Health Care Utilization and Cost Associated With Urinary Tract Infections in a Privately Insured Spinal Cord Injury Population

Abstract

Background

Urinary tract infections (UTIs) are the most common secondary medical complication following spinal cord injury (SCI), significantly impacting health care resource utilization and costs.

Objectives

To characterize risk factors and health care utilization costs associated with UTIs in the setting of SCI.

Methods

IBM’s Marketscan Database from 2000–2019 was utilized to identify individuals with traumatic SCI. Relevant ICD-9 and ICD-10 codes classified individuals into two analysis groups: having ≥ 1 UTI episode or no UTI episodes within 2 years following injury. Demographics (age, sex), insurance type, comorbidities, level of injury (cervical, thoracic, lumbar/sacral), and health care utilization/payments were evaluated.

Results

Of the 6762 individuals retained, 1860 had ≥ 1 UTI with an average of three episodes (SD 2). Younger age, female sex, thoracic level of injury, noncommercial insurance, and having at least one comorbidity were associated with increased odds of UTI. Individuals with a UTI in year 1 were 11 times more likely to experience a UTI in year 2. As expected, those with a UTI had a higher rate and associated cost of hospital admission, use of outpatient services, and prescription refills. UTIs were associated with 2.48 times higher cumulated health care resource use payments over 2 years after injury.

Conclusions

In addition to bladder management-related causes, several factors are associated with an increased risk of UTIs following SCI. UTI incidence substantially increases health care utilization costs. An increased understanding of UTI-associated risk factors may improve the ability to identify and manage higher risk individuals with SCI and ultimately optimize their health care utilization.

Introduction

Traumatic spinal cord injury (SCI) impairs sensorimotor and autonomic function, resulting in a constellation of life-long health complications.^1,2 As a result of increased multi-morbidity, persons with SCI frequently interact with the health care system at a higher rate than the general population.³ Long term, SCI poses a substantial personal and economic burden, with estimated lifetime costs ranging between $1.2M and $5.1M; this does not include indirect costs associated with lost wages, benefits, and productivity.⁴ Hospitalization due to secondary health complications accounts for one of the primary drivers of cost following SCI.⁵ During the first year of injury, approximately 26% to 39% of individuals with SCI are readmitted to the hospital.^5–8 More recently, it is estimated that the rate of rehospitalization within the first year of injury can be as high as 45% and even greater (63%) for those with injuries to the upper cervical spinal cord (C1–C4).⁹ Persons with SCI experience twice as many hospitalizations as the general population, with the highest rates of readmission occurring within the first 5 years after injury.^6,10 Genitourinary pathologies,⁴ such as urinary tract infections (UTIs), are the primary reasons for rehospitalization in the first year following SCI.^8,11–13 UTIs are a common chronic urologic problem in the outpatient setting and can persist long term, decreasing the overall quality of life after SCI.¹⁴

The prevalence of UTIs in the SCI population is high, with about 2.5 episodes per year.¹⁵ The inability to effectively empty the bladder and the use of a urinary catheter increase the risk of UTIs,^16,17 which are the single most common secondary medical complication after SCI.^18,19 Not only do UTIs account for an increased rate of emergency department visits, but they are also associated with extended hospital lengths of stay, predisposing individuals to nosocomial infections.¹³ The objective of this study was to determine the factors associated with an increased risk of UTI hospital readmission and to evaluate the health care utilization costs associated with UTIs in the setting of a privately paid insured SCI population. Understanding the incidence of rehospitalization can provide insights into the timing and development of more preventative measures to help decrease health care cost utilization and improve the quality of life for persons with SCI.

