Factors Affecting Readmission in Patients with Surgical Site Infection: A Graphical and Prediction Model-Based Approach

Abstract

Background:

Antimicrobial therapy is becoming less effective because of the rising microbial resistance. Surgical site infections (SSI) are one of the major complications that require modifications in the infection control policy for effective management.

Objective/Aim:

To develop a model for predicting the readmission rates post-SSI treatment and to identify prevalent microbial isolates and the respective trends in resistance patterns.

Methodology:

A retrospective study was carried out in a tertiary care setting in India. A total of 549 patients were diagnosed with SSI from January 1, 2016, to August 25, 2021, visiting orthopedics (n = 373), general surgery (n = 135), and neurosurgery (n = 41) departments were included in the study. Patient data and microbial isolate data were collected. Logistic regression with purposeful selection of covariates (p ≤ 0.25) was used to identify the predictors. The model fit was validated using the omnibus test. The area under the curve (AUC) was considered for the model discrimination. The resistance trend of microbial isolates was graphically represented.

Results:

One hundred thirty-seven (24.9%) were readmitted because of repeated infections. Readmission happened with a mean of 152 ± 32 days post-surgery was estimated. Uni-variable logistic regression showed 40 significant variables. The multi-variable logistic regression eliminated three variables because of insufficient comparator levels. Collinearity statistics further excluded two variables, i.e., reconstruction type of surgery and peripheral surgical area (variance inflation factor >10). The model showed an AUC of 0.77 and an accurate prediction of 77.8% (Akaike Information Criterion [AIC]: 568; Bayesian Information Criterion [BIC]: 722). Fifteen types of micro-organisms were isolated from 75.4% of readmitted patients. Methicillin-resistant Staphylococcus aureus (23.8%) was the primary isolate showing a resistance trend toward cloxacillin, ciprofloxacin, and ofloxacin (25.69%) equally, followed by erythromycin (18.4%) and gentamycin (6.25%).

Conclusion:

The current study predicted the post-SSI readmission rate and the microbial isolates along with their resistance patterns. The results of the study could serve as a tool for assessing and managing the factors leading to readmissions.

Microbial colonization at surgical sites is an ever-increasing threat. Antimicrobial therapy is becoming ineffective as the micro-organisms are gaining resistance. According to the World Health Organization, antimicrobial resistance (AMR) is one of the top 10 global public health threats. AMR can add to the increased liability caused by readmissions because it leads to prolonged hospital stays, thereby increasing the financial burden.¹ Surgical site infections (SSI) are prone to cause a substantial change in the infection control policy during the hospital stay.² Interventions for managing SSIs can lead to reduced expenditure, vitalizing the necessity for assessing the outcomes of patients. The complications of SSI include increased hospital readmission, reoperation, morbidity, and mortality.³

Factors affecting SSI can predict the chances of readmission because of its complications. This could be the result of the untoward outcome following the surgery or because of patient-related characteristics at the time of SSI diagnosis.⁴ The necessity for readmissions can differ considerably from one healthcare setting to another on the basis of the efficiency of the hospital in handling the SSI at the outpatient department and because of the differences in the expertise of the infection control team. Prediction algorithms for individual healthcare settings can give a focused outcome on the basis of the identification of relevant factors influencing readmission.^5,6 The current study aims to design a prediction model for readmission after being diagnosed with SSI and to identify the major microbial isolates, with their resistance trend from the infected sites.

Methodology

A retrospective study was conducted in a tertiary care setting in India after obtaining approval from the institutional ethics committee (IEC 731-2021). The medical records of patients diagnosed with SSI (ICD 2019: T 81.4) from January 1, 2016, to August 25, 2021, were reviewed. The data were abstracted from the medical records manually (the study site does not have an electronic medical record system). Patients older than 18 years who underwent surgeries in the departments of general surgery, orthopedics, and neurosurgery were included in the study. Oncology case records and incomplete case records were excluded from the study.

