Mexico has a divided health system where public health care is furthered divided into people with and without social insurance. Uninsured Mexicans’ health expenditure can be out of pocket or covered by the Seguro Popular (SP). The SP consists of well defined benefit packages and medicines that provides coverage only for certain services such as preventive medicine and primary care, national vaccination program, HIV infection care, and catastrophic medical expenditures associated with certain conditions such as critical care, neonatal care, congenital disorders, certain malignancies, hepatitis C infection, stem-cell, and solid-organ transplants (with different limitations regarding age).10

Four hospitals located in different states around the country that provide medical attention to patients without social security were included.

In all the hospitals, patients’ socioeconomic level (SEL) was evaluated upon admission and many of the expenses are charged according to the socioeconomic level.

The National Cancer Institute (INCan) of Mexico City is a 150-bed tertiary care center focused on the treatment of adult oncological patients who are referred mostly from the central part of the country for medical attention. For diseases not included in the SP, hospital expenses are paid according to the SEL assigned, but medication costs are absorbed 100% by the patient, regardless of SEL.

The Regional High Specialty Hospital of Oaxaca (HRAEO) is a 100-bed tertiary care center located in Oaxaca City, Southwest Mexico. It provides service only to adults, without gynecology and obstetrics department. The patients pay the hospital length of stay (LOS) and procedures according to the assigned SEL, the rest is absorbed by the hospital. The costs of drugs included in the basic hospital chart, are 100% absorbed by the hospital.

The University Hospital “Dr. José Eleuterio González” (HUNL) is a 667-bed general hospital in Nuevo León, Northeast Mexico. The hospital is part of the Autonomous University of Nuevo León, providing total hospital attention to around 50% of the patients without medical insurance in the entire state. Patients treated in the hospital are predominantly from Nuevo León and surrounding Northeast states. All expenses are paid according to the SEL.

The Regional High Specialty Hospital of El Bajío (HRAEB), in León Guanajuato, North Central Mexico, is a 184-bed tertiary care center, specialized in adult oncological patients as well as hematopoietic, renal and hepatic transplantation, and highly specialized surgery. Patients are referred from the middle and east part of the country. Almost 100% of them have SP, and more than half of them are covered by the catastrophic funding; total expenses of hospitalization and medication costs are absorbed 100% by federal government. In all the hospitals, patients with diseases not covered by SP or catastrophic diseases pay hospital expenses according to the SEL, and purchase medications not provided by the hospital at full cost.

This was a retrospective and cross-sectional study approved by each local institutional review board (IRB) with the following register numbers: HUNL IF17-00001, HRAEB CEI-08-17, INCAN CI/381/17 and HRAEO-CIC 003-17.

All IFIs occurring during 2016 were identified using the databases of each participating hospital. Clinical data was recovered from medical records. This included demographic information, previous comorbidities, type of IFI, antifungal type used, duration, and indication, as well as, outcomes such as development of acute kidney injury (AKI), liver damage (LD), ICU admission, hospital LOS, and mortality.

The following definitions were used:

• •

  AKI was defined as a ≥two-fold increase in serum creatinine compared to baseline level and/or presence of hypokalemia, or hypomagnesaemia during antifungal therapy.

• •

  LD was defined as a ≥two-fold increase in AST/ALT or total bilirubin compared to baseline level during antifungal therapy.

• •

  IFIs were classified according EORTC/MSG criteria.11 Possible IFIs were categorized as possible IFIs in neutropenic fever (NF) individuals or possible IFIs in non-NF individuals.

Data regarding costs of antifungal drugs, hospital, and ICU stay were provided by each participant hospital for each patient. In Mexico, hospitals buy certain authorized antifungal drugs at pre-established prices during public tenders (Supplementary material, Table S1). Antifungals such as liposomal amphotericin B (LAmB) and posaconazole are usually not part of the hospitals’ stock due to their high prices. Hence, LAmB should usually be purchased by the patients at drug stores or via a non-governmental organizations (NGOs). Estimate expenses for antifungals were calculated based on the prices paid by each hospital (Table S1).

All identified IFI-cases without clinical data available were not included for either clinical nor cost analysis, however they were included for incidence rates calculation.

Only proven and probable IFIs were used to estimate incidence rates. These were calculated by type and an overall incidence rate for all types of IFIs. The incidence rates were estimated using the number of discharges and patients-days during 2016.

