 To examine drug costs and entry and exit rates into the Part D coverage gap for beneficiaries with diabetes in Medicare Advantage managed care plans.

Cross-sectional observational study.  Methods: Study patients were Medicare Advantage Part D beneficiaries with diabetes from 2 large California health plans who were continuously enrolled in 2006 and had a drug coverage gap starting at $2250. Entry and exit into the gap, total drug costs, and out-of-pocket drug costs were determined using pharmacy databases.

In 2006, 26% of the 42,801 beneficiaries with diabetes reached the coverage gap; 2% of beneficiaries exited the gap and qualified for catastrophic coverage. Beneficiaries incurred a mean of $2182 in total drug costs during 2006. Drug expenditures remained stable over the year for beneficiaries who did not enter the gap. For beneficiaries who entered the gap, total drug costs were higher overall and decreased at year’s end as out-of-pocket expenses increased.

Fewer diabetes patients in this study entered the coverage gap than had been previously estimated, but the entry rate was much higher than that of the general Medicare Advantage Part D population. Patients entering the gap had lower subsequent monthly drug expenditures; this may be due to lower-than-expected drug prices and greater use of generics in managed care, or it may potentially signal poorer drug adherence. Future work should examine these hypotheses and explore risk factors for entering the Part D coverage gap.

Patients with chronic illnesses such as diabetes may be particularly vulnerable to “coverage gap” benefit designs such as those in Medicare Part D.

We observed that 26% of diabetic patients in major California Medicare Advantage Part D plans reached the drug coverage gap in 2006.

These patients’ overall drug costs dropped as their out-of-pocket costs increased, a pattern that may signal cost-related medication nonadherence.

Future work should examine the effects of Part D on medication nonadherence in patients with chronic illness and risk factors for entering the coverage gap.

The Medicare Part D drug benefit, introduced in January 2006, was designed to help address the needs of seniors with high out-of-pocket costs for medication.

 and to provide continuous access to needed medications for chronic illness.

 A unique feature of Medicare Part D is that under the standard benefit structure, enrollees have a potential coverage gap, commonly referred to as a “doughnut hole,” in prescription drug cost coverage. The standard Part D benefit in 2006 began with partial coverage for the first $2250 of total drug costs in 2006, followed by a period of no coverage until patients reached cumulative out-of-pocket costs of $3600 in 2006.

 At the end of the gap, catastrophic coverage begins; patients often pay only 5% of drug costs or a predetermined copayment from that time forward.

 An estimated 18.2 million beneficiaries (89% of Part D enrollees not receiving a low-income subsidy) in 2006 were enrolled in a Medicare Part D plan with a full coverage gap.

Patients with chronic illnesses such as diabetes are at particularly high risk for facing high drug and other out-of-pocket costs given their need for chronic drug therapy and frequent need for multiple drugs to treat comorbid conditions. Persons with diabetes in particular often require daily medication to control their blood sugar levels, plus concurrent chronic treatment for other conditions.

 The cost of managing chronic illness is high for the government as well, and the diabetes population is of particular interest. While approximately 20% of Medicare beneficiaries have diabetes, the Centers for Medicare & Medicaid Services has estimated that up to 32% of all Medicare spending could be attributed to patients with diabetes.

 The costs for drugs such as insulin and oral agents to control diabetes account for up to 12% of the healthcare expenditures attributable to the disease.

The high levels of clinical need for drug therapy combined with substantial out-of-pocket costs for patients with diabetes may make them more vulnerable to cost-related medication nonadherence.

 Studies suggest that elderly patients reduce their medication use when faced with limited prescription benefit coverage

 and experience adverse health consequences including nonelective hospitalizations and death.

 Results of current studies examining the effect of Medicare Part D on adherence and cost are mixed. Some studies suggest Medicare Part D’s introduction improved adherence and decreased out-of-pocket costs,

 while others found evidence that drug costs decreased medication adherence, especially for those entering the coverage gap.

 Such effects may be more pronounced in patients with comorbidities and worse health status

One recent study conducted in a Medicare Advantage Prescription Drug (MAPD) plan offered by an integrated delivery system included in this study reported that 8% of beneficiaries overall entered the coverage gap in 2006,18 a rate much lower than pre-2006 projections of gap entry rates for the general population.

 One pre—Part D implementation study of diabetic Medicare beneficiaries estimated that 64% would enter the coverage gap in the first year.

 However, there is currently no empirical information on how many seniors with diabetes enter and exit the doughnut hole of prescription drug coverage in Medicare Part D plans.

This study examined total drug costs, out-of-pocket drug costs, and the rates of entry and exit into the Medicare Part D coverage gap in an important subset of Medicare beneficiaries, those with diabetes, for patients from 2 large health plans offering MAPD benefits in California.

Patients from 2 large California health plans offering MAPD benefits in 2006 were included in the study if they were age 65 years or older, were identified as having diabetes, and were beneficiaries of the Medicare Part D program. The first plan (plan A) is a for-profit network model HMO serving patients primarily in Southern California. The second plan (plan B) is a not-for-profit integrated delivery system HMO model that serves patients in Northern California.

Patients were determined to have diabetes if they had filled a prescription for at least 1 glucose-lowering medication (eg, insulin, metformin) at some point during 2005. Fifteen percent of Medicare beneficiaries in both plans had diabetes based on this criterion. In order to be eligible for the study, these diabetes patients also were required to have been continuously enrolled in the health plan between January 1, 2006, and December 31, 2006, so that complete pharmacy data could be captured (11% of patients were excluded from the study on this basis). An additional 26% were excluded from the study if they had any type of drug cost coverage during the gap, including a supplemental retiree benefit plan or a low-income supplement provided by Medicaid or through the Part D program.

