Integrated healthcare delivery systems in the US—such as Kaiser Permanente (KP)—have relied on service co-location as a model of care to facilitate patients’ access to ambulatory care. In a model developed on co-location, the patient–provider visits occur in-person and at the same clinic where supporting ancillary services (diagnostic radiology, laboratory, and pharmacy) are readily available – “one-stop shopping”. Co-location eliminates some direct and opportunity costs to a patient that would otherwise be required to coordinate travel to multiple locations for completing a patient–provider consult and for fulfilling physician orders for associated diagnostic or therapeutic services.

In recent years, health care systems have implemented virtual visits as substitutes for in-person visits. A virtual model of ambulatory care delivery “dis-locates” the place and time of service delivery otherwise enabled in a co-location model of ambulatory care delivery. Physicians and patients are in different locations; patients are not (typically) at a clinic for a history and physical, and patients are remote from medical facilities where they could expeditiously fulfill ancillary services orders. Nevertheless, patients report acceptance, convenience, and satisfaction with the ability to choose a virtual or in-person visit.1–10 These favorable patient reports suggest virtual visits improve ambulatory care access by lowering direct or opportunity costs for some patients.1,4–6,8,9,11,12

Following the access to care framework of Aday, Andersen, and colleagues,13–15 we examined 3 factors—or moderators—of ancillary services order fulfillment: distance from residence to clinic, level of cost-sharing in service fulfillment defined by insurance plan benefits, and prior experience with pharmacy mail-order fulfillment as an alternative to facility-based fulfillment. In prior analyses, we studied whether visit mode was associated with patient fulfillment of ancillary services orders and found a modestly—sometimes significantly—lower frequency with virtual versus in-person visits for neck or back pain (NBP) or urinary tract infection (UTI).16,17 Our primary study objective was to evaluate if any of these 3 factors moderated the likelihood of patient fulfillment of ancillary services ordered by a physician on an ambulatory care visit differed by visit mode (virtual vs. in-person) or COVID-19 period [prepandemic (before beginning of the national emergency in April 2020) or recovery (after June 2020)].

CONCEPTUAL MODEL Distance

Distance from residence to the clinic is characterized by a number of features that affect healthcare access: transportation mode, travel distance, and travel time (including direct and opportunity costs). These features can affect the likelihood of service order fulfillment since many ancillary services (diagnostic radiology examinations, drawing a blood specimen or collecting a urine sample, dispensing a medication) remain facility-based and require the co-location of provider and patient. Distance from residence to the clinic has been associated with appointment keeping, preventive care rates, and medication adherence uptake and persistence.18–29

Our study provided an opportunity to evaluate if there might be a differential impact of distance on ancillary services fulfillment for virtual versus in-person visits. Our working hypothesis was greater distance (and, therefore, greater direct and opportunity costs) would be inversely associated with the frequency of ancillary services generally, and specifically for virtual versus in-person visits.

Cost-Sharing

Beginning with the RAND Health Insurance Experiment,30 many studies, particularly in prescribed medication utilization, have affirmed an inverse relationship between patient cost-sharing and medical service utilization.31–35 “High deductible health plans” (HDHPs), introduced during the 1980s, shifted a higher proportion of health care costs to patients. For exempt services (eg, vaccinations), HDHPs have minimal impact on medical services utilization compared with traditional HMO plans34–41; however, patient fulfillment of nonexempt services or services conditional on a nonexempt service with high patient cost-sharing may be lower in HDHPs.37

Our study provided an opportunity to evaluate if there might be a differential impact of HDHP enrollment on ancillary services fulfillment for virtual versus in-person visits. Our working hypothesis was that HDHP enrollees would have lower ancillary services fulfillment if the order was exempt from cost-sharing (eg, routine x-ray services) but significant decreases in ancillary services fulfillment for which patient cost-sharing in HDHPs was required and high (eg, CT imaging).

Prior Mail-Order Pharmacy Fulfillments

A mail-order alternative to in-person prescription fulfillment has been offered in recent years by retail pharmacies, the VA, and KP with the goal of improving primary and secondary adherence by making medication order fulfillment more convenient than in-person fulfillment.42–44 Mail-order fulfillment is often offered at a reduced cost to the patient (eg, waived or lower copayment than would be required of an in-person refill) and can be initiated by Internet or telephone requests. During the COVID-19 pandemic, the safety and convenience of mail-order medication fulfillment was widely marketed.

