Overall, 574 studies were found. After excluding duplicates, 547 underwent title and abstract reading, and 27 were read fully. Twelve were selected to be included. The Flowchart it is shown in Fig. 1.
Motivations for exclusion are, study about health associated pneumonia without PAV in four (Niederman et al., 2014) (Collins and Schwemm, 2015) (Cock et al., 2009)(Ambrose et al., 1999), three were revision (Plosker and Figgitt, 2005) (Wang and Barrett, 2006)(Dorman, 2016), comments (Niederman, 2001) (Vanni, 2006), community pneumonia acquired (Edwards, Wordsworth and Clarke, 2012), one study without site of infection description (Varón-Vega et al., 2022), three about hospital pneumonia with any differentiation (Cara, Zaidi and Suleman, 2018), (Huon et al., 2020), (Spektrumlu et al., 2020) and two of them did not available to access fully in any bases (Spektrumlu et al., 2020) (Shorr, Susla and Kollef, 2004).
Key details of the included studies are in Table 2. Year of publication varied from 2005 until 2022. The USA was the most-represented country, four studies (Mcgarry et al., 2010)(Kongnakorn et al., 2015)(Zilberberg et al., 2010)(Merchant et al., 2008), two in Italian (Mennini et al., 2022)(Tichy et al., 2020), one in India (Sathe et al., 2017), UK (Vlachaki et al., 2022), Brazil (Machado et al., 2005), Greece (Tsaganos et al., 2016), Spain (Del and Neipp, 2005) and Chine one of them (Zhang et al., 2019).
Four studies included adulthood diagnosed with VAP (Mcgarry et al., 2010) (Merchant et al., 2008)(Del and Neipp, 2005)(Zilberberg et al., 2010). Three also related health care associated pneumonia (HAP) (Mennini et al., 2022) (Tichy et al., 2020)(Kongnakorn et al., 2015). One referred VAP and sepse (Tsaganos et al., 2016), and one VAP, HAP and Skin infection (Zhang et al., 2019a). Sathe et al. (2017) included VAP for Acinetobacter baumannii, Valachaki et al. (2022) adulthood with infection for Klebisiella pneumoniae CRE and Machado et al. (2005) VAP for Staphylococcus aureus meticilin-resistent in adulthood.
Table 2
Characteristics of the included studies: authors, journal, country, population, year and comparison.
Reference | Journal | Country | Population | Year | Comparison |
Mcgarry, et al. | Medical of economics | USA | Adulthood with VAP | 2015 | Doripinem versus imipinem |
Kongnakorn, et al | Current Medical Research & Opinion | USA | Adulthood, with nosocomial pneumonia, including VAP | 2015 | Doripinem versus imipinem |
Sathe, et al | Journal of critical care | INDIA | VAP for Acinetobacter baumanii | 2017 | New antibiotic (AAEE), adjuvant treatment associated with carbapenem |
Zilberberg, et al | surgical infection | USA | Adulthood with VAP | 2010 | Doripinem versus imipinem |
Merchant, et al | Clinical therapeutics | USA | Adulthood with VAP | 2008 | Doripinem versus imipinem |
Grau, et al | Journal of chemotherapy | SPAIN | Adulthood with VAP | 2005 | Linezolid versus vancomycin |
Vlachaki, et al | European Journal of Health Economics | UK | Adulthood with infection for CRE-KPC, including VAP | 2022 | Meropenem-varbobactam versus BAT (carbapenem, aminoglycoside, polymyxin B, colistin, tigecycline or ceftazidime-avibactam) |
Machado, et al | Brazilian Journal of Infectious Diseases | BRAZIL | Adulthood with VAP for MRSA | 2005 | Linezolid versus vancomycin |
Tichy, et al | Clinical therapeutics | ITALY | Adulthood with VAP or HAP | 2020 | Ceftazidime-avibactam versus meropenem |
Mennini, et al | Global & regional health technology assessment | ITALY | Adulthood with VAP or HAP | 2022 | Ceftalozone-tazobactam versus meropenem |
Tsaganos, et al | Antimicrobial Agents and Chemotherapy | GREECE | Adulthood with VAP (and sepsis) | 2016 | Clarithromycin, adjuvant treatment |
Zhang, et al | Antimicrobial Resistance and Infection Control | CHINA | VAP, HAP and Skin infection | 2019 | Antibiotics used in complicated infections x vancomycin as a control |
There was considerable variation in the economic modelling reported, the most prevalent was cost-effectiveness, seven studies used this model (Zilberberg et al., 2010)(Del and Neipp, 2005)(Machado et al., 2005)(Tichy et al., 2020)(Mennini et al., 2022) (Zhang et al., 2019a)(Vlachaki et al., 2022). Cost-minimization was stated to McGarry et al. (2010) and Kongnakorn et al. (2015). Cost-benefit was showed in one case (Tsaganos et al., 2016). Two studies analyzed overall costs between comparations (Merchant et al., 2008)(Sathe et al., 2017).
