Inter-Hospital Transfer Is an Independent Risk Factor for Hospital-Associated Infection

Abstract

Background:

Regionalization of surgical care shifts higher acuity patients to larger centers. Hospital-associated infections (HAIs) are important quality measures with financial implications. In our ongoing efforts to eliminate HAIs, we examined the potential role for inter-hospital transfer in our cases of HAI across a multihospital system.

Hypothesis:

Surgical patients transferred to a regional multihospital system have a higher risk of National Healthcare Safety Network (NHSN)-labeled HAIs.

Patients and Methods:

The analysis cohort of adult surgical inpatients was filtered from a five-hospital health system administration registry containing encounters from 2014 to 2021. The dataset contained demographics, health characteristics, and acuity variables, along with the NHSN defined HAIs of central line-associated blood stream infection (CLABSI), catheter-associated urinary tract infection (CAUTI), and Clostridioides difficile infection (CDI). Univariable and multivariable statistics were performed.

Results:

The surgical cohort identified 92,832 patients of whom 3,232 (3.5%) were transfers. The overall HAI rate was 0.6% (528): 86 (0.09%) CLABSI, 133 (0.14%) CAUTI, and 325 (0.35%) CDI. Across the three HAIs, the rate was higher in transfer patients compared with non-transfer patients (CLABSI: n = 18 (1.3%); odds ratio [OR], 4.79; CAUTI: n = 25 (1.8%); OR, 4.20; CDI: n = 37 (1.1%); OR, 3.59); p < 0.001 for all. Multivariable analysis found transfer patients had an increased rate of HAIs (OR, 1.56; p < 0.001).

Conclusions:

There is an increased risk-adjusted rate of HAIs in transferred surgical patients as reflected in the NHSN metrics. This phenomenon places a burden on regional centers that accept high-risk surgical transfers, in part because of the downstream effects of healthcare reimbursement programs.

Hospital-associated infections (HAIs) are public and important quality measures with substantial financial implications.¹ Healthcare facilities are required to report HAIs to the U.S. Centers for Disease Control and Prevention (CDC) via the National Healthcare Safety Network (NHSN).^2,3 The information provided to NHSN can serve as a valuable guide for the development of infection prevention strategies.

Additionally, HAI rates are used by payers to determine financial incentives for performance.^2,4 The Centers for Medicare & Medicaid Services (CMS) receives required HAI event data through the Hospital Inpatient Quality Reporting Program.⁵ This includes hospital patient safety information regarding annual counts of: central line-associated blood stream infection (CLABSI); catheter-associated urinary tract infection (CAUTI); surgical site infection (SSI) for abdominal hysterectomy and colon procedures; methicillin-resistant Staphylococcus aureus (MRSA) bacteremia; and Clostridioides difficile infection (CDI).¹ These quality measures are evaluated by CMS through the Hospital-Acquired Condition Reduction Program (HACRP).^3,4 The HACRP is a CMS value-based purchasing (VBP) program that reduces overall Medicare fee-for-service (FFS) payments to a hospital based on the institution's nationally ranked performance on mitigating HAIs.^2,6 Previous studies have argued that this methodology for quantifying HAIs may be suboptimal as a quality measure and therefore a suboptimal driver of pay for performance.⁷

Inter-Hospital Transfer

Inter-hospital transfer is increasingly pertinent to the laudable goal of reducing HAIs. The regionalization of healthcare in the United States has driven a coalescence of specialty care into dense urban centers.^8–10 This increase in the centralization of care has resulted in a reliance on an informal and sometimes complex inter-hospital transfer process.^11–13 This can create relative geographical “deserts” where specialty care is absent.^9,14 Transfer occurs when patient needs surpass hospital capabilities resulting in approximately 322,000 annual surgical transfers.^15,16 Patients undergoing inter-hospital transitions of care are vulnerable to increased morbidity, mortality, and resource utilization.^16–18 The transfer process itself introduces delays and disruptions in care and communication, potentially exacerbating these patient vulnerabilities.^8,19,20 Tertiary and quaternary centers receive most of these resource-intensive surgical transfers.^10,16,21 The ramifications of this regionalized system require continued exploration to mitigate the risk to patients and healthcare institutions.

The increased risk of morbidity in transferred surgical patients may extend to HAIs because delays in treatment, prolonged antibiotic exposure, or delay in infection recognition may result in the receiving center being “credited” with the HAI.²² Similarly, sending facilities may not have robust or comprehensive prevention programs in place and patients may be well into developing an iatrogenic event prior to transfer. This deserves examination, because it potentially imbues receiving centers with public and financial penalties. We therefore examined the confluence of inter-hospital transfers and the occurrence of the CDC's NHSN-labeled HAIs that are reported to CMS for the VBP programs.

Hypothesis

We hypothesized that surgical patients transferred to a regional referral center have a higher risk of NHSN-labeled HAIs.