Methods

Data source

In this retrospective study, IBM MarketScan database records from 2000–2019 were acquired from Truven Health Analytics - IBM Watson Health and described previously.²⁰ Briefly, these data are from a set of large employers, health plans, government, and public organizations. They are grouped into Commercial Claims, Medicaid, and Medicare, are available to researchers for a fee, and have been used in the medical field since 1990.²⁰ They represent the health services of enrollees as they interact with the health care system, including hospital admissions, outpatient visits and services, outpatient prescribed and inpatient medication use, and laboratory and dental care procedures performed. The data also include information on the use of short- and long-term disability, workers’ compensations, and self-reported data related to medical conditions. Each individual has a unique identifier, which can track their trajectory. A neurological/neurosurgical custom database covering 63,354 incident SCI cases was used in this study. For each encounter, demographics are noted; medical conditions and procedures are recorded using diagnostic and procedure codes (International Classification of Diseases, 9th and 10th Revisions, Clinical Modification [ICD-9-CM before October 2015, ICD-10 after October 2015], Current Procedural Terminology, 4th edition [CPT-4]). Medications are coded with the National Drug Code (NDC) and the medication names. Although a convenience sample, the data are vast and cover the entire United States. This study used de-identified data from a national administrative database, thus informed consent was not required. The study was approved by the University of Louisville Institutional Review Board (IRB #10.0559).

Study cohort selection

From the inpatient database, cases of primary traumatic SCI were extracted using the ICD-9-CM codes 806, 952, and ICD-10-CM codes S14, S24, S34 with a defined level of injury. The primary field was chosen to ensure data were restricted to incident cases. For this reason, 12 months of look-back data were screened to exclude anyone with any claim of SCI tetraplegia or paraplegia (ICD-9-CM: 3440, 3441; ICD-10-CM: G82). Individuals with claims of UTI (ICD-9-CM: 590, 595, 598.0, 599.0; ICD-10-CM: N10-N16, N288.4-N288.6, N30, N351, N37, N390) during this time were also excluded to avoid evaluating UTIs that may be related to previous infections. Those with less than 12 months of continuous enrollment before injury were excluded. Preinjury enrollment time was calculated as the difference between the start enrollment date (or first claim in the data if missing) and the injury hospitalization admission date. The study cohort was restricted to individuals 18 years or older. Note that the classification of an individual’s injury using the American Spinal Injury Association Impairment Scale (AIS) by the International Standards for Neurological Classification of Spinal Cord Injury (ISNCSCI)^21,22 is unavailable for these data.

UTI episodes

Claims occurring up to 2 years postdischarge from hospitalization due to SCI were screened to flag those claims related to UTIs using ICD-9 and ICD-10 codes defined previously. Individuals who did not have 2 years postinjury continuous enrollment were excluded. Postinjury follow-up time was calculated as the difference between the end enrollment date (or last claim date if missing) and the injury discharge date. Those retained were classified into two analysis groups depending on whether they experienced at least one UTI episode or not. An episode was defined as a period of a month following a UTI claim. Any claim falling within this period was considered part of the episode. Any claim occurring more than a month after a previous one was considered to belong to a different episode.

Individual characteristics

Demographics (age, sex) and insurance type (Commercial, Medicare, Medicaid) were noted at the SCI hospitalization admission. The burden of comorbidities was accounted for by the Elixhauser score²³ using the Quan et al.²⁴ adaptation to ICD-9 and ICD-10 codes present in the SCI hospitalization. Level of injury (cervical, thoracic, lumbar/sacral) was defined from the SCI ICD-9-CM and ICD-10 codes’ decimal digits.

Outcomes

Health care utilization in hospital resources, outpatient services, and associated payments were the primary outcomes of interest. We evaluated hospital stays solely due to UTIs by restricting the data to those whose primary diagnosis was UTI. The number of occurrences and the hospital length of stay were also evaluated. The number of outpatient services (including all services provided in an outpatient facility, such as clinic visits, medical labs, and imaging) was totaled. Payments encompassed all sources (insurance, copay, co-insurance) and were adjusted to 2018 US dollars using the medical component of the Consumer Price Index, accessible through the US Bureau of Labor Statistics. Outcomes were studied during the first 2 years following injury discharge.

