Restricting Medicaid Payments for Transportation: Effects on Inner-City Patients’ Health Care

Restricting Medicaid Payments for Transportation: Effects on Inner-City Patients' Health Care WILLIAM M. TIERNEY, MD; LISA E. HARRIS, MD; DENISE L. GASKINS, BS; XIAO-HUA ZHOU, PHD; GEORGE J. ECKERT, MAS; ANN S. BATES, MD; FREDRIC D. WOLINSKY, PHD

ABSTRACT: Background: Prior approval programs have been used to control spiraling costs of Medicaid, but they are rarely formally assessed. We evaluated the effect of a change in Indiana Medicaid's policy (effective October 1, 1993) requiring prior approval to pay transportation costs. Methods: We performed a historical cohort study comparing the health care utilization of Medicaid patients during the first 6 months of 1993 versus the first 6 months of 1994. Subjects included all Medicaid patients who visited any inpatient or outpatient site affiliated with an inner-city public hospital in the first 6 months of 1993 (N = 23,015) and 1994 (N = 23,707). Results: These Medicaid patients made 82,961 visits in the first 6 months of 1993 and 79,809 visits in the first 6 months of 1994. Visits to hospital-based primary care clinics declined 16% (P < 0.001), which was partially offset by a 7% increase in visits to neigh-

borhood health centers (P :::5 0.001). Emergency and urgent visits fell by 8%; visits for medication refills fell by 18% (P < 0.001 for each). Hospitalizations increased slightly in 1994, with no change in the number of inpatient days. There was no change in inpatient or outpatient nontransportation charges. There were no systematic reductions in selected aspects of preventive care. However, there were fewer emergency and urgent visits among patients with reactive airway disease. Conclusions: Requiring prior approval for transportation was associated with reductions in visits for primary care visits and refilling prescriptions without measurable shortterm effects on charges or selected clinical parameters. Neighborhood health centers partially ameliorated the decline in primary care visits. KEY INDEXING TERMS: Medicaid; Health policy; Clinical epidemiology; Vulnerable populations [Am J Med Sci 2000;319(5):326-33.]

U

$150 billion in 1996, is the largest US social welfare program and creates hardships for many states because of its unfunded mandates. 2 By 1994, Medicaid costs consumed more than 20% of the average state's budget.3,4 In response, many states attempted to reorganize the delivery of Medicaid health services to lower their costs, yet the clinical and financial effects of such programs have rarely been evaluated. 2 ,5-7 In October 1993, Indiana Medicaid began limiting the number of health care visits for which transportation costs would be reimbursed without the physician and transportation carrier obtaining prior authorization. Using a comprehensive electronic medical record system serving a county-wide health care system that provides most of the care to the county's Medicaid patients, we assessed the effect of this intervention on the utilization of health services by Medicaid patients. Our a priori hypotheses were that limiting transportation payments would result in: 1) fewer outpatient visits, especially to primary care providers and for medication refills; 2) more emergency and urgent visits; 3) less preventive care;

S health care expenditures have reached 15% of gross domestic product, surpassing $1 trillion annually, and continue to expand faster than the US economy.1 Medicaid, which cost more than

From the Regenstrief Institute for Health Care (WMT, LEH, DLG, the Departments of Medicine (WMT, LEH, X-HZ, GJE, ABB) and Pediatrics (ABB), Wishard Memorial Hospital and The Indiana University School of Medicine, Indianapolis, Indiana; The Richard L. Roudebush Veterans Affairs Medical Center, Indianapolis, Indiana (WMT); and The School ofPublic Health, Saint Louis University, St. Louis, Missouri (FDW) Submitted May 20, 1999; accepted in revised form August 26, 1999. This work was supported by Grant PHB93-S1 from the Indiana State Board of Health and in part by Grants R01-HS07632, R01-HS07763, R01-HS07719, R01-HS08823, R01-HS09083 (all WMT, LEH), and R03-HS90217 (X-HZ) from the Agency for Health Care Policy and Research, and Grant R37-AG09692 (FDW) from the National Institute on Aging. LEH was also supported in part by a Scholar Award from the Picker-Commonwealth Program in Patient-Centered Care. Correspondence: William M. Tierney, M.D., Regenstrief Institute for Health Care, 6th Floor, Regenstrief Health Center, 1001 West Tenth Street, Indianapolis, IN 46202 (E-mail: [email protected]). X-HZ, ABB),