Patient age, sex, and Medicare Part D enrollment were determined from automated databases on demographics and coverage status maintained by both plans. Comorbidity data were determined by using 2005 claims data for patients in plan A and 2005 automated outpatient clinical records for patients in plan B.

Individual entry and exit into the Medicare Part D coverage gap were determined by using pharmacy claims data for patients in plan A and automated outpatient pharmacy records for patients in plan B. In both plans, patients reached the coverage gap after total drug costs exceeded $2250, which included drug acquisition costs and dispensing fees. Patients exited the gap once out-of-pocket costs reached a total of $3600 during the year.

This study was developed and approved by the Steering Committee of the Translating Research in Action for Diabetes (TRIAD) Study and conducted by researchers in 2 TRIAD Translational Research Centers. The study protocol was approved by the Kaiser Permanente Northern California and University of California at Los Angeles institutional review boards.

The MAPD plans had a total of 42,801 beneficiaries with diabetes who met study criteria. Almost half (47%) were between 65 and 74 years of age, 43% were between 75 and 84 years of age, and 10% were age 85 years and older. Fifty-one percent of study subjects were female. Most patients (81%) were using only oral hypoglycemics to manage their diabetes. Eighty percent had an 

International Classification of Diseases, Ninth Revision

 diagnosis of hypertension, and 65% had hyperlipidemia. The percentage of elderly patients, percentage of female patients, diabetes drug use patterns, and comorbidity burden were similar in both plans. During 2006, 26% of diabetes patients entered the coverage gap. Only 2% of all beneficiaries in the study had $3600 in total drug costs, therefore qualifying for catastrophic coverage drug cost assistance in 2006.

 shows the cumulative percentage of diabetes patients entering the coverage gap by month. Fewer than 1% of patients entered the prescription drug coverage gap in the first quarter of 2006 (January-March); by the end of the second quarter (June) only 5% had entered the coverage gap. More than half of the patients who entered the coverage gap during the year did so after August 31.

On average, diabetes patients incurred $2182 in total drug costs during 2006, with a mean of $807 in out-of-pocket drug costs to patients. 

 shows the amount of total and out-of-pocket drug expenditures for the MAPD beneficiaries in the study who did and did not reach the coverage gap in 2006. For those who did not reach the gap, both total drug expenditures and out-of-pocket drug expenditures remained stable over time during the year. For the 26% of MAPD patients in the study who reached the coverage gap in 2006, total drug costs were much higher overall and decreased toward the end of the year as out-of-pocket drug expenses increased. Analysis-of-variance tests of the differences in drug costs showed that the decrease in total drug costs between quarter 3 and quarter 4 of 2006 and the quarterly increases in out-of-pocket costs for these patients were significant (

Data from both Medicare Advantage plans in this study suggests that Medicare Advantage beneficiaries with diabetes are entering and exiting the doughnut hole of drug coverage at a far lower rate than originally expected. One study projected that 64% of patients with diabetes who were eligible for Medicare Part D would have drug costs exceeding $2250 and incur significant out-of-pocket spending in 2006.20 We found a much lower rate of gap entry than this original prediction (26% of patients entered the gap in 2006).

There are many potential explanations for the disparity between previous projections and the actual experience of diabetes patients in these Medicare Advantage plans. One possible reason is that patients in these managed care plans are receiving more generic therapies than those in the general diabetes population previously studied. These managed care plans may provide appropriate drug therapies in ways that lower costs through methods such as through formulary management, generic substitution, or multitiered copayments.

Another potential explanation for the difference between projected and actual rates of entering and exiting the coverage gap is that patients could use fewer medications up front to avoid entering the coverage gap, and then continue to use fewer medications once the doughnut hole is reached. Previous projections of gap entry rates were based on assumptions that patient drug spending would stay steady as patients approach the gap, as well as after the gap was reached.20 Our findings demonstrate that total drug costs were decreasing in the latter part of the year as out-of-pocket costs were rising, which suggests that medication use by diabetes patients could be decreasing once they hit the coverage gap. This observation would support the findings of other Medicare Part D studies, which found evidence of decreased drug adherence based on cost,1,18 especially for patients with worse health status such as those with diabetes.

 Because cost-related medication nonadherence is associated with poorer health status, greater rates of cardiac adverse events, and increased use of emergency services,

 particular attention should be paid to addressing this potential risk for diabetes patients in Medicare Part D plans.

This study has a number of limitations that should be noted. This study only examines patients in MAPD plans, which make up 28% of the total Part D population; most Medicare Part D enrollees are covered by stand-alone Prescription Drug Plans (PDPs).23 Further research will be required to examine gap entry rates, drug costs, and risk for cost-related medication nonadherence in PDP patients with diabetes. This study identified diabetes patients through drug claims data rather than diagnosis codes; however, a sensitivity analysis showed that requiring a diabetes diagnosis in addition to antidiabetic drug use resulted in a very small number of patient exclusions (<5%). This study only examined the experience of Medicare Part D enrollees in 2 California plans, who may differ from enrollees nationally. However, examination of plan A data on enrollees with diabetes in 7 other states (Arizona, Colorado, Nevada, Oklahoma, Oregon, Texas, and Washington) showed that the plan A gap entry rate for all western states combined was almost identical to that of California alone. This result suggests that the results presented here are representative of the overall MAPD enrollees’ experience.

