spectrum b-lactamase-producing bacteria and MRSA).19,20,41,42 In addition, the quality of antibiotic prescribing and adherence to guidelines was continuously scrutinized and improved.21,43 For example, adherence to the Trust antibiotic policy was improved after 200821,43 compared with prior periods.44 Thirdly, it was not possible to determine accurate data for the entire study period with regard to other types of infection (e.g. surgical site infections, ventilator-associated pneumonia) to assess its relation with antibiotic use and age-adjusted comorbidity score; this requires further prospective research in different hospitals. However, the presence of multiple comorbidities is an indicator of patients in suboptimal health who may be unable to mount an effective immune response to bacterial challenge, and as such might facilitate patient colonization/infection with pathogens; the age-adjusted comorbidity score provides a convenient measure for the latter. Fourthly, the Charlson index might be limited in its applicability to complex processes of care and functional outcomes since it does not incorporate physiological data to assess the severity of comorbid conditions and does not include functional impairment attributable to pre-existing comorbid conditions.17 Finally, further research is needed to compare antibiotic use by different hospital departments, and to evaluate the proposed benchmarking approach using external groups and involving several hospitals across different countries with high and low volumes of antibiotic use. In conclusion, the proposed modified antibiotic use unit of measurement provided an innovative approach to the measurement of actual trends in antibiotic use and the comparison of variations in antibiotic prescribing between hospitals, taking account of the effect of patient case mix. It facilitated the evaluation of antibiotic stewardship more accurately and can enable more informative benchmarking between hospitals. It is interesting to note that other identified predictors are considered modifiable factors, thus leaving room for quality improvement in antibiotic use. Age-adjusted comorbidity score calculations (utilizing the Charlson index) are relatively easy to apply, based on availability of
Aldeyab et al.
12 Filius PM, Liem TB, van der Linden PD et al. An additional measure for quantifying antibiotic use in hospitals. J Antimicrob Chemother 2005; 55: 805–8. 13 Cooke J. Trends in antimicrobial prescribing. J Antimicrob Chemother 2000; 46: 639. 14 Ibrahim OM, Polk RE. Benchmarking antimicrobial drug use in hospitals. Expert Rev Anti Infect Ther 2012; 10: 445– 57. 15 Aldeyab MA, McElnay JC, Scott MG et al. Hospital antibiotic use and its relationship to age-adjusted co-morbidity and alcohol-based hand rub consumption. Epidemiol Infect 2013; 9: doi:10.1017/S0950268813001052. 16 Charlson ME, Pompei P, Ales KL et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40: 373– 83.
Retrospective Data Collection (1997– 2002). J Antimicrob Chemother 2006; 58: 159–67. 30 Lesch CA, Itokazu GS, Danziger LH et al. Multi-hospital analysis of antimicrobial usage and resistance trends. Diagn Microbiol Infect Dis 2001; 41: 149–54. 31 Kislak JW, Eickhoff TC, Finland M. Hospital-acquired infections and antibiotic usage in the Boston City Hospital-January, 1964. N Engl J Med 1964; 271: 834–5. 32 Werner NL, Hecker MT, Sethi AK et al. Unnecessary use of fluoroquinolone antibiotics in hospitalized patients. BMC Infect Dis 2011; 11: 187. 33 Kuster SP, Ruef C, Bollinger AK et al. Correlation between case mix index and antibiotic use in hospitals. J Antimicrob Chemother 2008; 62: 837– 482. 34 Rogues AM, Placet-Thomazeau B, Parneix P et al. Use of antibiotics in hospitals in south-western France. J Hosp Infect 2004; 58: 187–92.
18 Peterson JC, Paget SA, Lachs MS et al. The risk of comorbidity. Ann Rheum Dis 2012; 71: 635– 7.
35 Polk RE, Hohmann SF, Medvedev S et al. Benchmarking risk-adjusted adult antibacterial drug use in 70 US academic medical center hospitals. Clin Infect Dis 2011; 53: 1100– 10.
19 Aldeyab MA, Devine MJ, Flanagan P et al. Multihospital outbreak of Clostridium difficile ribotype 027 infection: epidemiology and analysis of control measures. Infect Control Hosp Epidemiol 2011; 32: 210– 9. 20 Aldeyab MA, Kearney MP, Scott MG et al. An evaluation of the impact of antibiotic stewardship on reducing the use of high-risk antibiotics and its effect on the incidence of Clostridium difficile infection in hospital settings. J Antimicrob Chemother 2012; 67: 2988 –96. 21 Conlon G, Aldeyab MA, McElnay JC et al. Improving and maintaining adherence with hospital antibiotic policies: a strategy for success. J Hosp Infect 2011; 77: 88– 9. 22 Box GEP, Jenkins GM. Time Series Analysis: Forecasting and Control. San Francisco: Holden Day, 1976. 23 Box GEP, Jenkins GM, Reinsel GC. Time Series Analysis: Forecasting and Control. 3rd edn. Englewood Cliff, NJ: Prentice Hall, 1994. 24 Aldeyab MA, Monnet DL, Lo´pez-Lozano JM et al. Modelling the impact of antibiotic use and infection control practices on the incidence of hospitalacquired methicillin-resistant Staphylococcus aureus: a time-series analysis. J Antimicrob Chemother 2008; 62: 593–600. 25 Aldeyab MA, Harbarth S, Vernaz N et al. Quasiexperimental study of the effects of antibiotic use, gastric acid-suppressive agents, and infection control practices on the incidence of Clostridium difficile-associated diarrhea in hospitalized patients. Antimicrob Agents Chemother 2009; 53: 2082– 8. 26 Dickey DA, Fuller WA. Distribution of estimators for autoregressive time series with a unit root. JASA 1979; 74: 427– 31. 27 Liu L-M, Hanssens DM. Identification of multiple-input transfer function models. Commun Stat 1982; A11: 297–314. 28 Liu L-M. Time Series Analysis and Forecasting, Second Edition. Chicago, IL: Scientific Computing Associates Corporation, 2006. 29 Vander Stichele RH, Elseviers MM, Ferech M et al. Hospital consumption of antibiotics in 15 European countries: results of the ESAC
10 of 10
36 Jarlier V, Trystram D, Brun-Buisson C et al. Curbing methicillin-resistant Staphylococcus aureus in 38 French hospitals through a 15-year institutional control program. Arch Intern Med 2010; 170: 552–9. 37 Vernaz N, Sax H, Pittet D et al. Temporal effects of antibiotic use and hand rub consumption on the incidence of MRSA and Clostridium difficile. J Antimicrob Chemother 2008; 62: 601–7. 38 MacKenzie FM, Lopez-Lozano JM, Monnet DL et al. Temporal relationship between prevalence of meticillin-resistant Staphylococcus aureus (MRSA) in one hospital and prevalence of MRSA in the surrounding community: a time-series analysis. J Hosp Infect 2007; 67: 225– 31. 39 Burnett KM, Scott MG, Fleming GF et al. Effects of an integrated medicines management program on medication appropriateness in hospitalized patients. Am J Health Syst Pharm 2009; 66: 854–9. 40 Shardell M, Harris AD, El-Kamary SS et al. Statistical analysis and application of quasi experiments to antimicrobial resistance intervention studies. Clin Infect Dis 2007; 45: 901–7. 41 Aldeyab MA, Harbarth S, Vernaz N et al. The impact of antibiotic use on the incidence and resistance pattern of extended-spectrum b-lactamase-producing bacteria in primary and secondary healthcare settings. Br J Clin Pharmacol 2012; 74: 171– 9. 42 Aldeyab MA, Scott MG, Kearney MP et al. Impact of an enhanced antibiotic stewardship on reducing methicillin-resistant Staphylococcus aureus in primary and secondary healthcare settings. Epidemiol Infect 2013; 5: 1– 7. 43 Aldeyab MA, Kearney MP, McElnay JC et al. A point prevalence survey of antibiotic prescriptions: benchmarking and patterns of use. Br J Clin Pharmacol 2011; 71: 293– 6. 44 Aldeyab MA, Elshibly SM, McElnay JC et al. An evaluation of compliance with an antibiotic policy in surgical wards at a general teaching hospital in Northern Ireland. Infect Control Hosp Epidemiol 2009; 30: 921– 2.
