Preoperative and operation-related risk factors for postoperative nosocomial infections in pediatric patients: A retrospective cohort study.

BACKGROUND: Pediatric patients undergoing invasive operations bear extra risk of developing nosocomial infections (NIs). However, epidemiological evidence of the underlying risk factors, which is needed for early prevention, remains limited.
METHODS: Using data from the electronic medical records and the NI reporting system of a tertiary pediatric hospital, we conducted a retrospective analysis to identify preoperative and operation-related risk factors for postoperative NIs. Multivariable accelerated failure time models were fitted to select independent risk factors. The performance of these factors in risk stratification was examined by comparing the empirical risks between the model-defined low- and high-risk groups.
RESULTS: A total of 18,314 children undergoing invasive operations were included for analysis. After a follow-up period of 154,700 patient-days, 847 postoperative NIs were diagnosed. The highest postoperative NI rate was observed for operations on hemic and lymphatic system. Surgical site infections were the NI type showing the highest overall risk; however, patients were more likely to develop urinary tract infections in the first postoperative week. Older age, higher weight-for-height z-score, longer preoperative ICU stay, preoperative enteral nutrition, same-day antibiotic prophylaxis, and higher hemoglobin level were associated with delayed occurrence of postoperative NIs, while longer preoperative hospitalization, longer operative duration, and higher American Society of Anesthesiologists score showed acceleration effects. Risk stratification based on these factors in an independent patient population was moderate, resulting in a high-risk group in which 72% of the postoperative NIs were included.
CONCLUSIONS: Our findings suggest that pediatric patients undergoing invasive operations and at high risk of developing postoperative NIs are likely to be identified using basic preoperative and operation-related risk factors, which together might lead to moderately accurate risk stratification but still provide valuable information to guide early and judicious prevention.

Introduction

Nosocomial infections (NIs) pose a long-standing challenge to clinical practitioners and remain one of the leading causes of in-hospital mortality [1]. The overall prevalence of NIs varies from 7% in affluent countries to 15% in economically developing countries [2, 3], whereas the most common types of NIs are invariably surgical site infections (SSIs) and device-associated infections [2, 4, 5], suggesting that most of the NIs are attributable to invasive operations.

Pediatric patients undergoing invasive operations face extra risk of developing NIs because of their underdeveloped immune system. According to two European studies, the NI incidence was 2.5% in general pediatric wards and was 17% in surgical wards [6, 7]. From a prevention perspective, by recognizing the preoperative and operation-related factors that enhance patients’ susceptibility to postoperative NIs, healthcare providers would be able to identify vulnerable patients for closer observation and to initiate timely prophylactic treatment when necessary. For adult surgical patients, several epidemiological studies have revealed a wide variety of such factors [8-10], but detailed research in pediatric patients is scarce.

In the present study, we focused on pediatric patients undergoing invasive operations and aimed to identify preoperative and operation-related risk factors for the occurrence of postoperative NIs.

Methods

Study setting and design

The present study was a hospital-based retrospective cohort study conducted in a tertiary referral hospital—Guangzhou Women and Children’s Medical Center—in Guangzhou, China. The data source of this study was the electronic medical records (EMRs) of the pediatric inpatients who underwent invasive procedures between 2016 and 2018.

Extraction of data and ascertainment of study outcomes

Clinical data were derived from the EMRs and then linked to the hospital’s NI reporting system via patient identification numbers. In the EMRs, operative procedures were documented in both text and the procedural codes defined by the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM). According to the criteria of the NI reporting system, an infection was considered nosocomial if it occurred > 48 hours after admission. Neonatal infections were considered nosocomial as well if they were acquired during delivery. Diagnoses of specific NI types were made following the criteria from the Centers for Disease Control and Prevention/National Healthcare Safety Network [11]. The outcome of interest in the present study was NI diagnosed after invasive operation. For patients with multiple invasive operations and/or multiple postoperative NI episodes, only the first invasive operation and the first postoperative NI were analyzed.

The patient cohort

In the EMR database, we identified 18,314 patients who underwent invasive operations between 2016 and 2018 on one of the following specific systems: the nervous system (ICD-9-CM code: 01–05), respiratory system (ICD-9-CM code: 30–34), cardiovascular system (ICD-9-CM code: 35–39), hemic and lymphatic system (ICD-9-CM code: 40–41), digestive system (ICD-9-CM code: 42–54), urinary system (ICD-9-CM code: 55–59), and musculoskeletal system (ICD-9-CM code: 76–84). The main reason for us to focus on these systems was that the invasive operations performed on these systems except those on the hemic and lymphatic system were the most common ones in our institution. Invasive operations on the hemic and lymphatic system were relatively rare but were included because of the high postoperative NI rate after hemic and lymphatic surgeries.

