journolofHos~ital
Infection (1998) 39: 71-74
Risk factors for surgical diagnosed after hospital M. Lecuona, A. Torres-Lana, A. Sierra
site infections discharge
M. Delgado-Rodriguez*,
J. Llorca*
and
Department ofMicrobiology and Preventive Medicine, School ofMedicine, University ofh Laguna, Tenerife; and *Division of Preventive Medicine and Public Health, School ofMedicine, University of Cantabria, Santander; Spain
Summary:
A prospective cohort study on 1103 consecutive patients undergoing general surgery with a follow-up of up to 30 days was undertaken to analyse the risk factors for surgical-site infection (SSI). Relative risks (RRs), crude and multiple-risk factors adjusted for by logistic regression analysis, and their 95% confidence intervals were calculated. One hundred and four patients (9.4%) developed infection, 81 in hospital and 23 at home. Predictors for in-hospital SSI differed from those for post-discharge SSI. In a crude analysis, an increased risk of post-discharge SSI occurred after clean-contaminated surgery (but not contaminated surgery). Stepwise logistic regression failed to identify any significant predictor for post-discharge SSI. Keywords: Surgical-site
infection;
postdischarge
surveillance;
Introduction The increase in health-care costs has encouraged day surgery or early discharge. This may increase the number of un-noticed surgical-site infections (SSIs) and so underestimate the risks of health care. We have found only three previous studies on the SSI determinants after discharge: following hernia repair,’ and general surgery. 2,3None of the reports included an analysis of determinants stratified by SSI type (superficial incisional, deep incisional, organ/ space). If risk factors vary with type, it may Received 9 July 1997; revised manuscript accepted 3 December 1997. Correspondence to: Dr M Lecuona, Department of Medicine, Hospital and Preventive Microbiology Universitario de Canarias, 38071 -La Laguna (Tenerife), Spain. Tel: 34 22 31 91 73. Fax: 34 22 31 92 79.
0 195-670
I /98/05007
I + 04 $12.0010
risk factors.
explain some of the differences in the epidemiology of in-hospital SSIs and post-discharge SSIs. This was the object of our study.
Methods The study was carried out at the University of La Laguna Hospital, a 651-bed tertiary center. Patients attending for General Surgery between 1 May 1994, and 30 April 1995 were studied. Patients who died within 48 h of admission were excluded; 1103 patients were enrolled. The commonest operative procedures were: cholecystectomy and bile duct surgery (20*90/o), appendectomy (16.3%), hernia repair (ll.S%), colonic surgery (10.8%) endocrine system surgery (9.2%) and gastric surgery (6.3%).
0 1998 The Hospital
Infection
Society
72
Data on risk factors collected included condition at admission, and details of diagnostic and therapeutic procedures. The SENIC index4 and the National Nosocomial Infections Surveillance (NNIS) index’ for assessing intrinsic patient risk were computed. Patients with nosocomial infections were prospectively identified according to the Centers for Disease Control (CDC) criteria.6 Surveillance was extended to 30 days post-operation (usually 30 days postdischarge from hospital), based on a scheduled visit to the surgeon. Seven hundred and fortyone patients attended this visit and 35 died while in hospital. Follow-up was complete in 70.4% of patients. Surveillance for patients not attending follow-up was based on a review of the Accident/ Emergency Department records as in other studies.3 Statistical association was assessed as relative risk (RR). A stepwise logistic regression analysis was performed to identify the best independent predictors for SSI. The significance level to allow inclusion of a variable in the regression model was 0.2.
Results One hundred and four patients (9.4%) developed SSIs; in 23 (2.2%) the SSI was detected after hospital discharge. Ten post-discharge SSIs (43%) were detected within a week of discharge, eight (35%) within the second week, and the remaining five thereafter. SSIs clustered neither by surgeon nor surgical procedure. The distribution of in-hospital SSIs was: 55.6% superficial incisional, 25.9% deep incisional, and 18.5% organ/space. The figures for post-discharge SSIs were 21.7, 52.2 and 26.1% respectively. There was no significant difference in the bacteria isolated from either group. Many well-known risk factors for SSI showed a significant (usually PcO.01) correlation for inpatient SSI (of all types) including age, sex, diabetes mellitus, cancer, American Scientific Association (ASA) score, number of diagnoses, NNIS index, SENIC index, length of preoperative stay, emergency operation, duration of operation, type of wound, wound drainage and stay in the Intensive Care Unit. None of
M. Lecuona et al.
these variables were associated with post-discharge SSI. The only variable associated with the latter was clean-contaminated surgery (contaminated or dirty surgery were not-see Table 1). In order to ascertain the independent predictors of SSI, the aforementioned variables were included in a stepwise logistic regression analysis (Table II). Only two variables were selected for analysis of post-discharge SSI. Although none reached statistical significance (variables with P-values >5% were included) the only variables that were selected were chemoprophylaxis and male gender. These results contrasted with those for in-hospital SSI.
