The impact of frailty on the economic evaluation of geriatric surgery: hospital costs and opportunity costs based on meta-analysis

Abstract

Objectives: We used a systematic review and meta-analysis to analyze the difference in costs between surgery for frail and non-frail elderly patients. The opportunity cost of frailty in geriatric surgery is estimated using the results.

Methodology: Two literature reviews were carried out between 2000 and 2019: (1) studies comparing total hospital costs of frail and non-frail surgical patients; (2) studies evaluating the length of hospital stay and cost for surgical geriatric patients. We performed a meta-analysis of the items selected in the first review. We subsequently calculated the opportunity cost of frail patients, based on the design of a cost/time variable.

Results: Twelve articles in the first review were selected (272,717 non-frail and 16,461 frail). Fourteen articles were selected from the second review. Frail patients had higher hospital costs than non-frail patients (22,282.541 € and 16,388.844, p?<?.001) and a longer hospital stay (10.16 days and 8.4 (p?<?.001)). The estimated opportunity cost in frail patients is 1,019.56 € (cost/time unit factor of 579.30 €/day).

Conclusions: Frail surgical geriatric patients generate a higher total hospital cost, and an opportunity cost arising from not operating in the best possible state of health. Preoperatively treating the frailty of elderly patients will improve the use of health resources     

Medical costs in patients with heart failure after acute heart failure events: one-year follow-up study

Aims: This study investigated annual medical costs using real-world data focusing on acute heart failure.

Methods: The data were retrospectively collected from six tertiary hospitals in South Korea. Overall, 330 patients who were hospitalized for acute heart failure between January 2011 and July 2012 were selected. Data were collected on their follow-up medical visits for 1 year, including medical costs incurred toward treatment. Those who died within the observational period or who had no records of follow-up visits were excluded. Annual per patient medical costs were estimated according to the type of medical services, and factors contributing to the costs using Gamma Generalized Linear Models (GLM) with log link were analyzed.

Results: On average, total annual medical costs for each patient were USD 6,199 (±9,675), with hospitalization accounting for 95% of the total expenses. Hospitalization cost USD 5,904 (±9,666) per patient. Those who are re-admitted have 88.5% higher medical expenditure than those who have not been re-admitted in 1 year, and patients using intensive care units have 19.6% higher expenditure than those who do not. When the number of hospital days increased by 1?day, medical expenses increased by 6.7%.

Limitations: Outpatient drug costs were not included. There is a possibility that medical expenses for AHF may have been under-estimated.

Conclusion: It was found that hospitalization resulted in substantial costs for treatment of heart failure in South Korea, especially in patients with an acute heart failure event. Prevention strategies and appropriate management programs that would reduce both frequency of hospitalization and length of stay for patients with the underlying risk of heart failure are needed.     

Perioperative outcomes and hospital costs associated with flowable gelatin hemostatic matrix for lumbar surgeries in real world hospital setting

Abstract

Background: Efficient hemostasis during lumbar surgery (LS) is associated with better perioperative outcomes. Flowable gelatin hemostatic matrix (FGHM) is a new type of absorbable hemostatic agent, which is effective to control bleeding during spinal surgery. This study aimed to assess the impact of FGHM on perioperative outcomes and hospital costs associated with LS.

Methods: This study retrospectively analyzed medical and billing records of patients who underwent LS for spinal degenerative disease in a Chinese tertiary care hospital from 2014 to 2016. The identified patients were further stratified into a FGHM group (n?=?108) (using the combination of FGHM and gelatin sponge) and a historical control group (using oxidized cellulose and/or collagen, n?=?82) for the adjusted comparisons of the perioperative outcomes using a propensity score matching method. Multiple generalized linear regression was conducted to assess the impact of using FGHM on total hospitalization costs.