Ethical considerations

This study was approved by the ethics committee of the Guangzhou Women and Children’s Medical Center (2019–13600). This study utilized historical data collected during routine clinical practice and the findings of this study will be used only for academic activities; therefore requirement for informed consent from patients was waived.

Statistical analysis

The following candidate risk factors were considered: sex; age, nutritional status measured with weight-for-age z-score (WAZ), and blood test result at admission; lengths of preoperative hospitalization and intensive care unit (ICU) stay; preoperative enteral nutrition (EN) and parenteral nutrition (PN) support; operative duration; surgical implantation; antibiotic prophylaxis; the American Society of Anesthesiologists (ASA) score; surgical wound classification (SWC); and ICD-9-CM code. For patients younger than 10 years of age, WAZ was calculated using the WHO growth standards as the reference. For patients above that age, WAZ is not an appropriate measure of nutritional status and thus was treated as missing. Antibiotic prophylaxis was defined as use of antibiotics on the same day as the operation was performed. In order to retain the patients with missing values in analysis, binary indicators were created to denote data incompleteness, which was the case for WAZ, operative duration, and ASA score. Binary indicators were also created for preoperative ICU, blood test at admission, and surgical incision to distinguish between patients who did and who did not receive these procedures (see Table A in S1 File for a detailed description).

Time to event was defined as the duration from date of operation to date of NI diagnosis or date of discharge, whichever came first. After confirming that the log-transformed time to event was approximately normally distributed, accelerated failure time (AFT) models with log-normal distribution were fitted. In log-normal AFT models, the exponentiated coefficient (e) of a factor is interpreted as time ratio (TR) indicating whether this factor would decelerate (e >1) or accelerate (e <1) the occurrence of the event. The TR estimates can be converted into the number of days by which the time between operation and occurrence of postoperative NIs could be prolonged or shortened using the following formula: ΔT = T {exp[β(x-x)] −1}, where T denotes the time between the operation and the occurrence of postoperative NI for a reference patient, and X denotes the risk factor profile of the reference patient.

A reduced model was achieved using backward elimination with a threshold P-value of 0.1. Indicator variable and the variables related to it were handled as a block during variable selection; therefore we chose a less stringent P-value to decrease the possibility of excluding blocks that contain statistically significant risk factors. In order to control for the clinical heterogeneity of the operations, we predetermined that ICD-9-CM code should be included in the models regardless of the result of variable selection. We also built multivariable Cox regression models to estimate the hazard ratios (HRs), which are more commonly used to describe the strengths of risk associations.

In order to examine the clinical applicability of the identified risk factors for risk stratification purpose, we derived a postoperative NI risk score for another data set of 4,383 pediatric patients who underwent invasive operations between January 2019 and May 2019, of them 110 NIs were diagnosed postoperatively. The postoperative NI risk score was calculated by summing up the risk factors weighted by the AFT model coefficients and used the median of this risk score as a cutoff to divide the patients into low- and high-risk groups (Risk score calculation in S1 File). We compared the two groups regarding their empirical postoperative NI risks, which were estimated using the Nelson-Aalen method [12].

All the statistical tests were two-sided with P < 0.05 considered statistically significant. All the statistical analyses were performed using the “survival” package in R (R Foundation for Statistical Computing, Vienna, Austria).

Results

Baseline characteristics of the cohort, stratified by the subsequent NI status, are shown in Table 1. Compared with the patients who did not develop postoperative NIs, those who did were relatively younger, had a lower WAZ and longer preoperative hospitalization, and were more likely to have preoperative ICU stay and to receive preoperative nutrition support. The NI group had on average a lower hemoglobin level. Regarding operation-related characteristics, patients in the NI group had an averagely longer operative duration and were more likely to receive surgical implantation and to have higher ASA scores.

Table 1

Baseline characteristics of a retrospective cohort of pediatric patients who underwent invasive operations (n = 18,314), stratified by postoperative NI status, the Guangzhou Women and Children’s Medical Center, 2016–2018.

