Seasonal variations in healthcare-associated infection in neonates in Canada
- 1Department of Paediatrics, Mount Sinai Hospital, Toronto, Canada
- 2Department of Pediatrics, University of Regina, Saskatoon, Canada
- 3Department of Newborn and Developmental Pediatrics Sunnybrook Health Sciences Center, Toronto, Canada
- Correspondence to Prakesh S Shah, Department of Paediatrics, Rm 775 A, Mount Sinai Hospital, Toronto, Ontario M5G 1X5, Canada;
Contributors Prakesh S Shah initiated the concept, developed the draft, analysed data and wrote the manuscript. Woojin Yoon participated in the design, analysed the data, prepared the results and reviewed the manuscript. Kate Bassil participated in the design, concept, and revision and editing of the manuscript. Michael Dunn participated in the design, interpretation and revision of the manuscript. Zarin Kalapesi participated in the design, and contributed to interpretation of results and editing of the manuscript. Shoo K Lee participated in the design of the study, review of results and manuscript preparation.
- Received 27 October 2011
- Accepted 5 March 2012
- Published Online First 3 May 2012
Objective To assess the seasonal pattern of healthcare-associated infections (HCAI) among neonates and to describe the trend of HCAI.
Design Secondary analyses of database.
Setting The Canadian Neonatal Network database (2003–2009).
Participants Neonates with HCAI defined as blood/cerebrospinal fluid positive with pathogenic organism in a symptomatic infant after 2 days of age.
Main outcome measure The incidence rate for HCAI per 1000 days with a 95% CI, for the 4 warmest months (June–September) was compared with the remaining 8 months, to calculate the incidence rate ratio (IRR).
Results Of 75 629 total infants, 4305 (5.7%) had HCAI (3367 had 1 and 938 had >1 episodes). Infants who had HCAI were of lower gestation, birth weight and Apgar score; but had higher severity of illness scores and clinical chorioamnionitis. There was a borderline increase in all HCAI (IRR 1.05, 95% CI 1.00 to 1.11) and a significant increase in Gram-negative HCAI (IRR 1.20, 95% CI 1.04 to 1.39) during the summer months. Overall, there was a 20% reduction in HCAI from 4.45/1000 days in January 2003 to 3.54/1000 days in December 2009 (mean difference 0.91/1000 days (95% CI 0.89 to 0.92).
Conclusions Gram-negative infections were significantly increased during the summer months of the year compared with the rest of the year among neonates. Overall, there was a significant temporal reduction in HCAI rates over the study period.
What's already known on this topic
Population-based studies in adults have reported higher rates of infection during the summer months.
However, the impact of a warm external climate or other seasonal factors on neonatal infections has not been studied.
What this study adds
There was a borderline increase in overall infection rate and a significant increase in Gram-negative infection rate in the summer months in NICUs.
Overall, there was a decline in healthcare-associated infection in NICUs over 7 years in Canada.
Population-based studies have suggested that infection rates vary throughout the year. Seasonal variation in infections by some Gram-negative organisms, including Eschericia coli,1 ,2 Klebsiella,3 Acinetobacter,4 Aeromonas species,2 ,5 Burkholderia pseudomallei,6 Pseudomonas2 and Enterobacter cloacae2 has been observed, with the highest incidence rates occurring in the summer months. All of these organisms can be acquired through the environmental route and there may be seasonal differences in their colonisation rates. Majority of these studies are of community-acquired infection, except one,2 which is for hospitalised patients. It has been observed that the ‘thermal niche’, defined as the range of temperatures over which the growth of bacteria is at ≥75% of maximal for E coli was 28.5–41°C and for Staphylococcus enterica, it was 27.7–39.8°C.7 These temperature ranges are similar to those consistently achieved in the summer months (June–October) in Canada. An ecological study in the Lake Superior watershed in Northern Minnesota revealed a higher density of E coli in the summer months.8 Several recreational facilities and beaches are often closed during summer due to high coliform counts in the water.
