University Researcher and Law Enforcement Collaboration: Lessons From a Study of Justice-Involved Persons With Suspected Mental Illness

Abstract

In 2012, heads of local law enforcement agencies in Benton County, Oregon, contacted researchers at Oregon State University to discuss a problem: a sharp rise in the number of contacts between police and suspects displaying symptoms of mental illness. This initial inquiry led to an ongoing collaborative examination of the nature, causes, and consequences of the rise in police contacts. In this article, the authors describe this collaboration between researchers and law enforcement officials from the perspective of both parties, situating it within the context of mental illness in the U.S. criminal justice system. The collaborators draw on firsthand experiences and prior collaborations to discuss the benefits of, challenges in, and recommendations for university-police research collaborations. Although such collaborations may pose challenges (related to relationship definition, data collection and analysis, outputs, and relationship maintenance), the potential benefits-for researchers and law enforcement agencies-are substantial.
© The Author(s) 2015.

Full text

Criminal Justice Policy Review
2016, Vol. 27(1) 97 –114
© 2014 SAGE Publications
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0887403414559268
cjp.sagepub.com
Article
Law Enforcement Response
to “Frequent Fliers”: An
Examination of High-
Frequency Contacts Between
Police and Justice-Involved
Persons With Mental Illness
Scott Akins1, Brett C. Burkhardt1, and Charles Lanfear2
Abstract
This article examines a subset of justice-involved persons with mental illness who
have repeated contacts with law enforcement officers. Previous work has alluded
to this sub-population—often termed “frequent fliers”—but little research has
empirically examined its size and nature. This study proposes a method of identifying
frequent fliers that is based on the amount of time elapsed between multiple mental-
health-related contacts with police. Using more or less stringent thresholds, the
analysis defines several groups of frequent fliers, including rapid cyclers, those having
very frequent contacts with police. In considering policy responses to the problem
of justice-involved persons with mental illness, addressing the needs of the frequent
flier population proves to be a way of targeting limited resources for the most impact.
Keywords
criminal justice policy, habitual offenders, research and policy, treatment, police
decision making
The disproportionate rate of arrest and incarceration of people with mental illnesses
(PwMI) is an issue of growing concern of police, policymakers, and academic
researchers throughout the United States (Reuland, Schwarzefeld, & Draper, 2009;
1Oregon State University, Corvallis, USA
2University of Washington, Seattle, USA
Corresponding Author:
Scott Akins, Oregon State University, Fairbanks Hall, Corvallis, OR 97331, USA.
Email: sakins@oregonstate.edu
559268CJPXXX10.1177/0887403414559268Criminal Justice Policy ReviewAkins et al.
research-article2014
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