Results: Comparisons of 64 propensity score matched pairs showed a significantly lower blood transfusion rate (34.4% vs 64.1%, p?=?0.005), lower blood transfusion volume (182.7?±?312.4 vs 301.3?±?281.0?mL, p?=?0.045), reduced post-surgery drainage tube placement rate (82.8% vs 93.8%, p?=?0.046), and shorter post-operative days on antibiotics (6.0?±?2.6 vs 7.1?±?2.4 days, p?=?0.010) in the FGHM group. Although with a relatively high acquisition price, the use of FGHM for hemostasis in LS did not increase the total hospitalization costs (coefficient = ?0.001, p?=?0.972).

Conclusions: The use of FGHM in LS improved perioperative outcomes related to hemostatic effects without increasing overall hospital costs in a real-world hospital setting.     

Impact of heart failure on hospital activity and healthcare costs in Belgium

Summary

Heart failure (HF) is a serious public health problem worldwide. It has a high prevalence, affects mainly the elderly and causes high mortality or disability with high economic costs. The aim of the present study was to calculate the number of admissions for HF, the total in-hospital stay, the mean length of in-hospital stay and the in-hospital costs due to HF in Belgium.

Retrospective analysis of data from the national hospital registration system provided the following results. In 2001, there were 19,398 admissions with HF as a primary diagnosis, with a total in-hospital stay of 286,938 days. The mean in-hospital stay for HF was 14.8 days. The total in-hospital cost of HF as a primary diagnosis was € 94,113,827, representing 1.8% of the total hospital expenditure.

The limitations of this study are its mere focus on admissions and their characteristics in 2001, and the use of a retrospective analysis. Nevertheless, it led to the conclusion that HF was responsible for a significant number of in-hospital days, with a significant impact on healthcare costs in Belgium.     

Cost-effectiveness analysis of a sealing hemostat patch (HEMOPATCH) vs standard of care in cardiac surgery

Background: A recent randomized controlled trial showed that patients undergoing ascending aorta surgery treated with HEMOPATCH to control bleeding had a significantly better hemostasis success rate than with dry or wet gauze compression or similar standard of care (SOC).

Objective: To compare the cost-effectiveness using two different agents for hemostasis (HEMOPATCH vs dry or wet gauze compression or similar SOC) in cardiac surgery from the European hospital perspective.

Methods: A literature-based cost-effectiveness model estimating average cost per successful hemostasis event was developed based on the hemostasis efficacy difference (HEMOPATCH?=?97.6%, SOC?=?65.8%, p?<?.001). Additional clinically significant end-points studied in the trial (blood transfusions and surgical revisions) were also analyzed. It was assumed that each surgery utilized two units of HEMOPATCH (dimensions of 4.5?×?9?cm) and two units of SOC. Product acquisition costs for HEMOPATCH and SOC were included along with outcome-related costs derived from the literature and inflation-adjusted to 2017 EUR and GBP. Results are presented for an average hospital with an annual case load of 574 cardiac surgeries. One-way and probabilistic sensitivity analyses were performed.

Results: Considering only product acquisition cost, HEMOPATCH had an incremental cost-effectiveness ratio (ICER) of €1,659, €1,519, €1,623, and £1,725 per hemostasis success when compared to SOC for Italy, Spain, France, and the UK, respectively. However, when considering the cost and potential difference in the frequency of transfusions and revisions compared to SOC, the use of HEMOPATCH was associated with an annual reduction of six revisions and 60 transfusions, improving the ICER to €1,440, €1,222, €1,461, and £1,592, respectively. Sensitivity analysis demonstrated model robustness.

Conclusions: This analysis supports the use of HEMOPATCH over SOC in cardiac surgery in European hospitals to improve hemostasis success rates and potential cost offsets from reduced transfusions, complications, and surgical revisions.     

Economic burden of skeletal-related events in patients with multiple myeloma: analysis of US commercial claims database

Aims: To estimate incremental healthcare resource utilization (HRU) and costs associated with skeletal-related events (SREs) secondary to multiple myeloma (MM), and HRU and cost differences in patients with one vs multiple SREs.