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
17 Tobacman JK. Assessment of comorbidity: a review. Clin Perform Qual Health Care 1994; 2: 23– 32.
J Antimicrob Chemother doi:10.1093/jac/dkt458
A modified method for measuring antibiotic use in healthcare settings: implications for antibiotic stewardship and benchmarking Mamoon A. Aldeyab1,2*, James C. McElnay1, Michael G. Scott2, William J. Lattyak3, Feras W. Darwish Elhajji4, Motasem A. Aldiab5, Fidelma A. Magee2, Geraldine Conlon1 and Mary P. Kearney6 1
*Corresponding author. Clinical and Practice Research Group, School of Pharmacy, Queen’s University Belfast, Belfast BT9 7BL, Northern Ireland, UK. Tel: +44-28-90972033; Fax: +44-28-90247794; E-mail: [email protected]
Received 22 June 2013; returned 8 September 2013; revised 13 September 2013; accepted 22 October 2013 Objectives: To determine whether adjusting the denominator of the common hospital antibiotic use measurement unit (defined daily doses/100 bed-days) by including age-adjusted comorbidity score (100 bed-days/ age-adjusted comorbidity score) would result in more accurate and meaningful assessment of hospital antibiotic use. Methods: The association between the monthly sum of age-adjusted comorbidity and monthly antibiotic use was measured using time-series analysis (January 2008 to June 2012). For the purposes of conducting internal benchmarking, two antibiotic usage datasets were constructed, i.e. 2004– 07 (first study period) and 2008– 11 (second study period). Monthly antibiotic use was normalized per 100 bed-days and per 100 bed-days/ age-adjusted comorbidity score. Results: Results showed that antibiotic use had significant positive relationships with the sum of age-adjusted comorbidity score (P¼ 0.0004). The results also showed that there was a negative relationship between antibiotic use and (i) alcohol-based hand rub use (P ¼ 0.0370) and (ii) clinical pharmacist activity (P ¼0.0031). Normalizing antibiotic use per 100 bed-days contributed to a comparative usage rate of 1.31, i.e. the average antibiotic use during the second period was 31% higher than during the first period. However, normalizing antibiotic use per 100 bed-days per age-adjusted comorbidity score resulted in a comparative usage rate of 0.98, i.e. the average antibiotic use was 2% lower in the second study period. Importantly, the latter comparative usage rate is independent of differences in patient density and case mix characteristics between the two studied populations. Conclusions: The proposed modified antibiotic measure provides an innovative approach to compare variations in antibiotic prescribing while taking account of patient case mix effects. Keywords: antibiotic measure, age-adjusted comorbidity index, time-series analysis, pharmacoepidemiology
Introduction Evidence for a relationship between antibiotic use/misuse and the development of antibiotic-resistant bacteria, at both the individual and the population level, has been documented.1 – 4 This has subsequently increased morbidity, mortality and incurred additional hospital costs.5,6 Antibiotic resistance is a multifactorial problem that requires a multifactorial solution, of which optimizing antibiotic use has been considered a key component.7 – 9 The latter has been formalized through the establishment of hospital antibiotic
stewardship programmes in a range of healthcare settings, which consist of a set of coordinated strategies to improve the use of antibiotic medications with the goal of enhancing patient health outcomes.7 – 10 The surveillance of hospital antibiotic use and resistance patterns is considered a key aspect of high-quality antibiotic stewardship implementation, evaluation and performance. To measure antibiotic use, the use of defined daily doses (DDDs) per 100 bed-days has been recommended by the WHO Collaborating Centre for Drug Statistics and Methodology.10,11 In healthcare settings, expressing antibiotic use as DDDs per 100 patient-days has
# The Author 2013. Published by Oxford University Press on behalf of the British Society for Antimicrobial Chemotherapy. All rights reserved. For Permissions, please e-mail: [email protected]
1 of 10
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Clinical and Practice Research Group, School of Pharmacy, Queen’s University Belfast, Belfast BT9 7BL, Northern Ireland, UK; 2Pharmacy and Medicines Management Centre, Northern Health and Social Care Trust, Ballymena BT43 6DA, Northern Ireland, UK; 3Scientific Computing Associates Corp., River Forest, IL, USA; 4Faculty of Pharmacy, Applied Science Private University, PO Box 166, Amman 11931, Jordan; 5Faculty of Computing and Information Technology – North Jeddah Branch, King Abdulaziz University, PO Box 80221, Jeddah 21589, Saudi Arabia; 6 Microbiology Department, Northern Health and Social Care Trust, Ballymena, Northern Ireland BT43 6DA, UK
Aldeyab et al.
been used as an indicator for antibiotic selection density (the proportion of patients receiving antibiotics in the hospital) and the selection pressure. However, this measure does not provide information on the number of patients actually exposed to antibiotics, possibly indicating the need for additional measurement units that will allow meaningful understanding of hospital antibiotic use.12 Whereas there is a consensus regarding the role of antibiotic stewardship as a modality to improve patient care and healthcare outcomes,7 – 10 the most appropriate method for quantifying and reporting hospital antibiotic use is still debatable.12 – 14 Recently, an investigation with the objective of evaluating the effects of age-adjusted comorbidity and alcohol-based hand rub on monthly hospital antibiotic usage was carried out in the Antrim Area Hospital (Northern Ireland). The results showed that monthly antibiotic use was significantly related to the mean age-adjusted comorbidity index, a non-modifiable factor, indicating the possible utility of this variable in explaining hospital antibiotic use.15 The objective of this study was to determine whether adjusting the denominator of the common hospital antibiotic use measurement unit (DDDs/100 bed-days) by including age-adjusted comorbidity score (100 bed-days/age-adjusted comorbidity score) would result in more accurate and meaningful assessment of hospital antibiotic use. The utility of this method for better informed hospital antibiotic stewardship and benchmarking was explored.
usage datasets (two study periods), i.e. antibiotic usage for 2004 –07 and 2008– 11, were constructed. The determination of cut-off dates, at the end of period 1 and the start of period 2 in this second study phase, was based on the timing of an enhanced antibiotic stewardship programme, which was introduced in the study site hospital in January 2008. An overview of the study design and characteristics is provided in Figure S1 (available as Supplementary data at JAC Online).
