Abstract
Background
During the COVID‐19 pandemic, patients with chronic illnesses avoided regular medical care, raising concerns about long‐term complications. Our objective was to identify a population of older patients with chronic conditions who may be at risk from delayed or missed care (DMC) and follow their non‐COVID outcomes during the pandemic.
Methods
We used a retrospective matched cohort design using Medicare claims and electronic health records at a large health system with community and academic clinics. Participants included 14,406 patients over 65 years old with two or more chronic conditions who had 1 year of baseline data and up to 9 months of postpandemic follow‐up from March 1, 2019 to December 31, 2020; and 14,406 matched comparison patients from 1 year prior. Risk from DMC was defined by 13 indicators, including chronic conditions, frailty, disability affecting the use of telehealth, recent unplanned acute care, prior missed outpatient care, and social determinants of health. Outcomes included mortality, inpatient events, Medicare payments, and primary care and specialty care visits (in‐person and telehealth).
Results
A total of 25% of patients had four or more indicators for risk from DMC. Per 1000 patients annually, those with four or more indicators had increased mortality of 19 patients (95% confidence interval, 4 to 32) and decreased utilization, including unplanned events (−496 events, −611 to −381) and primary care visits (−1578 visits, −1793 to −1401).
Conclusions
Older patients who had four or more indicators for risk from DMC had higher mortality and steep declines in inpatient and outpatient utilization during the pandemic.
Keywords: COVID‐19 pandemic, delayed care, Medicare, multiple chronic conditions
Key points
Of a population of 14,406 Medicare patients, 25% had four or more indicators putting them at risk from DMC.
Patients with four or more indicators of risk from DMC had higher mortality and steep declines in inpatient and outpatient utilization during the pandemic.
Why does this paper matter?
Older patients who are at risk from DMC due to the pandemic may benefit from outreach and care coordination to ensure proactive management of their chronic conditions.
INTRODUCTION
Delaying or missing regular care for chronic conditions could exacerbate long‐term complications, 1 although we have limited tools to identify those patients who are most at risk. This concern became particularly acute during the COVID‐19 pandemic, as some patients avoided medical attention for their conditions for fear of catching the virus or because they were sheltering at home. 2 Health systems also rapidly transformed care delivery by delaying elective care and shifting outpatient care to telehealth, 3 although substantial barriers to telehealth exist for older adults. 4 By June 2020, 41% of U.S. adults had delayed or avoided medical care, 5 while admissions and emergency department (ED) visits had declined precipitously. 6 Fewer patients were presenting with acute cardiovascular conditions, 7 and cancer diagnosis and treatment had been delayed. 8 Most concerning, non‐COVID‐19 mortality increased during the pandemic for those with specific chronic conditions, including heart disease, Alzheimer disease–dementia, and diabetes. 9 Since we may need to cope with the pandemic for an extended period 10 as well as prepare for future pandemics, there is an urgent need to support the clinical identification of patients with chronic conditions to determine how they fared during the pandemic and to support identification of these patients at risk for poor outcomes such as increased mortality.
Older patients with multiple chronic conditions are likely at particular risk as they are major users of health care, often seeing multiple physicians, frequently visiting the ED, and having multiple admissions each year. 11 , 12 , 13 Missed outpatient appointments have been associated with increased admissions, 14 all‐cause mortality (particularly for those with mental health conditions), 1 and suboptimal glycemic control for diabetes. 15 Since missed care can lead to poor outcomes, health systems often identify patients with chronic conditions and provide outreach or case management to ensure the appropriate use of outpatient care. 16 This identification may use a tool or predictive model to find patients at high‐risk of hospital admissions or cost, 17 but these tools may miss others who would be at risk from delayed or missed care (DMC) during a pandemic such as those with mental health conditions.
Our goal was to identify a population of older patients with chronic conditions who may be at risk from DMC and follow their non‐COVID outcomes during the pandemic. We focused on patients either at risk for long‐term complications due to clinical factors, or who might have had difficulty in accessing the resources they need to manage their conditions due to socioeconomic vulnerability, for example, Medicaid patients. Medicaid is a means‐tested federal and state healthcare program that provides coverage for low‐income adults, pregnant women, and children. To achieve our goal, we first developed indicators to identify patients who were at high‐risk for poor health outcomes from DMC and validated that those indicators predicted higher rates of visits, hospital events, and payments in a baseline timeframe. Second, we examined the impact of the pandemic on utilization and mortality outcomes in the follow‐up timeframe for these patients according to how many indicators they had for high‐risk from DMC.
METHODS
Study design and setting
We used a 1:1 matched cohort study design with measures of the outcomes throughout baseline and follow‐up. Specifically, we used baseline characteristics to match a “pandemic cohort” of patients to a “comparison cohort” representing the year prior to the pandemic, where the start of the pandemic was defined as March 1, 2020 (Figure 1). The pandemic cohort included patients who received primary care during the 12 months prior to the pandemic, with patient baseline characteristics measured from March 1, 2019 to February 28, 2020. The comparison cohort included patients who received primary care during the 12 month period starting 2 years prior to the pandemic, with baseline characteristics measured from March 1, 2018 to February 28, 2019.
FIGURE 1.