Methods: Adults with MM diagnosis between January 1, 2010–December 31, 2014, with benefits coverage ≥12 months pre- and ≥6 months post-diagnosis were followed to last coverage date or December 31, 2015, excluding patients with prior anti-myeloma treatment or cancers. SREs were identified by diagnosis or procedure codes (pathological fracture, spinal cord compression, radiation, or surgery to the bone). SRE patients (index?=?first post-diagnosis SRE) were propensity score matched 1:1 to patients without SRE (assigned pseudo-index) using baseline characteristics, and ≥1 month of continuous enrollment after index/pseudo-index date was required. Per-patient-per year (PPPY) HRU and costs (2016?US$) were determined for inpatient, outpatient, emergency department (ED), and outpatient pharmacy services during follow-up. Wilcoxon signed rank for means and McNemar’s tests for proportions were used to assess differences. Negative binomial regression and generalized linear regression analyses estimated differences in HRU and costs, respectively, for the comparison of single vs multiple SREs.

Results: Each cohort included 848 patients (mean age?=?61 – 62 years, 57% male) with no significant differences in pre-index demographic or clinical characteristics between matched cohorts. Versus non-SRE patients, SRE patients had significantly higher PPPY use (p?<?.0001) of inpatient hospitalizations, ED visits, outpatient pharmacy, and higher direct medical costs ($188,723 vs $108,160, p?<?.0001). Adjusted PPPY total costs were $209,820 in patients with multiple SREs; $159,797 in patients with one SRE.

Limitations: SRE misclassification and residual confounding are possible.

Conclusions: Among patients with MM, average annual costs were substantially higher in patients with SRE compared with matched non-SRE patients. The economic burden of SRE increased further with multiple events.     

Cost-effectiveness analysis of bariatric surgery for morbid obesity in Belgium

Aims: This study presents the cost-effectiveness analysis of bariatric surgery in Belgium from a third-party payer perspective for a lifetime and 10-year horizon.

Materials and methods: A decision analytic model incorporating Markov process was developed to compare the cost-effectiveness of gastric bypass, sleeve gastrectomy, and adjustable gastric banding against conventional medical management (CMM). In the model, patients could undergo surgery, or experience post-surgery complications, type 2 diabetes, cardiovascular diseases, or die. Transition probabilities, costs, and utilities were derived from the literature. The impact of different surgical methods on body mass index (BMI) level in the base-case analysis was informed by the Scandinavian Obesity Surgery Registry and the Swedish Obese Subject (SOS) study. Healthcare resource use and costs were obtained from Belgian sources. A base-case analysis was performed for the population, the characteristics of which were obtained from surgery candidates in Belgium.

Results: In the base-case analysis over a 10-year time horizon, the increment in quality-adjusted life-years (QALYs) gained from bariatric surgery vs CMM was 1.4 per patient, whereas the incremental cost was €3,788, leading to an incremental cost-effectiveness ratio (ICER) of €2,809 per QALY. Over a lifetime, bariatric surgery produced savings of €9,332, an additional 1.1 life years and 5.0 QALYs. Bariatric surgery was cost-effective at 10 years post-surgery and dominant over conventional management over a lifetime horizon.

Limitations: The model did not include the whole scope of obesity-related complications, and also did not account for variation in surgery outcomes for different populations of diabetic patients. Also, the data about management of patients after surgery was based on assumptions and the opinion of a clinical expert.

Conclusions: It was demonstrated that a current mix of bariatric surgery methods was cost-effective at 10 years post-surgery and cost-saving over the lifetime of the Belgian patient cohort considered in this analysis.     

