How complicated skin and soft tissue infections are treated in Italy: economic evaluation of inpatient intravenous antibiotic treatment in seven hospitals

Abstract

Objectives: Complicated skin and soft tissue infections (cSSTIs) are a common cause of morbidity at hospital level. This study aimed to evaluate the costs and outcomes of inpatient intravenous antibiotic therapy for the treatment of cSSTI in seven Italian hospitals.

Methods: A total of 307 patients were enrolled in a retrospective, multicentre, incidence-based, observational study. The target population consisted of hospitalised patients eligible to receive intravenous antibiotic therapy for cSSTI. Direct hospital costs were measured through a microcosting approach.

Results: Failure of initial antibiotic therapy occurred in 23% of patients. Average antibiotic treatment lasted 12.2 days; the average full cost of admission totalled €5,530. If the initial antibiotic regimen fails to eradicate the infectious organism, the length of stay extends for 7 days and costs increase by €2,850 per patient. Nevertheless, when taking into account the lower intensity of care during the last days of treatment, savings reduce costs to €671 per patient. These could be increased by €74 for each hospital day avoided because of faster antibiotic action.

Conclusions: Efforts should be made to minimise the risk of selecting wrong antibiotics and to identify the quickest antibiotic in eradicating the infection.

Keywords::

• economic evaluation
• cSSTI
• antibiotic treatment
• microcosting

Introduction

Complicated skin and soft tissue infections (cSSTIs) are a common cause of morbidity at community and hospital levels and represent 18% of nosocomial infections1,2. If we consider that nosocomial infections are responsible for 4,500–7,000 deaths per year in Italy3,4, it becomes evident that cSSTIs are a significant problem for Italian hospitals. A SSTI is classified as complicated if the infection has spread to the deeper soft tissues, if surgical intervention is necessary or if the patient has a co-morbid condition hindering treatment response (e.g. diabetes mellitus or HIV)5,6. Examples of cSSTI include cellulites, abscesses, diabetic foot infections and surgical site infections, and they generally require hospitalisation, intravenous (iv) antibiotics and/or surgery7.

Standard antibiotic treatment of complicated infections has become difficult and costly to administer owing to treatment failure associated with the rise in bacterial resistance. cSSTIs may be caused by a wide range of pathogens, with Staphylococcus aureus isolated from 40% of SSTIs in the SENTRY Antimicrobial Surveillance Program8. Meticillin-resistant S. aureus (MRSA) has become a predominant pathogen, ranging from 30 to 70% of S. aureus in most developed countries9–15. Intravenous vancomycin is the first-line therapy against MRSA16,17, but reduced susceptibility to vancomycin is also becoming an emerging problem that will have an impact in the future18. New antibiotics (e.g. quinupristin/dalfopristin, linezolid and daptomycin) are being introduced onto the market and have been proved to be at least as effective as vancomycin against MRSA19–21. However, in times of resource constraints, incremental benefits need to be assessed against higher costs. Economic evaluation can therefore play an important role in supporting the decision-making process for infection control by identifying and measuring the costs and benefits of current therapeutic strategies to manage cSSTI22–25.

The objective of the present study was to evaluate the costs and outcomes of antibiotic therapy for the treatment of cSSTIs in Italy in order to provide decision makers with relevant base data that could be used to assess the introduction of newer antibiotics.

Patients and methods

A retrospective, multicentre, incidence-based, observational study was designed. The hospital perspective was considered. Consistent with the study perspective, direct hospital costs were identified and measured. Seven hospitals were selected across the country and each hospital was asked to recruit patients retrospectively. Five hospitals were identified amongst those specialised in the treatment of infectious diseases (Luigi Sacco Hospital (Milan), IRCCS San Matteo (Pavia), San Martino University Hospital (Genoa), IRCCS Spallanzani (Rome) and Gesù and Maria Hospital (Naples)) as well as two diabetic foot wards (San Donato Hospital (Arezzo) and Abbiategrasso Hospital (Milan)).