Pandemic and comparison cohort observation periods
We examined the impact of the pandemic on monthly utilization and mortality outcomes in the follow‐up year by comparing the two groups at different levels of DMC. The “follow‐up period” is 9 months or until death or censoring due to lack of data for both groups (April 1, 2020 to December 31, 2020 for the pandemic cohort and April 1, 2019 to December 31, 2019 for the comparison cohort). The month of March was excluded as it was a transition into the pandemic. We used electronic health records (EHR) linked to Medicare claims data from UW Health, a health system and Medicare Accountable Care Organization (ACO) with 30 community‐based and academic primary care clinics and 279 primary care providers (PCPs) across the state of Wisconsin. 18 , 19 , 20 , 21 , 22 This project was deemed exempt from Institutional Review Board oversight at University of Wisconsin‐Madison.
Pandemic and matched comparison patients
We included only patients aged 65 years and older who met the following criteria: (a) uninterrupted EHR and claims data available for at least 1 year prior to follow‐up; (b) assigned to a UW Health PCP; (c) assigned to the ACO during baseline and follow‐up periods; (d) at least 1 month of continuous EHR and claims data during follow‐up, and (3) at least two chronic conditions. 23 We excluded patients not enrolled in Medicare Part A or Part B or enrolled in Medicare Part C throughout the baseline and follow‐up periods. We excluded 575 patients diagnosed with COVID‐19 during follow‐up; those patients were retained in a sensitivity analysis.
We matched each patient in the pandemic cohort to the closest eligible patient in the comparison cohort using exact matching on 18 baseline variables, including sociodemographics (age, gender, race, rural–urban, disability, Medicaid), chronic conditions (diabetes, end‐stage renal disease [ESRD]), utilization (Medicare payments, hospice), and enrollment in case management or home‐based primary care programs. If multiple comparison patients matched the pandemic patients, we picked one comparison patient with the closest risk of hospitalization or death in the next 6 months. 19
The final sample included 14,406 patients for the pandemic cohort and 14,406 for the comparison cohort. Across both the comparison and pandemic cohorts, 2.2% of patients (N = 322) were identified as the exact match for themselves in the other cohort. We also ran a sensitivity analysis on our mortality outcome to remove the 2.2% of patients who used their own prior year as their comparison episode to assess for the possibility of survivorship bias and our conclusions did not change. However, all results for the mortality outcome are presented after removing these 2.2% of patients. We also note that 83% of the patients in the comparison cohort were also included in the pandemic cohort and followed throughout the pandemic. Because the pandemic and comparison cohorts were treated as independent, and not followed as a single cohort, patients in the pandemic cohort who were also in the comparison cohort, but not included as their own control, should not have an issue of survivorship bias.
Indicators of risk from DMC
We identified 13 indicators of possible risk from DMC that represented high‐risk chronic conditions, frailty, disability affecting the use of telehealth, recent unplanned acute care, prior missed outpatient care, and social determinants of health. Indicators were based on (1) a conceptual model of disability, frailty, and comorbidity, 24 (2) a conceptual model for episodes of acute unscheduled care, 25 (3) Centers for Disease Control and Prevention recommendations for subgroups of patients who required extra attention during the COVID‐19 pandemic, 26 and (4) evaluation of content validity by an expert review panel of five primary care physicians. 27 We operationalized these concepts using data accessible through EHR and/or claims, allowing for rapid and scalable deployment of the tool in a health system or ACO.
Specific DMC indicators included significant polypharmacy (defined as five or more unique prescribed medications ordered or billed); 28 any diagnosis of cardiovascular disease 29 or stroke 30 ; uncontrolled hypertension (systolic blood pressure [BP] >140 or diastolic BP >90 and a diagnosis of hypertension 31 , 32 ); uncontrolled diabetes (most recent HbA1c ≥9 and a diagnosis of diabetes 33 ); end‐stage liver disease (ESLD), 34 ESRD 35 and stage 4/5 chronic kidney disease (CKD), 35 and mental health condition (defined as bipolar disorder, schizophrenia, psychotic disorders; behavior and personality disorders; substance abuse; or mental health‐related hospitalization or ED visit 29 ). Frailty was operationalized using the list of frailty indicators from the Johns Hopkins Ambulatory Care Groups (e.g., incontinence, mobility, dementia–cognitive impairment, falls, malnutrition); a patient was considered frail if they had three or more frailty conditions. 36 We created an indicator for any hearing impairment‐related condition, 29 representing a prevalent disability in older adults that might affect the use of telehealth. Recent unplanned acute care was represented by unplanned hospitalizations or ED visits using established definitions. 29 We also created an indicator for possible poverty or homelessness using lack of–inadequate housing or inadequate material resources diagnosis codes or if the patient's social history documentation included the word “shelter” or “homeless.” 37 Lastly, we created an indicator for two or more unplanned missed appointments to any specialty. 38 Detailed information on these indicators is available at HIPxChange (https://www.hipxchange.org/DelayedMissedCare).
Outcome measures
Our outcome measures described the extent of hospital admissions, observations stays, ED visits, Medicare payments, and mortality from Medicare claims during the baseline and follow‐up period. ED visits that resulted in hospitalization were not counted as an ED visit but were counted as part of the hospitalization. Unplanned hospital events were defined as admissions, observation stays, or ED visits. We used total Medicare payments excluding payments for planned hospitalizations 39 , 40 and pharmacy payments. We also examined outpatient visits for face‐to‐face and telehealth (video or telephone) primary and specialty care visits. To construct repeated measures of our outcomes, we created a dataset with one observation per patient per month. The first month was 12 months prior to March and continued for a minimum of 1 month and a maximum of 9 months after April, unless the patient died or was otherwise censored due to lack of data.