The SSI patients were identified on the basis of the diagnosis of the patient medication chart. The inclusion criteria were restricted to superficial and deep incisional SSI. We excluded the organ space SSI admissions. The patients was admitted because previous infections at the site of surgery were considered as readmission cases. SSI-related admission was defined by the Center for Disease Control and Prevention’s National Healthcare Safety Network (Supplementary Data S1).⁷

Data analysis

Parameters related to patient demography, clinical parameters, treatment, infection site, type and area of surgery, type of surgical wound, and calendar details were retrieved from the medical records. The microbial culture information was retrieved from the laboratory information system. A predesigned Excel sheet was utilized for resistance pattern analysis. The AMR pattern for individual microbial isolates was compiled, and the resistance trend from 2016 to 2021 was analyzed descriptively.

The variables were represented as median ± inter-quartile range or mean ± standard deviation (SD) on the basis of the type of distribution. Frequency (%) was used for categorical variables. Purposeful selection of covariates was used to identify the significant variables using uni-variable logistic regression analysis.⁸ The variables with p value ≤0.25 were taken for multi-variable logistic regression analysis, which excluded variables on the basis of the absence of comparator levels. The collinearity regression model, accompanied by the omnibus test, was used to check for goodness of fit of the model. Collinearity statistics were assessed for variance inflation factor (VIF) greater than 10 to further exclude the variables in the model. The area under the receiver operating characteristic curve (ROC) was plotted to test for the model discrimination. Statistical analysis was carried out using Jamovi version 2.3.26.

Result

Demographic characteristics

Of the 549 SSI admissions, 137 were readmitted to the departments of orthopedics (n = 107), general surgery (n = 24) and neurosurgery (n = 6). The mean population age at the surgery time was 44.2 ± 15.26 years. There was a greater representation of males (78%) in the study. The maximum frequency of SSI-related admissions happened in 2016 (29%). Among the study population, the highest frequency of comorbidities was identified as diabetes mellitus (18%) and hypertension (16%). Smoking and alcoholism were observed to be 13% and 14%, respectively. The patient characteristics with a comparison of readmitted and non-readmitted with SSI are detailed in Table 1.

Table 1.

Patient-Related Characteristics

Characteristics	Readmission group (N = 137)	No readmission group (N = 412)
Age in years (mean ± SD)	43.6 ± 17.2	44.3 ± 14.5
Gender
Male	112 (81.8)	318 (77.2)
Female	25 (18.2)	94 (22.8)
Body mass index	23.5 ± 5.2	24.1 ± 4.9
Public assistance (yes)	44 (32.1)	111 (26.9)
Comorbidities
Diabetes mellitus	26 (19)	70 (17)
Hypertension	15 (10.9)	70 (17)
Cardiac disorders	3 (2.2)	11 (2.7)
Thyroid disorders	2 (1.5)	9 (2.2)
Rheumatoid disorders	3 (2.2)	3 (2.2)
Bronchial asthma	1 (0.7)	1 (0.7)
Others^a	10 (7.3)	24 (5.8)
Social habits
Smoking	15 (10.9)	58 (14.1)
Alcoholism	13 (9.5)	63 (15.3)

Rare comorbidities, frequency of two or less in the entire patients taken.

SD = standard deviation.

Surgery-related characteristics

The surgery-related characteristics included parameters relating to the initial surgery, the site of incision, the type of surgery carried out, wound type, anesthesia given, and the condition of the type of surgical infection. Surgical procedures were classified on the basis of the nature of the procedure done as ablative surgery, including the removal of one or more body parts by surgical means such as fingers and toes after crush injury; diagnostic surgery, including procedures to diagnose a disease condition such as laparoscopy; and palliative surgery, including procedures to cure uncontrolled pain or save a life-threatening condition such as abscess removal and reconstruction surgery, where methods for replacing a body part with an implant, such as total knee replacement, are carried out. Slough, discharge, and malodor used as parameters in the study were obtained from the medical records as the terms were clearly mentioned in the patient records. The parameters are detailed in Table 2.