Patients selection

The study period was from January 2004 to August 2006. The target population consisted of adult patients (≥18 years) eligible to receive antibiotic therapy for cSSTI. A primary inclusion criterion was the presence of a cSSTI that required hospitalisation and parenteral antimicrobial therapy for at least 96 hours. When appropriate, the sensitivity of an isolated bacterial strain was determined by an antibiogram. The following baseline diagnoses were included (from now on indicated as ‘clusters’):

• infected ulcer, not diabetic (e.g. ulcers associated with vascular insufficiency or decubiti);

• wound infection (e.g. surgical wounds, traumatic wounds and bites);

• infected diabetic ulcer of the lower extremity; and

• major abscess.

Patients were excluded from the study if they had minor or superficial infections (e.g. simple abscesses, impetigo and uncomplicated cellulites), perirectal abscesses, gangrene, multiple infected ulcers at distant sites, or infections of third-degree burns. Patients were also excluded if they had bacteraemia, required curative surgery (e.g. amputation) or had concomitant infections at another site (e.g. endocarditis, osteomyelitis or septic arthritis).

Data collection

Clinical and cost data were drawn from patients’ medical records and registered on individual case report forms (CRFs). More specifically, demographic (age, sex) and clinical (baseline cSSTI, co-morbidities) variables for the patients were gathered. Severity at admission was assessed through the American Association of Anesthetists (ASA) score, which groups patients into five subcategories by pre-operative physical fitness: ‘I’, a completely healthy patient; ‘II’, a patient with mild systemic disease; ‘III’, a patient with severe systemic disease that is not incapacitating; ‘IV’, a patient with incapacitating disease that is a constant threat to life; and ‘V’, a moribund patient who is not expected to live 24 hours with or without surgery26.

The CRF contained a specific part for identification of the infecting organism(s) isolated during the hospital stay, thus distinguishing between monomicrobial and polymicrobial infections.

Healthcare costs

The CRF was also designed to collect data on consumption of healthcare resources and their unit costs. In accordance with the hospital perspective, only direct treatment costs were considered. More specifically, for each single patient the following cost components were identified and measured:

• medication (i.e. antibiotics used for the treatment of cSSTI); and

• hospitalisation (hospital resources consumed during the admission because of cSSTI, excluding antibiotic consumption), including:

• ✓ disposables;

• ✓ specialist visits;

• ✓ laboratory tests;

• ✓ diagnostic tests;

• ✓ surgical procedures (e.g. revascularisation, incision and drainage, or wound debridement);

• ✓ personnel;

• ✓ equipment (depreciation);

• ✓ ordinary maintenance services;

• ✓ administrative costs (ward specific); and

• ✓ hotel costs.

Resources were considered only if correlated with cSSTI, and length of stay (LoS) was calculated as the number of days the patient received specific care for the infection while admitted in a dedicated ward (infectious diseases or diabetic foot wards). Resources devoted to the management of co-morbid conditions themselves were excluded from the analysis.

A bottom-up microcosting approach was used to measure consumption of antibiotics. This is the most precise approach as it goes through the patient and collects consumption data for each single component27–30 (i.e. the antibiotic(s) used to treat cSSTI). Drug costs for antibiotics were based on official ex-factory prices per unit in Italy31.

Resource consumption during hospitalisation was assessed using a top-down gross costing approach (i.e. costs were averaged out at hospital ward level). More precisely, cost components for each single recruited hospital ward were retrieved from the hospital's Accounting Department in order to identify all types of resources consumed within hospital admissions (i.e. variable, fixed and semi-fixed resources). The cost for antibiotics was taken out and replaced with the cost calculated through the microcosting approach as described above.

In addition, cost components were split into two categories, avoidable and non-avoidable costs, to adjust also for the intensity of care across the hospitalisation32. This stems from the consideration that there are costs that the hospital bears anyway (non-avoidable costs) either because they serve to make patient diagnoses (i.e. specialist visits, diagnostic tests; mostly performed during the early days of admission) or they relate to the normal functioning of the ward (i.e. depreciation and maintenance of capital equipment). Conversely, avoidable costs are those costs (either fixed or variable) that could actually be saved in response to a reduction in LoS (e.g. for treatment success), assuming that scarce resources could be reallocated to other uses. Through these accounting exercises, it was possible to calculate for each single patient: (i) the cost for antibiotics; (ii) the full (fixed and variable) cost of admission; (iii) the full avoidable cost for admission; and (iv) the variable avoidable cost for admission. Variable costs are important because they highlight the real savings hospitals can achieve from reducing LoS compared with full costs that conversely represent the potential savings for the hospital, at least in the short term.

