Missed or Delayed Medical Care Appointments by Older Users of Nonemergency Medical Transportation

Abstract

Purpose of the Study: This study identified factors associated with canceling nonemergency medical transportation appointments among older adult Medicaid patients. Design and Methods: Data from 125,913 trips for 2,913 Delaware clients were examined. Mediation analyses, as well as, multivariate logistic regressions were conducted. Results: Over half of canceled trips were attributed to client reasons (e.g., no show, refusal). Client characteristics (e.g., race, sex, functional status) were associated with cancelations; however, these differed based on the cancelation reason. Regularly scheduled trips were less likely to be canceled. Implications: The evolving American health care system may increase service availability. Additional policies can improve service accessibility and overcome utilization barriers.

Background

Transportation mobility among older adults is critical for accessing health care services and for aging-in-place. There is a growing concern that older adults will become increasingly transportation disadvantaged as the sociodemographic composition of the United States continues to change. Health care policies and other programs are critical for alleviating transportation barriers for older adults. As such, this study was funded by Federal Transit Administration with cooperative agreements with Easter Seals Project ACTION and the National Center on Senior Transportation to better understand the relationship between health and transportation by evaluating existing research to: (1) evaluate how Medicaid nonemergency medical transportation (NEMT) is being used; and (2) document these findings for future research, policy, and practice.

Transportation barriers have been associated with poorer health care access, differential treatment, missed or delayed medical appointments, and lower health status (Fitzpatrick, Powe, Cooper, Ives, & Robbins, 2004; Goodwin, Hunt, & Samet, 1993; Gwira et al., 2006; Horton & Johnson, 2010; Okoro, Strine, Young, Balluz, & Mokdad, 2005; Rittner & Kirk, 1995). Transportation has been recognized as a key element of health care access, which highlights its role to offset long-term health consequences. Delayed medical care may adversely affect patients’ quality of life and result in more depressive symptoms (Owsley et al., 2007). These delays can also lead to an increased need for emergency care and preventable hospitalizations and result in higher medical costs (Hughes-Cromwich & Wallace, 2006; Rust et al., 2008).

Approximately 3.6 million adults miss or delay nonemergency medical care due to transportation barriers (Hughes-Cromwich & Wallace, 2006). The population missing these visits may be particularly susceptible to poorer health status and higher cost of care. These transportation barriers are more likely to influence older individuals, females, minorities, and those with lower income levels (Hughes-Cromwich & Wallace, 2006; U.S. Government Accountability Office, 2003). Medicaid acknowledges the importance of transportation for vulnerable populations and the associated potential costs of delayed medical care (Kim, Norton, & Stearns, 2009). In fact, federal Medicaid requires that states “ensure necessary transportation for recipients to and from providers” to help meet the needs of this population and the intent of Medicaid (The Hilltop Institute, 2008). It is important to consider how we can improve health care access and utilization to reduce health disparities. These types of transportation services may improve healthy aging and allow for aging-in-place.

With health care reform through the Patient Protection and Affordable Care Act, Medicare and Medicaid recipients and new enrollees will expand and could result in a large increase in the number of NEMT riders (Community Transportation Association of America, 2010). To effectively and efficiently plan for future demand, it is important to consider the particular needs of certain populations and how they respond to such programs. Older adults are one vulnerable group that may need particular attention.

Older adults are disproportionally disadvantaged by transportation barriers, in part, because driving frequency declines with age (Ragland, Satariano, & MacLeod, 2004). Although this type of barrier to services is well documented (Leutz & Capitman, 2007; Whittier, Scharlach, & Dal Santo, 2005), transportation concerns among older adults are rising as this population segment is projected to grow from over 40 million in 2010 to over 88 million by the year 2050 (U.S. Census Bureau, 2012). The growth of the aging population will increase transportation service demands as many older adults do not have alternative transportation support (Choi, Adams, & Kahana, 2012). Further, many older adults experience age-related changes and impairments that may be a barrier to health care. For example, cognitive, physical, and sensory function often declines with age. Missed visits may cause disease progression and mismanagement and limits communication with physicians, which can hinder the patients’ ability to manage their conditions.

NEMT has been deemed cost-effective and cost-saving for preventive care (e.g., dental care, screenings for breast and colorectal cancers) and the treatment of chronic conditions (e.g., asthma, chronic obstructive pulmonary disease, diabetes, end-stage renal disease, heart disease, hypertension, mental health) (Hughes-Cromwich & Wallace, 2006; Kim, Norton, & Stearns, 2009). Although this service is particularly useful for older adults (Bellamy, Stone, Richardson, & Goldsteen, 2003), it is not utilized sufficiently among all those who need it. The Andersen model indicates that health service use is determined by predisposing factors (e.g., demographic, socioeconomic, and beliefs), enabling factors (e.g., family and community), and need (perceived and evaluated health status) (Andersen & Newman, 1973). As such, this study examined some of the predisposing factors and need as indicated by client and trip characteristics associated with canceled trips among those who intended to use NEMT in a Medicaid population. This analysis focused on the potential barriers among older adults (aged 65 and older). More specifically, the aims of this study were to: (1) identify characteristics of NEMT clients and trips in Delaware; and (2) assess the trip purposes, client characteristics, and client reasons associated with canceling NEMT trips among older adult Medicaid beneficiaries. Understanding reasons for cancelations and barriers to NEMT utilization can inform policy and additional programming to increase use of NEMT among older adults.

