الفهرس 
TitleOnly 14 pages are availabe for public view
 |  | from | 151 |  |  |
| | | from | 151 | | |
Abstract
Hospital-acquired infection (HAI) is a preventable adverse event that occurs with
relatively high frequency and leads to significant damage imposing substantial economic
outcomes on the health care system. In United Kingdom, National Health Service found that
approximately 15% of hospital acquired infections are preventable. So, reduction of HAIs has
been the main target of quality improvement to efficiently prevent HAIs. This makes
quantifying the cost of HAIs of increasing interest in health economics. Thus, the number of
healthcare economic studies that assessed the economic impact of HAIs has increased.
Reasonably accurate estimation of the additional costs of HAIs is essential to support
decisions related to the resources that should be assigned by hospitals to infection control and
the incentives that payers should pay to hospitals to decrease infection rates Different
methodologies in the literature have been used to produce accurate cost estimates. These
methodologies are case reviews, matched comparison analyses and regression analyses.
Although the latest guidelines allow researchers to use cost estimates from published
studies when assessing the economic impact of HAI control programs, it has been found that
these cost estimates may not be accurate and thus the decision to implement a program may
be misleading and may lead to undesirable outcomes. Also, these results may not be
necessarily comparable and cannot be applied to other settings. That is to say they are “not
transferable” due to differences in the conditions in which the original cost estimates were
produced.
Fukuda et al. conducted a systematic review of studies that estimated the cost of HAIs from
1980 to 2006 to assess the transferability of the cost estimates. Only eight papers (9%) out of 89
publications, had high transferability in which all cost components were mentioned, cost data in
each component and unit cost with actual costings.
This study aims to conduct a systematic review of published studies that estimated the
costs of hospital acquired infections, to determine the distribution and trend of the analytical
methodology used in estimating the costs of hospital acquired infections, to evaluate the quality
of the costing methods and the transferability of the cost estimates. Also, we will investigate if
studies of high transferability are correspondingly well regarded by other researchers.
In this systematic review, we searched MEDLINE to identify primary studies which
estimated original costs of hospital-acquired infections (HAIs) and published in English from
2007 to April 2021. We excluded studies which utilized existing cost obtained from other
published studies or included community acquired infections or that used infected patients in
the control group or studies in which the primary aim was economic evaluation of
intervention or infection control programs. We also excluded reviews, meta- analyses, letters,
conceptual papers, editorials, commentaries, and research protocols.
Identified studies were reviewed in two steps: an initial abstract review for screening,
followed by a full text review to confirm the eligibility of the studies, determine the analytical
methodology used in estimating the costs of HAIs and evaluate the transferability of studies.
This evaluation was based on two evaluation axes developed by Fukuda et al. which are the
Summary, Conclusions and Recommendations
70
transparency of cost estimates (Clarification of costing scope) and evaluating the costing
method. To investigate if other researchers more frequently use high transferability studies,
we compared number of citations by transferability axes using Poisson log-linear models.
After searching the MEDLINE and hand search the references of relevant studies then
title and abstract review, 171 were included.
Most studies were conducted in developed countries (150 studies, 87.7%). There were
83 studies conducted in the US, 40 in Europe, 31 in Asia, two in Africa and 14 in Australia,
Brazil, Canada and Mexico. Half (50.8%) of the studies were multicenter, 37 (21.6%) were
conducted in university hospitals and 25 (14.6%) in tertiary hospitals.
The most frequent analytical methodology used was matched comparison (71 studies,
41.5%), followed by regression models (58 studies, 33.9%), then regression on a matched
sample (19 studies, 11.1%), unmatched comparisons (16 studies, 9.4%), case reviews (8 studies,
4.7%), finally multistate modelling (3 studies, 1.8%) and survival analysis (one study, 0.6%).