Outcomes

Initial iv antibiotic therapy (single antibiotic or combination) was considered to lead to clinical success if it caused the resolution of signs and symptoms such that no further iv antibiotic therapy was required. Although patients were to receive only iv therapy, a change to oral medication was still considered a clinical success provided that the patient had received iv therapy for at least 4 days, there had been clear clinical improvement and the infecting organism was susceptible to an available oral therapy.

Initial iv antibiotic therapy was considered to be a failure if at any point during the hospitalisation the patient did not respond to it and required either another iv antibiotic therapy (if received) or a surgical procedure (other than revascularisation or minor debridement), or if they developed new signs and symptoms relevant to the pre-treatment infection site.

Statistical analysis

Continuous variables were analysed using descriptive statistics, one-way analysis of variance (ANOVA) or Student's t-test, as appropriate. Categorical variables were analysed using the χ2 test. As is frequently the case with medical cost data and LoS, the distribution of the treatment cost and LoS is positively skewed33. Values were therefore logarithmically transformed to achieve a more normal distribution and to permit the use of standard parametric statistical tests. Multiple variables were identified in order to perform multiple regression analysis with the objective of identifying the most relevant cost predictors: age; sex; number of concomitant pathologies; clinical success/failure; ASA score; single or multiple organism infection; infection clusters; Gram-positive/-negative; meticillin-susceptible S. aureus (MSSA)/MRSA; initial antibiotic therapy (single or combination); and enrolling centre. In the final model, using a stepwise backward procedure, associations were considered significant if p<0.05. Analyses were carried out with STATA version 9.0 (Stata Corp., College Station, TX).

Results

In total, 307 patients were enrolled in the study. Their demographic and clinical characteristics are presented in Table 1. The most frequently isolated organism was S. aureus (59%), comprising 16% MRSA, 40% MSSA and 2% unspecified S. aureus (Table 2).

Table 1.  Sample characteristics (n=307).

Characteristic	n (%)*
Age (years)
Mean (sd)	57 ± 17
Median (range)	61 (18–91)
≥65 years (n)	129 (42)
Gender
Female	103 (34)
Male	204 (66)
Baseline diagnosis
Infected ulcer, not diabetic	30 (10)
Wound infection	56 (18)
Infected diabetic ulcer	124 (40)
Major abscess	97 (32)
No. of co-morbidities
None	59 (19)
One	78 (25)
Two	87 (28)
Three	63 (21)
Four	20 (7)
Co-morbidities
Diabetes	139 (45)
Cardiopathy	84 (27)
Hypertension	68 (22)
Immunocompromised	54 (18)
Peripheral vascular disease	41 (13)
ASA score
I	44 (14)
II	156 (51)
III	94 (31)
IV	13 (4)
V	0

sd, standard deviation; ASA, American Association of Anesthetists.

* Data are n (%) unless otherwise stated.

Table 2.  Infecting organism at baseline.

Infection	n (%)
Single and multiple infections
No bacteria isolated at baseline	45 (15)
One isolated bacteria	211 (69)
Two isolated bacteria	32 (10)
Three isolated bacteria	19 (6)
Total	307 (100)
All isolated bacteria*
Staphylococcus aureus
All	180 (59)
Meticillin-susceptible S. aureus	124 (40)
Meticillin-resistant S. aureus	50 (16)
CoNS
All	24 (8)
Meticillin-susceptible CoNS	4 (1)
Meticillin-resistant CoNS	4 (1)
Pseudomonas aeruginosa	18 (6)
Enterococcus spp.	18 (6)
Streptococcus pyogenes	18 (6)
Streptococcus spp.	13 (4)
Enterococcus faecalis	11 (4)
Streptococcus agalactiae	6 (2)
Streptococcus viridans	5 (2)
Proteus mirabilis	4 (1)
Others	35 (11)

CoNS: coagulase-negative staphylococci.