Methods

Data and Population

This study is based on computerized records from a large NEMT broker that is active in 40 states and brokers more than 26 million NEMT trips per year. The NEMT service is typically provided curb-to-curb unless clients require additional assistance. The expected wait time is 30min (15min prior to 15min after the scheduled pick-up time). Reminder calls are not part of the standard procedure but are sometimes made for rural trips. Trips are assigned to the appropriate operating authority and clients are not transferred when crossing geographic boundaries.

Data for Medicaid eligible Delaware clients who were scheduled for a transportation appointment with the broker during 2008–2010 were analyzed. Scheduled appointments were analyzed because scheduling a transportation appointment indicates the patient intended to travel. Delaware data were selected for this exploratory analysis because of the relative diversity of the state and availability of complete data for the study period. In 2010, 14% of residents were aged 65 years or older, 52% were female, 8% were Hispanic or Latino, and 21% were black (U.S. Census Bureau, 2012).

Sociodemographic Data

Sociodemographic data such as, age, sex, and ZIP Codes for residential pick-up locations were available. Using ZIP Code, the percent black, percent poverty, and percent rural were obtained from the U.S. Census ZIP Code Tabulation Area data to provide a proxy for additional sociodemographic characteristics. Race data are from the Summary File 1 2010. Rural/urban and poverty data are from Summary File 3 2000. Urban is defined as core census block groups or blocks that have a population density of at least 1,000 people per square mile and surrounding blocks have an overall density of at least 500 people per square mile.

Trip Data

Data related to the trip were available. The treatment reason represents the medical purpose for the trip. Other trip characteristics include: trip date, scheduling status, miles from origin to destination, level of service, and whether the client would bring his or her escort. The scheduling status information provided categorized trips as urgent (requested within 24hr of use), prescheduled (i.e., repeated on a regular basis), or nonurgent (scheduled 2–30 days in advance). A leg of a trip includes a ride from one point to another.

Inclusion and Exclusion Criteria

A series of hierarchical inclusion/exclusion criteria were applied. Clients who were 65 years of age or older at their first scheduled trip for the 3-year study period were eligible for analysis. The database was structured as one record per user per trip date per leg. Scheduled legs for these clients include service that was: (1) not available (n = 696); (2) denied (n = 1,106); (3) no longer needed (n = 5,600); (4) a processing mistake (n = 845); (5) rescheduled (n = 3,967); or (6) canceled for administrative reasons (n = 2,912). Summarized at the trip level (user date level), there were 151,947 trips. Trips with more than two legs were excluded as there was concern that these trips may differ from one-way or round-way trips or that they may represent duplicate scheduled legs (n = 3,470, 2%). Of the remaining 148,477 (98%) trips, approximately 15% of the treatment types were for a higher level of care other than office visit (e.g., adult day care), did not have meaningful reasons, or were too few to represent a treatment category. Of the remaining, <1% had missing covariates of interest leaving 125,913 trips for 2,913 clients. Of these, 97% were scheduled as round trips (two-leg trips).

Primary Endpoints

The dependent variable for all analyses was cancelation (1 or yes vs. 0 or no). Four different canceled binary variables were derived from trip status (=“Canceled”) and cancelation reason. Cancelation reasons were categorized as:

• 1. All cancelations—this includes a range of reasons;

• 2. Client canceled and this includes the following reasons:

• Client no show

• Late cancelation

• Client canceled with sufficient notice

• Client refused

• 3. Client obtained alternative transportation and this includes the following reasons:

• Transported by family or friend

• Transported by other means

• Transported by another provider

• Client drove self

• Client walked

• 4. Client canceled due to health and this includes the following reasons:

• Client is sick

• Client is in the hospital

The dependent variables for #2–4 are mutually exclusive categories.

Independent Variables

Sex, age, ZIP Code percent black, Zip Code percent in poverty, ZIP Code percent rural, treatment reason, scheduling status, escort, and level of service were independent variables. The treatment of continuous variables (age, ZIP Code percent black, ZIP Code percent in poverty, and total miles of trip) was determined by evaluating the functional form of the log odds of each outcome for continuous variables categorized in deciles. Age was categorized as 65–74, 75–84, and 85+. ZIP Code percent black was categorized as <30% and 30%+. ZIP Code percent poverty was categorized as <10%, 10 to <15%, and 15%+. ZIP Code percent rural was also categorized as <51% and 51%+. Total miles was categorized as <30 and 30+ miles. Age, ZIP Code poverty, ZIP Code rurality, and total miles were also explored as a continuous variable with or without a quadratic term depending on the functional form. Numerous treatment reasons were provided and then categorized as: dialysis, doctor’s visit, mental health, rehabilitation, specialty, substance use, and testing or screening. Specialty encounters could include allergy, dental, dermatology, dietary, foot care, ear/nose/throat, vision, and so forth. Dummy variables were created for all categorical variables.

