1. Background
Invasive fungal infections (IFI) are major complications of severe neutropenia, especially in patients with hematologic malignancies (1). The frequency and severity of IFI has increased steadily, over the last decades, mainly due to host defense impairment, secondary to intensive cytotoxic chemotherapies, hematopoietic stem cell transplantation (HSCT), use of corticosteroids and new immunosuppressive agents (2-4). Despite progress in IFI diagnosis and antifungal therapy performance, mortality rates in patients with IFI remain high, ranging from 20% to 57% in patients with candidemia (5, 6) and reaching around 90% in patients with invasive mold infections (7, 8). The options of antifungal effective agents are limited and often results in high healthcare costs (1).
Fungal infections may be acquired from an endogenous or external source, such as the hands, water, food and air (9). The implementation of protective measures is essential to reduce the IFI incidence, among high-risk patients, mostly caused by Candida and Aspergillus species (3, 10). Several organizations, such as the Healthcare Infection Control Practices Advisory Committee and the center for disease control and prevention (CDC) developed guidelines to provide infection control recommendations for hospitals, including a set of prevention measures, termed protective environment (PE), directed to HSCT patients (11, 12). The PE refers to isolation practices, designed to decrease the risk of exposure to environmental fungal agents, also including standard and transmission-based precautions (12).
Because inhalation seems to be the main route of mold infection, efficiently controlled ventilation systems remain a major protective strategy (13), although their efficacy was demonstrated only in non-randomized studies (14, 15). A meta-analysis, including randomized and non-randomized studies (16), using high efficiency particulate air (HEPA) filters, with or without other control measures, showed only a modest effect on invasive mold infection that did not reach statistical significance.
2. Objectives
The objective of our investigation was to assess the impact of PE implementation on IFI incidence, consumption and cost of antifungal drugs among cancer patients with chemotherapy-induced neutropenia (CIN), in a general, tertiary teaching hospital in Southern Brazil.
3. Patients and Methods
We conducted a quasi-experimental study to evaluate an intervention in a hospital ward that consisted in simultaneous preventive measures, involving air-quality technologies installation, including HEPA filters, positive pressure maintenance and infection control routines. Those measures were implemented according to international recommendations, for PE wards (11, 12). Before the intervention, cancer patients were admitted in hospital wards, without special infection control measures. Only allogeneic HSCT patients were allocated in a special ward, with single rooms, restricted access to visitors and extensive hand disinfection practices. The preventive measures did not include air-quality technologies or positive pressure. Because of the high rates of infection and mortality (19), our institution developed a project to implement a PE ward to admit high-risk neutropenic patients.
The air entering the PE ward passes through a set of filters that remove particles ≥ 0.3 μm in diameter, with 99.97% efficiency, adjusted to 34 air exchanges per hour. Air pressures, air-supply ducts and filters were checked weekly. The ward comprises 25 beds, allocated in 15 rooms, each one equipped with HEPA filters and hand hygiene facilities (hand operated sinks). The air pressure within each room is maintained positive in comparison to the hallway. To maximize appropriate pressure relationships and HEPA filtration, the doors to individual rooms were kept closed, at all times. The ward has restricted access to visitors. Additional protective conditions are the presence of a monitored anteroom, with hand hygiene facilities, closed by a double door, to maintain positive pressure in the ward and to prevent the access of non-authorized people. Extensive hand disinfection is part of the staff routines and was introduced through educational sessions. Allogeneic HSCT patients remain alone in a room and neutropenic patients that require going outside the unit for diagnostic procedures are required to wear N95 masks. During the renovation, the ward was separated from the other hospital wards, by physical barriers, to prevent dust dissemination.
The renovation started on December 2005 and the ward has been available for patient admission since May 21, 2007. The PE ward admits high-risk neutropenic patients, submitted to high dose chemotherapy, or HSTC. All neutropenic patients [leukocyte counts up to 1.000/mm3 or absolute neutrophil counts (ANC) up to 500/mm3] admitted on PE until September 2008 were identified by prospective surveillance and were included in the intervention group. Neutropenic patients, admitted to any hospital ward from January to December 2006, were identified in the institution’s patient record system and were included in the historical control group. Patients had their clinical records reviewed, until discharge or death. Subsequent admissions of the same patient were included as separate episodes. Exclusion criteria were age less than 18 years, HIV positivity, neutropenic episodes, secondary to infection, and presence of febrile neutropenia, at the time of admission.