Patients and Methods

Data acquisition

Data for this retrospective cohort analysis were compiled from a health system administration database. The dataset contained all acute care inpatient encounters from 2014 to 2022 across five hospitals that serve urban metropolitan populations in southeastern Michigan. This dataset included validated information regarding demographics, health characteristics, case features, outcomes, and quality and operational metrics. Prior to analysis, personally identifiable information was removed resulting in anonymized encounter information. The project was deemed exempt from review by the Institutional Review Board of the healthcare system.

Data definitions

All data preparation and analyses were performed using version 4.0.3 of the R programming language (R Project for Statistical Computing; R Foundation) within RStudio version 1.2.1335. Analysis variables were classed, labeled, and reviewed for missingness in relation to patient demographics and transfer status. Categories for the variables: racial identity, ethnicity, and preferred language were defined by investigators beyond the stratum contained in the electronic health records based on the U.S. Office of Management and Budget's Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity. The dataset contained chart-abstracted indications for HAI events as defined by the CDC NHSN for reporting to the CMS HACRP. Labelling of HAI events was performed via Epic's Bugsy application for Infection control (Wisconsin) by certified infection control nurses. Surgical site infection for abdominal hysterectomy and colon procedures and MRSA bacteremia were not tracked for analysis due to insufficient cases. The occurrence of CLABSI, CAUTI, and CDI were included in the HAIs variable. Transfer status indicates individuals who were admitted from a separate hospital facility including patients arriving at the referral hospital from both inside and outside the system.

Study setting

The study cohort consisted of discharged adult surgical inpatients encountered by any hospital within a large midwestern health system between January 1, 2014, and December 31, 2021. This includes one quaternary teaching hospital, referred to as the academic referral hospital, with four smaller community hospitals. Admissions between February 1, 2020, and July 1, 2021, were removed because of inundation by severe acute respiratory syndrome coronavirus-2 (SARs-CoV-2) infections, although patients with coronavirus disease 2019 (COVID-19) were not removed as a whole and are included in the analysis for other time periods as appropriate. Surgical patients were defined as individuals who had one or more operating room encounter. No encounters were excluded based on diagnosis or procedure. Following filtering to the study cohort, cases with data missing at random were removed along with any empty factor levels.

Univariable analysis

Univariable analysis was performed on all clinically relevant variables. The analyzed variables were partitioned into tables for demographic information and health characteristics, general outcome attributes, and transfer status. Statistical tests were performed to highlight differences in variables across an HAI event. The statistical test was dependent on variable class with comparisons for binary continuous variables employing a Wilcoxon signed-rank test, and non-binary variables utilized a Kruskal-Wallis H-test. Differences across categorical variables were assessed with Fisher exact tests when computing allowed and were replaced with $χ^{2}$ tests where computationally required. Category percent were reported row wise. To adjust for the substantial number of statistical tests, significance values were corrected with the false discovery rate (FDR) method and reported as Q values. A p value of 0.05 or below was considered statistically significant when partnered to a q value ≤0.05.

Multivariable analysis

The relation between HAIs and patient variables was investigated on a multivariable level with a multinomial logistic regression. The HAIs flag was the dependent variable whereas all clinically important variables from univariable analysis were the independent variables. Redundant variables were determined via variance inflation factors (VIFs) and the performance of Wald's tests and were removed to avoid multicollinearity. The logistic regression reported odds ratios, 95% confidence intervals, and p values and q values. Relative risks were determined from the multivariable model. The absence of HAIs was the reference level for the regression model. A p value ≤0.05 was considered statistically significant when partnered to a q value ≤0.05.

Results

There were 92,832 surgical inpatients from 2014 through 2021 with the primary COVID-19 surge time periods removed. Patients with at least one HAI accounted for 544 (0.6%) of surgical encounters. Demographic composition differed across HAIs status at the univariable level (Table 1). The median age of patients was higher for those that experienced HAIs (65 years; interquartile range [IQR], 56–76 years) compared with patients without HAIs (61 years; IQR, 45–72 years; p < 0.001; Table 1). Individuals who were biologic females (279 patients; 0.51%; p = 0.005), white-identifying (327 patients; 0.52%; p = 0.009), or had private insurance (135 patients; 0.38%; p < 0.001) had lower rates of HAIs (Table 1). Aligning with NHSN guidelines, the inclusion of delivery and obstetrics/gynecology service lines resulted in 13,053 (24%) biologic female encounters which explains the imbalance in the surgical cases across biologic gender (Table 1). Dual-eligible beneficiaries (73 patients; 0.89%; p < 0.001), individuals who live alone (113 patients; 0.75%; p = 0.003), or those with an emergency department visit in the last six months (168 patients; 0.82%; p < 0.001) had an increased prevalence of HAIs (Table 1). All comorbidities comprising the Charlson comorbidity index score, excluding acquired immune deficiency syndrome/human immunodeficiency virus (AIDS/HIV), had higher rates of HAIs (Table 1). The additive count of comorbidities and the Charlson comorbidity index score were increased for patients who had HAIs (median, 3 comorbidities; IQR, 1–4 comorbidities; median, 4 Charlson; IQR, 2–6 Charlson) in relation to patients without HAIs (p < 0.001; p < 0.001; Table 1).

