Enterococci but not E coli counts in drinking

Enterococci but not E. coli counts in drinking water are positively associated with gastrointestinal disease risk in urban Zambia UNC water and Health Conference 2016 13 th October, 2016 Michelo Simuyandi Ph. D candidate: LSHTM-ITD-DCD Research Fellow: CIDRZ

Outline • • • Background Methods Results Discussion Conclusion

Background • Studies on the association between faecal indicator bacteria in drinking water and diarrhoeal disease risk have yielded mixed findings that range from no association to significant associations 1, 2 • This variation could be due to: – high variability of water quality over time 3 – Indicator organisms being weakly correlated to enteric pathogens 4 – Study design related: collecting water quality and health data at the same time may fail to determine direction of causality 5 – Acquired immunity to certain diarrhoea causing pathogens could affect generalisation of the associations found in some studies 6 – Other transmission routes of infection 7 1. 2. 3. 4. 5. 6. 7. Gundry et al 2004 Gruber et al, 2014 Levy et al. , 2009 Wu et al. , 2011, Khush et al. , 2013 Frost et al. , 2005 Luby et al 2015

Study design • We conducted a prospective study collecting repeated measures of water quality and health indicators and related covariates at both household and individual level in a periurban area south of Lusaka – As part of the pilot study of a field trial of a point of use HWTS intervention

Aim • To determine the association between the levels of Enterococci, E. coli and total coliforms in household drinking water and self-reported highly credible gastrointestinal illness (HCGI) experienced by household members using a 7 day recall period.

Definitions • The exposure was defined as – Safe = <1 MPN/100 m. L – Unsafe >1 MPN/100 m. L • Outcome measure used was self-reported highly credible gastrointestinal illness (HCGI) experienced by household members using a 7 day recall period – HCGI : defined as a report of at least one of: 1) vomiting; 2) watery diarrhoea; 3) soft diarrhoea and abdominal cramps both occurring during recall period; 4 ) nausea and abdominal cramps both occurring during recall period (Colford et al. , 2005) • The associations between Enterococci, E. coli, total coliforms in household drinking water and the HCGI experienced by household members were investigated using mixed-effects logistic regression

Methods • Data was collect by – household interviews using a survey, – household observations • Testing of household stored drinking water and source water for bacterial indicators of faecal contamination was done using the IDEXX quanti-tray, MPN Methods – Enterolert® Test for Enterococci – Colilert® Test for E coli and total coliforms

FACTS/stats • • • Intermittent water supply unsafe sanitation high fertility High Unemployment rate poor drainage

5 CEA Household clusters(1206 HH) 300 households recruited (60/cluster) 244 households included in analysis Inclusion criteria At least a child below 5 yrs, written consent, purposive sampling till agreed 60 hh /cluster 10 households lost to follow-up • 8 relocated to area outside study area. • 2 withdrew without reason 44 households excluded due to missing/incomplete data 244 households contributing to analysis 832 Individuals 466 females 196 children below 5 yrs