Finally, this study did not specifically demonstrate that decreases in drug spending lead to adverse outcomes such as increased mortality and morbidity. Future research is required to examine this important issue.

This study is the first to examine actual total and out-of-pocket drug costs, and empirical rates of entering and exiting the coverage gap, in a Medicare Advantage population with chronic illness. Our findings suggest that diabetes patients are more vulnerable to entering the coverage gap than MAPD beneficiaries in the general population, but are entering the gap at lower-than-expected rates for diabetes patients. These lower-than-expected costs and rates of experiencing the doughnut hole may reflect efforts by Medicare Advantage plans to provide appropriate drug therapies in ways that lower costs, or may suggest that seniors are significantly lowering their use of prescription medications in response to high out-of-pocket costs. This is a critically important question for policy makers, patients, and their providers, and further research should examine specific patient responses to gaps in prescription drug coverage in Medicare.

: From the Division of Research (JAS, VF, JH, BES, JTH), Kaiser Permanente Medical Care Program, Northern California, Oakland, CA, and the Division of General Internal Medicine and Health Services Research and the School of Public Health (SLE, NT, ESQ, CMM), University of California at Los Angeles, Los Angeles.

: This project was funded by Centers for Disease Control Contract no. U58/CCU923527-04-1; the Agency for Healthcare Research and Quality (AHRQ) and the National Institute on Aging (NIA) (R01-HS013902-01) and the National Institute on Aging (R01-AG029316-01). Dr Schmittdiel is supported by the Office of Research in Women’s Health Building Interdisciplinary Careers in Women’s Health K12 Career Development Award (K12HD052163). Dr Mangione is partially supported by the UCLA Resource Center for Minority Aging Research and by a grant from the National Institute on Aging (2P30AG021684-06).

: Dr Huang is an employee of Kaiser Foundation Research Institute, and he reports having received grants from NIH, AHRQ and Commonwealth Fund (CMWF). He reports receiving internal funding from the Kaiser Foundation Research Institute. Dr Hsu is an employee of Permanente Medical Group and he reports having received grants from NIH, AHRQ, and CMWF. He also reports having received funding from the Kaiser Foundation Research Institute and internal funding from Permanente Medical Group. The authors (JAS, SLE, VF, NT, ESQ, BES, CMM) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

: Concept and design (JAS, SLE, VF, BES, JTH, CMM); acquisition of data (JAS, JH, NT, ESQ, BES, JTH, CMM); analysis and interpretation of data (JAS, SLE, VF, JH, NT, BES, CMM); drafting of the manuscript (JAS); critical revision of the manuscript for important intellectual content (SLE, VF, ESQ, BES, CMM); statistical analysis (JH, NT); obtaining funding (BES, CMM); administrative, technical, or logistic support (ESQ, BES, CMM); and supervision (JTH, CMM).

: Julie A. Schmittdiel, PhD, Division of Research, Kaiser Permanente Medical Care Program, 2000 Broadway, Oakland, CA 94612. E-mail: julie.a.schmittdiel@kp.org.

1. Neuman P, Strollo MK, Guterman S, et al.

 Medicare prescription drug benefit progress report: findings from a 2006 national survey of seniors. 

 Prescription drug coverage and seniors: findings from a 2003 national survey. 

3. Soumerai SB, Pierre-Jacques M, Zhang F, et al.

 Cost-related medication nonadherence among elderly and disabled Medicare beneficiaries: a national survey 1 year before the Medicare drug benefit. 

 A prescription for a modern Medicare program. 

 The first months of the prescription-drug benefit—a CMS update. 

6. Stuart S, Briesacher BA, Shea DG, Cooper B, Baysac FS, Limcangco MR. 

Riding the roller coaster: the ups and downs in out-of-pocket spending under the standard Medicare drug benefit [published correction appears in 

7. Medicare Payment Advisory Commission (MedPAC).

Report to the Congress: Medicare Payment Policy

. March 2007: chap 4. http://www.medpac.gov/documents/

Mar07_EntireReport.pdf. Accessed June 10, 2008.

 Implications of the Diabetes Control and Complications Trial. 

 Standards of medical care for patients with diabetes mellitus. 

10. Pearson TA, Blair SN, Daniels SR, et al.

 AHA Guidelines for Primary Prevention of Cardiovascular Disease and Stroke: 2002 Update: Consensus Panel Guide to Comprehensive Risk Reduction for Adult Patients Without Coronary or Other Atherosclerotic Vascular Diseases. American Heart Association Science Advisory and Coordinating Committee. 

 Medicare coverage for patients with diabetes. A national plan with individual consequences. 

 Economic costs of diabetes in the U.S. in 2007. 

13. Jackson JE, Doescher MP, Saver BG, Fishman P.

 Prescription drug coverage, health, and medication acquisition among seniors with one or more chronic conditions. 

14. Ho PM, Rumsfeld JS, Masoudi FA, et al.

 Effect of medication nonadherence on hospitalization and mortality among patients with diabetes mellitus. 

 Unintended consequences of caps on Medicare drug benefits. 

 The impact of Medicare Part D on prescription drug use by the elderly. 

17. Yin W, Basu A, Zhang JX, Rabbani A, Meltzer DO, Alexander GC. 

The effect of the Medicare Part D prescription benefit on drug utilization and expenditures. 

18. Hsu J, Fung V, Price M, Huang J, Brand R, Hui R, Fireman B, Newhouse JP.

 Medicare beneficiaries’ knowledge of the Part D prescription drug benefit and response to drug costs. 