Analyses

It was expected that cancelations would vary by treatment reason and race (Leduc et al., 1998; Yeatts, Crow, & Folts, 1992). It was hypothesized that providing one’s own escort may be a mediator between scheduling status and cancelation outcomes. This was evaluated using Baron and Kenny’s approach (Baron & Kenny, 1986). Prescheduled trips are ones that are conducted on a repeated basis (e.g., every Monday) and it was hypothesized that these may be inherently different from other trips. Analyses were explored without the prescheduled trips. Frequency distributions were produced for categorical variables stratified by cancelation status. Bivariate and multivariate logistic regressions were conducted. The final models were as follows:

All types: Logit(Y) = β₀ + β₁Black + β₂Dr + β₃MentalHealth + β₄Rehab + β₅Specialty + β₆SubstanceUse + β₇Testing + β₈Presched + β₉Urgent + β₁₀Wheelchair + β₁₁Stretcher.

Client initiated: Logit(Y) = β₀ + β₁Black + β₂Dr + β₃MentalHealth + β₄Rehab + β₅Specialty + β₆SubstanceUse + β₇Testing + β₈Presched + β₉Urgent + β₁₀Wheelchair + β₁₁Stretcher.

Obtained alternative transportation: Logit(Y) = β₀ + β₁Black + β₂Dr + β₃MentalHealth + β₄Rehab + β₅Specialty + β₆SubstanceUse + β₇Testing + β₈Presched + β₉Urgent.

Health: Logit(Y) = β₀ + β₁Female + β₂Dr + β₃MentalHealth + β₄Rehab + β₅Specialty + β₆SubstanceUse + β₇Testing + β₈Presched + β₉Urgent + β₁₀Wheelchair + β₁₁Stretcher.

For those trips that were not prescheduled (n = 31,254), 13% percent were canceled, 8% were client initiated, <1% were due to alternative transportation, and 2% were for health reasons. Therefore, only cancelations for all reasons and client reasons were examined. The final models for these were as follows:

All types: Logit(Y) = = β₀ + β₁PovertyGroup2 + β₂PovertyGroup3 + β₃Dr + β₄MentalHealth + β₅Rehab + β₆Specialty + β₇SubstanceUse + β₈Testing + β₉Distance.

Client initiated: Logit(Y) = β₀ + β₁Black + β₂PovertyGroup2 + β₃PovertyGroup3 + β₄Dr + β₅MentalHealth + β₆Rehab + β₇Specialty + β₈SubstanceUse + β₉Testing + β₁₀Distance.

The inclusion/exclusion of the final covariates was determined using a series of Wald tests. The Huber White Sandwich estimator method was applied to produce robust standard errors adjusting for the correlation between observations on the same client. This differs from the typical standard errors produced based on maximum likelihood and gives more accurate estimates for clustered data of the sample-to-sample variability of the parameter estimates (Huber, 1967; White, 1980). Statistical significance levels were evaluated at the 0.05 level. All logistic regression analyses were conducted using Stata 12 (StataCorp, College Station, TX).

Results

Descriptive Analyses

Over a 3-year period, 125,913 trips were scheduled by 2,913 older adult clients (Table 1). Approximately two thirds of these trips were scheduled by females (66%), half were scheduled by persons younger than 75 (51%), and 10% were scheduled by persons aged 85 or older. Many of the older adults (42%) who intended to travel lived in ZIP Codes with 30% or more black residents. A small proportion of clients scheduled trips with escorts (3%). Seventy-five percent of all trips were scheduled for dialysis and a majority of those were prescheduled. The next leading reason for trip was for a doctor’s visit and <1% of these were prescheduled. As this service is for nonemergency medical care, only a small percentage (<1%) was considered urgent and 75% of all trips were prescheduled. Over 30% were scheduled for wheelchair level of service. Most trips were relatively close with the total trip distance a median of 10 miles and 98.8% of trips were within 60 miles (maximum = 280 total miles) (data not shown). Approximately 6% of all trips were canceled, and 55% of these were for client reasons (e.g., no show). There were 31,254 (25% of all trips) that were not prescheduled. Of these nearly 13% were canceled.

Table 1.

