Risk factors for infection in home health care: Analysis of national Outcome and Assessment Information Set data

Abstract

Patients in home health care (HHC), a rapidly growing healthcare sector, are at high risk for infections. This study aimed to identify risk factors for infections among HHC patients using the Outcome and Assessment Information Set (OASIS) data. We used a 5% random sample of the 2013 national OASIS data. Infections were identified if records indicated that patients were hospitalized or received emergency care for one of three types of infections (respiratory, wound site, and urinary tract infection). Multivariate logistic regression models were used to identify risk factors for each individual infection type. The final analysis included 128,163 patients from 8,255 HHC agencies nationwide. Approximately 3.2% of the patients developed infections during their HHC stay that led to hospitalization or emergency care treatment. We found that associations between demographics and infection risk are specific to the type of infection. In general, a history of multiple hospitalizations in past 6 months, comorbidity, having a severe condition at HHC admission, and impaired physical functioning increased HHC patients’ risk of infections. We also identified that HHC patients with caregivers who needed training in providing medical procedure or treatment are at higher risk for wound-site infections. Our findings suggest that patients with underlying medical conditions and limited physical function status are more likely to develop infection. The caregiver’s lack of training in providing needed care at home also places HHC patients at high risk for infection. Education for patients and caregivers should be tailored based on their health literacy level to ensure complete understanding.

1 |. BACKGROUND

Home health care (HHC), also known as “skilled home care,” is a type of home-based care. HHC includes a wide range of services provided in the patient’s home by skilled health care professionals including nurses, physical therapists, occupational therapists, and social workers (Landers et al., 2016). HHC has been one of the fastest growing health care sectors in the United States since the 1970s (Jarvis, 2001) because of Americans’ desire for care at home when possible (U.S. Department of Health and Human Services & Administration on Aging, 2012) and federal and state policies that expanded home-based health care services (Doty, 2010; Wiener, 2013). In 2017, Medicare spent over $17.7 billion on HHC services provided to approximately 3.4 million Medicare beneficiaries nationwide (Medicare Payment Advisory Commission, 2019).

In the U.S., HHC is a significant source of postacute care that is associated with cost savings. A recent study using 2010–2016 national Medicare claims data found that over 38% of Medicare beneficiaries received HHC after hospital discharge (Werner, Coe, Qi, & Konetzka, 2019). Compared with matched controls who were discharged from the hospital with “self-care,” the use of HHC was associated with a mean unadjusted savings of $15,233 per patient, or $6,433 after adjusting for covariates in the 365 days after the index hospital discharge (Xiao, Miller, Zafirau, Gorodeski, & Young, 2018). With the continuing decrease in hospital length of stay and the expanding of the bundled payment models that incentivize for lower-cost care outside of hospitals (Lewin Group, 2018), HHC will continue to grow.

Patients in HHC face special challenges in infection prevention and control (IPC). Unlike facility-based health care, HHC is provided in the patient’s home, an environment designed for daily living and thus with fewer health resources. IPC measures that are common in the hospitals may become difficult to implement at home. The presence of pets and/or pests in patients’ homes also increases the risk for infection (Markkanen et al., 2007). Furthermore, although supervised by HHC clinicians, much of the actual care is provided by patients, caregivers, and/or aides who have limited or no formal IPC training. These factors can increase an HHC patient’s risk for infections. Researchers have studied infections in HHC since the 1990s. In a systematic review of 25 research papers on infections in the HHC setting, the authors found a large variation of infection rates (ranging from 5% to 80%) due to different patient populations studied and how infections were identified and defined (Shang, Ma, Poghosyan, Dowding, & Stone, 2014).

In 2002, infection control experts (Manangan, Pearson, Tokars, Miller, & Jarvis, 2002) highlighted the significance of IPC in HHC; acknowledged the challenges facing HHC in building an infection surveillance system, such as the lack of an easily accessible laboratory for examination of infections; and suggested the use of existing datasets to serve the purpose. These existing datasets included the Missouri Alliance for Home Care, the Outpatient Parenteral Antimicrobial Therapy registry, and the Outcome and Assessment Information Set (OASIS), which is the only one that covers all U.S. states.

Using the national OASIS data, and considering a 48-hr infection incubation period, researchers found that 18% of unplanned hospitalizations in HHC patients were caused by four types of infections: respiratory, wound site, urinary tract, and intravenous [IV] catheter related (Shang, Larson, Liu, & Stone, 2015). While many HHC patients who develop infections during HHC episodes receive antibiotic treatment and recover, those who deteriorate and end up with a hospitalization or emergency department (ED) visit usually have more adverse health outcomes and higher healthcare costs. The focus on infection-related events such as hospitalizations or ED visits stresses the significance of IPC in HHC which, in fact, has been identified as one of the national patient safety goals by the Joint Commission for HHC (The Joint Commission Accreditation of Healthcare Organizations, 2020). The Centers for Medicare and Medicaid Services (CMS) requires HHC agencies to adhere to standard IPC practices and report any infection control breaches to state Departments of Health (Centers for Medicare & Medicaid Services, 2014).

A first step toward effective IPC is to identify risk factors for infection. However, previous studies on this topic are limited by small sample sizes, local scope of inquiry, targeting specific type of infection or subgroups of HHC patients, primarily focusing on underlying diseases or medical treatments, and failing to consider caregivers’ role (Osakwe, Larson, & Shang, 2019; Shang et al., 2014). Informal caregivers are essential for providing care to patients in the home (Cho, Kim, & Lee, 2013). In 2008, the Association for Professionals in Infection Control and Epidemiology, Inc. (APIC) and the Centers for Disease Prevention and Control’s Healthcare Infection Control Practices Advisory Committee published APIC – HICPAC Surveillance Definitions for Home Health Care and Home Hospice Infections, which emphasized the influence of caregivers on HHC infection control (APIC, 2008).

To provide a comprehensive assessment of the risk factors for infections, the objectives of this study were to (a) identify risk factors for three types of infection in HHC patients using the OASIS data and (b) explore the potential role of informal caregivers in prevention of infections in the HHC setting. We hypothesized that HHC patients with more social support, measured by informal caregiver’s availability/readiness, have a lower risk for infection events.