Household and individual baseline characteristics for the study population Household Characteristic Water Quality Ecoli count <1 1 -10 11 -100 101 + Total coliforms (Ecoli-media) <1 1 -10 11 -100 101 + Enterococci count <1 1 -10 11 -100 101 + Household Water Treatment No Yes Primary Water Source Tap inside house Stand pipe Other Containers per daya mean(SD) Containers per day category 0 -2 3 -5 6 -9 Distance to water source <200 meters 200+ metres How long to collect water 5 -14 minutes 15+ minutes Number of subjects (% of total) (n=635) 268(42. 2) 127(20. 0) 98(15. 4) 142(22. 4) (n=635) 81(12. 8) 48(7. 6) 18(2. 8) 488(76. 9) (n=570) 529(92. 8) 23(4. 0) 12(2. 1) 6(1. 1) (n=829) 661 (79. 73) 168 (20. 27) (n=569) 53 (9. 31) 244 (42. 88) 272 (47. 80) 3. 16 (2. 00) (n=766) 347 (45. 30) 307 (40. 08) 112 (14. 62) (n=791) 762 (96. 33) 29 (3. 67) (n=371) 306 (82. 48) 65 (17. 52) Proportion(%) with HCGI (95% CI)1 42/339 (12. 4) 15/180 (8. 3) 13/131 (9. 9) 19/185 (10. 3) (8. 8, 17. 1) (4. 1, 16. 3) (4. 8, 19. 4) (5. 9, 17. 2) 12/98 (12. 2) 12/65 (18. 5) 2/40 (5. 0) 63/632 (10. 0) (6. 6, 21. 5) (8. 2, 36. 5) (1. 3, 17. 7) (7. 4, 13. 3) 51/676 (7. 5) 5/23 (21. 7) 3/12 (25. 0) 2/7 (28. 6) 89/692 (12. 86) 21/280 (6. 98) 5/67 (7. 46) 52/371 (14. 02) 52/406 (12. 81) 40/472 (8. 47) 62/468 (13. 25) 22/181 (12. 15) 128/1101 (11. 63) 1/47 (2. 13) 40/468 (8. 55) 11/85(12. 94) (5. 3, 10. 6) (6. 2, 5. 4) (9. 1, 52. 6) (8. 2, 64. 3) (9. 92, 16, 51) (4. 45, 10. 77) (3. 94, 13. 70) (10. 12, 19. 09) (9. 17, 17. 60) (5. 62, 12. 59) (9. 86, 17. 58) (7. 86, 18. 33) (9. 42, 14. 26) (0. 26, 15. 25) (5. 74, 12. 55) (7. 70, 20. 99)

Household Characteristic How long to collect water 5 -14 minutes 15+ minutes Safe Sanitationb No Yes Good Hygienec No Yes Place to Wash hands in house No Yes Place to Wash hands near toilet No Yes Seasone Wet Dry Zone A Zone B Zone C Zone D Zone E Number of subjects (% of total) (n=371) 306 (82. 48) 65 (17. 52) (n=695) 622 (89. 50) 73 (10. 5) (n=821) 622 (75. 76) 199 (24. 24) (n=827) 640 (77. 39) 187 (22. 61) (n=829) 721 (86. 97) 108 (13. 03) (n=832) 113 (13. 58) 172 (20. 67) 249 (29. 93) 176 (21. 15) 122 (14. 66) Proportion(%) with HCGI 40/468 (8. 55) 11/85(12. 94) 107/920 (11. 63) 11/112 (9. 82) 102/900 (11. 33) 29/290 (10. 00) 95/934 (10. 17) 37/262 (14. 12) 106/1038 (10. 21) 26/160 (16. 25) 86/928(9. 27) 46/273(16. 85) 9/145(6. 21) 41/229(17. 90) 41/366(11. 20) 17/254(6. 69) 24/207(11. 59) (95% CI)1 (5. 74, 12. 55) (7. 70, 20. 99) (9. 23, 14. 55) (4. 50, 20. 11) (8. 91, 14. 30) (6. 56, 14. 96) (8. 05, 12. 78) (9. 15, 21. 16) (8. 10, 12. 80) (10. 00, 25. 31) (7. 21, 11. 83) (11. 54, 23. 94) (2. 67, 13. 76) (13. 03, 24. 09) (7. 49, 16. 44) (4. 07, 10. 82) (7. 22, 18. 10) Individual Characteristics Sex Female Male Age Mean(SD) Age Category <5 years >= 5 years Total (n=824) 358 (43. 45) 466 (56. 55) 19. 47 (16. 35) (n=825) 196 (23. 76) 629 (76. 24) 832 (100%) 82/677(12. 11) 48/510(9. 41) (9. 56, 15. 23) (7. 09, 12. 40) 37/268 (13. 81) 95/925 (10. 27) 132/1201 (10. 99) (10. 03, 18. 70) (8. 10, 12. 92) (8. 93, 13. 45)