Baseline variables
Sociodemographics included age (continuous), gender, race, Medicaid insurance (ever), Medicare disability entitlement, rural–urban residence, 41 and the Hierarchical Condition Categories score. 42 Chronic conditions were measured from the diagnosis codes associated with medical encounters in the EHR data and included 28 medical conditions defined by Elixhauser et al. using ICD‐9‐CM diagnosis codes along with a count of the conditions and an indicator variable for three or more conditions. 43
Analysis
Descriptive analyses compared means and proportions for baseline characteristics between pandemic and comparison patients, as well as average baseline and annualized follow‐up utilizations from Medicare claims. Pearson correlation was used to assess the association between the count of DMC indicators and baseline utilization.
Models included terms for the baseline time trend (except for mortality models), change in level between baseline and follow‐up times, and follow‐up time trends in our monthly events for both pandemic and comparison patients. To account for further possible changes in utilization trends during the pandemic, models also included follow‐up time period indicators for both cohorts (except for mortality models). In particular, the indicators are for the first 3 months and the first 6 months in the follow‐up period. For mortality, we conducted longitudinal binomial regression modeling of the risk‐adjusted difference in monthly death rate trajectories between the pandemic and comparison patients using patient‐month data in the follow‐up time frame. We conducted longitudinal regression modeling of the risk‐adjusted difference in monthly visit count, event count, and payment trajectories between the pandemic and comparison patients using patient‐month data for both baseline and follow‐up time frames with poisson (visit counts), zero‐inflated poisson (event counts), and zero‐inflated gamma (payments) regression modeling. We accounted for clustering at the patient‐level due to 80% of patients overlapping between the pandemic and comparison year. Models were stratified by DMC categories.
To improve interpretation, results were transformed into the annualized predicted difference in mortality rate, the number of event counts or visit counts, and Medicare payments for 1000 patients during the pandemic follow‐up time compared to the comparison follow‐up time. Bootstrapped confidence intervals (CI) were calculated using 200 replications. Analyses were carried out using SAS software (SAS Institute, Inc., Cary, North Carolina).
RESULTS
Baseline characteristics
Our comparison and pandemic cohorts have essentially identical characteristics in their baseline timeframes; the average age across the comparison and pandemic study cohort was 75 and patients were more likely to be female, (60%), white (96%), and urban (69%) (Table 1). The cohorts also had similar distributions of unplanned events, ED visits, hospitalizations, and Medicare payments. As expected, visits in the baseline were almost entirely face‐to‐face; telehealth visits were negligible. In both cohorts, patients had an average of six chronic conditions with the most prevalent chronic conditions, including hypertension, chronic obstructive pulmonary disease (COPD)–Asthma, CKD, and anxiety.
TABLE 1.
Sociodemographics, baseline utilization, and chronic conditions for age 65+ primary care patients with 2+ chronic conditionsa
| Baseline comparison cohort (pre‐COVID) N = 14,406 | Baseline pandemic cohort (pre‐COVID) N = 14,406 | |
|---|---|---|
| Baseline characteristic | 2018 | 2019 |
| Sociodemographics | ||
| Age, mean (SD) | 75.3 (6.6) | 75.4 (6.5) |
| Female | 60.7 | 60.7 |
| Race | ||
| American Indian | 0.2 | 0.2 |
| Asian | 1.2 | 1.1 |
| Black | 0.9 | 1.0 |
| Other–unknown | 0.9 | 0.9 |
| White | 96.7 | 96.7 |
| Medicaid Insurance Ever | 5.6 | 5.6 |
| Disability entitlement | 4.2 | 4.2 |
| Rural–urban | ||
| Urban code | 69.3 | 69.3 |
| Suburban | 18.8 | 18.8 |
| Large town | 10.7 | 10.6 |
| Small town–isolated rural | 1.2 | 1.4 |
| HCC score, mean (SD) | 1.2 (1.1) | 1.2 (1.1) |
| Events and payments | ||
| Unplanned events PMPM × 1000, mean (SD) | 46.3 (102) | 51.1 (111) |
| Unplanned event‐days PMPM × 1000, mean (SD) | 95.1 (273) | 106 (297) |
| ED visits PMPM × 1000, mean (SD) | 30.0 (72.3) | 33.9 (78.5) |
| Unplanned hospitalizations PMPM × 1000, mean (SD) | 12.0 (41.5) | 12.4 (42.1) |
| Days in hospital PMPM × 1000, mean (SD) | 60.8 (243) | 67.5 (260) |
| Observation stays PMPM × 1000, mean (SD) | 4.3 (20.4) | 4.7 (21.7) |
| Medicare payment amount ($), mean (SD) | 609 (1278) | 632 (1316) |
| Primary and specialty care visits | ||
| Primary care total visits PMPM × 1000, mean (SD) | 345 (270) | 353 (272) |
| Primary care face‐to‐face visits PMPM × 1000, mean (SD) | 341 (267) | 348 (267) |
| Primary care telehealth visits PMPM × 1000, mean (SD) | 0.01 (0.7) | 0.00 (0.00) |
| Specialty care total visits PMPM × 1000, mean (SD) | 220 (250) | 213 (248) |
| Specialty care face‐to‐face visits PMPM × 1000, mean (SD) | 219 (250) | 213 (248) |
| Specialty care telehealth visits PMPM × 1000 mean (SD) | 0.01 (1.0) | 0.00 (0.00) |
| Chronic conditions | ||
| Mean condition count | 6.0 (3.1) | 5.8 (3.0) |
| COPD–asthma | 25.2 | 25.6 |
| Chronic kidney disease | 25.0 | 26.2 |
| Anxiety | 29.6 | 31.3 |
| ESRD | 0.5 | 0.5 |
| Anemia | 15.4 | 15.8 |
| Rheumatoid arthritis–vasculitis | 8.4 | 8.9 |
| Chronic blood loss anemia | 1.8 | 1.8 |
| Coagulopathy | 4.7 | 4.9 |
| Depression | 18.5 | 18.7 |
| Diabetes with chronic complication | 13.9 | 14.3 |
| Diabetes without chronic complication | 9.0 | 9.7 |
| Hypertension | 68.3 | 70.4 |
| Hypothyroidism | 21.1 | 20.9 |
| Liver disease | 2.7 | 2.9 |
| Lymphoma | 1.7 | 1.6 |
| Fluid–electrolyte disorders | 17.0 | 17.7 |
| Metastatic cancer | 2.3 | 2.4 |
| Other neurological disorders | 13.8 | 14.3 |
| Obesity | 16.1 | 16.7 |
| Paralysis | 1.9 | 2.2 |
| Pulmonary circulation disease | 4.9 | 4.9 |
| Psychosis | 9.7 | 10.9 |
| Peripheral vascular disease | 14.9 | 15.1 |
| Renal failure | 15.4 | 15.6 |
| Solid tumor w/o metastasis | 10.5 | 11.1 |
| Valvular disease | 10.1 | 10.3 |
| Weight loss | 5.4 | 5.6 |
Abbreviations: COPD, chronic obstructive pulmonary disease; ED, emergency department; HCC, hierarchical condition categories; PMPM, per member per month.