2 |. METHODS

2.1 |. Theoretical framework

We used the Andersen’s Behavioral Model (ABM) for Vulnerable Populations (Andersen, 1968) to guide our study, which was originally developed to assist in understanding why people use health services. The ABM is a comprehensive model that recognizes the dynamic interactions between individual and contextual determinants of health service use and patient outcomes and has been used extensively in health services research (Babitsch, Gohl, & von Lengerke, 2012). Guided by ABM, the determinants in this study include (a) predisposing factors—individual patient attributes such as demographics and socio-structural characteristics that are usually not changeable; (b) need factors—the patient’s self-perceived and/or professionally assessed need such as underlying diseases, current health conditions, and functional status; and (c) enabling factors—resources available to the patient including his/her living arrangement and social support measured by the caregiver’s availability and readiness to provide needed care. Due to the limitations of the data set, the contextual determinants, which refer to HHC agency’s characteristics, were not included in this study; health service use was HHC; and outcome was hospitalization or ED visit due to infection.

2.2 |. Study design

This is a secondary data analysis of a 5% random sample of the 2013 national OASIS data.

2.3 |. Data source

OASIS was originally developed in 1999 and is primarily funded by CMS to enable a standardized, systematic, and comprehensive assessment of HHC patients’ conditions, care quality, and outcomes. It is the mandatory assessment for all adult (≥18 years old) nonmaternity patients receiving skilled HHC by Medicare certified HHC agencies. Using OASIS, HHC patients are assessed by HHC nurses and other clinicians such as therapists at least twice during an HHC episode, including (a) admission to HHC and (b) any change in health status indicated by transferring to an inpatient facility, death, discharge from HHC, or reassessment on the 60th day if none of the above scenarios have happened. Data collected for OASIS are the most comprehensive at HHC admission which is usually done by HHC nurses, including information on a patient’s sociodemographic status, health conditions, physical function, environment, living arrangement, and his/her informal caregiver’s availability and readiness. The comprehensive range of variables and longitudinal assessments of OASIS make it an ideal data source to study risk factors for adverse outcomes in HHC, including infections.

2.4 |. Population and sample

Patients who had complete HHC stays between 01/01/2013 and 12/31/2013 were included in the analysis. For patients who had multiple HHC stays during that period, only the first HHC stay was included.

2.5 |. Variables and measurement

2.5.1 |. Dependent variable: Hospitalization or emergency care due to infection

The dependent variables are infection events defined as a hospitalization and/or ED visit caused by any of the following three types of infections: respiratory, wound site, or urinary tract infection (UTI). We excluded IV catheter-related infections because they are relatively rare in the sample and the key indicator, presence of an IV catheter, is not well measured in OASIS admission data. Each of the infection events was determined by two OASIS items indicating the occurrence and reasons for the hospitalization and/or ED visit. These two OASIS items were collected at the referral to an inpatient facility or discharge from HHC to home. Considering the incubation period of infections and following the updated criteria for home care-onset healthcare-associated infection (McGoldrick, 2019), we included infection events that occurred 2 days after HHC admission to eliminate infections that might have been acquired in other settings and carried to HHC.

2.5.2 |. Individual determinants: Predisposing, enabling, and need factors

Predisposing factors include age (years), sex (female/male), and race/ethnicity (White/Black or African American/Hispanic or Latino/Others).

Need factors are organized in three groups (Table 1).

TABLE 1.

Basic characteristics of home care patients^a

	Total sample (n = 128,163)b	Respiratory infection (n = l,880)c	UTI (n = 1,268)c	Wound-site infection (n = 1,039)c	Comparison group (n = 124,062)d
Predisposing factors
Age (mean ± SD)	77.1 ±11.6	77.8 ±11.5*	79.4 ±11.0**	70.4 ±13.2**	77.1 ±11.6
Age ≥65 (%)	88.7	89.0	92.3**	73.4**	88.8
Female (%)	65.3	57.6**	68.4*	51.8**	65.5
Race/ethnicity (%)
White	79.9	85.5**	85.5**	78.3	79.7
Black or African American	11.1	8.3**	8.8**	12.6	11.1
Hispanic or Latino	6.6	3.8**	4.0**	7.4	6.7
Others	2.5	2.4	1.7**	1.6	2.5
Payer (%)
Medicare (fee-for-service)	77.0	79.8**	77.4	77.3	77.0
Need factors
Shared factors
Patient’s health history or diagnosis
Past inpatient facility discharge in prior 14 days (%)
Short stay acute hospital	46.0	52.5**	41.7**	52.6**	45.9
Skilled nursing facility	16.5	17.5	23.3**	15.4	16.5
Decline in mental, emotional, & behavioral condition	13.1	18.2**	21.8**	10.9	12.9
Multiple hospitalizations in past 6 months	24.7	41.9**	39.1**	36.0**	24.2
Charlson comorbidity index (mean±SD)	0.8 ± 1.2	1.5 ± 1.4**	1.0 ±1.3**	0.9 ± 1.2	0.8 ± 1.2
Patient’s current conditions
Overall status - in fragile or serious condition that may lead to death within a year	30.5	49.9**	47.5**	39.4**	30.0
Smoking	15.5	22.5**	13.4	21.1**	15.4
Cognitive functioning
Alert	56.1	47.3**	38.6**	64**	56.3
Need assistance in specific situations	40.1	46.9**	52.2**	31.3**	39.9
Total dependent	3.9	5.9**	9.2**	4.7**	3.8
Depressione,f	4.9	7.3**	9.0**	5.1	4.8
ADL/IADL
Need assistance in toilet hygiene	32.8	44.4**	55.4**	38.9**	32.4
Chair-fast, unable to ambulate	8.3	14.2**	24.2**	20.4**	8.0
Need assistance in preparing meal	36.9	50.6**	57.9**	39.3	36.5
Medicine managementf
Need assistance in managing oral medications	70.4	81.5**	83.6**	62.8**	70.2
Need assistance in managing injectable medications	12.8	16.4**	17.3**	17.2**	12.7
Infection type specific factors
Respiratory infection specific factors
Having shortness of breath
When walking more than 20 feet, climbing stairs	25.7	19.8**	NA	NA	25.8
With moderate exertion (like walking less than 20 feet)	28.4	33.6**	NA	NA	28.3
With minimal exertion	10.8	23.0**	NA	NA	10.6
At rest	2.4	9.5**	NA	NA	2.3
Receiving oxygen treatment at home	12.1	36.5**	NA	NA	11.8
Receiving continuous/bi-level positive airway pressure at home	2.6	3.3*	NA	NA	2.6
UTI specific factors
Treated for a UTI in past 14 days
No UTI treatment	90.3	NA	69.3**	NA	90.5
Treated for a UTI in past 14 days	9.0	NA	28.7**	NA	8.8
On prophylactic treatment	0.7	NA	2.1**	NA	0.7
Urinary incontinence
No incontinence or urinary catheter	47.9	NA	22.5**	NA	48.2
Having urinary incontinence	49.0	NA	60.2**	NA	48.9
Requires a urinary catheter	3.1	NA	17.4**	NA	2.9
Bowel Incontinence
No bowel incontinence	87.1	NA	70.7**	NA	87.4
<4 times/week	8.5	NA	15.0**	NA	8.3
≥4 times/week	3.9	NA	12.2**	NA	3.8
Having ostomy	0.5	NA	2.2**	NA	0.5
Wound-site infection specific factors
Having an unhealed pressure ulcer at least Stage II (%)	3.7	NA	NA	13.9**	3.5
Having a stasis ulcer (%)	1.3	NA	NA	6.4**	1.3
Surgical wound
No surgical wound	71.5	NA	NA	47.5**	71.4
Observable one or more (%)	24.8	NA	NA	48.1**	24.8
Surgical wound known but not observable due to nonremovable dressing (%)	3.8	NA	NA	4.4**	3.8
Having a skin lesion or open wound excluding ostomy (%)	20.3	NA	NA	46.4**	20.0
Enabling factors
Living arrangement (%)
Living alone	26.8	18.5**	18.5**	23.3*	27.1
Living with other(s) at home	63.3	72.0**	68.9**	67.2**	63.1
Living in a congregate situation	9.9	10.0	12.6**	9.5	9.9
Other variable
Home care length of stay
Median (IQR)	31 (19–55)	17 (8–29)	19 (10–33)	17 (9–30)	31 (19–55)

