The incidence and outcomes of hospital acquired- bloodstream infection and ventilator associated pneumonia and the impact of infectious diseases referral using data from the clinically-oriented antimicrobial resistance surveillance network for healthcare-associated infections (ACORN-HAI): An interim report Authors: Stephanie N. Co-Chang, Nicole Bernadette S. Terre, Cybele Lara R. Abad Published: 2026 ABSTRACT ObjectivesTo determine the incidence of hospital-acquired bloodstream infections (BSI), Ventilator-Associated Pneumonias (VAP) and the impact of Infectious Diseases (ID) referral on outcomes using data from the ACORN HAI surveillance network.MethodsAdults hospitalized ≥48 h in a tertiary hospital (3/2023-3/2024) were prospectively included if HA-BSI and VAP criteria were met. ID referral was defined as formal referral to, or consultation with an ID specialist. Clinical, laboratory parameters, and 28-day outcomes were collected. The incidence of HA-BSI and VAP was calculated using the number of patients with HA-BSI or VAP as numerator, and total hospitalized adult patients as denominator; incidence density per 1000 patient-days was computed using infection episodes as numerator, and total patient-days as denominator.ResultsAmong 115 patients (mean age 62.4 ± 18.1 years), 58.3% were male, and 33. 91% diabetic. HA-BSI incidence density was higher compared to VAP (1.03 vs. 46 per 1000 patient-days). Mortality (BSI 24.7%, VAP 50%, combined infection 57.1%) differed across infection types. Majority of infection episodes were referred to ID (119/143, 83.21%). Although late referrals (9/143, 6.29%), had numerically higher mortality and prolonged hospitalization, these were not statistically significant. The timing of ID referral had no statistically significant association with clinical and functional outcomes, mortality, and length of stay.ConclusionOverall HA-BSI and VAP rates were low, but still associated with high mortality and prolonged hospitalization. ID referral was not associated with better outcomes in this small study. Introduction The Clinically-Oriented Antimicrobial Resistance Surveillance Network for Hospital-Acquired Infections (ACORN-HAI) is a complementary study to the main ACORN protocol [1] which aims to incorporate and implement an efficient clinically-oriented antimicrobial resistance (AMR) surveillance system in hospitals in low- and middle-income country (LMIC) settings such as the Philippines. Surveillance definitions for ventilatory associated pneumonia (VAP) and bloodstream infections (BSI) can vary, but often used standardized, laboratory-based criteria, such as those established by the Center for Disease Control's (CDC) National Healthcare Safety Network (NHSN), to consistently identify, monitor, and report healthcare-associated infections [1]. However, these can be too rigorous, cumbersome, and difficult to follow especially in settings without well-established infection control programs or dedicated infection control practitioners. For example, VAP has now been broken down into tiered definitions including Ventilator-Associated Condition (VAC), infection-related Ventilator-Associated Complication (IVAC) and possible/probable VAP. As such, a more clinically oriented criteria may make it easier and faster to identify these infections. Hospital acquired infections (HAIs) complicate almost 43 million hospitalizations globally; up to 30% of patients requiring intensive care may be affected by HAIs [2]. LMICs are disproportionately affected, and report at least one HAI during the hospital stay of 15 out of every 100 patients in acute-care hospitals, which is 2 to 20 times higher incidence compared to high income countries [2]. Gravely affecting low-resource settings, about 4.95 and 1.27 million deaths were associated with, and attributable to bacterial AMR worldwide [3]. Among the HAIs, VAP and BSI, specifically, are associated with high economic costs and excess mortality of up to 24% and 29% respectively in Latin America, Asia and Africa among adult patients [4]. Effective infection control and antimicrobial stewardship strategies are key in addressing the threats of AMR and HAIs. Early referral to Infectious Diseases (ID) specialists via a formal consultation ensures improved adherence to quality of care standards, and can be a surrogate of stewardship activities, as expert consultation frequently leads to earlier optimization of antibiotic therapy, which is the primary goal of stewardship. This has led to better patient outcomes, including up to 40-50% reduction in in-hospital mortality by optimizing treatment from providing appropriate antibiotic choices, duration of therapy and route of delivery, thereby