Currently Viewing:
The American Journal of Managed Care February 2011
Effect of Multiple Chronic Conditions Among Working-Age Adults
James M. Naessens, ScD; Robert J. Stroebel, MD; Dawn M. Finnie, MPA; Nilay D. Shah, PhD; Amy E. Wagie, BA; William J. Litchy, MD; Patrick J. F. Killinger, MA; Thomas J. D. O'Byrne, BS; Douglas L. Wood, MD; and Robert E. Nesse, MD
Psychological Family Intervention for Poorly Controlled Type 2 Diabetes
Karen M. Keogh, PhD; Susan M. Smith, MD; Patricia White, PhD; Sinead McGilloway, PhD; Alan Kelly, PhD; James Gibney, MD; and Tom O'Dowd, MD
Abolishing Coinsurance for Oral Antihyperglycemic Agents: Effects on Social Insurance Budgets
Kostas Athanasakis, MSc; Anastasis G. Skroumpelos, MSc; Vassiliki Tsiantou, MSc; Katerina Milona, MSc; and John Kyriopoulos, PhD
Effects of Nonadherence With Prescription Drugs Among Older Adults
Richard J. Butler, PhD; Taylor K. Davis, BA; William G. Johnson, PhD; and Harold H. Gardner, MD
Behavioral Health Disorders and Adherence to Measures of Diabetes Care Quality
Gary Y. Leung, PhD; Jianying Zhang, MD, MPH; Wen-Chieh Lin, PhD; and Robin E. Clark, PhD
Timing of Follow-up After Abnormal Screening and Diagnostic Mammograms
Karen J. Wernli, PhD; Erin J. Aiello Bowles, MPH; Sebastien Haneuse, PhD; Joanne G. Elmore, MD, MPH; and Diana S.M. Buist, PhD, MPH
Excess Hospitalization Days in an Academic Medical Center: Perceptions of Hospitalists and Discharge Planners
Christopher S. Kim, MD, MBA; Anita L. Hart, MD; Robert F. Paretti, MD; Latoya Kuhn, MPH; Ann E. Dowling, BSN, RN; Judy L. Benkeser, BSN, RN; and David A. Spahlinger, MD
Health Insurance in India: Need for Managed Care Expertise
Thomas K. Thomas, MBA
Currently Reading
Outpatient Wait Time and Diabetes Care Quality Improvement
Julia C. Prentice, PhD; B. Graeme Fincke, MD; Donald R. Miller, ScD; and Steven D. Pizer, PhD

Outpatient Wait Time and Diabetes Care Quality Improvement

Julia C. Prentice, PhD; B. Graeme Fincke, MD; Donald R. Miller, ScD; and Steven D. Pizer, PhD
Longer primary care wait times were associated with small increases in glycated hemoglobin levels. Diabetes care quality improvement strategies should consider timely access to care.

Objective: To examine the relationship between glycated hemoglobin (A1C) levels and the number of days spent waiting for primary care appointments.

 

Study Design: Retrospective observational study that relied on Department of Veterans Affairs (VA) utilization data and Medicare claims data from 2001 to 2003. The outcome was A1C levels. The main explanatory variable of interest was facility-level primary care wait times measured in days.

 

Methods: Heckman selection models simultaneously predicted the presence of an A1C value and its level. Models were risk adjusted for prior individual health status. Separate models were estimated on the entire sample and on subsamples stratified by baseline A1C levels.

 

Results: Veterans who visited VA facilities with wait times of longer than 32.5 days had small significant increases in A1C levels of 0.14 percentage point for the whole sample, 0.07 percentage point for patients with baseline A1C levels less than 7%, 0.11 percentage point for patients with baseline A1C levels between 7% and 8%, and 0.18 percentage point for patients with baseline A1C levels greater than 8%.

 

Conclusions: Decreasing wait times has the potential to reduce A1C levels by 0.18 percentage point for patients with baseline A1C levels exceeding 8%. This effect is roughly one-third of what is achieved with the most successful existing quality improvement strategies. Ensuring timely access to outpatient care could be an important addition to future diabetes care quality improvement programs.

 

(Am J Manag Care. 2011;17(2):e43-e54)

Diabetes care quality improvement strategies have had limited success in lowering glycated hemoglobin (A1C) levels, with annual decreases of 0.50 percentage point in A1C levels considered successful. An overlooked element in quality improvement is timely access to healthcare.