Methods
Enhanced antibiotic stewardship
The study was carried out in the Antrim Area Hospital [Northern Health and Social Care Trust (NHSCT), Northern Ireland], a 426 bed district general teaching hospital serving a population of 420000. The hospital provides all acute, general medical and surgical services, supports a range of outpatient facilities and acts as a centre for the coordination of health service provision throughout a defined geographical area in Northern Ireland. The study was ecological in design (i.e. an observational study in which data are measured at the group level) and consisted of two phases. Firstly, a model aimed at measuring the association between the monthly sum of age-adjusted comorbidity score (a proxy measure for patient case mix) and monthly antibiotic use was developed. A proxy measure is an indirect measure or indicator that represents a certain outcome in the absence of the availability of a direct or exact measure. In developing this model, other predictors {i.e. proxy measures for infection control practices [alcohol-based hand rub] and clinical pharmacy activity [clinical pharmacist full-time equivalents (FTEs)]} were included. The model was built for the period January 2008 to June 2012. Secondly, and for the purpose of conducting internal benchmarking, two antibiotic
Monthly antibiotic usage data for hospitalized adult patients were obtained from the pharmacy computer systems retrospectively and were converted into DDDs. Monthly data for alcohol-based hand rub (measured in L), clinical pharmacist FTEs and occupied patient bed-days were determined retrospectively from the hospital information system. Monthly hospital adult admissions, with their corresponding age-adjusted comorbidity data, were obtained from the hospital episode statistics. Age-adjusted comorbidity scores were calculated by combining age with the Charlson comorbidity index.16 – 18 The Charlson score is a weighted index that includes 19 diagnoses, and each diagnosis is weighted according to a 1 year relative risk of mortality. The weighted index reflects the severity of each comorbid disease.16 – 18 With age-adjusted comorbidity, an additional score is added for age (i.e. one point is added for each decade over the age of 40 years). Following this, the summation of the monthly patient age-adjusted Charlson comorbidity score was calculated for inclusion in the analysis. The collection of data included the intensive care unit (one eight-bed unit). The paediatric department was excluded from the study.
Following a Clostridium difficile infection (CDI) outbreak, the Northern Health and Social Care Trust devised an antibiotic policy to minimize the use of high-risk antibiotics (second-generation cephalosporins, thirdgeneration cephalosporins, fluoroquinolones and clindamycin; January 2008), and classified other antibiotics as medium-risk (amoxicillin/clavulanic acid and macrolides) or low-risk antibiotics (all remaining antibiotics; September 2008).19,20 Clinical staff were encouraged to adhere to the hospital policy and their compliance with the hospital policy was observed and recorded using a standardized procedure form.21 The use of antimicrobials not included in the policy was monitored through exemption forms, which required authorization by consultant physicians. The exemption forms were assessed by the antimicrobial management team (AMT) as appropriate or inappropriate with a written explanation. Results were directly shared with the prescribing physicians.
Antibiotic measure calculations Antibiotic use was normalized per 100 bed-days and per 100 age-adjusted comorbidity score separately. A modified antibiotic usage measuring unit was devised to take into account both patient bed-days and patient
Table 1. Estimates of time-series analysis model for monthly antibiotic use (R 2 ¼0.58) Term Sum of age-adjusted comorbidity scores Alcohol-based hand rub (L/100 bed-days) Clinical pharmacist FTEs (100 bed-days) Moving average a
Lag timea
Coefficient (SEM)b
T ratio
P value
0 2 3 5
0.927000 (0.244047) 20.085489 (0.039737) 20.487303 (0.155512) 20.625627 (0.124044)
3.798450 22.151364 23.133538 25.043594
0.0004 0.0370 0.0031 ,0.0001
Represents the delay necessary to observe the effect (in months). Indicates the size and direction of the effect.
b
2 of 10
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Setting and study design
Hospital data
JAC
A modified method for measuring antibiotic use
Clinical pharmacist FTEs
May-12
Mar-12
Jan-12
Nov-11
Sep-11
Jul-11
May-11
Mar-11
Jan-11
Nov-10
Antibiotic use
0.6
180 160
0.5
140
0.4
120 100
0.3
80
0.2
60 40
0.1 Mar-12
May-12
Jan-12
Sep-11
Nov-11
Jul-11
May-11
Mar-11
Jan-11
Sep-10
Nov-10
Jul-10
Mar-10
May-10
Jan-10
Sep-09
Nov-09
Jul-09
May-09
Jan-09
Mar-09
Nov-08
Sep-08
Jul-08
May-08
Mar-08
20 Jan-08
0
Antibiotic use, DDDs/100 bed-days
May-12
Jan-12
Mar-12
Nov-11
Jul-11
Sep-11
May-11
Jan-11
Mar-11
Nov-10
Sep-10
Jul-10 Jul-10
May-10
Jan-10
Mar-10
Nov-09
Jul-09
Sep-09
May-09
Jan-09
Mar-09
Nov-08
Sep-08
Jul-08
May-08
(c)
180 160 140 120 100 80 60 40 20 0
Antibiotic use
Sep-10
Mar-10
May-10
Jan-10
Sep-09
Nov-09
Jul-09
Mar-09
May-09
Jan-09
Sep-08
Nov-08
Jul-08
May-08
Mar-08
Jan-08 Jan-08
Alcohol-based hand rub 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0
180 160 140 120 100 80 60 40 20 0
Antibiotic use, DDDs/100 bed-days
Antibiotic use
9000 8000 7000 6000 5000 4000 3000 2000 1000 0
Mar-08
Alcohol-based hand rub, L/100 bed-days
Time-series analysis methods, including the use of linear transfer function techniques, were applied to study the relationship of age-adjusted comorbidity index, alcohol-based hand rub and clinical pharmacy activity to
Sum of age-adjusted comorbidity scores
(b)
Clinical pharmacist FTEs/100 bed-days
Statistical analysis
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Sum of age-adjusted comorbidity scores
(a)
antibiotics involved in the restriction policy (i.e. second-generation cephalosporins, third-generation cephalosporins, fluoroquinolones, clindamycin, penicillins with extended spectrum, and combinations of penicillins including b-lactamase inhibitors) was considered. Antibiotic comparative usage rate was also calculated for the latter and presented separately.