19. Madden JM, Graves AJ, Zhang F, et al.

 Cost-related medication nonadherence and spending on basic needs following implementation of Medicare Part D. 

 Will the Medicare prescription drug benefit eliminate cost barriers for older adults with diabetes mellitus? 

 The effect of insurance type on prescription drug use and expenditures among elderly Medicare beneficiaries. 

. Impact of multitiered copayments on the use and cost of prescription drugs among Medicare beneficiaries. 

 Status report on Medicare Part D enrollment in 2006: analysis of plan-specific market share and coverage. 

Articles

TRENDS FROM THE FIELD

When diabetic patients in 2 Medicare Advantage Part D plans reached the 2006 coverage gap, overall drug costs dropped as out-of-pocket costs increased.

Medicare Part D Coverage Gap and Diabetes Beneficiaries

 To use a natural experiment to identify patient factors associated with the decision to follow one’s primary care provider (PCP) to a more distant location after the closure of a medical practice.

Case-control study.  Methods: Eight months after the closure of a practice in Dundalk, Maryland, we randomly selected 140 patients older than 60 years from each of the following groups: those who followed their PCP (continuity group) and those who transferred to a closer clinic (proximity group). We designed a survey instrument to collect information about demographics, duration of the patient-PCP relationship, transportation, self-assessed driving proficiency, and patients’ estimates of the distance in miles and the driving time in minutes from their homes to both practices. 

 Tests and logistic regression analyses were used to determine differences between the groups.

The response rate to the survey was 64%. More than 85% of patients in both groups had been with their original PCP for longer than 2 years. In multivariable analysis, the following 3 factors were associated with being a patient in the proximity group: living alone (adjusted odds ratio [OR], 3.14; 95% confidence interval [CI], 1.35-7.26), having greater physical disability (physical component summary score <40; OR, 2.14; 95% CI, 1.04-4.39), and perceiving that travel time from home to the farther clinic would require at least 10 minutes longer than the MapQuest estimate (OR, 4.08; 95% CI, 1.97-8.43).

Older patients who live alone and are weaker seem to be more likely to forgo continuity with their PCP for the sake of convenience when a barrier to access occurs such as relocation of the physician to a more distant office.

Patients who elect to maintain continuity with their primary care provider (PCP) even when a barrier such as relocation of the physician to a more distant office is imposed differ from those who forgo continuity of care for the sake of convenience.

This study adds to the understanding of the factors associated with decisions made by older patients related to maintaining continuity with a PCP. 

Organizations may want to be particularly mindful of and sensitive to patients’ needs when planning changes that will be disruptive to or affect continuity of care.

Sustained relationships between primary care providers (PCPs) and patients foster the development of meaningful connections that promote the delivery of high-quality comprehensive care.

Continuity of care is associated with higher patient satisfaction,

 improved care of chronic health conditions,

 enhanced patient adherence to medical regimens,6 and decreased emergency department use and hospitalizations.

 Specific subgroups of patients, namely, the elderly and patients with chronic medical conditions, seem to particularly value continuity with their physicians.

 Continuity of care may be threatened when practices downsize, relocate, or close because of economic conditions such as lower reimbursement schedules or higher malpractice premiums.

 Despite their valuing continuity in care, prior research suggests that few patients will spend additional time or money to maintain it.

 In particular, elderly patients (for whom longer distances present barriers of driving and public transportation) may find it more difficult to follow PCPs who relocate to geographically distant clinical sites.

In this study of patients whose primary care clinic closed, we examined the factors associated with older patients’ choosing to follow their PCPs to a more distant site compared with older patients who transferred their care to new PCPs at an affiliated clinic near the original clinic. We hypothesized that patients who were older and sicker and those who overestimated the distance between their home and the distant clinic would sacrifice continuity for the sake of accessing care closer to them.

We conducted a case-control study comparing older patients who followed their PCPs to a more distant clinic with older patients who transferred their care to a practice close to their original clinic. The former constituted the continuity group, and the latter constituted the proximity group.

The study population comprised patients served by 5 internal medicine PCPs at a freestanding clinic practice in Dundalk, Maryland. Five months before the closure of the clinic, all patients were invited to follow their PCPs to a new practice 11 miles from the original clinic. Alternatively, they could opt to transfer their care and records to an affiliated primary care clinic located in the same town 2 miles from the original clinic, or they could find another arrangement on their own. Both the distant and nearby primary care group practices were the same as the original clinic in terms of structure, operations, size, and acceptance of the same insurance types.

Eight months after the closure of the primary clinic, it became apparent where the patients had elected to pursue their ongoing care. We used electronic databases to generate the following 2 lists of patients older than 60 years from the original office: (1) those who had followed their PCP to the new farther clinic (continuity group) and (2) those who had transferred their care to a new PCP at the affiliated nearby clinic (proximity group). We chose to study older patients because they have more chronic medical conditions and because the barrier of transportation is more significant among older adults. Patients living in nursing homes or assisted-living communities were excluded from the study.

From each list, we randomly selected 140 patients for inclusion in the study. To do this, we used a random integer generator (http://www.random.org).

The research team used an evidence-based iterative process to develop the questionnaire. We performed a comprehensive literature review on the topics of clinic closings, physician relocation, continuity of care, and patient loyalty to physicians. We organized the questionnaire into the following sections: demographic characteristics, self-reported medical conditions, duration of the patient-PCP relationship, driving proficiency, and patient perceptions about distance and travel time between their home and the 2 clinic sites. The literature review and the experience of the research team (who have also undergone practice location moves) contribute evidence of the instrument’s content validity.

