Disparities in Access to Medications for Opioid Use Disorder in the Veterans Health Administration

Abstract

Objectives:

A variety of patients – including women, older, racial/ethnic minority, rural, homeless, and justice-involved patients – are vulnerable to experiencing poor healthcare access and quality, such as lower quality substance use disorder treatment, than other populations. The current study examined receipt of medications for opioid use disorder by vulnerable populations within Veterans Health Administration (VHA) facilities to determine whether there are patient and facility factors that are associated with disparities in care.

Methods:

Using national VHA clinical/administrative data from Fiscal Year 2017, we calculated receipt of medications for opioid use disorder using the American Society for Addiction Medicine quality measure specifications. A mixed-effects logistic regression model tested whether patient vulnerability (i.e., women, older age, racial/ethnic minority, rural residence, homeless, and justice-involved) and facility (e.g., regional location, availability of a methadone clinic) characteristics were associated with medication receipt.

Results:

Among the 53,568 veterans at VHA facilities diagnosed with opioid use disorder in Fiscal Year 2017, vulnerable populations – including women, older, Black, rural, homeless, and justice-involved veterans – had lower odds of receiving medications for opioid use disorder than their non-vulnerable counterparts. Veterans had higher odds of receiving medications at facilities with a higher proportion of patients with opioid use disorder, but lower odds of receiving medications at facilities in the Southern region compared to the Northeast region of the United States.

Conclusions:

Quality improvement efforts targeted at vulnerable populations are needed at the VHA to ensure these groups receive the same quality of substance use disorder treatment as other veterans.

Introduction

Medications are the most effective treatments for opioid use disorder with strong evidence supporting methadone and buprenorphine as first-line medications (Amato et al. 2005, Volkow et al. 2019) and naltrexone as a second-line medication (Department of Veterans Affairs & Department of Defense 2015). Positive outcomes associated with these medications include reduced heroin use, likelihood of repeat overdose, and mortality (Amato et al. 2005, Sordo et al. 2017, Larochelle et al. 2018). Medications for opioid use disorder are estimated to cost less than short-term, medically mandated withdrawal from opioid use with more quality-adjusted life-years accumulated (Krebs et al. 2018). Despite the benefits, these medications are underutilized (Volkow et al. 2019) and some populations may face heightened challenges accessing medications for opioid use disorder. The National Institute on Minority Health and Health Disparities Research Framework (2018) and the National Institute on Aging (Hill et al. 2015) propose frameworks with fundamental factors to guide health disparities research. Drawing from these frameworks, we identified six vulnerable populations – women, older veterans, racial and ethnic minority veterans, rural veterans, homeless veterans, and justice-involved veterans – and examined potential disparities in treatment for opioid use disorder.

Vulnerable populations may experience lower quality substance use disorder treatment than non-vulnerable populations. National studies of United States military veterans from Fiscal Year 2008 (FY08) and FY2012 indicated that vulnerable groups (e.g., women compared to men) had lower odds of receiving medications for opioid use disorder (Oliva et al. 2012, Finlay et al. 2016). Non-veteran studies have yielded similar results. Racial/ethnic minority patients and patients 30 years old and older in the United States had lower odds of receiving medications for opioid use disorder (Morgan et al. 2018, Evans et al. 2019). Given the evolving opioid epidemic and many public health initiatives to improve treatment access, more recent data on receipt of medications for opioid use disorder for vulnerable populations can be used to develop and improve treatment programs.

Facility or program factors may also be associated with lower utilization of medications for opioid use disorder by vulnerable populations; however, identified facility factors are limited. In a national study of substance use disorder treatment programs, a higher racial or ethnic minority patient mix at the program level was associated with less medication availability (Knudsen and Roman 2009). A national study of Veterans Health Administration (VHA) facilities found lower odds of receipt of medications for opioid use disorder at facilities where there was a longer waitlist to residential programs, but higher receipt where a methadone clinic was available or where medications for psychiatric disorders were provided from treatment program staff or via a contracted provider (Finlay et al. 2016).