Abbreviations: ADL, activity of daily living; HMO, health maintenance organization; IADL, instrumental activity of daily living; IQR, interquartile range; SD, standard deviation; UTI, urinary tract infection.

^at-tests and χ² were used to test the group comparisons.

^bTotal study sample after excluding from the initial sample (n = 150,867) those patients with home healthcare episode <3 days (n = 2,493) and patients hospitalized for other reasons after first 2 days (n = 20,211).

^cThere were 4,101 patients accounting for the total of 4,187 infections due to 86 patients having more than one type of infection.

^dComparisons are HC patients who were discharged to home, or continued staying in HHC after the first 60-day episode.

^eDepression is measured by OASIS item M1730, indicated by positive Patient Health Questionnaire-2 (≥3).

^fMissing data include 2.3% missing for “depression” and 0.26% missing for two medication management variables.

^*p < .05.

^**p < .01.

1. Patient’s health history or diagnosis. Related variables included (a) admission source that indicates at which facility the patients stayed 14 days before this HHC admission; (b) multiple hospitalizations (two or more) in past 6 months; and (c) prior and current diagnoses measured by the Charlson comorbidity index using OASIS diagnosis codes for the current HHC admission and previous inpatient stay within the last 14 days. The Charlson comorbidity index, originating in the inpatient setting, is a weighted index of selected comorbid diseases (Charlson et al., 2008; Deyo, Cherkin, & Ciol, 1992), which has been used in the HHC setting previously (Wang, Kearney, Jia, & Shang, 2016). A score of zero indicates no comorbidities; the higher the score, the more comorbidities.

2. Patient’s current conditions. Related variables included (a) patient’s overall health status, indicated by fragile or serious condition at HHC admission that may lead to death within a year; (b) lifestyle-related risk factors such as smoking; (c) decline in mental, emotional, and behavioral condition; (d) cognitive functioning that potentially affects patient’s ability of self-care; (e) depression that could potentially increase the patient’s risk for hospitalization. Depression is measured by the OASIS item M1730, which was added to OASIS-C in 2010 and incorporates the Patient Health Questionnaire-2 for the purpose of depression screening (Kroenke, Spitzer, & Williams, 2003; Sheeran et al., 2010); (f) physical function measured by activities of daily living (ADLs) and instrumental activities of daily living (IADL); and (g) ability to manage oral or injectable medicine.

3. Medical/health conditions that are specific to individual infection type. For example, respiratory status variables are for respiratory infections; elimination status variables such as urinary catheter presence are for UTIs; and integumentary status variables are for wound-site infections.

Enabling factors include a patient’s living arrangement (alone/with others at home/in a congregate situation) and the caregiver’s availability and readiness to provide assistance in seven activities (Table 2)—ADLs, IADLs, medication administration, medical procedure treatments such as wound dressing changes, medical equipment management, supervision and safety due to cognitive impairment, and advocacy or facilitation of patient’s participation in medical care. Each of the seven activities were collapsed into four categories: (a) no assistance needed in the area; (b) caregiver(s) currently provide assistance; (c) caregiver(s) need training; or (d) caregiver(s) not available, or unlikely, or unclear to provide assistance.

TABLE 2.