Cost-minimization is the appropriate to carry out when two medications have no significant difference in efficacy, then only the costs of the interventions need to be considered (Newby and Hill, 2003), commonly used data from non-inferiority trials (Mcgarry et al., 2010). Bias will be negligible for non-inferiority studies, cost-effectiveness analysis is almost always required to avoid biased estimation of uncertainty (Dakin and Wordsworth, 2013). We can also discuss about a ‘purchaser’ perspective, it can be argued that it is unacceptable to risk paying more for a drug that has failed to equivocally show superiority to an existing therapy. (Newby and Hill, 2003)
The choice of the modelling depends on several factors, such as who is the decision maker and purpose of analysis, how a decision maker values health outcomes, money, risks and resource available (Owens, 1998). For a new intervention to add to health, it must be more efficient per unit of resource than the least efficient currently funded intervention and ought to displace it, in whole or part (McCabe, et al., 2008). The acceptable cost-effectiveness limit is not clearly, in Brazil the value assumed to National Commission for the Incorporation of Technologies in the Unified Health System will be around R$ 1.361 – R$ 147.016 (Prado, 2015). It is difficult translate costs of health interventions in money units, so evaluations of cost-effectiveness and cost-utility substitute cost-benefit analyses (Prado, 2015).
Different modeling approaches were applied in pharmacoeconomic analysis. Tree decision model in three situations (Vlachaki et al., 2022)(Machado et al., 2005)(Zilberberg et al., 2010) and for short time horizon to Mennine et al. (2022). For long time horizon to Mennine et al. (2022) Markov cohort model was used. Discrete event micro simulation (DES) was used to Kongnakorn et al. (2015). The sequential simulation model was applied in one study (Tichy et al., 2020). Analytical decision model without any specification was cited to McGarry et al (2010), Grau et al (2005) and Zang et al (2019).
Ideally, all modeling entails abstraction from reality to capture essential details that are relevant to decisions, while ignoring details that add complexity without improving model reliability, validity or predictive accuracy. Simplicity is preferable whenever possible (Hay, 2004). The choose is dependent upon the complexity of the problem, the need to model outcomes over extended periods of time, and whether or not resource constraints and interactions of various elements in the model are required (Renée, 2021).
Most studies in our review showed two antibiotics as comparators, this fact makes it possible to choose a simpler model to analyze the impacts of treatments and her cost in VAP.
Tree Decision is a model more used in pharmacoeconomics studies of VAP found in this review. Decision trees and Markov cohort models are the most commonly used approaches in economic evaluation (Tsoi et al., 2015). Decision trees are the simplest modeling techniques and are most appropriate for modeling interventions in short time period (Haji Ali Afzali and Karnon, 2014), depicts patient health state changes with sequential probability branches from initial state or intervention, are extremely flexible and can be used to model nearly any pharmacoeconomic decision process to shorter-term (Hay, 2004), as cases of VAP when shorter-term outcomes are the aim and measure. Basic decision analysis considers only one sequence of events within one time-frame (Ademi et al., 2012), VAP has one stage commonly.
To overcome limitation about time-frame Markov models are particularly useful (Ademi et al., 2012), when a decision problem involves a risk that is ongoing over time (Sonnenberg et al., 1993), in Markov Models patients cannot transit between states. In VAP patients frequently changes between states, as cure to re-infection or adverse event to cure or sensibility to resistance of the treatment, it can be to motivate the less number of studies applied this modelling.
About discrete event micro simulation, this modelling is a stochastic approach, becoming more popular, nevertheless, focused in healthcare to address issues at an operational level mostly (Caro, 2005). Findings of a review bout DES in health care show a tendency to use this approach within emergency departments, patient clinical conditions, and medical centers seeking to allocate resources and improve times and efficiency (Peimbert-garc, 2021). Sequential simulation shows sequential treatments, which can be change for toxicity, efficacy or other reasons. This modelling is less used for investigated costs in infectious disease (Zheng, Pan and Sorensen, 2016).
