Detection of Undiagnosed Disease in Medicare Beneficiaries After a Clinical Home Visit

Abstract

Undiagnosed chronic conditions are a common and costly problem in Medicare patients. This study examined whether a clinical home visit program was associated with an increased future detection of undiagnosed diabetes, chronic obstructive pulmonary disease (COPD), and atrial fibrillation. Members of Medicare Advantage Plans (MAP), including Chronic Special Needs Plans (C-SNP), were identified who received a comprehensive geriatric home visit under United Health Group's HouseCalls program and those who did not. Members with no medical or prescription drug claim for diabetes, COPD, and atrial fibrillation in the 12 months prior to the visit were selected. New diagnoses were then identified based on claims for office visits and/or prescription drugs in the 6 months after the HouseCalls visit. Members who received a visit had a significantly higher rate of detection of previously undiagnosed diabetes and COPD, but not of atrial fibrillation. The detection rates for diabetes within 6 months of the visit were 2.8% versus 2.3% (P < 0.01) for MAP and 7.1% versus 5.6% (P < 0.01) for C-SNP members. For COPD, 2.5% versus 2.2% (P < 0.01) of members in MAP and 5.3% versus 4.3% (P < 0.01) of members in C-SNP were newly diagnosed. New diagnoses for atrial fibrillation were not significantly more common for members in MAP (1.4% versus 1.3%)) and C-SNP (1.9% versus 2.1%). These findings suggest that a home visit program, such as HouseCalls, is a promising avenue to address the hidden disease burden and unmet care needs in the Medicare population.

Background

More than 80% of people aged 65 years or older in the United States have at least 1 chronic condition, with the majority suffering multiple coexisting conditions.¹ Many of the chronic conditions that are common in older adults are underdiagnosed and thus untreated. It is estimated that nearly a quarter of diabetes cases, half of COPD cases, and one tenth of atrial fibrillation cases in older adults are undiagnosed.^2,3,4,5

Undiagnosed and untreated chronic conditions are costly. Mapel et al estimated considerably higher direct medical costs for patients with undiagnosed COPD than those with diagnosed COPD, primarily because of greater use of hospital care.⁶ If left untreated, atrial fibrillation can lead to potentially life-threatening and debilitating stroke, of which the average cost of hospital care ranged from $8000 to $23,000 in 2008 dollars, according to a review conducted by Demaerschalk et al.⁷ According to one study, the population attributable risk of stroke in subclinical atrial fibrillation is 13%.⁸ From the societal perspective, the aggregated economic burden of undiagnosed diseases in older adults has important policy implications, in particular for the Medicare program as the dominant payer for older adults.

It has been of interest to better manage chronic conditions for older adults.⁹ Several major carriers of Medicare Advantage Plans (MAP) offer preventive home visits, which feature in-home geriatric assessment conducted by health care professionals. Among them is the HouseCalls Program offered by United Health Group's Optum division for members of UnitedHealthcare Medicare Advantage members—the subject of this study.^10,11 Systematic literature reviews on the effectiveness of preventive home visits for community-dwelling older individuals have reported mixed findings.^12,13,14,15 But it is also worth noting that programs varied in different dimensions and thereby could lead to heterogeneous effects.

In contrast to prior publications, which focused on health care utilization, mortality, and functional decline, this study examines whether receiving a HouseCalls visit was associated with detection of undiagnosed diseases in subsequent visits to health care providers who were not affiliated with the HouseCalls program. For this study, the research team investigated the program effect of HouseCalls on detection of diabetes, COPD, and atrial fibrillation, all of which are commonly undiagnosed in the elderly.

Methods

Description of the intervention

The HouseCalls program was implemented in 2007 by XLHealth Corporation for their Medicare Advantage Chronic Special Needs Plan (C-SNP) members with diabetes, heart failure, and COPD. After XLHealth was acquired by United Health Group, the program was expanded to all members of qualified UnitedHealthcare MAPs irrespective of their underlying conditions.