Caregiver(s) availability/readiness in seven health care tasks^a

	Total (n = 128,163)b	Respiratory infection (n = 1,880)c	UTI (n = 1,268)c	Wound-site infection (n = 1,039)c	Comparison group (n = 124,062)d
ADL assistance (%)
No assistance is needed	10.0	6.1**	5.1**	11.7	10.1
Caregiver(s) provide assistance	62.6	65.6**	62.2	63.0	62.5
Caregiver(s) need training	20.1	22.2*	26.6**	17.3*	20.1
Caregivers not likely/unclear to provide care/no caregiver	7.3	6.1*	6.2	8.0	7.3
IADL assistance (%)
No assistance is needed	3.4	1.7**	2.1*	4.0	3.5
Caregiver(s) provide assistance	83.7	84.6	83.5	84.0	83.7
Caregiver(s) need training	8.3	10.5**	10.4**	7.2	8.3
Caregivers not likely/unclear to provide care/no caregiver	4.5	3.2**	3.9	4.7	4.5
Medication administration (%)
No assistance is needed	27.3	16.6**	13.4**	33.4*	27.6
Caregiver(s) provide assistance	51.4	60.5**	62.9**	46.7*	51.2
Caregiver(s) need training	17.6	19.8*	20.9**	15.3	17.6
Caregivers not likely/unclear to provide care/no caregiver	3.7	3.1	2.8	4.6	3.7
Medical procedures/treatments (%)
No assistance is needed	61.3	61.6	55.8**	18.7**	61.7
Caregiver(s) provide assistance	18.3	20.3*	22.1**	31.1**	18.1
Caregiver(s) need training	13.7	13.0	16.1**	28.9**	13.5
Caregivers not likely/unclear to provide care/no caregiver	6.8	5.1**	6.1	21.4**	6.7
Management of medical equipment (%)
No assistance is needed	72.6	51.7**	65.9**	71.3	73.0
Caregiver(s) provide assistance	17.8	34.3**	22.5**	18.5	17.5
Caregiver(s) need training	8.0	11.9**	10.1**	8.4	7.9
Caregivers not likely/unclear to provide care/no caregiver	1.5	2.1*	1.5	1.8	1.5
Supervision and safety (%)
No assistance is needed	37.6	29.7**	23.5**	42.6**	37.8
Caregiver(s) provide assistance	48.9	56.6**	61.9**	45.9	48.7
Caregiver(s) need training	10.8	11.4	12.5*	8.4	10.8
Caregiver not likely/unclear to provide care/no caregiver	2.8	2.2	2.1	3.1	2.8
Advocacy or facilitation (%)
No assistance is needed	6.7	4**	4.2**	7.4	6.7
Caregiver(s) provide assistance	84.9	87.4**	86.0	83.5	84.9
Caregiver(s) need training	5.4	6.1	7.3**	5.3	5.3
Caregivers not likely/unclear to provide care/no caregiver	3.1	2.5	2.4	3.6	3.1

Abbreviations: ADL, activity of daily living; IADL, intrumental activity of daily living; UTI, urinary tract infection.

^aχ² was used to test the group comparison.

^bTotal study sample after excluding from the initial sample (n = 150,867) patients with home healthcare episode <3 days (n = 2,493) and patients hospitalized for other reasons after first 2 days (n = 20,211).

^cAfter excluding infections occurred 2 days after HHC admission. There were 4,101 patients accounting for the total of 4,187 infections due to 86 patients having more than one type of infection.

^dComparisons are HHC patients who were discharged to home, or continued staying in HHC after the first 60‐day episode.

^*p < .05.

^**p < .01.

2.6 |. Variable selection and statistical analysis

Descriptive statistics were used to summarize sample characteristics as means (standard deviations [SD]) or median (interquartile range [IQR]) for continuous variables and frequency (%, [N]) for categorical variables. t-tests and χ² tests were conducted to compare differences between patients with each infection event and comparison group. The comparison group is HHC patients who were discharged to home or continued staying in HHC after the first 60-day episode. We also recoded variables based on the bivariate analysis of data and collapsed categorical variables by combining nearby categories.

Multivariate logistic regression models were used to identify risk factors for each type of infection event. We checked multicollinearity for each model and finalized the candidate risk factors based on both variance inflation factor and clinical relevance. For example, we removed the variable of prior short stay acute hospital from the models because it is highly correlated with multiple hospitalizations in past 6 months. In addition, some of the candidate factors might be meaningful only for a certain type of infection; therefore, these were only included in those specific models. For example, among all caregiver variables, we chose caregiver’s availability/readiness to provide medical procedures/treatment, specifically focusing on dressing changing for wound-site infection event and chose caregiver’s availability/readiness to assistance on management of equipment (including oxygen, catheter, and so forth) for respiratory infection and UTI events. In this study, patients clustered within the HHC agency. The clustering would affect the accuracy in estimating the standard errors of the parameter estimates. Therefore, we used robust logistic regressions that adjusted for such a clustering effect to identify factors associated with the risk for each infection event (Everitt, Landau, Leese, & Stahl, 2011). We also calculated C-statistics for the full models and each block of factors based upon the ABM. The C-statistics, also called “concordance” statistic, is a frequently used indicator of a logistic regression model’s discrimination power. It is calculated as the area under the receiver operating characteristic curve and ranges from 0.5 to 1, with 0.5 representing that the model is no better than a random chance in predicting an outcome, and 1 indicating the perfect model. A common rule is that a C-statistics >0.7 indicates a good model, and >0.8 indicates a strong model (Hosmer & Lemeshow, 2000). Stata 15.0 (StataCorp, College Station, TX) was used for all data analyses with a significance level of .05.

3 |. RESULTS

3.1 |. Sample characteristics

The initial sample included 150,867 HHC patients from 8,255 HHC agencies nationwide. Excluding patients with home healthcare episode less than 3 days (n = 2,493) and hospitalized for other reasons after first 2 days (n = 20,211), the final sample is 128,163, including patients with three types of infection and the comparisons. Among those 4,101 patients (3.2% of final sample or 2.7% of initial sample) were hospitalized or received ED care due to three types of infection, including 1,880 (1.5%) had respiratory infections, 1,268 (1.0%) had UTIs, and 1,039 (0.8%) had wound-site infections. Eighty-six patients had two types of infection, including 63 with respiratory infection and UTI, 15 with UTI and wound infection, and eight with respiratory and wound infections.

