A model of the costs of community and nosocomial pediatric respiratory syncytial virus infections in Canadian hospitals

BACKGrounD: Approximately one in 10 hospitalized patients will acquire a nosocomial infection (NI) after admission to hospital, of which 71% are due to respiratory viruses, including the respiratory syncytial virus (RSV). NIs are concerning and lead to prolonged hospitalizations. The economics of NIs are typically described in generalized terms and specific cost data are lacking. oBJeCtIve: To develop an evidence-based model for predicting the risk and cost of nosocomial RSV infection in pediatric settings. MethoDs: A model was developed, from a Canadian perspective, to capture all costs related to an RSV infection hospitalization, including the risk and cost of an NI, diagnostic testing and infection control. All data inputs were derived from published literature. Deterministic sensitivity analyses were performed to evaluate the uncertainty associated with the estimates and to explore the impact of changes to key variables. A probabilistic sensitivity analysis was performed to estimate a confidence interval for the overall cost estimate. resuLts: The estimated cost of nosocomial RSV infection adds approximately 30.5% to the hospitalization costs for the treatment of community-acquired severe RSV infection. The net benefits of the prevention activities were estimated to be equivalent to 9% of the total RSV-related costs. Changes in the estimated hospital infection transmission rates did not have a significant impact on the base-case estimate. ConCLusIons: The risk and cost of nosocomial RSV infection contributes to the overall burden of RSV. The present model, which was developed to estimate this burden, can be adapted to other countries with different disease epidemiology, costs and hospital infection transmission rates.


A model of the costs of community and nosocomial pediatric respiratory syncytial virus infections in Canadian hospitals
An RSV infection can be acquired either in the community (CRSV) or in hospital, which is known as a nosocomial RSV infection (NRSV).A nosocomial infection is defined as an infection that is not present at admission, but is acquired during hospital care (6).Important risk factors for NRSV include prolonged length of stay and admission to an intensive care unit (ICU) (6).Among the pediatric patient subgroups, those at highest risk for NRSV are infants younger than one year of age, children in the pediatric or neonatal ICU, and infants born prematurely and/or with low birth weight (<1000 g) (6,7).In 1999, a United States (US) national survey of pediatric ICUs found that 11.9% of patients acquired a nosocomial infection (8).In their study involving an outbreak of NRSV at the Royal Liverpool Children's Hospital-Alder Hey (Liverpool, United Kingdom) pediatric ICU, Thorburn et al (9) noted that 73% (11 of 15) of the patients who developed an RSV infection had at least one of the following comorbidities: congenital heart disease, chronic lung disease, airway abnormalities or immunosuppression.
NRSV is considered to be preventable.In 2000, Zoutman et al (10) surveyed Canadian hospitals with >80 acute beds to assess infection surveillance and control activities.Results from 172 hospitals (72% of total) identified deficits in key components of effective infection-control programs, which led the authors to conclude that greater investment in resources was required to meet recommended standards.The abovementioned study, however, did not examine the resources required to achieve reductions in NRSVs, or the expected impact of these preventive efforts.
The goal of the present study was to develop an economic model of pediatric RSV infection costs related to admissions for CRSV, infection-control measures and their costs, and additional treatment costs of NRSV cases.