leading to reduced length of hospital stay, lower treatment costs, and decreased development of AMR through appropriate empiric and targeted treatment and de-escalation strategies [5]. However, referral to ID specialists can also be a marker of complexity or severity of illness, as the referral is usually made when the patients are unresponsive to therapy, already in shock, or when a multi-drug resistant pathogen is identified. This study used a clinically-oriented AMR surveillance system of patients to describe the incidence of HA-BSI and VAP, and explore the association between ID referral and clinical outcomes. Methodology Study setting The Medical City (TMC) is a 526 tertiary level private hospital located within Metro Manila. Its Intensive Care Unit (ICU) has a total of 35 beds, divided into medical, surgical, cardiac, and stroke units. Patients are admitted in the ICU if they are assessed to be in need of continuous or specialized monitoring; other indications may include patients who are in, or in danger of, respiratory failure; patients with unstable cardiac status who require vasopressors; sudden organ failure; neurologic emergencies, severe infections, or major surgery requiring post operative care. The decision to be admitted in the ICU is made by the attending physician, together with critical care specialists. Population and sample All adult patients at risk, e.g. patients >/ = 18 years old hospitalized for more than 48 h in this tertiary level institution regardless of location (e.g. general medical floor, telemetry, intensive care unit) from March 2023 until March 2024 were systematically screened using a microbiology database. Those meeting CDC criteria for HA-BSI (e.g., prescription of an intravenous antibiotic and positive growth of bacterial or Candida sp. from blood specimen taken on or after the 3rd day of admission; bacterial isolates defined as a non-common commensal isolate or the same common commensal isolate from two or more samples collected on separate sites or occasions) and VAP (clinical suspicion of VAP, prescription of an intravenous antibiotic, AND growth of bacterial pathogen from respiratory samples taken on or after day 3 of ventilation) were enrolled [1]. Patients who had positive growth of organisms belonging to the predetermined list of common commensal organisms identified as typical causes of community-acquired infections, and are rarely or not known to be causes of healthcare associated infections, or are associated with severe immunosuppression (e.g. Pneumocystis jirovecii in HIV/AIDS, Aspergillus or Cryptococcus species in transplant recipients) were excluded. To estimate the incidence of HA-BSI and VAP, incidence per 1000 patients was derived using the number of patients with HA- BSI or VAP as the numerator, and total hospitalized adult patients admitted for more than 48 h during the study period as the denominator. Incidence density per 1000 patient-days was computed using number of infection episodes as numerator, and total patient-days as denominator. The total number of ventilator days was not captured but would have been optimal. The intended target sample size of the main ACORN HAI study for each study site is 250 per infection syndrome. A prior meta-analysis with a reported 20% prevalence for HA-BSI and VAP, and a relative uncertainty of. 05 served as basis for sample size computation [4]. Methods Informed consent was initially secured from patients (3/2023-10/2024) but this requirement was later waived after further ethics review, based on similar waivers from other study sites to eliminate sampling bias (e.g., exclusion of critically ill patients in the hospital). On enrollment, baseline clinical data including demographics, date of admission and ward location, date of mechanical ventilation, primary reason for admission and type (emergency or elective), comorbidities, healthcare exposure in the 3 months prior, and date of ID referral were extracted from the patient clinical records. In our institution, ID referral is physician dependent, and is mandated only when carbapenem resistance is identified. Clinical laboratory measurements including microbiological data and laboratory parameters, as well as hospitalization data which included date of extubation or reintubation (if applicable), clinical severity signs, and antibiotic prescriptions and appropriateness, were collected for each infection episode. For patients who developed multiple episodes of infection during a single admission, each episode of either HA-BSI and VAP developing within the same admission was counted as a new case. The most frequent organisms isolated and corresponding susceptibility patterns were