Discussion

Currently, no published data are available for or have been analyzed regarding PIVCR BSI rates in Latin America. This 8-year study is the first to determine PIVCR BSI rates per 1,000 device days in this region, including data from 10,120 ICU patients for 40,078 bed days and 38,262 short-term PIVC days, from 100 ICUs among 83 hospitals in 26 cities of 9 countries.^{Reference Alexandrou, Ray-Barruel and Carr15}

The pooled mean PIVCR BSI rate was 2.06 per 1,000 PIVC days (95% CI, 1.635–2.257). Similar PIVCR BSI rates have been reported by the INICC recently. In a pool of 42 countries worldwide, the PIVCR rate was 2.41 per 1,000 PIVC days.^{Reference Rosenthal, Bat-Erdene and Gupta7} In Asia, the rate was 2.65 per 1,000 PIVC days,^{Reference Rosenthal, Bat-Erdene and Gupta16} and in the Middle East, the rate was 2.32 per 1,000 PIVC days.^{Reference Rosenthal, Belkebir and Zand8} The incidence of PIVCR BSI has been determined using the number of PIVC days in 2 studies from more economically developed countries: a 2006 systematic review with data from the United States, Australia, and Italy, observed a rate of 0.5 PIVCR BSIs per 1,000 PIVC days,^{Reference Maki, Kluger and Crnich4} and a 2018 study conducted in pediatric and neonatal ICUs from Australia, noted a rate of 0.67 PIVCR BSIs per 1,000 PVC days.^{Reference Worth, Daley, Spelman, Bull, Brett and Richards17}

In a 2019 systematic review by Alliance for Vascular Access Teaching and Research (AVATAR) group, the selected studies did not report PIVC days as denominators of PIVCR BSI rates. In consequence, such data are not comparable to our present study.^{Reference Ray-Barruel, Xu, Marsh, Cooke and Rickard18} The cited AVATAR review did include studies that reported the following PIVCR BSI rates^{Reference Ray-Barruel, Xu, Marsh, Cooke and Rickard18}: 0.39 PIVCR BSIs per 10,000 occupied bed days in Australia^{Reference Rhodes, Cheng and McLellan19}; 3.04 PIVCR BSIs per 1,000 patient days in Germany^{Reference Salm, Schwab, Geffers, Gastmeier and Piening20}; 1.17 PIVCR BSIs per 10,000 patient days in Spain^{Reference Saliba, Hornero and Cuervo21}; and 0.05 PIVCR BSIs per 1,000 patient days,^{Reference Freixas and Bella22} and 0.0150 PIVCR BSIs per 100 patient days,^{Reference DeVries, Valentine and Mancos23} and 0.57 PIVCR BSIs per 1,000 patient days in the United States.^{Reference Duncan, Warden, Bernatchez and Morse24}

Many studies have reported on the adverse consequences of BSIs in ICUs and on the comparative infection risks of central lines (CLs) versus PIVCs, with CLs being much more prone to higher BSIs rates than PIVCs.^{Reference Miliani, Taravella and Thillard25}

A 2019 INICC study published data of CLABSIs in 45 countries, prospectively collected over 6 years (2012–2017), from 532,483 ICU patients hospitalized in 523 ICUs of 242 hospitals for an aggregate of 2,197,304 patient days. In the medical-surgical ICUs, the pooled CLABSI rate was 5.05 per 1,000 central-line days.^{Reference Rosenthal, Bat-Erdene and Gupta11} A comparison of these 2019 CLABSI data with the present study on PIVCR BSI indicates that the CLABSI rate is 45% higher than the PIVCR BSI rate. However, since ~80%–90% of the vascular catheters used worldwide are PIVCs, the raw number of BSIs resulting from PVCs is ~6 times higher than the number of BSIs resulting from central lines.^{Reference Mermel1,Reference Alexandrou, Ray-Barruel and Carr15}

The pooled mean of the distribution of crude mortality was 6.33% of PIVCR BSI cases in our ICUs, compared to 3.67% mortality of PIVC patients who were not infected. Nevertheless, the power of the sample was not enough to show significant difference. Mortality rates attributable to PIVCR BSI in recent studies in Spain and Japan were 13.2% and 12.9% respectively, both of which were higher than in our study.^{Reference Saliba, Hornero and Cuervo21,Reference Sato, Nakamura and Fujita26}

The added ALOS of patients with PIVCR BSI (5.29 days) was 33% higher than in patients without PIVCR BSI (3.95 days). Comparable data showing the extra ALOS attributable to PIVCR BSI are unavailable. Notwithstanding, in the aforementioned INICC study, patients who acquired CLAB had on average 9.4 extra days of hospital stay.^{Reference Rosenthal, Bat-Erdene and Gupta11}