The HouseCalls program provides an annual in-home visit by either physicians or nurse practitioners who are employees or contractors of companies related to the health plan. Members receive program information and a phone call from the program to schedule an appointment prior to the HouseCalls visit. The visit includes an evaluation of health history, medication review, physical examination, depression screening, nutritional assessment, identification of health risks, member education, and meets the requirements of the Medicare Annual Wellness Visit as defined by the Centers for Medicare & Medicaid Services (CMS). Upon completion of the HouseCalls visit, the member receives a letter, which lists care recommendations consistent with evidence-based guidelines, to be discussed with their primary care provider (PCP).

A key component of the HouseCalls program is care coordination with the member's PCP. After the visit, the results of the assessment are sent to the member's PCP, and the program coordinates follow-up visits with the PCP when health risks and gaps in care are identified. The practitioners also recommend lab tests, make referrals to other care providers such as dietitians and social workers, or obtain assistance for urgent health issues. In addition, program staff work with member services to provide assistance obtaining a PCP when the member does not have one.

Analytic strategy

This study is a retrospective analysis examining whether receiving a home visit under the HouseCalls program is associated with the detection of previously undiagnosed diabetes, COPD, and atrial fibrillation. Those 3 conditions were selected because they are known to be both prevalent and underdiagnosed in older adults, and because a thorough patient history and physical exam might raise the possibility of those conditions being present.

An individual was labeled as undiagnosed with the condition if he or she did not have any inpatient or outpatient encounters with any International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis code or prescription drug claims for the condition during the 12 months prior to his or her first HouseCalls visit. These patients were followed up to the 180^th day after their HouseCalls visit, and the number of first diagnoses that appeared on claims or encounter data (ie, made by health care providers who were not affiliated with the HouseCalls program) were counted. The cumulative percentage of patients having a new diagnosis within the 180 days since the visit was plotted against a comparison diagnosis rate derived from individuals who did not receive a HouseCalls visit. Diagnoses were excluded that were documented on the day of the home visit to minimize misclassification, as the exact time at which the visit occurred was not known and the possibility that the diagnosis was made earlier in the day of the visit could not be ruled out.

A comparison diagnosis rate was estimated based on patients who had never had a HouseCalls visit but were enrolled in the same health plan as patients in the intervention group. Because HouseCalls recipients completed the visit at different points in time throughout the course of a year, the issue of moving the intervention date was addressed by randomly selecting a “synthetic” visit date for these comparison patients. The same data cleaning procedure was then performed to form comparison cohorts of patients absent any diagnosis codes or prescription drug claims for the condition during the 12 months prior to the synthetic visit.

To compare the rate of new diagnosis of the intervention group versus the comparison group, a 2-sample proportions test was conducted to test whether the difference in new diagnosis between the 2 groups is statistically significant.

Data source and sample

Data for diagnoses and prescription drug claims came from MAP encounter data for 2008–2013 and were combined with administrative data on HouseCalls visit dates. The sample was limited to residents of Arkansas, Georgia, Missouri, South Carolina, and Texas, where the HouseCalls program had been operational since January 2008. The analytic file contained de-identified patient-level data for patients enrolled in either the C-SNP or MAPs. Individuals were excluded who did not have 12 months of consecutive enrollment in the health plan prior to the (actual or synthetic) HouseCalls visit. For each of the 3 conditions under study, an analytic sample was constructed consisting of patients who did not have a diagnosis for that condition during the 12 months prior to the (actual or synthetic) visit. Thus, the sample size varied for each condition.

Measurement

The research team identified any Evaluation and Management code with diagnosis of and any prescription drug claim for the 3 conditions before and after the HouseCalls visit. Of note, the team did not consider diagnosis made by the HouseCalls practitioner during a home visit, but only those made during a subsequent inpatient or outpatient encounter with a health care provider who was not affiliated with the HouseCalls program.