An estimation of the total expenditure in antifungal therapy during one year in Mexico was performed based on the number of hospital discharges during 2015 (last information available). This information, was gathered from the Organization for Economic Cooperation and Development (OECD) webpage, in December 2017.12 The rest of the estimations was calculated with data obtained in this study, such as incidence rate for proven/probable IFIs, mean cost of antifungal therapy, mean days of treatment. Costs were adjusted by the expected proportion of AKI and increasing cost of antifungal therapy due to AKI.

This study data is presented as proportion, median and interquartile range (IQR) depending on the type of data. For categorical data comparisons, Pearson's chi-square or Fisher's exact test were used as appropriate and for ordinal and quantitative variables, Mann–Whitney or Kruskal–Wallis tests were used as indicated. After univariate analysis, those variables with a p-value ≤0.05 were included in a binary logistic regression multivariate analysis of factors associated with death, AKI and LD, and adjusted for sex and age. Statistical analysis and plot construction were performed using SPSS 24 (IBM Chicago, US) and Prism 7 (Graph Pad, California, US) software. Mean and 95% confidence intervals were estimated as required for the yearly expenditure analysis and incidence rates.

A total of 238 cases were included in the analysis. Most of the cases had a hematological malignancy (34%), followed by a solid tumor (16%), diabetes mellitus (10%) and HIV infection (8%) (Table 1). The main indication for antifungal therapy in this sample was empiric in 59% (140/238) of the cases, most of them to treat possible IFIs in non-neutropenic individuals (101/140, 72%); in the remaining 41% the indication was to treat proven or probable IFI (Table 1). Ninety-four percent of the individuals with a NF-possible IFI (37/39) had a hematological malignancy and/or solid tumor. In contrast, only 33% of the individuals with a non-NF possible IFI had either these comorbidities (Table S2).

AKI was diagnosed in 16% (39/235) of the cases during antifungal therapy, whereas LD was identified in 25% (58/235), ICU stay was required in 35% (83/238), and the mortality rate was 23% (54/229 cases) (Table 1). No difference in mortality was seen by type of proven/probable IFI (Figure S1). However, compared with possible IFIs (NF and non-NF) the mortality rate was higher for proven/probable IFIs as group (29/91, 32% vs 25/138, 18%, p=0.025) (Table S2).

There were 98 cases of proven/probable and 140 possible IFIs, 39 in neutropenic and 101 in non-neutropenic individuals. The proven/probable IFIs were categorized into four groups: invasive candidiasis (IC), invasive aspergillosis (IA), endemic IFIs (coccidioidomycosis and histoplasmosis), and other IFIs (cryptococcosis, mucormycosis, fusariosis, non-specified IFIs). IC was the most frequent IFI overall (32/98, 33%), followed by IA (22/98, 23%), and histoplasmosis (19%) (Fig. 1A). Upon patients’ subgroups analysis, aspergillosis, was the most frequent proven/probable IFI among individuals with a hematological malignancy (18/40, 45%), and endemic IFIs (histoplasmosis and coccidioidomycosis) were more frequently seen in the HIV subpopulation (11/19, 58%) (Fig. 1B).

Fig. 1.

(A) Proven/probable infections identified during 2016 in four Mexican hospitals. Ninety-eight proven/probable IFIs were identified. Unspecified IFIs refers to diagnosis by histopathology without etiological identification. (B) Distribution of IFIs by clinical context. The frequency of the proven/probable IFIs variated depending on the comorbidity. Other IFIs includes: cryptoccocosis, mucormycosis, cladosporidosis, fusariosis, pneumocystosis, and unspecified IFIs.

(0.2MB).

The overall, IFI incidence was 4.8 cases (95% CI=0.72–8.92) per 1000 discharges and 0.7 cases (95% CI=0.03–1.16) per 1000 patients-days. IC had the highest incidence rates (1.93 cases per 1000 discharges, 95% CI=−1.01 to 2.84), followed by endemic IFIs (1.53 per 1000 discharges 95% CI=−3.36 to 6.40) and IA (1.25 per 1000 discharges, 95% CI=−0.90 to 3.45) (Table 2).

Based on these incidence rates, and according to the total number of hospital discharges (6,269,155) reported by the OCDE in Mexico during 2015, the number of new cases of IC per year would be 12,103 (95% CI=3621–20,584), for IA 7851 (95% CI=−5918 to 21,621), for endemic mycosis 9598 (95% CI=−21,034 to 40,265). Taking into account the total Mexican population (127.5 million), the incidence rate of these infections would be approximately 9.49 (95% CI=2.84–16.00), 6.15 (95% CI=−4.64 to 16.96), 7.52 (95% CI=−16.49 to 31.58) per 100,000 inhabitants for IC, IA, and endemic mycosis, respectively.