	Patients withoutpostoperative NI (n = 17,467)	Patients withpostoperative NI(n = 847)
Sex, male (%)	12,208 (64.9)	559 (63.9)
Age in months, median (IQR)	23 (6–59)	16 (3–47)
WAZ, indicator: yes (%)	16,292 (93.3)	795 (93.9)
WAZ median (IQR) a	−0.64 (−1.57–0.11)	−0.93 (−2.02–−0.03)
Preoperative hospitalization days, median (IQR)	2 (1–5)	3 (1–6)
Preoperative ICU stay, indicator: yes (%)	6,559 (37.6)	382 (45.1)
Preoperative ICU days, median (IQR) a	2 (1–4)	2 (1–6)
Preoperative EN, yes (%)	4,233 (24.2)	237 (28.0)
Preoperative PN, yes (%)	1,576 (9.0)	118 (13.9)
Antibiotic prophylaxis, yes (%)	11,400 (65.3)	521 (61.5)
Preoperative blood tests, indicator: yes (%)	15,039 (86.1)	765 (90.3)
Hemoglobin (g/L), median (IQR) a	117 (104–127)	109 (94–124)
WBC (109/L), median (IQR) a	9.5 (7.1–12.4)	9.2 (6.2–13.3)
Surgical implantation, yes (%)	2,926 (16.8)	175 (20.7)
Operative duration, indicator: yes (%)	14,481 (82.9)	515 (60.8)
Operative duration (min), median (IQR) a	130 (80–190)	175 (100–260)
ASA score, indicator: yes (%)	14,728 (84.3)	597 (70.5)
ASA score I, n (%) a	3,277 (22.2)	75 (12.6)
ASA score II, n (%) a	8,896 (60.4)	280 (46.9)
ASA score ≥III, n (%) a	2,555 (17.4)	242 (40.5)
SWC, indicator: yes (%)	8,242 (47.2)	310 (36.6)
Clean (%) a	6,302 (76.5)	229 (73.9)
Clean-contaminated (%) a	1,271 (15.4)	57 (18.4)
Contaminated (%) a	669 (8.1)	24 (7.7)

aLimited to the patients for whom the data were available or applicable. ASA (American Society of Anesthesiologists), CVC (central venous catheterization), EN (enteral nutrition), ICU (intensive care unit), IQR (interquartile range), NI (nosocomial infection), PN (parenteral nutritionSD standard deviation), UC (urinary catheterization), WAZ(weight-for-age z-score), WBC (white blood cell).

During a follow-up of 154,700 patient-days, 847 postoperative NIs were diagnosed, yielding an incidence rate of 5.5 per 1000 PDs. Overall, operations on hemic and lymphatic system were followed by the highest NI rate (11.0 per 1,000 PDs), and SSIs were the most common NI type (Table 2). However, when confined to the first postoperative week, the highest NI risk was observed for cardiovascular operations (Fig 1A), and the NI type with the highest risk was urinary tract infections (UTIs, Fig 1B).

Table 2

Incidence of major NIs in a retrospective pediatric patient cohort (n = 18,314) after invasive operations, by operative site (ICD-9-CM) and by NI type, the Guangzhou Women and Children’s Medical Center, 2016–2018.

	Patients	NIs (rate, per1,000 PDs)	Specific NI types
			SSI	UTI	URI	LRIa	GI	BSI	Others	Unknown
Total	18,314	847 (5.5)	182	148	122	115	78	63	83	56
By operative site (ICD-9-CM)
Nervous system (01–05)	2,353	182 (7.6)	81	15	12	15	5	10	33	11
Respiratory system (30–34)	1,875	74 (4.4)	13	4	24	22	6	3	2	0
Cardiovascular system (35–39)	3,233	209 (6.4)	20	55	23	45	20	10	11	25
Hemic and lymphatic system (40–41)	466	81 (11.0)	8	2	19	10	6	11	18	7
Digestive system (42–54)	5,497	171 (3.7)	48	16	21	14	21	27	14	10
Urinary system (55–59)	2,839	105 (5.6)	5	52	20	3	18	1	3	3
Musculoskeletal system (76–84)	2,051	25 (2.7)	7	4	3	6	2	1	2	0

aIncluding ventilator-associated pneumonia. BSI bloodstream infection. ICD-9-CM (International Classification of Diseases, 9th Revision, Clinical Modification), GI (gastrointestinal infection), LRI (lower respiratory tract infection), NI (nosocomial infection), PD (patient-day), SSI (surgical site infection), URI (upper respiratory tract infection), UTI (urinary tract infection).