Table 1.

Patient Demographics, Baseline Health, and Encounter Characteristics Across the Occurrence of at Least One HAI

	No CLABSI/CAUTI/CDI^a n = 92,304 (99)	CLABSI/CAUTI/CDI n = 528 (0.6)	p	q
Demographic information
Age, y, median (IQR)	61 (45–72)	65 (56–76)	< 0.001	< 0.001
Biologic gender, n (%)			0.005	0.006
Female	54,401 (99)	279 (0.5)
Male	37,903 (99)	249 (0.7)
Racial identity, n (%)			0.009	0.011
White	63,412 (99)	327 (0.5)
Asian	1,388 (99)	10 (0.7)
Black	20,440 (99)	149 (0.7)
Other	3,290 (99)	17 (0.5)
Unspecified	3,774 (99)	25 (0.7)
Ethnicity, n (%)			0.17	0.18
Not Hispanic or Latinx	84,422 (99)	478 (0.6)
Hispanic or Latinx	3,272 (100)	15 (0.5)
Unspecified	4,610 (99)	35 (0.8)
Insurance provider, n (%)			< 0.001	< 0.001
Private	35,589 (100)	135 (0.4)
Medicare	40,515 (99)	293 (0.7)
Medicaid	15,640 (99)	98 (0.6)
Other	560 (100)	2 (0.4)
Dual-eligible beneficiary, n (%)	8,169 (99)	73 (0.9)	< 0.001	<0.001
Preferred language, n (%)			0.009	0.011
English	87,510 (99)	485 (0.6)
Non-English	3,592 (99)	33 (0.9)
Unspecified	1,202 (99)	10 (0.8)
Health characteristics
Substance use, n (%)	3,417 (99)	18 (0.5)	0.82	0.87
Lives alone, n (%)	15,076 (99)	113 (0.7)	0.003	0.003
History of depression, n (%)	10,805 (99)	62 (0.6)	0.95	0.97
Recent ED visit (< 6 mo), n (%)	20,602 (99)	168 (0.8)	< 0.001	<0.001
Comorbidities
Myocardial infarction, n (%)	10,618 (99)	114 (1.1)	< 0.001	< 0.001
Congestive heart failure, n (%)	554 (99)	8 (1.4)	0.016	0.019
Peripheral vascular disease, n (%)	12,464 (99)	120 (1.0)	< 0.001	< 0.001
Cerebrovasculer disease, n (%)	9,491 (99)	125 (1.3)	< 0.001	< 0.001
Dementia, n (%)	5,048 (99)	76 (1.5)	< 0.001	< 0.001
Chronic obstructive pulmonary disease, n (%)	25,216 (99)	180 (0.7)	< 0.001	< 0.001
Rheumatoid disease, n (%)	3,239 (99)	29 (0.9)	0.018	0.020
Peptic ulcer disease, n (%)	4,836 (99)	50 (1.0)	< 0.001	< 0.001
Mild liver disease, n (%)	3,115 (99)	47 (1.5)	< 0.001	< 0.001
Diabetes mellitus without complications, n (%)	25,647 (99)	195 (0.8)	< 0.001	< 0.001
Diabetes mellitus with complications, n (%)	12,833 (99)	121 (0.9)	< 0.001	< 0.001
Hemiplegia or paraplegia, n (%)	2,208 (98)	49 (2.2)	< 0.001	< 0.001
Renal disease, n (%)	17,618 (99)	191 (1.1)	< 0.001	< 0.001
Cancer, n (%)	14,345 (99)	116 (0.8)	< 0.001	< 0.001
Moderate or severe liver disease, n (%)	2,026 (99)	27 (1.3)	< 0.001	< 0.001
Metastatic solid tumor, n (%)	4,675 (99)	50 (1.1)	< 0.001	< 0.001
AIDS/HIV, n (%)	121 (100)	0 (0)	>0.99	>0.99
Comorbidity count grouped, n (%)			< 0.001	< 0.001
None	30,071 (100)	54 (0.2)
One	22,635 (100)	91 (0.4)
>One	39,598 (99)	383 (1.0)
Comorbidity count, median (IQR)	1 (0–3)	3 (1–4)	< 0.001	< 0.001
Charlson comorbidity index score, Median (IQR)	1 (0–4)	4 (2–6)	< 0.001	< 0.001
Encounter characteristics
Hospital type, n (%)			< 0.001	< 0.001
Academic hospital	37,434 (99)	336 (0.9)
Community hospital	54,870 (100)	192 (0.3)
Service line, n (%)			< 0.001	< 0.001
Surgery	21,691 (99)	176 (0.8)
Orthopedics	21,116 (100)	28 (0.1)
Delivery	10,108 (100)	1 (< 0.1)
Cardiology/vascular/thoracic	9,584 (99)	72 (0.7)
Internal medicine	9,309 (99)	66 (0.7)
Neurosurgery	8,149 (99)	59 (0.7)
Other	5,524 (98)	109 (1.9)
Urology	3,881 (100)	15 (0.4)
Obstetrics/gynecology	2,942 (100)	2 (< 0.1)
Urgent/emergent surgery, n (%)	46,431 (99)	440 (0.9)	< 0.001	< 0.001
Transfer, n (%)	3,156 (98)	76 (2.4)	< 0.001	< 0.001
Mortality risk group, n (%)			< 0.001	< 0.001
Minor	48,952 (100)	26 (< 0.1)
Moderate	20,936 (100)	59 (0.3)
Major	14,738 (99)	160 (1.1)
Extreme	7,678 (96)	283 (3.6)
MS-DRG diagnosis weight, median (IQR)	2 (1–3)	5 (3–11)	< 0.001	< 0.001