Our study provided an opportunity to evaluate if there might be a differential impact of HDHP enrollment on prescription medication fulfillment for virtual versus in-person visits. During the national COVID-19 shutdown, when in-person services at KP clinics were curtailed, mail-order pharmacy fulfillment would have been an attractive option for medication order fulfillment.42,45–48 Our working hypothesis was that prior use of a KP mail-order pharmacy program would promote the fulfillment of prescription medication orders because patients with prior use would understand how to navigate the program and its benefits (no or waived copayments, the convenience of home ordering and delivery).

METHODS Study Settings

Eligible participants were identified from our study’s common data model (CDM) implemented at 3 KP Regions: Kaiser Permanente Colorado (KPCO), Georgia (KPGA), and Mid-Atlantic States (KPMAS). All 3 KP Regions are primarily group-model integrated delivery systems but differ in important patient population demographic characteristics and mode and timing of virtual visit implementation.41,42

Under a ceding arrangement, the IRBs of KPCO, KPGA, and KPMAS collectively reviewed and approved the study protocol.

Data Sources

The samples of patients with incident NBP or UTI visits for 2016-June 2021, visit modes and pandemic phases, and classes of ancillary services associated with incident NBP or UTI visits were extracted from each site’s identically configured CDM using distributed programming code.49

Study Sample

An incident NBP visit was defined as: a completed visit (ie, cancelled appointments and opened but incomplete visits were excluded); with an ICD-10 diagnosis identified as primary or principal; be provided as routine or urgent ambulatory care (defined by the department of service – such as the department of family medicine, general internal medicine, or urgent or after hours care or an equivalent virtual visit program such as KPCO synchronous chats or KPMAS “house calls”), provided to a patient 19 years or older of age, and, with no NBP visit within the prior 180 days.16 An incident UTI visit was similarly defined, except the interval between visits was set at 90 days.17

Measures Service Order Fulfillments

Medical service was considered to be fulfilled by the patient if at least 1 order in a medical service class was fulfilled within 30 days of the order date associated with an NBP or UTI visit. For NBP visits, service classes were: neck or back x-ray, neck or back CT or MRI, non-narcotic analgesic, narcotic analgesic, and skeletal muscle relaxant. For UTI visits, service classes were: urinalysis, urine culture, abdominal CT or MRI or ultrasound, first-line antibiotic, or second-line antibiotic. Order fulfillment within the service class is a dichotomous variable.

Moderator Variables

The distance was measured as a straight-line distance between the latitude and longitude of a person’s residence and the latitude and longitude of the nearest KP primary care clinic (PCC). We assumed the PCC would be the most likely facility where the patient would travel to fulfill related ancillary services if the incident NBP or UTI visit were virtual. The distance was dichotomized at the median distance of the empiric distribution for each KP Region: 3.3 miles at KPCO, 5.7 miles at KPGA, and 4.0 miles at KPMAS.

Cost-sharing was defined as whether or not the patient was enrolled in an HDHP versus a traditional HMO plan at the time of the NBP or UTI visit and up to 6 months prior. HDHP was a broad category inclusive of all deductible plans (with or without a federally qualified HSA) compared with a traditional HMO plan with first-dollar coverage of medical services and nominal copayments.

Prior use of the KP pharmacy mail-order option was defined as evidence in the EHR that the patient had fulfilled at least 1 prescription dispensing at any time for any medication before the NBP or UTI visit.

Visit Mode

Each incident NBP or UTI visit was classified as virtual or in-person. A virtual visit was defined as a technology-mediated (phone or computer) synchronous patient–provider interaction. Virtual visits included synchronous chats, telephone visits, and video visits. An in-person visit was defined as a face-to-face patient–provider interaction within an examination room at a KP clinic.