No differences in AKI or LD rates were found among the type of proven/probable IFIs, not even among survivors (Table S4). Intensive care unit admission was more frequent in patients with IC (17/32, 53%), endemic IFIs (27%, 6/22), and other IFIs (18%, 4/22). For aspergillosis, ICU stay was required only in 9% of the cases (2/22), p<0.002, no differences were found when only survivors were analyzed (Table S4).

Fluconazole and echinocandins were the two most frequently indicated antifungal drug as initial therapy (Table 1). Only three out of 238 individuals did not receive the indicated antifungal therapy. Seventy-three percent of the IC cases received an echinocandin as main therapy (22/30), while for IA, voriconazole (11/22, 50%) and amphotericin B deoxycholate (6/22, 27%) were the preferred options, as for the cases with endemic and other IFIs, AmBD was the leading therapy (20/43, 46%), in these cases lipid formulation of amphotericin B was used only in 7 of 43 cases (16%).

IC required the shortest duration of antifungal therapy (median 14 days, IQR 10–18) compared with the rest of the proven/probable IFIs groups (IA=median 23 days IQR 9–59; endemic IFIs=median 15 days IQR 10–91, and other IFIs=median 34 days IQR 17–68, p=0.05) (Fig. 2A). This difference was more evident (p<0.0001) when the antifungal therapy period was analyzed only among survivors. All in all, groups with longer antifungal therapy were endemic IFIs (median 91 days, IQR 14–259), followed by other IFIs (median 53 days, IQR 24–68), and IA (median 56 days, IQR 33–74) when compared with IC (median 14 days, IQR 11–22) (Fig. 2B).

Fig. 2.

(A) Time of antifungal therapy in proven/probable fungal infections in the total population. *p value=0.046. Invasive candidiasis n=32, aspergillosis n=21, endemic (histoplasmosis/coccidioidomycosis) n=21, other IFI n=21. (B) Time of antifungal therapy in proven/probable fungal infections among survivors. p value=0.002. *p=0.006 invasive candidiasis vs aspergillosis, **p=0.002 invasive candidiasis vs endemic, ***p=0.006. Candidiasis n=22, aspergillosis n=12, endemic n=11, other IFI n=14. Data presented as median and IQR. Data analyzed with Kruskal–Wallis test. Post hoc analysis was done with Dunn's test.

(0.16MB).

The development of AKI and LD was associated with higher mortality rate (39% vs 9% and 44% vs 18%, respectively, p<0.0001). These adverse events, in a multivariate binary regression analysis, were associated with death, independently of ICU stay, age, and sex (Table S5). In this model, AKI development was associated with a four-fold higher probability of dying (OR=4.29, 95% CI=1.80–10.02, p=0.001), and LD was associated with double the probability (OR=2.34, 95% CI=1.13–4.85, p=0.02).

In the overall population, the antifungal drug was modified in 80/235 (34%) cases due to different causes, such as AKI (20/80, 25%), de-escalation or adjusting to diagnosis (60/80, 75%) (Table 1).

In this study, survivors (n=175) were defined as those who received a complete antifungal therapy scheme. In this subgroup, AKI was more frequent in cases of proven/probable IFIs (11/62, 18%) than in cases of NF and non-NF possible IFIs (2/31, 6% and 3/82, 4%, p=0.01) (Table S3). This pattern was similar in all groups regarding LD during the antifungal therapy (28% vs 13% vs 12%, respectively, p=0.03) (Table S3). Among survivors who developed AKI, 31% had received AmBD and 25% LAmB as first antifungal, these proportions were higher compared with 14% and 2.6%, respectively, for the individuals without AKI (p<0.0001) (Table 3). Survivors who developed AKI received longer antifungal therapy (median 18 days IQR 14–50 vs median 10 days IQR 6–17, p=0.0008), had longer hospital stay (median 27 IQR 18–40 vs median 19 days IQR 12–31, p=0.05), and higher probability of ICU admission (63% vs 27%, p=0.008) (Table 3). The group developing LD also received longer antifungal therapy (median 15 days IQR 5–17 vs median 10 IQR 5–17, p=0.001), had longer hospital stay (median 28 IQR 20–38 vs 18 IQR 12–30, p=0.003), and ICU admission (48% vs 27%, p=0.03) (Table 4).