p values and q values reported from relevant univariable statistical tests for each variable.

HAI = hospital-associated infection; CLABSI = central line-associated blood stream infection; CAUTI = catheter-associated urinary tract infection; CDI = Clostridioides difficile infection; IQR = interquartile range; ED = emergency department; AIDS/HIV = acquired immune deficiency syndrome/human immunodeficiency virus; MS-DRG = Medicare severity diagnosis related groups.

Most HAIs were experienced by patients with a mortality risk score of extreme (283 patients; 54%) with only 26 (4.9%) of the patients with HAIs having minor risk (Table 1). In general, increased mortality risk groups were associated with the occurrence of HAIs (p < 0.001; Table 1). Patients diagnosed with an HAI had higher medical severity diagnosis-related group (MS-DRG) diagnosis weights (median, 5; IQR, 3–11) compared with those without HAIs (median, 2; IQR, 1–3; p < 0.001; Table 1). Patients with urgent or emergent surgical procedures had proportionally higher rates of HAIs (440 patients; 0.95%) in relation to elective cases (88 patients; 0.20%; p < 0.001; Table 1). The occurrence of HAIs at the academic hospital (336 patients; 0.90%) was higher than the rate at community hospitals (192 patients; 0.35%; p < 0.001; Table 1).

Hospital-associated infections were more frequent in the transfer population (76 patients; 2.4%) compared with non-transfer population (452 patients; 0.51%; p < 0.001; Table 1). Inter-hospital transfers accounted for 3,232 (3.5%) of surgical admissions (Table 2). The rate of CLABSI in transfer patients was 0.56% (18 patients), which differed from the 0.08% (68 patients) for non-transfers (p < 0.001; Table 2). The rate of CAUTI in transfer patients was 0.77% (25 patients) and was higher than the 0.12% (108 patients) rate for non-transfers (p < 0.001; Table 2). The prevalence of CDI was higher in transfer patients (37 patients; 1.1%) compared with non-transfer patients (288 patients; 0.32%; p < 0.001; Table 2).

Table 2.

The Rate of CLABSI, CAUTI, and CDI by Inter-Hospital Transfer Status and Overall

	Non-transfer, n = 89,600 (97)	Transfer, n = 3232 (3.5)	Overall, n = 92,832	p	q
HAIs
CLABSI, n (%)				< 0.001	< 0.001
No	89,532 (97)	3,214 (3.5)	92,746 (100)
Yes	68 (79)	18 (21)	86 (100)
CAUTI, (%)				< 0.001	< 0.001
No	89,492 (97)	3,207 (3.5)	92,699 (100)
Yes	108 (81)	25 (19)	133 (100)
CDI, n (%)				< 0.001	< 0.001
No	89,312 (97)	3,195 (3.5)	92,507 (100)
Yes	288 (89)	37 (11)	325 (100)

p value and q-value reported from relevant univariable statistical test for each HAI.

HAI = hospital-associated infection; CLABSI = central line-associated blood stream infection; CAUTI = catheter-associated urinary tract infection; CDI = Clostridioides difficile infection.

Multivariable analysis relating the presence of HAIs to the absence of HAIs found there were multiple correlations with independent variables (Table 3). The demographics of biologic gender (male odds ratio [OR], 0.77; 95% confidence interval [CI], 0.64–0.93; p = 0.005), and primary language (non-English OR, 1.71; 95% CI, 1.10–2.57; p = 0.013) had relations with HAIs (Table 3). The comorbidities of dementia (OR, 1.40; 95% CI, 1.07–1.81; p = 0.013), and mild liver disease (OR, 1.74; 95% CI, 1.13–2.59; p = 0.009) were associated with the occurrence of HAIs (Table 3). Patients with mortality risk groups of moderate (OR, 4.25; 95% CI, 2.66–6.97; p < 0.001), major (OR, 12.4; 95% CI, 7.99–19.9; p < 0.001), extreme (OR, 29.1; 95% CI, 18.7–46.9; p-value <0.001), and higher MS-DRG diagnosis weight (OR, 1.10; 95% CI, 1.08–1.11; p < 0.001) had increased odds of HAIs (Table 3). Hospital-associated infections were more frequent in urgent or emergent surgical cases (OR, 1.43; 95% CI, 1.11–1.86; p = 0.006) and less frequent at community hospitals (OR, 0.62; 95% CI, 0.50–0.77; p < 0.001; Table 3). Inter-hospital transfer patients were at increased odds for experiencing HAIs (transfer OR, 1.56; 95% CI, 1.18–2.04; p = 0.001) resulting in a relative risk of 1.49 (95% CI, 1.12–1.87; p = 0.010; Table 3).