Pandemic Periods

We defined 3 COVID-19 pandemic periods when significant shifts in choice and availability of visit mode in the 3 KP Regions occurred: prepandemic (before April 2020 when the availability of virtual visits was gradually increasing), “national shutdown” (April 2020-June 2020 when limited in-person ambulatory services were available), and recovery (July 2020-June 2021 when in-person primary care became widely available again). In our current study, we examine NBP and UTI visits only during the prepandemic and recovery periods, given the restricted availability of in-person visits during the short national shutdown period.

Patient Covariates

The study CDM was our source of data for measuring: patient age at the time of the visit, gender, race and ethnicity, Charlson comorbidity score based on ICD-10 diagnoses on any health care event in the 365 days before the visit,50–52 and national percentile of an area disadvantage index for the patient’s residence at the time of the visit.53,54

Statistical Analysis

We used inverse probability of treatment weighting (IPTW) to create weight sets for addressing the imbalance in patient case mix between visit modes and pandemic periods.55,56 Weight sets were created separately for combinations of clinical condition, KP Region, and ancillary service class for the clinical condition. Weight sets were estimated for 1 set of IPTW models with visit mode (virtual vs. in-person) as the dependent variable and pandemic period (recovery vs. prepandemic) as the dependent variable in another set. Independent variables in each IPTW model were the aforementioned “Patient Covariates”. This approach resulted in weight sets to balance patient case mix by visit mode or by pandemic period.

To evaluate a moderator effect on the fulfillment of an order within the service class, we cross-classified fulfillment (Yes/No) with a moderator: above versus below median distance to PCC, enrollment versus not in an HDHP, and prior pharmacy mail-order use versus not. Observations were weighted using the appropriate IPTW weight set. Distance to PCC and HDHP enrollment were considered moderators of all 5 ancillary service classes for NBP and UTI visits; prior pharmacy mail-order use was considered as a moderator only for patient fulfillment of medication orders (3 classes for NBP, 2 classes for UTI). The significance of a moderator effect in each cross-classification was assessed with a 2-tailed χ2 test and P value of ≤0.05.

We then conducted a series of analyses on the distribution of ratios of the percentage of order fulfillments in an attempt to ascertain if a moderator was associated with a disparity in order fulfillment for each clinical condition. Ratios were computed from the aforementioned IPTW cross-classifications (eg, 80% fulfillment of x-ray orders given above median distance and 90% fulfillment for below median distance would be 0.889). We compared the medians of ratios for visit mode within period (“Does a moderator affect ancillary services fulfillment due to visit mode independent of the pandemic period?”) and period within visit mode (“Does a moderator affect ancillary services fulfillment due to period independent of visit mode?”). A Wilcoxon test was used to compare medians of ratios between comparator sets (ie, virtual vs. in-person visit within period, or prepandemic vs. recovery within mode). The significance of a pattern of disparity in order fulfillment was assessed with a 2-tailed median test and P value of ≤0.05.

All data management and statistical tests were conducted using SAS version 9.4 (SAS Institute, Cary, NC).

RESULTS

Table 1 presents a summary, in heuristic terms, of IPTW adjusted association of order fulfillment with each of the moderators. For NBP and UTI, 15 cross-classifications (3 KP Regions and 5 service classes) were generated to assess the moderating effects of PCC distance and HDHP enrollment with order fulfillment. For NBP, 9 cross-classifications (3 KP Regions and 3 medication classes) were generated to assess the moderating effects of prior pharmacy mail-order, with medication order fulfillment; for UTI, 6 cross-classifications (3 KP Regions and 2 medication classes) were generated.

TABLE 1 - Qualitative Summary of Moderator Associations With Patient Fulfillment of Ancillary Service Orders Period Prepandemic Recovery Condition Mode Moderator Number of associations significant for P≤0.05 Fulfillment effect and direction of association Number of associations significant for P≤0.05 Fulfillment effect and direction of association NBP Virtual Above median distance to clinic 2/15 ↓2/2 1/15 ↓1/1 Enrollment in HDHP 1/15 ↓ 1/1 4/15 ↓4/4 Prior mail-order pharmacy use 3/9 ↑3/3 6/9 ↑ 6/6 In-person Above median distance to clinic 4/15 ↓2/4 3/9 ↓3/3 Enrollment in HDHP 9/15 ↓6/9 8/15 ↓5/8 Prior mail-order pharmacy use 9/9 ↑6/9 5/9 ↑ 2/5 UTI Virtual Above median distance to clinic 2/15 ↓2/2 1/15 ↓1/1 Enrollment in HDHP 6/15 ↓4/6 5/15 ↓5/5 Prior mail-order pharmacy use 5/6 ↑3/5 1/6 ↑1/1 In-person Above median distance to clinic 2/15 ↓1/2 1/15 ↓ 0/1 Enrollment in HDHP 7/15 ↓4/7 2/15 ↓ 2/2 Prior mail-order pharmacy use 2/6 ↑1/2 1/6 ↑ 1/1