Economic burden and healthcare utilization associated with castration-resistant prostate cancer in a commercial and Medicare Advantage US patient population

Objective: Prostate cancer is a leading cause of cancer death in men in the US. Castration-resistant prostate cancer (CRPC) is an advanced form of the disease and has a poor prognosis and limited treatment options. The objective of this study was to identify patients with CRPC from a medical claims database, and determine the prostate cancer-related economic burden and healthcare utilization of these patients.

Methods: This was a retrospective study using claims and enrollment information from a large US database linkable to laboratory data. Male patients aged 40 or older who were diagnosed with prostate cancer and received surgical or medical castration between July 1, 2001 and December 1, 2007 were considered for study inclusion. Patients with CRPC were initially identified based on at least two increases in prostrate-specific antigen (PSA) values. Due to the small number of patients with available PSA results data, logistic regression modeling using characteristics of patients with known CRPC was used to identify a larger set of patients with likely CRPC. Per-patient per-month healthcare utilization and costs were determined using medical and pharmacy claims data.

Results: The final sample of patients with likely CRPC as determined by regression modeling included 349 patients with known CRPC identified from the database on the basis of PSA results and an additional 2391 with likely CRPC. Within this final sample of 2740 CRPC patients, there was a per-patient per-month average of 1.43 prostate cancer-related ambulatory visits, 0.04 prostate cancer-related inpatient stays, and 0.01 prostate cancer-related ER visits. Average per-patient per-month prostate cancer-related costs were $1152 (SD = $2073) for ambulatory visits, $559 (SD = $2383) for inpatient stays, $72 (SD = $229) for pharmacy costs, and $1 (SD = $14) for ER visits. Total per-patient per-month prostate cancer-related costs were on average $1799 (SD = $3505), and these costs comprised about half of the all-cause healthcare costs for these patients.

Conclusions: CRPC is a costly disease, with ambulatory visits and inpatient care accounting for a substantial proportion of the economic burden. Limitations related to the use of retrospective claims data should be considered when interpreting these results.     

Cost-consequence analysis of two different active flowable hemostatic matrices in spine surgery patients

Objectives: A recently published retrospective analysis comparing two different active flowable hemostatic matrices (FLOSEAL and SURGIFLO Kit with Thrombin) showed significantly increased resource use and complications (surgery time, risk of blood product transfusion, and amount of matrix used) with SURGIFLO use compared to FLOSEAL in major spine surgery, and also significantly increased surgical time with SURGIFLO use in severe spine surgery. This analysis was developed as a follow-up to this prior analysis, to evaluate the cost-consequence of using FLOSEAL vs SURGIFLO in major and severe spine surgery.

Methods: A cost consequence model was constructed from a US hospital provider perspective. Model parameters combined clinical inputs from the published retrospective analysis with supplemental analyses on annual spine surgery volume using the 2012 National Inpatient Sample (NIS) database. Cost of hemostatic matrices, blood product transfusion, and operating room time were identified from published literature. Various one-way and probabilistic sensitivity analyses were performed.

Results: The base case for a medium volume hospital showed that, compared to SURGIFLO, patients receiving FLOSEAL required three fewer blood product transfusions and saved 27?h of OR time, resulting in annual savings of $151 per major and $574 per severe spine surgery. Additional scenarios for high and low volume hospitals supported cost savings in the base case. Probabilistic sensitivity analysis revealed FLOSEAL was cost-saving in 76% of simulations in major spine and 97% of iterations in severe spine surgery.

Conclusions: This economic analysis indicates that use of FLOSEAL instead of SURGIFLO hemostatic matrices to induce hemostasis in both major and severe spine surgery could potentially lead to sizable cost savings in US hospitals, regardless of spinal surgery case-mix.     

Healthcare resource utilization and costs by age and joint location among osteoarthritis patients in a privately insured population

Aims: To compare healthcare resource utilization and costs between patients aged 18–64 years with osteoarthritis (OA) and matched controls without OA in a privately insured population.