Table 3.

Multivariate Logistic Regression Relating Patient Demographic, Baseline Health, and Encounter Variables, Including Inter-Hospital Transfer Status, With the Occurrence of at Least One HAI

	OR	95% CI	p	q
Demographic information
Biologic gender
Female
Male	0.77 (0.64–0.93)	0.64–0.93	0.005	0.031
Age, y	1.01 (1.00–1.01)	1.00–1.01	0.16	0.39
Racial identity
White
Asian	1.46 (0.70–2.72)	0.70–2.72	0.27	0.47
Black	1.11 (0.89–1.39)	0.89–1.39	0.35	0.56
Other	0.84 (0.47–1.42)	0.47–1.42	0.53	0.80
Unspecified	0.76 (0.44–1.25)	0.44–1.25	0.29	0.48
Ethnicity
Not Hispanic or Latinx
Hispanic or Latinx	1.04 (0.56–1.79)	0.56–1.79	0.90	0.97
Unspecified	1.36 (0.86–2.08)	0.86–2.08	0.17	0.39
Insurance provider
Private
Medicare	0.98 (0.77–1.25)	0.77–1.25	0.87	0.97
Medicaid	1.36 (1.02–1.82)	1.02–1.82	0.036	0.11
Other	1.55 (0.25–5.07)	0.25–5.07	0.55	0.80
Dual-eligible beneficiary	1.01 (0.75–1.34)	0.75–1.34	0.95	0.98
Preferred language
English
Non-English	1.71 (1.10–2.57)	1.10–2.57	0.013	0.047
Unspecified	0.83 (0.39–1.59)	0.39–1.59	0.60	0.85
Health Characteristics
Comorbidities
Myocardial infarction	0.95 (0.76–1.19)	0.76–1.19	0.68	0.85
Congestive heart failure	1.01 (0.45–1.95)	0.45–1.95	0.98	0.98
Peripheral vascular disease	0.83 (0.66–1.03)	0.66–1.03	0.10	0.26
Cerebrovascular disease	0.95 (0.75–1.20)	0.75–1.20	0.69	0.85
Dementia	1.40 (1.07–1.81)	1.07–1.81	0.013	0.047
Chronic obstructive pulmonary disease	0.96 (0.79–1.16)	0.79–1.16	0.64	0.85
Rheumatoid disease	1.32 (0.87–1.91)	0.87–1.91	0.16	0.39
Peptic ulcer disease	1.22 (0.89–1.64)	0.89–1.64	0.19	0.40
Mild liver disease	1.74 (1.13–2.59)	1.13–2.59	0.009	0.042
Diabetes mellitus without complications	0.97 (0.78–1.19)	0.78–1.19	0.76	0.91
Diabetes mellitus with complications	1.00 (0.76–1.32)	0.76–1.32	0.98	0.98
Hemiplegia or paraplegia	1.39 (0.99–1.93)	0.99–1.93	0.052	0.15
Renal disease	1.05 (0.84–1.30)	0.84–1.30	0.68	0.85
Cancer	1.16 (0.90–1.48)	0.90–1.48	0.25	0.45
Moderate or severe liver disease	0.55 (0.32–0.94)	0.32–0.94	0.030	0.10
Metastatic solid tumor	1.23 (0.86–1.73)	0.86–1.73	0.25	0.45
Substance use	0.71 (0.42–1.13)	0.42–1.13	0.18	0.39
Lives alone	1.15 (0.92–1.43)	0.92–1.43	0.20	0.40
History of depression	0.98 (0.74–1.28)	0.74–1.28	0.90	0.97
Recent ED visit (< 6 mo)	1.02 (0.83–1.24)	0.83–1.24	0.88	0.97
Encounter characteristics
Hospital type
Academic hospital
Community hospital	0.62 (0.50–0.77)	0.50–0.77	< 0.001	< 0.001
Urgent/emergent surgery	1.43 (1.11–1.86)	1.11–1.86	0.006	0.031
Transfer	1.56 (1.18–2.04)	1.18–2.04	0.001	0.009
Mortality risk group
Minor
Moderate	4.25 (2.66–6.97)	2.66–6.97	< 0.001	< 0.001
Major	12.4 (7.99–19.9)	7.99–19.9	< 0.001	< 0.001
Extreme	29.1 (18.7–46.9)	18.7–46.9	< 0.001	< 0.001
MS-DRG diagnosis weight	1.10 (1.08–1.11)	1.08–1.11	< 0.001	< 0.001

p value and q value reported from the regression for each independent variable.

HAI = hospital-associated infection; OR = odds ratio; CI = confidence interval; ED = emergency department; MS-DRG = Medicare severity diagnosis related groups.