326

May 2000 Volume 319 Number 5

Tierney et al

and 4) more emergency department visits and hospitalizations for selected chronic conditions. Methods Approval for this study was obtained from Indiana University's institutional review board. The study site was Wishard Health Services, a comprehensive health care system that includes an urban public teaching hospital, and its inpatient, emergency, urgent, and outpatient services. Using data in the Regenstrief Medical Record System (described below), we first identified all patient visits to any inpatient or outpatient site between January 1 and June 30 of 1993 (inclusive) and the same dates of 1994 where the primary or secondary payer for the visit was listed as Indiana Medicaid. (The Regenstrief Medical Record System functioned in such a manner that Medicaid coverage was recorded for all visits during the period of time each patient had Medicaid coverage.) These 2 6-month periods were chosen because 1) the new Indiana Medicaid transportation reimbursement policy went into effect on October 1, 1993, and 2) comparing identical months in consecutive years controls for seasonal influences in the demand for health care. Prior studies 8 and several internal audits have shown that 95% of adult patients who kept at least one scheduled visit to Wishard Health Services or were hospitalized at Wishard Hospital received all of their care from this system. This is especially true for Medicaid patients, for whom Wishard Health Services is one of only 2 approved provider systems in Indianapolis. We included all Medicaid patients, regardless of age. The new Indiana Medicaid transportation reimbursement policy provided automatic approval for reimbursing 10 round trips per year. A rolling definition of a I-year period was used. That is, beginning October 1, 1993, when a patient requested transportation from one of the approved transporters, the round trip would be automatically reimbursed by Indiana Medicaid if the total number of paid round trips in the 365 days before the target visit was nine or fewer. If the number of paid round trips exceeded nine, prior authorization was required. It is important to note that when the policy went into effect, it counted trips that occurred prior to October 1, 1993. Obtaining prior approval required both the transporter and the physician to complete forms justifying the visit. That entire process usually took 6 weeks. Ambulance services were excluded from this policy. No other programmatic or policy changes in Indiana Medicaid occurred in 1993 or 1994. Similarly, there were no changes in the organization or delivery of care by Wishard Health Services during this time. During both study periods, and for years previously, the Wishard pharmacy rarely dispensed more than 2 months' supply of most medications, except in unusual instances when a larger amount was requested (and had to be justified) by the patient's physician. Therefore, just being compliant with daily medications required at least 6 round-trips per year. Data for this study were extracted from the Regenstrief Medical Record System, which has been described in detail elsewhere. 9 Briefly, it is a comprehensive inpatient and outpatient electronic medical record system that performs all data processing for all Wishard Health Services whose outpatient facilities include a single large, hospital-based multi specialty center that is adjacent to the hospital. At the time of this study, this center contained more than 65 clinics and physicians' practices, including the system's largest pediatric and adult general internal medicine practices. There were also 8 satellite primary care neighborhood health centers.l0 All hospital-based clinics and the neighborhood health centers have on-site pharmacies and exclusively use Wishard Hospital's diagnostic testing services. All physicians practicing at all sites are faculty or house staff employed full-time by the Indiana University School of Medicine. The Regenstrief Medical Record System maintains a permanent record for all patients visiting any site. 9 Patient records are never deleted, even those of inactive or deceased patients. We THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