* Sum exceeds 100% because of multiple infections.

With regard to the initial antibiotic therapy, 172 patients (56%) were administered a single antibiotic, whilst the remaining 135 patients (44%) received a combination of antibiotics. Five initial antibiotic regimens account for almost 50% of the observed treatments: piperacillin/tazobactam is the antibiotic used most frequently (19%), followed by amoxicillin/clavulanic acid (10%), ampicillin/sulbactam (6%), teicoplanin (6%) and ceftriaxone (6%). These data reflect the frequent initial use in clinical practice of broad-spectrum antibiotic therapies, following a prevalent empirical approach for the treatment of cSSTI34.

Failure of initial antibiotic therapy occurred in 23% of patients, with no statistically significant differences amongst the clusters (χ2 test, p=0.94).

On average, antibiotic therapy lasted 12.2 days (Table 3). The LoS is highly dependent on either success or failure of initial antibiotic therapy (Student's t-test, p<0.001), also being the strongest predictor of costs (see also Table 6). Days of therapy were on average 6 days shorter in the case of success compared with failure for any baseline diagnosis (average 11.1 vs. 16.9 days, respectively); this difference was significant for all subgroups except for those with wound infection. Moreover, if observation extends beyond days of therapy to also include the time patients remain in hospital after the conclusion of antibiotic therapy (i.e. ‘LoS after antibiotic therapy’), success of initial antibiotic therapy saves 1 additional day compared with failure: 0.3 vs. 1.1 days, respectively. Thus, in the case of success, 1 week of hospitalisation is saved for each patient (i.e. 13.4 vs. 20.3 days); this difference was significant for all subgroups.

Table 3.  Length of stay.

	Success (days)	95% CI	Failure (days)	95% CI	p-value*	Mean (days)^†	95% CI
Days of therapy	11.1	10.3–11.9	16.9	15.0–19.1	<0.001	12.2	11.4–13.0
Infected ulcer, not diabetic	10.6	8.5–12.8	18.8	13.0–27.3	0.02	12.1	9.9–14.9
Wound infection	10.9	9.4–12.7	15.0	10.0–22.6	0.09	11.6	10.0–13.5
Infected diabetic ulcer	13.1	11.7–13.8	16.5	13.9–19.6	0.04	13.8	12.6–15.2
Major abscess	9.1	8.0–10.3	17.9	14.8–21.6	<0.001	10.7	9.5–12.0
LoS after antibiotic therapy	0.3	0.2–0.4	1.1	0.7–1.7	0.003	0.4	0.3–0.6
Infected ulcer, not diabetic	0.4	0.1–1.6	0.5	0.0–9.7	0.89	0.4	0.1–1.4
Wound infection	0.3	0.1–0.8	1.0	0.6–1.8	0.24	0.4	0.2–0.9
Infected diabetic ulcer	0.6	0.3–1.1	1.4	1.0–2.0	0.15	0.8	0.5–1.2
Major abscess	0.1	0.0–0.2	1.0	0.4–2.5	0.008	0.2	0.1–0.3
Total LoS for treatment	13.4	12.5–14.3	20.3	18.2–22.7	<0.001	14.7	13.9–15.6
Infected ulcer, not diabetic	13.8	11.4–16.2	24.3	16.3–36.2	0.01	15.8	13.0–19.1
Wound infection	13.2	11.6–14.8	18.7	15.0–26.9	0.03	14.2	12.5–16.1
Infected diabetic ulcer	15.8	14.2–16.6	19.7	16.7–23.1	0.04	16.6	15.2–18.2
Major abscess	10.8	9.7–12.0	20.8	17.4–24.9	<0.001	12.6	11.3–14.0

LoS, length of stay; CI, confidence interval.

* Student's t-test.

^† Mean values are calculated by log-transforming the variable followed by antilog to correct for positive skewness.

Table 6.  Multiple regression analysis.