reported based on the phenotypic susceptibility pattern (pansusceptible, extended spectrum beta-lactamase [ESBL], or carbapenemase) stated in the final report released by the microbiology laboratory. Follow up at 28 days after onset of the last infection was carried out by chart review (if still hospitalized) or via telephone call (if already discharged) to determine final health and functional status through the use of the Functional Bloodstream Infection Score (FBIS). Definition of terms Multidrug-resistance (MDR) - non-susceptibility to at least one agent in three or more antimicrobial categories [6]. Infectious Disease (ID) referral - a formal referral or consultation with the infectious disease service. The reasons for referral are varied, and include antibiotic guidance, evaluation of diagnostic dilemmas, or managing complex or persistent infections. Although ID specialists are known as antimicrobial stewards, a formal consult is independent of the existing antimicrobial stewardship program of the hospital. Referral was defined as early, if facilitated within 3 days (72 h) [7] of clinical suspicion of infection onset, and late, if beyond 3 days (72 h), regardless of the availability of final culture data. The timing of ID referral was at the complete discretion of the attending physician. Ventilator associated Pneumonia - pneumonia occurring >48 h after endotracheal intubation, documented through diagnostic chest imaging and/or clinical worsening such as tachypnea, desaturation, or increased oxygen requirements. Tracheal aspirates are usually obtained once there is a suspicion of VAP. However, the diagnostic evaluation for VAP is not standardized and is left to the discretion of the healthcare team. Functional Bloodstream Infection Score (FBIS) - a novel performance score to determine patient outcomes in those who have been treated for blood stream infections, that highlights patients’ actual functional status. The score ranges from zero to seven, representing the worst and the best outcomes, respectively [8]. The FBIS is shown in Supplementary Table 1. Outcomes The primary outcome for this study was to determine the incidence of HA-BSI and VAP. Secondary outcomes included mortality and the impact of ID referral (e.g. on mortality, length of stay, FBIS score). Analysis Descriptive statistics were used to summarize the demographic, clinical, microbiological, and treatment-related characteristics of patients with hospital-acquired infections. Analyses were performed at both the patient and episode level, depending on the nature of the variable. Categorical variables were presented as frequencies and percentages, while continuous variables were summarized as mean ± standard deviation (SD) for approximately normally distributed data or median with interquartile range (IQR) for skewed data. Incidence per 1000 patients was derived using the total number of hospitalized adult patients admitted for more than 48 h during the study period as the denominator, while incidence density per 1000 patient-days was computed using total patient-days. Corresponding 95% confidence intervals (CIs) were estimated using Poisson approximation. For categorical variables, the chi-square test or Fisher's exact test was used, as appropriate. For continuous variables, comparisons were conducted using one-way analysis of variance (ANOVA) for normally distributed data or the Kruskal–Wallis test for non-normally distributed data. Missing variables were neither replaced nor estimated. The null hypothesis was rejected at a significance level of. 05 α. R-4.1.3 was used for data analysis. Results Patient profiles and baseline characteristics A total of 15,541 adult patients comprising 95,811/1000 patient days were admitted for more than 48 h during the study period. Microbiological data collected from 8642 patients involved 7862 blood and 4663 respiratory samples from 3939 and 4703 patients, respectively. Among these, 168 blood samples had positive growth, and after accounting for duplicates, 99 fulfilled criteria for HA-BSI. 44 out of 562 samples from ventilated patients fulfilled criteria for VAP. A total of 115 patients were enrolled - 77 (66.96%) with HA-BSI alone, 24 with VAP alone (20.87%), and 14 (12.17%) who developed both HAIs (Fig. 1).Fig. 1Flowchart of screened patients, blood and respiratory samples. Flowchart of screened patients, blood and respiratory samples. More than half of patients were male (58.26%) and mean age (62.37 ± 18.11) was 62.5 years old. The most common comorbidity was diabetes mellitus (33.91%). 