 

  • This study finds that longer wait times for primary care are associated with small increases in A1C levels (0.18 percentage point for patients with poor baseline control).

 

  • The size of the effect from wait times is roughly one-third of the effect found for the most successful quality improvement strategies.

 

  • Healthcare managers should consider adding to quality improvement programs tested interventions that ensure timely access to care.
Diabetes mellitus is a prevalent costly illness that significantly increases the risk of poor health outcomes and increased healthcare costs.1-3 An essential focus of diabetes management is maintaining glycemic control.4 Despite accepted evidencebased guidelines and consensus in the medical community about appropriate processes of care, many patients with diabetes do not have optimal glycemic control.5-7

The uneven implementation of processes of care in diabetes management has led to numerous quality improvement (QI) interventions. These interventions can target patients (eg, patient education) and providers (eg, physician reminders) or can reorganize the delivery of healthcare services (eg, case management use of pharmacists to adjust medication).3,5-8 Evaluations of these interventions have found that they succeed at improving the process of diabetes care (eg, measuring blood glucose levels) but that the effect of these interventions on glycated hemoglobin (A1C) levels is limited.3,5-8

A key to glycemic control that may be overlooked in current diabetes QI programs is the timeliness of access to care. Successful management of diabetes requires patients to consistently access care in a timely manner. For example, the progressive nature of diabetes means that glycemic control will likely deteriorate over time and that changes in glucose-lowering medication will be needed. Similarly, acute illnesses may destabilize a patient’s glycemic control.4,9 Therefore, patients who consistently have longer wait times for appointments may have poorer glycemic control and outcomes compared with patients who have better access to healthcare services.

This study is the first to date to examine the relationship between wait time for primary care services and glycemic control in a nationwide sample by capitalizing on unique data available through the Department of Veterans Affairs (VA). Distinct from other health payers and healthcare systems, the VA automatically monitors wait times for appointments nationwide. Geographic variations in the demand for VA care or differences between VA facilities in the implementation of policies and programs aimed at managing excess  demand result in varying wait times across VA facilities.10-12 We exploited this variation in wait times to examine the effect of wait time for healthcare services on glycemic control   and hypothesize that patients who visit VA facilities with longer wait times may have poorer glycemic control.

METHODS

Study Population

This study used administrative data from the VA and from Medicare. The sample was extracted from the VA Diabetes Epidemiology Cohorts database, a registry of all VA patients nationwide with diabetes mellitus since 1998. The criteria for identifying patients with diabetes included a prescription for diabetes medication in the current year or at least 2  diabetes codes for inpatient or outpatient visits (VA and Medicare) over a 2-year period.13

Our sample consisted of veterans diagnosed as having diabetes before 2001. Outcomes were measured in 2002 and 2003, with 2001 used as a baseline year for risk adjustment (see “Data Availability” section in Appendix A). American Diabetes Association14 guidelines at the time of the study called for A1C levels to be tested biannually for individuals with glycemic control and quarterly for individuals without glycemic control. We split the outcome period into the following 4 six-month observation periods: (1) the first half of 2002, (2) the second half of 2002, (3) the first half of 2003, and (4) the second half of 2003.

To ensure complete utilization, laboratory, and diagnosis data, we required the sample to be VA users and Medicare eligible. The most important criterion for entering the sample was that individuals relied exclusively on the VA for outpatient physician visits in 2001. Following methods developed by Gardner et al, we counted outpatient physician visits in comparable ways among the VA utilization and Medicare claims data.15 Veterans who had VA outpatient physician visits and no Medicare physician visits in the baseline year were eligible for the sample. We confirmed that these individuals continued to rely heavily on the VA during the outcome period (Table 1). This was important because we needed complete A1C laboratory values, and Medicare claims do not provide laboratory values.16 A second criterion for entering the sample was whether we had complete claims data for baseline risk adjustment. Ineligible individuals included those without baseline A1C levels for stratification or those enrolled in Medicare health maintenance organizations in 2001 (incomplete diagnosis codes for risk adjustment).

