key: cord-0889061-nvhwizpa
authors: Cvietusa, Peter J.; Goodrich, Glenn K.; Steiner, John F.; Shoup, Jo Ann; King, Diane K.; Ritzwoller, Debra P.; Shetterly, Susan M.; Bender, Bruce G.
title: Transition to virtual asthma care during the COVID pandemic:An observational study
date: 2022-03-06
journal: J Allergy Clin Immunol Pract
DOI: 10.1016/j.jaip.2022.02.027
sha: 37229fd6d8114d1ae89037bd72a7993f33913fd9
doc_id: 889061
cord_uid: nvhwizpa

Background The COVID-19 pandemic increased reliance on virtual care for patients with persistent asthma. Objective This retrospective cohort study assessed changes from in-person to virtual care during the pandemic in patients with persistent asthma compared the same time period prior to the pandemic. Methods Kaiser Permanente Colorado members aged 18-99 with persistent asthma, were evaluated during two time periods (March to October 2019 and March to October 2020). Comparison of asthma exacerbations (hospitalizations, emergency department visits, courses of oral prednisone) and asthma medication metrics were evaluated between the two time periods and by type of care received during the pandemic (no care, virtual care only, in-person care only or a mix of virtual and in-person care). Population characteristics by type of care received during the pandemic were also evaluated. Results Among 7,805 adults with persistent asthma, those who used more virtual care or sought no care during the pandemic were younger, had fewer co-morbidities, mental health diagnoses, or financial barriers. Exacerbations decreased (0.264 to 0.214, p <.001) as did courses of prednisone (0.213 to 0.169), asthma medication adherence (0.53 to 0.54, p <.001) and the Asthma Medication Ratio, a quality-of-care metric, (0.755 to 0.762, p <.001) increased slightly. Patients receiving a mix of in-person and virtual care had the highest rate of exacerbations (0.83) and lower AMR (0.74) despite having the highest adherence. Conclusions Despite an increase in virtual care, asthma exacerbations decreased except among individuals who received both in-person and virtual care, likely because they had more severe disease.

Given the rapid and unprecedented transition to virtual care, several studies highlighted the 147 difficulties associated with this historic transition.(5-7) The impact of virtual care on the management of 148 chronic diseases, including asthma, has been of particular concern because these conditions require 149 ongoing access to primary and specialty care to maintain quality and prevent adverse events such as 150

hospitalizations. Preliminary studies that assessed the impact of this transition on chronic disease 151 management showed variable outcomes.(8-11) The care of asthma patients during the pandemic was a 152 particular concern as the Center for Disease Control and Prevention, early in the pandemic, suggested that 153 patients with moderate to severe asthma could potentially be at higher risk for complications of COVID-154

In this study, we describe the transition to virtual care that took place in a large integrated health 156 care system with a population of nearly 8,000 adult patients with asthma, during the first 8 months of the 157 pandemic (March 2020 to October 2020). We examined the extent to which patients received care 158 through virtual channels and the effect of this transition on asthma outcomes, compared with the same 159 time period in 2019. 

The primary data source for this analysis was patient-level information from the KPCO electronic 179 health records (EHR), augmented by administrative and claims databases, that included procedure codes, 180 diagnosis codes and census-based measures of socioeconomic status, pharmacy dispensing data, as well 181 as all internal hospital and ambulatory encounters and claims for services outside KPCO. 182

The cohort for this study included adults 18 years and older who were continuously enrolled in 184 KPCO during the period before COVID (from March 2019 through October 2019), and during COVID 185 (from March 2020 through October 2020). Matched time periods were used to account for the usual 186 seasonality of asthma exacerbations(13). We included only patients who had persistent asthma for both 187 study periods, based on clinician diagnosis. Patients with diagnoses of chronic obstructive pulmonary 188 disease were excluded. We also required that the clinical diagnosis of persistent asthma, captured in both 189 time periods, be confirmed by at least one of the following: a visit (in-person or virtual) with an asthma 190 diagnosis, a fill of controller medication (e.g., inhaled corticosteroids, leukotriene receptor antagonist), a 191 fill of reliever medication (e.g., short acting beta-agonist), or an asthma exacerbation (prednisone fill, 192 emergency department (ED), hospital, or urgent care visit with an asthma diagnosis). 193

