Abstract
BACKGROUND:
Infusion therapy, used to administer medications for multiple chronic diseases, can be performed in hospital outpatient departments (HOPDs), patients’ homes, ambulatory infusion centers, and physicians’ offices. The site for administering infusion therapy can impact the cost of care. However, there is limited evidence on quality associated with the site of care (SOC).
OBJECTIVE:
To assess how patient utilization, cost, and adherence outcomes differ between infusions administered in HOPDs compared with alternative SOCs (ambulatory infusion centers, patients’ homes, and physicians’ offices).
METHODS:
This retrospective cohort study used administrative claims data to construct a sample of infusions administered in January 2022 to December 2023 to commercially insured adults across the United States, with infusions occurring in all 50 states and Washington, District of Columbia. These included 23 infusion agents used to treat 7 chronic conditions. To assess outcomes across SOCs, infusions in HOPDs were 1:1 matched to infusions in alternative SOCs using a combination of exact matching and propensity score matching on the infusion agent, treated disease, infusion sequence number, state of residence, patient demographics, baseline health status, and baseline medical utilization. Infusions were excluded when an HOPD was deemed medically necessary. Using logistic and Poisson regressions, outcomes were observed in two separate postinfusion time frames, including the infusion date: 1-day postinfusion and 7-day postinfusion. Outcomes included all-cause cost and utilization across inpatient, emergency department (ED), outpatient, and pharmacy services; mild adverse events; and severe adverse events. Using linear regressions, infusion therapy adherence outcomes were observed within 12 months of an index infusion for a subset of the study sample treated with 8 infusion agents.
RESULTS:
Of 52,760 infusions among 18,988 patients, within 1-day postinfusion, patients administered infusions in HOPDs had no significant differences in ED and inpatient utilization and costs compared with alternative SOCs. However, outpatient costs were 41.9% higher (P < 0.01) among patients treated in an HOPD compared with those treated in alternative SOCs. There were no significant differences in serious or mild adverse events between the two groups. Within 7 days postinfusion, outcomes were similar, except HOPD-treated patients had 8.6% lower odds of filling any prescription at a pharmacy (P < 0.01) and 45.2% higher odds of having an inpatient admission (P < 0.05). Among the 410 patients analyzed for infusion adherence, there were no significant differences in adherence outcomes 12 months after the index infusion.
CONCLUSIONS:
Patients receiving infusions in HOPDs have higher outpatient costs without a reduction in adverse events, inpatient admissions, or ED visits or an increase in infusion therapy adherence compared with SOCs, indicating that SOCs offer similar quality outcomes at lower costs.
Plain language summary
Infusion therapies can occur in hospital outpatient departments (HOPDs) or alternative sites of care (SOCs). When comparing the outcomes of patients who received infusions in HOPDs vs alternative SOCs, we did not find differences in adverse events or emergency department visits. However, those who received infusions in HOPDs had 42% higher outpatient costs compared with those who received infusions in alternative SOCs. When there is not a medical need to receive infusion therapy in an HOPD, receiving infusions in alternative SOCs may lower costs while maintaining quality care.
Implications for managed care pharmacy
Payers, providers, and policymakers should assess the necessity of HOPD settings for the administration of infusion treatments for certain patients and develop strategies to expand access to alternate care sites to reduce costs and maintain care quality.
Infusion therapy can be used for the treatment and management of several chronic diseases. Because many of these therapies are administered by a health care professional, they are performed in various sites of care (SOCs), including hospital outpatient departments (HOPDs), patients’ homes, ambulatory infusion centers (AICs), and physicians’ offices.
Much of the current literature on infusion therapy outcomes and SOCs focuses on oncologic treatments and has generally found infusions in alternative SOCs (patients’ homes, AICs, and physicians’ offices) to be associated with lower costs than HOPD infusions. Oncological infusion therapy studies that compared infusions in HOPDs to physicians’ offices found lower infusion-related and all-cause costs associated with physicians’ offices1 despite generally similar and sometimes better-quality outcomes than HOPDs.1–3 Similarly, nononcological infusion therapy studies also found lower costs associated with physicians’ offices4 and home infusion therapy5 than HOPDs.
Most prior studies on infusion therapy for treatment of nononcological chronic conditions are limited to comparing home infusion therapy to a single SOC. Among patients with primary immunodeficiency, home infusion therapy has been found to be associated with a lower risk of infection compared with HOPDs.6 However, when compared with physician offices (designated by Centers for Medicare & Medicaid Services [CMS] Place of Service Code 11), home infusion of biologic therapies has been associated with an increased risk of emergency department (ED) visits and inpatient admissions for certain immune-related conditions.7 Conversely, in a study among patients with multiple sclerosis, home infusion had similar rates of adverse events compared with infusion centers.8
In addition to clinical measures, measures of patient satisfaction also differ by SOC. Patients with one of several immune-mediated conditions who were receiving infusion therapy across multiple SOCs reported that the chance to interact with staff and other patients at their SOC was very valuable but that scheduling and traveling to their SOC could be challenging. They also reported higher satisfaction related to staff interaction, wait times, and ease and cost of parking in physicians’ offices compared with HOPDs. Over two-thirds did not have a choice in the SOC for infusion therapy.9 A study of patients receiving infusion therapy for multiple sclerosis reported that patients had higher rates of confidence in their nurses, feeling respected and safe during the infusion, and being comfortable in their surroundings when receiving home infusion therapy compared with their previous experiences at an infusion center.8 Finally, recipients of outpatient parenteral antibiotic therapy were more likely to recommend their SOC to others and less likely to have complaints about lapses in care and infection prevention when receiving infusions at home compared with in skilled nursing facilities.10
There is currently no known literature evaluating the relationship between infusion SOCs and health care outcomes for infusion therapies treating multiple nononcological chronic conditions. There also is currently no known literature evaluating infusions across all 4 types of SOCs. To address this gap, our study investigated how health care utilization and costs, adverse outcomes, and infusion therapy adherence differ between infusions in HOPDs and those in alternative SOCs (AICs, home, and physicians’ offices) for several nononcological conditions.