The current literature regarding seasonal variations in bloodstream infection rates stems mainly from adult patients in localised population settings.1 ,3,–,6 One study from Pediatric Intensive Care Units in the US reported no seasonal variation in device-associated nosocomial infection rates.9 No reports have been published so far exploring seasonal variation in HCAI among neonates. Neonates admitted to neonatal intensive care units (NICUs) in Canada represent a unique population in which to study whether or not such a pattern exists. In particular, there is a perception that there should not be any difference in the rates of infections among these neonates as temperature is stringently controlled in the NICUs. The environmental temperature in the NICU and isolette temperature in the incubators are rigorously controlled. We therefore hypothesised that there will be no difference in the incidence rates of infections in neonates admitted to NICUs between the summer months and the rest of the year, because of the climate control that is utilised in modern settings. The information gained could be very useful for NICUs in helping them determine whether they need to enhance or strengthen the resources allocated to infection control measures during certain months as it has been shown repeatedly that HCAI are associated with increased risk of adverse outcomes in neonates..10,–,12 Our primary objective in this study was to compare incidence rates of healthcare-associated infections (HCAI) among infants admitted to NICUs during the summer months with those admitted during the rest of the year. Our secondary objective was to describe the trend of HCAI in Canadian NICUs over a 7-year period.
This study was conducted using data on all neonates admitted to level three NICUs in Canada that were members of the Canadian Neonatal Network (CNN) from January 2003 to December 2009. This was a retrospective secondary analysis of data collected routinely on all admissions to participating NICUs. These centres included 20 perinatal centres and five children's hospital NICUs.
All neonates admitted within the first 3 days of birth were included. This was mandated to ensure that all episodes of HCAI were captured prior to discharge form NICUs.
We excluded neonates who were admitted after 3 days of age and those who were moribund on admission when a physician, in consultation with the parents, had made an explicit decision not to provide life support at the time of the admission. Cases of early-onset sepsis (within the 2 days of birth) were excluded.
Healthcare-associated infection was defined as isolation of a pathogenic organism in a blood culture or cerebrospinal fluid (CSF) culture in a symptomatic infant using criteria similar to Freeman et al.13 There was no mandatory requirement for two distinct blood cultures or lumbar puncture for diagnosis of infection. We excluded Corynebacterium species, Propionibacterium species and unidentified organisms, including reports of Gram-positive or Gram-negative bacteria whose species was not identified due to possibility of contaminated samples. The cases included both catheter-associated and non-associated infections. When infants had more than one episode of infection, an episode occurring >14 days after the previous one was considered a separate episode, and two or more episodes occurring <14 days apart either by the same type or a different type of organism were considered as one episode. Patients in whom multiple organisms were identified, their data were individually assessed by authors to distinguish contaminants from pathogenic organisms. If a neonate was infected with more than one pathogenic organism during a single episode, it was treated as one episode of infection. Presence of same or different organisms in both blood and CSF at the same time point was considered as one episode.
Data on baseline demographics and the results of blood cultures were abstracted from the dataset. The Score of Neonatal Acute Physiology version II (SNAP II) was calculated to assess severity of illness on admission for all infants.14 Diagnosis of clinical chorioamnionitis was based on clinical criteria (maternal fever, foul-smelling amniotic fluid with or without placental pathology). We did not have data on the temperatures maintained in the NICUs during the study period as there is no standardised guideline; however, we believe that it was kept constant and did not fluctuate throughout the year.
The incidence rate, expressed as the number of new cases of HCAI per 1000 patient days for each month was calculated with the assumption that the entire population of admitted neonates were at risk of contracting HCAI. For each neonate, the number of patient days for a particular month was calculated based on the date of admission and the date of discharge. Total patient days for a particular month were calculated based on the total number of days any infant stayed in the NICU in that particular month. Incidence of infection (diagnosed based on the date when blood/CSF culture was withdrawn) was calculated for each month using patient days in a particular month as the denominator and number of infections diagnosed during that month as the numerator. To enable comparisons between different months, the infection rate was converted to infection rate per 1000 patient days. A 95% CI for each incidence rate was estimated using a Poisson distribution.