Trip Characteristics by Cancelation Reason, One-Way or Round-Trip Users Age 65+ Residing in Delaware, 2008–2010 (N = 125,913)

	Overall		Cancelations— all reasons (n = 7,798)		Cancelations— client reasons (n = 4,323)		Cancelations— alternative transportation (n = 486)		Cancelations— health reason (n = 1,019)
	n	Col (%)	n	Row (%)	n	Row (%)	n	Row (%)	n	Row (%)
Female	83,112	66.01	5,607	6.75	3,200	3.85	349	0.42	772	0.93
Age
65 to <75	64,021	50.85	3,929	6.14	2,200	3.44	226	0.35	479	0.75
75 to <85	49,894	39.63	2,885	5.78	1,534	3.07	199	0.40	402	0.81
85+	11,998	9.53	984	8.20	589	4.91	61	0.51	138	1.15
Zip Code characteristics
30+ % black	52,312	41.55	3,514	6.72	2,054	3.93	274	0.52	355	0.68
<10% poverty	67,685	53.76	4,073	6.02	2,208	3.26	253	0.37	520	0.77
10% to <15% poverty	24,736	19.65	1,489	6.02	840	3.40	72	0.29	199	0.80
15%+ poverty	33,492	26.60	2,236	6.68	1,275	3.81	161	0.48	300	0.90
>50% rural	25,927	20.59	1,476	5.69	788	3.04	69	0.27	221	0.85
Escort	4,231	3.36	568	13.42	358	8.46	41	0.97	80	1.89
Treatment reason
Dialysis	94,838	75.32	3,684	3.88	1,720	1.81	239	0.25	365	0.38
Doctor’s	19,509	15.49	2,577	13.21	1,700	8.71	160	0.82	429	2.20
Mental health	1,875	1.49	221	11.79	127	6.77	12	0.64	18	0.96
Rehabilitation	2,893	2.30	343	11.86	149	5.15	11	0.38	87	3.01
Specialty	4,314	3.43	594	13.77	363	8.41	48	1.11	92	2.13
Substance use	1,064	0.85	202	18.98	157	14.76	5	0.47	0	0
Testing/screening	1,420	1.13	177	12.46	107	7.54	11	0.77	28	1.97
Trip scheduling status
Nonurgent	30,924	24.56	3,879	12.54	2,462	7.96	282	0.91	665	2.15
Prescheduled (repeat basis)	94,659	75.18	3,870	4.09	1,838	1.94	194	0.20	353	0.37
Urgent	330	0.26	49	14.85	23	6.97	10	3.03	1	0.30
Trip characteristic
Level of service
Ambulatory	77,544	61.59	4,228	5.45	2,272	2.93	307	0.40	410	0.53
Wheelchair	39,557	31.42	2,982	7.54	1,764	4.46	152	0.38	453	1.15
Stretcher	8,812	7.00	588	6.67	287	3.26	27	0.31	156	1.77
Trip distance
30+ miles	12,891	10.24	997	7.73	544	4.22	62	0.48	160	1.24

Logistic Analyses With all Trips

Sociodemographic and Social Support Characteristics

In the multivariate analysis, females were significantly more likely to cancel for health reasons (odds ratio [OR] = 1.4). In bivariate analyses, the oldest old (aged 85+) compared with those aged 65–74 were generally more likely to cancel (ORs ranged 1.4–1.5). However, these results did not hold in the multivariate analyses (Tables 2 and 3). In multivariate analyses, living in a ZIP Code 30% or more black residents was associated with increased odds of all canceling for any reason, except for health reasons (ORs ranged 1.2–1.9). Poverty was not significantly associated with cancelation outcomes in this population. In bivariate analyses, living in a predominately rural area was associated with a decreased odds of canceling due to alternative transportation (OR = 0.6); however, this finding did not hold in the multivariate analysis results. Being able to provide one’s own escort, an indicator of caregiver support, increased the odds of all cancelations (ORs ranged 2.5–2.8); however, after consideration of treatment reason and scheduling this relationship was not observed.

Table 2.

Bivariate Logistic Regression Modeling the Odds of One-Way or Round Trip Cancelation by Cancelation Reason, One-Way or Round-Trip Users Age 65+ Residing in Delaware, 2008–2010 (N = 125,913)

