Post-pandemic Review of anti-Influenza Drug Effectiveness (PRIDE - PowerPoint PPT Presentation

Post-pandemic Review of anti-Influenza Drug Effectiveness (PRIDE Study): An investigation of the impact of neuraminidase inhibitor antiviral use on pneumonia and length of hospital stay in hospitalized influenza A(H1N1)pdm09 patients: an individual participant data meta-analysis • Dr Stella Muthuri • Dr Puja Myles • Dr Sudhir Venkatesan • Dr Jo Leonardi-Bee • Dr Wei Shen Lim • Professor Jonathan Van-Tam • PRIDE Research Consortium 1/21/2015 The Fifth ESWI Influenza Conference , Riga, 17 th September 2014

Conflict of interest  This study is taking place through an unrestricted educational grant funding from F. Hoffmann-La Roche – The study is being undertaken fully independently of the company, which has had/will have: • no input to the project design; • no access to any of the data; • no role in analysis or data interpretation; • no preview of the study results; • no opportunity to preview or comment on this presentation or any manuscripts arising from the work. The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

Authors’ declarations • JVT: Between October 2007 and September 2010, ad hoc paid consultancy to both major NAI manufacturers (GlaxoSmithKline plc, F. Hoffman-La Roche). Former employee of SmithKline Beecham (now part of GSK) and Roche, 2000 - 2002. He has no outstanding interests related to shares, share options or accrued pension rights since 2004. • Other Nottingham authors: nil The Fifth ESWI Influenza Conference , Riga, 17 th September 2014 1/21/2015

Background • Influenza-related pneumonia was a common severe complication during the 2009-10 influenza pandemic • Prior to the 2009 pandemic, evidence of NAI effectiveness in seasonal influenza was strongest in terms of symptom alleviation in healthy outpatients, but less robust for reductions in complications (such as pneumonia) • Since then, several studies support the potential benefit of NAIs during 2009-10 but lack statistical power to be conclusive The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

Preliminary work Journal of Infectious Diseases, 2013: 207 (15 February) Study based on published manuscripts Outcomes studied: – Mortality – Severe outcomes: admission to HDU/ICU or death – Influenza(AH1N1)pdm09 related Pneumonia The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17th September 2014

Key findings from meta-analysis of published studies on Influenza-related pneumonia (Muthuri et al 2013) Limitations of meta-analysis: •High degree of heterogeneity between studies •Most studies did not provide adjusted risk estimates •Where adjusted risk estimates were available there were differences in the extent of adjustment or confounders included in final models •Inability to adjust for propensity to receive treatment The Fifth ESWI Influenza Conference , Riga, 17 th September 2014 1/21/2015

Objective To carry out an individual patient data (IPD) meta- analysis to investigate the impact of NAI use on influenza-related pneumonia and hospital length of stay among patients admitted to hospital with influenza A(H1N1)pdm09 Prospero Registration: CRD42011001273 The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

About the PRIDE Study • 80 Research groups* from 38 countries and 6 WHO regions have provided data on 168,117 individual patient observations * Two research groups provided exclusively outpatient data (excluded from current analysis). • Of these 29,259 patients with known data on NAI treatment were hospitalised. – Pneumonia analysis: 56 datasets (n=7,967; 27%) with data on pneumonia status confirmed using radiographs – Length of hospital stay: 72 datasets (n=20,446; 70%) with data on duration of hospital stay The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

Methodology Standardisation of datasets • Standardised definitions created for confounder variables such as ‘comorbidity’, ‘disease severity’ •Outcomes defined as: – Influenza-related pneumonia: no pneumonia (normal on radiograph); pneumonia (positive for pneumonia on radiograph) – Length of hospital stay: duration in days between admission & discharge/ death • Exposure defined as: – NAI treatment (at any time) vs no NAI treatment – Early NAI (≤2 days of symptom onset) treatment vs later NAI (>2 days after symptom onset) treatment – Early NAI (≤2 days of symptom onset) treatment vs no NAI treatment •Propensity scores (for receiving treatment) calculated using a regression model with ‘treatment’ as the outcome variable and covariates (minimum model): age, sex, comorbidity (yes/no), disease severity at admission (yes/no) The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

Statistical analysis • Pneumonia analysis: Multilevel logistic regression modelling† • Length of stay analysis: Multilevel mixed-effects negative binomial regression† • Adjustment carried out for: propensity score (by quintile), steroid use in hospital, antibiotic treatment • Stratified analyses: by age group (adults, children), pregnant women, and laboratory confirmed A(H1N1) • Analyses conducted using Stata ™ V.13 † baseline outcome rates were allowed to vary between the studies through using a random intercept The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

