The impact of recent hospitalization on surgical site infection after a pancreatectomy

Abstract

Background

Surgical site infections (SSI) are a major cause of increased morbidity and cost after a pancreatectomy. Patients undergoing a pancreatectomy frequently have had recent inpatient hospital admissions prior to their surgical admission (recent pre-surgical admission, RPSA), which could increase the risk of SSI.

Methods

The 2009–2011 Healthcare Cost Utilization Project California State Inpatient Database was used. Chi-square tests, Student's t-tests and multivariable logistic regression were used.

Results

Three thousand three hundred and seventy-six patients underwent a pancreatectomy, and 444 (13.2%) had RPSA. One hundred and eighty (40.5%) RPSAs were to different hospitals other than where patients' pancreatectomy took place. In univariate analysis, patients with RPSA had a significantly higher rate of post-operative SSIs, and this was associated with a longer length of post-operative stay, higher post-operative hospital costs and increased postoperative 30-day readmission rates (Table 1). In Multivariate analysis, RPSA was an independent predictor of post-operative SSI [odds ratio (OR) = 1.68, P = 0.013], and the risk of SSI increased with increasing RPSA length of stay (OR = 1.07 per day, P = 0.001).

Conclusions

Recent pre-surgical admission is an important risk factor for SSI after a pancreatectomy. Many patients with RPSA are not admitted pre-operatively to the same hospital where the pancreatectomy occurs; in such circumstances, SSI rates may not be a sole reflection of the care provided by operating hospitals.

Background

Surgical Site Infections (SSI) are a major cause of increased cost and morbidity after a pancreatectomy.¹ SSI after a major hepatopancreatobiliary (HPB) operation increases hospital costs by an average of $11 462 per event.² In addition, SSIs are considered a hospital-acquired condition and are being used as a quality metric for patient care. Hospitals currently receive financial penalties for high rates of SSI after certain types of surgery, such as coronary artery bypass, and it is likely that these penalties will be extended to other types of operations in the coming years. Furthermore, complications such as SSI can delay adjuvant therapy after oncological surgery, which could ultimately impact on survival.³

Risk factors for post-operative SSI can be divided into patient- and procedure-related.⁴ Many of these risk factors are targeted by methods such as the Surgical Care Improvement Program (SCIP) as well as through patient medical optimization.⁵ However, patients undergoing a pancreatectomy frequently have had recent inpatient hospital admissions prior to their surgical admissions (recent pre-surgical admission, RPSA), which could also increase the risk of post-operative SSI. Recently, hospitalized patients are more likely to be colonized by antibiotic resistant, virulent bacteria, such as methicillin-resistant Staphylococcus aureus (MRSA), which can predispose patients to infections.⁶ In addition, hospitalized patients are more likely to receive interventions which can predispose to infection, such biliary stenting^7,8 and a blood transfusion.⁹ The aim of this study was to examine whether patients who are hospitalized within 30 days prior to surgery for any reason are at a higher risk of post-operative SSI after a pancreatectomy.

Patients and methods

Study design and patient population

This study was a retrospective cohort study using the Agency for Healthcare Research and Quality (AHRQ) Healthcare Cost and Utilization (HCUP) State Inpatient Database (SID)¹⁰ for California from 2009 to 2011 to identify patients undergoing a pancreaticoduodenectomy [International Classification of Diseases, Ninth Revision, Clinical Modification (ICD9-CM) procedure codes: 52.51 and 52.7], a total pancreatectomy (ICD9-CM procedure code: 52.6) and a distal pancreatectomy (ICD9-CM procedure codes: 52.52). Variables are included in this database to track sequential visits for a patient within a state and across facilities and hospital settings. These variables include a person-level identifier (VisitLink) and a timing variable (DaysToEvent) which can be used to determine the days between hospital events for a patient. Using this approach, patients were identified with admission to any hospital from 1 to 30 days before their surgical admission and from 1 to 30 days post-surgical discharge. For simplicity, only the last admission prior to surgery and the first readmission after surgery were considered. RPSA was defined as inpatient stays > 48h with a date of discharge within 30 days prior to admission for a pancreatectomy. Patients under the age of 18 years and patients with missing VisitLink or DaysToEvent variables were excluded. Also excluded were patients who underwent hospital-to-hospital transfer for a pancreatectomy (n = 159) as this was thought to represent a separate phenomenon from RPSA.

