1. Introduction
Staphylococcus aureus is a common gram-positive bacterium of clinical significance causing skin and soft-tissue infections worldwide [
1,
2]. Approximately 10% to 30% of the population is estimated to be colonized with
S. aureus on the skin or in the nose [
3]. However, development of antibiotic-resistant strains such as methicillin-resistant
S. aureus (MRSA) has become a major health concern, especially for hospital settings and community-acquired infections [
4,
5]. MRSA is often found at higher incidence in healthcare settings as compared to community settings. In a U.K. study, nearly 2% of patients were colonized after admission [
6], while a U.S. study estimates that 4% of hospital inpatients are colonized [
7,
8]. Klevens et al. [
9] reported about 9000 observed cases of MRSA per year in the U.S., in which 58.4% were associated with healthcare settings and 26.6% were community-based. The national burden of MRSA infections in the U.S. in 2014 was about 72,000 infections [
10]. The Centers for Disease Control and Prevention Emerging Infections Program (EIP) population-based surveillance from 2009 to 2013 found a total of 4607 nursing-home onset and 4344 hospital-onset cases of invasive MRSA [
11].
The anterior nares is the primary reservoir of the
S. aureus in humans and the replication occurs followed by dispersal of the organism to the skin [
12,
13]. About 30% of all humans carry
S. aureus in their nose persistently, while another 20% to 30% carry intermittently [
14]. The typical transmission route of
S. aureus is from the nose to the hand of a person [
15], then to a surface (e.g., a door knob), and/or via the hand to the nose of a second person [
16,
17]. Activities involving close physical contact and the risk of minor injuries are positively correlated with
S. aureus spread and acquisition [
18]. Even a brief contact of fingers with a
S. aureus contaminated surface may cause the transfer of a large amount of organisms resulting in a potential infection hazard [
19]. The transfer rate is higher from moist contaminated surfaces than dry surfaces [
20,
21].
S. aureus can survive on dry surfaces between 2 and 4 days, and then can be easily transferred to hands and foods [
22]. Other experiments showed more than a day of survival in hospital fabrics (cotton, terry, blend, and polyester) to over 90 days of survival in polyethylene [
23]. These long survival times indicate a potential high risk of transmission of
S. aureus through the surface-to-hand pathway. Once
S. aureus is in the human body, it is believed to form biofilms, which makes the pathogen less vulnerable to host immune responses and allows them to cause colonization and local infections [
14].
S. aureus is an opportunistic pathogen and does not usually pose a fatal risk to humans even if it colonizes human mucosa or skin [
14]. However, in some cases,
S. aureus can cause severe or fatal infections.
S. aureus infections progress in five stages: colonization, local infection, systematic dissemination, metastatic infection, and toxinosis [
24]. Severe forms of
S. aureus infection include bacteremia, sepsis, pneumonia, endocarditis, and osteomyelitis [
25]. The causative agent of 50% of all cutaneous infections is
S. aureus [
26,
27]. Young children, the elderly population living in poor hygienic conditions, persons with diabetes and overweight conditions, and people living in high temperatures and humid conditions are particularly sensitive to
S. aureus infection [
28].
Quantitative microbial risk assessment (QMRA) is the process of characterizing health risk associated with pathogen exposures through environmental media [
29]. QMRA follows a four-step paradigm similar to chemical risk assessment which begins with hazard identification, followed by an exposure assessment to quantify the number of organisms a receptor (i.e., human) comes in contact with based on the fate and transport of the organisms across an exposure pathway (i.e., hand-to-surface-to-mouth). Dose –response models are generally developed from controlled animal or human trials to describe the mathematical relationship between a given exposure dose and the probability of an adverse health outcome (i.e., infection, illness, or death). Such models are quasi-mechanistic in that they are derived from mathematical models that describe the plausibility of biological processes resulting in a measurable health endpoint [
30] rather than the deep incorporation of mechanisms of in vivo physiological response. The final step in QMRA integrates the exposure dose prediction with the dose–response model to estimate risks with a characterization of the variability and uncertainty in the predicted values.
For the majority of pathogens with peer-reviewed dose–response models (primarily for ingestion, inhalation, and similar exposure routes), no manipulation of the exposed dose in human and/or animal trials is required to fit a dose–response relationship. Due to the testing procedures used to estimate
S. aureus infection—inoculation of the skin followed by occlusion which promotes growth—a transformation of the exposure dose is required prior to modeling the probability of infection [
28]. This article describes the development of a
S. aureus dose–response model using previously collected peer-reviewed data. The dose–response model is based on a model fit to a previously untested model to describe
S. aureus growth on skin that captures the
S. aureus growth and decay kinetics after inoculation (or exposure) with a low relative error and low correlation among estimated parameters as compared to the previous work in this area [
28]. As
S. aureus has recently risen to be among the leading causes of hospital-acquired infections, this new dose–response model should be a useful tool in estimating human
S. aureus risk in order to support risk management evaluation (e.g., test behavioral changes on risk reduction or surface decontamination strategies). While the best fit parameter of the dose–response model remains unchanged over the previous work, uncertainty bounds around this estimate were desirable and the previous fit of the kinetic model could not be reproduced, thereby generating an opportunity to illustrate the inverse problem parameter estimation approach in a novel context.