Values represent percents unless otherwise indicated; per member per month (PMPM) × 1000 is used for consistency with other tables that have varied follow‐up times.
DMC indicators
Across our DMC indicators, the comparison and pandemic cohorts were also very similar, and showed high rates of polypharmacy (79%), risk for cardiovascular disease or stroke (77% and 78%), frailty (13%), and hearing loss (20%) (Table 2). When we compared patients with three DMC indicators and those with four or more DMC indicators (data not shown), the patients with four or more indicators were more likely to experience polypharmacy (99% compared to 95%), have uncontrolled hypertension (34% compared to 22%), be at risk for cardiovascular disease or stroke (98% compared to 94%), have a mental health condition (25% compared to 9%), have three or more frailty conditions (42% compared to 9%), and suffer from hearing loss (40% compared to 25%). Approximately 25% of patients had four or more DMC indicators. As expected, the count of DMC indicators was highly correlated with baseline utilization (Table S1).
TABLE 2.
Baseline delayed or missed care (DMC) indicators for age 65+ primary care patients with 2+ chronic conditions
| Indicator | Baseline comparison cohort (pre‐COVID) | Baseline pandemic cohort (pre‐COVID) |
|---|---|---|
| N = 14,406 | N = 14,406 | |
| 2018 | 2019 | |
| Polypharmacy (5+) | 78.8 | 79.2 |
| Uncontrolled hypertension | 15.9 | 16.9 |
| Uncontrolled diabetes | 1.5 | 1.5 |
| Cardiovascular disease or stroke | 76.3 | 77.8 |
| Mental health condition | 9.7 | 9.7 |
| 3+ frailty conditions | 12.1 | 13.0 |
| 2+ no show appointments | 3.7 | 3.9 |
| Hearing loss | 19.9 | 19.7 |
| ESLD | 0.5 | 0.6 |
| ESRD or stage 4/5 CKD | 3.6 | 3.6 |
| Unplanned hospitalization | 10.4 | 11.0 |
| ED visit not resulting in admission | 23.0 | 25.4 |
| Possible poverty or homelessness | 0.7 | 0.6 |
| Any indication DMC | 94.3 | 94.5 |
| 0 or 1 DMC indicator | 23.1 | 22.4 |
| 2 DMC indicators | 30.7 | 29.2 |
| 3 DMC indicators | 23.7 | 23.6 |
| 4+ DMC indicators | 22.5 | 24.8 |
Note: Values represent percents.
Abbreviations: CKD, chronic kidney disease; ESLD, end‐stage liver disease; ESRD, end‐stage renal disease.
Pandemic mortality and healthcare utilization by DMC indicators
Mortality increased significantly (p = 0.025) during the pandemic for patients with four or more DMC indicators, while both unplanned events and Medicare payments declined for all patients (Table S2). As expected, the drop in utilization during the pandemic was evident across all utilization measures except for telehealth visits. Declines in utilization included ED visits, observation stays, and hospitalizations, as well total primary and specialty care visits and face‐to‐face primary and specialty care visits. In contrast, telehealth (including both video and telephone) visits increased with the higher rates of telehealth visits among patients with more DMC indicators. The steepest declines in both inpatient and outpatient utilization were among patients with four or more DMC indicators.
After modeling changes in mortality and utilization rates during the pandemic, these results did not change (Table 3; Figure S1). Overall, mortality increased by 13 patients per 1000 annually but this finding was due to an increase in mortality in patients with four or more DMC indicators of 19 patients per 1000 annually. The number of unplanned events and Medicare payments declined significantly overall; the most striking decline was seen in patients with four or more DMC indicators, with a decrease of almost 500 unplanned events per 1000 patients annually and a decrease in Medicare payments of $2.7 million per 1000 patients annually ($2732 Per Beneficiary Per Year). Total primary care visits and total specialty care visits also declined, with the greatest decline in patients with four or more DMC indicators representing decreases of over 1500 total primary care visits per 1000 patients annually and almost 900 total specialty care visits per 1000 patients annually. Results were essentially identical in sensitivity analysis that did not exclude 575 patients diagnosed with COVID‐19 during the pandemic follow‐up period.
