Nonsterile Glove Use in Addition to Hand Hygiene to Prevent Late-Onset Infection in Preterm Infants: Randomized Clinical Trial

Importance  Late-onset infections commonly occur in extremely preterm infants and are associated with high rates of mortality and neurodevelopmental impairment. Hand hygiene alone does not always achieve the desired clean hands, as microorganisms are still present more than 50% of the time. We hypothesize that glove use after hand hygiene may further decrease these infections.

Objective  To determine if nonsterile glove use after hand hygiene before all patient and intravenous catheter contact, compared with hand hygiene alone, prevents late-onset infections in preterm infants.

Design, Settings, and Participants  A prospective, single-center, clinical, randomized trial was conducted in infants admitted to the neonatal intensive care unit who weighed less than 1000 g and/or had a gestational age of less than 29 weeks and were less than 8 days old. There were 175 eligible infants, of which 120 were enrolled during a 30-month period from December 8, 2008, to June 20, 2011.

Interventions  Infants were randomly assigned to receive care with nonsterile gloves after hand hygiene (group A) or care after hand hygiene alone (group B) before all patient and intravenous line (central and peripheral) contact. Study intervention was continued while patients had central or peripheral venous access.

Main Outcomes and Measures  One or more episodes of late-onset (>72 hours of age) infection in the bloodstream, urinary tract, or cerebrospinal fluid or necrotizing enterocolitis.

Results  The 2 groups were similar in baseline demographic characteristics. Late-onset invasive infection or necrotizing enterocolitis occurred in 32% of infants (19 of 60) in group A compared with 45% of infants (27 of 60) in group B (difference, −12%; 95% CI, −28% to 6%; P = .13). In group A compared with group B, there were 53% fewer gram-positive bloodstream infections (15% [9 of 60] vs 32% [19 of 60]; difference, −17%; 95% CI, −31% to −1%; P = .03) and 64% fewer central line–associated bloodstream infections (3.4 vs 9.4 per 1000 central line days; ratio, 0.36; 95% CI, 0.16 to 0.81; P = .01).

Conclusions and Relevance  Glove use after hand hygiene prior to patient and line contact is associated with fewer gram-positive bloodstream infections and possible central line–associated bloodstream infections in preterm infants. This readily implementable infection control measure may result in decreased infections in high-risk preterm infants.

Trial Registration  clinicaltrials.gov Identifier: NCT01729000

As major advances in neonatal intensive care have increased the survival rates and numbers of extremely preterm infants, late-onset infections and necrotizing enterocolitis (NEC) continue to occur in many of these patients and are associated with high rates of mortality and neurodevelopmental impairment.^1-4 Studies of critically ill patients in the neonatal intensive care unit (NICU) indicate that direct infant contact occurs an average of 32 times per shift.^5,6 Even after appropriate hand hygiene, cultures from the hands of medical staff yield microorganisms more than 50% of the time.⁷ Standard hand hygiene lowers the number of microbes on health care professionals’ hands to a few colony-forming units (CFUs) and potentially eliminates transient flora, but permanent flora may persist.^6,8 These CFUs and permanent flora could still be a significant exposure to extremely preterm newborns owing to their immature immune system, especially their underdeveloped skin and mucosal barriers.⁹

In a study on the dynamics of bacterial transmission in the NICU, using gloves was associated with significantly lower CFU counts on the hands of health care professionals in all aspects of patient care, including skin contact, diaper changes, vascular catheter care, and suctioning of respiratory secretions.¹⁰ Adult studies have shown that gloves are also effective in preventing contamination of health care professionals’ hands with Clostridium difficile or vancomycin-resistant Enterococcus.^11,12 In a retrospective study of preterm infants with a birth weight less than 1500 g, Ng et al¹³ studied glove use during care and demonstrated a significant decrease in late-onset bloodstream infections (BSIs) and NEC compared with the previous period, when hand hygiene alone was the standard of care. Another study involving all pediatric units observed an association with their universal glove use policy during respiratory syncytial virus season and lower rates of bacteremia and central line–associated bloodstream infections (CLABSIs).¹⁴ These studies demonstrate that glove use in addition to hand hygiene results in lower CFUs and acquisition of pathogens on the hands of health care professionals and may be associated with lower rates of infections and NEC.