↓ indicates that the moderator suppressed patient fulfillment of provider order of a service (the expected effect of the moderator among associations significant for P≤0.05).

↑ indicates that the moderator promoted patient fulfillment of provider order of a service (the expected effect of the moderator among associations significant for P≤0.05).

HDHP indicates high deductible health plan; NBP, neck or back pain.

For both NBP and UTI, few of the associations were significant for mode or period (Table 1). For example, moderator effects for virtual NBP visits were significant during the prepandemic period for 2/15 associations on distance, 1/15 on cost-sharing, and 3/9 on prior mail-order pharmacy use. Where significant, however, most of those associations were in the expected direction. For example, significant moderator effects for virtual NBP visits during the prepandemic period indicated longer distance (2/2) and higher cost-sharing (1/1) suppressed fulfillment, whereas prior mail-order pharmacy use promoted medication fulfillment (3/3).

Details on fulfillment percentages, given a service order within class, are displayed in the Supplement for NBP (Tables S1–S6, Supplemental Digital Content 1, https://links.lww.com/MLR/C597) and UTI visits (Tables S7–S12, Supplemental Digital Content 1, https://links.lww.com/MLR/C597). Overall, fulfillment percentages were high for both NBP and UTI visits, often>70%. Fulfillment percentages were lowest for CT/MRI scans with NBP visits (several instances in the 40–50% range; Tables S4–S5, Supplemental Digital Content 1, https://links.lww.com/MLR/C597) and UTI visits (several instances in the 60–70% range; Tables S10-S11, Supplemental Digital Content 1, https://links.lww.com/MLR/C597).

Tables 2–5 display results of analysis of the ratio of the fulfillment percentages by mode and period separately for NBP and UTI visits. Ratios<1 indicate the moderator (above median PCC distance, HDHP enrollment, prior mail-order pharmacy use) suppressed fulfillment of the service order; ratios>1 indicate the moderator enhanced fulfillment.

TABLE 2 - Ratios of Percentages of Ancillary Service Order Fulfillments Associated With Incident NBP Visits: Comparing Visit Modes Within Period Moderator (median and interquartile range) Period Visit mode Distance (Impact of greater distance) HDHP (Impact of HDHP enrollment) Mail-order Rx (Impact of prior use) Prepandemic Virtual 0.990 [0.966, 1.014] 0.993 [0.973, 1.031] 1.059 [1.050, 1.062] In-person 0.997 [0.975, 1.007] 0.998 [0.990, 1.001] 1.020 [0.941, 1.021] P value 0.57 0.95 0.01 Recovery Virtual 0.996 [0.968, 1.013] 0.999 [0.978, 1.012] 1.052 [1.019, 1.062] In-person 0.996 [0.978, 1.014] 0.988 [0.979, 1.000] 1.016 [0.958, 1.018] P value 0.60 0.31 0.02

This table compares the marginal (incremental) effects of a moderator on percentages of order fulfillments between visit modes within a period.

TABLE 3 - Ratios of Percentages of Ancillary Service Order Fulfillments Associated With Incident UTI Visits: Comparing Visit Modes Within Period Moderator (median and interquartile range) Period Visit mode Distance (Impact of greater distance) HDHP (Impact of HDHP enrollment) Mail-order Rx (Impact of prior use) Prepandemic Virtual 0.988 [0.958, 1.058] 0.997 [0.982, 1.017] 1.011 [0.968, 1.027] In-person 0.999 [0.984, 1.004] 1.000 [0.996, 1.002] 1.008 [0.994, 1.028] P value 0.60 0.63 0.75 Recovery Virtual 0.974 [0.944, 1.004] 0.995 [0.969, 1.012] 1.024 [1.020, 1.036] In-person 0.997 [0.980, 1.008] 0.994 [0.987, 0.999] 1.011 [0.971, 1.014] P value 0.34 0.95 0.09

This table compares the marginal (incremental) effects of a moderator on percentages of order fulfillments between visit modes within a period.

