Transcript Slides - View the full AIDS 2016 programme

Pharmacovigilance to inform
policy: experience in South Africa
Karen Cohen
Division of Clinical Pharmacology
University of Cape Town
HIV programmatic pharmacovigilance
Pharmacovigilance: “Detection, assessment, understanding and
prevention of short and long term adverse effects of medicines”
• Clinical studies short
• Co-morbidities, concomitant medicines, genetic variability
• Risk versus benefit:
– early treatment initiation
– prevention
• Focus on serious adverse drug reactions (ADRs)
– Resulting in hospitalisation and death
– Treatment limiting ADRs- drug substitutions
South African context
• Largest ARV treatment programme in the world
– ± 3 million people on ART
• High rates of concomitant HIV and TB treatment
• Growing burden of non-communicable diseases
• Methods:
– Hospital-based surveys
– Sentinel Cohorts
– Spontaneous reporting
Hospital-based surveillance
• 4 South African hospitals in 2013
• 8.4% medical admissions in SA due to ADR (164/1951)
(Worldwide 5.3% of admissions)
– ART, TB treatment and/or co-trimoxazole implicated in 34%
• In 16% of in-hospital deaths ADR implicated (56/357)
– Most commonly implicated: tenofovir, rifampicin, co-trimoxazole
• ADR contributed to death of 2.9% of medical admissions
(Europe, UK, USA 0.05 to 0.32% of admissions)
Mouton et al 2015 Br J Clin Pharmacol 80(4); Mouton et al 2016 Medicine 95(9);Kongkaew Ann
Pharmacother. 2008; 42: 1017 ; Juntti-Patinen et al 2002 Eur J Clin Pharmacol 58; Davies et al 2009
PloS ONE 4:e4439; Pirmohamed et al 2004 BMJ 329 ; Lazarou et al 1998 JAMA 279: 1200
Hospital-based surveillance
Top 4 ADRs resulting in medical admission in South African hospitals (n=1951)
Renal
impairment
Hypoglycaemia
Liver injury
Haemorrhage
(n=22)
(n=20)
(n=19)
41
61
35
67
71%
5%
90%
16%
(n=24)
Median age (yrs)
HIV infection
Commonly
implicated drugs
tenofovir (46%)
ACE-I (38%)
insulin (64%)
TB drugs (60%)
sulfonylureas(50%) efavirenz (20%)
warfarin (68%)
NSAIDs (32%)
46%
18%
35%
16%
Median stay
9 days
6 days
10 days
6 days
Preventable
46%
77%
15%
58%
Mortality
• ADRs in HIV patients- high mortality, prolonged admission
• Importance of looking at all drug exposures
Mouton et al 2016 Medicine 95(9)
Sentinel cohorts
• Valuable resource for ADR surveillance
– Requires fewer resources than setting up cohorts solely
for toxicity surveillance (Cohort event monitoring)
• Robust denominator data
– Can determine incidence of treatment-limiting ADRs
– Can identify risk factors
• Cohort collaborations increase study power
– Important for studying rare but serious events
d4T
0.25
0.10
0.15
0.20
Adults on first-line ART in 2 Western Cape cohorts
(Khayelitsha and Gugulethu)
0.05
AZT
NVP
EFV
0.00
Proportionhaving
havingdrug
drug substituted
substituted
Proportion
Data on stavudine toxicity from sentinel
informed
policy
Kaplan-Meiercohorts
failure estimates,
by drug*
0
1
2
Time on
drug
in years
Time
onindividual
individual
drug
(yrs)
3
Changed by 3
months (%, 95% C
7.8 (5.9-10.
1.9 (1.3-2.8
7.6 (5.3-10.