Discussion

The occurrence of HAIs was increased in inter-hospital transfer patients at the univariable and multivariable levels (Table 2 and Table 3). Although univariable analysis highlighted differences across HAIs for multiple demographic and health variables, there was a limited number prevailing with multivariable analysis (Table 1 and Table 3). The highlighted increased risk experienced by biologic females for HAIs has been attributed to a higher incidence of CAUTI in this population by previous literature.^23–27 In concurrence with earlier studies Medicaid patients were at increased risk for HAIs.²⁵ Non-privately–insured patients experience increased health disparities, and are associated with increased hospital length of stay, which exacerbates known risks for HAIs.²⁵ Primary language discordance between patients and providers results in longer length of stay and lower patient satisfaction surrounding hospital transfer that possibly explain the increased risk for HAIs experienced by patients with primary languages beyond English.^28–30

In contrast to existing studies, patient age did not exhibit a statistically significate correlation with the occurrence of HAIs (Table 3).^24,31–34 Although mild liver disease increased the risk for HAIs, individuals with moderate to severe liver disease, previously understood to have a heightened risk for complications, experienced an aberrational decreased risk for HAIs (Table 3).^33,35 Intentional differences in treatment and improvements in monitoring and prevention could explain this discrepancy. Further analysis of this hepatic failure subgroup is required to determine their intrinsic risk for developing HAIs. Patients with dementia were at higher risk for HAIs, perhaps reflecting increased complication rates, workloads for providers, and increased lengths of stay that have been observed alongside this patient group (Table 3).^36–38 In agreement with previous research the primary variables for patient acuity, mortality risk score, and MS-DRG diagnosis weight were strongly related to the incidence of HAIs (Table 3).^31,39 Academic institutions, prominent learning environments, were found to be correlated with HAIs in light of controlling for demographics, acuity, and transfer status (Table 3).⁴⁰

There are several possible mechanisms for the observations we make in these data regarding the risk related to inter-hospital transfer. It may be that transferred patients are at higher risk because of delays inherent in the transfer process.²⁰ This may relate to time to definitive surgical intervention, prolonged exposure to antimicrobials or impairment in nutritional and immunologic status. Alternatively, it is possible that some of these HAIs had their genesis at the referring hospital and were merely identified at the receiving hospital, thus giving “credit” to the accepting center. Although there are defined methods for determining this fault in the definition of HAIs, they are by nature arbitrary and imperfect. A further possible mechanism is that clinicians who choose to transfer patients are very astute about risk, including nosocomial infections, and choose to move patients based on their need for complex care and medical resources.

We acknowledge there are limitations to our data that may affect the conclusions one might reasonably draw. Most of these are related to its structure as a retrospective review of an administrative dataset that is not highly curated. Although this approach provides the reality of the relation between transfers and HAIs it falls short of establishing causation. Nor does the analysis narrow down the occurrence of least one HAI to a specific aspect of the transfer process. Furthermore, the quality practices surrounding received transfer patients differ across institutions, limiting the ability to extrapolate these findings. However, the policies experienced by this analysis cohort are reflective of current medical guidelines and include the standard practice of immediately replacing the lines and drains of transferred patients.

An additional concern is our ability to risk adjust for severity of the acute and chronic components of illness in the absence of detailed clinical and laboratory data. These kinds of data such as vital signs, laboratory markers of infection, and laboratory markers of shock are frequently used in risk-adjustment for incidence of infection. We rely on MS-DRG diagnosis weight, the components of the Charlson comorbidity index, and the Mortality Risk Group calculator, which provides mortality risk as a function of primary diagnosis and procedure to adjust for severity of chronic and acute disease. There may well be other discrete variables that are not included that covary with transfer and risk for HAIs. We can surmise that there are material clinical differences between transferred and non-transferred patients that are not adjusted for in our model.

Acknowledging these limitations, our data suggest that serving as a regional referral center for surgical disease imparts a burden on a center as these transferred patients appear to have a roughly 150% increase in risk for reportable HAIs. This, in turn, has negative public image and financial consequence for the center that has made itself available to the patients and their referring providers. Tertiary and quaternary centers are, in effect, accepting downstream financial penalties while serving as a necessary resource from a patient care standpoint. This unintended negative consequence of benchmarking deserves further scrutiny, as it can produce misalignment of incentives for financial health and optimal patient care.

Conclusions

Inter-hospital transfer is an independent risk for the occurrence of at least one HAI in surgical patients. Transfer patients exhibit a combined five-fold increase in the rate of HAIs. With adjustment for pertinent variables, transfer patients have a 1.5 greater risk of HAIs relative to non-transfer patients. This medical process and the associated risk of infection occurs prior to the final hospital encounter and results in suboptimal patient care alongside potential financial and public penalties. Comprehensive quality improvement initiatives should target both the upstream and downstream facets of patient care. Improvements to the inter-hospital transfer process likely results in deceased rates of HAIs for receiving medical centers.