21 (18.1%) patients had surgery in the last 3 months before admission (Table 1). The most common empiric antibiotic used was meropenem (38.46%) (Supplementary Table 2). More patients with VAP were emergently admitted (22/24, 91.67%) compared with HA- BSI (56/77, 72.73%), whereas patients with both infections were all initially admitted as emergency cases (14, 12.17%) (Table 1). Mean days between admission to infection onset was shorter for HA-BSI and VAP alone at 10.5 (5-19) days and 11 (6.5-22.25) days, respectively, compared to those with both infections at 26 (12-41) days ( Supplementary Table 3).Table 1Baseline characteristics of patients with hospital-acquired infection (n = 115).Table 1All (n = 115)HA-BSI (n = 77, 66.96%)VAP (n = 24, 20.87%)Both (n = 14, 12.17%)pMean ± SD; Frequency (%)Age, years62.37 ± 18.1160.38 ± 16.5965.96 ± 21.7267.14 ± 18.99.303∗Sex.752ψ Female48 (41.74)34 (44.16)9 (37.50)5 (35.71) Male67 (58.26)43 (55.84)15 (62.50)9 (64.29)Comorbidities Cancer24 (20.87)18 (23.68)2 (8.33)4 (28.57).188† Chronic Pulmonary Disease6 (5.22)2 (2.60)2 (8.33)2 (14.29).071† Congestive Heart Failure12 (10.43)9 (11.69)2 (8.33)1 (7.14)>.999† Dementia3 (2.61)1 (1.30)1 (4.17)1 (7.14).253† Diabetes Mellitus39 (33.91)28 (36.36)8 (33.33)3 (21.43).578† Metastatic solid tumor5 (4.35)4 (5.19)1 (4.17)0>.999† Renal Disease18 (15.65)13 (16.88)1 (4.17)4 (28.57).093† Others76 (66.09)51 (66.23)15 (62.50)10 (71.43).854ψType of Admission.012† Elective23 (20.00)21 (27.27)2 (8.33)0 Emergency92 (80.00)56 (72.73)22 (91.67)14 (100.00)Ward location at admission<.001† General ward60 (52.17)54 (71.05)5 (20.83)1 (7.14) ICU46 (40.00)13 (17.11)19 (79.17)13 (92.86) Hematology-oncology8 (6.96)8 (10.53)00 Surgical unit1 (.87)1 (1.32)00Transfer from Another Hospital7 (6.09)2 (2.60)3 (12.50)2 (14.29).052†Prior healthcare exposure Surgery within the last 3 months prior to admission17 (14.78)11 (14.47)5 (20.83)1 (7.14).521† Overnight hospitalization in the last 3 months before admission47 (40.87)34 (44.74)8 (33.33)5 (35.71).555ψ Regular hospital contact in the last 3 months25 (21.74)19 (25.00)2 (8.33)4 (28.57).166†Episode number<.001† One episode94 (81.74)70 (90.91)24 (100.00)0 Two episodes15 (13.04)7 (9.09)08 (57.14) Three episodes5 (4.35)005 (35.71) Four episodes1 (.87)001 (7.14)Statistical Analysis Used: ∗-One-way ANOVA; ψ–Chi-square test; †–Fisher's Exact test.This table is patient-level. Each patient was counted once. Table 1Baseline characteristics of patients with hospital-acquired infection (n = 115).Table 1All (n = 115)HA-BSI (n = 77, 66.96%)VAP (n = 24, 20.87%)Both (n = 14, 12.17%)pMean ± SD; Frequency (%)Age, years62.37 ± 18.1160.38 ± 16.5965.96 ± 21.7267.14 ± 18.99.303∗Sex.752ψ Female48 (41.74)34 (44.16)9 (37.50)5 (35.71) Male67 (58.26)43 (55.84)15 (62.50)9 (64.29)Comorbidities Cancer24 (20.87)18 (23.68)2 (8.33)4 (28.57).188† Chronic Pulmonary Disease6 (5.22)2 (2.60)2 (8.33)2 (14.29).071† Congestive Heart Failure12 (10.43)9 (11.69)2 (8.33)1 (7.14)>.999† Dementia3 (2.61)1 (1.30)1 (4.17)1 (7.14).253† Diabetes Mellitus39 (33.91)28 (36.36)8 (33.33)3 (21.43).578† Metastatic solid tumor5 (4.35)4 (5.19)1 (4.17)0>.999† Renal Disease18 (15.65)13 (16.88)1 (4.17)4 (28.57).093† Others76 (66.09)51 (66.23)15 (62.50)10 (71.43).854ψType of Admission.012† Elective23 (20.00)21 (27.27)2 (8.33)0 Emergency92 (80.00)56 (72.73)22 (91.67)14 (100.00)Ward location at admission<.001† General ward60 (52.17)54 (71.05)5 (20.83)1 (7.14) ICU46 (40.00)13 (17.11)19 (79.17)13 (92.86) Hematology-oncology8 (6.96)8 (10.53)00 Surgical unit1 (.87)1 (1.32)00Transfer from Another Hospital7 (6.09)2 (2.60)3 (12.50)2 (14.29).052†Prior healthcare exposure Surgery within the last 3 months prior to admission17 (14.78)11 (14.47)5 (20.83)1 (7.14).521† Overnight hospitalization in the last 3 months before admission47 (40.87)34 (44.74)8 (33.33)5 (35.71).555ψ Regular hospital contact in the last 3 months25 (21.74)19 (25.00)2 (8.33)4 (28.57).166†Episode number<.001† One episode94 (81.74)70 (90.91)24 (100.00)0 Two episodes15 (13.04)7 (9.09)08 (57.14) Three episodes5 (4.35)005 (35.71) Four episodes1 (.87)001 (7.14)Statistical Analysis Used: ∗-One-way ANOVA; ψ–Chi-square test; †–Fisher's Exact test.This table is patient-level. Each patient was counted once. Baseline characteristics of patients with hospital-acquired infection (n = 115). Statistical Analysis Used: ∗-One-way ANOVA; ψ–Chi-square test; †–Fisher's Exact test. This table is patient-level. Each patient was counted once. HA-BSI and VAP Eighty-one patients developed 99 HA-BSI episodes; of these, the source of BSI was primary in 53, and secondary in 46. Secondary BSI was thought to arise from the following sources – abdominal (n = 14), pulmonary (n = 13), urinary (n = 11), skin and soft tissue (n = 7), or unclear/unknown (n = 1). BSI episodes occurred in the general medical ward (n = 55) or ICU (n = 26); only a few occurred while in the hematology-oncology (n = 8) or surgery units (n = 1). HA-BSI (69.23%) accounted for the more substantial proportion compared to VAP (30.77%), with a computed incidence density of 1.03 bloodstream infection per 1000 patient days. Thirty-four patients developed 44 VAP episodes, with an overall incidence density of. 46 per 1000 patient days (Supplementary Table 4). Majority of total VAP episodes (including those who developed more than 1 type of HAI) occurred while located in the ICU (n = 39); only a few were in the general medical ward (n = 5). Isolates and susceptibility patterns Among patients with BSI, Coagulase negative Staphylococcus (CoNS), Klebsiella pneumoniae, Staphylococcus aureus and Acinetobacter baumannii were the most common pathogens isolated (Supplementary Table 5). Majority of CoNS were methicillin resistant (12/13, 92.3%) while most of the S. aureus isolates were susceptible to methicillin (8/10, 80%). Majority of K. pneumoniae were pansusceptible (9/10, 81.82%) while many A. baumannii isolates were carbapenem-resistant (6/9, 66.67%). Among patients with VAP, Pseudomonas aeruginosa was the most commonly isolated organism (n = 15), of which 6/15 (40%) were carbapenem-resistant. Majority of K. pneumoniae (8/13, 61.5%) and A. baumannii (3/5, 60%) were also pansusceptible. Patient outcomes Mean length of hospitalization was longest for those with both HAIs at 49 (range 39.25-93.25) days, followed by VAP at 39 (range 22-50) days and HA-BSI at 25.5 (range 15-47.5) days. In-hospital mortality rate was highest among those who developed both infections (8/14, 57.14%) compared to those patients with VAP alone (12/24, 50%) or those with BSI alone (19/77, 24.68%). Mean FBIS score was lower among VAP [4] than HA-BSI [5] patients. (Table 2). At the time of censorship, there were no documented mortalities within 28 days after discharge.Table 2Outcomes of patients with HA-BSI and VAP (n = 115).Table 2All (n = 115)HA-BSI (n = 77, 66.96%)VAP (n = 24, 20.87%)Both (n = 14, 12.17%)pMedian (IQR); Frequency (%)Length of Hospital Stay, days31.50 (17-49)25.50 (15-45.75)39 (22-50)49 (39.25-93.25)<.001§ICU admission during hospitalization70 (60.87)32 (41.56)24 (100.00)14 (100.00)<.001ψICU length of stay, days17 (7.50-24)9 (4-16)20.50 (16-23.25)26.50 (19.25-46)<.001§FBIS Score5 (4-6)5 (5-6)4 (4-5)4.50 (4-5.75).046§In-hospital Mortality.011† Alive76 (66.09)58 (75.32)12 (50.00)6 (42.86) Dead39 (33.91)19 (24.68)12 (50.00)8 (57.14)Discharge status at 28 days.009† Alive67 (64.42)51 (75.00)10 (45.45)6 (42.86) Dead37 (35.58)17 (25.00)12 (54.55)8 (57.14)Statistical Analysis Used: §–Kruskal-Wallis test; ψ–Chi-square test; †–Fisher's Exact test.This table is patient-level. Each patient was counted once.Mortality was reported as all-cause in-hospital mortality. Infection-attributable mortality was not available. Table 2Outcomes of patients with HA-BSI and VAP (n = 115).Table 2All (n = 115)HA-BSI (n = 77, 66.96%)VAP (n = 24, 20.87%)Both (n = 14, 12.17%)pMedian (IQR); Frequency (%)Length of Hospital Stay, days31.50 (17-49)25.50 (15-45.75)39 (22-50)49 (39.25-93.25)<.001§ICU admission during hospitalization70 (60.87)32 (41.56)24 (100.00)14 (100.00)<.001ψICU length of stay, days17 (7.50-24)9 (4-16)20.50 (16-23.25)26.50 (19.25-46)<.001§FBIS Score5 (4-6)5 (5-6)4 (4-5)4.50 (4-5.75).046§In-hospital Mortality.011† Alive76 (66.09)58 (75.32)12 (50.00)6 (42.86) Dead39 (33.91)19 (24.68)12 (50.00)8 (57.14)Discharge status at 28 days.009† Alive67 (64.42)51 (75.00)10 (45.45)6 (42.86) Dead37 (35.58)17 (25.00)12 (54.55)8 (57.14)Statistical Analysis Used: §–Kruskal-Wallis test; ψ–Chi-square test; †–Fisher's Exact test.This table is patient-level. Each patient was counted once.Mortality was reported as all-cause in-hospital mortality. Infection-attributable mortality was not available. Outcomes of patients with HA-BSI and VAP (n = 115). Statistical Analysis Used: §–Kruskal-Wallis test; ψ–Chi-square test; †–Fisher's Exact test. This table is patient-level. Each patient was counted once. Mortality was reported as all-cause in-hospital mortality. Infection-attributable mortality was not available. Impact of infectious diseases referral HA-BSI Of the 81 patients with HA-BSI, 75 (92.59%) were formally referred to ID. 19/72 (26. 