Table 2.

Surgery-Related Characteristics

Characteristics	Total % (n = 549)
Blood transfusion	49 (8.9)
Implant at the surgical site	164 (29.9)
Anesthesia
General anesthesia	95 (17.3)
Regional anesthesia	218 (39.7)
Local anesthesia	33 (6)
Infected condition
Slough/discharge	411 (74.9)
Necrosis	137 (25)
Malodor	60 (10.9)
Surgical site
Abdominal area^a	94 (17.1)
Extremities^b	388 (70.7)
Spine/cranium	58 (10.6)
Breast/chest/face/neck	10 (1.8)
Surgical wound type
Clean surgical wound^c	41 (7.5)
Non-clean surgical wound^d	408 (92.5)
Type of surgery
Ablative	144 (26.2)
Diagnostic	5 (0.9)
Palliative	27 (4.9)
Reconstruction	373 (67.9)

Abdominal portions and organs of the body.

Upper and lower limbs.

Surgical wound without any contamination.

Surgical wound in contaminated condition.

Management, culture, and hospital admission-related characteristics

The parameters such as the antimicrobial therapy with its route of administration, ventilator admissions, procedure time during surgery, details on the pain medications, amputation procedures done in these patients during the infected conditions, culture samples taken from SSI site, types of infections in the patients, calendar parameters including hospital stay, days to procedures, readmission, and the number of admissions from surgery to readmission, be it with infection complications or for surgical examinations, were collected from the patient records. A wound sample from the infected surgical site was taken for culture testing in 75% of patients, among which 71% of samples retrieved isolates (these are the samples that gave isolate presence through culture testing using the samples from the sites of infection after developing SSI and not from the samples collected during the surgical procedure period), which were the pathogenic micro-organisms from the sites consisting of 27% gram-positive, 24% gram-negative, and 2% both gram-positive and gram-negative micro-organisms. A total of 7% of SSI were treated with antibiotic agents, without a retrievable pathogen. The parameters are detailed in Table 3.

Table 3.

Management, Culture, and Hospital Admission-Related Characteristics

Surgery characteristics		SSI characteristics		Discharge characteristics		Hospital admission-related characteristics
Characteristics	Total % (n = 549)	Characteristics	Total % (n = 549)	Characteristics	Total % (n = 549)	Characteristics	Total % (n = 549)
Antibiotic agents post-surgery		Antibiotic agents for SSI		Antibiotic agents for discharge		Type of infection
Glycopeptide	11 (2)	Nitroimidazole	55 (10)	Oxazolidinones	45 (8.2)	Infection	508 (92.5)
Quinolone	18 (3.3)	Oxazolidinones	18 (3.3)	Aminoglycosides	25 (4.6)	Reinfection	12 (2.2)
Beta-lactam	263 (49.9)	Aminoglycosides	84 (15.3)	Glycopeptides	9 (1.6)	Recurrence	12 (2.2)
Aminoglycoside	25 (4.6)	Quinolones	31 (5.6)	Nitroimidazole	10 (1.8)	Unknown	17 (3.1)
Lincosamide	6 (1.1)	Macrolides	2 (0.4)	Macrolides	3 (0.5)	Number of days
Oxazolidinone	4 (0.7)	Sulphonamides	27 (4.9)	Sulphonamides	84 (15.3)	DOA to DOPCR^a(median ± IQR)	2.0 ± 3.0
Tetracycline	1 (0.2)	Diaminopyrimidines	27 (4.9)	Diaminopyrimidines	84 (15.3)	DOA to DOD^b(median ± IQR)	7.0 ± 7.0
Nitroimidazole	16 (2.9)	Lincosamide	17 (8)	Lincosamide	34 (6.2)	DOS to DOA^c(median ± IQR)	32 ± 84.3
Ventilator admission	124 (22.4)	Glycopeptides	15 (2.7)	Quinolones	47 (8.6)	Number of admissions from surgery to present infection (mean ± SD)	1.1 ± 0.55
Procedure time (min) (median ± IQR)	60.0 ± 45.0	Tetracycline	1 (0.2)	Tetracycline	3 (0.5)
Route of administration		Macrolides	2 (0.4)	Isolated Organisms
Oral	53 (9.7)	Beta lactam	408 (74.3)
Intravenous	455 (82.9)	Nutraceutical	209 (38.1)	Culture sample
Local application	3 (0.5)	Pain medication		Wound sample	414 (75.4)
		Non-opioid drugs	459 (83.6)	Tissue	49 (8.9)
		Opioid drugs	131 (23.9)	Pus	41 (7.5)
		Pain score (mean ± SD)	2.7 ± 1.89	Others	33 (6.0)
		Route of administration		Culture positivity	389 (70.9)
		Oral	413 (75.2)	Gram staining
		Intravenous	58 (10.6)	Gram-positive micro-organisms	148 (27)
		Local application	5 (0.9)	Gram-negative micro-organisms	130 (23.7)
		No antibiotic agents	24 (14)
		Intervention type
		Conservative	186 (33.9)
		Non-conservative	363 (66.1)
		Amputation	6 (1.1)