Statistical analysis

Characteristics and outcomes were presented with summary statistics and compared between individuals who had at least one UTI and those who did not with the Wilcoxon rank sum test for continuous variables, as they all failed the normality test per Kolmogorov-Smirnov test, and chi-square test for categorical variables. UTI risk, in the first 2 years following injury, was evaluated with a logistic regression model, including all characteristics as independent variables. The effect was presented with odds ratio (OR) with an associated 95% confidence interval (95% CI), and the concordance statistic (C-statistic) was provided. A conditional logistic regression was used to analyze the odds of having UTI in year 2 not having had one in year 1 versus not having UTI in year 2 having had one in year 1. A separate logistic regression model was used to evaluate the risk of UTI in the chronic phase in relation to having UTI in the first year following injury. To evaluate the health care utilization and cost associated with UTI, the outcomes were compared between the two groups. Because the characteristics were unevenly distributed between the two groups, the inverse probability of treatment weight (IPTW) technique^25–27 was used first to balance the characteristics, eliminating the confounding variables due to observed differences. In this method, a propensity score (PS) was calculated using logistic regression with UTI in 2 years as the dependent variable and all the characteristics as the independent variables. For each individual, the weight was calculated as the sample size adjusted inverse of the propensity score of the group one belongs to, such that for those in the UTI group and for those in the non-UTI group. The balance after IPTW was evaluated with p values from the IPTW-weighted Kruskal-Wallis test for continuous variables and chi-square test for categorical variables. Then, within the resulting pseudo-cohort, in which characteristics were evenly distributed,²⁸ health care utilization rates were compared with the IPTW-weighted chi-square test, and payments, highly right-skewed, were evaluated with IPTW-weighted linear regression on log-transformed values. The estimate of health care use and payment attributable to UTI per individual was estimated with the ratio of least-square mean difference (estimate ratio [ER]) with associated 95% CI. The significance level was set to 0.05, and all tests were two-sided. Analyses were performed in SAS 9.4 (SAS Institute Inc., Cary, NC).

Results

Study sample description

In this study, 6762 individuals with traumatic SCI were included (Table 1). At the time of injury, the average age was 53 years (SD 21) with an approximately even distribution of sex (56% males, 44% females). The majority of individuals had commercial insurance (52%). Most injuries were sustained at the cervical spine (44%), followed by the thoracic spine (33%), and 66% had at least one comorbidity. Within 2 years following injury, 1860 individuals had an average of three UTI episodes (SD 2) (Table 2). Compared to those without any UTIs, this subset of the overall cohort was younger (49 vs. 54 years old, p <.0001) at the time of injury with a higher proportion of females (48% vs. 43%, p = .0007; Table 1). The most prevalent level of injury in both the non-UTI and UTI subsets was cervical (44%). However, the UTI group included more individuals with thoracic (38% vs. 30%) levels of injury, less lumbar/sacral (18% vs. 24%) injuries (p < .0001), and at least one comorbidity (74% vs. 62%, p <.0001) relative to the non-UTI subset. The distribution of insurance was different among the two groups (p < .0001), with the UTI group comprised of more individuals on Medicaid (30% vs. 19%) and fewer on Medicare (23% vs. 27%) and Commercial insurance (46% vs. 54%) (p < .001).

Table 1.

Clinical characteristics

			Raw data			IPTW-weighted data
		Traumatic SCI	No UTI	UTI		No UTI	UTI
Variables		(n = 6762)	(n = 4902)	(n = 1860)	p value	(n = 4896)	(n = 1874)	p value
Age	Mean (SD)	53 (21)	54 (20)	49 (23)	<.0001	53 (20)	53 (22)	.2394
	Median (IQR)	53 [37 – 68]	54 [41 – 68]	49 [26 – 66]		53 [38 – 68]	55 [36 – 71]
	Range, min-max	18 to 100	18 to 100	18 to 99		18 to 100	18 to 99
Sex	Male, n (%)	3767 (56%)	2793 (57%)	974 (52%)	.0007	2713 (55%)	1009 (54%)	.2443
	Female, n (%)	2995 (44%)	2109 (43%)	886 (48%)		2183 (45%)	865 (46%)
Insurance	Commercial, n (%)	3528 (52%)	2666 (54%)	862 (46%)	<.0001	2554 (52%)	974 (52%)	.9067
	Medicaid, n (%)	1489 (22%)	926 (19%)	563 (30%)		1071 (22%)	405 (22%)
	Medicare, n (%)	1745 (26%)	1310 (27%)	435 (23%)		1270 (26%)	496 (26%)
Elixhauser index	No comorbidity, n (%)	2314 (34%)	1838 (37%)	476 (26%)	<.0001	1678 (34%)	650 (35%)	.7441
	One or more, n (%)	4448 (66%)	3064 (63%)	1384 (74%)		3218 (66%)	1224 (65%)
Level of injury	Cervical, n (%)	2981 (44%)	2162 (44%)	819 (44%)	<.0001	2160 (44%)	827 (44%)	.9248
	Thoracic, n (%)	2195 (32%)	1491 (30%)	704 (38%)		1585 (32%)	599 (32%)
	Lumbar/Sacral, n (%)	1586 (23%)	1249 (25%)	337 (18%)		1151 (24%)	448 (24%)