39%) of those with early, and 2/4 (50%) late referrals died during the study period; there was no documented mortality among patients not referred to ID. Majority of BSI episodes with drug resistant pathogens (27/29, 93.10%) were referred. Length of stay was longest at 41 (range 29.5- 149.5) days in patients who had late ID referrals than those with early (26, range 16.5-48 days) or without (29, range 14-29 days) referral. FBIS score was lower for those referred to ID (5 vs. 6), but these were not statistically significant (Table 3).Table 3Association between timing of ID referral and outcomes.Table 3No referral (n = 20, 17.39%)Early referral (n = 90, 78.26%)Late referral (n = 5, 4.35%)pMedian (IQR); Frequency (%)Overall Patient-Level (n = 115)FBIS Score6 (4-6)5 (4-6)4.50 (4-5).358§In-hospital mortality.638† Alive15 (75.00)58 (64.44)3 (60.00) Dead5 (25.00)32 (35.56)2 (40.00)Discharge status at 28 daysn = 18n = 81.532† Alive13 (72.22)50 (61.73)4 (80.00) Dead5 (27.78)31 (38.27)1 (20.00)Length of Hospital Stay, days32.50 (22-51.75)31 (17-48)45 (37-73).516§HA-BSI Patient-Level (n = 81)No referral (n = 5, 6.17%)Early referral (n = 72, 88.89%)Late referral (n = 4, 4.94%)pMedian (IQR); Frequency (%)FBIS Score6 (6-6)5 (5-6)5 (4.50-5).399§In-hospital mortality.206† Alive5 (100.00)53 (73.61)2 (50.00) Dead019 (26.39)2 (50.00)Discharge status at 28 daysn = 5n = 63n = 3.500† Alive5 (100.00)45 (71.43)3 (75.00) Dead018 (28.57)1 (25.00)Length of Hospital Stay, days29 (14-29)26 (16.50-48)41 (29.50-149.50).690§VAP Patient-Level (n = 34)No referral (n = 15, 44.12%)Early referral (n = 18, 52.94%)Late referral (n = 1, 2.94%)pMedian (IQR); Frequency (%)FBIS Score4.50 (3.75-6)4 (4-5)4 (4-4).819§In-hospital mortality.037† Alive10 (66.67)5 (27.78)1 (100.00) Dead5 (33.33)13 (72.22)0Discharge status at 28 daysn = 13.072† Alive8 (61.54)5 (27.78)1 (100.00) Dead5 (38.46)13 (72.22)0Length of Hospital Stay, days39 (24.50-52.50)43 (39-52)73 (73-73).359Statistical Analysis Used: §–Kruskal-Wallis test; †–Fisher's Exact test.This table is patient-level. Each patient was counted once. Table 3Association between timing of ID referral and outcomes.Table 3No referral (n = 20, 17.39%)Early referral (n = 90, 78.26%)Late referral (n = 5, 4.35%)pMedian (IQR); Frequency (%)Overall Patient-Level (n = 115)FBIS Score6 (4-6)5 (4-6)4.50 (4-5).358§In-hospital mortality.638† Alive15 (75.00)58 (64.44)3 (60.00) Dead5 (25.00)32 (35.56)2 (40.00)Discharge status at 28 daysn = 18n = 81.532† Alive13 (72.22)50 (61.73)4 (80.00) Dead5 (27.78)31 (38.27)1 (20.00)Length of Hospital Stay, days32.50 (22-51.75)31 (17-48)45 (37-73).516§HA-BSI Patient-Level (n = 81)No referral (n = 5, 6.17%)Early referral (n = 72, 88.89%)Late referral (n = 4, 4.94%)pMedian (IQR); Frequency (%)FBIS Score6 (6-6)5 (5-6)5 (4.50-5).399§In-hospital mortality.206† Alive5 (100.00)53 (73.61)2 (50.00) Dead019 (26.39)2 (50.00)Discharge status at 28 daysn = 5n = 63n = 3.500† Alive5 (100.00)45 (71.43)3 (75.00) Dead018 (28.57)1 (25.00)Length of Hospital Stay, days29 (14-29)26 (16.50-48)41 (29.50-149.50).690§VAP Patient-Level (n = 34)No referral (n = 15, 44.12%)Early referral (n = 18, 52.94%)Late referral (n = 1, 2.94%)pMedian (IQR); Frequency (%)FBIS Score4.50 (3.75-6)4 (4-5)4 (4-4).819§In-hospital mortality.037† Alive10 (66.67)5 (27.78)1 (100.00) Dead5 (33.33)13 (72.22)0Discharge status at 28 daysn = 13.072† Alive8 (61.54)5 (27.78)1 (100.00) Dead5 (38.46)13 (72.22)0Length of Hospital Stay, days39 (24.50-52.50)43 (39-52)73 (73-73).359Statistical Analysis Used: §–Kruskal-Wallis test; †–Fisher's Exact test.This table is patient-level. Each patient was counted once. Association between timing of ID referral and outcomes. Statistical Analysis Used: §–Kruskal-Wallis test; †–Fisher's Exact test. This table is patient-level. Each patient was counted once. VAP Of the 34 patients with VAP, 19 (55.88%) were formally referred to ID. Patient deaths were highest among those with early referral to ID (13/18, 72.2%). Median length of hospital stay was longest for those with late ID referrals at 73 days, compared to early (43 days) and no referral (39 days). FBIS scores were similar across subgroups. Majority of VAP episodes with a drug resistant pathogen (14/18, 77.78%) were referred to ID (Table 3). Timing of infectious diseases referral Overall, the timing of ID referral was not significantly associated with the following patient outcomes -- FBIS score (p = .358), in-hospital mortality (p = .638), 28-day discharge status (p = .532), or length of hospital stay (p = .516). (Table 3). Timing of referral among HA-BSI episodes was not significantly associated with days from admission to infection onset, location at infection onset, ICU admission during hospitalization, ICU length of stay, or device exposure. However, late referrals showed a trend toward longer ICU stays and later infection onset, though these were not statistically significant (Supplementary Table 6). For referrals among those with episodes of VAP, late referrals to ID were associated with longer median time from admission to infection onset of 135 (range 