Study variables

Common post-operative complications and procedures using ICD9-CM diagnosis and procedure codes were used based on previous work in pancreatectomy patients.¹¹ The AHRQ comorbidity software, version 3.7 (AHRQ, Rockville, MD, USA) was used to identify comorbidities present at initial surgical admission by utilizing measures defined by Elixhauser et al.¹² based on ICD9-CM diagnosis codes. Hospital volume quartiles were identified using the total number of pancreatectomies performed at individual hospitals between 2009 and 2011. The median state income quartiles are calculated by HCUP using the patient's zip code and the state of California's income data, and patients are categorized accordingly. Total hospital charges were converted to costs using the HCUP hospital-specific cost-to-charge ratio file. Cells with fewer than 11 patients per variable were relabelled as ‘< 11’ in compliance with the HCUP data use agreement.

Statistical analysis

Chi-square and Student's t-tests were used for univariate comparisons. Multivariable logistic regression was used to examine the association of RPSA and SSI. Models included age, gender, race, state income quartile, comorbidities (congestive heart failure, chronic lung disease, cancer status, hospital pancreatectomy volume quartile, pre-operative endoscopic pancreatic or biliary decompression, biliary obstruction at surgery, surgical admission pre-operative length of stay and RPSA (or RPSA length of stay)]. All P-values were two-sided and values <0.05 were considered statistically significant in all analyses. All statistics were performed using SAS version 9.3 (SAS Inc. Cary, NC, USA).

Results

In total, from 2009 to 2011, 3376 patients underwent a pancreatectomy, and were included in the study (Table 1). The average age was 64 years, roughly half were female, and about two-thirds of pancreatectomies were performed for cancer. Approximately 13% of patients had RPSA. Patients with RPSA were more likely to be of a non-white race and a lower income status. In addition, patients with RPSA were more likely to have chronic pancreatitis as well as weight loss. RPSA patients were more likely to undergo a pancreaticoduodenectomy and less likely to undergo a distal pancreatectomy. Cancer and biliary obstruction were also more common among patients with RPSA.

Table 1.

Characteristics of pancreatectomy patients with and without recent pre-surgical admission (RPSA) (n = 3376)

Characteristic	No. (Column %) or Mean (SD)		P-value
	No RPSA (n = 2932, 86.9%)	RPSA (n = 444, 13.2%)
Age, year	63.5 (12.8)	64.3 (12.2)	0.210
Gender
Male	1429 (48.9%)	240 (54.1%)	0.108
Female	1476 (50.3%)	201 (45.3%)
Missing	27 (0.9%)	<11a
Race
White	1751 (59.7%)	234 (52.7%)	0.005
Non-white	1181 (40.3%)	210 (47.3%)
Comorbidities
Congestive heart failure	89 (3.0%)	17 (3.8%)	0.372
Chronic lung disease	341 (11.6%)	54 (12.2%)	0.745
Diabetes	669 (22.8%)	124 (27.9%)	0.179
Chronic renal failure	154 (5.3%)	30 (6.8%)	0.193
Chronic pancreatitis	285 (9.7%)	60 (13.5%)	0.014
Obesity	274 (9.4%)	34 (7.7%)	0.250
Weight loss	325 (11.1%)	83 (18.7%)	<0.001
Liver disease	167 (5.7%)	20 (4.5%)	0.307
Alcohol abuse	108 (3.7%)	16 (3.6%)	0.934
Median state income quartileb
1st quartile	484 (16.5%)	122 (27.5%)	<0.001
2nd quartile	632 (21.6%)	96 (21.6%)
3rd quartile	824 (28.1%)	109 (24.6%)
4th quartile	935 (31.9%)	107 (24.1%)
Missing	57 (1.9%)	<11a
Pancreatectomy type
Whipple	1944 (66.3%)	380 (85.6%)	<0.001
Total	102 (3.5%)	18 (4.1%)
Distal	886 (30.2%)	46 (10.4%)
Diagnosis
Malignant	1972 (67.3%)	360 (81.1%)	<0.001
Benign	960 (32.7%)	84 (18.9%)
Biliary obstruction	506 (17.3%)	136 (30.6%)	<0.001
Surgical admission preoperative length of stay, day	1.4 (4.4)	1.7 (4.7)	0.132
Hospital pancreatectomy volume quartile
1st quartile (<26 patients)	738 (25.2%)	133 (30.0%)	0.0724
2nd& 3rd quartile (26–149 patients)	1504 (51.3%)	206 (46.4%)
4th quartile (>149 patients)	690 (23.5%)	105 (23.7%)

^aCells with fewer than 11 patients per variable are relabelled as ‘<11’ in compliance with the HCUP data use agreement.