The primary outcome was the difference in asthma exacerbations in the period before COVID For health care outcomes before and during the COVID-19 pandemic, we used Poisson regression 224 models adjusting for over-dispersion and repeated measures to compare total prednisone exacerbations 225 and proportion of days covered. For binary outcomes (any hospitalization, any ED visit, and any urgent 226 care), we used log-binomial models with repeated measure adjustment.(17) Because AMR is a ratio of the 227 count of controller medications relative to the count of controller plus reliever medications, we analyzed 228 AMR using a binary logit model with a fractional dependent variable. All outcomes were adjusted for 229 person years except AMR which was time independent. 230

For outcomes across utilization groups during COVID, we used the same analytical methods 231 except for hospitalizations and the sum of ED and urgent care visits where we used Poisson regression 232 models accounting for over-dispersion. We did not account for multiple time points as we were using an 233 analysis of covariance framework, but calculated unadjusted models as well as models adjusting for age, 234 J o u r n a l P r e -p r o o f gender, race, and the baseline value of the outcome of interest. All rates were adjusted for person years 235 except AMR. 236

For patients with exacerbations in both time periods, we compared utilization across periods 237

using Poisson models and non-linear mixed models to derive confidence intervals for the difference 238 between the two time periods. To calculate rate ratios, the differences calculated from the model were 239

exponentiated. All utilization rates were adjusted for person years. vs 38% for the group as a whole). The portion of patients receiving virtual care was roughly equal across 262 racial and ethnic groups though Black patients received a relatively higher proportion of care virtually 263 (29.7% vs 25% for the group as a whole) and were less likely to be in the no care group (30.9% vs 38%). 264

Those requiring medical financial assistance were less likely to be in the no care group (31.3% vs 38%) 265

and more likely to have sought in-person care (27% vs 23%) or required a mix of virtual and in-person 266 care (22% vs 14%). Those with Medicaid did not seek care any differently than those in the larger 267 persistent asthma population (p=0.09). Those with a prior mental/behavioral health diagnosis (24.5% vs 268 23%) or those with a with a higher number of co-morbid conditions more often sought in-person care. 269

Courses of prednisone and asthma exacerbations requiring urgent care, ED care, or 270 hospitalization were all more common in those receiving a mix of in-person and virtual care (Table 2) . 271

Those receiving no care during COVID had the highest AMR (0.78 vs 0.76 for the group as a whole) 272 while those receiving a mix of care, the group with the highest rate of exacerbations during COVID, had a 273 lower AMR (0.74 vs 0.76 overall), despite having the highest PDC (0.57 vs 0.54) ( Table 2 ). We adjusted 274 for age, gender, race-ethnicity, and the baseline value of the outcome did not affect these findings (Table  275 3). for asthma exacerbations in the during COVID period were not statistically significant. There was a 295 statistically significant increase, comparing the period before COVID with the period after COVID, for 296 the whole population, for both PDC (0.53 to 0.54, p<0.001) and for AMR (0.75 to 0.76, p=.019). 297

However, these changes do not appear to be clinically meaningful. 298

In comparing patients who had one or more exacerbations in 2019 with those who had 1 or more 299 exacerbations during 2020, the mean number of total visits was equal. However, the portion of total visits 300 that were virtual for this population increased from 13% of all visits before COVID to 58% of all visits 301 during COVID (Table 5) . 302 303

In this retrospective cohort study of patients with persistent asthma, during the transition of care 305 delivery from before COVID to during COVID, we found no evidence for any adverse impact on the 306 important clinical outcomes of hospitalizations, ED visits and courses of prednisone for asthma flares or 307 on asthma medication adherence or AMR. While there was an expected and widely documented decrease 308 in volumes of care very early on in the pandemic, the volume of care recovered quickly and transitioned 309 from in-person care to virtual care. To a large extent, virtual care was adopted across age, race, income, 310 and insurance-type groups. At the same time, the use of virtual care tended to be lower among older 311 patients, those with lower household income and in those with a mental health diagnosis or a higher suggesting an easier transition for health conditions, that were not so dependent on in-person testing and 319 examinations. However, other studies showed that those in a lower social-economic status had more 320 difficulty making connections for virtual care.(14, 18) Our finding of a lower use of virtual care among 321 more economically vulnerable populations is consistent with this finding. Similar to our study findings, a 322 study conducted in the Department of Veteran's Affairs healthcare system documented an increase in 323 virtual care among those who were younger (age less than 45) but in contrast, found that those with lower 324 income, higher disability, and more chronic conditions were more likely to receive virtual care during the 325 pandemic.(30) Findings from a in an urban academic medical center were similar to our own study, 326

showing that virtual visits were less likely in men and in elderly patients but, in contrast to our study, 327 found fewer virtual visits among those with Medicaid sponsored insurance.(31) 328