The survey included the following 2 instruments: (1) the previously validated reliable 12-Item Short-Form Health Survey to measure physical disability (physical component summary [PCS] score) and mental disability

 and (2) Drivers 55 Plus Self-Rating Form, a measure designed by the American Automobile Association to assess driving proficiency and to enhance self-awareness about areas for improvement.

Two survey questions asked patients to approximate the distance in miles and the travel time in minutes from their home to the distant clinic and to the closer clinic. We also verified actual distances and times by entering the patient’s home address and both clinic addresses into an online map service, MapQuest.

The final version of the 4-page questionnaire took approximately 20 minutes to complete. To improve the clarity of questions, we conducted pilot testing among elderly patients from the original clinic who were not randomly selected for participation in the study.

We mailed the survey to the study sample 8 months after the closure of the original primary care practice. This time frame was selected such that participants would have interacted with and received care at their new site. We included a consent form to be signed and returned to the investigators. We used repeated mailings to encourage participation. A Johns Hopkins University institutional review board approved the study.

The continuity group (who followed PCPs to the distant clinic) was compared with the proximity group (who transferred to the closer clinic). 

 Tests were used for dichotomous variables and 

We also compared patients’ perceptions versus online estimates of distances (in miles) and times (in minutes) from their homes to the clinics. We categorized patients’ estimates as overestimates or as underestimates.

Logistic regression modeling was used to identify variables that independently distinguished between continuity and proximity patients. We included variables in the final model that had statistical associations of 

 <.10 in bivariate analyses. We applied stepwise backward selection logistic regression analysis and took into account the change in model 

 statistic with the addition and deletion of each variable. Final model variables were examined for evidence of colinearity and for interactions.

We estimated that a sample size of 168 patients (which assumed a 60% response rate from 280 patients surveyed) would be adequately powered to detect meaningful differences between the groups. For example, to detect a 5-point difference (SD, 10) in the PCS score of the 12-Item Short-Form Health Survey with 0.05 significance and 0.80 power requires 63 patients from each group. We conducted all statistical analyses using STATA 8.2 (StataCorp LP, College Station, TX).

Response Rate and Characteristics of Respondents

Surveys were completed by 104 of 140 (74%) continuity patients and by 76 of 140 (54%) proximity patients. Sex and racial/ethnic characteristics of the responders were similar to the population characteristics of the suburban town of Dundalk, and nonresponders were similar to respondents in terms of age, sex, and race/ethnicity (

Characteristics of the proximity and continuity patients are given in 

. In bivariate analyses, proximity patients were older than continuity patients (mean age, 75 vs 70 years; 

 <.001). Proximity patients were more likely to live alone (38% vs 19%, 

 = .003) and to have Medicare only or Medicare with secondary insurance (72% vs 54%, 

 = .01).Proximity patients also had lower 12-Item Short-Form Health Survey PCS scores, indicating greater disability (mean, 36.8 vs 40.3; 

The percentages of patients still driving (

 = .09) and their self-assessed driving proficiency scores (

 = .18) were high and were not significantly different between the groups (

). However, proximity patients were more likely to overestimate the driving time that it would take to reach the more distant

Characteristics that were associated with increased odds of being a proximity patient (

 <.10) were included in the multivariable model. 

 summarizes the modeling that included and adjusted for all such variables. The following 3 variables were independently associated with being a proximity patient: living alone (odds ratio [OR], 3.14; 95% confidence interval [CI], 1.35-7.26), having greater physical disability (PCS score <40; OR, 2.14; 95% CI, 1.04-4.39), and perceiving that travel time from home to the farther clinic would require at least 10 minutes longer than the MapQuest estimate (OR, 4.08; 95% CI, 1.97-8.43). No colinearity was found among the variables in the final model.

The objective of this study was to investigate, after the closure of a primary care clinic, the factors associated with older patients following their established PCP to a more distant clinic compared with older patients transferring their care to a closer clinic with a new PCP. We hypothesized that more frail patients and those with barriers to travel would not follow their PCP to the distant site. We found that living alone, having a lower PCS score (or more disability), and overestimating the time required to travel from one’s home to the disntant clinic were the only 3 variables that were independently associated with older patients who chose to forgo continuity with their PCPs for the practicality of a closer clinic site.

Prior research has suggested that the elderly and those with chronic medical conditions especially value continuity of care with their PCPs.

 In a longitudinal cohort study, Mold et al10 demonstrated that older patients have greater satisfaction in longer relationships with their PCPs and tend to stay with their PCPs until forced to change; the impetuses for change most commonly observed were insurance-related issues and physician retirement or relocation. Our study adds to the literature in that it focused in a natural experiment solely on older adults at a time when a change in their healthcare delivery options was forced on them.

The findings of the study suggest that older patients may forgo continuity because of perceptions of inconvenience in traveling to a relocated PCP. The elderly may have difficulty finding safe, affordable, and efficient transportation.

 Even among those who drive, the elderly may be less willing to travel greater perceived distances because of reduced driving confidence, higher crash rates among older adults, and increased morbidity and mortality from motor vehicle crashes.

 The prospect of having to drive longer distances to see one’s PCP when acutely ill and feeling sick may be particularly concerning for some. Individualized decisions about the relative value of proximity versus continuity are complex and require careful reflection about priorities.