Methods
DATA SOURCE AND STUDY POPULATION
This is a retrospective matched cohort study that used administrative medical and pharmacy claims from the Healthcare Integrated Research Database,11 a database managed by Elevance Health. Data from July 2021 to December 2023 were used to identify a nationally representative commercially insured adult population who received one of 23 infusion agents administered by providers in an outpatient setting, either HOPD or alternative SOCs (Supplementary Exhibit 2 (266.8KB, pdf) , available in online article). Infusions occurred in all 50 states and Washington, District of Columbia, and include individual, employer, and self-insured plans. Among the 23 infusion agents included, some are available in both intravenous and subcutaneous forms. For these, the subcutaneous forms were included only if they were administered by providers in either HOPDs or alternative SOCs. The 23 infusion agents were selected for this study because they treat at least 1 of 7 common nononcologic medical conditions: ulcerative colitis, Crohn disease, rheumatoid arthritis, plaque psoriasis, psoriatic arthritis, osteoporosis, and multiple sclerosis.
The primary analysis was conducted at the infusion level, and multiple infusions could be included for each patient. Identified patients were required to have continuous health plan enrollment for 6 months preinfusion and 1 week postinfusion and to have a medical claim with a diagnosis for 1 of the 7 medical conditions in the 6 months prior to the infusion. Infusions were excluded if the patient’s state of residence was missing; no other covariates had missingness in the data.
To enhance the likelihood that patients had similar preinfusion health, infusions were excluded if patients were being treated for 2 or more of the 7 common nononcologic conditions or if the infusions could be identified as medically necessary to occur in an HOPD (ie, the patient may be higher acuity). Receiving infusions in an HOPD may be necessary when a higher level of monitoring is required (such as when treatment is initiated or the regimen changed), individual clinical factors contribute to a risk or repeated history of adverse events, there is a risk of severe adverse events that require rapid and intensive intervention, or alternate SOCs are not accessible.12 HOPD infusions of patients enrolled in a care management program, which aims to provide members with high-quality care in the most cost-effective SOC according to their clinical needs, were excluded from the analysis as those infusions may be deemed medically necessary to occur in an HOPD.13 The care management program steered infusion recipients to in-network AICs or to home infusions from the HOPD. Additionally, infusion recipients were able to choose an infusion in a physician’s office as an alternative to HOPD.
A supplemental analysis of adherence outcomes was conducted at the patient-level based on the index (first) infusion date with continuous health plan enrollment 12 months prior to index infusion and 12 months post-index infusion. To avoid missing infusion adherence data and to assess differences in outcomes by HOPD vs alternative SOCs, we excluded 15 of the 23 infusion agents that could be dispensed by pharmacies directly to patients to be self-administered. Therefore, the adherence analysis focused on 8 of the 23 infusion agents because they are only available to patients when they are administered by a health care provider in either HOPD or alternative SOCs and are not available for patients to self-administer the infusion agents. Therefore, the adherence is measuring continuity of patients receiving provider-administered therapies over the 12 months to ensure adherence is not biased by patients who may switch to self-administered versions of an infusion therapy. Patients were also excluded if they had an occurrence of the same infusion agent in the 12 months prior to the index infusion date or if they had more than 1 SOC associated with the index infusion agent in the 12 months post-index infusion date. A sensitivity analysis was conducted using a 6-month follow-up period among a larger sample of patients with 6-month post-index infusion eligibility.
EXPOSURE
Infusions were identified as occurring in an HOPD if the infusion claims had CMS place of service code of 22. Additionally, infusions were identified as occurring in alternative SOCs if the CMS place of service codes were either 11 (physician office) or 12 (AIC if infusion claim included ‘SS’ procedure modifier code, else patient’s home).
MATCHING
For the primary analysis, infusions in an HOPD were 1:1 matched to infusions in alternative SOCs using patient and infusion characteristics and a combination of exact matching and propensity score matching. The infusion agent Healthcare Common Procedure Coding System (HCPCS) code and the medical condition being treated by the infusion agent were exactly matched. Infusion sequence number (first, second, or third and greater) was also exactly matched because adverse clinical outcomes are more likely after the initial infusion.12 Other covariates for exact matching included the presence of cancer in the 6 months prior to the infusion and the patient’s state of residence. Covariates in propensity score matching were age on infusion date, sex, rurality of residence, socioeconomic status index,14 race and ethnicity, Elixhauser comorbidity score15 6 months preinfusion, baseline medical utilization (inpatient admissions, outpatient services, ED visits, and office visits) during the 6 months preinfusion, quarter of year to account for seasonality, and an indicator for if the member had a claim for an infusion agent other than one of the 23 study infusion agents during the 6 months preinfusion.