While several studies looked at the rapid implementation of virtual care for chronic disease and While we did find that adherence improved overall for our patient population during the COVID 352 pandemic, though this statistically significant improvement was not likely clinically meaningful, since 353 changes in adherence need to be quite substantial to affect asthma outcomes.(37) The "no care" group had 354 the highest AMR, suggesting that their asthma remained more stable or was more mild. It is also possible 355 that they managed their exacerbations at home, rather than coming for an in-person visit, though these 356 exacerbations likely would have been mild. All asthma patients at KPCO receive automated reminders if 357 they are due to refill their controller medications, which has been shown to improve adherence.(37) 358

Those receiving virtual care had the lowest AMR, possibly suggesting less well controlled but perhaps 359 milder asthma that, with a little reassurance via virtual care, they were able to manage at home. Those 360 getting a mix of care had the highest rate of exacerbations and a lower AMR, despite having a higher 361 PDC. Driving down their AMR, despite the higher PDC, was a higher use of albuterol. Thus, this group 362 utilizing more care and more albuterol, likely had more severe asthma. This is consistent with previous 363 studies that have found the HEDIS (Healthcare Effectiveness Data and Information Set) AMR, and in 364 particular, albuterol use, to be a stronger correlate for adverse asthma outcomes.(38-40) 365

Our study has several limitations. It was conducted in an integrated healthcare system whose 366 clinicians shared an EHR and where modes of virtual care had already been developed, although these 367 care channels were infrequently used (15% of patient encounters, with < 1% done through video visits), 368

prior to the pandemic. Thus, it may not be generalizable to other healthcare delivery systems or 369 populations. We chose to use a clinician diagnosis of persistent asthma but no other criteria such as AMR, 370 frequent use of beta-agonists, frequency of asthma flares, or other criteria to define this population. An 371 asthma exacerbation was used as an inclusion criterion, which may have blunted the differences in 372 outcomes between populations. Nevertheless, we did find a difference in important outcomes both for the 373 population as a whole between the two time periods and within the subpopulations (divided by mode of 374 care) during COVID. Finally, our research was restricted to an analysis of EHR data, which does not 375 include perceptions of the patient regarding perceived or actual challenges to accessing care, their 376

perceptions of the quality of care during COVID, and motivational factors that may have influenced their 377 decisions regarding whether or how to access care. 378

Given the persistence of the COVID-19 pandemic, it is hard to predict the expanded use of virtual 379 care will be maintained. For patients with asthma, some amount of in-person care will be needed to 380 objectively assess the patient by exam and spirometry testing. The amount and mix of virtual and in-381 person care will depend on both patient and clinician comfort with virtual visits but also patient and 382 clinician resources to conduct virtual visits efficiently and effectively. Prior to the pandemic, one study, 383 looking at children with asthma living in rural locations, managed by a tertiary care allergy clinic, 384 demonstrated non-inferior outcomes with those managed virtually versus in person.(41) A 2015 Cochrane 385 systematic review examining the impact of telehealth involving remote monitoring or video conferencing 386 compared with in-person or telephone visits for chronic conditions, including diabetes and congestive 387 heart failure, found similar health outcomes for patients with these conditions.(42) Future research will 388 need to continue to examine outcomes and ensure maintenance of care quality. In addition, there will be a 389 need for research that surveys or interviews patients to facilitate the development of interventions to 390 address system-level and patient-level barriers to access in healthcare systems. Compensation for virtual 391 visits will also likely influence the outcome as will the potential cost savings, in both personal and office 392 space, derived from doing fewer in-person visits. 393

In summary, our study suggests that access to services for our large population of patients with 394 persistent asthma was maintained during the COVID-19 pandemic, despite the change from in-person to 395 virtual care. Though likely due to other mitigating factors imposed by the pandemic itself, this sudden 396 transition to virtual care did not lead to worse outcomes. Our study and others have highlighted a number 397 of issues around the on-going delivery of virtual care that will require thoughtful research in the future, 398 particularly when the pandemic has subsided. 

