1474. Use of Surveillance Data to Assess the Sample Size in Clinical Trials for Testing Staphylococcus aureus Vaccine Efficacy in Orthopedics
Session: Poster Abstract Session: HAI: Surgical Site Infections
Friday, October 28, 2016
Room: Poster Hall

Background:

Staphylococcus aureus is the main etiological agent of surgical site infections (SSI). Patients undergoing total hip arthroplasty (THA) could be a worthwhile population for testing a potential anti-staphylococcal vaccine. Our objective was to assess the sample size for evidencing significant vaccine efficacy (VE) in a clinical trial.

Methods:

We used THA data from the national surveillance network of the SSI in South-East of France between 2008-2011. The outcome was the occurrence of S. aureus SSI within 30 days after surgery. Statistical power was estimated by simulations. Two groups of patients, stratified by gender and National Nosocomial Infection Surveillance (NNIS) index, were sampled with replacement 5,000 times. In the virtually vaccinated group, infection status was drawn from a binomial distribution, with the number of infected in this group and the theoretical vaccine efficacy (VE) as parameters. VE was derived from the relative risk of infection estimated by log-binomial regression adjusted for gender and NNIS index. Simulations were repeated for theoretical VE ranging from 20%-100% by step of 20% and for 6 sample sizes ranging from 250 to 8,000 individuals per arm.

Results:

18,688 patients older than 50 years and undergoing THA were included; 58% were women; 66 (3.5‰) S. aureus SSI occurred. To be able to evidence positive significant VE of 80% at 2.5% alpha risk for a one-tailed test with a probability of at least 80%, the sample size required would be at least 4,000 per arm (table 1). With lower VE, inclusion of 4,000 procedures per arm would give insufficient study power for detecting a vaccine effect.

Table 1. Statistical power (in %) according to vaccine efficacy (VE) and sample size

 

 

Sample size

VE(%)

250

500

1,000

2,000

4,000

8,000

20

1.2

3.1

5.8

6.9

8.5

11.3

40

3.1

7.0

11.5

15.2

22.8

38.4

60

4.8

12.8

21.4

31.4

49.9

79.1

80

6.9

20.6

40.8

58.9

84.7

98.7

100

9.7

31.0

72.3

97.3

100.0

100.0

 

Conclusion:

Simulations using real-life data such as surveillance data allow estimating power for clinical trials when events are rare and analytical formulas cannot be applied. This method could be a useful tool for defining sample size for vaccine clinical trials, particularly for the development of S. aureus vaccine.

 

Marie-Paule Gustin, PharmD, PhD1,2,3, Robin Ohannessian, MPH1, Marine Giard, MD1,4, Emmanuelle Caillat-Vallet, MD4, Thomas Bénet, MD1,2, Anne Savey, MD1,4 and Philippe Vanhems, MD, PhD1,2,3, (1)International Center for Infectiology Research (CIRI), Laboratory of Emerging Pathogens, UCBL1, Lyon, France, (2)Infection Control and Epidemiology Unit, Edouard Herriot Hospital, Hospices Civils de Lyon, Lyon, France, (3)Innovative Clinical Research Network in VACcinology (iREIVAC), Lyon, France, (4)Coordination Center for Healthcare Associated Infection Prevention and Control in South-East of France (CCLIN Sud-Est), Saint Genis Laval, France

Disclosures:

M. P. Gustin, None

R. Ohannessian, None

M. Giard, None

E. Caillat-Vallet, None

T. Bénet, None

A. Savey, None

P. Vanhems, None

Findings in the abstracts are embargoed until 12:01 a.m. CDT, Wednesday Oct. 26th with the exception of research findings presented at the IDWeek press conferences.