Limitations

This analysis has several limitations. First, the true underlying burden of undiagnosed chronic disease in the intervention and comparison groups is unknown, which may bias the results. Although both groups were required to have no medical claim or prescription for the respective condition in the 12 months prior to the actual/synthetic visit, there was no information on risk factors for having those conditions, such as smoking, diet, and body weight, and it is conceivable that the groups had differential rates of undiagnosed disease at baseline. The fact that HouseCalls participation is voluntary can exacerbate the effect of any differential disease burden or disease risk. Second, new diagnoses cannot be directly attributed to the HouseCalls visit, because the research team did not have access to information gathered during the visit that would point to a suspected diagnosis. Third, this analysis focused on only 1 program. The results may not be generalizable to other programs and populations.

Results

Results are presented in Figures 1–1 to 3–2. Each figure shows the cumulative percentage of new diagnosis made subsequently to the actual or synthetic HouseCalls visits for MAP or C-SNP patients. The estimates and the sample size for intervention and comparison groups are presented in the table under each figure.

FIG. 1.

Medicare Advantage (MA) diabetes cumulative new diagnosis (%) since visit and Chronic Special Needs plan (C-SNP) diabetes cumulative new diagnosis (%) since visit.

FIG. 2.

Medicare Advantage (MA) chronic obstructive pulmonary disease (COPD) cumulative new diagnosis (%) since visit and Chronic Special Needs plan (C-SNP) COPD cumulative new diagnosis (%) since visit.

FIG. 3.

Medicare Advantage (MA) atrial fibrillation cumulative new diagnosis (%) since visit and Chronic Special Needs plan (C-SNP) atrial fibrillation cumulative new diagnosis (%) since visit.

Diabetes

As shown in Figure 1, the study data contained 21,024 HouseCalls recipients without a diagnosis of or drug treatment for diabetes in the 12 months prior to the visit in MAP and 11,304 in C-SNP. The respective comparison groups had 136,267 and 5440 patients.

Of MAP members, the percentage of new diagnosis for diabetes was 0.4% in the intervention group versus 0.3% in the comparison group by day 20 after the visit. The difference in cumulative rates grew to 0.3 percentage points by day 40 and remained at this level until day 120 (P < 0.01). By day 180, 2.8% of HouseCalls recipients and 2.3% of comparison patients received a diagnosis for diabetes (P < 0.01). The diagnosis rate in the intervention group was significantly higher than that in the comparison group at either the 0.05 level or the 0.01 level throughout the follow-up period.

Compared to MAP members, a greater percentage of C-SNP members, regardless of HouseCalls participation, were diagnosed with diabetes during the follow-up period. Similar to MAP members, the diagnosis rate among C-SNP HouseCalls recipients was higher than that in the comparison group, but the difference was larger. By day 20, the diagnosis rate was 0.4 percentage points or 50% higher among HouseCalls recipients (P < 0.05). By day 180, 7.1% and 5.6% of HouseCalls recipients and comparison patients, respectively, had received a diagnosis of diabetes (P < 0.01). Again, the difference in diagnosis rates was significant at either the 0.05 level or the 0.01 level over the follow-up period.

The results suggest that approximately 105 and 170 additional members with diabetes were identified in the MAP and C-SNP cohorts, respectively, after 180 days.

COPD

As shown in Figure 2, the analytic file for COPD contained 23,541 HouseCalls recipients without a diagnosis of or drug treatment for COPD in the 12 months prior to the visit in MAP and 20,128 in C-SNP. The respective comparison groups had 146,330 and 7835 patients.

As in the case of diabetes, C-SNP members had higher diagnosis rates for COPD than MAP members throughout the follow-up period, irrespective of HouseCalls participation. For both MAP and C-SNP members, the percentage newly diagnosed with COPD was greater in the intervention group than the comparison group. The difference in diagnosis rates was not statistically significant at the 0.05 level during the first half of the follow-up period.

Among MAP members, 0.3% of HouseCalls recipients and 0.2% of the comparison group received a diagnosis of COPD by day 20 since the visit. By day 180, 2.5% and 2.2% of HouseCalls recipients and comparsion individuals, respectively, were diagnoised with COPD. The difference between the intervention and the comparison groups increased from 0.1 percentage points by day 20 (P = 0.16) to 0.3 percentage points by day 180 (P < 0.01).