	Cancelations—all reasons (n = 7,798)				Cancelations—client reasons (n = 4,323)				Cancelations—alternative transportation (n = 486)				Cancelations—health reason (n = 1,019)
	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p	OR	95% CI		P
Female versus male	1.34	1.15	1.57	<.0001	1.49	1.18	1.87	<.01	1.31	0.83	2.08	.25	1.62	1.18	2.22	<.01
Age
65 to <75	ref				ref				ref				ref
75 to <85	0.93	0.80	1.10	.41	0.89	0.71	1.10	.27	1.12	0.68	1.86	.65	1.07	0.79	1.46	.66
85+	1.36	1.10	1.68	<.01	1.44	1.10	1.88	<.01	1.43	0.94	2.19	.10	1.53	0.98	2.39	.06
Race (1)
30+ % black versus <30%	1.16	0.99	1.37	.06	1.28	1.04	1.59	<.05	1.82	1.21	2.74	<.01	0.75	0.56	1.01	.06
Poverty (1)
<10%	ref				ref				ref				ref
10 to <15%	1.00	0.82	1.21	1.00	1.04	0.80	1.35	.76	0.78	0.51	1.18	.24	1.05	0.70	1.58	.82
15%+	1.12	0.92	1.36	.27	1.17	0.90	1.53	.24	1.29	0.72	2.29	.39	1.17	0.82	1.65	.38
Rurality (1)
>50% rural versus ≤50%	0.89	0.75	1.07	.23	0.86	0.67	1.09	.21	0.64	0.43	0.95	<.05	1.07	0.74	1.54	.72
Escort versus no escort	2.46	2.15	2.82	<.0001	2.75	2.31	3.26	<.0001	2.66	1.70	4.18	<.0001	2.47	1.85	3.30	<.0001
Treatment reason
Dialysis	ref				ref				ref				ref
Doctor’s	3.77	3.36	4.22	<.0001	5.17	4.32	6.17	<.0001	3.27	2.12	5.03	<.0001	5.82	4.46	7.58	<.0001
Mental health	3.31	1.33	8.25	<.05	3.93	1.24	12.49	<.05	2.55	0.95	6.81	.06	2.51	0.49	12.97	.27
Rehabilitation	3.33	2.49	4.46	<.0001	2.94	1.93	4.47	<.0001	1.51	0.48	4.71	.48	8.03	3.90	16.52	<.0001
Specialty	3.95	3.41	4.56	<.0001	4.97	4.04	6.11	<.0001	4.45	2.68	7.38	<.0001	5.64	3.94	8.05	<.0001
Substance use	5.80	2.93	11.46	<.0001	9.37	3.98	22.08	<.0001	1.87	0.43	8.17	.41	(2)
Testing/screening	3.52	2.89	4.29	<.0001	4.41	3.36	5.78	<.0001	3.09	1.51	6.29	<.01	5.20	3.34	8.10	<.0001
Trip scheduling status
Nonurgent	ref				ref				ref				ref
Prescheduled (repeat basis)	0.30	0.26	0.34	<.0001	0.23	0.19	0.28	<.0001	0.22	0.13	0.37	<.0001	0.17	0.13	0.22	<.0001
Urgent	1.25	0.87	1.81	.23	0.89	0.58	1.39	.62	3.35	1.67	6.73	<.0001	0.14	0.02	0.97	<.05
Trip characteristic
Level of service
Ambulatory	ref				ref				ref				ref
Wheelchair	1.41	1.22	1.64	<.0001	1.55	1.26	1.90	<.0001	0.97	0.65	1.44	.89	2.18	1.58	3.00	<.0001
Stretcher	1.24	0.95	1.62	.11	1.12	0.80	1.56	.50	0.77	0.44	1.34	.36	3.39	2.15	5.35	<.0001
Total trip distance
30+ miles versus <30	1.31	1.08	1.56	<.01	1.28	1.01	1.61	<.05	1.28	0.84	1.95	.24	1.64	1.16	2.32	<.01

Notes: (1) Zip Code characteristics were obtained from the U.S. Census and (2) No cancelations for this treatment type. CI = confidence interval; OR = odds ratio; and the standard errors were produced using the Huber White Sandwich estimator adjusting for the clustering by user (n = 2,913).

Table 3.

Multivariate Logistic Regression Modeling the Odds of One-Way or Round Trip Cancelation by Cancelation Reason, One-Way or Round-Trip Users Age 65+ Residing in Delaware, 2008–2010 (N = 125,193)

	Cancelations—all reasons (n = 7,798)				Cancelations—client reasons (n = 4,323)				Cancelations—alternative transportation (n = 486)				Cancelations—health reason (n = 1,019)
	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p
Female versus male													1.40	1.08	1.81	<.05
Race (1)
30+ % black versus <30%	1.24	1.09	1.40	<.01	1.37	1.14	1.62	<.01	1.92	1.29	2.85	<.01
Treatment reason
Dialysis	ref				ref				ref				ref
Doctor’s	2.11	1.62	2.75	<.0001	2.53	1.69	3.79	<.0001	0.65	0.44	0.94	<.05	1.64	1.12	2.39	<.05
Mental health	3.31	1.37	7.97	<.01	3.89	1.27	11.86	<.05	1.78	0.71	4.43	.22	2.92	0.57	15.08	.20
Rehabilitation	2.52	1.92	3.31	<.0001	2.07	1.36	3.16	<.01	0.54	0.20	1.44	.22	3.90	2.17	7.03	<.0001
Specialty	2.24	1.71	2.95	<.0001	2.46	1.63	3.71	<.0001	0.84	0.53	1.33	.46	1.76	1.13	2.75	<.05
Substance use	5.17	2.47	10.83	<.0001	7.92	3.06	20.51	<.0001	0.83	0.13	5.17	.84	(2)
Testing/screening	1.99	1.45	2.72	<.0001	2.21	1.41	3.46	<.01	0.56	0.28	1.12	.10	1.55	0.91	2.65	.11
Trip scheduling status
Nonurgent	ref				ref				ref				ref
Prescheduled (repeat basis)	0.56	0.43	0.73	<.0001	0.49	0.32	0.74	<.01	0.16	0.09	0.27	<.0001	0.30	0.21	0.45	<.0001
Urgent	1.59	1.08	2.33	<.05	1.16	0.73	1.86	.53	3.14	1.59	6.19	<.01	0.15	0.02	1.01	0.05
Trip characteristic
Level of service
Ambulatory	ref				ref								ref
Wheelchair	1.29	1.14	1.45	<.0001	1.40	1.18	1.66	<.0001					1.77	1.34	2.35	<.0001
Stretcher	1.13	0.94	1.35	.18	1.00	0.80	1.25	.99					2.91	2.02	4.18	<.0001
p for goodness of fit	<.0001				<.0001				<.0001				<.0001

Notes: (1) Zip Code characteristics were obtained from the U.S. Census and (2) No cancelations for this treatment type. Shading indicates that the independent variable was not included in the model; CI = confidence interval; OR = odds ratio; and the standard errors were produced using the Huber White Sandwich estimator adjusting for the clustering by user (n = 2,913).