Note: IPTW = inverse probability of treatment weight; SCI = spinal cord injury; UTI = urinary tract infection.

Table 2.

Number of UTI episodes among those who had at least one UTI within 2 years of injury

	UTI (n = 1860)
Mean ± SD	3 ± 2
Median [IQR]	2 [1 – 3]
Range, min-max	1 to 14

Note: IQR = interquartile range; UTI = urinary tract infection.

Factors associated with UTI

All the measured descriptors were significantly associated with the risk of having a UTI (Table 3; model C-statistic = 0.64). In particular, each year younger was associated with 2.1% increased odds of having a UTI episode (OR, 1.021; 95% CI, 1.017–1.026). There were 31% higher odds of developing a UTI for females (OR, 1.31; 95% CI, 1.17–1.47), and 24% and 71% higher odds for those with a thoracic level of injury compared to cervical and thoracic compared to lumbar/sacral (OR, 1.24; 95% CI, 1.10–1.41; and OR, 1.71; 95% CI, 1.46–1.99). Those individuals on Medicaid and Medicare had respectively 50% and 78% higher odds of having at least one UTI in 2 years following injury than those with Commercial insurance (OR, 1.50; 95% CI, 1.31–1.72; and OR, 1.78; 95% CI, 1.47–2.15). The risks for Medicaid and Medicare were comparable. Having at least one comorbidity at the time of injury was associated with 80% increased odds of UTI (OR, 1.80; 95% CI, 1.59–2.04).

Table 3.

Factors associated with developing a UTI within 2 years following injury

Variables		OR (95% CI)
Age	1 year younger	1.021 (1.017–1.026)
Sex	Female	1.31 (1.17–1.47)
Insurance	Medicaid vs. Commercial	1.5 (1.31–1.72)
	Medicare vs. Commercial	1.78 (1.47–2.15)
	Medicaid vs. Medicare	0.84 (0.68–1.04)
Elixhauser index	One or more vs. none	1.8 (1.59–2.04)
Level of injury	Thoracic vs. Cervical	1.24 (1.1–1.41)
	Lumbar/Sacral vs. Cervical	0.73 (0.63–0.84)
	Thoracic vs. Lumbar/Sacral	1.71 (1.46–1.99)

Note: 95% CI = 95% confidence interval; OR = odds ratio; UTI = urinary tract infection.

In the first year after injury, 1437 (21%) had at least one UTI episode (Figure 1). In the second year, a total of 1139 (17%) had at least one UTI episode. Four hundred and twenty-three (6%) of those who did not have a UTI in the first year had a UTI in the second year. In the first year, 721 (11%) of those who had a UTI did not have a UTI in the second year. This represents a 41% lower likelihood of transitioning from no-UTI to UTI compared to transitioning from UTI to no-UTI (OR, 0.59; 95% CI, 0.52–0.66; p < .0001). However, a total of 50% of individuals who had a UTI in year 1 also had a UTI in the second year compared to only 8% of those with no UTI in the first year (p < .0001). This corresponds to 11 times higher odds of experiencing a UTI, having had one the year before (OR, 11; 95% CI, 10–13; C-statistic = 0.75; Figure 1).

Figure 1.

Risk of urinary tract infection (UTI) within the first 2 years of injury. Directed graph indicates that those having a UTI within the first year of injury are 50% more likely to experience UTI in the second year of injury. OR = odds ratio.