Discussion The risk factors found for in-hospital SSI agree with those already reported. We failed however to identify any significant risk factors for postdischarge SSI. This implies as has been sugsurveillance on gested,3 that conventional patients with higher risk of infection, would not achieve the goal of identifying potential for nosocomial infections developing post-discharge. A possible drawback of the study is in the low numbers of post-discharge SSIs found, which might affect the statistical power to detect weak associations. Also 30% of patients failed to attend for follow-up; although these did not differ (P>O*l) in sex, mean age, mean ASA, NNIS Index or post-operative stay from the remainder. We included a review of the Emergency Department attendances in case an infected patient had attended for relief of symptoms.3 However some patients may have been treated by their community health services. Such patients would have been likely to have had superficial incision infections, and this could account for the low numbers found in the study. Patients with severe infection would have been much more likely to return to hospital for treatment. However no relationship between intrinsic risk of infection (measured by both SENIC and NNIS indices) and post-discharge SSI was observed.
Risk factors
Table
I
for surgical infections
73
Risk factors fir deep surgical site infeaion Total patients
(DSSI) existing at admission In-hospital DSSIV n (%)
Age (years) 130 31-65 66-75 >75 Test for trend Cancer No Yes ASA score I II Ill IV/v Test for trend Number of diagnoses I 2 3+ Test for trend Duration of operation 160 min 61-120 121-180 >I80 Test for trend Type of surgical wound Clean Cleanlcont. Contaminated Dirty Test for trend Wound drainage No Yes Open Closed Surgeon* Low Medium High Test for trend SENIC index 21 I 2 3 Test for trend
211 620 157 II5
4 I8 6 8
I.9 (2.9) (3.8) (7.0)
899 204
22 (2.4) I4 (6.9)
456 445 172 30
6 12 I4 4
(1.3) (2.7) (8.1) (13.3)
673 255 175
II (1.6) I2 (4.7) I3 (7.4)
639 326 109 29
9 I3 IO 4
(1.4) (4.0) (9.2) (13.8)
Post-discharge DSSI n (%)
2 II 3 2
(1.0) (1.8) (2.0) (1’9)
I5 (1’7) 3 (1’7) 9 4 5 0
(2.0) (0.9) (3.3) (0.0)
I3 (1.9) 2 (0.8) 3 (1.9)
8 6 4 0
(1.3) (1.9) (4.0) (0.0)
RR for inhospital DSSI (95% Cl)
RR for postdischarge DSSI (95% Cl)
I I .5 (0.54.5) 2.0 (04-70) 3.7 (1.1-l 1’9) P=O.O14
I I.9 (04-8.6) 2. I (O+ I 2.4) 2.0 (0.3-l 3.7) P=O.513
I 2.8 (I .5-5.4)
I I .o (0.3-3.4)
I 2.0 (0.8-5.4) 6.2 (24-l 5.8) IO. I (3.0-34.0) PCO.00 I
I 0.5-I .5) I .7 (06-4.9) P=O.871
I 2.9 (I .3-6.4) 4.5 (2.1-10.0) PCO.00 I
I 0.4 (0. I - I ‘9) I .o (0.3-3.4) P=O.658
I 2.8 (I .2-6.6) 6.5 (2.7-l 5.7) 9.8 (3.2-29.9) PCO.00 I
I I .5 (0.5-4.3) 3.2 (0.9- 10.3) P=O.206
358 316 247 182
2 I2 I2 IO
(0.6) (3.8) (4.9) (5.5)
3 IO I 4
(0.8) (3.2) (0.4) (2.3)
I 6.6 (I .5-29.2) 8.7 (2.0-38.5) 9.8 (2.2-44.4) PCO.00 I
I 3.8 (1.1-13.7) 0.5 (0.1-4.9) 2.7 (061-l 1.8) P=O.637
676 427 172 255
9 27 I7 IO
(I.3 (6.3) (9.9) (3.9)
IO 8 4 4
(1.5) (2.0) (2.5) (1.6)
I 4.7 (2.3-l 0.0) 7.4 (34-l 6.4) 2.9 (I .2-7.2)
I I .3 (0.5-3.3) I .7 (0.5-5.2) I I (0.3-3.4)
525 361 216
IO (1.9) I5 (4.2) I I (5.1)
I 2.2 (I Gl.8) 2.7 (I .2-6.2) P=O.OlS
I 0.6 (0.2-I ‘8) I .o (0.3-3. I) P=O.790
618 353 I04 28
5 (0.8) I9 (5.4)
I 6.7 (2.5-l 7.7) 9.6 (3.3-28. I) 22. I (6.8-7 1.9) P
I I .2 (04-3. I) 0.6 (0. I-5. I ) 2.4 (0.3- 18.0) P=O.509
7 (6.7) 5 (17.9)
VAfter excluding 35 in-hospital deaths. * Risk levels according to previously adjusted NNis patient
IO (2.0 4 (I.1 4 (1.9)
9 7 I I
(1.5) (2.0) (1.1) (4.2)
risk for each surgeon.