Treatment and Trip Level of Service.

Reason for treatment was significantly associated with cancelations. In multivariate analyses, trips for a wide variety of treatment reasons were more likely to be canceled compared with dialysis encounters except for obtained alternative transportation cancelations. Doctor’s encounter trip had 2.1 and 2.5 times the odds for all reasons cancelations and client cancelations, respectively. Substance use trips had 5.2 and 7.9 times the odds, respectively; however, the estimates are less precise (95% confidence interval = 2.5–10.8 and 3.1–20.5, respectively). No substance use trips were canceled for health reasons. In multivariate analyses, more frail individuals as indicated by need for wheelchair and stretcher had 1.8 and 2.9 times the odds of canceling for health reasons compared with ambulatory level of service, respectively. Needing a wheelchair was associated with increased odds of canceling for any reason and for client reasons by 29% and 40%, respectively, compared with ambulatory level of service.

Trip Scheduling and Characteristics

Prescheduled trips were less likely to be canceled compared with regularly scheduled trips (ORs ranged 0.2–0.6). Urgent trips were associated with an increased odds of canceling due to obtaining alternative transportation (OR = 3.1) and a decreased odds of canceling due to health (OR = 0.2). In bivariate analyses, longer trips were associated with an increased odds of being canceled (ORs ranged 1.3–1.6) compared with shorter trips; however, this finding did not hold when treatment reason and scheduling were considered.

Logistic Analyses Without Prescheduled Trips

Among trips that were not scheduled on a regular repeat basis, cancelations did not differ by urgent versus nonurgent stats (data not shown). In multivariate analyses, black race and poverty were related to client cancelations (Table 4). Clients living in higher poverty were associated with a reduced odds of 19%–25% of client canceling compared with the lowest poverty group while controlling for other factors. The odds of cancelations for all reasons and client reasons decreased with increase in trip distance (in miles). Trips for treatments other than dialysis were more likely to be canceled except for substance use which had a reduced odds.

Table 4.

Multivariate Logistic Regression Modeling the Odds of One-Way or Round Trip Cancelation by Cancelation Reason, One-Way or Round-Trip Users Age 65+ Residing in Delaware (Excluding Prescheduled Trips), 2008–2010 (N = 31,254)

	Bivariate								Multivariate
	Cancelations—all reasons (n = 3,928)				Cancelations—client reasons (n = 2,485)				Cancelations—all reasons (n = 3,928)				Cancelations—client reasons (n = 2,485)
	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p	OR	95% CI		p
Female versus male	1.13	1.00	1.27	<.05	1.15	1.00	1.31	<.05
Age
65 to <75	ref				ref
75 to <85	1.03	0.91	1.16	.64	1.05	0.90	1.22	.57
85+	1.16	1.01	1.34	<.05	1.17	0.99	1.38	.07
Race (1)
30+ % black versus <30%	1.03	0.92	1.17	.58	1.14	0.99	1.31	.06					1.24	1.06	1.44	<.01
Poverty (1)
<10%	ref				ref				ref				ref
10 to <15%	0.85	0.71	1.00	.06	0.81	0.67	0.97	<.05	0.83	0.70	0.98	<.05	0.75	0.62	0.90	<.01
15%+	0.93	0.82	1.05	.25	0.90	0.75	1.08	.24	0.90	0.79	1.01	.08	0.81	0.68	0.97	<.05
Rurality (1)
>50% rural versus ≤50%	0.95	0.82	1.10	.48	0.86	0.73	1.01	.06
Escort versus no escort	1.09	0.96	1.24	.17	1.08	0.93	1.26	.29
Treatment reason
Dialysis	ref				ref				ref				ref
Doctor’s	1.73	1.46	2.06	<.0001	1.82	1.48	2.24	<.05	1.78	1.50	2.11	<.0001	1.89	1.54	2.32	<.0001
Mental health	1.84	0.96	3.50	.06	2.10	1.02	4.32	<.05	1.88	0.98	3.60	.06	2.06	1.01	4.19	<.05
Rehabilitation	1.83	1.35	2.48	<.0001	1.14	0.64	2.04	.65	1.89	1.39	2.56	<.0001	1.24	0.73	2.13	.43
Specialty	1.81	1.49	2.21	<.0001	1.76	1.39	2.22	<.0001	1.88	1.55	2.29	<.0001	1.82	1.44	2.30	<.0001
Substance use	0.42	0.27	0.67	<.0001	0.17	0.05	0.55	<.01	0.43	0.27	0.67	<.0001	0.16	0.05	0.51	<.01
Testing/screening	1.62	1.28	2.05	<.0001	1.55	1.16	2.08	<.01	1.64	1.29	2.08	<.0001	1.56	1.17	2.08	<.01
Trip characteristic
Level of service
Ambulatory	ref				ref
Wheelchair	1.14	1.02	1.28	<.05	1.18	1.02	1.36	<.05
Stretcher	1.07	0.90	1.28	.45	1.02	0.84	1.23	.85
Trip distance	0.998	0.996	0.999	<.05	0.997	0.995	0.999	<.05	0.997	0.995	0.999	<.01	0.997	0.995	0.999	<.01
p for goodness of fit									<.0001				<.0001