^bMedian State Income Quartiles are calculated by HCUP using the patient's zip code and the state of California's income data.

The details surrounding RPSA hospital stays were examined (Table 2). The two most common non-cancer diagnoses at RPSA were biliary obstruction and pancreatitis. Approximately 41% of patients underwent pancreatic or biliary decompression during RPSA. The median length of RPSA stay was 5 days, and the average interval between RPSA discharge until surgical admission was 17 days. Roughly 40% of RPSA were to hospitals other than where patients' pancreatectomy took place (outside hospitals, OSH).

Table 2.

Details of recent pre-surgical admissions within 30-days of a pancreatectomy

Variable	No (%) or median (range)
Pancreatitis	164 (36.9%)
Biliary obstruction	255 (57.4%)
Pancreatic or biliary decompression	183 (41.2%)
Length of stay, day	5 (3–39)
Interval post discharge until surgery, day	17 (1–30)
Admission to outside hospital	180 (40.5%)

In univariate analysis, patients with RPSA had higher rates of both superficial/deep incisional SSI as well as organ space SSI (Table 3). In addition, patients with RPSA had longer median hospital stays, higher median hospital costs and a higher rate of 30-day readmission. In multivariate analysis, weight loss, having a pancreaticoduodenectomy or total pancreatectomy (with distal pancreatectomy as the reference group), surgical admission length of stay and RPSA were all independently associated with an increased risk of post-operative SSI (Table 4). Furthermore, the risk of SSI increased with increasing RPSA length of stay (Table 5).

Table 3.

Post-operative outcomes of patients with and without recent hospital admissions prior to a pancreatectomy (n = 3376)

Variable	No recent admission (n = 2932)	Recent admission (n = 444)	P-value
Superficial/Deep incisional surgical site infection (%)	288 (9.8%)	64 (14.4%)	0.003
Organ space surgical site infection (%)	163 (5.6%)	36 (8.1%)	0.034
Any surgical site infection (%)	339 (11.6%)	77 (17.4%)	<0.001
Post-operative length of hospital stay, median days (SD)	9 (10.8)	11 (11.5)	<0.001
Post-operative hospital costs, median $ (SD)	40 927 (44 342)	53 909 (54 986)	<0.001
Post-operative 30-day readmission (%)	591 (20.8%)	115 (27.1%)	0.003

Surgical site infections defined by ICD-9-CM codes¹¹

Table 4.

Significant factors associated with post-operative surgical site infection after a pancreatectomy in multivariable logistic regression with recent pre-surgical admission (RPSA) as a categorical variable

Characteristic	OR	95% CI		P-value
Comorbidities
Weight Loss	1.47	1.11	1.95	0.008
Pancreatectomy type
Whipple	3.54	2.50	5.01	<0.001
Total	2.69	1.42	5.10	0.002
Distal	[reference]
Surgical admission preoperative length of stay, per day	1.05	1.03	1.07	<0.001
RPSA	1.68	1.12	2.52	0.013

Table 5.

Significant factors associated with post-operative surgical site infection after a pancreatectomy in multivariable logistic regression with recent pre-surgical admission (RPSA) length of stay as a continuous variable

Characteristic	OR	95% CI		P-value
Comorbidities
Weight loss	1.45	1.10	1.93	0.010
Pancreatectomy type
Whipple	3.57	2.52	5.06	<0.001
Total	2.64	1.39	5.01	0.003
Distal	[reference]
Surgical admission preoperative length of stay, per day	1.05	1.03	1.07	<0.001
RPSA length of stay, per day	1.07	1.03	1.12	0.001

Discussion

In the present study, RPSA was associated with an increased risk of SSI after a pancreatectomy. Additionally, many patients with RPSA are admitted to OSH, which may have consequences for operating hospitals beyond their control. The impact of hospital exposure on infection rates is well documented; ¹³ however, little is known about the impact that recent hospitalization has on the future risk of post-operative infection at a later admission. RPSA is associated with an increased risk of SSI after a pancreatectomy, but it is unknown how long after discharge patients remain at an increased risk. Future work should examine whether delaying the interval between RPSA discharge and surgery affects this risk. Currently, efforts are being focused on reducing time in the hospital after surgery by promoting decreased post-operative length of stay and reduced readmission rates. However, the current study raises the possibility that perhaps reducing hospitalization or implementing interventions to reduce the risk of SSI as well as other complications even before the surgical admission should also be a priority. It has recently been proposed that the ability of patients to receive post-operative oncological therapy (return to intended oncological treatment, RIOT) should be used as a quality indicator of surgical care.¹⁴ As complications such as SSI directly affect RIOT, reducing RPSA could potentially improve the rate of RIOT after a pancreatectomy.