Table 1 presents the characteristics of the final sample, overall and by infection type. The sample was aged 77 years on average (SD = 12), with 88.7% aged 65 years or older. Most of the final sample was female (65.3%), White (79.9%), had Medicare fee-for-service as the primary payer (77.0%), and lived with other(s) at home (63.3%). In the 14 days before the index HHC admission, 46.0% of the sample had an acute care hospital stay, and 16.5% had a skilled nursing facility stay. About a quarter of patients (24.7%) had more than one hospitalization in past 6 months. The median length of HHC stay for the total sample is 33 days (IQR = 19–55); patients who developed infection events had shorter length of HHC stay (Table 1). There are missing data in three variables, including depression (2.3% missing) and management of oral and injectable medications (both 0.26% missing).

HHC patients had different levels of need for assistance on seven tasks, with the highest needs for assistance in IADLs (96.6%). Assistance was provided to patients by caregiver(s) most frequently for advocacy or facilitation (84.9%), followed by IADLs (83.7%), ADLs (62.6%), and medication administration (51.4%). Caregiver training was most needed in ADL (20.1%), followed by medication administration (17.6%) and medical procedures/treatments (13.7%). Only a small proportion of patients had caregiver(s) who were not available/unlikely/unclear to aid in the seven tasks (ranging from 1.5% in management of equipment to 7.3% in ADLs; Table 2). Compared with the comparison group (n = 124,062), patients with a UTI event had more unmet caregiver needs (caregivers need training) on all seven tasks while patients with a respiratory infection event were more likely to need caregiver training on ADL, IADL, medication administration, and medical equipment management. Patients with a wound-site infection event had more unmet caregiver needs than the comparison group on only one task, medical procedures or treatments. However, their unmet needs were more than double that of the comparison group (28.9% vs. 13.5%), and these patients also were three times more likely to have no caregivers to help on this task than the comparison group (21.4% vs. 6.7%).

3.2 |. Identified risk factors for infection-related hospitalizations or emergency department visits

Table 3 presents the risk factors for each of the three types of infection events by predisposing factors (three variables for all models), need factors (13 shared variables, the numbers of individual infection specific factors are three for respiratory infection, three for UTI, and four for wound-site infection), and enabling factors (three variables for respiratory infection and UTI, and two for wound-site infection).

TABLE 3.

Results of multivariate logistic regression

	Respiratory infection model (na = 122,748)		UTI model (na = 122,118)		Wound-site infection model (na = 121,933)
	OR (95% Cl)	p-value	OR (95% Cl)	p-value	OR (95% Cl)	p-value
Predisposing factors
Age	1.01 (1.00–1.01)	.037*	1.01 (1.00–1.01)	.003**	0.98 (0.97–0.98)	.000**
Female	0.81 (0.73–0.89)	.000**	1.23 (1.08–1.41)	.002**	0.77 (0.67–0.88)	.000**
Race/ethnicity
White	Ref		Ref		Ref
Black or African American	0.74 (0.62–0.88)	.001**	0.70 (0.57–0.87)	.001**	1.09 (0.89–1.34)	.425
Hispanic or Latino	0.62 (0.48–0.81)	.000**	0.60 (0.45–0.80)	.000**	1.11 (0.86–1.45)	.424
Others	0.99 (0.73–1.35)	.954	0.54 (0.34–0.86)	.010*	0.69 (0.42–1.16)	.164
Need factors
Shared factors
Patient’s health history or diagnosis
Skilled nursing facility stay in prior 14 days	1.09 (0.95–1.24)	.219	1.24 (1.07–1.45)	.005**	1.06 (0.87–1.28)	.580
Multiple hospitalizations in past 6 months	1.46 (1.32–1.62)	.000**	1.48 (1.29–1.69)	.000**	1.42 (1.23–1.64)	.000**
Charlson comorbidity index	1.20 (1.17–1.24)	.000**	1.10 (1.06–1.15)	.000**	1.04 (0.99–1.09)	.142
Patient’s current conditions
Overall status - in fragile or serious condition that may lead to death within a year	1.28 (1.15–1.42)	.000**	1.14 (1.13–1.47)	.000**	1.42 (1.13–1.51)	.000**
Smoking	1.13 (1.00–1.27)	.048*	0.95 (0.80–1.13)	.565	1.10 (0.93–1.29)	.268
Decline in mental, emotional, & behavioral condition	1.10 (0.97–1.25)	.153	1.14 (0.98–1.33)	.084	0.99 (0.80–1.22)	.914
Cognitive functioning
Alert	Ref		Ref		Ref
Need assistance in specific situations	1.04 (0.93–1.16)	.458	1.18 (1.02–1.35)	.024*	0.89 (0.77–1.04)	.132
Total dependent	1.18 (0.92–1.51)	.190	0.90 (0.68–1.20)	.468	0.84 (0.57–1.25)	.391
Depressionb,c	1.17 (0.96–1.41)	.119	1.19 (0.96–1.48)	.121	0.93 (0.68–1.28)	.664
ADL/IADL
Need assistance in toilet hygiene	1.11 (1.00–1.24)	.060	1.19 (1.02–1.38)	.027*	1.03 (0.88–1.20)	.729
Chair-fast, unable to ambulate	1.43 (1.23–1.67)	.000**	1.69 (1.42–2.01)	.000**	2.15 (1.77–2.60)	.000**
Need assistance in preparing meal	1.22 (1.10–1.36)	.000**	1.27 (1.10–1.46)	.001*	0.98 (0.85–1.13)	.826
Medicine managementc
Need assistance in managing oral Medications	1.08 (0.88–1.33)	.443	0.85 (0.66–1.08)	.185	0.72 (0.71–1.15)	.403
Need assistance in managing injectable medications	1.08 (0.94–1.23)	.270	1.20 (1.01–1.41)	.036*	1.16 (1.06–1.54)	.010*
Infection type specific risk factors
Respiratory infection specific risk factors
Having shortness of breath
No shortness of breath	Ref		NA		NA
When walking more than 20 feet,climbing stairs	1.63 (1.38–1.94)	.000**	NA		NA
With moderate exertion (like walking less than 20 feet)	1.95 (1.67–2.28)	.000**	NA		NA
With minimal exertion	2.44 (2.05–3.92)	.000**	NA		NA
At rest	3.15 (2.50–3.97)	.000**	NA		NA
Receiving oxygen treatment at home	2.12 (1.88–2.41)	.000**	NA		NA
Receiving continuous/bi-level positive airway pressure at home	0.78 (0.60–1.03)	.078	NA		NA
UTI specific risk factors
Treated for UTI in past 14 days
No UTI treatment	NA		Ref		NA
Treated for a UTI in past 14 days	NA		2.64 (2.14–2.86)	.000**	NA
On prophylactic treatment	NA		2.03 (1.32–3.15)	.001**	NA
Urinary incontinence
No incontinence or urinary catheter	NA		Ref		NA
Having urinary incontinence	NA		1.50 (1.49–2.03)	.000**	NA
Requires a urinary catheter	NA		5.53 (5.20–7.98)	.000**	NA
Bowel incontinence					NA
No bowel incontinence	NA		Ref		NA
<4 times/week	NA		1.13 (0.94–1.35)	.193	NA
≥4 times/week	NA		1.39 (1.11–1.73)	.004**	NA
Having ostomy	NA		2.06 (1.32–3.23)	.002**	NA
Wound-site infection specific risk factors
Having an unhealed pressure ulcer at least Stage II (%)	NA		NA		2.01 (1.63–2.48)	.000**
Having a stasis ulcer (%)	NA		NA		3.46 (2.63–4.55)	.000**
Surgical wound
No surgical wound	NA		NA		Ref
Observable one or more	NA		NA		2.32 (1.99–2.69)	.000**
Surgical wound known but not observable due to nonremovable dressing	NA		NA		1.60 (1.14–2.23)	.006**
Having a skin lesion or open wound excluding ostomy (%)	NA		NA		2.14 (1.87–2.45)	.000**
Enabling factors
Living arrangement
Living alone	Ref		Ref		Ref
Living with other(s) at home	1.36 (1.19–1.54)	.000**	1.24 (1.05–1.46)	.009**	0.95 (0.81–1.12)	.565
Living in congregate situation	1.23 (1.01–1.49)	.040	0.78 (0.76–1.22)	.778	1.47 (1.14–1.89)	.003**
Caregiver’s availability and readiness
Medication administration
No assistance is needed	Ref		Ref		NA
Caregiver(s) provide assistance	1.14 (0.92–1.42)	.232	1.54 (1.18–2.02)	.001**	NA
Caregiver(s) need training	0.93 (0.73–1.17)	.523	1.36 (1.01–1.83)	.041*	NA
Caregivers not likely/unclear to provide care/no caregiver	0.92 (0.64–1.31)	.640	1.30 (0.84–2.01)	.246	NA
Medical procedures/treatments
No assistance is needed	NA		NA		Ref
Caregiver(s) provide assistance	NA		NA		3.26 (2.69–3.94)	.000**
Caregiver(s) need training	NA		NA		3.62 (2.97–4.42)	.000**
Caregivers not likely/unclear to provide care/no caregiver	NA		NA		5.54 (4.49–6.83)	.000**
Management of equipment
No assistance is needed	Ref		Ref		NA
Caregiver(s) provide assistance	1.20 (1.06–1.36)	.005	0.90 (0.78–1.05)	.176	NA
Caregiver(s) need training	1.16 (0.98–1.38)	.092	0.92 (0.74–1.14)	.445	NA
Caregivers not likely/unclear to provide care/no caregiver	1.41 (0.99–2.01)	.057	1.09 (0.67–1.78)	.717	NA
C-statistics
Predisposing factors	0.567		0.568		0.663
Need factors	0.748		0.764		0.660
Enabling factors	0.645		0.606		0.734
Full model	0.756		0.771		0.827