Methods: Patients with OA were selected from de-identified US-based employer claims (Q1:1999–Q3:2011). The index date was defined as the first OA diagnosis indicated by ICD-9-CM codes. One year before and after the index date were defined as the baseline and study periods, respectively. A second OA diagnosis during the study period was also required. Patients with OA were matched one-to-one on age, gender, index date, and minimum length of follow-up to controls without OA. Baseline characteristics and study period resource utilization and costs (2016 USD) were compared between cohorts.

Results: This study identified 199,539 patients with OA (knee: 87,271, hip: 19,953, hand: 15,670, spine: 12,496). The average age was 54 years, and 58% were female. OA patients had higher healthcare resource utilization than matched controls in inpatient, emergency room, and outpatient settings (p?p?Limitations: This sample, obtained using claims data, only includes patients who were actively seeking care for OA and were likely symptomatic. Asymptomatic patients would likely not be captured in this analysis.

Conclusions: Patients with OA incur greater healthcare resource utilization and costs than patients without OA, with substantial variation by joint location.     

Economic modeling to evaluate the impact of chronic myeloid leukemia therapy management on the oncology care model in the US

Abstract

Objective: To develop an economic model to evaluate changes in healthcare costs driven by restricting usage of branded tyrosine kinase inhibitors (TKIs) through substitution with generic imatinib among chronic myeloid leukemia (CML) patients in a typical Oncology Care Model (OCM) practice, and examine the impact on Performance-Based Payment (PBP) eligibility.

Methods: An Excel-based economic model of an OCM practice with 1,000 cancer patients during a 6-month episode of care was developed. Cancer types and proportions of patients treated in the practice were estimated from an OCM report. All-cause healthcare costs were obtained from published literature. It was assumed that if a practice restricts usage of branded TKIs for newly-diagnosed CML patients, 80% of the market share of branded imatinib and 50% of the market shares of 2^nd-gen TKIs would shift to generic imatinib. Among established TKI-treated patients, it was assumed that 80% of the market share of branded imatinib and no patients treated with 2^nd-gen TKIs would shift to the generic.

Results: Four CML patients were estimated for a 1,000-cancer patient OCM practice with a total baseline healthcare cost of $51,345,812 during a 6-month episode. If the practice restricts usage of branded TKIs, the shift from 2^nd-gen TKIs to generic imatinib would reduce costs by $12,970, while shifting from branded to generic imatinib lowers costs by $25,250 during a 6-month episode. Minimum reductions of $3,013,832 in a one-sided risk model and $2,372,010 in a two-sided risk model are required for PBP eligibility; the shift from 2^nd-gen TKIs to generic imatinib would account for 0.4% and 0.5% of the savings required for a PBP, respectively.

Conclusions: This analysis indicates that the potential cost reduction associated with restricting branded TKI usage among CML patients in an OCM setting will represent only a small proportion of the cost reduction needed for PBP eligibility.     

Burden of rotavirus gastroenteritis and potential benefits of a pentavalent rotavirus vaccination in Belgium

Abstract

Objective and Methods: A decision analytic model was built to assess the paediatric rotavirus gastroenteritis (RVGE) burden and potential benefits associated with the introduction of RotaTeq^®? (pentavalent rotavirus vaccine) in Belgium.

Results: In the absence of a rotavirus (RV) immunisation programme, paediatric RVGE was estimated to account for about 5,860 hospitalisations, 1,720 cases of nosocomial infections, 9,410 cases treated by general practitioners/paediatricians (GP/P) and 10,790 cases not seeking medical care for a birth cohort followed up to 5 years of age. Paediatric RVGE was estimated to cost about €9.0 million from the Belgian healthcare provider perspective and €15.3 million to society. Given a 90% RV vaccination coverage rate, the pentavalent RV vaccine would have a high impact on RV burden by preventing more than 4,850 hospitalisations, 995 cases of nosocomial infections, 7,145 cases treated by GP/P and 8,190 cases not seeking medical care, and reduce RVGE costs by €7.1 million from the Belgian healthcare provider perspective and €12.0 million to society.     