Number of days from date of admission (DOA) to date of SSI therapeutic procedure (DOPCR).

Number of days from DOA to date of discharge (DOD) for SSI admission (hospital stay).

Number of days from date of original surgery (DOS) to DOA for SSI.

SSI = surgical site infections; IQR = inter-quartile range; SD = standard deviation.

Resistance trend of the microbial isolates

The collected data showed 15 types of microbial isolates. The antimicrobial sensitivity test was carried out before an antimicrobial was prescribed for the treatment of SSI. Multidrug-resistant Staphylococcus aureus (MRSA) i.e., 24% was the most prevalent microbial isolate, followed by other Staphylococcus species (18%), Escherichia coli (12%), Klebsiella (9%), Enterobacter species (5%), and Pseudomonas aeruginosa (5%). Microbial isolates such as Providencia stuartii, Citrobacter species, Proteus mirabilis, Streptococcus pyogenes, Acinetobacter baumannii complex, Morganella morganii, Serratia marcescens, Aeromonas hydrophila, and Enterococcus were also isolated rarely. Escherichia coli showed the highest resistance toward ampicillin and amoxicillin. MRSA showed increased resistance toward cloxacillin, ciprofloxacin, and ofloxacin, followed by erythromycin and gentamycin. Staphylococcus species showed greater resistance to ciprofloxacin and ofloxacin, piperacillin, tazobactam, and erythromycin. A detailed graphical representation of these microbial isolates is given in Figure 1.

FIG. 1.

Resistance trend of the microbial isolates.

Readmission in SSI patients

The model for readmission included categories of parameters representing public assistance, comorbidities, social habits, implant at the surgical site, anesthesia, infected condition, surgical site, surgical wound type, type of surgery, antibiotic agents post-surgery, route of administration, antibiotic agents for SSI, pain medication, pain score, nutraceutical, route of administration, antibiotic agents for discharge, type of infection and number of admissions from surgery to present infection. Surgery of extremities and reconstruction surgical sites were excluded from the model according to the VIF >10 of collinearity statistics. The final model had a prediction accuracy of 79.6%. The area under the curve (AUC) between the predicted probability and the original clinical cases of readmission was 0.77 (sensitivity: 0.950; specificity: 0.32), suggesting an acceptable discrimination capacity. The odds ratio and corresponding p value of the variables used in the model are given in Table 4 (refer to Supplementary Data S2 for the variables used in the uni-variable analysis). The ROC representing the model is given in Figure 2.

FIG. 2.

Receiver operating characteristic curve.

Table 4.