Ripa et al^{Reference Ripa, Morata and Rodriguez-Nunez27} found the following incidences of microorganism in patients with PIVCR BSI: gram-positive cocci, 75% (Staphylococcus aureus 46%, Coagulase-negative staphylococci 25%, Enterococcus spp 3.7%); gram-negative bacilli, 22.8% (Klebsiella spp 5.6%, Pseudomonas aeruginosa 4.5%, Escherichia coli 4.1%, Enterobacter spp 3.9%, Serratia spp 0.7%, Acinetobacter spp, 0.7%) and Candida spp, 1.3%.^{Reference Ripa, Morata and Rodriguez-Nunez27} A 2019 study reported a significant increase in the proportion of gram-negative infections with an interval of 20 years: 22.6% in 1992–1996 versus 33.2% in 2012–2016.^{Reference Ripa, Morata and Rodriguez-Nunez27} Enterococcus faecalis was 100% sensitive to vancomycin, contrary to the findings of the aforementioned India study in which vancomycin was 100% sensitive to PVCR-BSI Enterococcus spp.^{Reference Dalai, Padhi, Padhi and Parida14}

PIVCR BSI surveillance by number of device days is essential to reducing the hospitalized patients’ risk of infection because it accurately describes the threat of PIVCR BSIs. Additionally, multifaceted and surveillance programs aimed toward PIVCR BSI prevention and control must be implemented. To this end, INICC bundles for insertion and maintenance of PIVCs have been published and applied in limited-resource settings over the past 3 years.²⁸ Likewise, antimicrobial resistance should be addressed and susceptibility to antimicrobials of PIVCR-BSI–associated pathogens should be reported to effectively prevent the transmission of resistant strains.^{Reference Ray-Barruel, Xu, Marsh, Cooke and Rickard18,Reference Saliba, Hornero and Cuervo21}

In the present study, we focused exclusively on the ICU setting. This is the healthcare environment with the highest HAI rates because ICU patients have critical medical conditions and are most often exposed to invasive devices.^{Reference Rosenthal, Maki and Salomao29} The INICC was worked tirelessly over the last 20 years and across the 6 WHO regions in an effort to combat the burden of HAIs. Increasing hand hygiene compliance and improving compliance with infection control bundles have proven to be successful strategies to prevent CLABSIs as described in several INICC publications.^{28,Reference Alkhawaja, Saeed and Rosenthal30–Reference Higuera, Rosenthal, Duarte, Ruiz, Franco and Safdar39} The present data can guide the implementation of prevention strategies and other quality improvement efforts for the reduction of PIVCR BSI rates and their adverse consequences.

This study has several limitations. First, we have not provided insights into the influence of INICC interventions,^{28,Reference Rosenthal, Duenas and Sobreyra-Oropeza34–Reference Rosenthal, Maki and Rodrigues38} such as the implementation of INICC multidimensional approach and ISOS.^{Reference Rosenthal10,28,Reference Rosenthal, Duenas and Sobreyra-Oropeza34–Reference Rosenthal, Maki and Rodrigues38} Second, trends in data over time are not presented for this 8-year study. Moreover, changes in compliance of healthcare workers with preventive measures were not considered. Furthermore, most PIVCs were inserted in the ICU, which may have affected the PIVCR BSI rate. In addition, this study was limited to a benchmark comparison with a few studies that also report PIVCR BSIs by PIVC days. Furthermore, due to limited resources, cultures taken were probably less than ideal, which likely influenced PIVCR BSI rates. Also, resistance rates cannot be generalized due to the small simple size. Additionally, we did not obtain data on the illness severity score at patient admission to the ICU, which is likely associated with crude mortality. Finally, to define ALOS in patients with and without BSI, the time of origin was counted from the first day of admission, and it was not possible to determine whether the longer ALOS was the cause or the consequence of BSI.

In conclusion, we have presented the only comprehensive data on PIVCR BSIs per 1,000 PIVC days from Latin America currently available. Consequently, the benchmark comparison of our findings was limited to the results of 2 studies from more economically developed countries (a 2006 systematic review of data from the United States, Australia, and Italy and an Australian study published in 2018^{Reference Worth, Daley, Spelman, Bull, Brett and Richards17}) and to 4 studies INICC conducted in countries of limited resources in Asia^{Reference Aygun, Yasar and Yilmaz6} and the Middle East^{Reference Rosenthal, Belkebir and Zand8} and in a pool of 42 countries.^{Reference Rosenthal, Bat-Erdene and Gupta7} The PIVCR BSI rates in this study were higher than those of economically developed countries. Thus, it is clear that PIVCR BSIs in ICUs from limited-resource countries are a detriment to patient safety. The systematic surveillance of PIVCR BSI and prevention programs, such as antibiotic resistance reports, should be implemented widely to reduce the incidence of PIVCR BSI and its adverse consequences worldwide.

One key finding of this research is that, in Latin America, healthcare workers should conduct surveillance of PIVC BSI rates, extra length of stay, and extra mortality. They should develop and implement bundles for insertion and maintenance of PIVC, including proper hand hygiene before insertion and care, select the insertion site with less risk, apply aseptic technique, use chlorhexidine skin antisepsis, use a sterile dressing, use needle-free connectors instead of 3-way stopcock, scrub the access to the catheter before use, use of prefilled syringes, change administration sets every 4 days unless use for blood transfusion or lipids administration, do not change PIVC at fixed intervals, remove PIVC when they are not needed. Finally, they should monitor compliance with bundles.