2. Methods for AMR Surveillance in Communities – lessons from the Durban site Gray AL and Essack SY Department of Pharmacology, Nelson R Mandela School of Medicine and School of Pharmaceutical Sciences, University of KwaZulu-Natal, Durban, South Africa

3. Summary of the Durban pilot project  Objective  To investigate the association between antibiotic use and resistance over time in respiratory tract infections in the Inner West metropolitan area of Durban  Methods  Sputum specimens from consenting patients with self- reported cough, with or without fever, at 4 convenience sampled sites  Retrospective prescription audit (2 weeks’ Rx per month) from 7 randomly selected private pharmacies, 7 convenience sampled private dispensing practitioners and 7 randomly selected primary health care clinics  Results  No direct relationship between resistance levels and antimicrobial usage; feasibility of establishing a system to generate data of this sort demonstrated

4. Methodological issues - resistance  Grand aim: “to determine the incidence of resistant infections among the total number of infections in a population”  Overcome biases of hospital-based and treatment failure associated data  Need to choose a common infection with easily accessed clinical material – in our case:  respiratory tract infections  sputum specimens (vs. oropharyngeal swab) - minimally invasive  ? carriage vs. infection

5. Problems encountered  Negotiating access in both the public and for-profit private sectors  had to use convenience sample  Low return  small % of positive sputa (521/3556) – 14.7%  preponderance of some isolates - M. catarrhalis resistance could not be characterised over time  H. influenzae – 387/570 (67.9%)  S. pneumoniae – 137/570 (24.0%)  M. catarrhalis – 46/570 (8.1%)  Time consuming and expensive  3 fieldworkers, travelled 9 945km in 12 months

7. Reasons for declining returns …  Fieldworker motivation- repetitive task, dealing with difficult patients  Refusal by some patients to give repeated specimens when no immediate clinical benefit was discerned  Potential solutions  Rotating sites – difficult to negotiate  Community feedback – easier in public sector?  Different target infection/carriage

8. Methodological issues - usage  Grand aim: enable “early action to optimize prescribing patterns and to reduce inappropriate use”  move beyond hospital-level utilisation reviews  cover all possible sources of community access:  informal (markets) –  assumed not to be a major source in South Africa  formal –  on-prescription sales by retail (community) pharmacies  on-prescription sales by dispensing medical practitioners  issues by state-operated primary health care clinics (largely nurse practitioners)

9. Initial challenges – negotiating access (1)  Negotiating access - pharmacies  willing to co-operate – allowed random sampling  stratified by socio-economic status of area  Data source –  original prescriptions; computerised  accessible, good data on the prescription – all necessary details  sparse clinical data

10. Initial challenges – negotiating access (2)  Negotiating access – dispensing doctors  Initially reluctant to co-operate – had to resort to convenience sampling  ongoing policy battles around the “right” to dispense  currently sell prescription data – source of income for the independent practitioner association (IPA)  stratified by socio-economic status of area  Data source –  clinical records  variable quality of data

11. Initial challenges – negotiating access (3)  Negotiating access – PHC clinics  protracted negotiations with provincial and local authorities – allowed random sampling  stratified by size to include 2 large community health centres (CHCs)  mixed medical practitioner and nurse prescribers  Data source –  daily clinic registers (“tick registers”)  Sparse data

12. Data sources - clinics

13. Problems encountered …  Small numbers of antimicrobial prescriptions in smaller pharmacies, practices and clinics  Large number of “tick registers” in larger clinics (CHCs) – inability to access all data accurately  Solutions implemented  returned to collect extra week of data per site (2 weeks’ Rx)  deleted all AM usage data from one problematic CHC (left with 20 sites)

14. Further concerns …  Missing data -  clinics usually dispense original packs, so quantities could be assumed – difficult when practices change  e.g. increased prescribing of cotrimoxazole for PCP prophylaxis  Choice of denominator  usually as DDD/1000 pop/unit time  not possible without a “catchment population” or complete coverage  mobile population  no “registration” with a provider  using both sectors interchangeably  Used Defined Daily Doses (DDD) per 100 patients seen (doctors/clinics) or prescriptions dispensed (pharmacies)

15. Antimicrobial use - cotrimoxazole

16. Time and expense  2 fieldworkers (full-time M.Pharm student, ½ day nurse) for medicine utilisation review  travelled 15 578km (from Mar ’03 to Feb ’04)  3 fieldworkers for sputum collection  travelled 9 945km  Feasibility as an ongoing venture?  commitment of health authorities  viability of the District Health Systems model  routine data vs. periodic (survey) approach

17. Possible alternative sources of medicine use data (problems)  Pharmacies  Wholesaler and distributor sales records  Wide range of possible sources, locally and across the country/ direct purchase from manufacturers – impact of new pricing regulations?  Doctors  IPA data (currently revenue generating)  Impact of dispensing license regulations and data privacy regulations?  Clinics  Depot issue records  Clinic (CHC) to clinic supplies – impact of the DHS and nature of future contracts with local authorities (municipal health services)?

18. Conclusions  Although no direct relationship between resistance levels and antimicrobial usage could be shown, the feasibility of establishing a system to generate data of this sort was demonstrated  Given the differences in antimicrobial use patterns in different settings, interventions to contain the development of resistance will have to be carefully tailored for each setting  Choose a different target infection or site of carriage; rotate collection between different sites; need to characterise resistance separately for different settings?  Need to measure AM usage in different settings; could perhaps limit to a few selected months of the year (some seasonal variation)

19. Acknowledgements  WHO/EDM for funding this pilot project  Kathy Holloway (WHO, Geneva) and Thomas Sorenson (Statens Serum Institut, Denmark) for technical advice and support  Our co-investigators (Wim Sturm, Fathima Deedat), the fieldworkers and laboratory staff, for their hard work and insights into the process  The staff at the facilities, for allowing us access to patients and/or data  The patients, for providing us with sputum specimens