Abbreviations: ADL, activity of daily living; CI, confidence interval; IADL, intrumental activity of daily living; OR, odds ratio; UTI, urinary tract infection

^an indicates the sample for the model.

^bDepression is measured by OASIS item M1730, indicated by positive Patient Health Questionnaire-2 (≥3).

^cMissing data include 2.3% missing for “depression,” and 0.26% missing for two medication management variables.

^*p < .05.

^**p < .01.

3.2.1 |. Respiratory infection event

There were a total of 22 variables included in the model, and 13 variables (three predisposing, eight need, and two enabling) significantly increased HHC patients’ risk for respiratory infections. Predisposing factors such as older age and male gender increased HHC patients’ risk for respiratory infections. White patients were more likely to develop a respiratory infection than Black and Hispanic HHC patients. Need factors variables associated with increased risk for respiratory infections included multiple hospitalizations in past 6 months, comorbidity measured by the Charlson comorbidity index, having serious progressive conditions that may lead to death within a year, smoking, chair-fast, and needing assistance in preparing meal. We also identified OASIS items specific to respiratory infections risk. Experiencing shortness of breath when walking more than 20 feet or climbing stairs (odds ratio [OR] = 1.63, 95% confidence interval [CI] = 1.38–1.94), with moderate exertion (OR = 1.95, 95% CI = 1.67–2.28), with minimal exertion (OR = 2.44, 95% CI = 2.05–3.92), or at rest (OR = 3.15, 95% CI = 2.50–3.97) are associated with higher risk for respiratory infection. Receiving oxygen treatment at home is also related to higher respiratory infection risk (OR = 2.12, 95% CI = 1.88–2.41). Among the enabling factors, living with others at home or living in a congregate situation both are associated with higher odds of developing a respiratory infection event compared to living alone; patients receiving assistance in managing equipment were more likely to develop a respiratory infection event than those who did not need such assistance (OR = 1.20, 95% CI = 1.06–1.36).