Several limitations of this study should be considered. First, this study relied exclusively on self-report for demographic, health, and driving characteristics. Second, this study included older patients affected by the closure of a single clinic in a specific geographic area, and our findings may not be generalizable to other populations. Third, the timing of the inquiry purposively occurred 8 months after the change so that patients had the opportunity to settle into a new primary care home. However, this delay may have introduced some recall bias. Fourth, all actual driving times and distances were estimated using an online map service. These services default to the most direct and quickest routes; however, the older patients in this cohort may take alternate courses (such as avoiding highways) and may travel at slower speeds such that their estimates may in fact be informed by their own driving practices and  experiences. Fifth, cross-sectional studies

demonstrate that associations and causality cannot be implied from the results. Sixth, we did not specifically assess the patients’ perceptions of the quality of the relationship that they had with their original PCP. However, approximately 90% had been seeing that PCP for more than 2 years, and this may be viewed as a proxy for satisfaction.

In conclusion, this study adds to our understanding of the patient factors associated with decisions made by older patients related to maintaining continuity with a PCP. Perhapsmore should be done to help older patients to make accurate assessments of the actual distance and time required to reach a new clinic location when their PCPs are relocated. Future research could investigate how to support continuity of care for older adults and other vulnerable populations in the face of major changes in the healthcare delivery systems. Organizations may want to be particularly mindful of and sensitive to patients’ needs when planning changes that will be disruptive to or affect continuity of care.

Dr Wright is an Arnold P. Gold Foundation Professor of Medicine and a Miller-Coulson Family Scholar. We are grateful for the support from The Osler Center for Clinical Excellence at the Johns Hopkins University School of Medicine.

: From the Division of Medicine (AG), Franklin Square Hospital Center, Baltimore, MD; the Division of General Internal Medicine (MAN), Columbia University Medical Center, New York, NY; and the Division of General Internal Medicine (AG, NR, JAC, SMW), Johns Hopkins University School of Medicine, Baltimore, MD.

: The authors (AG, MAN, NR, JAC, SMW) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

: The authors received support from The Osler Center for Clinical Excellence at the Johns Hopkins University School of Medicine.

: Concept and design (MAN, SMW); acquisition of data (MAN, SMW); analysis and interpretation of data (AG, NR, JAC, SMW); drafting of the manuscript (AG, NR, SMW); critical revision of the manuscript for important intellectual content (AG, MAN, NR, JAC, SMW); statistical analysis (AG, JAC); obtained funding (SMW); administrative, technical, or logistic support (SMW); and supervision (SMW).

: Scott M. Wright, MD, Division of General Internal Medicine, Johns Hopkins University School of Medicine, 5200 Eastern Ave, Ste 2300, Baltimore, MD 21224. E-mail: smwright@jhmi.edu.

 Does continuity of care improve patient outcomes? 

 Length of patient-physician relationship and patients’ satisfaction and preventive service use in the rural south: a cross-sectional telephone study. 

3. De Maeseneer JM, De Prins L, Gosset C, Heverick J.

 Provider continuity in family medicine: does it make a difference for total health care costs? 

4. Love MM, Mainous AG III, Talbert JC, Hager GL

. Continuity of care and the physician-patient relationship: the importance of continuity for adult patients with asthma. 

There are differences between patients who elect to maintain continuity with their PCP versus those who choose to change their PCP for convenience.

Patient Factors Associated With Following a Relocated Primary Care Provider Among Older Adults

 To examine attitudes, experiences, and preferences regarding advance directives (ADs) among adults of all ages.

Subjects were surveyed regarding end-of-life (EOL) care wishes. The survey included items measuring knowledge, experiences, and attitudes regarding ADs, as well as preferences toward initiation of discussions.   Methods: Subjects included a random sample of patients (age range, 20 to >80 years) from a large midwestern managed care organization stratified by age decade. Descriptive statistics were used to summarize variables, and c2 tests were performed to examine differences by age category.

One hundred eighty-seven surveys were completed. The likelihood of having completed an AD increased with age (P <.001). Of those without an AD, 44% reported having talked with someone about their wishes. Many (62%) subjects thought that it was up to them to raise the topic of EOL care. However, 70% of subjects reported that they would be comfortable if their provider initiated discussion. Participants 60 years or older were more likely to report being very comfortable with their provider bringing up the subject compared with those younger than 60 years (60% vs 39%, P = .02). Few differences were found by age.

Room for improvement exists for increasing the number of patients who complete an AD or engage in discussion of their wishes. Ways to involve healthcare providers in the process should be explored, as it seems that patients are receptive to physician-initiated discussions of ADs.

Most research on advance directives (ADs) has focused on seniors or patients with terminal illness. The findings on this topic among adults of all ages highlight the following:

There is much room for improvement in increasing the number of patients who complete ADs or engage in discussion of wishes.

Ways to involve providers in the process should be explored, as it seems that patients of all ages are receptive to physician-initiated discussion of ADs.

Managed care organizations should build on their unique strengths, including multiple points for patient interaction and comprehensive documentation systems, to assist patients in advance care planning.

The media coverage surrounding the Terri Schiavo case

 highlighted the need to explore ways to encourage patients of all ages to think about and document end-of-life (EOL) care wishes. Before and after this publicized and controversial case, there was much discussion about how to increase the number of people who have an advance directive (AD), as well as how to best facilitate the process.

 However, much of the research on knowledge and attitudes related to ADs has been among seniors,

 or patients with critical or terminal illness.