For the supplemental analysis, patients with an index infusion in an HOPD were 1:1 matched to patients with an index infusion in alternative SOCs. This included a combination of exact matching and propensity score matching similar to the primary analysis matching with preinfusion characteristics based on the 12 months prior to the index infusion instead of 6 months preinfusion.
OUTCOMES
For the primary analysis, adverse events included mild adverse events (fever and chills, dyspnea, pruritus, and urticaria) and severe adverse events (anaphylaxis, convulsions, and hypotension). Additionally, mild and severe adverse events were combined along with drug-specific and other adverse events associated with infusions to measure all potential infusion-related adverse events (Supplementary Exhibit 2 (266.8KB, pdf) ). The utilization outcomes studied were the presence of at least 1 all-cause inpatient admission, ED visit, emergency hospitalization, or observation visit and any pharmacy claim. The cost outcomes studied were total cost, infusion claim cost, and costs associated with inpatient admissions, ED visits, outpatient services (including the cost of the infusion), and medication fills at a pharmacy. Costs were adjusted to the 2024 Consumer Price Index. Outpatient, total, and infusion claim costs were winsorized at the 5th and 95th percentiles to account for large outliers. ED, pharmacy, and inpatient costs were not winsorized as more than 93% of those costs consisted of zeros. All outcomes were studied in two different time frames: 1 day and 7 days postinfusion; each period included the infusion date.
For the supplemental analysis, we calculated the observed vs expected refill ratio (OvERR) with and without a 7-day gap between the expected dosing regimens16 (see Supplementary Table 1 (266.8KB, pdf) ). OvERR quantifies adherence by dividing the actual number of infusions received by the number expected according to the dosing regimen, with values of 1.0 suggesting that patients received infusions in accordance with the dosing regimen. OvERR with a 7-day gap allows for an additional 7 days from one scheduled dose to the next scheduled dose. Traditional adherence measures like medication possession ratio and proportion of days covered were not used because the infusion therapies in this study are administered by providers and lack days supply data in administrative claims.
STATISTICAL ANALYSIS
Unadjusted logistic regressions were used for binary utilization outcomes, unadjusted Poisson regressions for cost outcomes, and unadjusted linear regressions for continuous outcomes to evaluate the relationship between the site for administering infusion therapy and a given outcome. SEs were clustered at the matched-pair level. Statistical significance was evaluated using P < 0.05.
ETHICS
This observational study, conducted under the research exception provisions of Privacy Rule 45 CFR 164.514(e), was exempt from institutional board review because researchers accessed a limited dataset for analysis that was devoid of individual patient identifiers and complied with all relevant provisions of the Health Insurance Portability and Accountability Act. Institutional review board exemption was not necessary because the study was an analysis of the managed care organization’s membership data for the purposes of health plan treatment, planning, and operations. All data were anonymized before being used.
Results
POPULATION
The prematch sample consisted of 302,574 infusions with 29,739 infusions in an HOPD (among 7,481 patients) and 272,835 infusions in alternative SOCs (among 52,451 patients). For the primary analysis, the matched sample consisted of 26,380 infusions in the HOPD setting (6,958 patients) and 26,380 infusions in alternative SOCs (12,849 patients) with 3,359 HOPD infusions unmatched and 246,455 alternative SOC infusions unmatched. Among the alternative SOC infusions, 21,462 occurred in an office, 4,393 occurred in a home, and 525 occurred in an AIC. Infusion sample attrition is presented in Table 1.
TABLE 1.
Inclusion Criteria for Analysis Cohort
| Adverse events, health care utilization, and cost sample attrition | Count of infusions |
|---|---|
| Infusions to treat ≥1 of 7 chronic conditions between January 1, 2022, and December 31, 2023 | 604,644 |
| Infusions of patients with ≥6 months of continuous medical enrollment prior to infusion date and ≥7 days of continuous medical enrollment postinfusion date | 389,290 |
| Infusions of patients with only 1 study infused agent on infusion date and only 1 of the 7 chronic conditions in prior 6 months | 302,574 |
| Propensity score–matched infusions for 1-day and 7-day postinfusion cost and utilization outcomes | 52,760 |
| Adherence sample attrition | Count of infusions |
| First infusion in sequence with 8 of the 23 infused agents, including in medication adherence analysis | 6,358 |
| Infusions of patients with 12 months’ prior and 12 months’ post–continuous enrollment | 3,485 |
| Propensity score–matched infusions for 6-month medication adherence analysis | 594 |
| Propensity score–matched infusions for 12-month medication adherence analysis | 410 |
In the prematch sample, significant differences in patient characteristics were observed between those receiving infusions in HOPDs compared with alternative SOCs. Compared with the alternative SOC group, patients receiving infusions in an HOPD were younger (mean age 48.8 vs 50.9 years), more likely to be female (68.5% vs 67.8%), more likely to reside in rural areas (14.2% vs 11.8%), and more likely to be Black (9.4% vs 5.8%). Following matching, all standardized mean differences for the covariates were below 0.07 except for the indicator for an infusion agent other than the study infusion agents in the prior 6 months (standardized mean difference = 0.145, 62.5% of HOPD infusions vs 55.4% of alternative SOC infusions) (Table 2).