3.2.2 |. Urinary tract infection

There were a total of 22 variables included in the model, and 17 variables (3 predisposing, 12 need, and 2 enabling) significantly increased HHC patients’ risk of UTI. HHC patients who were older, female, and White were more likely to develop a UTI than other age, gender, and racial/ethnic groups. Need factors that were associated with increased risk for UTI included skilled nursing facility stay before this HHC admission, multiple hospitalizations in the past 6 months, comorbidity measured by the Charlson comorbidity index, having serious progressive conditions that may lead to death within a year, cognitive dysfunction, chair-fast, having difficulty managing toileting hygiene or preparing meal, and inability to manage injectable medication. OASIS items specific to UTI risk were receiving UTI or prophylactic treatment 14 days before this HHC episode (OR = 2.64, 95% CI = 2.14–2.86; OR = 2.03, 95% CI = 1.32–3.15, respectively), having urinary incontinence or requiring a urinary catheter (OR = 1.50, 95% CI = 1.49–2.03; OR = 5.53, 95% CI = 5.20–7.98, respectively), having bowel incontinence less than four times a week (OR = 1.39, 95% CI = 1.11–1.73), or having an ostomy (OR = 2.06, 95% CI = 1.32–3.23). Among the enabling factors, living with others at home is associated with higher odds of developing a UTI event compared to living alone; HHC patients who needed assistance with medication administration (OR = 1.54, 95% CI = 1.18–2.02) and had caregivers who needed training (OR = 1.36, 95% CI = 1.01–1.83) were more likely to develop a UTI event than those who did not need such assistance (Table 3).

3.2.3 |. Wound-site infection

There were a total of 22 variables included in the model, and 12 variables (two predisposing, eight need, and two enabling) significantly increased HHC patients’ risk of wound-site infection. Younger age and male gender were associated with higher odds of developing a wound-site infection event. Need factors related to wound-site infection included multiple hospitalizations in the past 6 months, having serious progressive conditions that may lead to death within a year, chair-fast, and inability to manage injectable medication. OASIS items specific to wound-site infection risk were having an unhealed pressure ulcer at Stage II or higher (OR = 2.01, 95% CI = 1.63–2.48), having a stasis ulcer (OR = 3.46, 95% CI = 2.63–4.55), having an observable surgical wound (OR = 2.32, 95% CI = 1.99–2.69) or an unobservable surgical wound (OR = 1.60, 95% CI = 1.14–2.23), and having a skin lesion or open wound (OR = 2.14, 95% CI = 1.87–2.45). Patients who lived in a congregate setting were more likely to develop a wound-site infection event than those who lived alone. Compared with patients who did not need assistance with medical procedures, such as changing wound dressing, those who needed such help (OR = 3.26, 95% CI = 2.69–3.94) and had caregivers who needed training on medical procedures (OR = 3.62, 95% CI = 2.97–4.42) or had no caregivers to assist (OR = 5.54, 95% CI = 4.49–6.83) were significantly more likely to develop a wound-site infection event (Table 3).

The wound-site infection model shows strong discrimination (C-statistics >0.8) while the other two models show good discrimination (C-statistics >0.7). For both respiratory infection and UTI models, the need factors explain the most variance (C-statistics >0.7) while in wound-site infection model, the enabling factors, including living arrangement and caregiver’s support, explained the most variance with a good discrimination (C-statistics >0.7).

4 |. DISCUSSION

Using the national OASIS data and guided by the Andersen’s Behavioral Model, we identified predisposing, need, and enabling risk factors of three types of infection in HHC leading to hospitalizations or ED visits. Our findings indicate that caregivers play a key role in IPC in HHC.

Our examination of demographic factors suggested that the relationship between these variables and development of infection events are specific to individual types of infection. Some of our findings are consistent with established evidence including the fact that women tend to get an UTI more frequently than men (Turck & Stamm, 1981), and advanced age is associated with higher risk for respiratory infections (Shiono et al., 2007) and UTI (Turck & Stamm, 1981). We found that male HHC patients had higher risk for wound-site and respiratory infections, which may be explained by the fact that a patient’s skin colonization or anatomical differences may be related to gender differences (Cohen et al., 2013). We also found that White HHC patients had higher risk of developing UTI and respiratory infection than Black and Hispanic patients. In our post-hoc analysis, we found these White patients were older, more likely to have a history of UTI treatment, have urinary incontinence, need a urinary catheter, or receive respiratory treatments at home. These underlying conditions can increase the patient’s risk of developing a UTI or respiratory infection.

Estimates from our models indicate that need factors are critical in predicting infection events in HHC as evidenced by model discrimination indicators (i.e., C-statistic). We identified shared or common need risk factors for all infections as well as risk factors that were specific to an infection type. We found that multiple hospitalizations in past 6 months and having a fragile or serious condition that may lead to death within a year increased HHC patients’ risk for all three types of infections. Comorbidity measured by the Charlson comorbidity index was associated with increased risk for respiratory infection and UTI. These factors all indicate poor health status. We also found that, although less than 20% of HHC patients needed injectable medications, difficulty in managing these medications at home can increase risk for UTI and wound-site infection. Although we cannot examine the pathway between this variable and infections without knowing the exact type of injectable medications these patients were taking, this may be indicative of the patient’s self-care ability.

Based on initial findings from a study that found an association between dependency on others for physical functions and UTI risk (Osakwe et al., 2019), we examined the associations between infections and specific functional capacities. We found that mobility limitations and difficulty in maintaining toilet hygiene and preparing meals were associated with increased risk for infections, which is consistent with previous literature (Rogers et al., 2008) and corroborated by qualitative interviews with HHC nurses (Russell & Shang, 2019) who expressed concerns about the HHC patient’s ability to maintain adequate nutrition as it affects their immune system and ability to fight infection. Mobility limitation increases the risk for pressure ulcer development, delays toileting and causes urinary stasis, and can limit lung expansion, thereby increasing a patient’s risk for wound-site infection, UTI and respiratory infection. Physical therapy can help HHC patients to improve their mobility, which can help reduce the risk for infections.