The impact of the COVID-19 407 pandemic on outpatient visits: practices are adapting to the new normal

Trump administration finalizes permanent expansion of Medicare telehealth services and 412 improved payment for time doctors spend with patients

Trends in 414 outpatient care delivery and telemedicine during the COVID-19 pandemic in the US

Who is (and is not) receiving 417 telemedicine care during the COVID-19 pandemic

Telemedicine and healthcare 419 disparities: a cohort study in a large healthcare system in New York City during COVID-19

Telehealth use among older adults during COVID-422 19: associations with sociodemographic and health characteristics, technology device ownership, and 423 technology learning

Impact of rapid 425 transition to telemedicine-based delivery on allergy/immunology care during COVID-19

LoVE in a time of 428 CoVID: Clinician and patient experience using telemedicine for chronic epilepsy management

Successful distancing: telemedicine in gastroenterology and hepatology 431 during the COVID-19 pandemic

Preliminary estimates of the prevalence of selected underlying health conditions among patients 435 with coronavirus disease 2019 -United States

Asthma exacerbations· 1: epidemiology

Standardizing terminology and 439 definitions of medication adherence and persistence in research employing electronic databases. Med 440 Care

The controller-to-442 total asthma medication ratio is associated with patient-centered as well as utilization outcomes

Generalized Linear Models

Patient use and clinical 450 practice patterns of remote cardiology clinic visits in the era of COVID-19

Rapid 453 implementation of virtual neurology in response to the COVID-19 pandemic

COVID-19: pivoting from in-person to virtual orthopedic 456 surgical evaluation

Comparison of patient 458 satisfaction between virtual visits during the COVID-19 pandemic and in-person visits pre-pandemic

Evaluation of telephone and virtual 461 visits for routine pediatric diabetes care during the COVID-19 pandemic

What do 464 asthmatic patients think about telemedicine visits?

Virtual group visits: hope for improving chronic disease management 466 in primary care during and after the COVID-19 pandemic

Reflections on virtual care for chronic conditions 468 during the COVID-19 pandemic

Rapid transition to telehealth 470 and the digital divide: implications for primary care access and equity in a post-COVID era

Rapid conversion of an 473 outpatient psychiatric clinic to a 100% virtual telepsychiatry clinic in response to COVID-19

Abbreviations: M=mean; SE=standard error; N=number; ED=emergency department; UC=urgent care; 614 PDC=proportion of days covered; AMR=Asthma medication ratio 615 * Poisson model accounting for over-dispersion used to obtain p-value

† Poisson model accounting for over-dispersion used to obtain p-value Mean is total number of days 617 covered by total number of days enrollment in study period

Logit binomial model with fractional dependent variable (controller divided by sum of controller and 619 reliever) model used to determine

* Poisson model accounting for over-dispersion used to obtain p-value

† Poisson model accounting for over-dispersion used to obtain p-value Mean is total number of days 622 covered by total number of days enrollment in study period

Logit binomial model with fractional dependent variable (controller divided by sum of controller and 624 reliever) model used to determine

Abbreviations: N=number; SE=standard error; ED=emergency department; PDC=Portion of Days 646 Covered; AMR=Asthma Medication Ratio 647 * Poisson model accounting for repeated measures unadjusted for covariates, ratio with confidence interval 648 is incidence density ratio

Generalized linear model using log-binomial model accounting for repeated measures unadjusted for 650 covariates. Reported ratio and confidence interval is relative risk 651 ‡ Rates are adjusted for person years except for AMR which is not dependent on time

Odds ratio for logit model with fractional dependent variable accounting for repeated measures 653

Abbreviations: M=mean; SE=standard error; N=number; ED=emergency department; UC=urgent care; 671 PDC=proportion of days covered

Each outcome is adjusted for age, gender, race, and baseline value of outcome

* Poisson model accounting for over-dispersion used to obtain p-value

† Poisson model accounting for over-dispersion used to obtain p-value Mean is total number of days 675 covered by total number of days enrollment in study period

Logit binomial model with fractional dependent variable (controller divided by sum of controller and 677 reliever) model used to determine

Comparison of mean number of visits in patients with one or more exacerbations

Abbreviations: N=number; M=mean; SE=standard error; IP=in-person; VV=virtual visit; CI=confidence 692 interval 693 *Means adjusted for person years, and standard errors derived from poisson regression 694 ** Confidence intervals presented derived from poisson regression and non-linear mixed models 695

Confidence intervals presented derived from poisson regression

Figure 1: Waterfall diagram for venues of care received by patients with persistent asthma during COVID