0
Antibiotic use, DDDs/100 bed-days
comorbidity score, i.e. DDDs/100 bed-days/age-adjusted comorbidity {[number of antibiotic DDDs in a specific month/(patient days×sum of age-adjusted comorbidity score in that month)]×100}. Antibiotic comparative usage rate was calculated by dividing the average monthly antibiotic use during the second study period by the average monthly antibiotic use during the first study period. The latter was carried out considering all antibiotics, i.e. total antibiotics for systemic use (J01). However, to address possible effects with regard to changes in antibiotic volumes due to the restriction policy (i.e. an increase in amoxicillin and amoxicillin/clavulanic acid agents, which may lead to an apparent increase in antibiotic use regardless of increased patient activity), another analysis that excluded all
Figure 1. Monthly antibiotic use versus (a) sum of age-adjusted comorbidity score, (b) use of alcohol-based hand rub and (c) clinical pharmacist FTEs, January 2008 to June 2012, Antrim Area Hospital.
3 of 10
Aldeyab et al.
explain the use of antibiotics (see the Statistical analysis section in the Supplementary data available at JAC Online).22 – 28 The series were checked for stationarity (i.e. having a constant mean and variance) using the augmented Dickey– Fuller test for unit roots.26 All variables were logarithmically transformed and first-order differencing was applied. The coefficient of determination (R 2), which corresponds to the percentage of the variance of the observed time-series explained by the model, was determined. Significance tests for parameter estimates were used to eliminate the unnecessary terms in the model. All analyses were performed using EViews 6 software (QMS, Irvine, CA, USA) and the SCA Statistical System software (River Forest, IL, USA). All results are reported based on the EViews software estimation summaries.
Results Analysis of data showed that monthly antibiotic use had significant positive relationships with the sum of age-adjusted comorbidity score (concomitant effect; coefficient, 0.927; P¼ 0.0004; Table 1). This means that a level change in the sum of age-adjusted comorbidity score by 1% would cause a level change in antibiotic use by 0.927% (direct relationship). The results also showed that there was a negative relationship between antibiotic use and (i) alcohol-based hand rub use (2 months delay; coefficient, 20.085; P¼ 0.0370; Table 1) and (ii) clinical pharmacist FTEs (3 months delay; coefficient, 20.487303; P ¼ 0.0031; Table 1). This means that a level change in alcohol-based hand rub use and clinical pharmacist FTEs by 1% would cause a level change in antibiotic use by 0.085% and 0.487% (inverse relationship), respectively. Based on examination of the residuals of the final model, adjustment was made for two outliers (i.e. February 2009 and March 2011) by inclusion of binary dummy variables (i.e. a value of 1 if an outlier is present; 0 otherwise.) in the model. For the final model, R 2 was 0.58, i.e. 58% of the variation in the monthly antibiotic use in the study site hospital was explained by the model. Graphs for antibiotic use versus sum of age-adjusted comorbidity score, alcohol-based hand rub and clinical pharmacist FTEs are shown in Figure 1.
Modified measure and benchmarking Age-adjusted comorbidity scores were calculated for 167576 admissions. Distributions of comorbidity and age-adjusted comorbidity score are shown in Table 2. Data showed a strong association between the monthly sum of age-adjusted comorbidity score and the monthly number of admissions (Pearson’s correlation, 0.92, P,0.0001). Over the two study periods (2004 –07 and 2008 –11), the average observed monthly antibiotic use was 8062.7 and 11407.6 DDDs, respectively, with an increasing trend in antibiotic use being observed (P,0.0001; Tables 3 and 4). Similar increasing trends were observed for both study periods when antibiotic use was normalized per 100 bed-days (P,0.0001; Table 4). Characteristics and trends in monthly antibiotic use are shown in Tables 3 and 4. The average monthly sums of age-adjusted comorbidity scores for both study periods were 4194.3 and 5061.5, respectively; an increasing trend was observed (P,0.0001; Table 4). When antibiotic use was normalized per age-adjusted comorbidity and per bed-days/age-adjusted comorbidity, a significant increasing trend was observed for the period 2004–07 (P,0.0001 and
4 of 10
2004– 07 (n¼76479) frequency (%)
number of patients
frequency (%)
Comorbidity score 0 51 174 1 12 874 2 7092 3 2400 ≥4 2939
66.9 16.8 9.3 3.1 3.8
56393 17235 9513 3832 4124
61.9 18.9 10.4 4.2 4.5
Age-adjusted comorbidity score 0 27 653 1 7194 2 6368 3 6705 4 7867 5 8169 6 5921 7 3079 8 1616 ≥9 1907
36.2 9.4 8.3 8.8 10.3 10.7 7.7 4.0 2.1 2.5
28306 8898 7514 8111 9463 10637 8117 4785 2490 2776
31.1 9.8 8.2 8.9 10.4 11.7 8.9 5.3 2.7 3.0
Term
Score
number of patients
2008 –11 (n¼91 097)
P,0.0027, respectively). However, antibiotic use during the period 2008–11 for both DDDs/age-adjusted comorbidity score and DDDs/100 bed-days/age-adjusted comorbidity score remained stable (P,0.1311 and P,0.6857, respectively; Table 4). Plots for monthly antibiotic use (DDDs) versus use of antibiotics normalized for different denominators (i.e. bed-days, age-adjusted comorbidity score and bed-days/age-adjusted comorbidity score) are presented in Figure 2. Graphs for antibiotic use DDDs/100 bed-days versus (i) antibiotic use DDDs/age-adjusted comorbidity scores and (ii) DDDs/100 bed-days/age-adjusted comorbidity score are shown in Figure 3. The average observed monthly antibiotic use during the second study period was higher (115.8 DDDs/100 bed-days, 2008 –11) than during the first study period (88.7 DDDs/100 bed-days; Table 3). This has contributed to a comparative usage rate of 1.31, i.e. after considering patient density, antibiotic use during the second period of the study was 31% higher than that during the first study period (Table 5). Normalizing antibiotic use per age-adjusted comorbidity score resulted in a comparative usage rate of 1.06, i.e. after taking into account the effect of patient case mix, antibiotic use during the second period of the study was 6% higher than that during the first study period. Considering a denominator that took into account both patient density and patient case mix characteristics led to a comparative usage rate of 0.98 (Table 5), i.e. the average antibiotic use was 2% lower in the second study period. Importantly, the latter comparative usage rate is independent of differences in patient density and case mix characteristics between the two studied populations. The exclusion of specific antibiotic agents showed similar patterns of decreasing overall antibiotic use when considering the studied adjustment units (Table 5).