Six studies compared total cost between HAI cases and controls using different analytical
methodologies. Four of them used matching as well as well as regression model. All of them
reported that the attributable cost of HAIs calculated by matching and regression were the same
except one study. Another study estimated the cost attributed to hospital- acquired clostridium
difficle infection using two time-fixed models as well as a time varying model and concluded that
the cost from time fixed models is similar. However, time fixed models resulted in cost which is
six times higher than that of the time varying analysis. Also, another study compared total cost
between HAI cases and controls using Linear regression model with logarithmic transformation
and Tobit (censored regression) model and concluded that the presence of nosocomial
bloodstream infections was associated with 62% increase in costs in adjusted model (Linear
regression model) and 65% increase in costs in adjusted and truncated model. (Tobit (censored
regression) model. We found the overall distribution of the studies through the different analytical
methodologies was not significantly different during both periods 2007-2013 and 2014- 2020.
Less than third of the studies (48 studies, 28.1%) accounted for the length of stay (LOS)
when estimating the cost. Seven (14.6%) studies used the whole LOS for adjustment while 28
(58.3%) studies correctly used the pre-infection LOS i.e., time to infection. Also, 13 (27.1%)
studies adjusted for only part of time to infection. The most common time varying methods
used to account for time to infection were matching and regression that used time to infection
or part of it to adjust for the time dependent bias of HAI. Less commonly used time varying
methods were multistate modelling and survival analysis.
More than half of the included studies did not state the economic perspective. The cost
of HAI was estimated from hospital/ provider perspective in 43 studies (25.1%) and from
payer perspective in 24 (14%) studies. Cost from patient perspective was reported in four
studies. In the developed countries, payer, patient and societal perspective were more
common than in developing countries.
Clarification of costing scope was graded from A to D. Only three studies (1.7%) were
A, 54 (31.6%) studies were B, 23 were C (13.5%), and 91 were D (53.2%). The most
frequent costing methods were the use of charges (65 studies, 38 %), 40 studies were with
unknown costing method (23.4%), 36 studies used actual costs (21.1%), 31 studies used
RCCs (18.1%), and one study (0.6%) used RVUs. from a total of 171 studies included in our
sample, only 16 studies (9.4%) had high transferability level in which the level of clarification
of the costing scope was A/B and costing method was α.
Summary, Conclusions and Recommendations
71
The overall distribution of the studies through the levels of the scope of costing was not
significantly different during both periods 2007-2013 and 2014- 2020. For the 60 studies
where economic perspective was stated, and with known costing method, 72.2% of the studies
that adopted provider/hospital perspective used actual costs directly or indirectly by
converting charges to actual cost using RVU or RCC, while 70.8% of the studies that
considered payer or patient, or societal perspective used unmodified charges.
The majority of the studies (60.2%) reported the mean or total cost. Median cost was
reported in 31 studies (18.1%). Thirty- three studies (19.3%) reported both median and
total/mean cost.
All studies reported inpatient costs. Outpatient costs were stated in 23 (13.5%) studies,
capital costs in 19 (11.1%), opportunity costs in 10 (5.8%) studies, out-of-pocket costs in 7
(4.1%) studies, and productivity losses in 6 (3.5%)
Eighty-three (48.5%) studies did not report price year. It is not clear whether they have
adjusted for inflation or not. Eighty (49.1%) studies adjusted for the effects of inflation. There
were different currencies across studies, US dollars (53.8%), Euros (19.3%), Japanese yen (5.3%),
Canadian dollars (4.1%) and British pound sterling (2.9%). Thirty-seven (21.6%) studies
converted original currency into US dollars and mentioned the method of conversion and
exchange rate. However, only four studies used the purchasing power for currency conversion.
Seventeen studies undertook sensitivity analysis to explore the impact of the uncertainty
characterizing the input values on the costs.
We concluded that over the past 14 years, matching and regression analysis were mostly
and likely used as analytical methodologies to control for confounding factors to minimize
bias. Most of the review studies did not account for the time dependent bias of HAI by using
time fixed methods. Among the time varying methods, matching and regression were the
most common used to adjust for time to infection while multistate modelling and survival
analysis were less likely used. Although there was no change in the proportion of the studies
that adjusted for LOS, researchers started to correctly use ―time to infection‖ instead of
”entire LOS” for adjustment. Researchers mainly adopted hospital perspective and neglected
patient and societal perspectives in the post discharge setting. After the sentimental article of
Fukuda et al., there was no significant improvement in the transferability of published studies
as still few studies had high transferability. charge was the most frequent method used for
costing and level D was the most common scope.