20.8 (16.2-26.
n
2007:
30
mg all weights,
point of care 295
lactate meters, avoidance
in
AZT
676
469
126
EFV
1,613education HCW
858
334
74
obesity,
NVP
1,062
376
75
44
Dramatic
rates, severity
mortality
d4T
1,996 decrease referral
782
137 at admission,15
Boulle et al 2007 Antivir Ther;12:753; Schutz 2010 AIDS Res Ther, 7:13
Percentage with new onset diabetes
ARV exposure and diabetes
Diabetes incidence:
13.24 per 1000 PYFU
Associations with diabetes:
• Older ager
• Higher BMI
• Efavirenz, zidovudine and
stavudine exposure
• Other diabetogenic meds
Time on antiretroviral therapy (yrs)
Karamchand et al 2016 Medicine 95(9)
Spontaneous reporting
• Does not give prevalence/incidence
• Signal detection
• ADRs that trouble HCWs
– Guide HCW training and clinical support
– Nurse-driven services
• Need responsive systems
– Telephonic; online reporting in addition to paper-based
– Prompt, individualised feedback and clinical support
• SA National HIV & TB HCW hotline most frequent ADRs:
–
–
–
–
–
Rash (efavirenz)
Liver injury (tuberculosis treatment and efavirenz)
Kidney injury (tenofovir)
Gynaecomastia (efavirenz)
Neuropsychiatric (efavirenz)
Chugley et al 2015 AIDS 29:503. Njuguna et al 2015 Drug Saf doi:10.1007/s40264-015-0359-8
Conclusions
• Pattern of ADRs is different in SA
• Large burden of serious ADRs in PLWHA and/or TB
• ADRs due to ARVs AND concomitant medicines
• Resource limited settings
– Create systems that can address multiple questions
• Communication and feedback
Acknowledgements
• National Dept of Health SA
– Mukesh Dheda
– Yogan Pillay
• CDC and PEPfAR
– Ehimario Igumbor
– Getahun Aynalem
• UCT Division of Clinical
Pharmacology
–
–
–
–
–
–
–
–
Hannes Mouton
Christine Njuguna
Annemie Stewart
Melony Fortuin- de Smidt
Marc Blockman
Gary Maartens
Annoesjka Swart
Jackie Jones
• Medicines Information Centre
• Hospital sentinel sites
–
–
–
–
Douglas Wilson
Andy Parrish
Peter Raubenheimer
Karl Technau
• UCT School of Public Health
–
–
–
–
Mary-Ann Davies
Renee De Waal
Ushma Mehta
Andrew Boulle
• IeDEA-SA data centre staff
• Patients
• Health care workers who
report
Top 5 ADRs causing admission
Median age (yrs)
HIV infection
Commonly
implicated drugs
Mortality
Median stay
Renal
impairment
Hypoglycaemia
Liver injury
41
61
35
67
41
71%
5%
90%
16%
79%
tenofovir (46%)
ACE-I (38%)
insulin (64%)
sulfonylureas (50%)
TB drugs (60%)
efavirenz (20%)
warfarin (68%)
NSAIDs (32%)
cotrimoxazole
(29%)
46%
18%
35%
16%
14%
9 days
6 days
10 days
6 days
6 days
Haemorrhage Blood dyscrasia
Pharmacovigilance
Detection, assessment, understanding and prevention of short
and long term adverse effects of medicines
Adverse drug reaction (ADR)
A response to a medicine which is noxious and
unintended, and which occurs at doses normally used in
humans (WHO).
Tenofovir and renal function monitoring
100
90
80
eGFR (mL/min)
110
• Incidence of eGFR<30mL/min: 15/1000 person years (95%CI 14 to 17)
• Older patients, advanced disease, baseline renal impairment, PI at risk
• Implication for monitoring guidelines
0
3
6
9
15
12
Time (months)
Baseline eGFR>=90 mL/min (n=7 335)
Baseline eGFR<90 mL/min (n=3 858)
De Waal et al TUPEB035
18
21
24
Sentinel cohort data for ADR analyses
• Data cleaning including record review to resolve
queries
– Requires resources
– Drug starts and stops not always accurately recorded
– Reasons for single drug substitutions
• Linkage outpatient cohorts to hospitalisation events
• Improved recording of pregnancy data in cohorts