Data regarding the underlying disease, length and severity of neutropenic episodes, central venous catheter (CVC) use, antifungal prophylaxis use until fever, corticosteroid use, graft-versus-host disease (GVHD), IFI, antifungal use, length of hospital stay and mortality were recorded for all included patients.
Severe neutropenia was defined by ANC up to 100/mm3. Neutropenia recovery was defined as the first day on which leukocyte counts and ANC exceeded 1.000/mm3 and 500/mm3, respectively, for 2 consecutive days. To identify possible differences in risk characteristics between groups, patients were classified in four risk categories: 1- autologous HSTC in the actual hospital admission; 2- allogeneic HSCT in the actual hospital admission; 3- acute myeloid leukemia (AML); 4- other diseases.
The primary endpoint of the study was IFI incidence during hospitalization, defined according to the consensus criteria, proposed by the European Organization for Research and Treatment of Cancer (2008) and the National Institute of Allergy and Infectious Diseases (NIAID) Mycosis Study Group (20). Briefly, proven disease requires histopathologic or microbiologic documentation of disease, from biopsied tissues. Probable infection is considered if the fungus is identified from cultures of bronchoalveolar lavage (BAL) fluid or sputum, when consistent signs and symptoms are present, whereas possible infection requires at least one host factor criterion and one microbiologic or clinical criteria from abnormal site, consistent with infection. Invasive mold infection was considered as probable nosocomial, if the first clinical symptoms occurred > 7 days after admission.
The secondary endpoints were consumption and cost of antifungal drugs. The same price was considered for each antifungal formulation, in the periods before and after the renovation, to avoid false comparisons due to antifungal price adjustment, along the 33 months (1.00 USD = 1.9 Brazilian Real, in May 2007).
The sample size needed to detect a reduction of 50% in the risk of IFI, with power of 80% and α error of 5%, was 247 patient hospitalizations, for each group. Data were analyzed with SPSS 16.0 (SPSS Inc., Chicago, IL, USA) and a P ≤ 0.05 was considered. Chi-square or exact tests (Fisher’s or Monte Carlo’s exact tests) were used in the comparison of categorical variables and Student’s t test or Mann-Whitney U were applied to compare continuous variables. Adjusted odds ratio (OR) for probable or proven IFI was computed in a logistic regression model, including antifungal prophylaxis. The study was approved by the Institution’s Review and Ethics Committee.
4. Results
A total of 190 and 181 hospital admissions were included in the intervention and control groups, corresponding to 128 and 124 patients, respectively. Most patients were admitted only once (71.9% in intervention and 78.2% in control group). Characteristics of the studied sample are shown in Table 1. The rate of multiple myeloma (MM) was higher in the intervention group, while acute lymphoid leukemia (ALL), non-Hodgkin’s lymphoma (NHL) and solid tumors rates were significantly lower. There was a trend to higher rates of antifungal prophylaxis use in the PE group (48.9% vs. 39.8 %; P = 0.08).
| Protected Environment (n = 190) | Control (n = 181) | P Value | |
|---|---|---|---|
| Male | 90 ± 47.4 | 85 ± 47.0 | 0.94c |
| Age, y | 46.3 ± 15.1 | 46.3 ± 15.0 | 0.97d |
| Underlying disease | 0.001e | ||
| Acute myeloid leukemia | 69 (36.3) | 54 (29.8) | |
| Chronic myeloid leukemia | 9 (4.7) | 6 (3.3) | |
| Acute lymphoid leukemiaf | 14 (7.4) | 26 (14.4) | |
| Chronic lymphoid leukemia | 6 (3.2) | 4 (2.2) | |
| Multiple myelomaf | 49 (25.8) | 20 (11.0) | |
| Hodgkin disease | 5 (2.6) | 10 (5.5) | |
| Non-Hodgkin’s lymphomaf | 21 (11.1) | 33 (18.2) | |
| Myelodysplastic syndrome | 6 (3.2) | 5 (2.8) | |
| Other hematologic malignancies | 2 (1.1) | 3 (1.7) | |
| Aplastic anemia | 6 (3.2) | 6 (3.3) | |
| Solid tumorsf | 2 (1.1) | 11 (6.1) | |
| Others | 1 (0.5) | 3 (1.7) | |
| Risk categories | 0.19c | ||
| Autologous HSCT | 50 (26.3) | 37 (20.4) | |
| Allogeneic HSCT | 19 (10.0) | 18 (9.9) | |
| Acute myeloid leukemiag | 63 (33.2) | 52 (28.7) | |
| Other diseases | 58 (30.5) | 74 (40.9) | |
| Neutropenia causes | 0.14c | ||
| Chemotherapy | 179 (94.2) | 163 (90.1) | |
| Othersh | 11 (5.8) | 18 (9.9) | |
| Length of neutropenia – median days (P25-P75) | 12 (8 - 17) | 12 (8 - 18) | 0.85i |
| Severe neutropenia | 170 (89.5) | 168 (92.8) | 0.26c |
| Central venous catheter use | 171 (90.0) | 154 (85.1) | 0.15c |
| Antifungal prophylaxis use | 93 (48.9) | 72 (39.8) | 0.08c |
| Corticosteroid use | 42 (22.1) | 49 (27.1) | 0.27c |
| GVHD | 6 (3.2) | 12 (6.6) | 0.12c |
aAbbreviations: HSCT, hematopoietic stem cell transplantation; GVHD, graft-versus-host disease.