TABLE 4 - Ratios of Percentages of Ancillary Service Order Fulfillments Associated With Incident NBP Visits: Comparing Periods Within Visit Mode Moderator (median and interquartile range) Visit mode Period Distance (Impact of greater distance) HDHP (Impact of HDHP enrollment) Mail-order Rx (Impact of prior use) Virtual Prepandemic 0.990 [0.966, 1.014] 0.993 [0.973, 1.031] 1.059 [1.050, 1.062] Recovery 0.996 [0.968, 1.013] 0.999 [0.978, 1.012] 1.052 [1.019, 1.062] P value 0.72 0.95 0.63 In-person Prepandemic 0.997 [0.975, 1.017] 0.998 [0.990, 1.001] 1.020 [0.941, 1.021] Recovery 0.996 [0.978, 1.014] 0.988 [0.979, 1.000] 1.016 [0.958, 1.018] P value 0.72 0.07 0.31

This table compares the marginal (incremental) effects of a moderator on percentages of order fulfillments between periods within a visit mode.

TABLE 5 - Ratios of Percentages of Ancillary Service Order Fulfillments Associated with Incident UTI Visits: Comparing Periods within Visit Mode Moderator (median and interquartile range) Visit mode Period Distance (Impact of greater distance) HDHP (Impact of HDHP enrollment) Mail-order Rx (Impact of prior use) Virtual Prepandemic 0.988 [0.958, 1.058] 0.997 [0.982, 1.017] 1.011 [0.968, 1.027] Recovery 0.974 [0.944, 1.004] 0.995 [0.969, 1.012] 1.024 [1.020, 1.036] P value 0.40 0.52 0.34 In-person Prepandemic 0.999 [0.984, 1.004] 1.000 [0.996, 1.002] [1.0080.994, 1.028] Recovery 0.997 [0.980, 1.008] 0.995 [0.987, 0.999] [1.0110.971, 1.014] P value 0.88 0.03 0.72

This table compares the marginal (incremental) effects of a moderator on percentages of order fulfillments between periods within a visit mode.

Overall, the medians of the ratio distributions suggest decreased order fulfillment with greater distance and higher cost-sharing in HDHPs for service fulfillment (ratios generally<1) and increased medication order fulfillment with prior mail-order pharmacy use (ratios generally>1), on percentages of service order fulfillments (Tables 2–5). With few exceptions (eg, NBP medication orders), the moderator effects on service order fulfillment are small. For example, among virtual NBP visits in the prepandemic period, the ratio of fulfillments for above compared with below median distance to PCC is 0.990 – a relative ±0.010 or 1% difference (Table 2).

Prior use of mail-order prescriptions significantly promoted medication order fulfillments on virtual NBP visits compared with in-person NBP visits in the prepandemic period (5.9% vs. 2.0%, P=0.01; Table 2) and in the recovery period (5.2% vs. 1.6%, P=0.02; Table 2). Otherwise, we observed no significant moderator effect on ancillary services fulfillment due to visit mode independent of the pandemic period for either NBP or UTI visits (Tables 2 and 3, respectively). We observed no significant moderator effect on ancillary services fulfillment due to the pandemic period independent of visit mode for either NBP or UTI.52,56

DISCUSSION

Overall, we found relatively high patient fulfillment of ancillary service orders associated with an NBP or UTI visit. Typically >70–80% of orders in the ancillary service categories that we studied were fulfilled within 30 days of the service order date. This percentage is similar to the 80% cut point often used to declare “adherence” in prescription medication studies.57,58