Health care utilization and cost within 2 years after SCI

The IPTW balanced the characteristics between the group of those who did not have any UTIs to those who had at least one UTI in the first 2 years after injury (all ps > .05 for IPTW weighted sample; Table 1). In the weighted groups (Table 3), having a UTI was associated with 17% higher hospital readmission rates (37% vs. 20%, p < .0001) leading to about $4247 higher hospital payments, a 65% difference (ER, 1.65; 95% CI, 1.34–2.03). The use of outpatient services was about 6% higher (98% vs. 93%, p < .0001) with an associated $8204 higher in outpatient services payments corresponding to a twofold difference (ER, 2.29; 95% CI, 2.06–2.56). The UTI group had 9% higher rates of outpatient prescription refills (89% vs. 80%, p < .0001) and an estimate of $1425 higher payments for them, a 31% difference (ER, 1.31; 95% CI, 1.16–1.49). Overall, having UTIs was associated with a median estimate of $15,131, representing 2.48 times higher cumulated health care resource use payments over 2 years after injury (ER, 2.48; 95% CI, 2.21–2.77).

Discussion

Most individuals with SCI require bladder management primarily by either intermittent catheterization or through indwelling catheters to empty the bladder. Despite advances in bladder and medical management strategies, urinary complications persist and are a significant cause of morbidity and mortality among individuals with SCI. UTIs are one of the leading causes of hospital readmission following inpatient rehabilitation.^{7–9,11,12,29} In the present study, we evaluated the factors associated with UTIs, the incidence of UTIs within the first 2 years following a traumatic SCI, and the health care utilization costs for inpatient and outpatient services within this time frame using the MarketScan research database.

At the time of SCI, the national average age is 43 years old, with 78% of new cases being males.⁴ Studies evaluating SCI commonly reflect a sample distribution comprised of 70% to 80% males, which may create some concern regarding the ability to adequately evaluate specific outcomes in female subgroups. In the general population, women experience more UTIs than men due to anatomical differences, such as a shorter urethra with closer proximity to the rectum, and hormonal considerations, such as changes in estrogen levels.³⁰ In the current study, which involved a relatively more balanced distribution of sex, we found that females had a 31% increased risk of UTI. Increased risk among females has previously been observed in SCI; however, most studies have been insufficiently powered to detect clinically significant differences.^31–33 Although a few studies have suggested an increased UTI risk among males,^34,35 there has generally been no association identified concerning sex.^32,36–38 We believe that our findings may more closely reflect observations in the general population where UTI incidence is far greater in females than in males.³⁹

Similar to sex prevalence, the risk presented by age has also been mixed. The present study demonstrated decreasing age to be a significant risk factor for UTI in individuals with SCI. In a study evaluating the management of individuals with neurogenic bladder, Chaudhry et al. observed a complimentary finding of increasing age being associated with decreased odds (7% per year) of UTI.⁴⁰ However, other studies have been unable to identify any such association with age.^34,38

Individuals having Medicaid and Medicare coverage were additional factors associated with increased rates of UTI. Given the reported differences in patient outcomes based on payer status, these findings are not surprising. Readmission rates of individuals with SCI have been described as higher among those with Medicaid and Medicare insurance.⁴¹ Additionally, individuals having Medicaid and Medicare insurance have been found to have limited access to certain services and resources.^42,43 In contrast, a recent cross-sectional survey of a prospective SCI registry with neurogenic bladder found no difference in bladder symptoms between groups based on insurance type. However, the authors indicate that significantly more publicly insured individuals experienced frequent (four or more) UTIs per year than privately insured individuals (30.4% vs. 24.5%).⁴⁴

Our study also found that comorbid conditions increase the risk of UTI; in this case, having at least one comorbidity increased the odds of UTI by 80%. Research supports the fact that individuals with SCI are at increased odds of developing immune impairment, elevated inflammatory markers, and increased risk of infections.^45,46 Moreover, the SCI population is at higher risk of developing metabolic derangements affecting their ability to respond to infections such as pneumonia.^47,48 Several studies have also shown a three times greater rate of diabetes mellitus among individuals in the SCI population,^45,46 leading to an increased risk for infections, including UTIs. Thus, given the commonly high rates of comorbidities, including diabetes, and the high risk of developing UTI associated with SCI,⁴⁹ it is not unexpected to observe such a relationship.