All of them pharmacoeconomic modelling are based on conjecture, they rely on multiple data inputs, and as the number of inputs and modelling steps increase, so too will uncertainty in outputs, this uncertainty needs be expressed in the modelling outputs (Ademi et al., 2012). Recognize the relevance of statistical uncertainty and dealing with it appropriately are required to obtain unbiased results from health economic modelling studies (Nuijten, 2004). The simplest way to perform probability sensitivity analysis in practice is by Monte Carlo methods (Hagan, Stevenson and Madan, 2007), influence of the parameters on the final result is identified (Carrera-hueso, 2011).
Monte Carlo simulation was used for sensitivity analysis in five studies in our review (Mcgarry et al., 2010)(Kongnakorn et al., 2015)(Zilberberg et al., 2010) (Tichy et al., 2020)(Zhang et al., 2019). Deterministic and probabilistic analysis were applied to Mennine et al. (2022), Vlachaki et al. (2022) used deterministic analysis and Univariate simple to Grau et al. (2005). Four studies did not evaluate sensitivity (Sathe et al., 2017)(Merchant et al., 2008)(Machado et al., 2005)(Tsaganos et al., 2016). Details are given in Table 3.
Table 3
Summary of the methodologies and sensitivity analyzes applied in studies.
Reference | Year | Model type | Methodologies | Sensitivity analyzes |
Mcgarry, et al | 2015 | Cost-minimization | Analytical decision model | Montecarlo simulation |
Kongnakorn, et al | 2015 | Cost-minimization | Discrete event simulation | Montecarlo simulation |
Sathe, et al | 2017 | Overall costs | Dose-based Time of treatment value | Not assess |
Zilberberg, et al | 2010 | Cost-effectiveness | Tree decision model | Montecarlo simulation |
Merchant, et al | 2008 | Overall costs | Overall costs | Not assess |
Grau, et al | 2005 | Cost-effectiveness | Analytical decision model | Simple univariate (of sensitivity) |
Vlachaki, et al | 2022 | Cost-effectiveness | Tree decision model | Deterministic sensitivity analyzes |
Machado, et al | 2005 | Cost-effectiveness | Tree decision model | Not assess |
Tichy, et al | 2020 | Cost-effectiveness | Sequential simulation model | Montecarlo simulation |
Mennini, et al | 2022 | Cost-effectiveness | Markov-model (long time) Tree decision model (short time) | Deterministic and probabilistic sensitivity analysis |
Tsaganos, et al | 2016 | Cost-benefit | Cumulative cost was expressed as mean and standard error (SE). Comparisons between the two groups were done on each separate day by the Mann-Whitney U test | Not assess |
Zhang, et al | 2019 | Cost-effectiveness | Analytical decision model | Montecarlo simulation |
The outcomes present involved mortality, clinical cure, efficacy, hospitalization days, UCI days, time in ventilation mechanic, costs with adverse events, costs and life-years (LYs) saved, quality-adjusted life year (QALYs) gained and antibiotic resistance.
Costs were measure like total costs in hospital to McGarry et al. (2010), Zilberg et al. (2010), Merchant et al. (2008), Grau et al. (2005), Vlachaki et al. (2022) and Tsaganos et al., 2016. Three studies considered only costs with medication (Sathe et al., 2017)(Machado et al., 2005)(Zhang et al., 2019). Direct costs were evaluated to Tichy et al. (2022) without further description about which costs refer. Direct costs (treatment, specialist, hospitalization and adverse events) and indirect costs (social perspective and human capital) were analyzed in only one study (Mennini et al., 2022). Kongnakorn et al. (2015) does not make clear how costs were used and how were obtained.
All but four studies evaluated empirical therapy (Mcgarry et al., 2010)(Zilberberg et al., 2010)(Del and Neipp, 2005)(Tichy et al., 2020). Four related guided therapy (Sathe et al., 2017)(Vlachaki et al., 2022)(Machado et al., 2005)(Zhang et al., 2019). Mennini et al. (2022) performed analysis taking both cases a possible, empirical and guided therapy. This information does not available in three studies (Tsaganos et al., 2016)(Merchant et al., 2008)(Kongnakorn et al., 2015).