Hospitalization adversely impacts post-operative outcomes after a pancreatectomy. Pratt et al.¹⁵ demonstrated that patients who were hospitalized pre-operatively and underwent a pancreatectomy during the same admission had worse postoperative outcomes. The present study differed in that it examined the impact of remote hospitalization on SSI after a pancreatectomy. This is an important distinction for a few reasons. First of all, the present study suggests that the adverse effects of hospitalization may in fact persist beyond the immediate period of hospitalization. Thus, a recently discharged patient who is later admitted for a pancreatectomy still appears to be at increased risk of an adverse outcome. In addition, the length of stay prior to surgery during patients' surgical admissions is easily accounted for in risk-adjustment given that they are receiving care in the same hospital in which their surgery occurs. However, RPSA is not accounted for by current methods of post-operative risk prediction, such as the NSQIP risk calculator.¹⁴ Accounting for RPSA in risk-adjustment is especially difficult given that many patients are not admitted to the same hospital where their pancreatectomy takes place – 40% in the present study. RPSA to outside hospitals may adversely affect outcomes of the operating hospital, and erroneously impact the perceived quality of care at tertiary referral hospitals.

Previous studies have identified pre-operative malnutrition and weight loss as independent predictors of SSI after a pancreatectomy.^16,17 The current study confirmed this finding at the population level. The finding that a pancreaticoduodenectomy and a total pancreatectomy are associated with increased SSI (using distal pancreatectomy as the reference group) is not surprising. Distal pancreatectomies are increasingly being done laparoscopically, which can reduce superficial SSI rates.¹⁸ In addition, the enteric and biliary anastomoses involved with a pancreaticoduodenectomy and a total pancreatectomy likely also increase the incidence of post-operative SSI. Surgical admission pre-operative length of stay was also an independently associated with post-operative SSI in the present study. This finding along with the association of RPSA length of stay with increased SSI risk supports the overarching conclusion of this work – that recent pre-operative hospital exposure increases the risk of SSI.

Others have shown that pre-operative biliary decompression is associated with an increased risk of SSI.^7,8 The present study did not find this association. One reason for this could be that the database was limited to inpatient admissions to acute care hospitals, and, therefore, could not identify patients undergoing stenting procedures on an outpatient basis. It is becoming increasingly common for patients presenting electively with obstructive jaundice and a peri-ampullary mass to undergo an outpatient endoscopic ultrasound with biopsy; at which time, many gastroenterologists simultaneously perform an endoscopic retrograde cholangiopancreatogram with stenting. Additionally, perhaps some stenting procedures were performed outside the pre-operative 30-day timeframe, and, therefore, were not detected in the present study.

The present study has some limitations. The HCUP SID is an administrative database and relies on ICD-9CM codes; thus, there is the potential for coding error or bias. This study utilized population level data and, therefore, could not examine many of the details surrounding admissions and surgery. Owing to the nature of the database, only inpatient data were able to be examined and any SSI diagnosed in the outpatient setting could not be detected. Unfortunately, microbiological data are not available in the database used in the present study. It would be interesting to see if the organisms responsible for the SSI were the same as those cultured in the bile or if they were more likely to be ‘hospital-acquired’ organisms, such as MRSA. Finally, as with any retrospective study, residual confounding owing to variables not included in the database could also bias the results.

In conclusion, RPSA is an important risk factor for SSI after a pancreatectomy and the risk of SSI increases with increasing RPSA length of hospital stay. Furthermore, many patients with RPSA are not admitted to the same hospital where the pancreatectomy occurs; in such circumstances, SSI rates may not be a sole reflection of the care provided by operating hospitals. Future attempts to predict, risk-adjust for, and prevent SSI should take into account RPSA as an indicator of increased SSI risk, both in a pancreatectomy and perhaps in other surgical procedures.

Conflicts of interest

None declared.