We found weak moderator effects for distance to PCC and HDHP enrollment. Longer distance to PCC and higher cost-sharing from HDHP enrollment did not suppress diagnostic radiology or laboratory order fulfillment for virtual compared with in-person visits for either NBP or UTI visits. The weak moderator effect for these moderators was observed in both the prepandemic and recovery period for both NPB and UTI visits. We suspected that longer the distance to PCC and higher cost-sharing from HDHP enrollment might present access barriers due to service dislocation. But, in all 3 KP Regions, diagnostic radiology and laboratory services were offered at PCCs relatively close to patients’ residences. The exception might be advanced diagnostic radiology imaging, such as CT/MRIs, which are offered at a limited number of PCCs in each KP Region (and often require future appointments, thus necessitating additional travel) and may have a high patient cost (especially in HDHPs). This might explain the overall low percentage of CT/MRI order fulfillment; however, order fulfillment was indifferent by visit mode or pandemic period.

For NBP visits, prior mail-order pharmacy use was significantly associated with higher percentages of nonnarcotic and narcotic analgesic order fulfillments. The effect was greater for virtual visits (5–6% in the prepandemic and recovery periods) than for in-person visits (1–2%). This is perhaps unsurprising insofar as mail-order pharmacy use mitigates access challenges from disruption of service dislocation in the case of virtual visits. Furthermore, copayments are frequently reduced or waived when prescription orders are fulfilled by mail. Thus, patients who had prior experience with the benefits of KP’s mail-order pharmacy program would realize its benefits to achieving medication adherence.

A few studies have examined patient fulfillments of ancillary services orders on virtual visits or during the COVID-19 pandemic period and found lower fulfillment rates than on in-person visits or before the pandemic.59–63 It is important to note that our approach to the study of patient adherence separates the order process from the fulfillment process; and all of our reported fulfillment percentages are contingent on an order for an ancillary service. In other studies, we found that order likelihood is lower for virtual than for in-person visits.16,17

In this study, we found that 3 moderators that might influence patient treatment adherence did little to account for fulfillment disparities by visit mode.

Findings on patient fulfillment of ancillary services appear to be robust to differences in the delivery system, medical condition, ancillary services classes, and pandemic period. Although each of the 3 sites are part of the Kaiser Permanente Medical Care Program, each KP Region operates independently and adapts national EHR, insurance products, and care pathways to their local market and patient population. For example, during the study period, virtual visit development and deployment in KPCO focused on synchronous chats, but in KPGA and KPMAS, the focus was on video visits.

Limitations

Our measures of moderators are imperfect, though easily measured from EHR data. Straight line distance from a patient’s location on departure to a primary care clinic is an imprecise measure of travel time and cost. We based the distance between the residence and the nearest PCC where the patient presumably could fulfill an ancillary service order. Travel time and costs to access primary care are affected by various circumstances, both persistent and variable over time which we cannot measure (eg, preference for travel from work vs. residence, transportation mode and availability, traffic patterns on any given day at any hour). HDHP enrollment indicates the potential for relatively high patient cost-sharing in services; however, the actual patient cost can vary over the course of a year, depending on whether or not the patient or family has exceeded the annual deductible. Although copayments and coinsurance in HDHPs are fairly uniform and higher than in traditional HMO plans, annual deductibles can vary from $500–$1,000 to over $10,000.

We dichotomized the distance and HDHP moderators and, consequently, may have overlooked other linear or non-linear associations of order fulfillment percentages, given that these measures have more-or-less continuous distributions. Dichotomized mediators helped simplify model specifications, presentations, and interpretations – particularly where NBP or UTI ancillary service orders were infrequent (eg, CT/MRIs).

We studied moderator effects for only 2 conditions—NBP and UTI—which are occasionally chronic, generally not life-threatening, and frequently resolved in the short term. Our findings may not be generalizable to other persistent chronic conditions, serious life-threatening conditions, or preventive care. For example, a patient undergoing chemotherapy for stage 3 or 4 cancer may perceive “long distance” or “high cost” as an access barrier quite differently from a patient with lower back pain who responds to OTC analgesics.

Our use of IPTW to balance patient case-mix by mode or by period incompletely controls for selection effects insofar as it accounts for only observable factors. Other unmeasured and unobserved patient factors might affect a patient’s decision to have presented for care by virtual or in-person mode or during the recovery period versus before the COVID-19 pandemic.