Multi-Variable Logistic Regression

Variables		Category		Uni-variable logistic regression		Multi-variable logistic regression
Name	Category	p value	Crude odds ratio	95% Confidence interval	p value	Adjusted odds ratio	95% Confidence interval
Public assistance
Public assistance	No	0.244	1.28	(0.84, 1.95)	0.130	0.68	(0.41, 1.12)
Comorbidities
Hypertension	Yes	0.093	0.60	(0.33, 1.09)	0.199	0.6464	(0.33, 1.26)
Rheumatoid disorders	Yes	0.175	0.33	(0.07, 1.64)	0.292	2.695	0.43, 17.06
Social habits
Alcoholism	Yes	0.095	1.71	(0.91, 3.22)	0.143	0.595	0.29, 1.19
Implant at the surgical site
Implant at the surgical site	Yes	0.001	1.95	(1.30, 2.92)	0.109	1.5085	(0.91, 2.49)
Anesthesia
Regional anesthesia	Yes	0.003	0.56	(0.38, 0.83)	0.698	1.1263	(0.62, 2.05)
Local anesthesia	Yes	0.041	0.29	(0.09, 0.95)	0.574	0.6884	(0.19, 2.53)
Infected condition
Malodor	Yes	0.005	0.45	(0.26, 0.77)	0.006	2.5441	(1.31, 4.93)
Surgical site
Abdominal area	Yes	0.007	2.33	(1.25, 4.34)	0.228	5.5526	(0.34, 89.88)
Extremities	Yes	<0.001	2.84	(1.71, 4.72)	0.124	8.9501	(0.55, 146.18)
Extremities^a	Yes	<0.001	2.84	(1.71, 4.72)	0.124	8.9501	(0.55, 146.18)
Spine/cranium	Yes	0.042	0.45	(0.21, 0.97)	0.322	4.4423	(0.23, 85.12)
Surgical wound type
Surgical wound type	Non-clean surgical wound	0.230	1.67	(0.72, 3.86)	0.76	1.187	(0.39, 3.56)
Type of ssurgery
Ablative	Yes	0.008	0.52	(0.32, 0.84)	0.926	1.1209	(0.10, 12.46)
Palliative	Yes	0.049	0.23	(0.05, 0.98)	0.509	0.3854	(0.02, 6.51)
Reconstruction^a	Yes	<0.001	2.18	(1.38, 3.46)	0.89	1.1809	(0.11, 12.57)
Antibiotic agents post-surgery
Glycopeptide	Yes	0.032	3.73	(1.12, 12.42)	0.726	0.7277	(0.12, 4.29)
Beta-lactam	Yes	0.209	0.78	(0.53, 1.15)	0.557	0.8392	(0.47, 1.51)
Lincosamide	Yes	0.175	3.05	(0.61, 15.30)	0.432	2.1148	(0.33, 13.72)
Route of administration
Oral	Yes	0.162	0.59	(0.28, 1.24)	0.518	1.3609	(0.53, 3.47)
Antibiotic agents for SSI
Nitroimidazole	Yes	0.065	0.48	(0.22, 1.05)	0.583	0.7619	(0.29, 2.01)
Aminoglycosides	Yes	0.032	1.99	(1.06, 3.71)	0.138	0.57	(0.28, 1.20)
Quinolones	Yes	0.055	3.26	(0.97, 10.89)	0.337	0.52	(0.14, 1.96)
Sulfonamide	Yes	0.220	0.51	(0.17, 1.50)	0.312	0.53	(0.16, 1.81)
Glycopeptides	Yes	0.004	0.21	(1.66, 13.62)	0.008	9.13	(1.77, 47.09)
Pain medication
Non-opioid drugs	Yes	0.088	1.66	(0.93, 2.95)	0.64	1.19	(0.56, 2.54)
Opioid drugs	Yes	0.133	1.40	(0.90, 2.17)	0.97	0.99	(0.55, 1.79)
Pain score
Pain score		0.220	1.07	(0.96, 1.18)	0.071	0.88	(0.77, 1.01)
Nutraceutical
Nutraceutical	Yes	0.076	1.43	(0.96, 2.11)	0.159	1.42	(0.87, 2.32)
Route of administration
Oral	Yes	0.162	0.58	(0.28, 1.24)	0.868	0.95	(0.52, 1.73)
Local application	Yes	0.142	6.09	(0.55, 67.68)	0.943	1.10	(0.07, 16.48)
Antibiotic agents for discharge
Aminoglycosides	Yes	0.178	1.78	(0.77, 4.13)	0.078	2.59	(0.89, 7.50)
Glycopeptides	Yes	0.043	3.95	(1.05, 14.94)	0.967	0.96	(0.12, 7.42)
Macrolides	Yes	0.137	6.23	(0.56, 69.23)	0.105	8.38	(0.64, 109.79)
Type of infection
Infection	Yes	<0.001	0.16	(0.08, 0.32)	0.05	0.20	(0.04, 1)
Reinfection	Yes	0.013	4.38	(1.36, 14.04)	0.632	0.65	(0.11, 3.88)
Recurrence	Yes	0.054	3.10	(0.98, 9.77)	0.41	0.47	(0.08, 2.84)
Number of admissions from surgery to present infection
Number of admissions from surgery to present infection		<0.001	0.34	(0.21, 0.55)	0.248	0.67	(0.34, 1.32)