Reimbursement schemes for hospitals: the impact of case and firm characteristics

We contribute to the debate on high-powered versus low-powered incentives in regulation by studying their heterogeneous impacts on different subpopulations, using data from the introduction of a high-powered prospective payment system (PPS) for hospital reimbursement in Germany. While no overall effect on quality or cost saving is found, our results support hypotheses drawn from an incentive and selection perspective: PPS reduces the length of stay of older relative to younger patients, of more severe relative to less severe cases, and in smaller relative to larger hospitals. Hospitals which adopted PPS earlier provide higher quality under PPS as proxied by the case-specific readmission rate. Our study also contributes to the health economic literature on hospital reimbursement as our data permits us to identify the treatment effect via different timings of adoption of PPS and to use a more accurate quality measure by following patients even when readmitted to other hospitals.     

A real-world analysis of healthcare costs and relative risk of hyperprolactinemia associated with antipsychotic treatments in the United States

Abstract

Aims: Antipsychotic medications are associated with an increased risk of hyperprolactinemia, but differ in their propensity to cause this complication. This study aimed to assess the economic burden of hyperprolactinemia, and to compare its risk among adult patients using atypical antipsychotics (AAs) with a mechanism of action associated with no/low vs high/moderate prolactin elevation.

Methods: This retrospective cohort study was based on US Commercial and Medicaid claims databases. Healthcare costs were compared between matched hyperprolactinemia and hyperprolactinemia-free cohorts using a two-part model. Risk of hyperprolactinemia was compared between patients receiving AAs with a mechanism of action associated with no/low (no/low prolactin elevation cohort) vs high/moderate prolactin elevation (high/moderate prolactin cohort) using logistic regression.

Results: In the commercially insured sample, compared to the hyperprolactinemia-free cohort (n?=?499), the hyperprolactinemia cohort (n?=?499) was associated with incremental total healthcare costs of $5,732 ($20,081 vs $14,349; p?=?.004), and incremental medical costs of $3,861 ($13,218 vs $9,357; p?=?.040), mainly driven by hyperprolactinemia-related costs. In the Medicaid-insured sample, compared to the hyperprolactinemia-free cohort, the hyperprolactinemia cohort was associated with incremental total healthcare costs of $10,773 ($30,763 vs $19,990; p?=?.004), and incremental medical costs of $9,246 ($20,859 vs $11,613; p?=?.004), mainly driven by hyperprolactinemia-related and mental health-related costs. The odds of hyperprolactinemia in the no/low prolactin elevation cohort were 4–5-times lower than that in the high/moderate prolactin elevation cohort (odds ratio =0.21; p?<?.001).

Limitations: Hyperprolactinemia may be under-reported in claims data.

Conclusions: Hyperprolactinemia is associated with substantial healthcare costs. AAs associated with no/low prolactin elevation reduce the risk of hyperprolactinemia by 4–5-times compared to AAs associated with moderate/high prolactin elevation. Treatment options with minimal impact on prolactin levels may contribute to reducing hyperprolactinemia burden in AA-treated patients.     

Healthcare costs and resource utilization in patients with severe aplastic anemia in the US

Abstract

Purpose: This study aimed to evaluate the healthcare resource utilization (HCRU) and costs for patients with severe aplastic anemia (SAA) using US claims data.

Methods: This retrospective, observational database study analyzed claims data from the Truven MarketScan databases. SAA patients aged ≥2?years identified between 2014 and 2017 who were continuously enrolled for 6?months before their first SAA treatment or blood transfusion, with a ≥6-month follow-up, were included. Baseline demographics and comorbidities were evaluated. Monthly all-cause and SAA-related HCRU and direct costs in the follow-up period were analyzed and differences were presented for all patients and across age groups.