bData are presented as mean ± SD or No. (%).
cPearson’s Chi-squared test.
dStudent’s t test.
eMonte Carlo’s exact test.
fP < 0.05 (adjusted residual).
gnot submitted to HSCT in the actual hospital admission.
hinclude underlying disease or other immunosuppressor agents.
iMann-Whitney’s test.
Invasive fungal infection occurred in 14 (7.4%) and 33 (18.2%) patient hospitalizations, in the PE and control groups, respectively, with a crude relative risk (RR) of 0.40 (95% CI: 0.22 ‒ 0.73, P = 0.002) (Table 2). Considering only the proven and probable infections, the IFI incidence was significantly reduced in the PE group (1.6% vs. 8.3%), with a crude RR = 0.19 (95% CI: 0.06 ‒ 0.65, P = 0.003). Adjusting for antifungal prophylaxis use, the OR was of 0.17 (95% CI: 0.05 ‒ 0.60, P = 0.006). There was a decreasing trend in proven and probable mold infections in the intervention group (0.5% vs. 3.3%; P = 0.06), as well as in proven yeast infections (1.1% vs. 4.4%; P = 0.057). Candida spp. was the most frequent yeast and among the seven molds, four were not identified by culture (Table 3). The single mold infection in PE group was identified through galactomannan test. Characteristics of the seven patients with mold infection are shown in Table 4. Three patients (42.8%) had diagnoses of acute leukemia, four (57.1%) underwent HSTC and the mortality rate was 57.1%. Except for only one mold, all IFI in the control group were probably hospital acquired.
Table 2. Invasive Fungal Infections in Intervention and Control Groups
aData are presented as No.(%).
bOne histoplasma capsulatum infection.
cPearson’s chi-squared test.
dProven and probable infections.
eFisher’s exact test.
Table 3. Fungal Distribution of Proven and Probable Invasive Infections
aone histoplasma capsulatum infection was identified in control group.
bGalactomannan positivity in serum (index ≥ 0.5 twice) without culture isolation of causative species.
cHistopathologic or cytopathologic evaluation, without culture isolation of causative species.
Table 4. Characteristics of Patients with Invasive Mold Infectionsa
| Group, Patient n. | Age | Diagnosis | Microbiological Criteria | Fungi Detected | Level | Outcome |
|---|---|---|---|---|---|---|
| Control | ||||||
| 1 | 31 | AML | Positive biopsy for A. niger | A. niger | Proven | Died |
| 2 | 26 | AAb | Positive biopsy for Rhizopus spp. | Rhizopus spp. | Proven | Died |
| 3 | 57 | NHLc | Positive blood culture for Fusarium spp. | Fusarium spp. | Proven | Alive |
| 4 | 38 | CML | Positive biopsy for hyphae. | NS | Proven | Alive |
| 5 | 43 | ALL | Positive biopsy for hyphae. | NS | Proven | Died |
| 6 | 37 | AMLc | Positive findings of cytopathologic evaluation for hyphae from BAL specimens. | NS | Probable | Alive |
| Protected environment | ||||||
| 1 | 60 | MDSb | Positive result for galactomannan in > 2 blood samples. | NS | Probable | Died |
aAbbreviations: AA, aplastic anemia; ALL, acute lymphoid leukemia; AML, acute myeloid leukemia; ANC, absolute neutrophil count; BAL, bronchoalveolar lavage; CML, chronic myeloid leukemia; MDS, myelodysplastic syndrome; NHL, non-hodgkin’s lymphoma; NS, non-speciated.
bAllogeneic hematopoietic stem cell transplant in previous admissions.
cAutologous hematopoietic stem cell transplant.