53.75-257.25) days, and a trend towards longer median stay in the ICU of 40.5 (range 23-57) days (Supplementary Table 7). Appropriateness of empiric antimicrobial therapy HA-BSI Majority of empiric antimicrobials for BSI were prescribed by an ID specialist (85.86%). Of these, most were deemed appropriate for the organism/s eventually isolated from blood cultures with 64/85 (85.88%) from those referred early, and 4/5 (80%) from those referred late. In contrast, only 6/9 (66.67%) of empiric antimicrobials were deemed appropriate for HA-BSI isolates among those not referred to ID (Supplementary Table 6). VAP 44 antimicrobials were initiated as empiric treatment, around half of which were prescribed by an ID specialist (25/44, 56.82%) of which majority were deemed appropriate; 64% (16/25) were from those referred early, and 100% (4/4) among those referred late. About 11/15 (73.3%) of all antimicrobials prescribed as empiric treatment among those not referred to ID were deemed appropriate based on the organisms isolated from respiratory cultures (Supplementary Table 7). Discussion Interim data from this prospective cohort study highlighted the following – first, the incidence of nosocomial BSI and VAP were low, but were associated with high mortality and prolonged hospitalization, with patients developing both HA-BSI and VAP demonstrating the poorest outcomes. Second, Gram-negative pathogens were frequent, with the majority of Acinetobacter sp., exhibiting multidrug-resistance. Third, ID referral was more commonly performed when drug-resistance was present, and was more frequently carried out for BSI compared to VAP; and finally, there was a lack of association between ID referral and improved patient outcomes. This study showed an HA-BSI rate (1.03 per 1000 patient days) comparable to prior studies in other LMICs such as Thailand with a rate of 2.4 per 1000 patient days [9]. This relatively low incidence may mean effective implementation, good adherence, and compliance with infection prevention and control practices. For example, overall compliance with the central line insertion bundle, which includes hand hygiene, use of personal protective equipment (PPE), skin antisepsis, and optimal catheter site selection, was high in the ICU, at 90% based on a prior study [10]. However, the low incidence needs to be interpreted with caution as it may reflect under reporting due to incorrect sample collection techniques leading to false negative results (e.g. blood collection done after loading antimicrobial doses). The incidence of VAP in this study was low (.47 per 1000 patient days). This may be due to the improved adherence of healthcare staff to the VAP bundle of care; a prior study from the same institution showed overall median bundle compliance of 84.6% [11]. However, we were unable to calculate incidence using ventilator days, which is a better measure, and this needs to be interpreted cautiously. One study that evaluated the incidence of VAP in Asia based on national income levels reported that low to upper-middle income countries, including the Philippines, have higher incidence density of VAP at 18.5 and 15.2 per 1000 ventilator days respectively, and high-income countries have the lowest at 9.0 per 1000 ventilator days. This lower incidence may be reflective of the surveillance, education and tailored interventions and prevention bundles, while the higher incidence among lower income countries was attributed to the lack of enforcement or regulation of healthcare [12]. While the majority of A. baumannii in this study were carbapenem-resistant, the majority of K. pneumoniae were pansusceptible. These findings mirror the national Antimicrobial Resistance Surveillance Program (ARSP) resistance rates, particularly for those obtained from ICU patients (Imipenem resistance at 30% vs. 52.5%, respectively). Conversely, the K. pneumoniae institutional resistance rates fell below half of the reported ARSP rates (ESBL rate at 17% vs. 40.8%, respectively) [13,14]. This difference in susceptibility patterns suggest that the emergence of carbapenem-resistance in A baumannii may occur among patients with prolonged hospitalization and consequently, substantial antimicrobial use. Resistance to antimicrobials emerges because of a combination of exposures to different classes of antibiotics and the exchange of resistance genes between the microbes themselves [15]. As such, more than reliance on country level data, the institutional antibiogram is important in guiding initial therapies, and identifying which pathogens are more likely the cause of infection syndromes within the institution. While the positive impact of a formal ID referral is usually discussed