Fig 1

Risk of developing NIs in the first postoperative week, stratified by operative site (plot A) and by infection site (plot B). BSI (bloodstream infection), GI (gastrointestinal infection), LRI (lower respiratory tract infection), NI (nosocomial infection), SSI (surgical site infection), URI (upper respiratory tract infection, (UTI) urinary tract infection.

In the full multivariable AFT model, older age, higher WAZ, longer preoperative ICU stay, preoperative EN, antibiotic prophylaxis, and higher hemoglobin level were statistically significantly associated with delayed occurrence of postoperative NIs (Table 3). Longer preoperative hospitalization, blood test at admission, and longer operative duration were statistically significantly associated with advanced occurrence of the events. Backward elimination led to a reduced model including all the prior statistically significant risk factors as well as high ASA score (≥ III). According to the reduced AFT model and given a reference patient who was defined as follows: underwent an invasive operation on the hemic and lymphatic system, age = 12 months, WAZ = −3, days of preoperative hospitalization = 7, preoperative ICU = 0, preoperative EN = 0, preoperative antibiotic prophylaxis = 0, hemoglobin = 100g/L, WBC = 10×109/L, operative duration = 120 minutes, and ASA score = II, the estimated time between invasive operation and the occurrence of postoperative NIs was 56 days. Therefore, antibiotic prophylaxis = 1, preoperative EN = 1, and one year older in age would extend this time by approximately 24, 14, and 4 days, respectively, for otherwise comparable patients; whereas preoperative hospital stay for one more week, one-point increase in ASA score, and one-hour increase in operation duration would advance the occurrence of the postoperative NIs by 11, 10 and 5 days, respectively.

Table 3

Associations between baseline factors and the development of postoperative NIs in multivariable AFT log-normal models, the Guangzhou Women and Children’s Medical Center, 2016–2018.

	Full modela			Reduced modela
	TR	95% CI	P	TR	95% CI	P
Sex, male vs. female	1.07	0.94–1.21	0.29
Age, per year increase	1.07	1.04–1.10	< 0.01	1.07	1.04–1.10	< 0.01
WAZ
Indicator, yes (WAZ = 0) vs.NA/missing	1.31	0.91–1.89	0.14	1.31	0.91–1.88	0.14
Per unit increase	1.05	1.00–1.09	0.05	1.05	1.00–1.09	0.04
Preoperative hospitalization
Per day increase	0.97	0.96–0.98	< 0.01	0.97	0.96–0.98	< 0.01
Preoperative ICU stay
Indicator, yes (days = 1) vs. no	1.03	0.89–1.18	0.69	1.04	0.90–1.20	0.59
Per day increase	1.03	1.01–1.04	< 0.01	1.03	1.01–1.04	< 0.01
Preoperative EN, yes vs. no	1.22	1.04–1.43	0.01	1.24	1.06–1.45	0.01
Preoperative PN, yes vs. no	1.08	0.89–1.31	0.45
Antibiotic prophylaxis
Yes vs. no	1.43	1.23–1.66	< 0.01	1.42	1.23–1.65	<0.01
Preoperative blood test
Indicator, yes (hemoglobin = 100,WBC = 10) vs. no	0.82	0.67–1.00	0.04	0.82	0.68–1.00	0.05
Hemoglobin, per 5 g/L increase	1.03	1.02–1.05	< 0.01	1.03	1.02–1.05	< 0.01
WBC, per 5×109/L increase	0.99	0.97–1.00	0.10	0.99	0.97–1.00	0.10
Surgical implantation, yes vs. no	0.89	0.74–1.07	0.22
Operative duration
Indicator, yes (operative duration= 1 hour) vs.missing	2.01	1.50–2.69	< 0.01	2.01	1.51–2.69	< 0.01
Per hour increase	0.91	0.87–0.95	< 0.01	0.90	0.86–0.94	< 0.01
ASA score
Indicator, yes (ASA score = I) vs.missing	0.67	0.47–0.95	0.02	0.67	0.47–0.96	0.03
ASA score II vs. I	0.98	0.79–1.20	0.81	0.98	0.79–1.20	0.82
ASA score ≥III vs. I	0.79	0.63–1.00	0.06	0.79	0.62–1.00	0.05
SWC
Indicator, yes (SWC = clean) vs.no	1.09	0.93–1.27	0.30
Clean-contaminated vs. clean	0.96	0.74–1.24	0.75
Contaminated vs. clean	0.87	0.59–1.30	0.50