For C-SNP members, 0.7% of HouseCalls recipients and 0.6% of the comparison group received a diagnosis of COPD by day 20 since the visit. The cumulative diagnoisis rate increased to 5.3% for HouseCalls recipients and 4.3% for comparison individuals by day 180. The difference between the intervention and the comparison groups went from 0.1 percentage points by day 20 (P = 0.42) to 1 percentage point by day 180 (P < 0.01).

The results suggest that around 71 and 201 additional COPD patients were identified in the MAP and C-SNP cohorts, respectively, after 180 days.

Atrial fibrillation

As shown in Figure 3, there were 27,401 HouseCalls recipients with no prior diagnosis or treatment of atrial fibrillation in MAP and 26,756 in C-SNP. The comparison groups included 165,893 patients in MAP and 10,162 in C-SNP.

HouseCalls recipients enrolled in MAP were no more likely than the comparison group to have received a diagnosis for atrial fibrillation through day 40 after the visit. The cumulative diagnosis rate was slightly higher in the intervention group after day 40. The difference in diagnosis rates between the 2 groups was statistically significant by day 120 (P < 0.05) and day 140 (P < 0.05). However, the effect dissipated afterward. By day 160, the differences were no longer statistically significantly different.

In contrast, the percentage of new diagnosis for atrial fibrillation was lower in the intervention group than the comparison group for C-SNP members. As in all other cases, the cumulative diagnoisis rate in both the intervention and the comparison groups continued to increase as time progressed further away from the day of visit. By day 180, the diagnosis rate in the intervention group was 1.9%, and was 0.2 percentage point lower than that in the comparison group (P = 0.24). However, the difference was not statistically significant by day 180.

Discussion

The research team performed a retrospective analysis to estimate how many additional previously undiagnosed conditions non-HouseCalls health care providers detected in Medicare beneficiaries after a single clinical home visit by a HouseCalls practitioner. The team looked at diabetes, COPD, and atrial fibrillation as conditions that are both common and frequently underdiagnosed in Medicare beneficiaries. Overall, the team found both MAP and C-SNP members to be more likely to be diagnosed with diabetes and COPD, but not with atrial fibrillation, after a home visit.

For both MAP and C-SNP members, there was a small but statistically significant association between receiving a HouseCalls visit and being diagnosed with diabetes or COPD, but not with atrial fibrillation, during the 180 days after the visit. The differences became apparent within the first 20 days of the visit for diabetes and within 80 and 120 days of the visit for C-SNP and MAP members, respectively.

The research team finds the association between HouseCalls visits and new diagnoses plausible for 2 reasons. First, the rates of additional new diagnoses increase with the underlying prevalence of disease and population risk. More cases were identified for diabetes, which is more common than COPD. In fact the inability to detect a difference for atrial fibrillation may have to do with the lower prevalence of this condition, which decreases the power to ascertain any changes. Also, more new cases are seen in the higher-risk C-SNP cohort than in the MAP cohort.

Second, the number of additional cases increases over time. The research team has shown in an earlier analysis that an important effect of the HouseCalls program is to increase patient engagement with the health care system by raising the number of physician office visits in the first year after the home visit.¹⁶ This effect would explain that much of the difference in case finding occurs well after 80 days post visit (ie, at a point in time when one cannot plausibly assume that the new diagnosis is directly related to the home visit). In other words, getting patients to see their regular providers may be an important mechanism by which the clinical home visit program finds undiagnosed diseases, consistent with CMS' expectation that home visit programs can be used to improve care planning and care coordination.¹⁰

Clearly, this exploratory research needs to be validated by future work. The research team will need to correlate information gathered during the HouseCalls visit with the new diagnoses and determine whether indeed the patients with additional office visits were more likely to be diagnosed. The team also needs to ascertain that patients receive proper treatment post diagnosis and experience better disease control and health outcomes. But the initial findings point toward a home visit program, like HouseCalls, as a promising avenue to address the hidden disease burden and unmet care needs in the Medicare population.