The median number of antifungals used per hospitalization did not differ between groups (one drug, P = 0.74). However, we recorded a lower frequency of liposomal amphotericin B (1.6% vs. 7.2%; P = 0.008) and a non-significant reduction in deoxycholate and lipid complex amphotericin B, in the intervention group (Table 5). Although not reaching statistical significance, the rates of fluconazole (both oral and intravenous formulation) use were higher in the PE group: it was used for prophylaxis in 43.0% of admissions in the PE vs. 34.8% in the control group (P = 0.10).
Table 5. Frequency and Duration of Antifungal use in Intervention and Control Groupsa
| Antifungal | Frequency of Useb | Days of Use - Median Days (P25 - P75) | ||||
|---|---|---|---|---|---|---|
| PE (n = 190) | Control (n = 181) | P Value | PE (n = 190) | Control (n = 181) | P Valuec | |
| Amphotericin B (deoxycholate) | 24 (12.6) | 32 (17.7) | 0.17d | 5.5 (3 - 10.5) | 6 (3 - 10.7) | 0.87 |
| Amphotericin B (liposomal) | 3 (1.6) | 13 (7.2) | 0.008d | 7 (3 - 10) | 7 (2.5 - 9.5) | 0.89 |
| Amphotericin B (lipid complex) | 0 | 2 (1.1) | 0.24e | - | 5.5 (4 - 7) | - |
| Caspofungin | 2 (1.1) | 1 (0.6) | 1.00e | 8.5 (3 - 14) | 7 (7 - 7) | 1.0 |
| Fluconazole (PO) | 96 (50.5) | 78 (43.1) | 0.15d | 12 (5 - 20) | 11.5 (6 - 18) | 0.81 |
| Fluconazole (IV) | 77 (40.5) | 59 (32.6) | 0.11d | 6 (1 - 1) | 5 (1 - 11) | 0.99 |
| Itraconazole | 12 (6.3) | 15 (8.3) | 0.47d | 12.5 (3.5 - 28.7) | 21 (8 - 24) | 0.96 |
| Voriconazole (PO) | 8 (4.2) | 8 (4.4) | 0.92d | 15.5 (10.2 - 22.5) | 13.5 (9.5 - 33) | 0.87 |
| Voriconazole (IV) | 3 (1.6) | 4 (2.2) | 0.72e | 2 (1 - 10) | 5 (2.5 - 6.7) | 0.59 |
aAbbreviations: PO, per Os; IV, intravenously.
bData are presented as No. (%).
cMann-Whitney’s test.
dPearson’s chi-squared test.
eFisher’s exact test.
The final cost with antifungal agents was lower after intervention (78347.37 USD vs. 154176.60 USD). However, the median direct cost of all antifungal drugs by admission did not differ between groups (14.46 USD in PE vs. 6.11 USD, under control; P = 0.55). Considering only admissions with proven, probable or possible IFI, compared to admissions without IFI in all sample, the median costs were, respectively, 1461.03 USD (P25 = 86.84; P75 = 6720.52) and 3.37 USD (P25 = 0.0; P75 = 34.42) with a P < 0.001. For proven or probable molds IFI, the figures were 3617.29 USD (P25 = 2990.17; P75 = 10653.70) and 6.26 USD (P25 = 0.0; P75 = 41.21), P < 0.001.
The median of days of hospital stay did not differ between groups (28 days; P = 0.38). On the other hand, the median length of hospital stay in admissions with proven, probable or possible IFI identified either in intervention or control groups was significantly higher than in admissions without IFI (48 days; P25 = 29 and P75 = 66 vs. 27 days; P25 = 22 and P75 = 35; P < 0.001).
5. Discussion
This paper suggests that the implementation of a PE ward in a general, tertiary and university affiliated hospital, in Southern Brazil, reduces the incidence of IFI in patients with CIN. In the control group, the incidence was higher than in a large Italian cohort study of patients with hematologic malignances, whereas it was lower after the renovation (3). After intervention, the rate of IFI was significantly reduced (7.4% vs. 18.2%, P = 0.002), even considering only proven or probable infections (1.6% vs. 8.3%, P = 0.003). This benefit remained after adjusting to antifungal prophylaxis (OR = 0.17, 95% CI: 0.05‒0.60). The GVHD is considered a main risk factor for invasive candidiasis and aspergillosis, in allogeneic HSCT patients (1, 21), and although not reaching statistical significance, it was more commonly identified in the control group (3.2% vs. 6.6%, P = 0.12). When prophylaxis and GVHD were included in the logistic regression model, GVHD was not independently associated with IFI incidence, and did not modify the PE effect. Multicolinearity with antifungal prophylaxis may explain this result. The small incidence of IFI precluded adjustment for other possible confounders, but our analysis is conservative since the intervention group is likely to be of higher risk (Table 1).