Water quality: A=Source vs HH stored and B =Storage container type

Univariate associations with Highly Credible Gastrointestinal Illness Variable Number of Obs. (N=1, 201) 1, 903 1, 187 Odds Ratio 1. 52 1. 40 95% CI (0. 95, 2. 44) (0. 92, 2. 11) P-value 0. 08 0. 12 Unsafe E. coli count Unsafe total coliforms 270/835 0. 59 0. 79 (0. 33, 1. 06) (0. 35, 1. 79) 0. 08 0. 57 Unsafe Enterococci count 250/718 15. 25 (2. 27, 102. 27) 0. 01 993 0. 41 (0. 22, 0. 75) <0. 01 Containers collected per day 2 1, 121 1. 09 (0. 96, 1. 25) 0. 19 Distance to Source > 200 meters 2 1, 148 0. 16 (0. 02, 1. 51) 0. 11 553 1. 85 (0. 66, 5. 14) 0. 24 Narrow mouth container 1 1, 023 0. 98 (0. 53, 1. 84) 0. 96 Good Hygiene 2 1, 190 0. 85 (0. 47, 1. 54) 0. 60 Place to wash hands in house 2 1, 196 1. 46 (0. 83, 2. 60) 0. 20 Place to wash hands near toilet 2 1, 198 2. 12 (1. 08, 4. 17) 0. 03 Safe Sanitation 2 1, 032 1. 03 (0. 40, 2. 64) 0. 95 Age 1 < 5 years Sex (female) Water quality Treat Water 1 >14 minutes to collect water 2 Containers collected per day= number of 20 litre containers of water collected per day. Good Hygiene=Have soap (shown to interviewer) and household respondent asserts washes with soap. Safe sanitation=Household respondent asserts that the household uses a private latrine. 1 As reported at visit corresponding to HCGI outcome. (i. e. “Did you treat this water you are currently drinking? ”) 2 As reported at baseline (first available visit for which response was available). Mixed effects logistic regression accounting for household and individual-level clustering. All associations with p-value < 0. 05 are underlined

Mixed effects logistic regression accounting for household and individual-level clustering (Unadjusted) Variable Adjusted n Odds Ratio 95%CI P-value Unsafe Enterococci count 528 26. 55 (1. 45, 486. 04) 0. 03 31. 33 (2. 13, 461. 73) 0. 01 Age < 5 years 528 2. 36 (0. 95, 5. 84) 0. 06 2. 92 (0. 85, 10. 03) 0. 09 Treat Water 528 0. 41 (0. 14, 1. 22) 0. 11 Place to wash hands near toilet 528 1. 87 (0. 46, 7. 52) 0. 38 Mixed effects logistic regression (accounting for household and individual-level clustering) using a backwards selection procedure; nested models were evaluated using likelihood ratio tests with a p<0. 1 threshold. ‘Treat water’ and ‘Place to wash hands near toilet’ were dropped from final model. P-values <0. 05 are underlined.

Discussion • Within the limitations of the study design, data and analysis – Enterococci counts and not Ecoli and total coliforms is potentially associated with HCGI. • Risebro et al (2012) reported a strong association between Enterococci counts and infectious intestinal disease as opposed to E. coli counts in England – Reported household treatment of drinking water was negatively associated with HCGI only in the unadjusted analysis 0. 41 (0. 22, 0. 75) <0. 01 – Place to wash hands near toilet was positively associated with HCGI only in the unadjusted analysis (OR 2. 12 (1. 08, 4. 17)

Conclusion • Given higher percentage of samples that were unsafe (E. coli and total coliforms) makes these potentially poor predictors (i. e. , there is E. coli everywhere! • In this urban setting, Enterococci counts have a potential association with HCGI than E. coli or total coliform counts. • Need to validate Enterococci counts in this and other settings as stronger indicator for risk of HCGI

Acknowledgements • Funding – Research funding NIH-through my supervisor: Joe Brown – Travel scholarship- BMGF through UNC Water & Health Institute • Tropical Gastroenterology group – Research host site/study research team • LSHTM-EHG – Host research group for my Ph. D • Centre for Infectious Disease Research in Zambia – Enteric Disease and Vaccines Research Unit • Employers, local Ph. D advisory team • Misisi community – Study participants

Comments and Questions?