Delayed initiation of antibiotic therapy has often been cited as a major risk factor for excess mortality. The infectious by multidrug-resistant organisms show the need not to delay antibiotic and to choose the right one (Chaïbi et al., 2022). Validation studies of the American guidelines suggest that empiric antibiotic therapy based on risk factors may lead to an overuse of broad-spectrum antibiotics (Dominedo, 2020).These fact supports the need and interest of empirical therapy in VAP, especially when we talk about Gram-negative bacteria, including carbapenem-resistant Enterobacterales, multidrug-resistant Pseudomonas aeruginosa and Acinetobacter species, which ones has complicated the selection of appropriate antimicrobials and contributed to treatment failure (Miwa et al., 2022).
A total of 8 articles analyzed antibiotics for Gram-negative infection (Mcgarry et al., 2010)(Kongnakorn et al., 2015)(Sathe et al., 2017)(Zilberberg et al., 2010)(Merchant et al., 2008)(Vlachaki et al., 2022)(Tichy et al., 2020)(Mennini et al., 2022). Four of them related comparation about Gram-positive spectrum (Del and Neipp, 2005)(Machado et al., 2005)(Tsaganos et al., 2016)(Zhang et al., 2019)(Merchant et al., 2008).
Linezolid was compared with vancomycin on two occasions (Del and Neipp, 2005)(Machado et al., 2005). To Zhang et al. (2019) Vancomycin was a control, compared with Teicoplanin, Linezolid and others antibiotics. Clarithromycin was used as adjuvant therapy in VAP and sepse to reduce hospital costs in 45 days. Ceftalozane-tazobactam and Ceftazidima-avibactam were analyzed and compared with Meropenem (Mennini et al., 2022) (Tichy et al., 2020).Association between Meropenem-varbobactam was tested comparing the costs to Vlachaki et al. (2022). New antibiotic, ceftriaxone + sulbactam + disodiumedetate (AAEE), had pharmacoeconomic performance tested in use associated with meropenem to Acinetobacter baumannii VAP. The most common therapy compared was Doripinem versus imipinem, in four studies (Mcgarry et al., 2010)(Kongnakorn et al., 2015)(Zilberberg et al., 2010). The summary of findings is providing in Table 4.
For MRSA pneumonia, the efficacy of linezolid and vancomycin were better than other antibiotics, the ICER per LY saved for using linezolid over vancomycin was US$2185, showed these two antibiotics might be recommended for treatment of MRSA pneumonia, based on results of Zhang et al. (2019). Linezolid was more cost-effective than vancomycin with an additional cost per QALY/LYS gained below the acceptable threshold in Spain of € 30,000 (Del and Neipp, 2005). This finding was also related to Machado et al. (2005), despite the higher price per unit, linezolid was more cost-effective.
The adjuvant treatment with Clarithromycin showed reduction in hospital costs in 45 days to Tsaganos et al. (2016). The mean cumulative costs on day 25 in the placebo group and in the clarithromycin group were €14,701.10 and €13,100.50. On day 45 were €26,249.50 and €19,303.10 per patient staying alive. The authors suggest in conclusion this result was associated with the savings from drugs other than antimicrobials.
About adjuvant treatment with ceftriaxone + sulbactam + disodiumedetate (AAEE) in cases of VAP caused to MDR A. baumannii (Sathe et al., 2017), treatment shortened the duration of carbapenem therapy, hence reducing the cost. The mean treatment durations were 12.4 ± 2.1, 13.2 ± 2.4 and 14.3 ± 2.1 days for AAE, meropenem + colistin and AAE + colistin treatment groups. In AAE susceptible patients, the mean treatment duration and cost could be reduced by 23–24% and 43–53% if used empirically. When considered resistant patients the cost could be reduced by 21% and 26% if used AAE + colistin regime is used empirically instead of meropenem followed by AAE + colistin.
Doripinem was more economic in all estimates (Mcgarry et al., 2010)(Kongnakorn et al., 2015)(Merchant et al., 2008)(Zilberberg et al., 2010). Zilberberg et al. (2010) cite attributable hospital costs of $12,000 per case, remains impressively cost-effective at under $6,000 per QALY, ranging from cost savings of $7,000 to a maximum cost of $14,000 per QALY.
Patients treated with Ceftalozane-tazobactam showed results cost-effectiveness in both scenario of treatment, like empirical therapy CER: €1,913/QALY from the NHS perspective and €2,203/QALY from the social perspective and in confirmed scenario ICER: €6,163/QALY from the SNS perspective and €6,597/QALY from the NHS perspective Social, as demonstrated by Mennini et al. (2022).