extracted the following data from patients' electronic records for each clinical encounter during the 2 6-month study periods: the date and site of the visit, the type of visit (e.g., scheduled, unscheduled, urgent, emergent, or only for a medication refill), the primary billing diagnosis (stored as both clinical diagnoses in patients' outpatient problem lists and as ICD-9-CM codes l l ), and the charges for facilities, drugs dispensed, procedures, and diagnostic tests. We also extracted the primary discharge diagnoses, lengths of stay, and charges for all inpatient stays. From registration data, we extracted each patient's sex, race, and age as of their first visit in 1993 and/or 1994. Because the hospital also operates the county ambulance service, those charges were also included. Charges for transportation provided by independent transporters, however, were not .available. Actual charges for physicians' services were also not available because each specialty uses a separate billing company. However, we estimated physicians' charges for inpatient and outpatient visits and procedures using the International Classification of Disease, 9th Revision, Clinical Modification (lCD-9-CM)11 and Physicians' Current Procedural Terminology (CPT4)12 procedure codes to assign relative value units (RVUs) to each visit and procedure.1 3 •14 We then multiplied the RVUs by the regional allowable charge for 1993. However, we had no information on where in the hospital each patient received inpatient care. Therefore, we could not assign different RVUs for physician visits to patients on different wards (eg, intensive care versus medical-surgical wards). We therefore applied a constant daily physician visit charge to each inpatient day and added physicians' charges for any procedures performed. Because Wishard Health Services increased charges for each revenue center by 5% in 1994, we divided 1994 charges by 1.05 to analyze all charges as constant 1993 dollars. Charges, rather than costs, were analyzed inasmuch as this study takes the perspective of the payer (Indiana Medicaid), which was entirely fee-for-service at the time ofthis study (and reimbursed the allowable charges stored in the Regenstrief Medical Record System). We categorized charges as outpatient, inpatient, and total. Beginning in January 1994, Indiana Medicaid began allowing payment for mental health visits in cases in which a physician was not seen. This policy substantially increased the recording of these nonphysician outpatient mental health visits during only the second of our 2 study periods. Because the Regenstrief Medical Record System could not identify similar outpatient mental health visits in 1993 where physicians were not seen, all visits to mental health clinics were excluded from the analysis. To assess possible effects of the change in Indiana Medicaid's transportation payment policy on the provision of preventive care, we used the U.S. Preventive Services Task Force recommendations 15 to identify patients who were eligible for selected preventive care services, specifically cervical Papanicolaou smears, mammograms, and pneumococcal vaccinations. Wishard Health Services' primary care physicians routinely received computergenerated reminders to perform indicated preventive care procedures at the time of each patient's scheduled primary care visit. 9 ,16 We also assessed blood pressure control for all adult patients who had hypertension recorded in their active outpatient problem lists by calculating the mean of all systolic and diastolic blood pressure readings, routinely recorded during all primary care visits and visits to selected specialty clinics (eg, subspecialty medicine), in each study period. To assess the possible effects of the change in Indiana Medicaid's transportation reimbursement policy on clinical outcomes, we selected reactive airway disease as an example because its outcomes are likely to be sensitive to interruptions in care and it is both common and morbid among Wishard patients.1 7,18 We therefore identified all patients who, at the beginning of each study period, had evidence of reactive airway disease. Reactive airway disease was considered to be present in any patient who previously had the diagnosis of asthma, emphysema, or chronic obstructive pulmonary disease recorded in the problem list from any prior inpatient or outpatient encounter. We also included patients without such diagnoses who had a previous prescription for theophylline, oral or inhaled /3-adrenergic receptor agonists,

327

Restricting Medicaid Transportation Payments

Table 1. Patient Characteristics Patient Demographic Variables

January-June 1993

January-June 1994

Percent Change

Number of Medicaid patients Sex (% women) Race (% black) Age (years:!: S.D.) Age < 1 year" Age 1-4 Age 5-12 Age 13-17 Age 18-39 Age 40-64 Age 2:65

23,015 66 60 20.3 :!: 21.0 2507 (11) 5303 (23) 3051 (13) 1463 (6) 7052 (31) 2286 (10) 1353 (6)

23,707 66 61 20.9 :!: 21.2 2517 (11) 5199 (22) 3430 (15) 1483 (6) 7176 (30) 2482 (10) 1420 (6)

+2.9 0.0 +2.0* +2.9* -0.2 -1.2 +1.2 -0.1 -0.3 +0.6 +0.1

*P <

0.01. Age at the time of first visit during each 6-month study period. Figures are number (percentage of all subjects in that column) during that 6-month period. a

inhaled corticosteroids, or inhaled cromolyn. We then searched these patients' electronic medical records for all hospitalizations and visits to the emergency department or the urgent visit center during the first 6 months of 1993 and 1994 where the primary reason for the visit was reactive airway disease, respiratory failure, bronchospasm, or less severe respiratory conditions such as bronchitis. We also assessed these same patients' emergency care for nonrespiratory problems. We adjusted for severity of illness using Ambulatory Care Groups (ACGs).19 ACGs use outpatient ICD-9-CM codes to assign all of the 34 ambulatory diagnostic groups (ADGs) that apply to each visit. ADGs are grouped according to the age of the patient and whether the problem is acute or chronic, likely to recur, or psychosocial in nature. Despite our excluding visits to mental health clinics, psychosocial problems are common problems in primary care, and psychosocial ADGs would be expected to be well represented in this group. The ADGs for each patient were then combined within each 6-month study period, along with the patient's age, to assign a single ACG (of 51 possible) to each patient. ACGs have been shown to predict utilization and costs in a number of populations, including Medicaid patients. 2o We used Student's t-tests to compare the number of times Medicaid patients visited primary care sites, the emergency department or urgent visit center, the pharmacy for prescription refills, and all outpatient sites in the first 6 months of 1993 versus the first 6 months of 1994. We also used corrected ¥' tests to compare the 2 6-month study periods with respect to compliance with preventive care protocols (ie, whether the appropriate test or vaccination was given or not), ACG distributions, and other categorical variables. For all analyses, primary care visits were further divided into those to the hospital-based clinics versus the 8 neighborhood health centers. We repeated all of the above analyses for patients in the following subgroups based on age: infants « 1 year old), toddlers (1-4 years old), children (5-12 years old), adolescents (13-17 years old), young adults (18-39 years old), middle-aged adults (40-64 years old), and older adults (2:65 years old). A p-value of less than 0.05 was accepted as significant in all analyses except in those instances (noted in the Results section and tables) where we had a priori planned multiple hypothesis testing. In those cases we used Bonferroni's correction. 21 The data on total charges and subcategories of charges were highly skewed, which causes particular problems in data analysis. 22 One means of dealing with this skew is to use the logarithm of charges as the dependent variable. However, this causes problems in interpretation of results when, as occurred with this study, the intergroup differences for the raw charge results have the opposite sign ofthe intergroup differences in the 10garithms. 22 Additional problems occur when there are substantial numbers of