Preventing infections is critical to improving rates of survival without neurodevelopmental impairment and decreasing duration of hospital stays and costs in this population. Although hand hygiene infection control and prevention measures have been the topics of multiple epidemiologic and retrospective studies, they have not been subjected to randomized clinical trials in the NICU.^15-18 Our objective was to perform a randomized clinical trial to examine whether nonsterile glove use after hand hygiene prior to all direct patient and catheter contact, compared with hand hygiene alone, prevents late-onset infections or NEC in preterm infants who weigh less than 1000 g and/or have a gestational age of less than 29 weeks.

Hand hygiene followed by nonsterile glove use prior to all patient and intravenous (central or peripheral) catheter contact will result in decreased late-onset infections or NEC.

Quiz Ref IDWe conducted a prospective, unmasked, randomized clinical trial of 120 preterm infants. All inborn or outborn infants admitted to the University of Virginia NICU were eligible for the study if they had a birth weight of less than 1000 g or gestational age of less than 29 weeks and were less than 8 days old. The protocol was approved by the institutional review board at the University of Virginia. Written informed consent was obtained by the clinical research nurse coordinator or another member of the local study team for all study participants.

Infants were randomized (using block randomization with randomly varying block sizes of 2 and 4) into one of the 2 groups using a randomization table generated by our biostatisticians. During the study period, health care professionals, study coordinators, and principal investigators were unmasked to the study intervention. The biostatistician was masked to the study assignment during and after the study period for all data analysis.

The study intervention was nonsterile glove use after hand hygiene (group A) or hand hygiene alone (group B) prior to all patient, bed, and/or catheter contact. Signs were placed on a stand at the bedside of all enrolled patients (with a box of gloves if in group A) indicating group assignment and protocol. Intervention was specified to be performed during all contact with the patient, inside the bed area, and with all central and peripheral venous catheters. Contact with central and peripheral catheters was defined as whenever there was catheter contact and when making or breaking a connection with the hub when (1) giving medications or flush, (2) changing tubing, (3) accessing an injection port, and (4) adding a device.

The NICU is a 45-bed unit divided into 7 pods with 6 to 9 infant bed spaces. Each bed space is between 5.4 and 7.2 m² and has its own alcohol hand rub dispenser within arm’s reach of the patient; each pod has 1 to 2 sinks. Independent of group assignment, all health care professionals followed the 5 moments of hand hygiene from the World Health Organization guidelines^19(pp99-123) for hand hygiene in health care, used nonsterile gloves for contact with body fluids, and used sterile gloves for aseptic procedures. For both groups, nonsterile gloves were used when accessing arterial lines. For all other instances of patient or catheter (central or peripheral) contact, the study intervention was performed according to group assignment. Hand hygiene was defined as using alcohol hand rub or washing hands with antimicrobial soap (eg, 2% chlorhexidine gluconate). Auditing of compliance was performed throughout the study.

Additional infection control and prevention practices in our NICU included a CLABSI bundle for placement, maintenance, and removal of catheters; fluconazole prophylaxis for all infants who weighed less than 1000 g at birth and/or had a gestational age of less than 28 weeks, or any infant with NEC or gastroschisis; antibiotic stewardship including limited use of third- and fourth-generation cephalosporins and carbapenems; limited use of postnatal corticosteroids, histamine H₂ receptor blockers, and proton pump inhibitors; and weekly changing of all nasogastric and orogastric tubes. All patients with NEC in both groups were placed in contact isolation in which gowns and nonsterile gloves were used while patients were receiving antimicrobials.