Characteristics of hospitalised patients included in analyses Overall sample Pneumonia Length of stay (All hospitalised analysis (LOS) analysis Characteristics patients) n= 7,967 (27%) n=20,446 (70%) n= 29,259 (100%) Age, years; median (IQR) 26 (11- 44) 29 (13 - 48) 27 (11-46) Males 14,442 (49%) 4268(54%) 10,219 (50%) Population groups Adults (ages ≥16 years) 19,834 (68%) 5,692 (71%) 14,169 (69%) Children (ages <16 years) 9,225 (32%) 2,258 (28%) 6,205 (30%) Pregnant women (ages:13-54 years) § 2,167 (23%) 374 (16%) 1,348 (21%) Laboratory confirmed A(H1N1)pdm09 25,026 (86%) 7,603 (95%) 16,710 (82%) Clinically diagnosed A(H1N1)pdm09 4,233 (14%) 364 (5%) 3,736(18%) Symptom onset to hospital admission; 2 (1-5) 3 (1 - 5) 3 (1-5) days, median (IQR) NAI (at any time) 18,811 (64%) 6,764 (85%) 13,722 (67%) Early NAI (≤2 days) 5,996 (20%) 1,782 (22%) 4,060(20%) Mortality 2,784 (10%) 706 (9%) 2,229 (11%) § proportions based on women between ages 13 & 54 years The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

Results: Influenza-related pneumonia Characteristics of pooled sample of hospitalised patients (N= 7,967) Total in No Pneumonia Characteristics pneumonia pneumonia n=4521 analysis n= 3446 Population groups Adults (ages ≥16 years) 5,692 (71%) 2,371 (69%) 3,321 (73%) Children (ages <16 years) 2,258 (28%) 1,066 (31%) 1,192 (26%) Pregnant women (ages:13-54 years) § 374 (16%) 192 (19%) 182 (14%) Laboratory confirmed A(H1N1)pdm09 7,603 (95%) 3,238(94%) 4365 (97%) Clinically diagnosed A(H1N1)pdm09 364 (5%) 208 (6%) 156 (3%) Males 4268(54%) 1852 (54%) 2416 (53%) Age, years; median (IQR) 29 (13 - 48) 25 (11 - 45) 33 (14 - 50) Symptom onset to hospital admission; 3 (1 - 5) 2 (1 - 4) 3 (1-6) days, median (IQR) Mortality 706 (9%) 92 (3%) 614 (14%) § proportions based on women between ages 13 & 54 years The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

Treatment characteristics of hospitalised patients included in pneumonia analysis Exposure definition Total in No pneumonia Pneumonia pneumonia n= 3,446 n=4,521 analysis n= 7,967 NAI (at any time) 6,764 (85%) 2,760 (80%) 4,004 (89%) Oseltamivir (at any time) 6,705 (99%) 2,713 (98%) 3,992 (100%) Zanamivir (at any time) 129 (2%) 65(2%) 64 (2 %) Peramavir (at any time) 47 (0.6%) 5 (0.2%) 42 (1%) Time of treatment after symptom onset Early NAI (≤2 days) 1782 (26%) 1,027 (37%) 755 (19%) Later NAI (>2 days) 2829 (42%) 968 (35%) 1861 (46%) NAI timing data missing 2153 (32%) 765 (28%) 1,388 (35%) 3 (2-6) 2 (1-4) 4 (2-7) Symptom onset to antiviral treatment; days, median (IQR) The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17 th September 2014

Outcome: Influenza related pneumonia Exposure: NAI treatment at anytime vs. none Adjusted † OR (95% CI) Population Subgroups Crude OR (95% CI) Lab confirmed cases (all ages) 1.62 (1.35 – 1.94) 1.28 (1.05 - 1.55) Lab confirmed and clinical 1.65 (1.39 - 1.96) 1.31 (1.09 - 1.58) diagnoses (all ages) Adults (16 years and above) 1.56 (1.26 – 1.94) 1.25 (1.00 - 1.57) Children (under 16 years) 1.42 (1.03 – 1.98) 1.30 (0.93 – 1.83) Pregnant (13 - 54 years) 1.19 (0.52 – 2.71) 0.80 (0.31 - 2.07) ICU patients Adults (16 years and above) 1.62 (1.02 – 2.59) 1.49 (0.92 – 2.42) Children (under 16 years) 1.55 (0.95 – 2.52) 1.47 (0.89 – 2.40) * Radiologically confirmed pneumonia vs no pneumonia on CXR † Adjusted for propensity score quintile, steroid use in hospital, antibiotic treatment in hospital Statistically significant results in bold The Fifth ESWI Influenza Conference , 1/21/2015 Riga, 17th September 2014