Footnotes

Author Disclosure Statement

Mr. Han, Ms. Wilks, and Dr. Mattke declared no conflicts of interest with respect to the research, authorship, and/or publication of this article.

The authors received the following financial support for this article: this article was funded by a contract with UnitedHealth Group. The funder reviewed an earlier version of the article and provided comments, but the authors made the final decisions on research design, interpretation of the findings, the text of the article, and the decision to submit.

References

Ward

, Schiller

, Goodman

. Multiple chronic conditions among US adults: A 2012 update. Prev Chronic Dis, 2014; 11:E62.

National Center for Chronic Disease Prevention and Health Promotion. National Diabetes Statistics Report: Estimates of Diabetes and Its Burden in the United States, 2014. Atlanta, GA: US Department of Health and Human Services; 2014.

CDC/National Center for Health Statistics. Early release of selected estimates based on data from the 2014 National Health Interview Survey: Diagnosed Diabetes. June 2015. Available at: <http://www.cdc.gov/nchs/nhis/released201506.htm#14>. Accessed February 29, 2016 .

Mannino

, Homa

, Akinbami

, Ford

, Redd

. Chronic obtructive pulmonary disease surveillance—United States, 1971–2000. MMWR Surveill Summ, 2002; 51(6):1–16.

Turakhia

, Shafrin

, Bognar

, et al. Economic burden of undiagnosed nonvalvular atrial fibrillation in the United States. Am J Cardiol, 2015; 116:733–739.

Mapel

, Robinson

, Dastani

, Shah

, Phillips

, Lydick

. The direct medical costs of undiagnosed chronic obstructive pulmonary disease. Value Health, 2008; 11:628–636.

Demaerschalk

, Hwang

H-M

, Leung

. US cost burden of ischemic stroke: A systematic literature review. Am J Manag Care, 2010; 16:525–533.

Healey

, Connolly

, Gold

, et al. Subclinical atrial fibrillation and the risk of stroke. N Engl J Med, 2012; 366:120–129.

Nelson

. Lessons from Medicare's demonstration projects on disease management and care coordination: Working paper 2012–01. January 18, 2012. Available at: <https://www.cbo.gov/publication/42924>. Accessed February 29, 2016 .

10.

Centers for Medicare & Medicaid Services. 2016 Advance Notice. February 20, 2015. Available at: <https://www.cms.gov/Medicare/Health-Plans/MedicareAdvtgSpecRateStats/Announcements-and-Documents-Items/2016Advance.html>. Accessed February 29, 2016 .

11.

Kronick

, Welch

. Measuring coding intensity in the Medicare Advantage program. Medicare Medicaid Res Rev, 2014; 4(2):E1–E19.

12.

Mayo-Wilson

, Grant

, Burton

, Parsons

, Underhill

, Montgomery

. Preventive home visits for mortality, morbidity, and institutionalization in older adults: A systematic review and meta-analysis. PLoS ONE, 2014; 9(3):e89257.

13.

van Haastregt

JCM

, Diederiks

JPM

, van Rossum

, de Witte

, Crebolder

HFJM

. Effects of preventive home visits to elderly people living in the community: Systematic review. BMJ, 2000; 320(7237):754–758.

14.

Stuck

, Egger

, Hammer

, Minder

, Beck

. Home visits to prevent nursing home admission and functional decline in elderly people: Systematic review and meta-regression analysis. JAMA, 2002; 287:1022–1028.

15.

Huss

, Stuck

, Rubenstein

, Egger

, Clough-Gorr

. Multidimensional preventive home visit programs for community-dwelling older adults: A systematic review and meta-analysis of randomized controlled trials. J Gerontol A Biol Sci Med Sci, 2008; 63:298–307.

16.

Mattke

, Han

, Wilks

, Sloss

. Medicare home visit program associated with fewer hospital and nursing home admissions, increased office visits. Health Aff, 2015; 34:2138–2146.