The benefits of PE implementation in our institution in reducing febrile neutropenia (P = 0.009), overall mortality (P = 0.001) and 30-day adjusted mortality (P = 0.02) have already been described (18).
The benefit of PE in reducing fungal infections was described in a prospective study conducted in Porto, Portugal. The building of new individual rooms with central HEPA filtration in the hematology unit significantly reduces the rates of proven or probable IFI (1.5% vs. 0%, P < 0.001). The consumption of antifungals was reduced and the final cost with antifungal therapy was reduced by 17.4% (15).
Evidence about infection control measures, involving ventilation and air-quality technologies, is mostly restricted to assessment of the impact of PE wards on invasive mold infections incidence, among high-risk patients undergoing HSCT or intensive chemotherapy for acute leukemia (15, 22-24). The benefit of rooms with positive pressure isolation in reducing invasive aspergillosis (IA) was registered in an adult hematological intensive care unit in a university hospital in France. The incidence of IA decreased from 13.2% before environmental modification to 1.6% after modification (14). Construction and renovation are a well-known risk factor for fungal infections, mainly caused by Aspergillus spp. (25, 26). This is plausible because these activities had been shown to dramatically increase the amount of airborne fungal spores (25). The benefit of preventive strategies in reducing spores counts, especially during renovation or construction, is known (13, 23, 24, 27). In our study, we observed a decreasing trend of invasive mold infections in the intervention group (0.5% vs. 3.3%, P = 0.057), in agreement to rates identified in studies that evaluated similar strategies (15, 22-24). Although patients in the control group were admitted in the period of PE ward construction, several others renovations continued to be made in our institution, after PE inauguration, and the intervention group also could have been exposed to dust. Although patients admitted to the PE ward may be adequately protected, they often required transfer to other areas of the hospital, either for diagnostic or therapeutic interventions, when they are exposed to the risk of acquiring invasive mold infections. Patients that required going outside the PE unit were required to wear N95 masks. The benefit of this strategy in preventing nosocomial aspergillosis during hospital construction was already demonstrated (28) and, furthermore, we could not expect that HEPA filtering facility would completely remove the risk of infection.
No patient treated in the PE ward developed proven invasive mold infection. Only one patient presented probable Invasive Aspergillosis after the renovation, and the microbiological criteria used in diagnosis were two sequential positive results for serum galactomannan. This indirect diagnostic test was implemented in our institution on June 2007, and was available only during the intervention group admissions. Although it is not considered an ideal test, gallactomannan is more sensitive than culture and allows early diagnosis of invasive aspergillosis (29). The use of this new diagnostic tool may considerably increase the incidence of probable invasive aspergillosis (5). If we have used the same diagnostic tests in both groups, no patient would be diagnosed with invasive mold infection in the intervention group. Even without this tool, six mold infections were identified in the control group. One proven IFI was caused by Aspergillusniger. Two biopsies and one cytopathologic evaluation were positive for hyphae, without species identification, although, according to pathologic evaluations, those findings were suggestive of Aspergillus spp. Although aspergillosis causes the most common invasive mycoses in highly immunosuppressed individuals, the other molds identified in control group are actually also expected in such patients (2, 3, 30). One patient in the control group developed fusariosis, diagnosed through positive blood cultures. In contrast to aspergillosis and most other invasive mold infections, fungemia is a common manifestation of disseminated fusariosis. The principal portal of entry for Fusarium spp. is the airway, followed by the skin at site of tissue breakdown, and, possibly, the mucosal membranes (31). Also, we recorded one IFI caused by Rhizopus spp., in the control group. This infection, termed zygomycosis, has risen significantly over the past decade, mainly in high-risk patients. The major mode of disease transmission for the zygomycetes is presumed to be via inhalation of spores from environmental sources (32), although ingestion and percutaneous exposure are also important in causing these infections (33). In this scenario, preventive interventions, involving environment control measures, remain a major protective factor.