CONCLUSION

Patients who presented for ambulatory management of incident NBP or UTI fulfilled ~70–80% of orders for ancillary (diagnostic radiology, laboratory, pharmacy) services associated with their visit. These relatively high fulfillment percentages differed little by visit mode (virtual vs. in-person) or period (prepandemic vs. recovery). Fulfillment percentages appeared to be little affected by distance from residence to primary care clinic or HDHP enrollment. Prior use of KP’s mail-order pharmacy program may have promoted patient fulfillment of medication orders.

Implementation of virtual visits, alone or in conjunction with the COVID-19 pandemic, contributed to the dis-location of the KP model of co-location of provider visits and ancillary services. Dis-location, if any, did not appear to affect patient fulfillment of ancillary services orders in ambulatory NBP or UTI care.

REFERENCES 1. Cottrell MA, Hill AJ, O’Leary SP, et al. Patients are willing to use telehealth for the multidisciplinary management of chronic musculoskeletal conditions: a cross-sectional survey. J Telemed Telecare. 2018;24:445–452. 2. Greenhalgh T, Shaw S, Wherton J, et al. Real-world implementation of video outpatient consultations at macro, meso, and micro levels: mixed-method study. J Med Internet Res. 2018;20:e150. 3. Greenhalgh T, Vijayaraghavan S, Werton J, et al. Virtual online consultations: Advantages and limitations (VOCAL) study. BMJ Open. 2016;6:e009388. 4. Kruse CS, Krowski N, Rodriguez B, et al. Telehealth and patient satisfaction: a systematic review and narrative analysis. BMJ Open. 2017;7:e016242. 5. Palcu P, Munce S, Jaglal SB. Understanding patient experiences and challenges to osteoporosis care delivered virtually by telemedicine: a mixed methods study. Osteoporos Int. 2020;31:351–361. 6. Polinski JM, Barker T, Gagliano N, et al. Patients’ satisfaction with and preference for telehealth visits. J Gen Intern Med. 2016;31:269–275. 7. Powell RE, Henstenburg JM, Cooper G, et al. Patient perceptions of telehealth primary care video visits. Ann Fam Med. 2017;15:225–229. 8. Powell RE, Stone D, Hollander JE. Patient and health system experience with implementation of an enterprise-wide telehealth scheduled video visit program: mixed-methods study. JMIR Med Inform. 2018;6:e10. 9. Tasneem S, Kim A, Bagheri A, et al. Telemedicine video visits for patients receiving palliative care: a qualitative study. Am J Hosp Palliat Care. 2019;36:789–794. 10. Welch BM, Harvey J, O’Connell NS, et al. Patient preferences for direct-to-consumer telemedicine services: a nationwide survey. BMC Health Serv Res. 2017;17:784. 11. Ashwood JS, Mehrotra A, Cowling D, et al. Direct-to-consumer telehealth may increase access to care but does not decrease spending. Health Aff (Millwood). 2017;36:485–491. 12. Courneya PT, Palattao KJ, Gallagher JM. HealthPartners’ online clinic for simple conditions delivers savings of $88 per episode and high patient approval. Health Aff (Millwood). 2013;32:385–392. 13. Aday LA, Andersen R. A framework for the study of access to medical care. Health Serv Res. 1974;9:208–220. 14. Andersen RM, Davidson PL Andersen RM, Rice TH, Kominski GF. Improving access to care in America: Individual and contextual indicators. Changing the U.S. Health Care System: Key Issues in Health Services Policy and Management San Francisco, CA: Jossey-Bass; 2007:3–31. 15. Wolinsky FD. Assessing the effects of predisposing, enabling, and illness-morbidity characteristics on health service utilization. J Health Soc Behav. 1978;19:384–396. 16. Roblin DW, Goodrich GK, Davis TL, et al. Management of neck or back pain in ambulatory care: did visit mode or the COVID-19 pandemic affect provider practice or patient adherence? Med Care. 2023;61:S30–S38. 17. Gander J, Goodrich G, McDonald B, et al. Virtual care and urinary tract infection management: comparing ancillary service orders and patient order fulfillments between