TABLE 3.
Annualized difference in adjusted predicted outcomes per 1000 patients during the pandemic year compared to the prior year, overall and by level of risk from delayed or missed care (DMC)
| Overall (N = 28,812) | ||
|---|---|---|
| Prediction | 95% CI | |
| Annualized difference in mortality per 1000 patients | 12.5 | (8.2, 16.9) |
| Annualized difference in unplanned events per 1000 patients | −87 | (−114.5, −61.5) |
| Annualized difference in Medicare payments per 1000 patients | −315,446 | (−618,471, 25,109) |
| Annualized difference in total primary care visits per 1000 patients | −909 | (−970.9, −853.8) |
| Annualized difference in total specialty care visits per 1000 patients | −422 | (−467.7, −376.6) |
| 0 or 1 DMC indicator (N = 6559) | ||
|---|---|---|
| Prediction | 95% CI | |
| Annualized difference in mortality per 1000 patients | 0.1 | (−3.9, 3.7) |
| Annualized difference in unplanned events per 1000 patients | 18.7 | (−16.9, 52.3) |
| Annualized difference in Medicare payments per 1000 patients | 96,334 | (−255,367, 463,559) |
| Annualized difference in total primary care visits per 1000 patients | −460 | (−555.1, −358.8) |
| Annualized difference in total specialty care visits per 1000 patients | −136 | (−227.2, −46.4) |
| 2 DMC indicators (N = 8626) | ||
|---|---|---|
| Prediction | 95% CI | |
| Annualized difference in mortality per 1000 patients | −1.2 | (−7.6, 4.6) |
| Annualized difference in unplanned events per 1000 patients | −69.9 | (−122.2, −17.3) |
| Annualized difference in Medicare payments per 1000 patients | −339,992 | (−829,398, 275,075) |
| Annualized difference in total primary care visits per 1000 patients | −798 | (−945.3, −662.5) |
| Annualized difference in total specialty care visits per 1000 patients | −223 | (−322.4, −112.1) |
| 3 DMC indicators (N = 6810) | ||
|---|---|---|
| Prediction | 95% CI | |
| Annualized difference in mortality per 1000 patients | 2.8 | (−6.4, 10.7) |
| Annualized difference in unplanned events per 1000 patients | −69.3 | (−157.0, 1.7) |
| Annualized difference in Medicare payments per 1000 patients | −840,734 | (−1,814,662, 142,904) |
| Annualized difference in total primary care visits per 1000 patients | −1138 | (−1301.2, −974.9) |
| Annualized difference in total specialty care visits per 1000 patients | −588 | (−722.5, −457.9) |
| 4 or more DMC indicators (N = 6817) | ||
|---|---|---|
| Prediction | 95% CI | |
| Annualized difference in mortality per 1000 patients | 19.2 | (4.4, 32.2) |
| Annualized difference in unplanned events per 1000 patients | −495.5 | (−611.3, −380.5) |
| Annualized difference in Medicare payments per 1000 patients | −2732,498 | (−4,592,609, −1,318,849) |
| Annualized difference in total primary care visits per 1000 patients | −1578 | (−1792.5, −1401.1) |
| Annualized difference in total specialty care visits per 1000 patients | −889 | (−1036.1, −755.1) |
DISCUSSION
These results suggest that older patients with chronic conditions and four or more indicators for risk from DMC had higher risk of mortality during the pandemic and the steepest declines in inpatient and outpatient utilization when compared to those with three or fewer indicators for risk from DMC. As the results did not change when patients diagnosed with COVID‐19 were included, this increase in mortality was not due to deaths from COVID‐19.
Our results suggest that patients with four or more indicators of risk from DMC had increased mortality. This is consistent with a study that identified excess non‐COVID‐19 mortality from heart disease, Alzheimer disease–dementia, and diabetes during pandemic surges. 9 These three conditions are prevalent among older adults but it is unlikely that these three conditions are the only conditions associated with increased risk for non‐COVID‐19 mortality. 44 Vulnerable patients who have comorbidities, disability, and/or frailty have increased healthcare needs 24 and are at risk for complications if their conditions are not managed. The CDC (2020) had recommendations for multiple subgroups of patients who required extra attention during the COVID‐19 pandemic, including patients with disabilities, 45 patients with developmental and behavioral disorders, 46 those experiencing homelessness, 47 and patients with drug use and substance use disorder. 48 Increased adoption of telemedicine for chronic disease management offers another avenue for care coordination, but implementation of telemedicine may increase disparities in healthcare access for certain subgroups of patients, including older adults who may have a digital literacy barrier. 49
Managing patients at risk from DMC might involve identifying these subgroups of patients with complex needs and providing additional services such as outreach or case management to ensure the appropriate use of outpatient care. Because many case management programs involve a significant telephonic component, 50 case managers are well‐positioned to continue or expand their outreach activities during the pandemic to coordinate needed care for vulnerable patients. Older patients with multiple chronic conditions frequently see multiple physicians, 11 suggesting that, in addition to outreach, care coordination across these physicians might be a critical activity during the COVID‐19 pandemic as patients avoided seeing both primary and specialty care physicians. These patients also may have difficulty in accessing the resources they need to manage their conditions due to the COVID‐19 pandemic as, for example, patients with elevated health risks are overburdened by transportation barriers and may need extra help to access health care. 51
As with any study, our study has limitations. Unmeasured confounding is a limitation of all observational studies. However, given our extensive matching process and similarity of our matched populations, it is unlikely that any remaining small differences explain our findings. We only followed outcomes for 9 months after the start of the pandemic, although the major effects of the pandemic on utilization happened almost immediately. In addition, we were limited to evaluating the impact of the pandemic in a single large health system with both academic and community clinics. This health system did participate in Medicare ACO programs, indicating that they had a strong base of primary care patients. 52 Finally, our data did not include detailed information on patient‐level income or education.