Footnotes

Acknowledgments

The authors would like to thank Henry Ford Health and the patient community for providing resources and support towards this research.

Authors' Contributions

Conceptualization (lead): Gardner. Conceptualization (supporting): Johnson. Methodology (lead): Gardner. Methodology (supporting): Johnson. Methodology: Rubinfeld. Formal analysis: Gardner. Data curation: Gardner. Writing—original draft: Gardner. Writing—review and editing: Johnson. Writing—review and editing (supporting): Rubinfeld, Gupta. Visualization: Gardner. Validation: Rubinfeld. Resources: Rubinfeld. Supervision (lead): Johnson. Supervision (supporting): Rubinfeld. Project administration (lead): Johnson. Project administration (supporting): Gupta. Funding acquisition: Johnson.

Funding Information

This research was made possible by the Dr. Roger Smith Surgical Research Endowment at Henry Ford Hospital. The authors have no further funding to disclose regarding this research.

Author Disclosure Statement

The authors have no conflicts of interest to disclose regarding this research.

References

Division of Healthcare Quality Promotion (DHQP). National Healthcare Safety Network (NHSN) Patient Safety Component Manual. 2021.

Centers for Medicare and Medicaid Services. Hospital Value-Based Purchasing Program. Available from: https://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/HospitalQualityInits/Hospital-Value-Based-Purchasing [Last accessed: April 19, 2022].

Centers for Medicare and Medicaid Services. Hospital-Acquired Condition Reduction Program. Available from: https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/HAC-Reduction-Program [Last accessed: April 19, 2022].

Centers for Medicare and Medicaid Services. Hospital-Acquired Condition Reduction Program (HACRP) Overview. Available from: https://qualitynet.cms.gov/inpatient/hac [Last accessed: April 19, 2022].

Healthcare Facility HAI Reporting Requirements to CMS via NHSN-Current or Proposed Requirements. 2019.

The NHSN Standardized Infection Ratio (SIR): A Guide to the SIR. 2021.

Young

, Knepper

, Moore

, et al. Surgical site infection after colon surgery: National Healthcare Safety Network risk factors and modeled rates compared with published risk factors and rates. J Am Coll Surg, 2012; 214(5):852–859; doi: 10.1016/j.jamcollsurg.2012.01.041

Mueller

, Fiskio

, Schnipper

. Interhospital transfer: Transfer processes and patient outcomes. J Hosp Med, 2019; 14(8):486–491; doi: 10.12788/jhm.3192

Diaz

, Schoenbrunner

, Pawlik

. Trends in the geospatial distribution of inpatient adult surgical services across the United States. Ann Surg, 2021; 273(1):121–127; doi: 10.1097/SLA.0000000000003366

10.

França

, McManus

. Transfer frequency as a measure of hospital capability and regionalization. Health Serv Res, 2017; 52(6):2237–2255; doi: 10.1111/1475-6773.12583

11.

Iwashyna

. The incomplete infrastructure for interhospital patient transfer. Crit Care Med, 2012; 40(8):2470–2478; doi: 10.1097/CCM.0b013e318254516f

12.

Herrigel

, Carroll

, Fanning

, et al. Interhospital transfer handoff practices among US tertiary care centers: A descriptive survey. J Hosp Med, 2016; 11(6):413–417; doi: 10.1002/jhm.2577

13.

Reichheld

, Yang

, Sokol-Hessner

, et al. Defining best practices for interhospital transfers. J Healthc Qual, 2021; 43(4):214–224; doi: 10.1097/JHQ.0000000000000293

14.

Lesher

, Gavrilova

, Ruggiero

, et al. Surgery and the Smartphone: Can technology improve equitable access to surgical care?. J Surg Res, 2021; 263:1–4; doi: 10.1016/j.jss.2020.12.066

15.

Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUPnet). Available from: https://hcupnet.ahrq.gov/ [Last accessed: April 19, 2022].

16.

Ingraham

, Wang

, Havlena

, et al. Factors associated with the interhospital transfer of emergency general surgery patients. J Surg Res, 2019; 240:191–200; doi: 10.1016/j.jss.2018.11.053

17.

Newton

, Fralic

. Interhospital transfer center model: Components, themes, and design elements. Air Med J, 2015; 34(4):207–212; doi: 10.1016/j.amj.2015.03.008

18.

Teng

, Davis

, Rosengart

, et al. Assessment of hospital characteristics and interhospital transfer patterns of adults with emergency general surgery conditions. JAMA Netw Open, 2021; 4(9):e2123389; doi: 10.1001/jamanetworkopen.2021.23389

19.

Lucas

, Ejaz

, Haut

, et al. Interhospital transfer and adverse outcomes after general surgery: Implications for pay for performance. J Am Coll Surg, 2014; 218(3):393–400; doi: 10.1016/j.jamcollsurg.2013.11.024

20.

Kummerow Broman

, Hayes

, Kripalani

, et al. Interhospital transfer for acute surgical care: Does delay matter?. Am J Surg, 2016; 212(5):823–830; doi: 10.1016/j.amjsurg.2016.03.004

21.