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Time-series analysis
Table 2. Distribution of comorbidity and age-adjusted comorbidity score in Antrim Area Hospital, 2004– 11
Average monthly antibiotic use, 2004 – 07
DDDs/100 bed-days
111.1 2.1 280.36
1.22 0.22 3.09
0.0268 0.0005 0.0676
0.0003 0.000005 0.0007
1.38 0.25 3.50
600.1 1.8 1465.46
5.98 0.22 14.61
0.1019 0.0003 0.2479
0.0010 0.000003 0.0025
193.33
2.11
0.0457
0.0005
2.39
205.77
2.09
0.0364
0.0004
1.80
371.09
4.07
0.0887
0.0010
4.61
567.3
5.71
0.0993
0.0010
4.93
3101.36
33.96
0.7373
0.0081
38.46
3994.11
40.99
0.7316
0.0075
35.39
52.25
0.58
0.0126
0.0001
0.66
20.62
0.21
0.0038
0.00004
0.18
295.64
3.26
0.0722
0.0008
3.69
14.8
0.15
0.0028
0.00003
0.13
63.53
0.7
0.0154
0.0002
0.79
43.45
0.44
0.0078
0.0001
0.38
1.66 77.46 187.35
0.02 0.83 2.06
0.0004 0.0176 0.0450
0.000004 0.0002 0.0005
0.02 0.94 2.33
93.02 248.61 334.18
0.95 2.51 3.38
0.0167 0.0437 0.0583
0.0002 0.0004 0.0006
0.82 2.17 2.92
24.71
0.27
0.0060
0.0001
0.31
53.68
0.53
0.0090
0.0001
0.46
1322.75 65.4 0.63 94.54 939.57 163.3 106.83
14.4 0.71 0.01 1.04 10.31 1.78 1.17
0.3108 0.0154 0.0002 0.0226 0.2239 0.0384 0.0257
0.0034 0.0002 0.000002 0.0003 0.0025 0.0004 0.0003
16.31 0.80 0.01 1.18 11.68 2.02 1.33
1909.12 72.3 1.04 253.09 197.25 518.2 155.49
19.52 0.73 0.01 2.55 1.99 5.21 1.58
0.3466 0.0124 0.0002 0.0440 0.0348 0.0902 0.0279
0.0036 0.0001 0.000002 0.0004 0.0004 0.0009 0.0003
536.6
5.89
0.1282
0.0014
6.67
471.63
4.69
0.0829
0.0008
4.05
0.4
0.0087
0.0001
0.45
138.18
1.38
0.0232
0.0002
1.19
0.59 88.7
0.0082 1.9179
0.00009 0.0211
0.67 100.00
58.32 11407.55
0.38 115.8
0.0102 2.0320
0.00010 0.0207
36.73 34.48 8062.67
DDDs/100 bed-days/ age-adjusted comorbidity score
% of J01 use (based on DDDs/100 bed-days)
DDDs
DDDs/100 bed-days
DDDs/ age-adjusted comorbidity score
DDDs/100 bed-days/ age-adjusted comorbidity score
% of J01 use (based on DDDs/100 bed-days) 5.16 0.19 12.62
16.86 0.63 0.01 2.20 1.72 4.50 1.36
0.33 100.00
JAC
5 of 10
Tetracyclines (J01A) Amphenicols (J01B) Penicillins with extended spectrum (J01CA) b-Lactamase-sensitive penicillins (J01CE) b-Lactamase-resistant penicillins (J01CF) Combinations of penicillins including b-lactamase inhibitors (J01CR) First-generation cephalosporins (J01DB) Second-generation cephalosporins (J01DC) Third-generation cephalosporins (J01DD) Monobactams (J01DF) Carbapenems (J01DH) Trimethoprim and derivatives (J01EA) Combination of sulphonamides and trimethoprim (J01EE) Macrolides (J01FA) Lincosamides (J01FF) Streptomycins (J01GA) Aminoglycosides (J01GB) Fluoroquinolones (J01MA) Glycopeptide (J01XA) Steroid antibacterials (J01XC) Imidazole derivatives (J01XD) Nitrofuran derivatives (J01XE) Other antibacterials (J01XX) Antibiotics for systemic use, total (J01)
DDDs
DDDs/ age-adjusted comorbidity score
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Antimicrobial class (ATC group)
Average monthly antibiotic use, 2008 –11
A modified method for measuring antibiotic use
Table 3. Characteristics of monthly antimicrobial use in Antrim Area Hospital, 2004 –11
Aldeyab et al.
Table 4. General characteristics and trends in antibiotic use adjusted per patient days and per age-adjusted comorbidity score, Antrim Area Hospital, 2004– 11 Trend (2004– 07)
Trend (2008–11)
Term
coefficient
P value
coefficient
P value
Admissions Patient bed-days Sum of age-adjusted comorbidity scores Total antibiotic DDDs Antibiotic use (DDDs/100 bed-days) Antibiotic use (DDDs/age-adjusted comorbidity score) Antibiotic use (DDDs/100 bed-days/age-adjusted comorbidity score)
9.296407 21.80346 23.60427 104.7636 0.924069 1.391882 0.000104
,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 0.0027
5.977366 24.77079 44.18167 112.3938 0.834901 0.388966 20.000011
,0.0001 ,0.0001 ,0.0001 ,0.0001 ,0.0001 0.1311 0.6857
The process of developing policies and identifying targets for quality improvement in healthcare systems depends largely on the adequate collection and interpretation of a range of quality indicators. Different sources of data, with several advantages and limitations, are available for quantifying hospital antibiotic use, including pharmacy purchases, patient prescription profiles and pharmacy deliveries to wards. The latter data have been widely used for surveillance since they are easily accessible and readily available. The use of DDDs as a unit of measurement provides a reflection of the quantitative volumes of antibiotic usage and helps to enable comparisons between different institutions and countries. However, concerns have been expressed over which denominator should be used in order to express hospital antibiotic usage.12 – 14 More importantly, it has been shown that antibiotic use varies widely between different hospitals,29,30 with several studies suggesting excessive use, thus making it difficult to conduct proper benchmarking across hospitals.3,31,32 High antibiotic use in a particular setting is often justified by clinicians as relating to differences in patient morbidities.33 – 35 The measurement of patient comorbidity may therefore play a major role in explaining variations in hospital antibiotic use and subsequently facilitating the conduct of more realistic benchmarking. The results of the present research showed that monthly antibiotic use was significantly related to the age-adjusted comorbidity score. The importance of measuring comorbidity lies in its ability to predict morbidity, mortality, cost, hospitalization and the burden of the disease.16 – 18 As such, this factor can provide risk-adjustment criteria for patient case mix purposes and can have a vital role in explaining trends and variation in hospitals’ antibiotic use.15 Interestingly, changing age-adjusted comorbidity score had a much higher impact within the model when compared with the two other significant variables (Table 1). The latter observation, together with the fact that the role of patients’ age-adjusted comorbidity as a proxy measure for the disease burden has been proven,16 – 18 suggested the possible utility of this non-modifiable variable as an appropriate denominator for the calculation and presentation of hospital antibiotic use. The findings of the study also showed that an increase in the use of alcohol-based hand rub was associated with a decrease in antibiotic use. Antibiotics administered to patients in hospital are used to treat community- and healthcare-acquired infections.