TABLE 2.
Baseline Characteristics of HOPD and Alternative SOC Infusions Before and After Matching
| Before matching | After matching | |||||
|---|---|---|---|---|---|---|
| Baseline characteristic | HOPD (n = 29,739) | Alternative SOC (n = 272,835) | SMD | HOPD (n = 26,380) | Alternative SOC (n = 26,380) | SMD |
| Patient baseline characteristics | ||||||
| Age on infusion date (in years) | 48.75 | 50.87 | −0.13 | 48.38 | 48.96 | −0.03 |
| Sex | ||||||
| Female | 69% | 68% | 0.02 | 68% | 70% | −0.03 |
| Male | 31% | 32% | −0.02 | 32% | 30% | 0.03 |
| Race and ethnicity | ||||||
| Asian | 2% | 3% | −0.08 | 2% | 3% | −0.03 |
| Black | 9% | 6% | 0.12 | 10% | 8% | 0.06 |
| White | 76% | 73% | 0.05 | 76% | 78% | −0.04 |
| Hispanic | 5% | 6% | −0.04 | 5% | 5% | 0.03 |
| Other/unknown | 8% | 12% | − | 7% | 7% | − |
| Census regiona | ||||||
| Northeast | 29% | 14% | − | 29% | 29% | − |
| Midwest | 25% | 21% | − | 27% | 27% | − |
| South | 32% | 40% | − | 29% | 29% | − |
| West | 14% | 25% | − | 15% | 15% | − |
| Rural | 14% | 12% | 0.07 | 14% | 13% | 0.04 |
| SES index | 2.95 | 3.01 | −0.06 | 2.96 | 2.97 | −0.01 |
| Treated chronic condition | ||||||
| Rheumatoid arthritis | 12% | 26% | −0.43 | 13% | 13% | 0.00 |
| Psoriatic arthritis | 2% | 5% | −0.21 | 2% | 2% | 0.00 |
| Multiple sclerosis | 26% | 9% | 0.39 | 22% | 22% | 0.00 |
| Crohn disease | 23% | 24% | −0.03 | 25% | 25% | 0.00 |
| Plaque psoriasis | 0% | 1% | −0.07 | 0% | 0% | 0.00 |
| Osteoporosis | 21% | 17% | 0.09 | 21% | 21% | 0.00 |
| Ulcerative colitis | 16% | 18% | −0.07 | 17% | 17% | 0.00 |
| Health status | ||||||
| ECI = 0 | 24% | 26% | −0.03 | 26% | 24% | 0.04 |
| ECI = 1 | 28% | 27% | 0.04 | 28% | 31% | −0.05 |
| ECI = 2 | 19% | 18% | 0.03 | 19% | 18% | 0.00 |
| ECI ≥ 3 | 29% | 30% | −0.03 | 27% | 27% | 0.01 |
| Lymph | 1% | 1% | 0.06 | 1% | 1% | 0.00 |
| Metastatic | 2% | 1% | 0.07 | 1% | 1% | 0.00 |
| Infusion baseline characteristics | ||||||
| Quarter of year | ||||||
| Infusion Q1 | 23% | 23% | 0.00 | 24% | 24% | 0.00 |
| Infusion Q2 | 25% | 24% | 0.01 | 25% | 25% | 0.00 |
| Infusion Q3 | 26% | 26% | 0.00 | 26% | 25% | 0.01 |
| Infusion Q4 | 26% | 26% | −0.01 | 26% | 26% | −0.01 |
| Utilization 6 months preinfusion | ||||||
| Inpatient visit | 0.10 | 0.07 | 0.06 | 0.10 | 0.07 | 0.05 |
| Outpatient visit | 11.31 | 11.33 | 0.00 | 11.07 | 11.21 | −0.01 |
| OV visit | 5.42 | 5.45 | −0.01 | 5.30 | 5.31 | 0.00 |
| ED visit | 0.17 | 0.15 | 0.04 | 0.17 | 0.14 | 0.05 |
| Other infusion agent | 61% | 52% | 0.18 | 63% | 55% | 0.15 |
| Infusion sequence number | ||||||
| Infusion number = 1 | 32% | 26% | 0.14 | 33% | 33% | 0.00 |
| Infusion number = 2 | 16% | 14% | 0.05 | 16% | 16% | 0.00 |
| Infusion number ≥ 3 | 52% | 60% | −0.17 | 52% | 52% | 0.00 |
| Infusion agent | ||||||
| J0129 | 4% | 8% | −0.24 | 4% | 4% | 0.00 |
| J0202 | 0% | 0% | 0.03 | 0% | 0% | 0.00 |
| J0717 | 1% | 4% | −0.39 | 1% | 1% | 0.00 |
| J0897 | 12% | 12% | 0.00 | 13% | 13% | 0.00 |
| J1602 | 1% | 6% | −0.41 | 1% | 1% | 0.00 |
| J1628 | 0% | 0% | −0.02 | 0% | 0% | 0.00 |
| J1745 | 21% | 26% | −0.13 | 23% | 23% | 0.00 |
| J2323 | 15% | 6% | 0.25 | 12% | 12% | 0.00 |
| J2327 | 1% | 1% | 0.05 | 1% | 1% | 0.00 |
| J2350 | 11% | 4% | 0.23 | 10% | 10% | 0.00 |
| J3111 | 2% | 3% | −0.04 | 3% | 3% | 0.00 |
| J3245 | 0% | 0% | −0.10 | 0% | 0% | 0.00 |
| J3262 | 1% | 3% | −0.17 | 2% | 2% | 0.00 |
| J3358 | 1% | 1% | 0.04 | 1% | 1% | 0.00 |
| J3380 | 14% | 18% | −0.12 | 16% | 16% | 0.00 |