aBoth models were adjusted for operative site (ICD-9-CM code). AFT (accelerated failure time model), ASA (American Society of Anesthesiologists), EN (enteral nutrition), ICD-9-CM (International Classification of Diseases, 9th Revision, Clinical Modification), ICU (intensive care unit), NA (not applicable), PN (parenteral nutrition), SWC (surgical wound classification), TR (time ratio), WAZ (weight-for-age z-score), WBC (white blood cell).

The multivariable Cox model (Table B in S1 File) yielded largely consistent results except for the statistically significant positive association for WBC. In the reduced Cox model, the proportional hazards assumption was not met for age at operation, antibiotic prophylaxis, and operative duration.

Risk stratification based on the reduced AFT model was applied to the patients who underwent invasive operations between January 2019 and May 2019 (n = 4,383). Fig 2 shows the empirical postoperative NI risks in the resulting low- and high-risk groups. Within the first 30 days after operation, the high-risk group (2,191 patients including 79 NIs) showed a consistently higher postoperative NI risk than the low-risk group (2,192 patients including 31 NIs). The 30-day postoperative NI rates were 3.1 and 5.4 per 1,000 PDs for the two groups, respectively, and 72% of the postoperative NIs occurred in the high-risk group.

Fig 2

Postoperative NI risk in the low-risk group (2,192 patients including 21 NIs) and the high-risk group (2,191 patients including 79 NIs), stratified using the median of the risk score derived from the reduced AFT modela.

aAnalysis was done in an independent data set of patients undergoing invasive operation between January and May, 2019 (n = 4,383). NI (nosocomial infection).

Discussion

In this retrospective cohort of pediatric inpatients undergoing invasive operations, older age, higher WAZ, longer ICU stay, preoperative EN support, antibiotic prophylaxis, and higher hemoglobin level were associated with delayed occurrence of postoperative NIs, while longer preoperative hospitalization, longer operative duration, and higher ASA score might accelerate the occurrence of the postoperative NIs. Risk stratification based on these factors in the same cohort was moderately accurate.

In this study, NI rate following hemic and lymphatic operations was the highest, probably due to immunodeficiency commonly seen in patients with hematologic diseases. In two studies including one in pediatric surgical patients, the most frequent NI types were SSI [9, 13], -consistent with our results. In the present study, however, the risk of developing UTIs within the first postoperative week was higher than the risks of developing other types of NIs, suggesting that the pathogens causing UTIs might have a relatively short incubation period.

Malnutrition increases the risk of infection by impairing the immune system [14, 15], accounting for the inverse association between WAZ and postoperative NI risk in this study. Increased NI risk in relation to malnutrition has been reported in prior studies, where malnutrition was defined using different measures such as low birth weight and low body mass index [16, 17]. As a biomarker of malnutrition, low hemoglobin level may compromise non-specific immunity and increase susceptibility to infection [18], which explains its inverse association with postoperative NI risk in our cohort.

For the reference patient we defined, we found that one-day increase in ICU stay before operation was associated with a delay of approximately 2 days in the occurrence of postoperative NIs: this was likely to result from the benefits of intensive care, such as close observation and nutrition support. Previous studies reported a positive association between length of ICU stay and NI risk among ICU patients [19, 20], which however was not surprising given the fact that the risk of developing NIs in ICU always accrues as the ICU stay extends. Furthermore, those studies were flawed by reverse causality (i.e. occurrence of NIs in ICU extends ICU stay) and their problematic use of logistic regression to analyze time-to-event data.

In our cohort, longer preoperative hospitalization was associated with an increased postoperative NI risk: one-week longer preoperative hospital stay would accelerate the occurrence of postoperative NIs by 11 days for the reference patient, in line with a previous study in adult surgical patients [21]. Length of preoperative hospitalization is largely a proxy of the severity and complexity of the underlying disease. Moreover, prolonged preoperative hospitalization means extended exposure to pathogens and enhanced possibility of developing hospital-acquired malnutrition [22, 23]: both increase patients’ susceptibility to infections.