328

patients with zero charges, as in our subanalysis of inpatient charges. Therefore, for each patient, we took the ratio of the charges in 1994 divided by the charges in 1993 and, using a bootstrap method,23 assessed whether the logarithm of this ratio was significantly different from zero. In this instance, statistical significance was defined as the 95% confidence interval of the charge ratio in the bootstrap analyses excluded zero. Because the cost analysis took the perspective of Medicaid (the payer), we imputed a charge of $0 for patients with no visits under Medicaid during one of the 2 6-month study periods. Such patients may have actually had encounters with Wishard Health Services and intendant charges during an observation period during which they had no Medicaid charges, due to transient or permanent loss of Medicaid coverage. For each period, we identified "Medicaid visits" by the payer for that visit, so that visits that occurred after loss of Medicaid coverage were not included in this study.

Results

In the first 6 months of 1993, 23,015 patients made 82,961 inpatient or outpatient visits for which Indiana Medicaid insurance was listed as the payer. In the first 6 months of 1994, 23,707 Medicaid patients made 79,809 visits. Only 10,853 Medicaid patients (47% of each group) made Medicaid-covered visits to Wishard Health Services in both observation periods. As shown in Table 1, 1994 patients were slightly but significantly older and more often Mrican Americans than patients in the 1993 cohort. There were significant but small differences between the 2 cohorts in only 4 ofthe 51 ACGs (Table 2). Thus, there seemed to be no meaningful differences in demographic characteristics or severity of illness between the 2 6-month study periods. As shown in Table 3, Medicaid patients made 6.4% fewer total outpatient visits in the first 6 months of 1994 than during the same months in 1993 (P < 0.001), including 7.8% fewer primary care visits (P < 0.001). There was a 15.6% decline in visits to the hospital-based primary care clinics (P < 0.001); however, this was partially offset by the 7.1% increase in primary care visits to the neighborhood May 2000 Volume 319 Number 5

Tierney et 01

Table 2. Distribution of Ambulatory Care Groups in the Two Study Periods

Ambulatory Care Group 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 31. 30. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51.

Acute minor, age <1 year Acute minor, age 2-5 Acute minor, age 6+ Acute major Likely to recur, w/o allergies Likely to recur, w/ allergies Asthma Chronic medical, unstable Chronic medical, stable Chronic specialty Ophthalmological/dental Chronic specialty, unstable Psychosocial, w/o psychosocial major Psychosocial, w/ psychosocial major, w/o psychosocial minor Psychosocial, w/ psychosocial major, w/ psychosocial minor Preventive/administrative Pregnancy Acute minor and acute major Acute minor, likely to recur discrete, age <1 Acute minor, likely to recur discrete, age 2-5 Acute minor, likely to recur discrete, age >5, w/o allergy Acute minor, likely to recur discrete, age >5, w/ allergy Acute minor and chronic medical. Stable Acute minor and eye/dental Acute minor and psychosocial w/o psychosocial major Acute minor, psychosocial major, w/o psychosocial minor Acute minor, psychosocial w/ psychosocial major and minor Acute major and likely to recur discrete Acute minor/major, likely to recur discrete, age <2 Acute minor/major, likely to recur discrete, age >6-11 Acute minor/major, likely to recur discrete, age 2-5 Acute minor/major, likely to recur discrete, age >5, w/o allergy Acute minor/major, likely to recur discrete, age >5, w/ allergy Acute minorllikely to recur discrete/eye & dental Acute minorllikely to recur discrete/psychosocial Acute minor/major, likely to recur discrete/eye and dental Acute minor/major, likely to recur discrete/psychosocial 2-3 other ADG combinations, age <17 2-3 other ADG combinations, men, age 17-34 2-3 other ADG combinations, women, age 17-34 2-3 other ADG combinations, age >34 4-5 other ADG combinations, age <17 4-5 other ADG combinations, age 17-44 4-5 other ADG combinations, age >44 6-9 other ADG combinations, age <6 6-9 other ADG combinations, age 6-16 6-9 other ADG combinations, men, age 17-34 6-9 other ADG combinations, women, age 17-34 6-9 other ADG combinations, age >34 10+ other ADG combinations No ADGs assigned