In the overall cohort, cervical level of injury was more prevalent. Higher rates of UTI have been found among this subgroup and have been attributed to reduced functional independence compared to thoracic or lumbar injuries.^50,51 However, in our study, the thoracic level of injury predominated among individuals with UTI. Rabadi and Aston, along with others, have observed a similar pattern involving thoracic level of injury and increased UTI frequency.⁵¹ Although the relationship between the level of injury and risk of UTI is unclear,³⁶ several reports appear to agree that the level of injury can lead to a level-dependent (i.e., ISNCSCI grade) SCI infection correlation, affecting the frequency of SCI infections, including UTIs.^34,35,47,52 While bladder management methods are not variables obtained from MarketScan database records, it represents a key component influencing the development and progression of catheter-associated UTIs and may be impacted by the level of injury. In concordance with the clinical characteristics of this study, the majority of SCIs occur in the cervical region, resulting in tetraplegia.⁵³ The use of a suprapubic (SP) catheter is often implemented to continuously drain the bladder for individuals with limited hand function, adductor spasticity where access to the female urethra is challenging, frequent incontinence, and/or if caregiver assistance to perform scheduled intermittent catheterization is lacking. While SP catheters are regarded by many consumers as a preferred, effortless alternative to a more demanding urethral catheterization management protocol,⁵⁴ insertion of an indwelling catheter into the bladder increases the susceptibility of an individual to infection, introducing opportunistic organisms into the urinary tract. Indwelling catheters provide a surface for the attachment of bacterial cells, which develop a series of adhesion proteins and other virulence factors, crucial to bacterial survival under unique environmental conditions.⁵⁵ Over time, the development of a bacterial biofilm from catheter usage enables microorganisms to resist penetration by both antibiotics and the body’s own immune mechanisms.⁵⁶ Collectively, these factors pose a prevalent medical dilemma in addressing appropriate urological treatment, where increased utilization of indwelling bladder management approaches persists.⁵⁷

Two years following injury, individuals with UTI demonstrated 2.48 times higher cumulative health care resource utilization. This utilization occurred with greater hospital readmission rates, outpatient service payments, and prescription refills. Similarly, Dryden and colleagues found that individuals with SCI had over 2.5 times the hospital readmission rate and outpatient physician contact resulting in higher health care resource utilization in years 2 to 6 following injury.³ In our opinion, this is an essential finding. The higher cost of health care utilization due to genitourinary disorders after SCI can be detrimental for both the individual and the health care system. Developing strategies to optimize surveillance, such as a navigation system of care, may help target those at higher risk of using more health care services.^58–60

Limitations

It is important to note that MarketScan is based on a large convenience sample; the sample is not random and thus may contain biases or fail to generalize to the target population. However, our sample size is large, bias is reduced, and results represent real-world occurrences. It is important to note that MarketScan is from large employers; therefore, results may not generalize to small and medium firms. Additional limitations include the observational claims of the database, making it prone to missing data, inaccurate or nonuniformity of coding practices as well as missing variables (injury severity, and bladder management method, indwelling vs. intermittent catheterization vs. condom catheter; behavioral factors, such as hygiene and catheter changes). These variables are not necessary for claims but are potentially crucial for health care use and cost. Despite these limitations, results found here represent real-world evidence of the cost associated with SCI with a strength of long-term follow-up.

Conclusion

Using data from a national database, we show that individuals with SCI share specific factors that increase UTI risk. Each variable measured was significantly associated with an increased risk of UTI (younger age, female, thoracic injury, Medicaid and Medicare, at least one comorbidity, and developing a UTI within the first year of injury). Additionally, having UTIs demonstrated a significant difference in total health care resource payments amassed compared to the non-UTI cohort. These results emphasize the clinical importance of long-term screening and constant surveillance access after SCI to optimize health care resource utilization and improve patient clinical outcomes. A clear understanding of the extent of health care utilization due to UTIs can also assist SCI specialists in developing tailored educational programs to prevent hospital readmissions and promote well-being and independence.

Funding Statement

Financial Support This research was supported by the Foundation for Physical Therapy Research Promotion of Doctoral Studies I Award (2017-2018) and the Craig H. Neilsen Foundation Allied Health Professional Research Award of ASIA (2018-2019).