Previously identified barriers to medications for opioid use disorder include patient, provider, and system level factors (summarized by Oliva et al. 2011). In addition to vulnerability, patient barriers include transportation difficulties, homelessness and other challenging living situations, stigma towards medications, and a preference not to use medications (Olsen and Sharfstein 2014, Fox et al. 2015, Godersky et al. 2019). Provider-level barriers include a lack of interest or education and negative attitudes towards medications (Gordon et al. 2011, Aletraris et al. 2016). System-level barriers include policies against providing medications and a lack of treatment facilities or programs (Friedmann et al. 2012, Quest et al. 2012). These barriers have been documented for many years in a variety of healthcare settings and systems.

Vulnerable populations within the Veterans Health Administration (VHA) should experience similar care quality as non-vulnerable populations because access to medications for opioid use disorder is mandated by VHA (Department of Veterans Affairs 2008) and treatment services are largely provided for free or with low co-pays for qualified veterans. However, prior research suggests that vulnerable populations may experience disparities in receipt of medications for opioid use disorder, sometimes due to facility characteristics (Knudsen and Roman 2009, Finlay et al. 2016). In this study, we examined the association between vulnerable populations, facility characteristics, and receipt of medications for opioid use disorder. We first examined the prevalence of opioid use disorder diagnosis to determine if there were any differences among different vulnerable populations. We then examined the prevalence of receipt of medications for opioid use disorder by vulnerability and facility characteristics. We hypothesized that vulnerable populations would have lower receipt of medications for opioid use disorder than non-vulnerable populations. We also hypothesized that facilities that had higher (compared to lower) opioid use disorder prevalence or had more (compared to less) available substance use disorder treatment services would have higher receipt of medications.

Method

Study Setting and Data Source

This study used national patient-level VHA clinical/administrative records from the Corporate Data Warehouse, which contains all VHA or VHA-contracted community care received by United States military veterans. The VHA provides healthcare to service-connected veterans in VHA facilities located throughout the United States. This study was approved by the Stanford University Institutional Review Board and the VA Palo Alto Research & Development Committee.

Study Population

All veterans who received care at VHA facilities or care paid for by VHA in FY17 (October 1, 2016 to September 30, 2017), as indicated by their patient records, were included in the study. Six vulnerable populations were defined: women, veterans age 35 and older, veterans from racial/ethnic minority groups, veterans who lived in rural areas, homeless veterans, and justice-involved veterans. Sex was coded as women or men. Age was coded as < 35, 35–44, 45–54, and 55+ to align with a prior study of veterans’ receipt of medications for opioid use disorder (Finlay et al. 2016) so that we could follow changes over time for VHA patients. Based on the United States Bureau of Census categories, race/ethnicity was coded as Hispanic, non-Hispanic American Indian/Alaskan Native, non-Hispanic Asian, non-Hispanic Black, and non-Hispanic White. Based on a patient’s last known address, urban areas were defined as having an urban core of at least 1,000 residents per square mile or 50,000 or more people in the urban nucleus and rural areas were defined as non-urban areas (U.S. Census Bureau definition; Ratcliffe et al. 2016). Based on a prior study (Finlay et al. 2016), homeless status was defined as utilizing services for homeless veterans (clinic codes: 501, 504, 507, 508, 511, 515, 522, 528, 529, 530, 555, 556, 590; bed section codes: 28, 29, 37, 39) or International Classifications of Diseases (ICD)-10th Edition-CM codes for housing and homelessness (Z59.0). Justice-involved status was defined by clinic codes (591, 592) or homeless outreach records indicating contact with the criminal justice system (Blue-Howells et al. 2013).

Measures

Prevalence of opioid use disorder.

Veterans were classified as having an opioid use disorder if they received an ICD-10 diagnosis code for opioid use disorder during an outpatient or inpatient visit in the Fiscal Year (Harris et al. 2016). Any veteran with a current diagnosis was included, whereas veterans with only in-remission diagnoses were excluded. The prevalence of opioid use disorder was calculated as the number of veterans diagnosed with opioid use disorder divided by the total number of veterans in the study population, separately for each vulnerable population.

Receipt of medications for opioid use disorder.

For each veteran, receipt of medications for opioid use disorder was coded as a binary variable, based on the American Society for Addiction Medicine’s (ASAM) version 2 specifications (Harris et al. 2016). Veterans had received medications for opioid use disorder if they had at least one VHA or VHA-contracted care methadone clinic outpatient visit (clinic code 523) with a concurrent opioid use disorder diagnosis and/or at least one pharmacy prescription fill for buprenorphine or naltrexone (oral or injectable) during the same Fiscal Year as their opioid use disorder diagnosis. The prevalence of receipt of medications for opioid use disorder was coded as the number of veterans who received medications for opioid use disorder divided by number of veterans diagnosed with opioid use disorder, separately for each vulnerable population.