Quiz Ref IDTo target the period during which preterm infants are at highest risk of infection or NEC and their related mortality, a minimum duration of intervention of 4 weeks after birth was selected. The intervention was extended beyond the 4 weeks for any infant who still required intravenous access (peripheral or central). In addition, if the line was removed and then subsequently needed, the study intervention assigned to that patient was followed again until intravenous access was discontinued. The presence of an intravenous line was selected for the intervention period (beyond the initial 4 weeks), as it correlates with the presence of risk factors of infection, including parenteral nutrition, lipid infusion, use of broad-spectrum antibiotics, and the intravenous line itself (if a central catheter).²⁰ This strategy focuses on the individual infant’s risk of infection, which potentially limits exposure to the intervention and its associated costs.

Quiz Ref IDLate-onset invasive infection (>72 hours after birth) was defined as 1 or more episodes per patient of a BSI, urinary tract infection, meningitis, and/or NEC associated with clinical signs and symptoms of infection and treated with antimicrobials. Blood, urine, or cerebrospinal fluid infections were defined as growth of bacteria or fungi from 1 or more cultures.²¹ Presence of NEC was defined as stage II or greater using the modified criteria of Bell et al.²² Blood cultures were obtained in aerobic culture bottles during examination for infection, with instructions to obtain 1 mL of blood. If gastrointestinal disease was suspected, anaerobic blood cultures were also obtained. Urine cultures were obtained by sterile bladder catheterization, and urinary tract infection was defined as growth of 10 000 or more CFU/mL. In addition to the Centers for Disease Control and Prevention definition of CLABSI, we also evaluated possible CLABSIs defined as detection of 1 or more blood cultures of any organism, including coagulase-negative Staphylococcus, and the presence of a central line within 72 hours in the absence of another source of infection. We included this definition because symptomatic BSIs (defined as growth in 1 or more blood cultures and treated), even when caused by coagulase-negative Staphylococcus organisms, are associated with increased neurodevelopmental impairment and white matter injury in preterm infants.^1,23-25 Criteria for determining evaluation for infection included but were not limited to temperature instability, hemodynamic compromise, respiratory deterioration, increased fraction of inspired oxygen, changes in vital signs and physiologic monitoring, and increased apnea, bradycardia, and desaturation events. Length of hospital stay, other infection-related outcomes, common neonatal morbidities, mortality, and cost of hospitalization were assessed as secondary outcomes.

The sample size of 120 infants was based on the assumption that the study would have a 2-sided type I error rate of .05 or less and at least 80% power to detect an absolute difference of 25% in the cumulative incidence of late-onset infection, including NEC, between the 2 groups, given a pretrial incidence of 60% based on local and national data.^20,26

The χ² test was used to compare categorical outcomes between groups; 95% CIs for the differences in proportions were calculated with the method of Miettinen and Nurminen.²⁷ The nonparametric Mann-Whitney test was used to compare continuous outcomes. Poisson regression using the appropriate number of days (line days or study days) as the offset was used to compare the groups with respect to rates per 100 days.

During the 30-month study period (December 2008 through June 2011), 175 eligible infants were screened (Figure 1). Seventeen were excluded because they were older than 7 days, and 161 met the criteria for enrollment in the study. Thirty-four were not enrolled owing to death prior to consent, parents being unavailable, known need for contact isolation for the entire hospitalization, or study refusal. A total of 124 infants were randomized. Four infants were randomized but never enrolled because they required continued contact isolation owing to methicillin-resistant Staphylococcus aureus colonization or the potential for maternal vancomycin-resistant Enterococcus acquisition for their entire hospitalization. Of the 120 remaining randomized infants, 60 were randomized to the group who received care with gloves after hand hygiene (group A) and 60 to the group who received hand hygiene alone (group B).

Demographic data and risk factors were similar between the groups (Table 1). The mean (SD) interval from birth until patient enrollment was 3.3 (1.7) days in group A vs 3.6 (1.9) days in group B (P = .29). The mean (SD) study intervention duration was 47 (34) days and 42 (23) days (P = .69) in groups A and B, respectively. Central venous catheters (umbilical, broviac, or peripherally inserted central catheter) were in place for a mean (SD) of 40 (34) days and 35 (22) days for groups A and B, respectively (P = .91).