Dependent variable Ln (full cost of admission)	Final model (stepwise backward pr.05)
Independent variables	β-coefficient	t-value	p-value
Clinical
No. of co-morbidities	0.075	2.79	0.006
Clinical failure (vs. success)	0.472	7.18	<0.001
Initial antibiotic therapy (combination vs. single)	0.142	2.26	0.025
Infection cluster
Major abscess (vs. non-diabetic infected ulcer)	–0.223	–2.25	0.025
Wound infection (vs. non-diabetic infected ulcer)	–0.120	–1.06	0.291
Infected diabetic ulcer (vs. non-diabetic infected ulcer)	–0.020	–0.16	0.870
Enrolling centre
Hospital 1 (vs. hospital 7)	0.043	0.33	0.743
Hospital 2 (vs. hospital 7)	0.110	0.83	0.409
Hospital 3 (vs. hospital 7)	0.420	3.61	<0.001
Hospital 4 (vs. hospital 7)	0.229	1.99	0.047
Hospital 5 (vs. hospital 7)	0.327	2.21	0.028
Hospital 6 (vs. hospital 7)	–0.037	–0.28	0.781
Constant	8.264	55.21	<0.001
R2	0.284
Adjusted R2	0.254
Number of observations	307
F (12;294)	9.96
Prob. >F	<0.001

Variables removed from the initial model: age; gender; American Association of Anesthetists score; multiple organism infection; presence of meticillin-susceptible S. aureus; presence of meticillin-resistant S. aureus; Gram-positive-only infection.

The average full cost of admission per patient amounted to €5,530, with the average antibiotic cost itself being a small percentage (6.4%) (Table 4). With regard to differences amongst different types of cSSTI, the average treatment cost is significantly lower for major abscesses compared with all the other clusters (ANOVA p<0.001). Most of this difference is explained by the average shorter LoS for major abscesses compared with other types of cSSTI: 12.6 vs. 14.7 days, respectively (ANOVA p<0.001) and the lower cost for antibiotics: €274 vs. €356, respectively (ANOVA p<0.001). Amongst the remaining three clusters, no significant differences in treatment costs and LoS were observed.

Table 4.  Full cost of admission.

	Success (€)	95% CI	Failure (€)	95% CI	p-value*	Mean (€)^†	95% CI
Antibiotic cost	282	243–328	777	655–922	<0.001	356	312–405
Infected ulcer, not diabetic	180	107–301	925	506–1,691	0.00	263	162–428
Wound infection	375	273–515	642	404–1,018	0.13	417	317–549
Infected diabetic ulcer	359	283–456	826	664–1,027	<0.001	436	357–533
Major abscess	200	157–256	749	529–1,059	<0.001	274	217–345
LoS cost (antibiotic not included)	4,564	4,277–4,870	6,933	6,202–7,749	<0.001	5,020	4,731–5,328
Infected ulcer, not diabetic	4,718	3,904–5,401	8,307	5,577–12,374	0.01	5,384	4,452–6,510
Wound infection	4,515	3,974–4,330	6,391	4,450–9,179	0.03	4,834	4,252–5,496
Infected diabetic ulcer	5,382	4,846–5,676	6,714	5,713–7,890	0.04	5,667	5,180–6,201
Major abscess	3,680	3,302–4,102	7,102	5,934–8,501	<0.001	4,301	3,860–4,792
Total treatment cost	4,989	4,672–5,328	7,835	7,019–8,747	<0.001	5,530	5,206–5,874
Infected ulcer, not diabetic	5,024	4,136–6,103	9,384	6,266–14,056	0.006	5,813	4,770–7,084
Wound infection	5,046	4,420–4,961	7,153	5,032–10,167	0.04	5,404	4,743–6,157
Infected diabetic ulcer	5,922	5,335–6,273	7,598	6,462–8,933	0.02	6,277	5,736–6,869
Major abscess	3,968	3,555–4,429	8,054	6,766–9,586	<0.001	4,693	4,201–5,243

CI, confidence interval; LoS, length of stay.

* Student's t-test.

^† Mean values are calculated by log-transforming the variable followed by antilog to correct for positive skewness.

With regard to clinical assessment, average full cost of admission per patient varies from €4,989 to €7,835 in the case of success or failure, respectively, resulting in a potential average resource savings of approximately €2,850, assuming an efficient re-allocation of resources inside the hospital (i.e. efficient bed turnover, reduction of waiting lists, etc). Nevertheless, if only full avoidable costs of admission (see Table 5) are considered, the hospitals could save €2,640 on average per patient in the case of success. If only variable avoidable costs for admission are considered, average savings reduce to €671 per patient; these could be ultimately increased by €74 avoidable costs for each hospital day avoided because of faster antibiotic action (Table 5).