 Although some work has been done among outpatients,

 information is limited about the association between age and knowledge or attitudes, as well as about patient preferences regarding initiation of EOL care discussions. We surveyed managed care organization (MCO) patients 20 years or older about knowledge, attitudes, experiences, and approaches to initiation of ADs.

Subjects included an age-stratified random sample of patients from a large MCO with approximately 700,000 members. The sample was drawn from patients who had obtained care in 1 of the plan’s 19 owned clinics and who were members as of July 2005. Subjects were identified via electronic administrative records and were then stratified by 7 age decades (20-29, 30-39, 40-49, 50-59, 60-69, 70-79, or ≥80 years). Patients were included if they had 3 years of continuous enrollment before identification, were 20 years or older, had a valid telephone number, and were not found on research exclusion lists. Targeted subjects in each age stratum were randomly selected (numbers were computer generated) from a pool of 58,569 who met eligibility criteria. The sampling plan was to stop surveying participants in a given age category when we reached 20 respondents. Although sex was not a primary variable in our analysis, approximately half of the randomized patients targeted for survey were female, and half were male. Because the survey included questions about whether the issue of ADs was raised by a healthcare provider, selection criteria included a minimum enrollment of at least 3 years. This study was approved by the MCO’s institutional review board, and participation served as a subject’s consent to use his or her data.

The survey included 54 items measuring patient knowledge of terms, experiences, and attitudes related to EOL care decision making and preferences toward initiation of advance care planning discussions (a copy of the survey is available from the corresponding author). Items considered to be possible correlates of having an AD other than age (ie, self-rated health, having experienced the prolonged death of someone close, having made EOL care decisions for someone, marital status, and education) were also included. Our selection of variables was based on an extensive literature review, followed by input from health plan leaders about factors to help the MCO encourage completion of an AD or discussion of wishes. Items from the experiences component were adapted from those used by Silveira et al31 in examining outpatients’ knowledge of their EOL options. Items measuring knowledge and preferences were created based on work done by others in these areas,

 as well as on experience of the study team (eg, expertise in hospice care and law). Items measuring attitudes were designed by Beck et al

 to examine attitudes and beliefs regarding life-sustaining treatment among senior MCO enrollees. The item measuring perceived health status was selected from the 

SF-12 Physical and Mental Health Summary Scales

 To ensure clarity and ease of administration among respondents of all ages, we pilot tested the survey with 14 subjects (1 man and 1 woman from each age stratum).

Subjects were mailed an invitation to participate in a 1-time telephone survey regarding EOL care wishes. All telephone surveys were conducted by a single interviewer (JMJ). At least 3 attempts were made to reach potential subjects via telephone. Initially, we planned to complete all surveys via telephone; however, we found it difficult to reach subjects, particularly those in the younger age groups (20-29 and 30-39 years). Therefore, after completing half of the targeted number of surveys by telephone, the remaining surveys were solicited by mail (mailed survey, followed by a reminder postcard 2 weeks later to nonresponders). The telephone survey was conducted between September 2005 and January 2006, and mail surveys were conducted from mid-March through May 2006.

Descriptive statistics were used to summarize variables overall and by age. Because this was an exploratory examination, numerous bivariate comparisons were made. Contingency tables (7 age categories by typically 2 or 3 categories for the other variables tested) and overall 

 tests (Pearson product moment correlation and Mantel-Haenszel test) for the entire table assessed differences in knowledge, experiences, attitudes, and preferences by age, marital status, and education. To simplify presentation, age groups were recoded as younger than 60 years and 60 years or older. Comparisons were made between experiences and the following possible correlates: self-rated health, having experienced the prolonged death of someone close, and having made EOL care decisions for someone. Comparisons between attitudes and these 3 variables were also made. To examine the effect of data collection mode, differences in experiences and knowledge were examined by whether the survey had been completed by telephone versus mail. For most items, missing values and “do not know” or “not sure” responses were removed for analysis purposes. However, because of the substantial proportion of subjects answering “do not know” for the attitude items, analysis of attitude statements retained uncertain responses. The results of all bivariate comparisons that were found to be statistically significant or suggestive are presented. In addition, unconditional logistic regression analysis was used to predict 2 experience variables (having completed an AD and having talked with someone about wishes) from age (<60 vs  ≥60 years, added in the first model of the analysis), self-rated health and education (added in the second model), having experienced the prolonged death of someone close, and having made EOL care decisions for someone (added in the third model). An additional analysis included these predictors along with all 2-way interactions with age to test for possible moderation.

Two hundred three members were initially selected as participants for the telephone survey. Thirty members could not be contacted because of disconnected telephone numbers or wrong addresses, and 6 members were unable to communicate via telephone. Of the remaining 167 members, 58 could not be reached after 3 attempts. Of 109 reached, 39 refused participation, and 70 surveys were completed, for a response rate of 64%.