Primary and secondary outcomes are listed in Table 2 and Table 3. The primary outcome of late-onset infection was 32% (19 of 60) in group A compared with 45% (27 of 60) in group B (P = .13). Infections were predominantly due to gram-positive organisms. There were 53% fewer gram-positive BSIs in group A (15% [9 of 60]) than in group B (32% [19 of 60]; P = .03). The number of Centers for Disease Control and Prevention–defined CLABSIs were similar between groups (1.7 per 1000 central line days in group A and 1.9 per 1000 central line days in group B; P = .88). Possible CLABSIs were reduced by 64% in group A compared with group B, with an incidence of 3.4 vs 9.4 per 1000 central line days, respectively (P = .01). There were no invasive fungal infections in either group. Hand hygiene compliance, assessed monthly during the study period, was 79% (2675 of 3385 observations). The difference in proportion between groups (group A minus group B) with 95% CIs is shown in Figure 2.

Quiz Ref IDThis study demonstrates that nonsterile glove use after hand hygiene prior to patient and line contact is associated with fewer gram-positive BSIs and CLABSIs in this group of extremely preterm infants. Our primary outcome of decreased late-onset infections was lower in group A but did not reach statistical significance. The number needed to treat to prevent 1 gram-positive BSI is 6 and to prevent 1 possible CLABSI is 7.

We studied the effects of glove use after hand hygiene on BSI, urinary tract infection, meningitis, and NEC rates in patients of less than 29 weeks’ gestation and/or weighing less than 1000 g. Our study aimed to reduce more than 1 type of infection, making comparative data difficult because most neonatal infection–related studies focus on 1 type of infection (eg, bloodstream) or patients of different preterm gestational age or birth weight (eg, <1500 g). Data from the Vermont Oxford Network for the years of the study (2008, 2009, 2010, and 2011) indicate that the rates of any late infection (3 or more days of age) in infants born at gestational ages of 22 to 29 weeks were 28.4%, 20.8%, 16.3%, and 13.3%, respectively, in our NICU and were lower compared with those of similar type C Vermont Oxford Network NICUs, in which the average rates were 25.9%, 23.8%, 21.9%, and 20.3%, respectively.²⁸ The intervention of glove use after hand hygiene in infants randomized to group A may account for some of the lower rates of infection seen in our NICU compared with similar NICUs in the Vermont Oxford Network.

Our findings are consistent with studies of different infection control measures to decrease infections in both older pediatric and adult patients.^29-31 Assessment of infection involves microorganism exposure, CFUs, and virulence in relation to host susceptibility. Standard hand hygiene lowers the microbial number on health care professionals’ hands to a few CFUs and potentially eliminates transient flora, but permanent resident flora may persist.^6,8 The underdeveloped skin and mucosal barriers of extremely preterm infants may allow for even a few CFUs to lead to an infection.⁹ In infants born at 26 weeks’ gestation, the stratum corneum of the skin is only 3 layers thick, with a very thin keratin layer, while the skin of full-term infants is 15 cell layers thick with several layers of keratin.³² These characteristics facilitate microbial growth and invasion as the skin is more permeable to microorganisms, more easily injured, and moist owing to a high degree of insensible water loss. Similarly, the gastrointestinal tract is susceptible to infection or microbe translocation because it has a thin mucin layer, few T cells, low amounts of secretory IgA, decreased gastric acid production, and decreased and disorganized motility. We hypothesize that nonsterile glove use after hand hygiene may have been effective by preventing the initial colonization by pathogens of patients and their lines, keeping the CFUs of microorganisms below a threshold that would lead to infection, and decreasing microbes on health care professionals’ hands.

In addition, despite the importance of hand hygiene in health care settings, scientific studies have not established the extent to which counts of microorganisms on health care professionals’ hands need to be reduced to minimize transmission of pathogens.⁸ Most studies observe that the average duration of handwashing by health care professionals is less than 15 seconds, but efficacy data are based on 30 to 60 seconds of handwashing. Antiseptic hand rubs require 2 applications of 3 mL of product for 30 seconds each to be efficient. These methods of evaluating the effectiveness of hand hygiene do not reflect actual use patterns among health care professionals. Evaluation is often with non–health care professionals and may not reflect hand flora in health care settings. For these reasons, hand hygiene followed by glove use may help prevent infections for certain patient populations.