| J3489 | 7% | 2% | 0.17 | 6% | 6% | 0.00 |
| J9312 | 2% | 2% | −0.01 | 2% | 2% | 0.00 |
| Q5103 | 4% | 3% | 0.09 | 4% | 4% | 0.00 |
| Q5104 | 1% | 1% | 0.05 | 1% | 1% | 0.00 |
| Q5115 | 0% | 0% | 0.03 | 0% | 0% | 0.00 |
| Q5119 | 1% | 0% | 0.07 | 0% | 0% | 0.00 |
| Q5121 | 1% | 1% | 0.01 | 0% | 0% | 0.00 |
| Q5123 | 0% | 0% | 0.02 | 0% | 0% | 0.00 |
Infusions were exact-matched on state of residence.
ECI = Elixhauser comorbidity index; ED = emergency department; HOPD = hospital outpatient department; OV = office visit; SOC = site of care; SES = socioeconomic status; SMD = standardized mean difference.
ADVERSE EVENTS, HEALTH CARE UTILIZATION, AND COST
Although adverse events are rare, there were no statistically significant differences in serious, mild, or any infusion-related adverse events among patients receiving infusions in HOPDs compared with alternative SOCs for 1-day or 7-day measures (Table 5). Most adverse events occurred soon after infusion, with 6,652 occurring within 1 week and 3,539 (or 53.2%) within 1 day of infusion.
TABLE 5.
Adverse Event Outcomes 1 Day and 7 Days Postinfusion Date
| Infusions in HOPDs | Infusions in alternative SOCs | ||||
|---|---|---|---|---|---|
| Outcome measures | Patients with event, n (%) | Patients with event, n (%) | Estimate, odds ratio (95% CI) | P value | |
| 1 Day postinfusion | Mild | 24 (0.09) | 21 (0.08) | 1.14 (0.64-2.05) | 0.6550 |
| Severe | 4 (0.02) | 6 (0.02) | 0.67 (0.19-2.36) | 0.5299 | |
| Any adverse event | 172 (0.65) | 173 (0.66) | 0.99 (0.81-1.22) | 0.9563 | |
| 7 Days postinfusion | Mild | 94 (0.36) | 83 (0.31) | 1.13 (0.84-1.52) | 0.4060 |
| Severe | 15 (0.06) | 25 (0.09) | 0.60 (0.32-1.14) | 0.1178 | |
| Any adverse event | 337 (1.28) | 316 (1.2) | 1.07 (0.92-1.24) | 0.4025 |
HOPD = hospital outpatient department; SOC = site of care.
Within 1 day of receiving an infusion, the odds of having an inpatient admission (P = 0.81), ED visit (P = 0.90), emergency hospitalization (P = 0.78), observational stay (P = 0.21), and medication fill (P = 0.19) were similar for patients receiving infusions in an HOPD compared with alternative SOCs. Results were similar within 7 days postinfusion, except that infusions in HOPDs were associated with lower odds of a medication fill (odds ratio [OR], 0.91; P < 0.01) and higher odds of an inpatient admission (OR, 1.45; P = 0.04) compared with infusions in alternative SOCs (Table 3).
TABLE 3.
Utilization Outcomes 1 Day and 7 Days Postinfusion Date
| Infusions in HOPDs | Infusions in alternative SOCs | ||||
|---|---|---|---|---|---|
| Outcome measures | Patients with event, n (%) | Patients with event, n (%) | Estimate, odds ratio (95% CI) | P value | |
| 1 Day postinfusion | ED visit | 31 (0.12) | 30 (0.11) | 1.03 (0.63-1.71) | 0.8981 |
| Inpatient visit | 9 (0.03) | 8 (0.03) | 1.13 (0.43-2.92) | 0.8085 | |
| Emergency hospitalization | 7 (0.03) | 6 (0.02) | 1.17 (0.39-3.47) | 0.7817 | |
| Pharmacy fill | 1,639 (6.21) | 1,712 (6.49) | 0.95 (0.89-1.02) | 0.1905 | |
| 7 Days postinfusion | ED visit | 163 (0.62) | 155 (0.59) | 1.05 (0.84-1.31) | 0.6537 |
| Inpatient visit | 74 (0.28) | 51 (0.19) | 1.45a (1.02-2.08) | 0.0408 | |
| Emergency hospitalization | 210 (0.8) | 193 (0.73) | 1.09 (0.89-1.33) | 0.3972 | |
| Pharmacy fill | 4,474 (16.96) | 4,818 (18.26) | 0.91b (0.87-0.96) | 0.0001 |
ED = emergency department; HOPD = hospital outpatient department; SOC = site of care.