6 of 10
The use of alcohol-based hand rub was shown to be an effective method to reduce the spread and acquisition of healthcareacquired infections in hospitals; the latter are associated with significant increases in length of hospital stay and antibiotic treatment.24,36,37 It is interesting to note that variations in hospital pathogens’ prevalence might be followed by similar variations in pathogens’ prevalence in the surrounding community. MacKenzie et al.38 showed that the monthly prevalence of methicillinresistant Staphylococcus aureus (MRSA) in the community was strongly related to the monthly MRSA prevalence observed 1 month previously in the hospital, suggesting that the reason for the increase in MRSA prevalence in the community was a hospital MRSA outbreak. Thus, increasing use of alcohol-based hand rub, which is a proxy measure for adequately practised infection control, might be leading to a subsequent reduction in hospital antibiotic use. However, further prospective research is needed to assess the impact of enhancing different infection control practices (e.g. alcohol-based hand rub, surveillance activity, education, etc.) on antibiotic use in healthcare settings. In addition, the number of clinical pharmacist FTEs was used as a proxy measure for clinical pharmacy activity. The results showed that an increase in the clinical pharmacy services provided across the hospital was associated with a decrease in antibiotic use. This is in line with other local research that showed that medicine appropriateness is improved via improved medication management by the clinical pharmacy team.39 In Antrim Area Hospital, a major CDI outbreak occurred in 2008.19 The CDI outbreak was ended following the implementation of an action plan that involved improving communication, infection control practices, environmental hygiene and surveillance, and enhanced antibiotic stewardship.19 The latter involved the implementation, led by the AMT, of prospective auditing of antimicrobial use with direct interaction and feedback to the prescriber, formulary restriction and a preauthorization strategy.19,21 With such high awareness of the problem of antibiotic use and resistance in this hospital, the statistically significant increasing usage trend and the observed 31% increase in average antibiotic use (DDDs/100 bed-days) during the implementation of enhanced antibiotic stewardship (second study period; Tables 4 and 5) contradicted the natural expected outcomes. Nevertheless, adjusting antibiotic usage DDDs per age-adjusted comorbidity score resulted in only a 6% antibiotic use increase during the enhanced antibiotic stewardship period, highlighting the value of considering patient
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Discussion
JAC
A modified method for measuring antibiotic use
Antibiotic use, DDDs/100 bed-days
Sep-11
May-11
Jan-11
Sep-10
Jan-10
May-10
Sep-09
Jan-09
May-09
Sep-08
May-08
Jan-08
Sep-07
May-07
Jan-07
Sep-06
Jan-06
May-06
May-05
Jan-05
Sep-04
May-04
18 000 16 000 14 000 12 000 10 000 8000 6000 4000 2000 0
2.5 2 1.5 1 0.5
Antibiotic use, DDDs/ age-adjusted comorbidity
3
Sep-11
May-11
Jan-11
Sep-10
Jan-10
May-10
Sep-09
May-09
Jan-09
Sep-08
May-08
Jan-08
Sep-07
May-07
Antibiotic use, DDDs/100 bed-days/age-adjusted comorbidity 0.035
18 000 16 000 14 000 12 000 10 000 8000 6000 4000 2000 0
0.03 0.025 0.02 0.015 0.01 0.005 Sep-11
May-11
Jan-11
Sep-10
May-10
Jan-10
Sep-09
May-09
Jan-09
Sep-08
May-08
Jan-08
Sep-07
May-07
Jan-07
Sep-06
Jan-06
May-06
Sep-05
May-05
Jan-05
Sep-04
May-04
Jan-04
0
Antibiotic use, DDDs/100 bed-days/ age-adjusted comorbidity
Antibiotic use, DDDs
Jan-07
Sep-06
Jan-06
May-06
Sep-05
Jan-05
May-05
Sep-04
May-04
0
(c)
Antibiotic use, DDDs
180 160 140 120 100 80 60 40 20 0
Antibiotic use, DDDs/age-adjusted comorbidity
Antibiotic use, DDDs
Jan-04
Antibiotic use, DDDs
(b)
Antibiotic use, DDDs/100 bed-days
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Jan-04
Antibiotic use, DDDs
18 000 16 000 14 000 12 000 10 000 8000 6000 4000 2000 0
Sep-05
Antibiotic use, DDDs
(a)
Figure 2. Monthly hospital antibiotic use adjusted for different denominators, January 2004 to December 2011, Antrim Area Hospital. January 2008 represents the introduction of the enhanced antibiotic stewardship.
case mix when expressing hospital antibiotic use. Such findings were consistent with the non-significant antibiotic use trend, which was observed during the second study period (DDDs/ age-adjusted comorbidity; Table 4). However, taking into account a more robust unit of measurement (i.e. DDDs/100 bed-days/ age-adjusted comorbidity score) has shown a reduction in the average antibiotic use by 2% (during the second study period). This reduction in antibiotic use, together with the significant reduction in high-risk antibiotics (described elsewhere),19,20 demonstrates
the success of this antibiotic stewardship in the study site hospital and confirms its value in the management of hospital antibiotic use. The study has several strengths. Firstly, the employed statistical regression technique has taken cognizance of the problem of dependency existing between consecutive observations and for identifying lag effects between predictors and the outcome series, thus measuring significant relationships accurately.40 Secondly, the study used robust risk-adjustment criteria for the patient case
7 of 10
Aldeyab et al.
3
160
2.5
140 120
2
100
1.5
80 60
1
40
0.5
20
Antibiotic use, DDDs/ age-adjusted comorbidity
180
0 Sep-11
May-11
Jan-11
Sep-10
May-10
Jan-10
Sep-09
May-09
Jan-09
Sep-08
May-08
Jan-08
Sep-07
May-07
Antibiotic use, DDDs/100 bed-days/age-adjusted comorbidity
180
0.035
160
0.03
140
0.025
120 100
0.02
80
0.015
60
0.01
40
0.005
20
0
0
Antibiotic use, DDDs/100 bed-days/ age-adjusted comorbidity
Jan-07
Sep-06
May-06
Jan-06
Sep-05
May-05
Jan-05
Sep-04
May-04
Antibiotic use, DDDs/100 bed-days
Figure 3. Monthly hospital antibiotic use (DDDs/100 bed-days) versus (a) antibiotic use (DDDs/100 age-adjusted comorbidity score) and (b) antibiotic use (DDDs/100 bed-days/age-adjusted comorbidity score), January 2004 to December 2011, Antrim Area Hospital. January 2008 represents the introduction of the enhanced antibiotic stewardship. Table 5. Comparative antibiotic usage rates adjusted per patient days and per age adjusted comorbidity scores, Antrim Area Hospital, 2004–11 Average use Term
2004–07
2008 –11a
Comparative rate
Estimates including all antibiotics antibiotic use (DDDs/100 bed-days) antibiotic use (DDDs/age-adjusted comorbidity) antibiotic use (DDDs/100 bed-days/age-adjusted comorbidity)
88.7 1.918 0.0211
115.8 2.032 0.0207
1.31 1.06 0.98
Estimates excluding specific antibioticsb antibiotic use (DDDs/100 bed-days) antibiotic use (DDDs/age-adjusted comorbidity) antibiotic use (DDDs/100 bed-days/age-adjusted comorbidity)
36.66 0.786 0.0086
56.90 0.995 0.0101
1.55 1.27 1.17
a
This period represents the introduction of the enhanced antibiotic stewardship. Antibiotics excluded are second-generation cephalosporins, third-generation cephalosporins, fluoroquinolones, clindamycin, penicillins with extended spectrum, and combinations of penicillins including b-lactamase inhibitors.