Tichy et al. (2022) found similar result when compared meropenem with ceftazidime-avibactam, ceftazidime-avibactam provided better health outcomes, shorter times in-hospital, and thus higher QALYs per patient, at an acceptable cost, support the use of CAZ-AVI as an alternative treatment in patients with HAP/VAP.
The analyze of association meropenem-varbobactam demonstrated incremental cost of € 5.165, in 5 years. The incremental QUALYs was 0.366, resulting in an ICER of € 14,133 per QUALY gained compered to carbapenem, aminoglycoside, polymyxin B, colistin, tigecycline or ceftazidime-avibactam. Considered 28 days the results gained the same impact. Second Vlachaki et al. (2022) meropenem-varbobactam could be considered a cost-effective treatment option compared with therapy used as standard in patients with CRE-KPC infections.
Table 4
Summary of the antibiotic and find assess.
Reference | Model type | Comparation | Find |
Mcgarry, et al | Cost-minimization | Doripinem versus imipinem | Doripinem is superior |
Kongnakorn, et al | Cost-minimization | Doripinem versus imipinem | Doripinem is superior |
Sathe, et al | Overall costs | New antibiotic (AAEE), adjuvant tratment associated with carbapenem | Treatment with AAE decreases treatment duration and cost |
Zilberberg, et al | Cost-effectiveness | Doripinem versus imipinem | Doripinem is superior. Hospital costs of $12,000 per case, remains impressively cost-effective at under $6,000 per QALY, ranging from cost savings of $7,000 to a maximum cost of $14,000 per QALY, justifying DOR as an empiric choice when PA is suspected |
Merchant, et al | Overall costs | Doripinem versus imipinem | Doripinem is superior |
Grau, et al | Cost-effectiveness | Linezolid versus vancomicina | Linezolid was more cost-effective than vancomycin in the treatment of VAP in Spain, with an additional cost per QALY/LYS gained below the acceptable threshold in Spain of € 30,000 for new therapies |
Vlachaki, et al | Cost-effectiveness | Meropenem + varbobactam versus BAT (carbapenem, aminoglycoside, polymyxin B, colistin, tigecycline or ceftazidime-avibactam) | In 5 years, the incremental costs of Vaborem were £5165 and the incremental QALYs was 0.366, resulting in an ICER of £14,113 per QALY gained compared to BAT. In 28 days too. Vaborem can be considered a cost-effective treatment option compared to BAT for adult patients with CRE-KPC infections in the UK. |
Machado, et al | Cost-effectiveness | Linezolid versus vancomycin | Despite the higher price per unit, linezolid was more cost-effective than vancomycin |
Tichy, et al | Cost-effectiveness | Ceftazidime + avibactam versus meropenem | When compared with meropenem, it provided better health outcomes (ie, clinical cures), shorter times in-hospital, and thus higher QALYs per patient, at an acceptable cost. CAZ-AVI was demonstrated as a cost-effective alternative to meropenem in HAP/VAP in Italy, based on the willingness-to-pay threshold, resistance levels, and efficacy data. These findings further support the use of CAZ-AVI as an alternative treatment in patients with HAP/VAP |
Mennini, et al | Cost-effectiveness | Ceftalozone + tazobactam versus meropenem (empiric and guided) | Cost-effective alternative both in the empirical treatment setting (ICER: €1,913/QALY from an NHS perspective and €2,203/QALY from a social perspective) and in the confirmed treatment setting (ICER: €6,163/QALY from an NHS perspective and € 6,597 / QALY in the social perspective) in both analysis perspectives |
Tsaganos, et al | Cost-benefit | Clarithromycin, adjuvant treatment | IV clarithromycin as an adjunct in sepsis and VAP reduced hospital costs assessed at 45 days |
Zhang, et al | cost-effectiveness | Antibiotics used in complicated infections x vancomycin as a control | For MRSA pneumonia, the efficacy of linezolid and vanco0mycin were better than other antibiotics |
Economic evaluations alongside trials have any limitations, as short time horizons, meaning that ultimate costs and benefits are not fully captured, inability to consider all relevant options and some limited generalizability (Wagner et al., 2020).
Our study has several limitations, as different countries involved, consequently, many perspectives, likely payers and different costs assumed to be more profitable in a health context. Long-term costs are not measure in our findings. The Quality of Health Economic Studies were not accessed.