Table 5.  Per day full cost, and per day avoidable and non-avoidable average costs.

Avoidable cost components	Per day average (€)	%	Non-avoidable costs	Per day average (€)	%
Antibiotics	32.84	9	Diagnostic tests	12.70	3
Disposables	7.75	2	Specialist visits	7.76	2
Laboratory tests	39.60	11
Surgical procedures	8.04	2
Subtotal	88.23	24	Subtotal (variable)	20.46	5
Personnel	191.60	51
Administrative costs*	25.99	7	Equipment (depreciation)	2.26	1
Hotel costs^†	41.32	11	Ordinary maintenance services	4.41	1
Subtotal	258.91	69
Total	347.14	93	Subtotal (fixed)	6.67	2
Per day full cost	374.27	100	Total (non-avoidable)	27.13	7

* Administrative costs include gas, power and water consumption, laundry and waste management.

^† Hotel costs include food, beverages and other support non-healthcare services.

With the use of a multiple linear regression analysis, (log-transformed) full cost per patient was found, coherently with the abovementioned analyses, to be predicted by the clinical failure of initial antibiotic therapy, by the presence of co-morbidities, by the initial antibiotic regimen (combination or monotherapy), by cluster and by the enrolling centre variables (Table 6). In particular, clinical failure of initial antibiotic therapy was the strongest predictor (p<0.001) of the full cost per patient: failure of antibiotic would increase the costs by 47%. Enrolment centres were also strong predictors of cost (ANOVA p<0.01): depending on the hospital where patients were treated cost differentials ranged between 22 and 42% per case. A different cost predictor was the initial antibiotic regimen (p<0.05): combination of antibiotics as initial therapy resulted in increased costs of 14%. Amongst the cluster variable, confirming the validity of the abovementioned analyses, the presence of major abscesses was associated with a strong decrease in costs (-22%). Age, ASA score, gender, multiple organism infection and the presence of MSSA or MRSA per se were not significant additional predictors of cost.

Discussion

The present study represents the first multicentre, observational study aimed at evaluating the current inpatient clinical practice used in Italian hospitals to treat cSSTIs. Data on antibiotic therapies have been gathered in full detail to show the types and quantity of each antibiotic administered to cSSTI patients as well as the use of other healthcare resources that have been used inside each enrolling hospital. This approach has allowed investigators to calculate different configurations of costs (i.e. variable, full and avoidable costs) as well as to explain cost variability between patient groups and among patient characteristics. The results clearly show that patients’ stay highly depends on success or failure of the initial antibiotic treatment, which is the strongest cost predictor in a complete regression model, adjusted for all the other relevant variables (patient's prior condition, enrolling centre, age, sex, etc): this is true of all types of cSSTI. LoS extends to almost 1 additional week if antibiotics fail to treat the infection, as happens in 23% of patients, with no statistically significant differences amongst the clusters in the analysis. Amongst different types of cSSTI, major abscess represents a cluster per se, being significantly cheaper on average to treat compared with all the other clusters, whilst all the others would cost the same amount of money to the hospital both in the case of success and failure of the initial antibiotic regimen. The study also highlighted significant differences in resource consumption amongst different hospitals for the same pathologies, which could be a measure of the efficiency of Italian hospital care supply: the more efficient they are the lower the costs associated with similar treatments. Moreover, the presence of MRSA, MSSA or multiple organism infections did not significantly affect healthcare costs in the analysed institutions, being at the time of the study efficiently and effectively treated with standard antibiotic regimens.

Taking into account that the Italian NHS adopted a prospective reimbursement system (diagnosis-related group system) since 1995, these considerations unveil potential relevant savings for the hospital that could save up to €2,850 per patient had the percentage of antibiotic failures been reduced, at the same reimbursement rate. In times of resource constraints and cost containment efforts, these are important data and remain relevant even in the most conservative scenario; that is when savings are limited to the very short term and ‘avoidable’ costs scenario only. The results also suggests that, among success cases, the shorter the LoS the bigger the savings for the hospital. This consideration leads to another point: the rapidity of action of antibiotics.