Our study has uniquely identified associations between caregivers and infection events, confirming the important roles caregivers play in HHC, which was emphasized in the 2009 National Research Council workshop on “The Role of Human Factors in Home Health Care” (Schulz & Tompkins, 2009). Over 80% of elderly HHC patients received care from informal caregivers; the assistance provided by these unpaid caregivers ranges from basic ADL/IADL tasks to complex medical and treatment procedures (Cho et al., 2013; Schulz & Tompkins, 2009). Medical/nursing tasks, such as wound care, injections, and managing medical equipment, usually provided by health care providers trained with IPC, may present safety concerns when completed by patients and informal caregivers who may have limited health literacy and IPC knowledge. By focusing on specific tasks that are most relevant to the development of individual infection events, we found patients who needed assistance in or whose caregivers had unmet needs on medication administration and management of medical equipment are at high risk of developing infections. More specific, we found that a caregiver’s availability or readiness for performing medical procedures such as wound care significantly increased an HHC patient’s risk for a wound-site infection. Home care patients have complex wounds that require caregivers to have considerable assessment skills to timely identify signs of infection, properly clean the wound, and apply wound care treatment (Kirkland-Kyhn, Generao, Teleten, & Young, 2018). However, in a study of family caregivers, 66% of family caregivers providing wound care reported challenges, 42% stated learning wound care on their own, 38% needed additional wound care training, 36% received instruction from the hospital staff, and 25% received instructions from HHC nurses (Reinhard, Levine, & Samis, 2012). This underscores the significance of and need for training and educating informal caregivers in wound care in HHC.

The Caregiver Advise Record and Enable (CARE) Act, implemented by most U.S. states and territories, requires training and education on medical tasks to informal caregivers during hospital discharge planning (Coleman, 2016; Reinhard & Ryan, 2017). While the CARE Act is effective in reducing the risk of hospital readmission (Rodakowski et al., 2017), it may be not adequate for caregivers providing assistance to patients needing HHC. A similar but more tailored policy should be considered for HHC.

The engagement of patients and caregivers that has been a part of the HHC plan of care is now critical in the current HHC policy landscape. With a proposed 5% base rate reduction in HHC payment (Medicare Payment Advisory Commission, 2017), fewer HHC visits would be paid for, and more of the care burden would be shifted to the shoulders of informal caregivers. It is critical for HHC clinicians to assess whether caregiver(s) are available and capable of providing the necessary care.

The Patient-Driven Groupings Model (PDGM), which became effective on January 1, 2020, encourages value over volume and reduces the number of therapy visits (CMS, 2020). This new model will prioritize case management and engagement of caregivers in postacute HHC. With PDGM, it will be imperative for case managers to (a) make data-savvy decisions and use evidence-based risk factors to identify patients at high risk for infections, (b) incorporate infection prevention training into individualized care plans, and (c) engage caregivers and provide training programs to reduce the number of home visits while keeping patients safe. Findings from our study can inform policies and guide HHC agencies in improving their performance under the PDGM.

Novel approaches in utilizing technology in HHC are needed to identify signs and symptoms of local and systemic infection, manage complex wound care regimens, and teach infection prevention strategies. For those patients with underlying conditions that increase their risk for infections, such as an unhealed wound or presence of urinary catheter, special assessments are needed to evaluate the knowledge and practices of patients and caregivers related to IPC. While many HHC agencies already focus on patient and family education, training interventions need to be tailored to patients and caregivers based on their health literacy levels, as well as resources available to them. Continuous assessments of patients and their caregivers in the subsequent home visits are also necessary as a patient’s functional status and their caregiver’s availability and competency may change over the HHC episode, which may result in a change in their care plan. Future research should also focus on the burden placed upon informal caregivers under the shifting care model and explore how they can be best supported. With the advance of healthcare information technology, mobile health technology may be one of the solutions by remotely monitoring a patient’s condition and interacting with patients and caregivers through mobile devices. The use of technology and mobile devices to provide training interventions for patients and caregivers related to IPC should also be explored in HHC. Furthermore, time spent on such training and education for caregivers by HHC providers should be appropriately reimbursed, as it will help decrease health costs over time.

4.1 |. Study limitations

First, the range of variables examined in this study was limited to the variables collected. OASIS is the most comprehensive data set of HHC patients, yet it is not collected specifically for infection-related studies. The infection events in our study are based on the assessment of reasons for hospitalization or emergency care by an HHC clinician. Due to the data limitations, we only studied severe infection events and cannot determine the exact incubation period for each infection type. Objective validation of infection-related hospitalizations or ED visits is needed. Future research may link OASIS data with claims data to better examine the infection-related outcomes. With the advances in data science and the connection of data from different healthcare sectors, future researchers may be able to have a precise diagnosis of HHC infections. Researchers can also study risk factors for infections that developed during an HHC episode but did not lead to hospitalization or ED visits, which may not be costly or severe but can still cause physical and psychosocial burden to HHC patients and their caregivers. Second, assessment of caregivers’ availability and readiness in OASIS thus far is not infection-specific. Direct measure of patients’ and caregivers’ knowledge and competency in IPC is needed. Other important socio economic and environmental factors that may be related to IPC, such as income, home environment, patient/caregiver education, and health literacy levels are not measured in OASIS. Future research should examine these factors through direct data collection by survey, observation, or qualitative interviews. Third, OASIS data is not as comprehensive as inpatient discharge data for the calculation of the Charlson comorbidity index and tends to underestimate the patient’s complexity. Linking OASIS to claims data can enhance the index. Finally, we only included an assessment of caregivers’ availability and their capacity to provide the required care when patients are (re)admitted to and discharged from HHC, but not for patients who stay in HHC for more than 60 days; these individuals should also be examined to improve infection control practices.

5 |. CONCLUSION

We used national OASIS data to examine risk factors for hospitalizations or ED visits related to three types of infection in the HHC setting. The comprehensiveness and multiple data collection time points of the OASIS data set allowed us to identify a variety of conditions present upon HHC admission that can increase HHC patients’ high risk for infections. While OASIS is limited by lacking a precise measure of infection and some key factors, policy makers should consider better utilizing OASIS by adding or revising OASIS items to improve the assessment of care quality and outcomes. For HHC clinicians, our findings can help identify patients at higher risk for infections and assist with planning HHC interventions accordingly. Our findings also underscore (a) the significant role that caregivers play in IPC in HHC within the current policy landscape and (b) the need for future research to focus on caregiver burden in HHC. Based on our findings, HHC clinicians should routinely assess infection control knowledge among patients and their caregivers and provide retraining when needed to streamline care planning and enhance patient outcomes in HHC.