Two hundred eighty-seven members were targeted for the mailed survey. Three persons were unreachable by mail, and 2 were unable to participate (non—English speaking or physically unable). Of 282 deemed eligible, 117 returned the survey, for a 41% response rate (4 declined participation, 2 returned surveys with too little information to be included, and 159 never responded). Although we intended to obtain 20 subjects per age stratum, for a total of 140 completed surveys, we kept all eligible surveys (N = 187) returned for analysis. We obtained at least 20 respondents in 5 of 7 age groups. However, for those aged 20 to 29 years, we received 14 completed surveys, and we received 17 for those aged 40 to 49 years. Patients younger than 50 years were less likely to respond compared with those 50 or older (26% vs 47%, P <.001). As an indicator of the precision of our findings, proportions computed on the totalof 187 had a 95% confidence interval of 0.43 to 0.06 (for a proportion of 0.50).

 gives a breakdown of characteristics of participants overall and by age (<60 vs  ≥60 years). Most participants were of white race/ethnicity. Approximately two-thirds were married, and 75% had at least some college or technical school education. Slightly more than half were female. There were slightly more men among the nonresponders compared with those who completed the survey (54% male and 46% female nonresponders vs 46% male and 54% female responders, 

 = .08). More than 90% reported their health as good or better. Subjects who completed the survey by telephone were more likely to report having talked with someone about their wishes (62% vs 35%, 

 = .005). No other differences were found related to mode of administration.

Subject awareness of terms related to EOL care planning was highly varied (

). When asked which term they most closely associate with provision of directions for EOL care, 66% indicated living will. Almost all subjects were familiar with their right to accept or refuse treatment (ie, along the full continuum of care), and 81% knew that they had the right to accept or refuse life-sustaining treatments. Almost two-thirds were aware that creating an EOL care document does not require a lawyer. Approximately 92% recognized that documented decisions could be changed, and 72% knew that more than 1 person may be named to speak for them. No differences in knowledge of AD terms or logistics were found related to age (<60 vs  ≥60 years).

Subjects thought that they and their families should have input regarding EOL care (

). More than one-third were unsure about their physician’s knowledge of their wishes. Conducting some surveys via telephone provided the opportunity to probe for further clarification when a “no” or “do not know” response was given. Those further questioned reported that they had not discussed the issue with their physician or that they did not have a regular care provider with whom they thought they could discuss such topics. No differences were found in attitudes by age (<60 vs ≥60 years).

Completion of an AD was reported by 32%, and 5% stated that they were in the process or were thinking about creating one. Completion increased incrementally with age (

 <.001). More than half of the subjects 60 years or older, 24% of subjects aged 50 to 59 years, 12% of subjects aged 40 to 49 years, and 4% of subjects aged 30 to 39 years reported having an AD. No subjects aged 20 to 29 years had completed a document. This strong association between completion of an AD and older age was consistent regardless of self-rated health. Age continued to be a predictor of having an AD in the regression analysis, with those 60 years or older being more likely to have completed an AD compared with those younger than 60 years (

). The nested models given in Table 4 also indicate that the odds ratio for age changes little when other variables are included. No other variables in the equation were significant, and all 2-way interactions of other covariates with age were also not significant. Of those without an AD, 44% reported having talked with someone about their wishes. The proportion of subjects who had a discussion did not differ significantly by age. When the same regression models were used in predicting whether subjects had talked with someone about their wishes, no variables were significant.

Asked in what setting and with whom they would like to discuss ADs, 54% reported a preference of discussing decisions with family rather than a provider. More than one-third (38%) reported no preference between these 2 choices (no significant differences by age). Completion of AD forms via mail was preferred by 60%, while 51% reported that they would consider attending an educational session on ADs, and 45% reported that they would consider using telephone assistance to complete forms. Only about 40% of participants thought that they would complete AD forms if available on the Internet. Sixty-two percent reported that it was up to them (or the person for whom an AD was being created) to bring up the subject of EOL care wishes. However, 70% of respondents reported that they would be comfortable if their provider brought up the subject of an AD. Those 60 years or older were more likely to report being very comfortable with their provider bringing up the subject compared with those younger than 60 years (60% vs 39%, 

Correlates of Knowledge, Attitudes, Experiences, and Preferences

In addition to our main independent variable of interest (age), we examined differences in experiences and attitudesby self-reported health, having experienced the prolonged death of someone close, having made EOL care decisions for someone, marital status, and education. Directional differences were found, but none reached statistical significance.  Of subjects without an AD, those who reported better health status were slightly more likely to have talked with someone about their wishes (

). Subjects who perceived their health to be less than very good were more likely to answer “do not know” on several of the items measuring attitudes (

 <.05 for all). Having experienced the prolonged death of someone close was not significantly associated with creating a document or having a discussion. Subjects who had made EOL care decisions for someone were slightly more likely to have completed an AD.

More than half of the married respondents would consider attending an educational session on ADs (Table 5). Subjects with more education were more likely to have heard of the terms advance directive and healthcare directive. Subjects with a high school education were slightly more likely to report that more than 1 person can be named to speak on their behalf. There was a significant difference in the number of subjects with a high school education or less compared with those with more education who were unsure about whether their wishes should determine what life-sustaining treatment they receive. Subjects with more education were more likely to report that they would consider using telephone assistance in completing an AD and were more likely to report that they would be very likely to complete AD forms online. There were no differences found in experiences with ADs by marital status or education.

Adults 20 years or older were surveyed regarding advance directives (ADs); one third had completed an AD. Subjects were comfortable with provider initiation of the topic.

March 2009

Medicare Part D Coverage Gap and Diabetes Beneficiaries

Patient Factors Associated With Following a Relocated Primary Care Provider Among Older Adults

Knowledge, Attitudes, and Preferences Regarding Advance Directives Among Patients of a Managed Care Organization

Curtailing Laboratory Test Ordering in a Managed Care Setting Through Redesign of a Computerized Order Form

VHA Pharmacy Use in Veterans With Medicare Drug Coverage

Voice Response System to Measure Healthcare Costs:A STAR*D Report

The Vermedx® Diabetes Information System Reduces Healthcare Utilization

List of Peer Reviewers (2008)