Few recent international economic studies have been published on these themes, focusing in particular on diabetic lower extremity ulcers and healthcare-acquired infections35–38. For the former, the results show that costs attributable to the diabetic infection ranges between €11,300 and €14,100 (corrected for inflation, 2006 exchange rate) in inpatient settings of care, which is much higher when compared with these study results (€6,277, year 2006 data)35,36. As for the latter, amongst healthcare aquired infections, surgical site infections would cost on average €24,000 (corrected for inflation, 2006 exchange rate), with a very high variability (standard deviation ± €37,400), which is much higher than the infected wounds cluster average treatment cost registered in the present study (€5,404, year 2006 data), assuming the reported mean as a representative measure of the distribution38. Nevertheless, these results are difficult to extrapolate and to compare internationally because they are strongly related to country-specific treatment practices, price levels and reimbursement systems36,39. In a recent Italian study40, the cost of central venous catheter bloodstream infection in the hospital setting was calculated and made equivalent to approximately €9,000 (year 2006 data). These results, even though they refer to a different type of hospital infection, do share the same context as in the present study (i.e. Italy) as well as the same costing methodology (i.e. bottom-up microcosting), making the comparison more meaningful than with the previous international studies.

However, this study has some limitations. The results of this study cannot be seen as representative of the clinical practice used in all Italian hospitals since those that have been recruited in the present study do have a dedicated unit/department for infectious diseases or diabetic foot units. Therefore, whilst the results can be considered fairly representative of clinical practice in highly specialised hospitals, they instead need to be considered cautiously when the hospital setting is different from that represented in this study (i.e. hospitals without a dedicated unit/department for infectious diseases where patients are treated in ‘general medicine’ wards). However, it must also be noticed that the types of infections that have been considered in the present study are normally treated in highly specialised hospitals (i.e. those with a dedicated unit/department for either infectious diseases or diabetic foot units). The authors are therefore fairly confident that the clinical practice and the relevant costs that have been estimated here can reflect the true patient pathway in Italy with regard to the treatment of cSSTI. These results would, however, need to be further confirmed in a larger multicentre study that should encompass a wider variety of hospital units across the entire country to become fully representative of the Italian practice, also taking into account outpatient parenteral therapy regimens. More specifically, the study does not account for the cost of care attributable to the infection that could extend into the outpatient setting, focusing exclusively on the inpatient setting, which could underestimate the ultimate resource consumption for cSSTIs.

Nevertheless, it must be noted that the type of approach used in this study (i.e. microcosting) can easily allow decision makers to use the present results, in terms of consumption of resources, as a baseline to compare clinical practice in hospitals/units different from those enrolled for this study (e.g. general medicine) in order to check what the difference is in the performance of other units in cSSTI treatment (e.g. rate of failure/success, types of antibiotics used as first line).

Conclusions

cSSTIs are becoming a greater issue for hospitals because of the increasing prevalence of pathogens resistant to current antibiotics. This implies that physicians switch from one antibiotic to another in the search for the most effective and that patients stay longer in hospitals thus being exposed to higher risks of nosocomial infections. This also has implications for the hospital cost function, which is highly sensitive to whether antibiotics fail or succeed in treating cSSTI patients. The correct choice of first-line antibiotics can save up to €2,850 per patient in the case of failure avoided, of which €671 are avoidable costs, that is they can actually be saved in the very short term. These could be ultimately increased by €74 for each hospital day avoided because of faster antibiotic action. Results therefore indicate that efforts must be made to reduce the failure rate of the initial antibiotic option as much as possible and to choose the quickest antibiotics, whilst keeping safety and effectiveness as high as those currently available.

Acknowledgements

Declaration of interest: This research project has been sponsored by Chiron Biopharmaceuticals/Novartis, Italy. The authors would like to thank M Fabbi, S Leone, L Minoli, S Noviello, N Petrosillo, E Righi and F Zanini for their contribution in patient selection, as well as the referees for the valuable comments made to the manuscript.