Current evidence regarding the association between EN and NI risk is inconclusive [24, 25]. Thus the inverse association between preoperative EN and postoperative NI risk in our cohort warrants confirmation by future studies. EN may cause NIs via contaminated feedings and/or tracheal colonization of gastric organisms; however, it may also reduce infectious complications by maintaining gut structure [26]. It has been well proved that PN increases the risk of BSI because of CVC involvement [27, 28]. In the present study, only a small fraction of PN (<2%) was administered via CVC, which might partly explain the low incidence of BSIs in the present study as well as the null association between PN and the overall NI risk.

Reduced SSI rates due to antibiotic prophylaxis has been reported [29], supporting our finding that antibiotic prophylaxis might delay the occurrence of postoperative NIs by more than 20 days. However, increasing evidence also suggests that antibiotic prophylaxis will cause unnecessary side effects if it is given without indication [30]. In our cohort, antibiotic prophylaxis was common, but lack of detailed data did not allow us to define antibiotic prophylaxis following the established protocols [31, 32] or to determine its necessity and appropriateness in terms of type, timing, and dosage.

The present study supports the finding of a prior study that higher ASA scores might raise the risks of specific NI types including SSI, UTI, and nosocomial pneumonia [9]. Our results also confirm the existing evidence that prolonged operative duration might increase the risks of postoperative SSIs and other complications [33]. As for SWC, the present study only had a low number of patients with contaminated wounds and had none with dirty wounds. It has also been suggested that surgical wounds might be substantially misclassified in clinical practice [34], making it even more difficult to compare our result with those from previous studies, where a positive association between SWC and the overall postoperative NI risk might exist given the increased SSI risk among patients with contaminated or dirty wounds [35, 36].

Epidemiologic studies on postoperative NI incidence and its risk factors are inadequate, especially for pediatric patients; therefore, our findings would enrich the existing knowledge and would inform healthcare providers of a patient’s risk of developing postoperative NIs even before invasive operations. As suggested by our results, a risk score based on basic preoperative and operation-related risk factors might lead to moderately accurate risk stratification. However, using such a risk score would be rewarded with early identification of vulnerable patients, timely prevention, and cautious selection of the postoperative treatment or procedures that might further increase the NI risk.

The strengths of the present study include its relatively large sample size, a variety of possible risk factors to be considered, and proper statistical analysis. From a methodological perspective, despite the largely consistent results, the AFT model was superior to the Cox model in our case, given that the proportional hazards hypothesis could be legitimately violated: for example, the effect of antibiotic prophylaxis will diminish rather than remaining constant over time. Several limitations of the present study should be noted. First, like many other hospital-based studies, this study was subject to selection bias in particular the referral bias, which might limit the generalizability of our results. Second, we were concerned that using ICD-9-CM code at its lowest level of specificity might not be sufficient to control for the invasiveness of the operation and the severity of the underlying disease. Third, several putative risk factors for NIs were either precluded from our analysis or could not be studied thoroughly due to lack of data, such as use of antibiotics and immunosuppressants. Finally, we could not include the NIs that occurred after discharge as no systematic post-hospital follow-up was performed to collect the data.

Conclusions

Our findings suggest that pediatric patients undergoing invasive operations and at high risk of developing postoperative NIs are likely to be identified using basic preoperative and operation-related risk factors, which together might lead to moderately accurate risk stratification but still provide valuable information to guide early and judicious prevention.

(DOC)

Click here for additional data file.

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

21 Aug 2019

PONE-D-19-18261

Preoperative and operation-related risk factors for postoperative nosocomial infections in pediatric patients: a retrospective cohort study

PLOS ONE

Dear Dr. Liang,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by Oct 05 2019 11:59PM. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Agnieszka Rynda-Apple, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please note that all PLOS journals ask authors to adhere to our policies for sharing of data and materials: https://journals.plos.org/plosone/s/data-availability. According to PLOS ONE’s Data Availability policy, we require that the minimal dataset underlying results reported in the submission must be made immediately and freely available at the time of publication. As such, please remove any instances of 'unpublished data' or 'data not shown' in your manuscript and replace these with either the relevant data (in the form of additional figures, tables or descriptive text, as appropriate), a citation to where the data can be found, or remove altogether any statements supported by data not presented in the manuscript.

3. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

4. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information

Additional Editor Comments:

Dear Dr. Liang,

We have completed our review of your manuscript PONE-D-19-18261 entitled "Preoperative and operation-related risk factors for postoperative nosocomial infections in pediatric patients: a retrospective cohort study". While our review indicated that the manuscript is not suitable for publication in PlosONE in its current state, the work itself seems to be sound. With the necessary revisions, this manuscript could be reconsidered. The comments of the reviewers are included at the bottom of this letter. While responding to the comments and criticisms voiced by the reviewers please provide more detailed description of statistical models and analysis used.

Per PlosONE policy the data underlying the findings of the manuscript needs to be fully available; either included as supplementary information or deposited to the public repository. If there are any restrictions on publicly sharing data, those need to be specified.

Yours sincerely,

Agnieszka Rynda-Apple

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: I Don't Know

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The paper is well written, methods appear appropriate, and the results provide useful information on risk factors for post-operative infection following paediatric surgery. The reference list needs tidying for consistent formatting.

Reviewer #2: This manuscript provides a retrospective analysis of nosocomial infections in pediatric patients undergoing invasive operations.

Major Comments:

- The statistical inference lacks scientific context beyond stating statistical significance. The American Statistical Association's statement on p-values highlights this point: \\emph{A p-value, or statistical significance, does not measure the size of an effect or the importance of a result.} For instance, line 269 states ``We found that longer preoperative ICU stay was associated with delayed occurrence of postoperative NIs:...`` How influential was this in terms of the occurrence probability? The entire discussion lacks this effect size discussion.

- L 109 - 111. What are the implications of only the first invasive operation / postoperative NI being analyzed? Does this result in any bias? Is it expected that follow up surgeries would have different rates of NI occurrence? My understanding is that if the first invasive operation did not result in a NI, then this data point would be included and all later invasive operations would be excluded. Is this correct?

- L 139 - 140. Be specific about how the variables were treated as `missing'. Missing data is a loaded word with a whole subfield of statistics devoted to it. How does the missing data and associated binary indicator impact the statistical models?

- L 148 - 149. How do the standard discharge rates (days after surgery) differ between the surgery types? Figure 1 shows the risk up to 7 days post operation. How large would the sample sizes be for 7 - days post operation for all surgery types? I'm curious if the analysis with incidence rate per patient days is a good comparison if many (most!) patients of a certain surgery type are released from the hospital shortly after surgery. How is this treated in the statistical models? Is this type of analysis standard, if so include supporting evidence/citations, otherwise, would an analysis of likelihood of experiencing an NI for a given surgery be appropriate?

- How are the ICD-9-CM codes used in the statistical models? Table 2 only states both models were adjusted for operative site - how so? Are these intercept values or would a more general model (such as a hierarchical model) that allows different responses to explanatory variables across the groups be more appropriate?

- Line 234. How are the high-risk and low-risk groups defined? I don't see this in the text, but the caption for Figure 2 states, "median of the risk score derived from the reduced AFT model". Why is this grouping useful or interesting? It is not surprising that the high-risk group would have higher NI incidence rates, but this does suggest the overall model is doing something useful.

Minor Comments:

- Use commas to separate five or more numbers, such as 18314 $\\rightarrow$ and 18,314.

- L 80, From prevention perspective $\\rightarrow$ from a prevention perspective.

- L 155 - 159. It is not clear how the block structures works in practice. Do the p-values for all variables in a block need to be under the specified threshold?

- P 9. ``averagely'' could be reworded/replaced.

- Table 1 contains a lot of information. One thing that might help would be lines or something to signify sub-categories (within SWC and ASA score).

- Table 2, typo ``paediatric''

- Table 3. Is the full model necessary?

Reviewer #3: This review focuses on the *statistical aspects* of the submitted manuscript, exclusively.

In essence, the manuscript describes the statistical analysis of a hospital data-set containing pediatric peri-operative data and nosocomial infection status; this reviewer understands the goal of the study as identification of pre-operative and operation-related risk factors for NI.

The authors chose "time to event" as primary variable of interest, where event specifies either NI or discharge. (see major comment, below). The main model is "Accelerated failure time" (AFT). A (more standard, but measuring something slightly different) Cox model was used for comparison (but data only reported in appendix 2). The authors state that non-linear associations of continuous factors were considered, but none retained (if this reviewer understands correctly).

A reduced model was obtained by backward elimination. Identified time-ratios of factors were used to classify patients into low- and high-risk, respectively, with an enrichment of 70% of NI in the high-risk group.