Facility characteristics.

Facility variables were coded using FY17 patient-level data aggregated at the facility level and selected based on availability in VHA data and prior studies (Knudsen and Roman 2009, Finlay et al. 2016). Four facility variables measured patient composition. Facility size was the number of VHA veterans served at that facility. The prevalence of veterans with opioid use disorder was the number of veteran patients diagnosed with opioid use disorder divided by the total number of veteran patients at the facility. The prevalence of women was the number of women veteran patients at the facility divided by the total number of veteran patients at the facility. The prevalence of racial/ethnic minority patients was the number of racial or ethnic minority patients (American Indian/Alaskan Native, Asian, Black, Hispanic) divided by the total number of patients at the facility. Two variables measured facility location: rural versus urban and geographic region where the facility was located (Northeast, South, Midwest, West). Five facility variables were related to substance use disorder treatment. The average number of substance use disorder visits was the number of substance use disorder visits that occurred in FY17 divided by the number of veteran patients diagnosed with a substance use disorder. The prevalence of medications for alcohol use disorder was the number patients who received medications for alcohol use disorder (acamprosate, topiramate, naltrexone, or disulfiram; coded as a dichotomous variable) divided by the number of patients diagnosed with alcohol use disorder, coded separate by facility and then averaged across facilities. Any facility that had records of bed section codes 72 (Alcohol Dependency-High Intensity), 73 (Drug Dependency-High Intensity), and 74 (Substance Abuse-High Intensity) were coded as having available inpatient detox. Any facility that had records of bed section codes 85 (Domiciliary), 109 (Psychiatric Residential Rehabilitation Programs), 86 (Domiciliary Substance Abuse), and 111 (Substance Abuse Residential Programs) were coded as having available substance use disorder residential treatment. Any facility that had a record of clinic stop code 523 (Opioid Substitution) was coded as having an available methadone clinic.

Covariates.

In a prior study, marital status, service-connected disability rating, and co-occurring conditions (psychiatric, substance use disorder, and medical) explained differences in medication receipt (Baxter et al. 2011, Finlay et al. 2016, Morgan et al. 2018). Covariates in this study included marital status (married, not married), service-connected disability rating (i.e., a rating of conditions caused or exacerbated by military service, which serves as a proxy for socio-economic status), co-occurring psychiatric conditions (depression, post-traumatic stress disorder, anxiety, bipolar, schizophrenia, other psychosis, or personality disorders), co-occurring substance use disorders (alcohol, cocaine, cannabis, amphetamine, sedative, and other drug use disorders), co-occurring medical conditions (measured by the Deyo comorbidity index; Deyo et al. 1992), and patient receipt of substance use disorder residential treatment (record of bed section codes 85, 109, 86, or 111).

Analyses

Differences between vulnerable and non-vulnerable groups in the prevalence of opioid use disorder diagnosis and the prevalence of receipt of medications for opioid use disorder were examined using chi-square tests. Mixed-effects logistic regression models, with a random effect for facility (n = 129) and exchangeable covariance structure, were used to test associations between patient and facility characteristics and the dichotomous outcome of receipt of medications for opioid use disorder (yes/no). Veterans are clustered by VHA facility and receipt of medications for opioid use disorder vary by facility (Finlay et al. 2018a). Therefore, a mixed-effects model, rather than a fixed-effects model, was chosen to preserve degrees of freedom, because of the multi-level nature of our predictors (patient and facility characteristics), and because we were not interested in specific facility effects. The first model was an unadjusted mixed effects logistic regression model that tested whether the primary variables of interest - vulnerable groups and facility characteristics (entered into the model simultaneously with a random effect for facility) - were associated with receipt of medications for opioid use disorder. The second model conducted was an adjusted mixed effects analysis that included additional patient-level covariates. There were 2,990 (6%) patients excluded from the logistic regression models due to missing data. A significance level of α = .05 was used. The models were conducted using SAS software version 9.04 and the PROC GLIMMIX command (SAS Institute Inc. 2013).