The benefit of PE in reducing rates of invasive mold infections that were probably hospital-acquired was also observed (0.5% vs. 2.8% in PE and control groups, respectively; P = 0.11). It is difficult to determine the origin of those infections. Given the fact that no consensual definition of hospital-acquired invasive mold infection exists, we adopted the criteria in accordance with most studies published in the field of nosocomial aspergillosis (23, 26, 34, 35). The respective importance of each protective measures implemented in PE ward is difficult to assess. We hypothesized that the implementation of measures, involving ventilation and air-quality technologies installation, including HEPA filters, were the main causal factor. However, other characteristics that influenced individual exposure cannot be formally excluded, including the underlying disease and undergoing treatment. Most of mold IFI occurred in the high-risk HSTC patients or with acute leukemia (Table 4) in accordance to what was expected (2-4, 36).
There was a trend to lower incidence of yeast infections after the renovation, especially caused by Candida spp. These infections are usually acquired from the patient’s own gastrointestinal or mucocutaneous flora (1), several of which being of endogenous origin (37). However, yeasts may be passed to patients by healthcare professionals (10). This assumption is based on the fact that a large proportion of medical professional carry Candida spp. on their hands (10) and that hand hygiene compliance is low (38). Extensive hand disinfection was part of the staff routine in the PE ward and visitors were monitored about hand hygiene, when entering in the unit. This strategy could significantly minimize the risk of exposure to exogenous yeasts (1). Although we did not monitor hand hygiene compliance before and after the intervention, studies showed that strategies like hand hygiene promotion programs and availability of hand washing facilities are associated with higher compliance to hand hygiene practices and better outcomes (39, 40). Another possible explanation for the lower incidence of Candida infections in the PE group was the rising trend in antifungal prophylaxis with fluconazole, a preventive strategy with known impact in reducing invasive candidiasis and improving outcomes in high-risk patients (41).
During several admissions, antifungal agents were not used. There was no difference in the median number of agents used between groups, probably explained by increased antifungal prophylaxis in the PE group. The final direct cost of antifungal was reduced by approximately 50%, after the intervention. However, the median costs did not differ, maybe because few cases of proven or probable IFI occurred in the sample, which substantially increases the costs of treatment (42, 43). The rates of possible IFI were similar in the intervention and control groups (5.8% vs. 10.0%; P = 0.14). This level of probability of the diagnosis of IFI included multiple questionable cases, particularly those involving neutropenia, nonspecific pulmonary infiltrates and persistent fever, refractory to broad-spectrum antibiotics, without microbiological evidence of fungi. Most patients with this diagnosis received antifungal treatment and the cost of these cases represents an important proportion of the total cost with antifungal agents (data not shown). According to the revised definitions of IFI from the European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group and the National Institute of Allergy and Infectious Diseases Mycoses Study Group (44) the scope of category “possible” has been diminished. If these criteria were applied in clinical practice, it is probable that several patients with low chance of having invasive mycosis would not be treated. Considering the absolute reduction of proven, probable or possible IFI of 10.8%, after the intervention, an average of 10 cases of IFI could be prevented in neutropenic patients, if 100 admissions occurred in the PE ward, leading to savings of approximately 15000.00 USD (1461.03 USD minus 3.37 USD per infection). Although the median days of hospital stay did not differ between groups, admissions with proven, probable or possible IFI, in the entire sample, were associated with lengthy admission (P < 0.001). Considering that hospitalizations costs are commonly described as important in the overall healthcare costs, the impact of PE implementation in reducing IFI incidence could be associated with long-term economic benefit. Future research should evaluate the overall costs of hospital care after implementation of a PE ward, in a developing country.
The present study was limited by the quasi-experimental design and the non-contemporaneous control group. Although these types of studies can provide valuable information regarding the effectiveness of various interventions, several factors decrease the certainty of attributing improved outcomes to a specific intervention. These include difficulties in controlling for important confounding variables and the simultaneous use of multiple interventions (12). However, it was not possible to implement a randomized controlled trial because of ethical and logistical considerations. The sample size was estimated considering all IFI and was not reached. However, the studied sample was able to show a significant reduction in the primary endpoint. Because of the small incidence of events, the sample was underpowered for secondary and subgroup analysis. Since we analyzed neutropenic episodes, about 25% of patients were admitted more than once. However, exclusion of repeated patients did not modify the observed benefits. The economic analysis was limited to direct costs of antifungals and further evaluations, considering the burden of IFI, are desired.
In conclusion, this study showed that preventive measures including a PE implementation reduces IFI in neutropenic patients, especially in those with hematologic malignancies, admitted in a general, tertiary teaching hospital, in a developing country. It suggests that those strategies may overcome their costs on the long-term, by saving expenditures associated with fungal infections.