The manuscript is well written, concise, and well-structured. There are a few items where this reviewer feels that the description should be more complete (verbose, but potentially also include a few math-stats expression for precision). Overall, this reviewer believes that with appropriate revisions the manuscript should be published (unless revisions bring major statistical flaws to light).

Comments that the authors should to address in revision:

1) major one: The authors are dealing with a highly censored data set. A majority of patients does not exhibit NI (or not on record, anyways). It is absolutely not clear to this reviewer, how the authors are handling censored events (i.e. distinguish time to event between NI and discharge). In particular since NI patients are in the minority, the main event is discharge---is the AFT model therefore primarily fitting time to discharge? It is this reviewer's strong belief that this "detail" should be described and discussed in much more detail, here. What are the exact assumptions? What does the AFT model look like (a mathematical expression or two would be appreciated)? How "log-normal" are the data (show it?)? How do NI events differ from discharge events? Is the AFT model fit to patients with NI, only, or the entire collection? etc.... Maybe the authors are doing everything correct, already, but from the description it is near impossible to know (or possible reproduce). This is the central element of the study and should be presented with sufficient detail (despite space limitations...).

2) The primary outcome of the AFT model is that certain factors accelerate or decelerate the progression of a patient through progress from operation to event (NI/discharge). How does this relate to NI *risk* (=what the cox model captures)? A description of the exact relationship between these "measures" should be provided.

3) The non-linear fractional polynomial part is dealt with in twice half a sentence --- either this part is important enough to deserve some more details (then some more details on this should be provided) --- or it was just a "stump track" and should parenthetically be noted as such.

4) it is this reviewer's understanding that after model fitting, significant factors were aggregated through non-linear averaging (median) to produce a "predictive score" indicating low or high NI risk. This reviewer would appreciate some more details in the description, as this is a key contribution.

5) One of the presented results is that after risk classification, the high-risk patients exhibit twice as big empirical risk for NI compared to low-risk group. Immediately resulting thereof, the high-risk group accounts for a bit over 2/3s of the NI occurrences. Can the authors compare this figure against other models that are already "on the market"? (a random grouping into equally large groups would result in a roughly 50% split; how much "better than random" is the observed 70%?)

6) Is there a compact way to graphically visualize the content of table 3 (and possibly appendix 2)? TR and HR CI, point estimates and p-values could be shown as "boxplot"-like bars relative to the neutral 1-axis. In this way, significant and large impacts will be more intuitively identifiable from the large crowd.

7) Discussion: the results are presented as "delayed occurrence" versus "enhanced postoperative risk". Again, in the light of point 1 and 2, it is not entirely clear to this reviewer what is talked about: delayed occurrence of NI (or discharge?!)? How is post-operative risk quantified, here? (= accelerated occurrence of NI/discharge?) A priori delayed occurrence and risk are two different things that can not immediately be compared. The description should be a bit more verbose and complete, here.

8) line 256: consisting => consistent

9) This reviewer is afraid that the list of possible (likely and relevant) confounders should be longer.

10) Cross-validation: since the data are used twice---first to learn the AFT model parameters, second to stratify into risk-groups---it would be indicated to use separate parts of the data set for these tasks, to control "overfitting". How good is the risk-stratification outcome when applied on 50% of the data that was NOT used to fit the AFT model used to identify risk factors? The differences may be marginal, but it would be prudent to check.

11) one in ten rule (and variants thereof): given 847 un-censored events how many predictor parameters can reasonably be expected to be fit? The authors might want to include a note on this.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

Reviewer #3: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PlosReview.pdf

Click here for additional data file.

10 Oct 2019

Please refer to attached the rebuttal letter

Submitted filename: Rebuttal Letter.pdf

Click here for additional data file.

8 Nov 2019

Preoperative and operation-related risk factors for postoperative nosocomial infections in pediatric patients: a retrospective cohort study

PONE-D-19-18261R1

Dear Dr. Liang,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Agnieszka Rynda-Apple, Ph.D.

Academic Editor

PLOS ONE

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

13 Dec 2019

PONE-D-19-18261R1

Preoperative and operation-related risk factors for postoperative nosocomial infections in pediatric patients: a retrospective cohort study

Dear Dr. Liang:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Agnieszka Rynda-Apple

Academic Editor

PLOS ONE