Notes: (1) Zip Code characteristics were obtained from the U.S. Census. Shading indicates that the independent variable was not included in the model; CI = confidence interval; OR = odds ratio; and the standard errors were produced using the Huber White Sandwich estimator adjusting for the clustering by user (n = 2,874).

Discussion

This study explored indicators of predisposing factors and health care necessity associated with canceling subsidized transportation for health care among Delaware Medicaid older adults who intended to travel. Findings suggest that client- and trip-related factors were associated with canceled trips; however, these factors differed slightly based on the reason for cancelation. The Medicaid subsidized NEMT provides access to preventive and potentially end of life care for vulnerable populations with transportation barriers. This study can assist in improving this program which could increase use and reduce missed appointments.

Within this study, a few patterns emerged suggesting that missed or delayed medical care was greater by certain predisposing factors. Black race was associated with increased odds of not showing/canceling short notice and canceling due to securing a transportation alternative for the appointment. It should be noted that this characteristic may not reflect the respondent’s own race but, in some cases, may serve as a measure for ZIP Code socioeconomic status and transportation opportunities. In terms of having transportation alternatives, it may be that among this population there were additional caregiving resources or that this proxy for race, which is based on ZIP Code, represents some measure of residential urbanity where other transportation resources were available. Age was not an important factor in cancelations and this is consistent with other research where health and function and not age predicted specialized transportation service use among older adults (Nasvadi & Wister, 2006). It is possible that the oldest old have care attendants or others that help them with their scheduling so that age itself is not a barrier to health care utilization. Females were more likely to cancel for health reasons. Sociodemographic characteristics are often associated with health care utilization. Such characteristics may reflect beliefs and attitudes (Andersen & Newman, 1973). In the United States, these characteristics may also be related to health and social status that present additional challenges to managing one’s health care.

Indicators of health and functional status were also associated with cancelations while controlling for sociodemographic characteristics. Compared with dialysis trips, other medical encounters were more likely to be canceled by the client. However, wheelchair and stretcher level of service increased the odds of health cancelations and it is not clear if medical appointments were missed or if the condition was elevated to emergency medical services. Interestingly, trips for substance use encounters were much more likely to be canceled. For both health care and substance use treatment compliance is typically lower among substance users (Dubinsky, 1986). The perceived or diagnosed severity of one’s health condition and the perceived benefits of treatment may be associated with health care utilization (Andersen & Newman, 1973; Leduc et al., 1998).

Policy and Program Implications

In this population, very few participants canceled due to securing another type of transportation. Other studies have indicated that many older adults do not have alternative transportation support (Choi, Adams, & Kahana, 2012). Affordable health care and transportation are important enabling factors for health service utilization. Medicaid subsidized NEMT has numerous implications for health, quality of life, and medical costs for older adults. It also allows older adults to age in place. This service is important for a healthy population and for reducing inequities in health. However, this service may be underutilized. In 2010, 43% of Delaware older adults who were eligible for Medicaid NEMT enrolled in this program. (Prohaska et al., 2012). Some populations may face persistent barriers to using this service that may be addressed through additional programming at various policy levels.

National Policies.

U.S. Centers for Medicare and Medicaid, the Federal Transit Administration, and the Administration for Community Living are federal sources of funding for specialized transportation for older adults and adults with disabilities. However, funding varies by state and by federal funding source (Lynott, Fox-Grage, & Guzman, 2013). The scope and size of many Medicare and Medicaid programs will expand with the passing of Health Care Reform in the United States (Community Transportation Association of America, 2009). These changes will introduce millions of newly eligible beneficiaries, many of whom were previously without access to health care services and programs. However, this increase in service availability does not guarantee that eligible beneficiaries can access needed services and programs. Because of the impending growth of NEMT demand in forthcoming years, Health Care reform requires that States provide NEMT to beneficiaries without transportation to obtain medical care (Community Transportation Association of America, 2008). Further, under Reform, States are granted flexibility to establish NEMT brokerage programs and receive the Federal medical assistance percentage matching rate for such services (Community Transportation Association of America, 2008). With the passing of Health Care Reform, the role of NEMT becomes increasingly important. There will be an increase in eligibility for NEMT and, with more NEMT utilization, this will also result in an increase in the absolute number of canceled NEMT trips, which has financial and health ramifications for the beneficiary, their families, and greater society.