M. Lecuona et al.
74
Table
II
Order entry
of
Results from the stepwise logistic regression ono/yses Variable
In-hospital surgical site infection (SSI) Wound drainage*: Open I Closed 2 Male gender Protocolized chemoprophylaxis 3 Wound contamination (per level) 4 No. of diagnoses 5 Duration of operation 6 (continuous in hours) In-hospital deep SSI SENIC index (continuous) I Wound drainage**: Open 2 Closed ASA score (continuous) 3 Emergency intervention 4 Protocolized chemoprophylaxis 5 Preoperative stay (continuous in 6 weeks) Duration of operation (per hour) 6 Post-discharge SSI Chemoprophylaxis I 2 Male gender Post-discharge deep SSI I Male gender
OR* (95% Cl)
the classic risk factors post-discharge SSIs.
for SSI
do not
apply
to
References 3.8 2.1 I .5 0.6 I .5 I.5 I .3
(2.1-6.9) (l~l-4~1) (0.9-2.5) (0.3-0.9) (I .2-2.0) (1.1-2.1) (0.9-I .7)
I .6 3.8 2.1 I .4 2.7 0.6 I .3
( I . I-2.4) (2.1-6.9) (1.1-4.1) (0.9-2.2) (1.1-6.8) (0.2-I .3) (0.9-I .9)
I .3 (0.7-l ‘7) 2.6 (0.7-8.7) I .8 (0.4.4) 2. I (0.8-5.6)
* All OR (odds ratio) estimates are mutually adjusted for the variables included in the model. ** No drainage = reference standard (I).
Our rate of post-discharge SSI was 22.10/o, a figure greater than the 14% reported in one study,” but less than the 40-70% found in others.7,R These differences may be reconciled in part by the different surveillance methods Similar findings have been reported previously, but these did not include an analysis of wound type. Others reported that most post-discharge SSIs are found after clean one report found the insurgery,2X3Z7 although cidence to be higher after clean-contaminated surgery.” Our isolates (all Gram-negative) from post discharge SSIs bear out the role of cleancontaminated surgery. In conclusion-most of
1. Simchen E, Wax Y, Galai N, Israeli A. Discharge from hospital and its effect on surgical wound infections. The Israeli Study of Surgical Infections (ISSI). J Clin Epidemiol 1992; 45: 11551163. 2. Weigelt
JA, Dryer D, Haley RW. The necessity and efficiency of wound surveillance after discharge. Arch Surg 1992; 127: 77-82. M, Sillero-Arenas M, Martinez 3. Medina-Cuadros Gallego G, Delgado-Rodriguez M. Surgical wound infections diagnosed after discharge from hospital. Epidemiological differences with inhospital infections. Am J Infect Control 1996; 24: 421428.
4. Haley RW, Culver DH, Morgan WM, White JW, Emori TG, Hooton TM. Identifying patients at high risk of surgical wound infection: a simple multivariate index of patient susceptibility and would contamination. Am J Epidemiol 1985; 121: 206-215.
DH, Horan RC, Gaynes RP, Mortone 5. Culver WJ, Jarvis WR, Emori TG, et al. Surgical wound infection rates by wound class, operative procedure, and patient risk index. Am J Med 1991; 91 (suppl 3B): 152S-157s. 6. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. Centers for Diseases Control (CDC) definitions for nosocomial surgical site infections, 1992: a modification of CDC definitions of surgical wound infections. Infect Control Hasp Epidemiol 1992; 13: 606-608. Reimer K, Gleed C, Nicolle LE. The impact of postdischarge infection on surgical wound infection rates. Infect Control 1987; 8: 237-240. Burns SJ, Dippe SE. Postoperative wound infections detected during hospitalization and after discharge in a community hospital. Am J Infect Control 1982; 10: 60-65. Rosendorf LL, Octavia J, Estes JP. Effect of method of post-discharge wound infection surveillance on reported infection rates. Am J Infect Control 1983; 11: 226-229.