b
mix (i.e. the Charlson index), which is considered as the gold standard to assess comorbid risk in clinical research and uses readily available hospital databases.16 – 18 Thirdly, analysis showed a
8 of 10
strong association between the monthly sum of age-adjusted comorbidity score and the number of monthly admissions. This is of importance since age-adjusted comorbidity score can reflect
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
Jan-04
0
(b) Antibiotic use, DDDs/100 bed-days
Antibiotic use, DDDs/age-adjusted comorbidity
Antibiotic use, DDDs/100 bed-days
Jan-04 Apr-04 Jul-04 Oct-04 Jan-05 Apr-05 Jul-05 Oct-05 Jan-06 Apr-06 Jul-06 Oct-06 Jan-07 Apr-07 Jul-07 Oct-07 Jan-08 Apr-08 Jul-08 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10 Jan-11 Apr-11 Jul-11 Oct-11
Antibiotic use, DDDs/100 bed-days
(a)
JAC
A modified method for measuring antibiotic use
clinical diagnosis.17 In addition to expressing antibiotic use per 100 bed-days, hospitals should be encouraged to express their antibiotic use per 100 bed-days/age-adjusted comorbidity score in order to allow the calculation of a comparative usage rate, while taking into account the effect of patient case mix. Following the determination of the percentage difference in antibiotic use, based on the identified antibiotic comparative usage rate, antibiotic policy makers should investigate the reasons behind a high or low antibiotic usage rate, thus identifying targets for quality improvement in the former and presenting best antibiotic prescribing practice for adoption by others in the latter.
Funding The study was carried out as part of our routine work.
Transparency declarations None to declare.
Supplementary data Figure S1 and a Statistical analysis section are available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/).
References 1 Patrick DM, Hutchinson J. Antibiotic use and population ecology: how you can reduce your “resistance footprint”. CMAJ 2009; 180: 416–21. 2 Goossens H. Antibiotic consumption and link to resistance. Clin Microbiol Infect 2009; 15 Suppl 3: 12– 5. 3 Hecker MT, Aron DC, Patel NP et al. Unnecessary use of antimicrobials in hospitalized patients: current patterns of misuse with an emphasis on the antianaerobic spectrum of activity. Arch Intern Med 2003; 163: 972–8. 4 Gould IM. Antibiotic policies to control hospital-acquired infection. J Antimicrob Chemother 2008; 61: 763–5. 5 Wilcox MH, Dave J. The cost of hospital-acquired infection and the value of infection control. J Hosp Infect 2000; 45: 81 –4. 6 Gould IM. The clinical significance of methicillin-resistant Staphylococcus aureus. J Hosp Infect 2005; 61: 277– 82. 7 Fishman N. Antimicrobial stewardship. Am J Med 2006; 119 Suppl 1: S53– 61. 8 Griffith M, Postelnick M, Scheetz M. Antimicrobial stewardship programs: methods of operation and suggested outcomes. Expert Rev Anti Infect Ther 2012; 10: 63–73. 9 Dellit TH, Owens RC, McGowan JE Jr et al. Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America guidelines for developing an institutional program to enhance antimicrobial stewardship. Clin Infect Dis 2007; 44: 159– 77. 10 Davey P, Brown E, Fenelon L et al. Interventions to improve antibiotic prescribing practices for hospital inpatients. Cochrane Database Syst Rev 2005; issue 4: CD003543. 11 WHO Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC Classification and DDD Assignment, 2012. Oslo, 2011. http://www. whocc.no/atc_ddd_publications/guidelines/ (20 September 2012, date last accessed).
9 of 10
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
monthly changes in hospital admissions. Therefore, expressing antibiotic use in terms of age-adjusted comorbidity score may possibly account for variations in admission rates and thus provide a meaningful understanding and interpretation of hospital antibiotic use.12 Fourthly, antibiotic use and the age-adjusted comorbidity score were calculated for all hospitalized adult patients; thus, information and selection bias were unlikely. Finally, adjustment was made for the confounding effect of different healthcare systems and hospital policies since the study used an internal benchmarking approach in the same study site hospital. However, the study has the following limitations. Firstly, although it is possible to argue that the increase in antibiotic use during the second study period was an apparent increase due to the increase in amoxicillin and amoxicillin/clavulanic acid, the exclusion of these agents showed similar patterns of decreasing overall antibiotic use when considering the presented adjustment measures (Table 5). This indicates that the increase in antibiotic use was genuine and possibly related to different reasons. Secondly, it is possible that other potential drivers for increased antibiotic use could be related to hospital infection rates and duration of treatment. Research from the study site Trust showed that the intervention introduced in 2008 (i.e. restricting high-risk antibiotics) had a positive impact on reducing the incidence rates of specific hospital pathogens (i.e. CDI pathogens, extended-spectrum b-lactamase-producing bacteria and MRSA).19,20,41,42 In addition, the quality of antibiotic prescribing and adherence to guidelines was continuously scrutinized and improved.21,43 For example, adherence to the Trust antibiotic policy was improved after 200821,43 compared with prior periods.44 Thirdly, it was not possible to determine accurate data for the entire study period with regard to other types of infection (e.g. surgical site infections, ventilator-associated pneumonia) to assess its relation with antibiotic use and age-adjusted comorbidity score; this requires further prospective research in different hospitals. However, the presence of multiple comorbidities is an indicator of patients in suboptimal health who may be unable to mount an effective immune response to bacterial challenge, and as such might facilitate patient colonization/infection with pathogens; the age-adjusted comorbidity score provides a convenient measure for the latter. Fourthly, the Charlson index might be limited in its applicability to complex processes of care and functional outcomes since it does not incorporate physiological data to assess the severity of comorbid conditions and does not include functional impairment attributable to pre-existing comorbid conditions.17 Finally, further research is needed to compare antibiotic use by different hospital departments, and to evaluate the proposed benchmarking approach using external groups and involving several hospitals across different countries with high and low volumes of antibiotic use. In conclusion, the proposed modified antibiotic use unit of measurement provided an innovative approach to the measurement of actual trends in antibiotic use and the comparison of variations in antibiotic prescribing between hospitals, taking account of the effect of patient case mix. It facilitated the evaluation of antibiotic stewardship more accurately and can enable more informative benchmarking between hospitals. It is interesting to note that other identified predictors are considered modifiable factors, thus leaving room for quality improvement in antibiotic use. Age-adjusted comorbidity score calculations (utilizing the Charlson index) are relatively easy to apply, based on availability of
Aldeyab et al.