Cost model
The target variable of the model was the average cost of activities related to the treatment and infection control of RSV infection in hospital.The model was expressed in terms of the average confirmed CRSV hospital admission.The model explicitly included costs related to the treatment of CRSV cases, infection-control activities and the treatment of nosocomial infections that were secondary to a CRSV hospital admission (Table 1).
The variable average cost of a CRSV hospital admission (C r ) in Table 1 represents all hospital and physician costs related to the treatment of a CRSV admission, and implicitly includes the cost of treating patients who require treatment within ICUs.C r was based on the product of the average Alberta (Canada) per diem cost for a case mix group related to respiratory infections and inflammations (11) and the average length of stay (ALOS) of CRSV hospital admissions (12,13).Associated physician costs were estimated based on the Alberta medical fee schedule (14).
Intensive infection-control activities are performed in Canadian pediatric wards and/or hospitals when RSV infection cases are detected, resulting in a reduced number of nosocomial cases than would have otherwise occured.Infection control includes strict hygiene, isolation of infected patients, diagnostic testing, and education for staff members and parents.The infection-control cost variable (C ic ) included the average costs of all diagnostic testing per CRSV case as well as costs related to contact isolation including supplies (gloves and gowns) and administration costs (nursing salaries and benefits).The estimates for all cost variables related to infection control involved with a CRSV admission were derived from a 1996 US study by Macartney et al (15), which reported the results of an implemented infection-control protocol; the patient numbers from the study by Macartney et al were used to obtain average prevention costs.The average cost of diagnostic testing was estimated as the product of the unit cost of an RSV test and the average number of all RSV tests undertaken (including negative results) per CRSV admission.Supply costs were estimated as the product of reported per diem supply costs and the estimated Canadian ALOS for CRSV.It should be noted that a small portion of infection-control costs are semi-fixed and hospitalwide and, as such, were considered to be variable.
For each CRSV, there is a probability that the infection will be transmitted to patients who were not infected with RSV and that a nosocomial case will occur.The resulting NRSV contributes to a longer stay and greater cost.The average cost of treating nosocomial cases per CRSV admission variable (C n ), was calculated as the product of the estimated cost of treating a single NRSV case and the rate of transmission.This additional cost per case of NRSV was derived by multiplying C r by an estimated ratio of the additional cost of treating NRSV cases to the cost of a CRSV admission (11,16).
To avoid mixing US and Canadian cost data, the nosocomial cost ratio was calculated using a ratio of two ratios.The numerator was an estimate of the additional cost (US$) to treat a NRSV divided by the cost of a typical hospital case (3.12) (16).The denominator was C r (Canadian$) divided by the cost of a typical hospital case (0.78) (11).The resulting ratio provides an estimate of the additional cost of treating a NRSV case compared with the cost of treating a CRSV admission alone (4.0=3.12/0.78).
The rate of transmission, defined as the expected number of NRSV admissions per CRSV admissions, was estimated to be 6.1% from results reported by Simon et al (7) from a study involving German hospitals.The rate of 6.1% is similar to a previous Canadian estimate of 6.4% (17).
The model presented in the present study reflects a balance between the increase in infection-control costs and the reduction in nosocomial cases and associated costs.All costs were estimated in 2007/2008 Canadian dollars.Prevention costs were adjusted to 2007/2008 prices using a US all-items price index (18) and converted to Canadian dollars using the purchasing power parity method (19).All other costs, already in Canadian dollars, were adjusted to the reference year using the all-items Canadian consumer price index (20).

Deterministic sensitivity analyses
Three sensitivity analyses were undertaken.A tornado analysis was performed in which four key variables (ALOS, hospital inpatient cost, NRSV cost ratio and transmission rate) were varied equally by ±10% to assess their relative impact on average cost.
A second analysis explicitly accounted for the proportion of CRSV patients admitted to the ICU and the relatively higher cost of treatment within an ICU unit.Randolph et al (21) reported that 7.4% of CRSV patients admitted to hospital were subsequently admitted to the ICU and had an ALOS of 8.6 days; four days on the pediatric ward and 4.6 days in the ICU (21).A recent Canadian study reported that the ICU per diem cost ($2,335) was approximately 2.5 times greater than the pediatric ward cost ($957) (22).These data were used to re-estimate the overall cost of a CRSV admission, including infection prevention and nosocomial treatment costs.
The final deterministic analysis assessed the impact of infection control on average cost.Previous research has shown that an active prevention and infection-control program resulted in a 39% reduction in nosocomial cases (15).Conversely, this implies that the transmission rate in the absence of prevention would be 64% (1/[1-0.39])higher than the rate following the adoption of infection-control practices.Thus, in the absence of an infection-control policy, it was assumed that the expected transmission rate would increase to 10.5%, compared with 6.1% with infection control.

Probabilistic sensitivity analysis
Using Monte Carlo simulation, a probabilistic sensitivity analysis (PSA) was performed to simultaneously assess the uncertainty related to all estimates used in the model.This analysis, based on 5000 replications, provided a 95% CI for average cost.For all variables, except hospital inpatient cost, a triangular distribution was used because estimates of SD were unavailable.Because the estimates were based on source studies with large samples (n≥100) or large administrative databases, it appeared reasonable to specify a maximum/minimum range of ±10% from the mean.For the nosocomial cost ratio, a range of ±50% was used to reflect the greater variability expected, because this variable was derived from two other ratios.Because hospital inpatient cost

resuLts
Base-case analysis (table 3) C r was $4,069, based on the per diem cost of $1,287 (after adjustment for inflation), an inpatient ALOS of three days and $207 of physician fees.
C ic was $247 per CRSV case, which included an average cost of $196 for diagnostic testing, $32 for the cost of supplies and $19 for administration over the three-day ALOS.
C n was four times C r ($4,069), or $16,276 per case.The rate of transmission of NRSV was found by Simon et al (12) to be 6.1% of CRSV hospitalizations; multiplying the total additional costs by this rate produces a C n of $993.Thus, C ic and C n added approximately 30.5% to the inpatient cost of each CRSV case, bringing the total average per case cost of treating an RSV hospital admission up to $5,309, from $4,069.