Results

Opioid Use Disorder Diagnosis

Among patients seen at VHA facilities in FY17, 53,568 (2%) were diagnosed with opioid use disorder. The prevalence significantly differed for all vulnerable compared to non-vulnerable groups (Table 1). The prevalence of opioid use disorder was notably higher among homeless compared to housed veterans (10% versus 2% respectively, p < .001), and justice-involved compared to non-justice-involved veterans (10% versus 2% respectively, p < .001).

Table 1.

Prevalence of Opioid Use Disorder and Receipt of Medications by Vulnerable Groups and Other Patient Characteristics in the United States Veterans Health Administration in Fiscal Year 2017

	Prevalence of Opioid Use Disordera	p	Prevalence of Receipt of Medicationsb	p
	N (%)		N (%)
Overall	53,568 (2%)		22,124 (41%)
Gender		< .001		0.702
Women	3,814 (1%)		1,564 (41%)
Men	49,754 (2%)		20,560 (41%)
Age		< .001		< .0001
<35	10,569 (3%)		5,585 (53%)
35–44	7,436 (2%)		3,789 (51%)
45–54	7,980 (2%)		3,215 (40%)
55+	27,583 (1%)		9,535 (35%)
Race/Ethnicity		< .001		< .0001
American Indian/Alaskan Native	533 (2%)		194 (36%)
Asian	528 (1%)		204 (39%)
Black	10,568 (2%)		4,150 (39%)
Hispanic	2,964 (1%)		1,224 (41%)
White	36,864 (2%)		15,552 (42%)
Residence		< .001		< .0001
Rural	15,466 (2%)		5,649 (37%)
Urban	37,124 (4%)		16,141 (43%)
Homelessness		< .001		< .0001
Homeless	22,982 (10%)		10,083 (35%)
Housed	30,586 (2%)		12,041 (49%)
Justice status		< .001		< .0001
Justice-involved	7,672 (10%)		3,508 (46%)
Not justice-involved	45,896 (2%)		18,616 (41%)

^aPrevalence of opioid use disorder was calculated as the number of patients diagnosed with opioid use disorder divided by the total number of patients, separately by each characteristic.

^bPrevalence of receipt was calculated as the number of patients who received medications for opioid use disorder divided by the number of patients diagnosed with opioid use disorder, separately for each characteristic.

Receipt of Medications for Opioid Use Disorder by Patient and Facility Characteristics

The overall prevalence of receipt of medications for opioid use disorder was 41% in FY17 (Table 1). In a fully adjusted mixed effects logistic regression model each vulnerable population had lower odds of receiving medications for opioid use disorder compared to their non-vulnerable counterparts (Table 2). Women (adjusted odds ratio [AOR] = 0.87, 95% confidence interval [CI] = 0.81–0.94), veterans ages 45–54 (AOR = 0.57, 95% CI = 0.53–0.61) and 55 and older (AOR = 0.44, 95% CI = 0.41–0.46), Black veterans (AOR = 0.89, 95% CI = 0.84–0.94), rural veterans (AOR = 0.81, 95% CI = 0.77–0.85), homeless veterans (AOR = 0.69, 95% CI = 0.66–0.72), and justice-involved veterans (AOR = 0.80, 95% CI = 0.76–0.85) all had lower odds of receiving medications.

Table 2.

Adjusted Odds Ratios of Receipt of Medications for Opioid Use Disorder by Vulnerable Groups and Facility Characteristics in the United States Veterans Health Administration in Fiscal Year 2017