12 Filius PM, Liem TB, van der Linden PD et al. An additional measure for quantifying antibiotic use in hospitals. J Antimicrob Chemother 2005; 55: 805–8. 13 Cooke J. Trends in antimicrobial prescribing. J Antimicrob Chemother 2000; 46: 639. 14 Ibrahim OM, Polk RE. Benchmarking antimicrobial drug use in hospitals. Expert Rev Anti Infect Ther 2012; 10: 445– 57. 15 Aldeyab MA, McElnay JC, Scott MG et al. Hospital antibiotic use and its relationship to age-adjusted co-morbidity and alcohol-based hand rub consumption. Epidemiol Infect 2013; 9: doi:10.1017/S0950268813001052. 16 Charlson ME, Pompei P, Ales KL et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40: 373– 83.
Retrospective Data Collection (1997– 2002). J Antimicrob Chemother 2006; 58: 159–67. 30 Lesch CA, Itokazu GS, Danziger LH et al. Multi-hospital analysis of antimicrobial usage and resistance trends. Diagn Microbiol Infect Dis 2001; 41: 149–54. 31 Kislak JW, Eickhoff TC, Finland M. Hospital-acquired infections and antibiotic usage in the Boston City Hospital-January, 1964. N Engl J Med 1964; 271: 834–5. 32 Werner NL, Hecker MT, Sethi AK et al. Unnecessary use of fluoroquinolone antibiotics in hospitalized patients. BMC Infect Dis 2011; 11: 187. 33 Kuster SP, Ruef C, Bollinger AK et al. Correlation between case mix index and antibiotic use in hospitals. J Antimicrob Chemother 2008; 62: 837– 482. 34 Rogues AM, Placet-Thomazeau B, Parneix P et al. Use of antibiotics in hospitals in south-western France. J Hosp Infect 2004; 58: 187–92.
18 Peterson JC, Paget SA, Lachs MS et al. The risk of comorbidity. Ann Rheum Dis 2012; 71: 635– 7.
35 Polk RE, Hohmann SF, Medvedev S et al. Benchmarking risk-adjusted adult antibacterial drug use in 70 US academic medical center hospitals. Clin Infect Dis 2011; 53: 1100– 10.
19 Aldeyab MA, Devine MJ, Flanagan P et al. Multihospital outbreak of Clostridium difficile ribotype 027 infection: epidemiology and analysis of control measures. Infect Control Hosp Epidemiol 2011; 32: 210– 9. 20 Aldeyab MA, Kearney MP, Scott MG et al. An evaluation of the impact of antibiotic stewardship on reducing the use of high-risk antibiotics and its effect on the incidence of Clostridium difficile infection in hospital settings. J Antimicrob Chemother 2012; 67: 2988 –96. 21 Conlon G, Aldeyab MA, McElnay JC et al. Improving and maintaining adherence with hospital antibiotic policies: a strategy for success. J Hosp Infect 2011; 77: 88– 9. 22 Box GEP, Jenkins GM. Time Series Analysis: Forecasting and Control. San Francisco: Holden Day, 1976. 23 Box GEP, Jenkins GM, Reinsel GC. Time Series Analysis: Forecasting and Control. 3rd edn. Englewood Cliff, NJ: Prentice Hall, 1994. 24 Aldeyab MA, Monnet DL, Lo´pez-Lozano JM et al. Modelling the impact of antibiotic use and infection control practices on the incidence of hospitalacquired methicillin-resistant Staphylococcus aureus: a time-series analysis. J Antimicrob Chemother 2008; 62: 593–600. 25 Aldeyab MA, Harbarth S, Vernaz N et al. Quasiexperimental study of the effects of antibiotic use, gastric acid-suppressive agents, and infection control practices on the incidence of Clostridium difficile-associated diarrhea in hospitalized patients. Antimicrob Agents Chemother 2009; 53: 2082– 8. 26 Dickey DA, Fuller WA. Distribution of estimators for autoregressive time series with a unit root. JASA 1979; 74: 427– 31. 27 Liu L-M, Hanssens DM. Identification of multiple-input transfer function models. Commun Stat 1982; A11: 297–314. 28 Liu L-M. Time Series Analysis and Forecasting, Second Edition. Chicago, IL: Scientific Computing Associates Corporation, 2006. 29 Vander Stichele RH, Elseviers MM, Ferech M et al. Hospital consumption of antibiotics in 15 European countries: results of the ESAC
10 of 10
36 Jarlier V, Trystram D, Brun-Buisson C et al. Curbing methicillin-resistant Staphylococcus aureus in 38 French hospitals through a 15-year institutional control program. Arch Intern Med 2010; 170: 552–9. 37 Vernaz N, Sax H, Pittet D et al. Temporal effects of antibiotic use and hand rub consumption on the incidence of MRSA and Clostridium difficile. J Antimicrob Chemother 2008; 62: 601–7. 38 MacKenzie FM, Lopez-Lozano JM, Monnet DL et al. Temporal relationship between prevalence of meticillin-resistant Staphylococcus aureus (MRSA) in one hospital and prevalence of MRSA in the surrounding community: a time-series analysis. J Hosp Infect 2007; 67: 225– 31. 39 Burnett KM, Scott MG, Fleming GF et al. Effects of an integrated medicines management program on medication appropriateness in hospitalized patients. Am J Health Syst Pharm 2009; 66: 854–9. 40 Shardell M, Harris AD, El-Kamary SS et al. Statistical analysis and application of quasi experiments to antimicrobial resistance intervention studies. Clin Infect Dis 2007; 45: 901–7. 41 Aldeyab MA, Harbarth S, Vernaz N et al. The impact of antibiotic use on the incidence and resistance pattern of extended-spectrum b-lactamase-producing bacteria in primary and secondary healthcare settings. Br J Clin Pharmacol 2012; 74: 171– 9. 42 Aldeyab MA, Scott MG, Kearney MP et al. Impact of an enhanced antibiotic stewardship on reducing methicillin-resistant Staphylococcus aureus in primary and secondary healthcare settings. Epidemiol Infect 2013; 5: 1– 7. 43 Aldeyab MA, Kearney MP, McElnay JC et al. A point prevalence survey of antibiotic prescriptions: benchmarking and patterns of use. Br J Clin Pharmacol 2011; 71: 293– 6. 44 Aldeyab MA, Elshibly SM, McElnay JC et al. An evaluation of compliance with an antibiotic policy in surgical wards at a general teaching hospital in Northern Ireland. Infect Control Hosp Epidemiol 2009; 30: 921– 2.
Downloaded from http://jac.oxfordjournals.org/ at Grand Valley State Univ on November 16, 2013
17 Tobacman JK. Assessment of comorbidity: a review. Clin Perform Qual Health Care 1994; 2: 23– 32.