January-June 1993

January-June 1994

713 (2.9) 1286 (5.3) 2348 (9.7) 2017 (8.3) 2490 (10.3) 35 (0.1) 171(0.7) 326 (1.3) 438 (1.8) 32 (0.1) 232 (1.0) 139 (0.6) 86 (0.4) 1446 (6.0) 41 (0.2) 3116 (12.9) 1653 (6.8) 668 (2.8) 534 (2.2) 530 (2.2) 374 (1.5) 9 (0.0) 129 (0.5) 35 (0.1) 20 (0.1) 165 (0.7) 3 (0.0) 341 (1.4) 166 (0.7) 33 (0.1) 133 (0.5) 99 (0.4) 1 (0.0) 12 (0.1) 68 (0.3) 36 (0.1) 34 (0.1) 571 (2.4) 85 (0.3) 1019 (4.3) 1137 (4.7) 154 (0.6) 268 (1.1) 385 (1.6) 17 (0.1) 7 (0.0) 8 (0.0) 21 (0.1) 163 (0.7) 7(0.0) 382 (1.6)

637 (2.5)* 1175(4.7) 2290 (9.2) 2231 (9.3)* 2707 (10.8) 54 (0.2) 182 (0.7) 378 (1.5) 457 (1.8) 26 (0.1) 302 (1.2) 137 (0.5) 90 (0.4) 1727 (6.9)* 44 (0.2) 3623 (14.5)* 1534 (6.1) 604 (2.4) 497 (2.0) 403 (1.6)* 334 (1.3) 11 (0.0) 104 (0.4) 53 (0.2) 18 (0.1) 190 (0.8) 4 (0.0) 359 (1.4) 141 (0.6) 37 (0.1) 91 (0.4) 88 (0.3) 2(0.0) 18 (0.7) 56 (0.2) 34 (0.1) 35 (0.1) 577 (2.3) 65 (0.3) 960 (3.8) 1318 (5.3) 136 (0.5) 236 (0.9) 345 (1.4) 13 (0.1) 7 (0.0) 5(0.0) 23 (0.0) 110 (0.4)* 13 (0.0) 431 (1. 7)

* P < 0.001 (ie, P < 0.05 with Bonferroni's correction for testing 51 hypotheses. ADG, ambulatory diagnostic group, assigned to visits using ICD-9-CM codes. 19

health centers (P = 0.01). Visits for medication refills decreased by 17.8% in the first 6 months of 1994, and emergency/urgent visits fell by 8.1% (P < 0.001 for each). Although there was a significant (P = 0.044) but modest increase of 8.2% in the number of patients hospitalized during the second 6-month study period, there was no difference in the THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

number of inpatient days (a 0.2% increase in 1994). Moreover, there was no difference in outpatient, inpatient, or total charges. Figure 1 shows that changes in the various types of outpatient visits were not evenly distributed across different age groups. Total visits decreased significantly for all groups except for infants less

329

Restricting Medicaid Transportation Payments

Table 3. Differences between January-June 1993 and 1994 for all study patients

Patients All visitsa Primary care visits Hospital-based primary care visits Neighborhood health center visits Emergency and urgent visits Medication refill visits Outpatient charges a Total Mean ± SD Median Patients hospitalized Inpatient days/patient Inpatient chargesa Total Mean ± SD Medianc Total charges a Total Mean ± SD Mediand

January-June 1993

J anuary-June 1994

Percent Change

23,015 82,974 (3.6 ± 4.3) 38,116 (1. 7 ± 2.4) 25,170 (1.1 ± 1.9) 12,946 (0.56 ± 1.7) 19,895 (0.86 ± 1.2) 10,256 (0.45 ± 1.4)

23,707 79,890 (3.4 ± 4.0) 36,177 (1.5 ± 2.2) 21,989 (0.9 ± 1.6) 14,279 (0.60 ± 1.7) 18,692 (0.79 ± 1.1) 8,850 (0.37 ± 1.2)

+3.0 -3.7 -6.4 -5.7 -7.8 -2.6 -15.6 +10 +7.1 -6.0 -8.1 -14 -17.8

$14,036,312 ($610 ± 1246) $222 2783 (5.9%) 1.21 ± 0.64

$14,448,477 ($609 ± 1398) $208 3012 (6.4%) 1.22 ± 0.63

+2.9 -0.1

$24,119,720 ($1048 ± 5878) $8057

$25,864,337 ($1091 ± 6081) $6885

+7.2 +4.1

$51,612,619 ($2243 ± 7870) $239

$53,574,635 ($2260 ± 8098) $220

+3.8 +0.8

+8.2 +0.2

P-value for 95% CI

<0.001 <0.001 <0.001 0.010 <0.001 <0.001

0.044 NS

NSe

Figures in parentheses are visits or charges ±SD per patient for all patients in each 6-month period. NS, not significant. a Includes zero charges for patients no hospitalized; does not include mental health visits (see text). b Ratio of log(1994 cost/1993 cost) is -0.024 (95% CI = -0.060 to 0.012). C Median only among those patients hospitalized. d Ratio oflog(1994 costl1993 cost) is 0.034 (-0.083 to 0.143). e Ratio of log(1994 costl1993 cost) is -0.003 (-0.071 to 0.064).