Infusion claim costs were 41.8% ($4,356; P < 0.01) higher for infusions occurring in an HOPD than infusions occurring in an alternative SOC. Subsequently, patients receiving infusions in an HOPD had 41.9% ($4,337; P < 0.01) higher outpatient costs within 1 day of the infusion and 41.5% ($4,357; P < 0.01) higher outpatient costs within 7 days of the infusion compared with infusions in alternative SOCs. This difference in costs is also observed in total costs 1 day (41.7% higher; P < 0.01) and 7 days (40.5% higher; P < 0.01) postinfusion. There were not statistically significant differences in 1-day or 7-day ED, inpatient, or prescription costs (Table 4).
TABLE 4.
Cost Outcomes 1 Day and 7 Days Postinfusion Date
| Infusions in HOPDs | Infusions in alternative SOCs | ||||
|---|---|---|---|---|---|
| Outcome measures | Average cost mean (SD) | Average cost mean (SD) | Estimate, relative risk (95% CI) | P value | |
| 1 Day postinfusion | Total cost | 10,415 (10,449) | 7,352 (9,691) | 1.42a (1.40-1.43) | <0.0001 |
| Outpatient cost | 10,346 (10,415) | 7,290 (9,675) | 1.42a (1.40-1.43) | <0.0001 | |
| Infusion claim cost | 10,172 (10,271) | 7,172 (9,588) | 1.42a (1.40-1.43) | <0.0001 | |
| ED cost | 3.64 (132.2) | 3.83 (151.2) | 0.95 (0.5-1.82) | 0.8776 | |
| Inpatient cost | 2.33 (181) | 1.47 (102.3) | 1.59 (0.45-5.6) | 0.4719 | |
| Pharmacy cost | 29.21 (548.2) | 25.79 (492.8) | 1.13 (0.82-1.55) | 0.4417 | |
| 7 Days postinfusion | Total cost | 10,726 (10,640) | 7,633 (9,838) | 1.41a (1.39-1.42) | <0.0001 |
| Outpatient cost | 10,508 (10,521) | 7,428 (9,740) | 1.41a (1.40-1.43) | <0.0001 | |
| ED lost | 19.46 (345.5) | 18.78 (317.1) | 1.04 (0.77-1.39) | 0.8122 | |
| Inpatient cost | 50.41 (1,271) | 31.5 (956.6) | 1.6b (0.99-2.58) | 0.0531 | |
| Pharmacy cost | 95.41 (1,025) | 106.7 (1,316) | 0.89 (0.73-1.09) | 0.2658 |
Costs are winsorized at the 5th and 95th percentiles. ED, pharmacy, and inpatient costs are not winsorized as over 93% of those costs are zero.
P = 0.01.
P = 0.10.
ED = emergency department; HOPD = hospital outpatient department; SOC = site of care.
ADHERENCE
For the supplemental analysis on adherence, the matched patient-level sample consisted of 410 patients with a qualified index infusion, with 205 patients receiving their index infusion and all post-index infusions in an HOPD matched to 205 patients receiving all infusions in alternative SOCs. Measuring adherence 1 year following first infusion, there were not statistically significant differences; OvERR without a gap was 0.76 for infusions in HOPD and 0.72 for infusions in alternative SOC (P = 0.22) and OvERR with a gap was 0.87 for infusion in HOPD and 0.83 for infusion in alternative SOC (P = 0.21) (Supplementary Table 2 (266.8KB, pdf) ). However, given our study’s sample size and we only have 80% power to detect a difference in OvERR of 0.089 (at 5% significant level) between patients who received infusions in an HOPD and those who received infusions in an alternative SOC. A sensitivity analysis with index infusions among patients with 6 months pre–continuous enrollment and post–continuous enrollment had similar results (Supplementary Table 2 (266.8KB, pdf) ). A more detailed analysis of the relationship between SOC and continuity of care was beyond the scope of this study.
Discussion
This study is among the first to evaluate the relationship between the infusion SOC and patient outcomes for nononcological biologic and specialty drug infusions across multiple chronic conditions. Prior research has focused primarily on comparisons between a single alternative SOC and HOPDs. In contrast, our study examined outcomes across HOPDs and a combined group of alternative SOCs—AICs, patients’ homes, and physicians’ offices—using a nationally representative, exact and propensity score–matched administrative claims dataset. This broader analysis offers new insights into how infusion therapies occurring in an HOPD may impact health care utilization, cost, short-term clinical outcomes, and infusion therapy adherence compared with alternative SOCs among a diverse population of individuals with chronic conditions.