Deterministic sensitivity analyses
The tornado analysis (Figure 1) determined the relative impact of four key input variables on the overall average cost when they were varied by an identical percentage.The variations in the estimates of ALOS and inpatient cost per day had the greatest impact, because the cost of the CRSV admission was the major component in the overall average cost.The nosocomial cost ratio and the transmission rate had a modest impact on the overall average cost.
The second sensitivity analysis explicitly accounted for the proportion of CRSV patients admitted to the ICU, the higher average length of stay of those patients and the relatively higher per diem cost associated with the ICU unit.The result of these changes was an overall average cost of $5,155 per CRSV admission, which was only approximately 3% less than the base-case analysis, indicating that the basecase analysis was reasonably robust.In the final deterministic analysis, the transmission rate was increased to 10.5% to reflect what the rate may be if no infectioncontrol activities were undertaken.In the absence of a hospital-based RSV infection-control program, C n would increase by $716 (14%) per CRSV hospital admission.However, when C ic was deducted, the net increase in overall cost was $469 per CRSV admission, or approximately 9%.Conversely, this estimate would represent the average savings, per CRSV admission, due to infection-control activities.

PsA
The PSA produced a near normal distribution around the base-case result of $5,309, providing a 95% CI of $4,597 to $6,097, or approximately ±14% (Table 4).This is a reasonably tight 95% CI; however, the interval is highly dependent on the dispersion of the sampling distributions of the input variables.The SE of many of the distributions was based on an assumption, which should be considered when evaluating the 95% CI.

extrapolation of C n to the national level
In the base-case analysis C n was $993.By multiplying this value by the projected number of RSV cases in the Canadian pediatric population <1 year of age, the cost to treat NRSV infections in the hospital setting was estimated to be approximately $7.9 million per year (Table 2).

DIsCussIon
We estimated the full cost associated with pediatric CRSV hospital admissions, including the costs of the original CRSV cases, and the additional costs to prevent and treat NRSV cases.Using primarily Canadian data, we estimated that NRSV adds $993 to each CSRV case; the original cost was approximately $4,000.Using a sensitivity analysis, we estimated that infection-control procedures, costing $247 per case, lead to a reduction in transmitted treatment costs of $469.
Although a number of studies have estimated RSV-related costs, only one included transmitted cost estimates (15).We presented our analysis in the form of a model, which enabled estimates to be made in a variety of environments and situations.Furthermore, we applied a variety of sensitivity analyses, which add considerable confidence to our results.Many hospital planning departments do not collect data in the format that was used in the present study.However, most of the variables in our model are readily available from different departments in the hospital.These include RSV testing from the hospital infection control department, infection-control supplies from the hospital inventory, and length of stay and treatment costs from the hospital financial department.We took into consideration the best evidence available on respiratory infectious disease transmission rates and noted that the 6.1% applied in our model is within published ranges for nosocomial infection transmission (24)(25)(26).
Generally speaking, resource utilization rates and unit cost vary considerably among countries.We attempted to use Canadian cost data as much as possible to consistently reflect the cost experience of one jurisdiction.It was a limitation of the present study that we were unable to use Canadian cost data for all the input variables.Where it was necessary to use US data, we incorporated it into relative measures in as many instances as possible.When absolute data were used, we converted the US costs to Canadian costs using the purchasing power parity method (19).
A specific limitation of our estimate of C ic was that the source study we used did not include the use of masks in the infection prevention program, whereas the use of masks has become a standard component of more recent infection control protocols (9).Because the cost of masks would represent a very small additional amount, relative to the total cost per CRSV admission, we have not attempted to adjust the data obtained from our source study.
Most hospitals should have the data required to perform these basic cost-effectiveness calculations; those that do not can use the estimates that we provide.This, of course, will limit the applicability to a specific hospital.However, this is a very simple model, and should be well within the grasp of most infection-control and finance departments.

ConCLusIon
Results of our model indicated that the cost of treating NRSV infection significantly contributed to the overall cost of CRSV hospital care.Public health prevention strategies to reduce the prevalence of severe pediatric RSV infection should be considered and their impact on costs evaluated.

ACKnoWLeDGeMents: The authors acknowledge SNELL Medical
Communication for their assistance with literature searches, manuscript editing and proofing.Writing and subsequent content revisions were performed entirely by the authors, who take full responsibility for the article.
Nosocomial infection: A risk factor for a complicated course in children with respiratory syncytial virus infection -results from a