Variables excluded on the basis of VIF value >10 in the multi-variable analysis.

SSI = surgical site infections; VIF = variance inflation factor.

Discussion

Readmissions in patients with SSI are an outcome of deteriorated treatment measures in healthcare settings.⁹ The present study identified factors affecting readmission because of SSI complications and gave a predictive modeling approach for the same. Our results described the significance of demographic and treatment-related parameters for readmission.

According to Shah et al. in 2017, males (Heart Rate [HR]: 1.25; 95% confidence interval [CI]: [1.07–1.46]; p < 0.001), class II or III obesity (HR: 1.33; 95% CI: [1.16–1.53]; p < 0.001) contaminated wound class (HR: 1.68; 95% CI: [1.03–2.73]; p = 0.03), American Society of Anesthesiologists (ASA) class III (HR: 1.22; 95% CI: [1.11–1.35]; p < 0.001), smoking (HR: 1.22; 95% CI: [1.1–1.36]; p < 0.001), steroid use (HR: 1.28; 95% CI: [1.10–1.49]; p < 0.001), surgery for perforation or obstruction (HR: 1.41; 95% CI: [1.17–1.69]; p < 0.001), disseminated cancer (HR: 1.46; 95% CI: [1.26–1.68]; p < 0.001), serum albumin concentration less than 3 g/dL (HR: 1.22; 95% CI: [1.06–1.4]; p = 0.004), long operative period (p < 0.001), younger patients, and open surgery were the substantial factors affecting readmission in SSI patients.⁶ A retrospective study identified 3.8% of readmissions in surgical wards of two hospitals among the patients diagnosed with SSI. In patients admitted to the general surgery ward, diabetes mellitus as a comorbidity, low serum albumin, and those who underwent contaminated surgery and longer surgical duration were statistically significant in patients with readmission.¹⁰ In a retrospective study on 3,663 patients, readmission occurred in 54% of patients diagnosed with SSI after discharge (p = 0.025). A total of 31% of patients with superficial SSI, 92% with organ space SSI, and 87% with multiple space infections were readmitted, 6% of patients were admitted to non-emergency units for SSI treatment (p < 0.001), and 54.5% of patients had SSI diagnosed between 5 pm and 8 am (p < 0.001).¹¹ The current study showed a substantial association between orthosurgery, general surgery, use of implants, regional and local anesthesia, malodor wounds, abdominal and extremity sites of surgery, spine/cranium, ablative, palliative and reconstruction surgeries, glycopeptide antibiotic agents for SSI therapy, aminoglycoside, quinolone and glycopeptide antibiotic agents used for surgical prophylaxis, glycopeptide antibiotic agents prescribed as discharge medication after SSI hospitalization, number of days from surgery to present infection, infection, recurrence, and reinfection states of SSI admission with readmission after being diagnosed with SSI.