Mueller

, Zheng

, Orav

, et al. Interhospital transfer and receipt of specialty procedures. J Hosp Med, 2018; 13(6):383–387; doi: 10.12788/jhm.2875

22.

Babu

, Nahed

, Demoya

, et al. Is trauma transfer influenced by factors other than medical need? An examination of insurance status and transfer in patients with mild head injury. Neurosurgery, 2011; 69(3):659–667; discussion 667; doi: 10.1227/NEU.0b013e31821bc667

23.

, Song

, Xu

, et al. Risk factors for catheter-associated urinary tract infection among hospitalized patients: A systematic review and meta-analysis of observational studies. J Adv Nurs, 2019; 75(3):517–527; doi: 10.1111/jan.13863

24.

Puro

, Joseph

, Zengul

, et al. Predictors of hospital-acquired Clostridioides difficile infection: A systematic review. J Healthc Qual, 2020; 42(3):127–135; doi: 10.1097/JHQ.0000000000000236

25.

Elman

, Williams

, Bearden

, et al. Healthcare-associated urinary tract infections with onset post hospital discharge. Infect Control Hosp Epidemiol, 2019; 40(8):863–871; doi: 10.1017/ice.2019.148

26.

Stamm

. Scientific and clinical challenges in the management of urinary tract infections. Am J Med, 2002; 113(Suppl 1A):1S-4S; doi: 10.1016/s0002-9343(02)01053-7

27.

Foxman

. Epidemiology of urinary tract infections: incidence, morbidity, and economic costs. Am J Med, 2002; 113(Suppl 1A):5S-13S; doi: 10.1016/s0002-9343(02)01054-9

28.

Diamond

, Izquierdo

, Canfield

, et al. A systematic review of the impact of patient-physician non-English language concordance on quality of care and outcomes. J Gen Intern Med, 2019; 34(8):1591–1606; doi: 10.1007/s11606-019-04847-5

29.

Rayan

, Admi

, Shadmi

. Transitions from hospital to community care: the role of patient–provider language concordance. Isr J Health Policy Res, 2014; 3(1):24; doi: 10.1186/2045-4015-3-24

30.

Jacobs

, Sadowski

, Rathouz

. The impact of an enhanced interpreter service intervention on hospital costs and patient satisfaction. J Gen Intern Med, 2007; 22(2_suppl):306–311; doi: 10.1007/s11606-007-0357-3

31.

Bignardi

. Risk factors for Clostridium difficile infection. J Hosp Infect, 1998; 40(1):1–15; doi: 10.1016/s0195-6701(98)90019-6

32.

Barbut

, Petit

. Epidemiology of Clostridium difficile-associated infections. Clin Microbiol Infect, 2001; 7(8):405–410; doi: 10.1046/j.1198-743x.2001.00289.x

33.

Moffatt

CRM

, Kennedy

, O'Neill

, et al. Bacteraemia, antimicrobial susceptibility and treatment among Campylobacter-associated hospitalisations in the Australian Capital Territory: A review. BMC Infect Dis, 2021; 21(1):848; doi: 10.1186/s12879-021-06558-x

34.

Zilberberg

, Nathanson

, Sadigov

, et al. Epidemiology and outcomes of Clostridium difficile-associated disease among patients on prolonged acute mechanical ventilation. Chest, 2009; 136(3):752–758; doi: 10.1378/chest.09-0596

35.

Vidal

, Torre-Cisneros

, Blanes

, et al. Bacterial urinary tract infection after solid organ transplantation in the RESITRA cohort. Transpl Infect Dis, 2012; 14(6):595–603; doi: 10.1111/j.1399-3062.2012.00744.x

36.

Zelnik Yovel

, Glotsman

, Love

, et al. Indwelling urinary catheter and dementia are associated with enterococcal urinary tract infections in hospitalized patients. Isr Med Assoc J IMAJ, 2021; 23(11):708–713.

37.

Bail

, Goss

, Draper

, et al. The cost of hospital-acquired complications for older people with and without dementia: A retrospective cohort study. BMC Health Serv Res, 2015; 15:91; doi: 10.1186/s12913-015-0743-1

38.

Szlejf

, Farfel

, Curiati

, et al. Medical adverse events in elderly hospitalized patients: A prospective study. Clin Sao Paulo Braz, 2012; 67(11):1247–1252; doi: 10.6061/clinics/2012(11)04

39.

Rodríguez-Acelas

, de Abreu Almeida

, Engelman

, et al. Risk factors for health care-associated infection in hospitalized adults: Systematic review and meta-analysis. Am J Infect Control, 2017; 45(12):e149–e156; doi: 10.1016/j.ajic.2017.08.016

40.

Tan

, Wiseman

, Betihavas

, et al. Patient, provider, and system factors that contribute to health care-associated infection and sepsis development in patients after a traumatic injury: An integrative review. Aust Crit Care Off J Confed Aust Crit Care Nurses, 2021; 34(3):269–277; doi: 10.1016/j.aucc.2020.08.004