Ng et al¹³ found an association between glove use and a decrease in late-onset BSIs and NEC in a retrospective study of infants who weighed less than 1500 g. They observed that hand hygiene followed by glove use plus the application of a 70% alcohol rub to the gloved hands prior to patient contact decreased rates of late-onset sepsis from 53% to 31% and NEC from 25% to 7% compared with hand hygiene alone. This infection control intervention helped make their late-onset infection and NEC rates consistent with reported rates for infants with a birth weight of less than 1500 g and raised the question of whether glove use in addition to hand hygiene can lower infection rates in other NICUs. We did not observe a difference in NEC rates; potentially this and another study, which used several infection control measures in addition to using contact isolation for NEC cases, prevented cases in which bacteria or viruses were major factors in the multifactorial disease of NEC.^13,33 In addition, our study differed in that we did not apply alcohol hand rub after donning gloves.

Nonsterile glove use after hand hygiene with patient and line care may also benefit other patient groups, and further study of those patients should be considered. A recent study¹⁴ examined the effect of a universal glove use policy during respiratory syncytial virus season on health care–associated infections between 2002 and 2010 in all pediatric units, including NICUs and the general wards. They found that the risk of health care–associated infections was 25% lower during the mandatory glove use periods compared with those months without. Specifically, BSIs and CLABSIs were reduced in the NICU, Pediatric Intensive Care Unit, and Pediatric Bone Marrow Transplant Unit during the universal glove use months.

Quiz Ref IDOur study has some limitations. It is a single-center study and may be affected by local infection control practices. Our sample size was based on a higher incidence of infections than were observed in our control group, so we were underpowered for our primary outcome. We had hoped for more than 90% compliance with hand hygiene. In addition, parents had the choice to use gloves if their child was randomized to group A. There is a possibility that, if hand hygiene, glove use by parents, and study protocol compliance were higher, infection rates may have been lower in 1 or both groups.

Study and hand hygiene compliance are important factors in this trial. Nonsterile glove use after hand hygiene may have served as an observable sign or marker that hand hygiene occurred because gloves are a visible sign, while after standard hand hygiene, there is no noticeable sign that hand hygiene occurred. To control for this possible effect, each bed space has its own mounted alcohol hand rub dispenser, signs for the study intervention were present at each enrolled patient’s bedside in both groups, and hand hygiene rates were audited and similar throughout the study.

This randomized clinical study demonstrated that nonsterile glove use after hand hygiene prior to patient and line contact was associated with fewer gram-positive BSIs and possible CLABSIs in preterm infants. This readily implementable control measure to reduce infections in preterm infants while they have central or peripheral venous access warrants further study in this and other patient populations.

Accepted for Publication: April 30, 2014.

Corresponding Author: David A. Kaufman, MD, Division of Neonatology, Department of Pediatrics, University of Virginia School of Medicine, 3768 OMS Hospital Dr, Charlottesville, VA 22908 (dak4r@virginia.edu).

Published Online: August 11, 2014. doi:10.1001/jamapediatrics.2014.953.

Author Contributions: Dr Kaufman and Ms Blackman had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Kaufman, Blackman, Sinkin.

Acquisition, analysis, or interpretation of data: Kaufman, Blackman, Conaway.

Drafting of the manuscript: Kaufman, Conaway.

Critical revision of the manuscript for important intellectual content: Kaufman, Blackman, Sinkin.

Statistical analysis: Conaway.

Obtained funding: Kaufman.

Administrative, technical, or material support: Kaufman, Blackman.

Study supervision: Kaufman, Sinkin.

Conflict of Interest Disclosures: None reported.

Funding/Support: The study was funded in part by grants from the University of Virginia Institute of Quality and Patient Safety, Cardinal Health Foundation, and University of Virginia Children’s Hospital Grant’s Program (Dr Kaufman).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We acknowledge the neonatal intensive care nurses and other health care professionals at the University of Virginia who were instrumental in conducting this study.