Consistent with prior literature, we found that infusions administered in HOPDs were associated with statistically significant higher outpatient costs within both 1 and 7 days following infusion. These findings affirm that HOPDs are a higher-cost setting in the management of chronic conditions, such as rheumatoid arthritis, ulcerative colitis, and multiple sclerosis. These cost differences persisted after adjusting for infusion characteristics, patient-level demographics, comorbidities, and baseline health care utilization. Higher total costs also resulted in higher out-of-pocket costs; in our study, patients with infusions in HOPD paid 21% more out of pocket than patients with infusions in alternative SOCs (data not shown). Therefore, alternative SOCs are not only less expensive for insurers, employers, and health care systems but also for patients.
In contrast to cost outcomes, we observed few statistically significant differences in short-term clinical quality or health care utilization–related outcomes between HOPDs and alternative SOCs. Patients receiving infusions in alternate SOCs were not more likely to experience ED visits, emergency hospitalizations, or infusion-related adverse events compared with those in HOPDs, suggesting that the higher cost of HOPD-administered infusions is not associated with fewer infusion-related adverse events or reduced emergent follow-ups within the 1- to 7-day observation window. These findings align with previous oncology-focused research that found similar or better-quality health care outcomes in physician offices compared with HOPDs.2,17 Our study expands this conclusion to a broader, nononcological population and to additional alternative SOCs.
Among the statistically significant differences, HOPDs were associated with lower odds of a medication fill (7 days postinfusion) as well as higher odds of an inpatient admission (7 days postinfusion). Alternative SOCs, such as physicians’ offices, may offer patients an opportunity to address additional health care needs, which may be related to the increased odds of a medication fill. In these settings, providers may not only be administering their infusions but also managing the infusion-related condition and attending to other health issues. For example, past studies have found that those receiving chemotherapy infusions in physicians’ offices received medications to proactively manage treatment side effects2,3 at higher rates than those who received infusions in HOPDs. This may include prescriptions for other treatments and routine outpatient care that are not included when receiving infusions in HOPDs.
It should be noted that 1-day inpatient admissions did not differ between groups, and this first day is when differences in outcomes directly associated with the infusion would be most apparent. Nevertheless, we observed higher odds of admission in the HOPD cohort during the week following infusion. One possible reason for this is potential bias from confounding due to differences in unobserved characteristics not included in the matching procedure. However, our analysis of utilization 2-4 weeks postinfusion does not show increases in hospitalizations, making this reason unlikely. As previously described in the methods, patients’ prior health status and health care utilization, including hospital admissions and ED visits, were accounted for in this analysis, to minimize the difference in health status between groups, yet some differences may remain. Future studies should further explore these differences.
Our study provides additional evidence supporting the broader adoption of alternative SOCs for infusion delivery in clinically appropriate circumstances. Given the comparable health care outcomes and statistically significant lower costs in these settings, policymakers and payers should implement strategies to leverage their value. These findings also underscore the importance of assessing medical necessity before directing infusions into HOPDs and suggest that expanding access to alternative SOCs may offer patient-level and system-level savings opportunities. To this end, some payers have already begun implementing care programs that require documentation of medical necessity to receive coverage of infusion therapy at HOPDs and transition care to alternate SOCs when there is not a documented clinical need to receive infusion therapy at a more intensive site of care.18–20
LIMITATIONS
There are several limitations to this study. First, this is a retrospective cohort study using administrative data, which makes it subject to the limitations typical of observational studies, such as selection bias, unmeasured confounding, and measurement error. Second, this analysis compares infusion therapy in HOPDs to a group of 3 alternative SOCs; findings may not be applicable to each individual alternative SOC and additional analysis should further explore this. Third, this analysis combines multiple chronic conditions and infused agents and results may not apply to individual medical conditions or infused agents. Fourth, the adverse event outcomes were evaluated at 1 day and 7 days postinfusion, and we did not evaluate delayed reactions following the infusion. Fifth, the infusion adherence outcomes are based on a small sample with 2 years of continuous enrollment and may benefit from a more robust sample, particularly as our study was underpowered to detect moderate differences in adherence. Sixth, this study only included individuals insured in commercial insurance plans, so we do not know how these results would apply to those insured by Medicaid and Medicare. Finally, we may not have excluded all infusions that were deemed medically necessary to occur in HOPDs.
Conclusions
Individuals receiving infusion therapies in HOPDs experience higher outpatient costs but demonstrate no associated clinical benefit in terms of short-term adverse events or reduced ED utilization. Infusions delivered in alternative SOCs—home, physician offices, and AICs—achieve comparable quality outcomes at a lower cost for a range of common chronic conditions. These findings support ongoing efforts to assess the medical necessity of HOPD-based infusions and to expand access to lower-cost, high-quality alternative settings where appropriate.