The resistance to antibiotic agents is an increasing threat to the treatment of SSI. This was studied by Hope et al. in 2019, which concluded that high resistance of gram-positive isolates to ampicillin (84.3%), oxacillin (81.25%), and ceftriaxone (78.13%) in all the microbial isolates except Enterococci, in a cross-sectional study of 83 patients diagnosed with SSI. Gram-negative isolates showed resistance to ampicillin (100%), ceftriaxone (88.89%), and sulfamethoxazole/trimethoprim (88.89%). Klebsiella species showed resistance toward all the antibiotic agents.¹² In a prospective cohort study of 338 patients admitted with SSI, E. coli showed 100% resistance toward ciprofloxacin, levofloxacin, ceftriaxone, cefuroxime, cefuroxime axetil, and ceftazidime. In addition to E. coli, Klebsiella pneumoniae, Pseudomonas species, A. baumannii, Klebsiella oxytoca, P. mirabilis, and M. morganii were also studied for their resistance trend toward antibiotic agents used for SSI. Most of the isolated strains showed resistance toward penicillin, aminoglycosides, cephalosporins, and quinolones.¹³ A prospective cohort study was carried out in general surgery, obstetrics and gynecology, and orthosurgery departments on 518 patients diagnosed with SSI. The study aimed at briefing the microbial profile of the infected sites. Gram-negative bacilli showed the most resistance toward ampicillin (90.1%), cefazolin (85.9%), and cefepime (61%). The most resistance for gram-positive isolates occurred toward 76.9% penicillin and 56.9% ampicillin.¹⁴ Through the analysis, we could identify MRSA, Staphylococcus species, and E. coli as the most prevalent microbial isolates. They showed maximum resistance toward ciprofloxacin/ofloxacin, cloxacillin, ampicillin/amoxicillin, cefuroxime, cefotaxime/ceftriaxone, and erythromycin.

Readmission in patients can be predicted with a statistical model using effective algorithms according to the variables utilized. A supervised gradient-boosting algorithm was used in predicting readmission after lumbar laminectomy in 26,869 patients by Kalagara et al. in 2019. The model showed an accuracy of 95.33% and a good discrimination ability of AUC = 0.8059. It also showed a positive predictive value of 49.6%.¹⁵

A prediction model using a logistic regression algorithm was put forward in the study using retrospective patient data from the general surgery, neurosurgery, and orthopedics departments, and the model had an acceptable discrimination ability of AUC = 0.77 (AIC: 568; BIC: 722), prediction accuracy: 77.8%.

Studies on risk factors in SSI have been a topic of interest. The study gave a prediction model approach for readmission in a cohort of patients admitted because of SSI. To our knowledge, only a few studies have been published on this concept. Initiatives involving weekly wound surveillance during clinical visits and highly structured discharge counseling are recommended for better outcomes in patients treated for SSI.¹⁶ Limitations of the study are that the authors could not classify superficial and open surgical wounds. Although the model showed an acceptable discriminative capacity, it did not give an excellent result, which can be because of the imbalanced data. The study also had the limitation of being retrospective design, as it can give a chance of information bias which is an inherent character of this study design.

Conclusion

The study identified the factors affecting readmission because of SSI complications. These included demographic, treatment, and hospital-related characteristics. Antibiotic resistance was observed in both gram-positive as well as gram-negative micro-organisms. A prediction model with acceptable discriminative capacity was designed using the relevant factors. Further studies focusing on the readmission of SSI can make an effective contribution to accurately identifying substantial factors contributing to complications in patients diagnosed with SSI.

Footnotes

Acknowledgments

The authors thank the Manipal Academy of Higher Education for giving the resources for the successful completion of the study.