The cost per unit of IV fluconazole and oral formulation varied between 0.4 and 1.2 USD, which represents the antifungal with the lowest cost per unit. The most expensive antifungals, per unit, were posaconazole ($558–766) and LAmB ($154–230) (Table S1).

The cost per day of antifungal per person was $11 USD (IQR 1–104). However, the cost varied depending on the type of antifungal drug used. The median costs per treated person were $7 (IQR 4–16) for fluconazole, $654 (218–1817) for voriconazole, $731 (IQR 522–1044) for caspofungin, $612 (IQR 349–1136) for anidulafungin, $96 (IQR 36–375) for AmBD, and $6840 (IQR 3040–9120) for LAmB (p<0.0001).

When the cost of antifungal therapy was analyzed by type of proven/probable IFI, among survivors, the group of other IFIs had the highest cost per treated person ($1843, IQR 567–46,543) followed by IA ($1382, IQR 645–2924), and IC (1072, IQR 528–1461), whereas endemic IFIs showed the lowest cost per treated person ($266, IQR 163–546), p=0.015 (Table S4).

Patients with AKI needed a second antifungal more frequently (30%) than those without AKI (54%) [p=0.008], and this difference was more evident in individuals who survived (30% vs 62%, p=0.01) (Table 3). The presence of AKI increased 4.3-fold the cost of antifungal therapy (median $246 IQR 9874) when compared with individuals without AKI (median $1048 IQR 729–6543) [p<0.0001] (Table 2). This increase in cost was mainly due to the cases with possible IFIs, between 7- (for cases of NF possible IFIs) and 18-fold ((for cases of non-NF possible IFIs) when all cases were analyzed (Table 5). However, the costs increased between 12- and 31-fold when the analysis was restricted to survivors, for non-NF and NF possible IFIs, respectively (Table 5).

Also, LD had a 4-fold increase in the cost of antifungal therapy (Table 3). Similarly, to AKI, the costs increased for non-NF and NF possible IFIs cases between 30- to 58-fold, and between 35- and more than 200-fold increase among survivors with non-NF and NF possible IFIs cases, respectively.

Information about the costs of the antifungal therapy and length of hospital stay provided by each hospital and case-specific allowed to determine the mean cost of a day of antifungal therapy and hospital-stay (Table 6). These values could vary, depending on the type of antifungal drug (more expensive for liposomal amphotericin B and less expensive for generic formulations of fluconazole), required doses, ICU stay, and if the patient died or survived. In order to have an approximated mean cost of the antifungal therapy, all different drug costs for all patients were taken into account, without distinction with regard to the final outcome (Table 6).

The total number of hospital discharges in Mexico was obtained from the OECD webpage.12 The last information available corresponded to 2015 for the hospital discharges by diagnostic categories. A total of 6,260,155 hospital discharges were reported by OECD during that year. Taking into account, the proven/probable IFI incidence rate estimated in this study, 4.8 per 1000 discharges (95% CI 0.7–8.9), we estimate that in one year the total number of proven/probable IFIs treated with an antifungal drug to be approximately 30,091 patients (95% CI 4396–55,785) (Table 6). However, based on our data (showed above), these numbers correspond only to 41% of the cases receiving an antifungal drug (Table 1). Hence, the total number of individuals who received an antifungal drug due to any type of IFI (proven/probable/possible) in one year would be 66,510 individuals (95% CI, 3930–130,602). This estimated burden represents an approximation to the total use of antifungal therapy in Mexico.

Taking into account the data expressed in this study, the estimated mean duration of one antifungal therapy for any type of IFI was 25 days (95% CI 17–33). The mean cost of one treatment, therefore was $1100 USD (95% CI 600–4488). However, the duration is usually longer in proven/probable IFIs (Table 6), especially for IA and endemic IFIs (>400 days in some cases) (Fig. 2). Hence, this estimate shown above could be even higher.

The total expenditure in antifungal therapy for all type of IFI, without adjusting for AKI, LD and ICU stay, would be approximately $154,636,516 USD (95% CI $4,075,571–$534,423,801 USD). After adjustment only for the proportion of expected individuals to develop AKI and the increasing in antifungal cost in such cases, the expenditure in antifungal therapy for all type of IFI would ascend to $233,435,536 USD (95% CI $6,224,993 to $773,810,330).