References
- 1.Winn AN, Keating NL, Trogdon JG, Basch EM, Dusetzina SB. Spending by commercial insurers on chemotherapy based on site of care, 2004-2014. JAMA Oncol. 2018;4(4):580-1. doi: 10.1001/jamaoncol.2017.5544 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Fisher MD, Punekar R, Yim YM, et al. Differences in health care use and costs among patients with cancer receiving intravenous chemotherapy in physician offices versus in hospital outpatient settings. J Oncol Pract. 2017;13(1):e37-46. doi: 10.1200/JOP.2016.012930 [DOI] [PubMed] [Google Scholar]
- 3.Mahendraratnam N, Farley JF, Basch E, Proctor A, Wheeler SB, Dusetzina SB. Characterizing and assessing antiemetic underuse in patients initiating highly emetogenic chemotherapy. Support Care Cancer. 2019;27(12):4525-34. doi: 10.1007/s00520-019-04730-3 [DOI] [PubMed] [Google Scholar]
- 4.Levin JS, Zhao X, Whaley C. Impact of hospital-physician vertical integration on physician-administered drug spending and utilization. Health Econ. 2025;34(2):345-67. doi: 10.1002/hec.4909 [DOI] [PubMed] [Google Scholar]
- 5.Alvarez E, Nair KV, Tan H, et al. Real-world cost of care and site of care in patients with multiple sclerosis initiating infused disease-modifying therapies. J Med Econ. 2023;26(1):494-502. doi: 10.1080/13696998.2023.2194185 [DOI] [PubMed] [Google Scholar]
- 6.Wasserman RL, Ito D, Xiong Y, Ye X, Bonnet P, Li-McLeod J. Impact of site of care on infection rates among patients with primary immunodeficiency diseases receiving intravenous immunoglobulin therapy. J Clin Immunol. 2017;37(2):180-6. doi: 10.1007/s10875-017-0371-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Baker MC, Weng Y, Fairchild R, Ahuja N, Rohatgi N. Comparison of adverse events among home- vs facility-administered biologic infusions, 2007-2017. JAMA Netw Open. 2021;4(6):e2110268. doi: 10.1001/jamanetworkopen.2021.10268 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Barrera B, Simpson H, Engebretson E, et al. Safety and patient experience with at-home infusion of ocrelizumab for multiple sclerosis. Ann Clin Transl Neurol. 2023;10(4):579-88. doi: 10.1002/acn3.51745 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Bolge SC, Eldridge HM, Lofland JH, Ravin C, Hart PJ, Ingham MP. Patient experience with intravenous biologic therapies for ankylosing spondylitis, Crohn’s disease, psoriatic arthritis, psoriasis, rheumatoid arthritis, and ulcerative colitis. Patient Prefer Adherence. 2017;11:661-9. doi: 10.2147/PPA.S121032 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mansour O, Arbaje AI, Townsend JL. Patient experiences with outpatient parenteral antibiotic therapy: results of a patient survey comparing skilled nursing facilities and home infusion. Open Forum Infect Dis. 2019;6(12):ofz471. doi: 10.1093/ofid/ofz471 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Carelon Research . Carelon Research’s Healthcare Integrated Research Database (HIRD). Carelon Research; 2024. Accessed July 2, 2025. https://www.carelonresearch.com/content/dam/digital/carelon/pdf/Carelon_Research_HIRD_technical_white_paper_March_2024_fv.pdf
- 12.American Academy of Allergy, Asthma, and Immunology . Guidelines for the site of care for administration of IGIV therapy. 2011. Accessed July 2, 2025. https://www.aaaai.org/Aaaai/media/Media-Library-PDFs/Practice%20Management/Practice%20Tools/Guidelines-for-the-site-of-care-for-administration-of-IGIV-therapy.pdf
- 13.CarelonRx . Managing inflammatory conditions and specialty drug trend. 2023. Accessed July 2, 2025. https://www.carelonrx.com/perspectives/inflammatory-conditions-specialty-drug-trend
- 14.International RTI . Creation of New Race-Ethnicity Codes and Socioeconomic Status (SES) Indicators for Medicare Beneficiaries. Agency for Healthcare Research and Quality; 2008. Accessed July 2, 2025. https://citeseerx.ist.psu.edu/document?repid=rep1&type=pdf&doi=abd8d0c761bc4f0fcf9593727e93ef8c04cf1c0e [Google Scholar]
- 15.Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130-9. doi: 10.1097/01.mlr.0000182534.19832.83 [DOI] [PubMed] [Google Scholar]
- 16.Tkacz J, Ingham MP, Brady BL, Meyer R, Ruetsch C. Novel adherence measures for infusible therapeutic agents indicated for rheumatoid arthritis. Am Health Drug Benefits. 2015;8(9):494-505. 10.1097/01.mlr.0000182534.19832.83 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Hopson S, Casebeer A, Stemkowski S, et al. Does site-of-care for oncology infusion therapy influence treatment patterns, cost, and quality in the United States? J Med Econ. 2018;21(2):152-62. doi: 10.1080/13696998.2017.1384736 [DOI] [PubMed] [Google Scholar]
- 18.Aetna . Drug infusion/Injection site of care policy. Accessed November 11, 2025. https://www.aetna.com/health-care-professionals/utilization-management/drug-infusion-site-of-care-policy.html
- 19.UnitedHealthcare . Provider administered drugs – site of care. UnitedHealthcare Commercial Medical Benefit Drug Policy. 2025. Accessed November 11, 2025. https://www.uhcprovider.com/content/dam/provider/docs/public/policies/comm-medical-drug/provider-administered-drugs-soc.pdf
- 20.Cigna Healthcare . Medication administration site of care. Drug coverage policy. 2024. Accessed November 11, 2025. https://static.cigna.com/assets/chcp/pdf/coveragePolicies/pharmacy/ph_1605_coveragepositioncriteria_site_of_care.pdf