In conclusion, we determined that older patients who had four or more indicators for risk from DMC had higher mortality during the pandemic, along with steep declines in inpatient and outpatient utilization. These patients may benefit from outreach and care coordination to ensure proactive management of their chronic conditions, particularly as future pandemic scenarios suggest that countries, communities, and individuals may need to cope in the longer‐term with the possibility of a continued threat from COVID‐19 and its variants. 10
CONFLICT OF INTEREST
The authors declare no conflicts of interest.
AUTHOR CONTRIBUTIONS
Study concept and design: Maureen Smith, Mary Vaughan Sarrazin, Menggang Yu. Acquisition, analysis, or interpretation of data: All authors. Drafting of the manuscript: All authors. Critical revision of the manuscript for important intellectual content: All authors. Final approval of the version to be published: All authors.
SPONSOR'S ROLE
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, PCORI, or other funders.
Supporting information
Table S1. Pearson correlation and p‐values for the relationship between the count of delayed or missed care (DMC) indicators and baseline utilization for age 65+ primary care patients with 2+ chronic conditions.
Table S2. Follow‐up utilization for age 65+ primary care patients with 2+ chronic conditions, for pandemic and comparison cohorts, by level of risk from delayed or missed care (DMC).
Figure S1. Annualized difference in adjusted predicted outcomes per 1000 patients during the pandemic year compared to the prior year, by level of risk from delayed or missed care (DMC).
Smith M, Vaughan Sarrazin M, Wang X, et al. Risk from delayed or missed care and non‐COVID‐19 outcomes for older patients with chronic conditions during the pandemic. J Am Geriatr Soc. 2022;70(5):1314‐1324. doi: 10.1111/jgs.17722
Funding informationThis project was supported by grant PCORI Grant # HSD‐1603‐35039. Additional support was provided by the University of Wisconsin School of Medicine and Public Health's Health Innovation Program (HIP), the Wisconsin Partnership Program, and the Community‐Academic Partnerships core of the University of Wisconsin Institute for Clinical and Translational Research (UW ICTR), grant 9U54 TR000021 from the National Center for Advancing Translational Sciences (previously grant 1 UL1 RR025011 from the National Center for Research Resources).
REFERENCES
- 1. McQueenie R, Ellis DA, McConnachie A, Wilson P, Williamson AE. Morbidity, mortality and missed appointments in healthcare: a national retrospective data linkage study. BMC Med. 2019;17(1):2. doi: 10.1186/s12916-018-1234-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Anderson KE, McGinty EE, Presskreischer R, Barry CL. Reports of forgone medical care among US adults during the initial phase of the COVID‐19 pandemic. JAMA Netw Open. 2021;4(1):e2034882. doi: 10.1001/jamanetworkopen.2020.34882 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Whaley CM, Pera MF, Cantor J, et al. Changes in health services use among commercially insured US populations during the COVID‐19 pandemic. JAMA Netw Open. 2020;3(11):e2024984. doi: 10.1001/jamanetworkopen.2020.24984 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Kruse C, Fohn J, Wilson N, Nunez Patlan E, Zipp S, Mileski M. Utilization barriers and medical outcomes commensurate with the use of telehealth among older adults: systematic review. JMIR Med Inform. 2020;8(8):e20359. doi: 10.2196/20359 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Czeisler ME, Marynak K, Clarke KEN, et al. Delay or avoidance of medical care because of COVID‐19‐related concerns ‐ United States, June 2020. MMWR Morb Mortal Wkly Rep. 2020;69(36):1250‐1257. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Oseran AS, Nash D, Kim C, et al. Changes in hospital admissions for urgent conditions during COVID‐19 pandemic. Am J Manag Care. 2020;26(8):327‐328. doi: 10.37765/ajmc.2020.43837 [DOI] [PubMed] [Google Scholar]
- 7. Bhatt AS, Moscone A, McElrath EE, et al. Fewer hospitalizations for acute cardiovascular conditions during the COVID‐19 pandemic. J Am Coll Cardiol. 2020;76(3):280‐288. doi: 10.1016/j.jacc.2020.05.038 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Richards M, Anderson M, Carter P, Ebert BL, Mossialos E. The impact of the COVID‐19 pandemic on cancer care. Nat Cancer. 2020;1(6):565‐567. doi: 10.1038/s43018-020-0074-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Woolf SH, Chapman DA, Sabo RT, Zimmerman EB. Excess deaths from COVID‐19 and other causes in the US, March 1, 2020, to January 2, 2021. JAMA. 2021;325(17):1786‐1789. doi: 10.1001/jama.2021.5199 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Skegg D, Gluckman P, Boulton G, et al. Future scenarios for the COVID‐19 pandemic. Lancet. 2021;397(10276):777‐778. doi: 10.1016/S0140-6736(21)00424-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Pham HH, Schrag D, O'Malley AS, Wu B, Bach PB. Care patterns in Medicare and their implications for pay for performance. N Engl J Med. 2007;356(11):1130‐1139. doi: 10.1056/NEJMsa063979 [DOI] [PubMed] [Google Scholar]