State and Local Policies.

To improve the effectiveness of this service, NEMT may need to be integrated into a whole system of care. First, increasing local public support can lead to better funding for specialized transportation. For example, at the state level, gas taxes could support transportation efforts. Support at the state level is also necessary for the successful funding of Medicaid NEMT (Lynott, Fox-Grage, & Guzman, 2013).

Second, involving the medical community could assist in reducing missed appointments through better coordination with transportation. In this study, regularly scheduled trips were less likely to be canceled. There is research to indicate that the time between scheduling and the medical appointment is associated with missed appointments, reminders can reduce missed appointments, and that this may be particularly important for older adults whose memory may be declining with age (Parikh et al., 2010). In these data clients can schedule up to 30 days in advance and reminder phone calls are not required. Engaging medical staff to ensure that patients have transportation for their appointments could improve utilization and reduce longer term costs. Currently, the Affordable Care Act provides an incentive for hospitals to reduce readmissions (Lynott, Fox-Grage, & Guzman, 2013). Similarly, efforts should be introduced that incentivize preventive care, which has potential to reduce the prevalence of canceled NEMT appointments.

Third, NEMT providers are, potentially, an important part of the health care provider team. NEMT agencies may be able to improve use by offering passenger incentives and utilizing technology to manage no-shows and late cancelations (Transit Cooperative Research Program, 2005). In addition, efforts to improve trust and satisfaction with NEMT providers could improve health care utilization. Consistent and supportive health care relationships are important to ongoing improved health care goals. There is research to indicate that connection to one’s health care provider is important for maintaining that relationship. In fact, some of the reasons that patients do not show for medical appointments include perceived disrespect (Parikh et al., 2010).

Finally, the effectiveness and efficiency of this program should be continually evaluated at the state level. It has been suggested that costs should be better monitored and that Centers for Medicare and Medicaid should make state Medicaid NEMT program data publicly available for policy research (Lynott, Fox-Grage, & Guzman, 2013). To address the service efficiency and cost, many paratransit agencies have implemented policies to reduce client no-shows and late cancelations. Some agencies have observed declines in no-show and late cancelation rates and these declines are associated with system cost savings (Transit Cooperative Research Program, 2008). However, client outcomes associated with these policies should be evaluated.

Limitations and Future Research

This study shows the feasibility of using this database to examine Medicaid NEMT. Delaware, while a diverse state, represents a manageable opportunity to investigate these issues. However, these findings may not be generalizable to other geographic areas. In addition to differences in clients, policies may differ by region and agency. Among these data, approximately 6% of all trips were canceled and nearly 13% of prescheduled trips were canceled. No-show and late cancelation rates can vary by agency. They can differ in how these types of cancelations are defined and how they are addressed. One study of urban agencies, observed rates of 3%–10% prior to implementing any no-show and late cancelation policies (Transit Cooperative Research Program, 2008). The factors associated with cancelations should be studied in other states and regions.

In addition, it is important to note that the analyses are somewhat limited as complete sociodemographic information was not available for each client. As previously noted, beliefs and family and community resources may be important factors for health care utilization (Andersen & Newman, 1973). These data do not include enabling factors about one’s living situation or caregivers. In addition, there is no information about who schedules these NEMT appointments. Some clients who were no-shows may have been categorized differently if they had family or care attendants who were able to schedule and cancel appointments on their behalf. Also, the data provided did not include the number of days in advance one’s trip was scheduled. There may be important differences in trips that were scheduled a week in advance versus a month in advance. Further, the data are not linked to claims or medical records so a complete assessment of health and functional status cannot be determined.

Within these limitations, there are several avenues for future research. First, linking NEMT data to medical records and other databases would allow for evaluating comorbidities, complete health care utilization, and health outcomes for monitoring the effectiveness of NEMT. In this study, clients who were stretcher level of service had increased odds of canceling for health reasons compared with ambulatory level of service. However, it may be that emergency medical service provided transportation and that medical care was not missed but this could not be evaluated. Second, future research should include other geographic regions and better measures of alternative transportation services. It was expected that clients who lived in rural areas may be particularly reliant on NEMT as alternative transportation (e.g., bus, taxi) may not be available. In this study, a small proportion of clients lived in highly rural areas and there were relatively few cancelations due to alternative transportation. Finally, future research should focus on better understanding the complexities of persistent barriers that cannot be captured in service and claims data. Specifically, there may be special needs and concerns among lower income, older adults. For example, factors such as support networks, weather, and transportation service provider trust may play an important role in service use. All in all, NEMT has a significant role in health care reform as it relates to health care cost and population health.

Funding

The study is part of a project that is a collaboration of Easter Seals Project Action, the American Medical Association, and LogistiCare and was funded by the U.S. Department of Transportation Federal Transit Administration cooperative agreements Easter Seals Project Action and the National Center on Senior Transportation (PI: Dr. T. R. Prohaska). Ms. MacLeod was partially supported by Award Number T32AA007240, Graduate Research Training on Alcohol Problems, from the National Institute on Alcohol Abuse and Alcoholism.