Results: With an average follow-up period of 21.5?months, 939 patients were included in the study. Monthly all-cause and SAA-related HCRU [mean (SD)] were 1.65 days (2.61 days) and 0.18 days (0.70 days) for length of stay, 0.18 (0.23) and 0.01 (0.04) for hospital admissions, 0.25 (0.30) and 0.02 (0.07) for ER visits, 2.24 (1.40) and 0.46 (0.99) for office visits, and 2.90 (2.64) and 0.55 (1.31) for outpatient visits, respectively. On average, SAA patients received 0.15 (0.57) blood transfusions per month. Mean monthly all-cause direct costs were $28,280 USD ($36,127) [US dollars, mean (SD)]. Direct costs related to admissions were $11,433 USD (SD $25,040), followed by $624 USD ($1,703) for ER visits, $528 USD ($694) for office visits, $7,615 USD ($13,273) for outpatient visits, and $5,998 USD ($11,461) for pharmacy expenses. Monthly SAA-related direct costs averaged $7,884 USD (SD $16,254); of these costs, $1,608 USD ($7,774) were from admissions, $47 USD ($257) from ER visits, $127 USD ($374) from office visits, $1,462 USD ($4,994) from outpatient visits, and $4,451 USD ($10,552) from pharmacy expenses.

Conclusion: SAA is associated with high economic burden, with costs comparable to blood malignancies, implying that US health plans should consider appropriately managing SAA while constraining the total healthcare costs when making formulary decisions.     

Major bleeding risk and healthcare economic outcomes of non-valvular atrial fibrillation patients newly-initiated with oral anticoagulant therapy in the real-world setting

Aims: This study compared the risk for major bleeding (MB) and healthcare economic outcomes of patients with non-valvular atrial fibrillation (NVAF) after initiating treatment with apixaban vs rivaroxaban, dabigatran, or warfarin.

Methods: NVAF patients who initiated apixaban, rivaroxaban, dabigatran, or warfarin were identified from the IMS Pharmetrics Plus database (January 1, 2013–September 30, 2015). Propensity score matching (PSM) was used to balance differences in patient characteristics between study cohorts: patients treated with apixaban vs rivaroxaban, apixaban vs dabigatran, and apixaban vs warfarin. Risk of hospitalization and healthcare costs (all-cause and MB-related) were compared between matched cohorts during the follow-up.

Results: During the follow-up, risks for all-cause (hazard ratio [HR]?=?1.44, 95% confidence interval [CI]?=?1.2–1.7) and MB-related (HR?=?1.57, 95% CI?=?1.0–2.4) hospitalizations were significantly greater for patients treated with rivaroxaban vs apixaban. Adjusted total all-cause healthcare costs were significantly lower for patients treated with apixaban vs rivaroxaban ($3,950 vs $4,333 per patient per month [PPPM], p?=?.002) and MB-related medical costs were not statistically significantly different ($100 vs $233 PPPM, p?=?.096). Risk for all-cause hospitalization (HR?=?1.98, 95% CI?=?1.6–2.4) was significantly greater for patients treated with dabigatran vs apixaban, although total all-cause healthcare costs were not statistically different. Risks for all-cause (HR?=?2.22, 95% CI?=?1.9–2.5) and MB-related (HR?=?2.05, 95% CI?=?1.4–3.0) hospitalizations were significantly greater for patients treated with warfarin vs apixaban. Total all-cause healthcare costs ($3,919 vs $4,177 PPPM, p?=?.025) and MB-related medical costs ($96 vs $212 PPPM, p?=?.026) were significantly lower for patients treated with apixaban vs warfarin.

Limitations: This retrospective database analysis does not establish causation.

Conclusions: In the real-world setting, compared with rivaroxaban and warfarin, apixaban is associated with reduced risk of hospitalization and lower healthcare costs. Compared with dabigatran, apixaban is associated with lower risk of hospitalizations.