- 12. Stange KC. The problem of fragmentation and the need for integrative solutions. Ann Fam Med. 2009;7(2):100‐103. doi: 10.1370/afm.971 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Yoon J, Zulman D, Scott JY, Maciejewski ML. Costs associated with multimorbidity among VA patients. Med Care. 2014;52(Suppl 3):S31‐S36. doi: 10.1097/MLR.0000000000000061 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Williamson AE, McQueenie R, Ellis DA, McConnachie A, Wilson P. 'Missingness' in health care: associations between hospital utilization and missed appointments in general practice. A retrospective cohort study. PLoS One. 2021;16(6):e0253163. doi: 10.1371/journal.pone.0253163 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Karter AJ, Parker MM, Moffet HH, et al. Missed appointments and poor glycemic control: an opportunity to identify high‐risk diabetic patients. Med Care. 2004;42(2):110‐115. doi: 10.1097/01.mlr.0000109023.64650.73 [DOI] [PubMed] [Google Scholar]
- 16. Berry‐Millett R, Bodenheimer TS. Care management of patients with complex health care needs. Synth Proj Res Synth Rep. 2009;19:52372. [PubMed] [Google Scholar]
- 17. Meek JA. Affordable Care Act: predictive modeling challenges and opportunities for case management. Prof Case Manag. 2012;17(1):15‐21; quiz 22‐13. doi: 10.1097/NCM.0b013e318234e7dd [DOI] [PubMed] [Google Scholar]
- 18. Smith MA, Nordby PA, Yu M, Jaffery J. A practical model for research with learning health systems: building and implementing effective complex case management. Appl Ergon. 2020;84:103023. doi: 10.1016/j.apergo.2019.103023 [DOI] [PubMed] [Google Scholar]
- 19. Huling JD, Yu M, Liang M, Smith M. Risk prediction for heterogeneous populations with application to hospital admission prediction. Biometrics. 2018;74(2):557‐565. doi: 10.1111/biom.12769 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Huling JD, Yu M, Smith MA. Fused comparative intervention scoring for heterogeneity of longitudinal intervention effects. Ann Appl Stat. 2018;13(2):824‐847. doi: 10.1214/18-AOAS1216 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Huling J, Smith M, Chen G. A two‐part framework for estimating individualized treatment rules from semi‐continuous outcomes. J Am Stat Assoc. 2021;116(533):210‐223. doi: 10.1080/01621459.2020.1801449 [DOI] [Google Scholar]
- 22. Yu M, Kuang C, Huling J, Smith M. Diagnosis‐group‐specific transitional care program recommendations for thirty‐day rehospitalization reduction. Ann Appl Stat. 2021;15(3):1478–1498. [Google Scholar]
- 23. Magnan EM, Bolt DM, Greenlee RT, Fink J, Smith MA. Stratifying patients with diabetes into clinically relevant groups by combination of chronic conditions to identify gaps in quality of care. Health Serv Res. 2016;53:450‐468. doi: 10.1111/1475-6773.12607 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Fried LP, Ferrucci L, Darer J, Williamson JD, Anderson G. Untangling the concepts of disability, frailty, and comorbidity: implications for improved targeting and care. J Gerontol A Biol Sci Med Sci. 2004;59(3):255‐263. [DOI] [PubMed] [Google Scholar]
- 25. Pines JM, Lotrecchiano GR, Zocchi MS, et al. A conceptual model for episodes of acute, unscheduled care. Ann Emerg Med. 2016;68(4):484‐491 e483. doi: 10.1016/j.annemergmed.2016.05.029 [DOI] [PubMed] [Google Scholar]
- 26. Centers for Disease Control and Prevention . Different Groups of People, 2020. Accessed August 25, 2021. https://www.cdc.gov/coronavirus/2019-ncov/need-extra-precautions/other-at-risk-populations.html
- 27. Grant JS, Davis LL. Selection and use of content experts for instrument development. Res Nurs Health. 1997;20(3):269‐274. doi: [DOI] [PubMed] [Google Scholar]
- 28. Gnjidic D, Hilmer SN, Blyth FM, et al. Polypharmacy cutoff and outcomes: five or more medicines were used to identify community‐dwelling older men at risk of different adverse outcomes. J Clin Epidemiol. 2012;65(9):989‐995. doi: 10.1016/j.jclinepi.2012.02.018 [DOI] [PubMed] [Google Scholar]
- 29. Agency for Healthcare Research and Quality . Clinical Classifications Software Refined (CCSR). Healthcare Cost and Utilization Project (HCUP). 2021. Accessed September 8, 2021. www.hcup-us.ahrq.gov/toolssoftware/ccsr/ccs_refined.jsp
- 30. Newton KM, Wagner EH, Ramsey SD, et al. The use of automated data to identify complications and comorbidities of diabetes: a validation study. J Clin Epidemiol. 1999;52(3):199‐207. doi: 10.1016/s0895-4356(98)00161-9 [DOI] [PubMed] [Google Scholar]
- 31. Chobanian AV, Bakris GL, Black HR, et al. Seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure. Hypertension. 2003;42:1206‐1252. doi: 10.1161/01.HYP.0000107251.49515.c2 [DOI] [PubMed] [Google Scholar]