Vulnerable Groups	Unadjusted Odds Ratio	p	95% Confidence Interval	Adjusted Odds Ratio	p	95% Confidence Interval
Women (ref: men)	0.89	.00	0.83–0.96	0.87	< .001	0.81–0.94
Age (ref: <35)
35–44	0.93	.02	0.87–0.99	0.94	.06	0.88–1.00
45–54	0.56	<.001	0.52–0.60	0.57	< .001	0.53–0.61
55+	0.41	<.001	0.39–0.43	0.44	< .001	0.41–0.46
Race/Ethnicity (ref: White)
American Indian/Alaskan Native	0.86	.12	0.71–1.04	0.85	.11	0.70–1.04
Asian	0.85	.10	0.70–1.03	0.87	.16	0.71–1.06
Black	0.88	<.001	0.83–0.93	0.89	< .001	0.84–0.94
Hispanic	0.97	.49	0.89–1.06	0.94	.20	0.86–1.03
Rural (ref: urban)	0.80	<.001	0.76–0.84	0.81	< .001	0.77–0.85
Homeless (ref: housed)	1.13	<.001	1.09–1.17	0.69	< .001	0.66–0.72
Justice status (ref: none)	0.98	.40	0.92–1.03	0.80	< .001	0.76–0.85
Facility Characteristics
Facility size	1.08	.20	0.96–1.22	1.10	.16	0.96–1.25
Prevalence of opioid use disorder	1.27	.00	1.11–1.45	1.24	< .001	1.08–1.43
Prevalence of women	0.91	.33	0.75–1.10	0.91	.39	0.74–1.12
Prevalence of racial/ethnic minority patients	1.12	.17	0.95–1.33	1.09	.34	0.91–1.30
Rural location (ref: urban)	0.79	.08	0.62–1.02	0.78	.08	0.60–1.03
Facility region (ref: Northeast)
Southern region	0.65	.01	0.47–0.90	0.66	.02	0.48–0.92
Midwest region	0.88	.49	0.62–1.26	0.86	.40	0.60–1.22
West region	0.65	.03	0.45–0.95	0.72	.10	0.49–1.05
Average of substance use disorder visits	1.08	.23	0.95–1.23	1.05	.49	0.92–1.20
Prevalence of medications for alcohol use disorder	1.13	.09	0.98–1.30	1.12	.14	0.96–1.30
Available inpatient detox	0.92	.74	0.57–1.50	0.93	.78	0.55–1.57
Available substance use disorder residential treatment	1.04	.73	0.83–1.30	0.93	.55	0.74–1.18
Available methadone clinic	1.14	.26	0.91–1.43	1.16	.23	0.91–1.48

Note: Cases with missing data (n=2,990; 6%) were excluded from the mixed-effects logistic regression model. The model was adjusted for additional patient characteristics of marital status, service-connected disability rating, co-occurring mental health disorder, co-occurring substance use disorder, Deyo comorbidity index, and receipt of substance use disorder residential treatment.

Most facility characteristics, such as facility size and rural location, were not significantly associated with receipt of medications for opioid use disorder (Table 2). Compared to facilities with a lower prevalence of patients with opioid use disorder, veterans at facilities with a higher prevalence of patients with opioid use disorder had higher odds of receiving medications (AOR =1.24, 95% CI = 1.08–1.43). Facility location in the Southern region of the United States was associated with lower odds of receipt of medications compared to the Northeast region (AOR = 0.66, 95% CI = 0.48–0.92).

Discussion

Among all VHA patients with opioid use disorder nationally, several vulnerable populations had lower receipt of medications, including women veterans, older veterans, Black veterans, rural veterans, homeless veterans, and justice-involved veterans. These patient-level results are consistent with prior studies (Oliva et al. 2012, Finlay et al. 2016), suggesting that differences in receipt of medications for opioid use disorder have not changed over time for vulnerable populations. Facility factors were largely unrelated to increased receipt of medications, except for facilities that served more veterans with opioid use disorder or were in the Northeast compared to the South. While the association with region was consistent with a prior study (Morgan et al. 2018), the lack of association with other facility factors is a notable change from prior research, which found that more available substance use disorder treatment services and the prevalence of racial/ethnic minority patients were associated with higher odds of receiving medications for opioid use disorder (Knudsen and Roman 2009, Oliva et al. 2012, Finlay et al. 2016). Examination of the moderating effects of facility-level characteristics on patient-level disparities may be a potential avenue for future research.

Since the last national study of VHA patients using FY2012 data (Finlay et al. 2016), the prevalence of receipt of medications for opioid use disorder increased overall from 33% to 41%. There have been numerous efforts to increase uptake of these medications within VHA, such as an educational campaign by the Office of Academic Detailing, sponsored trainings to increase the number of physicians with buprenorphine waivers, and a national consult service to improve receipt of buprenorphine (see Wyse et al. 2018 for full summary). These efforts may explain the increase in use of medications observed. More broadly, a national emphasis on the opioid epidemic and rapid changes in policies related to medications for opioid use disorder may have also impacted use of this treatment modality. However, there was little change from five years ago in the difference between vulnerable and non-vulnerable populations.

