Appropriate measure for outpatients antibiotic use in Europe. Ann Versporten (Belgium)

Appropriate measures for outpatient antibiotic use in Europe

Ann Versporten On behalf of Robin Bruyndonckx, Samuel Coenen, Herman Goossens

3rd Joint meeting of the Antimicrobial Resistance and Healthcare-Associated Infection (ARHAI) Networks

Stockholm, 11-13 February 2015

Outpatient antibiotic use in Belgium

Expressed as: N defined daily doses, N packages, N treatments, N insured individuals, reimbursed per 1000 inhabitants per day.

Coenen S, Gielen B, Blommaert A, Beutels P, Hens N, Goossens H. Appropriate international measures for outpatient antibiotic prescribing and consumption: recommendations from a national data comparison of different measures. J Antimicrob Chemother 2014;69:529-534.

Explained discrepancies between the different measurement units:

• Increased N DDD/pack for:

Amoxicillin (up to 50%, from 7.1 in ’02-’03 to 10.7 in ‘08-’09)

Co-amoxiclav (up to 70%, from 8.7 to 14.2)

These substances present 54% of all outpatient AB use in DDD in 2008-2009

Increase of N units/pack & amount active substance/unit

From 16 to 20 units/pack

From 500mg to 1000mg/unit

Outpatient antibiotic use in Belgium

Background – Research questions

What is the trend of outpatient antibiotic use in Europe?

What is the relationship between outpatient antibiotic use and antimicrobial resistance in Europe?

Methods – Data used

Outpatient antibiotic use:

• IMS data; all ATC J01 (antibacterials for systemic use); 31 EU countries; quarterly data; years 2000-2007

N defined daily doses/1000 inhabitants/day (DID)

N packages/1000 inhabitans/day (PID)

Antimicrobial resistance:

• EARSS data; 27 EU countries; years 2000-2009

Proportion penicillin-non-susceptible S. pneumoniae (PNSP)

Proportion erythromycin-non-susceptible S. pneumoniae (ENSP)

Methods – Data analysis

1. Outpatient AB use (IMS data)

Quarterly measurements per country mixed effects model

Seasonal fluctuation nonlinear mixed model

2. Association of outpatient AB use in DID, PID and both; and resistance

Generalized linear mixed model with 0, 1 or 2 years time lag between AB use and resistance

Predictions of resistance for decreasing AB use

Results – Outpatient AB use in DID

Results – Outpatient AB use in PID

Results – Change in DDD per package

Average dose per package (DDD/pack) in year 2000 differs between AB subgroups and between countries

Average change in DDD/pack over time (increase 2000-2007) differs substantially between countries and AB subgroups No increase for quinolones (J01M)

Results – Change in DDD per package (DP)

Overall increase in DDD/package over time • Quarterly increase ranged between 0.01 and 0.08 DDD/package • Yearly increase ranged from 0.04 to 0.31 DDD/package

Results – Estimated change in DDD per package

β0 (intercept), predicted average (standard errors) DDD per package in the first quarter of 2000; β1 (slope), predicted average (standard errors) increase (if positive)/decrease (if negative) in DDD per package per quarter. *P<0.05. **P<0.0001.

Results – Association AB use and resistance

PNSP and β-lactam use (penicillin, cephalosporin)

Best model fit for β-lactam use in PID & lag time=0 year

Sign. increase of odds of PNSP with increasing β-lactam use in PID (OR=1.96;1.57-2.44) which did not changed sign. over time

Predictions of PNSP if decrease of β-lactam use

• If decrease of β-lactam use in DID, then prediction of stable PNSP • If decrease of β-lactam use in PID or in PID & DID, then prediction of decrease of PNSP

Results – Association AB use and resistance

ENSP and TMLS use (tetracycline, macrolide, lincosamide, streptogramin)

Best model fit for TMLS in PID and DID & lag time=1 year

Sign. increase of odds of ENSP with increasing TMLS use in PID (OR=3.68;1.27-10.72) and decreasing TMLS use in DID (OR=0.78;0.65-0.93), with no sign. changes over time

Predictions of ENSP if decrease of TMLS use

• If decrease of TMLS use in DID or in PID & DID, then prediction of increase of ENSP • If decrease of TMLS use in PID, then prediction of decrease of ENSP

Conclusions – Measuring outpatient AB use

Present both DID and PID because:

Variable increase in dose/pack over time

(Country and AB subgroup)

Contrasting AB use trends between DID and PID

Inconsistent associations and predictions of resistance whether AB use expressed in DID or PID

Model fit depend on time lag between use & resistence

Conclusions – Measuring outpatient AB use

Better understanding and interpretation of outpatient antibiotic use and its relation to

resistance

Consider time lags and also data expressed in PID when exploring AB-use & resistance relations