- 32. Tu K, Campbell NR, Chen ZL, Cauch‐Dudek KJ, McAlister FA. Accuracy of administrative databases in identifying patients with hypertension. Open Med. 2007;1(1):e18‐e26. [PMC free article] [PubMed] [Google Scholar]
- 33. Hebert PL, Geiss LS, Tierney EF, Engelgau MM, Yawn BP, McBean AM. Identifying persons with diabetes using Medicare claims data. Am J Med Qual. 1999;14(6):270‐277. [DOI] [PubMed] [Google Scholar]
- 34. Vong S, Bell BP. Chronic liver disease mortality in the United States, 1990‐1998. Hepatology. 2004;39(2):476‐483. doi: 10.1002/hep.20049 [DOI] [PubMed] [Google Scholar]
- 35. Foley RN, Murray AM, Li SL, et al. Chronic kidney disease and the risk for cardiovascular disease, renal replacement, and death in the United States Medicare population, 1998 to 1999. J Am Soc Nephrol. 2005;16(2):489‐495. [DOI] [PubMed] [Google Scholar]
- 36. Weiner J. Development and Evaluation of the Johns Hopkins University Risk Adjustment Models for Medicare+Choice Plan Payment. The Johns Hopkins University Health Services Research and Development Center; 2003. [Google Scholar]
- 37. Oreskovic NM, Maniates J, Weilburg J, Choy G. Optimizing the use of electronic health records to identify high‐risk psychosocial determinants of health. JMIR Med Inform. 2017;5(3):e25. doi: 10.2196/medinform.8240 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Dantas LF, Fleck JL, Cyrino Oliveira FL, Hamacher S. No‐shows in appointment scheduling ‐ a systematic literature review. Health Policy. 2018;122(4):412‐421. doi: 10.1016/j.healthpol.2018.02.002 [DOI] [PubMed] [Google Scholar]
- 39. Centers for Medicare & Medicaid Services . ACO #36 Risk‐Standardized Acute Admission Rates for Patients With Diabetes, 2016. Accessed September 22, 2021. https://www.cms.gov/Medicare/Medicare‐Fee‐for‐Service‐Payment/sharedsavingsprogram/Downloads/ACO‐36.pdf
- 40. Centers for Medicare & Medicaid Services . ACO #38 Risk‐Standardized Acute Admission Rates for Patients With Multiple Chronic Conditions, 2016. Accessed September 22, 2021. https://www.cms.gov/Medicare/Medicare‐Fee‐for‐Service‐Payment/sharedsavingsprogram/Downloads/ACO‐38.pdf
- 41. Rural Health Research Center . Rural‐Urban Commuting Area Codes (RUCAs), 2018. Accessed September 8, 2021. http://depts.washington.edu/uwruca/ruca-data.php
- 42. Pope GC, Kautter J, Ellis RP, et al. Risk adjustment of Medicare capitation payments using the CMS‐HCC model. Health Care Financ Rev. 2004;25(4):119‐141. [PMC free article] [PubMed] [Google Scholar]
- 43. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8‐27. doi: 10.1097/00005650-199801000-00004 [DOI] [PubMed] [Google Scholar]
- 44. Salive ME. Multimorbidity in older adults. Epidemiol Rev. 2013;35:75‐83. doi: 10.1093/epirev/mxs009 [DOI] [PubMed] [Google Scholar]
- 45. Centers for Disease Control and Prevention . People with Disabilities, 2021. Accessed August 31, 2021. https://www.cdc.gov/ncbddd/humandevelopment/covid-19/people-with-disabilities.html
- 46. Centers for Disease Control and Prevention . Guidance for Direct Service Providers, Caregivers, Parents, and People with Developmental and Behavioral Disorders, 2021. Accessed August 31, 2021. https://www.cdc.gov/ncbddd/humandevelopment/covid-19/developmental-behavioral-disorders.html
- 47. Centers for Disease Control and Prevention . Homeless Populations: Resources to Support People Experiencing Homelessness, 2021. Accessed August 31, 2021. https://www.cdc.gov/coronavirus/2019-ncov/community/homeless-shelters/
- 48. Centers for Disease Control and Prevention . COVID‐19 and People at Increased Risk, 2021. Accessed August 31, 2021. https://www.cdc.gov/drugoverdose/resources/covid-drugs-QA.html
- 49. Nouri S, Khoong C, Lyles R, Karliner L. Addressing equity in telemedicine for chronic disease management during the COVID‐19 pandemic. NEJM Catalyst Innovations in Care Delivery. 2020.
- 50. Beland S, Dumont‐Samson O, Hudon C. Case management and telehealth: a scoping review. Telemed J E Health. 2021;28:11‐23. doi: 10.1089/tmj.2021.0012 [DOI] [PubMed] [Google Scholar]
- 51. Chen KL, Brozen M, Rollman JE, et al. How is the COVID‐19 pandemic shaping transportation access to health care? Transp Res Interdiscip Perspect. 2021;10:100338. doi: 10.1016/j.trip.2021.100338 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Schulz J, DeCamp M, Berkowitz SA. Medicare Shared Savings Program: public reporting and shared savings distributions. Am J Manag Care. 2015;21(8):546‐553. [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Pearson correlation and p‐values for the relationship between the count of delayed or missed care (DMC) indicators and baseline utilization for age 65+ primary care patients with 2+ chronic conditions.
Table S2. Follow‐up utilization for age 65+ primary care patients with 2+ chronic conditions, for pandemic and comparison cohorts, by level of risk from delayed or missed care (DMC).
Figure S1. Annualized difference in adjusted predicted outcomes per 1000 patients during the pandemic year compared to the prior year, by level of risk from delayed or missed care (DMC).