Contrary to a study with Black veterans indicating they were more likely to initiate and engage in specialty addiction psychosocial treatment compared to White veterans (Bensley et al. 2017), our study demonstrated that Black veterans had lower odds of receiving medications. These results suggest there may be unique barriers to initiation of medications for opioid use disorder among Black veterans, such as a preference for non-medication treatment. For veterans in rural areas, the lower receipt of care may be due to a lack of prescribers available in their communities (Quest et al. 2012). Telehealth is becoming a more popular treatment modality (Benavides-Vaello et al. 2013), but requirements of initial in-person induction of buprenorphine may limit uptake for patients with opioid use disorder.

Notable, both homeless and justice-involved veterans had a much higher prevalence of opioid use disorder (10%) than other vulnerable groups (1–4%). Some people become homeless or engage in criminal activity because of their opioid use (Chatterjee et al. 2018), underscoring the importance of providing treatment for these populations. However, both groups experienced lower receipt of medications for opioid use disorder compared to housed and non-justice-involved veterans. One promising strategy to improve initiation of medication use for homeless veterans is mobile opioid agonist treatment units, which have enrolled more homeless adults than traditional fixed-site methadone clinics (Hall et al. 2014). Shared medical appointments for buprenorphine, where patients with opioid use disorder were seen together in a small group setting, have been shown to retain homeless patients in treatment (Doorley et al. 2017) and have been used effectively in VHA for medications for alcohol use disorder (Robinson et al. 2013). For justice-involved veterans, medications for opioid use disorder are prohibited in some criminal justice settings (Olsen and Sharfstein 2014) and they may be reluctant to start medications that they may have to withdraw from if they become incarcerated (Fox et al. 2015). Finally, some buprenorphine providers stated that they are willing to terminate treatment among patients for whom they have concerns about selling or non-prescribed use of medications for opioid use disorder (Lin et al. 2018), concerns that may be elevated for homeless or justice-involved veterans. Future research identifying the mechanisms that explain differences in medication use for homeless and justice-involved veterans are particularly important to developing interventions that address the treatment barriers they experience.

Facility factors indicate that facilities that diagnose more patients with opioid use disorder are more effective at connecting veterans to medications for opioid use disorder. These facilities may be more experienced at caring for veterans with opioid use disorder or may have developed more staff and resources to address the treatment of opioid use disorder. A consultation program to improve adoption of clinical guidelines for opioid prescribing was shown to be feasible and acceptable to primary care physicians (Quanbeck et al. 2018); a similar program could be developed for medications for opioid use disorder to improve prescribing, with physicians at high diagnosing facilities consulting with less experienced facilities. Expansion of the availability of these medications in primary care as well as specialty care settings is recommended (Volkow et al. 2019). Although availability of a substance use disorder residential treatment program was not a significant predictor, participation in a residential program may provide more opportunities for a veteran to learn about and be connected with medications (Finlay et al. 2018b). Direct-to-patient materials and other advertising efforts may also help promote medication use.

Limitations

There are several limitations to our study. First, our sample only included veterans with an opioid use disorder diagnosis in their electronic health record. There may be vulnerable populations within the VHA system that are under- or over-diagnosed for opioid use disorder. For example, 15% more women screened positive for unhealthy alcohol use when a lower gender-tailored threshold was used (Hoggatt et al. 2018), but it is unknown whether women are under-diagnosed for opioid use disorder. Second, our study only included medications that were received in or paid for by the VHA health care system. Veterans who received medications outside the VHA and that was not paid for by the VHA as part of contracted care may have been missed and the extent to which medication use outside VHA varies by different populations is unknown. Third, pharmacy records only permit us to determine which veterans filled a prescription, but we cannot determine if they used the medications as directed. Fourth, we used only one year of data; however, because we have alignment with a prior publication, there is some extrapolation we can make about changes over times. Finally, our results may not generalize to veterans who do not use VHA health care nor to the general population.

Conclusions

Vulnerable populations of veterans at the VHA appear to be underserved in their receipt of medications for opioid use disorder. For homeless and justice-involved veterans who have a substantially higher prevalence of opioid use disorder, quality improvement efforts to increase their use to these medications are crucial. Monitoring of vulnerable populations to ensure they are receiving the same care quality as other populations served at the VHA will also be valuable.