in the context of S. aureus bacteremia, and where a significant reduction in in-hospital mortality has been reported, recent studies of other infectious syndromes have also shown its favorable effect on reducing length of hospitalization, especially among patients who are evaluated early in the course of their admissions [3,5,16]. Despite studies that demonstrate that early referral to ID is associated with decreased odds for mortality in patients with HA-BSI, shorter hospital stay, lower healthcare costs, and lower mortality at under 1.5% across the index stay and post-discharge period, compared to those referred late in the course of admission and those without any ID intervention at all [17], this study was unable to show a significant association between timing of ID referral and clinical outcomes. In fact, this study reports poor outcomes even among patients referred to ID (e.g., longer length of stay, lower FBIS score, higher mortality), although these were not statistically significant. These poor outcomes likely reflect the severity of illness and complexity of cases at the time of ID referral. In addition, the threshold for referral may vary according to the provider (e.g. more experienced physicians may refer less often, or at a later time compared to inexperienced ones), and we did not evaluate or stratify for this. In general, however, ID referrals are placed when a patient worsens, or when use of broadspectrum antimicrobials are warranted; that majority of HA-BSI events were referred to the ID service, or that referrals occurred when infections occurred later in the course of admission or the ICU stay are reflective of this indication bias (e.g. tendency to refer when a patient is critically ill or deteriorating). This was also observed in a prior study where patients receiving ID consultation had a significantly higher patient clinical complexity level (PCCL) and case mix index (CMI) which led to deviations in the expected mean length of stay primarily due to the need for further treatment even after discharge [18]. Limitations This study has several limitations. The sample size is small, and data is preliminary. Informed consent was initially required during the first few study months, which affected the sample size, and may have led to sampling bias, with exclusion of critically ill patients (e.g family members of critically ill patients or those with poor prognosis [e.g. terminal cancer] were more inclined to refuse enrolment). In addition, use of microbiologically defined VAP as part of the case definition may have under-estimated the true incidence of infection, as some VAPs are syndromic and culture negative. As a single-institution study, results may not be generalizable to other small hospitals in the country. Attributable mortality as an outcome would have been helpful, but was not captured, as it was not an outcome of interest of the original ACORN-HAI protocol. Finally, findings may not accurately reflect the true impact of early ID referral given the small sample size, the likelihood of confounding by severity, indication bias, and referral practices. Nevertheless, this interim data provides necessary information on HAIs, and can provide some early guidance on therapeutic choices and subspecialty referral. Conclusion The rates of HA-BSI and VAP in this study are low, but were associated with high mortality and prolonged hospitalization. Gram-negative pathogens were also frequent and drug-resistant A. baumannii common, highlighting the need to prevent these nosocomial infections. Benefits of early ID referral were not observed in this study, but firm conclusions cannot be made given the small sample size, and other limitations of this study. Other possible benefits of early ID referral such as timely de-escalation of therapy and shorter duration of treatment were not assessed, and need to be studied further. Ethical approval statement The study has been reviewed by the Oxford Tropical Research Ethics Committee, NHG Domain Specific Review Board, and has been approved by the Institutional Review Board of TMC(CDT MED 2022-013, January 2023). Funding source This study has been funded by 10.13039/100010269Wellcome Trust (Grant number 360G-Wellcome-225457_Z_22_Z, September 2022). CRediT authorship contribution statement Stephanie N. Co-Chang: Data curation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. Nicole Bernadette S. Terre: Data curation, Formal analysis, Investigation, Writing – original draft, Writing – review & editing. Cybele Lara R. Abad: Conceptualization, Methodology, Supervision, Writing – original draft, Writing – review & editing. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.