Author Disclosure Statement

All authors report no conflicts of interest relevant to this article.

Authors’ Contributions

Study design, Data collection, Data analysis Manuscript writing: Somakumar. Data collection, Data analysis: Basheer. Data analysis, Manuscript editing: Vijayanarayana K. Data analysis, critical evaluation: Lakshmi R. Critical evaluation: Bhat. Critical evaluation: Rodrigues. Critical evaluation: Menon R. Critical evaluation, Manuscript editing: Raj S. Study design, Manuscript editing, Critical evaluation: Rajesh V.

Funding Information

No funding was received for this article.

Supplementary Material

References

Antimicrobial resistance [Internet]. Available from: https://www.who.int/news-room/fact-sheets/detail/antimicrobial-resistance [Last accessed: October 20, 2024].

Andersen

. Prevention of postoperative wound infections. Prevention and Control of Infections in Hospitals, 2018; 377.

Mukagendaneza

, Munyaneza

, Muhawenayo

, et al. Incidence, root causes, and outcomes of surgical site infections in a tertiary care hospital in Rwanda: A prospective observational cohort study. Patient Saf Surg, 2019; 13(1):10.

Gibbons

, Wang

, Ma

. Determination of pharmaceutical and personal care products in wastewater by capillary electrophoresis with UV detection. Talanta, 2011; 84(4):1163–1168.

Goldberg

, Elazar

, Glatt

, et al. Perioperative interventions to reduce surgical site infections: A review. Aorn J, 2021; 114(6):587–596.

Shah

, Pavey

, Ju

, et al. Evaluation of readmissions due to surgical site infections: A potential target for quality improvement. Am J Surg, 2017; 214(5):773–779.

ssi-checklist-508.pdf [Internet]. Available from: https://www.cdc.gov/nhsn/pdfs/checklists/ssi-checklist-508.pdf [Last accessed: October 20, 2024].

Bursac

, Gauss

, Williams

, et al. Purposeful selection of variables in logistic regression. Source Code Biol Med, 2008; 3(1):17.

Fischer

, Lingsma

, Mheen

PJM

, et al. Is the readmission rate a valid quality indicator? A review of the evidence. PLoS One, 2014; 9(11):e112282.

10.

Napolitano

, Tomassoni

, Cascone

, et al. The Collaborative Working Group. Evaluation of hospital readmissions for surgical site infections in Italy. Eur J Public Health, 2018; 28(3):421–425.

11.

Gibson

, Tevis

, Kennedy

. Readmission after delayed diagnosis of surgical site infection: A focus on prevention using the American college of surgeons national surgical quality improvement program. Am J Surg, 2014; 207(6):832–839.

12.

Hope

, Ampaire

, Oyet

, et al. Antimicrobial resistance in pathogenic aerobic bacteria causing surgical site infections in Mbarara regional referral hospital, Southwestern Uganda. Sci Rep, 2019; 9(1):17299.

13.

Lakoh

, Yi

, Sevalie

, et al. Incidence and risk factors of surgical site infections and related antibiotic resistance in Freetown, Sierra Leone: A prospective cohort study. Antimicrob Resist Infect Control, 2022; 11(1):39.

14.

Deka

, Kalita

, Mahanta

, et al. High prevalence of antibiotic-resistant give bacteria causing surgical site infection in a tertiary care hospital of Northeast India. Cureus, 2020; 12(12):e12208.

15.

Kalagara

, Eltorai

AEM

, Durand

, et al. Machine learning modeling for predicting hospital readmission following lumbar laminectomy. J. Neurosurg.: Spine SPI, 2019; 30(3):344–352.

16.

Miletic

, Taylor

, Martin

, et al. Readmissions after diagnosis of surgical site infection following knee and hip arthroplasty. Infect Control Hosp Epidemiol, 2014; 35(2):152–157.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.02 MB

0.06 MB

0.00 MB