than 1 year old and children between 5 and 12 years old. Visits to hospital-based primary care clinics declined significantly for all adults more than 18 years old, whereas there was no pattern to the changes in visits to the primary care neighborhood health clinics, except for a significant increase among infants less than 1 year old. Emergency and urgent visits decreased significantly for all patients less than 40 years old. Medication refills trended downward for all groups except children less than 1 year old, but the decrease was statistically significant only for patients more than 40 years old. Provision of selected preventive care services during the 2 6-month study periods are shown in Table 4. There was no change in the proportion of eligible women who received cervical Papanicolaou smears or mammograms, but there was a significant (P = 0.008) but small decrease in the percent of eligible patients reCeIvmg pneumococcal vaccinations (which at the time of this study were available only in physicians' offices). Among patients with hypertension making visits during the 2 study periods, the

330

mean systolic blood pressure was significantly higher in 1994 and the mean diastolic blood pressure was significantly lower in 1994. However, these differences (less than 1 mm Hg) have dubious clinical significance. Among patients with reactive airway disease, there was a statistically and clinically significant decline in the number of visits to the emergency department and the urgent visit center (P < 0.001) for all diagnoses and for lower-respiratory problems. There was also a 15% decline in the number of emergency/urgent visits specifically for acute reactive airway disease, but this difference did not reach statistical significance (P = 0.14). Discussion

Mter implementation of a Medicaid prior approval program for payment of transportation costs, there was a clinically and statistically significant reduction in primary care visits. This effect, however, was driven mainly by a reduction in primary care visits to the hospital-based health center located in downMay 2000 Volume 319 Number 5

Tierney et 01

• <1 year old .1-4 years .5-12 years • 13-17 years • 18-39 years • 40-64 years .65+ years

All visits

Primary care

Hospital clinics

•••

I~;t

Neighborhood clinics

Emergency/urgent

Medication refill -30 -20 -10

0

10

20

30

Percent Change in Visits Between the Two Six-Month Study Periods, by Age Group

• Significantly different from zero (p<0.05 with Bonferroni's correction for multiple hypothesis testing)

Figure 1. Percent change in visits per patient by age groups: from top to bottom in each cluster of bars: <1 year old, 1-4 years old, 5-12 years old, 13-17 years old, 18 -39 years old, 40 - 64 years old, and '2: 65 years old.

town Indianapolis. A concomitant increase in primary care visits to satellite neighborhood health centers, located closer to patients' homes, partially offset the overall reduction in primary care visits. This makes sense: when transportation is restricted, it should affect trips to centralized health care venues and encourage patients to seek venues closer to

home. This ameliorating effect of the neighborhood health centers is consistent with previous studies; for example, the availability of satellite community mental health centers resulted in a reduction in inpatient psychiatric care. 24 Despite the significant reduction in outpatient visits in general, and primary care visits in particular, there was no concomitant increase in emergency or urgent visits. This is somewhat surprising because ambulance services were specifically exempted from Indiana Medicaid's new transportation prior approval policy. Instead, visits to the emergency department and urgent visit center (located adjacent to the emergency department) actually declined. Among patients with reactive airway disease, the decline was primarily for conditions that may not have required emergency or urgent care. Prior studies of the relationship to barriers to care and emergency services have been mixed. In one study,25 a major barrier to the receipt of care by Medicaid patients in physicians' offices (ie, low reimbursement) had no effect on the use of hospital outpatient services. However, barriers to the receipt of routine outpatient care have resulted in greater use of hospital outpatient services. 26 Although reductions in the number of Medicaid visits occurred between the 2 6-month study periods, there was no change in charges for care. This suggests that more care was compressed into the visits that did occur. Because our system captures charges for ambulance rides (which were excluded from the change in transportation policy) but no other transportation, the effect of the new policy on transpor-

Table 4. Selected Preventive Care for Eligible Patients, Blood Pressure Control in Hypertensive Patients, and Emergency and Urgent Visits for Patients with Reactive Airway Disease