To evaluate differences in the infection rates between the summer months and the rest of the year, the incidence rate for the 4 warmest months (June 1 through September 30) and the incidence rate for the 8 remaining months (October 1–May 31) were compared. The incidence rate ratio (IRR) for the 4 warmest months relative to the IRR for the remaining 8 months, and its 95% CI were calculated. The IRR were calculated for overall HCAI and then for Gram-positive and Gram-negative HCAI separately. All analyses were performed using the SAS statistical software package (version 9.2, SAS Institute, Cary, North Carolina, USA).
The Canadian Neonatal Network Database collection was approved by either local Research Ethics Board or appropriate quality improvement committee in each centre.
From a total of 25 participating centres, 75 629 neonates were included in the database of the CNN, of which 71 059 met the eligibility criteria for this study. Of these neonates, 4305 (6.0%) developed late-onset sepsis. These included 4194 (97.4%) infants who had only sepsis, 52 (1.2%) had sepsis and meningitis and 59 (1.4%) had meningitis alone. Overall, 3367 infants had one episode of infection (4.7%) and 938 (1.3%) infants had multiple episodes (2271 episodes; figure 1). Baseline characteristics of patients who developed sepsis and those who did not are reported in table 1. Infants who developed infection were younger, of lower birth weight, more likely to have been born to mothers with clinical chorioamnionitis and had higher severity of illness on admission. There was no difference in the rates of clinical chorioamnionitis (6.9% vs 6.7%; p=0.36), number of infants admitted to NICUs (admission rates 862/month vs 838/month; p=0.58) or number of infants who were low birth weight (20.7% vs 21.3%; p=0.09) between summer months and the rest of the year, respectively. Organisms responsible for these infections are outlined in Appendix 1.
Individual monthly rates of infection/1000 patient days reveal fluctuations in the rate over years and months (table 2). Comparison of the rates of infection on a year by year basis from 2003 to 2009 revealed a significant reduction over the 7 years from 4.45/1000 (95% CI 4.08 to 4.85) patient days to 3.54/1000 (95% CI 3.32 to 3.78) patient days. The mean difference between 2003 and 2009 was 0.91/1000 days, 95% CI 0.89 to 0.92. (p<0.001 by Cochran-Armitage test, figure 2). Comparison of infection rates (table 3) during the summer months against the rest of the year revealed borderline significance for an overall increase in HCAI (IRR 1.05, 95% CI 1.00 to 1.11). However, Gram-negative organisms were significantly increased during the summer months compared with the rest of the year (IRR 1.20, 95% CI 1.04 to 1.39). When only the first episode of infection was considered in the analysis, the results remained similar (IRR 1.04, 95% CI 0.98 to 1.11) for all infection, IRR for Gram-negative infections 1.19, 95% CI 1.00 to 1.42) and IRR for Gram-positive infection was 1.01, 95% CI 0.94 to 1.09). Similar to overall trend, higher IRR were observed for summer in 6 out of 7 years in the study; however, individually IRR were not statistically significant. In a post hoc analysis, we fitted an ARIMA (autoregressive integrated moving average) model with 84 infection rates from 84 consecutive months over the study period. The autocorrelation function in the model had the pattern characteristic of a first order moving average process for general time trend (lag 1) and a seasonal moving average process (lag 12). The resulting ARIMA (0,1,1)×(0,1,1)12 model estimated the moving average parameters for general time trend to be 0.75 (95% CI 0.59 to 0.91) and seasonal trend to be 0.53 (95% CI 0.31 to 0.75). This indicates that similar pattern of seasonality is repeated on year-to-year basis.
In this large, national sample of infants representing 90% of all infants admitted to NICUs Canada over the study period, we identified that there was a significant increase in Gram-negative infections during the summer months compared with the rest of the year. Overall, over the 7 year duration of the study, there was a significant reduction in HCAI rates in Canadian NICUs.