We further excluded individuals who died during the outcome period (because of changes in A1C levels right before death), individuals with missing race/ethnicity data, and individuals in a hospital or nursing home during the entire outcome period. The final sample size totaled 84,244 individuals (Table 2).

Wait Time Measure

The main explanatory variable of interest was the mean wait time until the next available primary care appointment for patients new to the primary care clinic at a VA facility. This was an overall measure of congestion at the facility, as VA managers assume that almost all new patients want the next available appointment.17

Previous studies18,19 demonstrated that wait times based on services that an individual actually used are inappropriate for studying the effect of wait times on outcomes. Unobserved individual health status affects individual wait times and outcomes because sicker patients are triaged to receive care faster.

Because we cannot use the individual wait time, we use a proxy measure. Our measure is a facility-level mean wait time that is exogenous to the individual and is similar to a measure used in previous research.18,19 Each individual’s zip code was obtained from the Medicare denominator file in 2002 and 2003 and was used to assign him or her to the nearest VA facility. All individuals who were closest to the same VA facility were assigned the same exogenous wait time corresponding to that facility. The limitations of relying on this proxy wait time measure, including the possibility of biasing results toward the null, are considered in the “Discussion” section.

Facility and Seasonal Fixed Effects

A key aspect of our study design was to include dummy variables (fixed effects) for each facility in the models. The facility fixed-effects specification removed between-facility variation in wait times and outcomes, alleviating concerns about case-mix differences across facilities and controlling for time-invariant facility quality.

The remaining variation in wait times and outcomes came from within-facility changes over time, so individual patients served as their own controls. In effect, we compared the A1C level of an individual in one observation period with the A1C level of the same individual in other observation periods, which eliminated concerns about case-mix differences between facilities.

Similarly, facility fixed effects control for all aspects of facility quality that remain constant over time. For example, facilities with managerial inefficiencies may provide poor quality of care and have longer wait times. The poorer quality of care also increases the risk of poorer health outcomes independent of wait times.18 Facility fixed effects eliminated this potential confounding.20,21

Finally, we included dummy variables to control for the seasonality of A1C levels and any overall increase or decrease in A1C levels over time. This statistical design, featuring a predetermined cohort of patients and time and facility fixed effects, means that any estimated relationship between wait time and A1C level was identified exclusively by within-facility variations over time that were independent of national trends.

Risk Adjustment and Other Explanatory Variables

Models were risk adjusted to control for observable differences in prior individual health status. Explanatory variables included age, sex, race/ethnicity, and diagnosis-based variables. Other explanatory variables included distance to VA care and VA priority status (1-3 vs >3). Longer driving distances have been found to be associated with poorer glycemic control,22 and veterans with a service-connected disability (priority status 1-3) receive priority access and likely experience shorter wait times.

We extracted diagnosis codes from all inpatient and outpatient encounters financed by the VA and Medicare during the 2001 baseline period and used the International  

Classification of Diseases, Ninth Revision, Clinical Modification
(ICD-9-CM) diagnosis codes listed by Elixhauser et al23 to define 28 comorbidity indicator variables, which included a wide variety of physical and mental conditions.

The Diabetes Complications Severity Index developed by Young et al24 was used to control for diabetes severity. This index included measures of complications from retinopathy, nephropathy, neuropathy, cerebrovascular disease, cardiovascular disease, peripheral vascular disease, and metabolic disease. The ICD-9-CM codes from VA and Medicare  outpatient and inpatient utilization and claims files for 2001 were used to determine each individual’s Diabetes Complications Severity Index.

Outcomes and Analyses

Data were analyzed using commercially available statistical software (STATA 10.0; StataCorp LP, College Station, TX). We modeled the mean 6-month A1C level during each observation period. The mean wait time for the previous 6 months predicted the A1C level in the current 6-month period. Therefore, each individual contributed up to 3 outcome observations because his or her first 6-month period had no lagged data available. Standard errors were clustered on individuals to account for the lack of independence between observations from the same individual.

Patients in a hospital or nursing home during the wait time measurement period should not be affected by outpatient wait times. Therefore, we censored observation periods if the veteran was in a hospital or nursing home for all 6 months during the wait time measurement period.

 
Copyright AJMC 2006-2018 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
x
Welcome the the new and improved AJMC.com, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up
×

Sign In

Not a member? Sign up now!