Mammography Eligible active patients No. (%) receiving mammogram Cervical Papanicolaou smears Eligible active patients No. (%) receiving cytology smear Pneumococcal vaccine Eligible active patients No. (%) receiving vaccine Blood pressure control Number of active hypertensive patients Mean systolic pressure (mm Hg) Mean diastolic pressure (mm Hg) Reactive airway disease Number of active patients with reactive airways disease Emergency and urgent visits per patient Emergency and urgent visits for lower respiratory problems Emergency and urgent visits for reactive airways disease

January-June 1993

January-June 1994

Percent Change

P-Value

1691 369 (21.8)

1423 341 (24.0)

+2.1

0.16

5789 880 (15.2)

5522 818 (14.8)

-0.4

0.56

2417 90 (3.7)

2342 56 (2.4)

-0.3

0.008

7571 114.1 ::+: 21 68.0::+: 13

8324 115.0::+: 22 67.4 ::+: 12

+0.7 -0.9

0.01 0.003

3277 1.1 ::+: 1.4 0.12 ::+: 0.39

3086 1.0 ::+: 1.3 0.09 ::+: 0.34

-10 -27

<0.001 <0.001

0.07::+: 0.23

0.06::+: 0.27

-15

0.14

Except where indicated, figures are mean per patient ::+:SD. THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

331

Restricting Medicaid Transportation Payments

tation costs could not be determined. It is likely, however, that transportation costs were lower after the change in policy, given the changes we observed in the number of outpatient visits. Despite the meaningful reduction in primary care visits after the new transportation policy was implemented, we detected no deleterious short-term effects on selected clinical outcomes (ie, blood pressure control among hypertensive patients or acute exacerbations among patients with reactive airways disease). This is somewhat reassuring, but it may be misleading. The effect of such policy changes on clinical outcomes may not manifest injust 6 months. Support for this caveat comes from the substantial decrease we observed in visits for refilling medications and the number of eligible patients receiving pneumococcal vaccinations. Prior research on the effect of logistic barriers to care among Medicaid patients has yielded mixed results. In a comprehensive literature review, Soumerai et al 27 found that prescription limits, set by some states to reduce unnecessary polypharmacy, reduced the consumption of both essential and nonessential medications. This paradoxically resulted in higher costs 28 and more health care utilization, especially by older, chronically ill patients. Nevertheless, targeted interventions can reduce costs with little risk to patients. For example, when Tennessee Medicaid instituted a prior-authorization program for expensive nonsteroidal anti-inflammatory drugs, costs for those drugs dropped precipitously, mostly because of the substitution of less costly generic nonsteroidal anti-inflammatory drugs. 29 Our study benefited from having an electronic medical record system that captured comprehensive patient data for more than 160,000 visits for almost 36,000 separate patients. Even so, this study may not have had sufficient power or detailed data to detect important effects on particularly vulnerable persons, such as those with multiple chronic diseases, disabling conditions, or inadequate social support. However, our main results should not be taken as definitive evidence that the change in Indiana Medicaid's transportation reimbursement policy had no adverse patient effects. In this regard, the reductions in primary and preventive care as well as visits for refilling prescriptions should be viewed as potentially ominous harbingers of adverse events yet to come. This study has several limitations. It represents the experience of one health care system and does not capture out-of-system care. However, Wishard Health' Services is the largest Medicaid provider in central Indiana and the primary "safety net" provider. Raising barriers to transportation is not likely to preferentially divert care from this system into other providers of care. Because Medicaid roles were not available, it was also not possible for us to

332

discriminate between patients who had lost their Medicaid coverage and those who had Medicaid coverage but did not present for care during one of the study periods. However, it is unlikely that, during 1993 and 1994, there were different rates of losing Medicaid coverage. The similarity between the number of patients in each cohort offers further reassurance that such an effect, if extant at all, is small. Finally, this was not a randomized, controlled trial, which is the only study design capable of controlling for cointerventions or coincidental environmental changes that could have confounded the results we observed. Unfortunately, almost all health policy interventions are not performed as concurrently controlled trials and, most disturbingly, the effects of such interventions on the processes and outcomes of care are almost never assessed. Although studies such as ours are not proof of the effects of such interventions, they do show what occurred after such policy changes and reinforce the notion that policy decisions can have broad-and sometimes unintended-effects. Future administrative and legislative initiatives to lower the costs of Medicaid, Medicare, or other entitlement programs should therefore be carefully scrutinized. Evaluating such "uncontrolled experiments" may require analysis of claims data, primary data collection (which is expensive), and (where available) the analysis of electronic medical records. For example, we have previously used the Regenstrief Medical Record System to document a reduction in the quality of inpatient care of patients with hip fractures after the institution of Medicare prospective payments. 30 Such assessments are necessary if patients, especially those who are the most vulnerable, are to be protected from unintended negative clinical effects of cost-containment efforts. References 1. Burner ST, Waldo DR. National health expenditure projections. 1994-2005. Health Care Finane Rev 1995;16:221-42. 2. Riley T Medicaid. The role of the states, JAMA 1995;274: 267-70. 3. Holahan J, Coughlin T, Ku L, et al. Insuring the poor through Section 1115 Medicaid waivers. Health Aff (Millwood) 1995;14:199-216. 4. Kelly L, Melcher RA. Power to the states. Business Week 1995; Aug 7: 48-54. 5. Freund DA, Rossiter LF, Fox PD, et al. Evaluation of the Medicaid competition demonstrations. Health Care Finane Rev 1989;11:81-97. 6. Stuart ME, Weinrich M. Beyond managing Medicaid costs: restructuring care. Milbank Q 1998;76:251-80. 7. McCall N, Wrightson CW, Paringer L, et al. Managed Medicaid cost savings: the Arizona experience. Health Aff (Millwood) 1994;13:234-45. 8. Smith DM, Weinberger M, Katz BP, et al. Post-discharge care and readmissions. Med Care 1988;26:699-708. 9. McDonald CJ, Tierney WM, Martin DK, et al. The Regenstrief Medical Record System: 20 years' experience in hospital outpatient clinics and neighborhood health centers. MD Comput 1992;9:206-17. May 2000 Volume 319 Number 5