No previous study has evaluated the association between season and rates of HCAI in children or neonates, and thus we could not directly compare our data. However, population-based local cohort studies that have been conducted in adults have reported higher rates of various organisms during the summer months.1,–,6 Plausible, although highly speculative, explanations for this phenomenon were presented in these adult populations. From these, we can make purely speculative attempts to identify the reasons underlying this phenomenon in our population.
Probability of ‘thermal niche’ concept8 is an unlikely explanatory argument as temperature is very strictly controlled in most of the nurseries via set parameters. In addition, the majority of the neonates are cared for in isolettes, where even stricter temperature regulation is in effect. Another theory is that relative changes in the humidity of the external environment could increase colonisation of bacteria in the nursery. High humidity, particularly when it is associated with heat, can create a receptive environment for Gram-negative organisms to stabilise and grow.15 Another possible explanation could be that, during the summer months, some units may get busier than during the rest of the year, as a result of which routine practices to prevent HCAI (hand hygiene, use of antiseptics prior to contact of ‘patient area’ and protective gear use) may become suboptimal. Another explanation could be transmission from adults (healthcare providers or family members) to neonates. There could be ‘thermal effect’ on maternal colonisation with Gram-negative organisms. Could this be an effect of new trainees starting in July in majority of units? It is possible that new trainees are less aware of stringent aseptic precautions. Could this be due to reduced number of staffing (support services) and overburdened remaining staff with lapses in infection control? Could this be due to relaxed attitude of caregivers in the summer or whether they were stressed because of influx of new trainees or reduced staffing? These are all possibilities; however, they need to be tested.
The significant overall reduction in HCAI in Canadian NICUs over the past 7 years is an important finding. It reflects an overall co-ordinated effort to identify, recognise and implement care practices to combat infections in the NICUs. In particular, in Canada, a concerted effort to combat infection and chronic lung disease was undertaken between 2003 and 2005, which resulted in a significant reduction in HCAI.16 The demonstration that the infection rate is higher in the summer months is also very crucial. Although we do not have any concrete explanation for this phenomenon, it highlights a need for greater vigilance during these months in NICUs. Until we identify and rectify the reason(s) underlying the higher rates of infection in the summer months, care providers in NICUs should be made aware of this association and must take additional routine precautions during these months. Additionally, all future before/after studies of HCAI should adjust for season or average temperature before concluding the effectiveness of a change in practice.
Strengths of our study include the large cohort, the near population-based nature of the sample, careful data collection and data from a high-risk population. Limitations of our study include the absence of data on practices during this period regarding antibiotic use, central catheter use, nutritional practices and the results of audits of hand hygiene or other practices. However, we do not have any reason to believe that these practices would be different during the summer months as opposed to the rest of the year. We evaluated whether these differences could be explained by variation between centres and identified that although there was some variability, no individual centre had significantly higher rates than other centres. We also evaluated whether the increase in HCAI during the summer months could be explained by higher rates of clinical chorioamnionitis, higher severity of illness or higher numbers of small and immature infants during the summer months as opposed to the rest of the year, and failed to identify such differences. It must be recognised that these data only reflect neonates admitted to NICUs and therefore do not include all neonatal infections in the country. Finally, we did not have data on the actual temperature in the city on the day of infection to correlate it directly with rates of infection.
These results open up a significant opportunity for researchers to identify reasons for higher rates of infection with Gram-negative bacteria in the summer. Our result corroborates observations from studies in adult population, and identifies an area in which improvement can be made in the outcomes of tiny vulnerable infants.
The authors thank Sarah De La Rue from the Maternal-Infant Care Research Center, Mount Sinai Hospital, Toronto, for editorial help with this manuscript. The Maternal-Infant Care Research Center is supported by the Ministry of Health and Long-term Care, Ontario, Canada.
Competing interests None.
Ethics approval Research Ethics Board or Quality improvement committee at each hospital.
Provenance and peer review Not commissioned; externally peer reviewed.