Tierney et 01

10. Weinberger M, Darnell JC, Tierney WM, et al. Self-rated health as a predictor of hospital admissions and nursing home placement in elderly public housing tenants. Am J Public Health 1986;76:457-459. 11. International classification of disease, 9th revision, clinical modification. Washington (DC): Department of Health and Human Services; 1980. 12. CPT 98. Physicians' current procedural terminology. Chicago: American Medical Association; 1998. 13. Relative Value Studies, Inc. The McGraw-Hill complete RBRVS. New York: McGraw-Hill; 1996. 14. Medicare program; revisions to payment policies and adjustments to the relative value units under the physician fee schedule for calendar year 1996. Fed Reg 1995;60:63124-357. 15. US Preventive Services Task Force. A guide to clinical preventive services. 2nd ed. Washington (DC): US Department of Health and Human Services; 1996. 16. McDonald CJ. Action-oriented decisions in ambulatory medicine. Chicago: Yearbook Medical Publishers; 1981. 17. Murray MD, Stang P, Tierney WM. Health care utilization by inner-city patients with asthma. J Clin Epidemiol 1997; 50:167-74. 18. Tierney WM, Takesue BY, Vargo DL, et aI. Using electronic medical records to predict mortality in primary care patients with heart disease: prognostic power and pathophysiologic implications. J Gen Intern Med 1996;11:83-9l. 19. Starfield B, Weiner J, Mumford L, et at Ambulatory care groups. A categorization of diagnoses for research and management. Health Serv Res 1991;26:53-74. 20. Weiner JP, Starfield BH, Steinwachs DM, et at Development and application of a population-oriented measure of ambulatory care case-mix. Med Care 1991;29:452-72. 21. Benjamini Y, Hochberg Y. Controlling the false discovery

THE AMERICAN JOURNAL OF THE MEDICAL SCIENCES

22. 23. 24. 25. 26. 27. 28.

29.

30.

rate: a practical and powerful approach to multiple testing. J R Stat Soc 1995;57:289-300. Zhou XII, Gao S, Hui SL. Methods for comparing the means oftwo independent log-normal populations. Biometrics 1997; 53:1129-35. Efron B, Tibshirani RJ. An introduction to the -bootstrap. New York: Wiley; 1993. Gronfein W. Incentives and intentions in mental health policy: a comparison of the Medicaid and community mental health programs. J Health Soc Behav 1985;26:192-206. Fossett JW, Choi CH, Peterson JA. Hospital outpatient services and Medicaid patients' access to care. Med Care 1991;29:964-76. Cohen JW. Medicaid policy and the substitution of hospital outpatient care for physician care. Health Serv Res 1991;24: 33-66. Soumerai SB, Ross-Degnan D, Fortress EE, et al. A critical analysis of state drug reimbursement policies: research in need of discipline. Milbank Q 1993;71:217-52. Soumerai SB, McLaughlin TJ, Ross-Degnan D, et al. Effects oflimiting Medicaid drug-reimbursement benefits on the use of psychotropic agents and acute mental health services by patients with schizophrenia. N Engl J Med 1994;331: 650-66. Smalley WE, Griffin MR, Fought RL, et at Effect of a prior-authorization requirement on the use of nonsteroidal antiinfiamatory drugs by Medicaid patients. N Engl J Med 1995;332:1612-7. Fitzgerald JF, Fagan LF, Tierney WM, et at Changing patterns of hip fracture care before and after implementation of the prospective payment system. JAMA 1987;258:218221.

333