98 Criminal Justice Policy Review 27(1)
Teller, Munetz, Gil, & Ritter, 2006).1 While those with mental illnesses that severely
compromise unassisted living constitute at most 5% of the general population, they are
disproportionately represented at multiple levels of the justice system. The prevalence
of mental-health-related police contacts has been found to vary significantly by locale;
several studies have found PwMI to be involved in between 7% and 10% of all police
contacts (e.g., Borum, Swanson, Swartz, & Hiday, 1997), though others have found
the proportion of such contacts to be significantly lower (Engel & Silver, 2001) or
higher (White, Goldkamp, & Campbell, 2006). In addition, PwMI represent at least
16% of the U.S. prison and jail population (Torrey, Kennard, Eslinger, Lamb, & Pavle,
2010). The overrepresentation of this population at various levels of the justice system
has been attributed to several factors, including deinstitutionalization (Lamb, 1998;
Slovenko, 2013), cutbacks in federal mental health funding (Teplin, 2000), and
enforcement consequences of the war on drugs (Honberg & Gruttadaro, 2005; Lurigio,
2001). While there is some disagreement over the relative importance of the causes,
observers agree that the failure to coordinate the services of local mental health, sub-
stance use, and criminal justice agencies is an important factor exacerbating the ongo-
ing problem of justice-involved PwMI (Honberg & Gruttadaro, 2005; Lurigio, 2001;
Reuland et al., 2009).
Research informing contacts between law enforcement and PwMI is important for
a number of reasons. From the perspective of police, such contacts are often frustrat-
ing, time-consuming and, on occasion, may escalate into volatile and potentially vio-
lent situations, placing all parties at risk (Reuland et al., 2009). Law enforcement
officials have long been called on as first responders to situations in which people are
having crises related to mental illness (Bittner, 1967), but the prevalence of such con-
tacts appears to be increasing (Santos & Goode, 2014; Teplin & Pruett, 1992), and the
nature of these interactions is distinct from those more commonly handled by police
(Hoover, 2007). Although the police are charged with the responsibility to protect the
safety and welfare of the public by removing dangerous persons from the community,
they are also charged with providing protection for vulnerable citizens, including
those with mental illness or those in a state of mental crisis (Teplin & Pruett, 1992).
When responding to mental crisis calls, police typically have three options: they may
execute a formal arrest, they may detain the person and transport him or her to a men-
tal health facility, or they can resolve the situation informally. Determining which
response is most appropriate often places police in the role of a “street-corner psychia-
trist” (Teplin, 1984), something police often report feeling ill-prepared to do (Franz &
Borum, 2011). From the perspective of those accessing mental health services and
their loved ones, the limited options available to individuals needing help can place
PwMI at heightened risk of justice system involvement and, most tragically, situations
in which people are injured or killed when due to illness they fail to comply with
police commands and/or present a perceived threat to officer safety (Police Executive
Research Forum, 2012; Santos & Goode, 2014). Across parties, there appears to be
agreement that the “traditional police response” to persons in mental crisis neither
improves the mental state of the person being contacted nor facilitates the safe and
controlled resolution of the call for service (Reuland, Draper, & Norton, 2010).
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

Akins et al. 99
Frequent Fliers
The focus of the present study is on a subset of justice-involved PwMI: those who
have repetitive and frequent (sometimes very frequent) contacts with police due to
their mental illness. Commonly referred to as frequent fliers2 in law enforcement cir-
cles (Santos & Goode, 2014), these individuals often cycle between jail, halfway
houses, hospital emergency rooms, to the streets, and back again. Frequent fliers are
thought to be a relatively small subset of the broader justice-involved PwMI popula-
tion (Reuland et al., 2009). They may be disproportionately likely to be homeless
(Green, 1997) and dual diagnosis mental health and substance use disorder (White
et al., 2006) as compared with other persons contacted by police for mental health
reasons.
Although anecdotally reported as a population of particular concern (Santos &
Goode, 2014; Szabo, 2014), few studies have empirically analyzed the size and nature
of the frequent flier population. Green (1997) documented that a majority (63.5%) of
police contacts with PwMI in Honolulu were with individuals “known on sight” by
police, likely indicating some level of repetitive contact. Similarly, the Los Angeles
Police Department identified 67 PwMI involved in a total of 536 calls for service in an
8-month span in 2004 (in Reuland et al., 2009). The Houston Police Department
(2010) identified 30 PwMI who generated 194 offense reports and 165 Emergency
Detention Orders in a span of 6 months. The most rigorous analysis of frequent fliers
comes from White et al. (2006), who randomly sampled individuals taken into police
custody for either an arrest, a protective custody hold (commonly intoxication), or a
mental health hold in Santa Fe, New Mexico. They found that those individuals with
multiple prior holds, and those with mental health and substance abuse problems, were
significantly more likely to experience an arrest or an involuntary hold in the future
(White et al., 2006).3
Although the available information on the topic is primarily anecdotal, there is
evidence to suggest that the frequent flier population is comparatively small but gener-
ates a high, sometimes very high, frequency of contacts (e.g., Houston Police
Department, 2010). Because all police contacts with PwMI take significantly longer to
resolve, and often require more specialized training than “traditional” police contacts
(Reuland et al., 2009), the frequent flier population may generate substantial cost in
terms of officer hours invested and expenses related to incarceration (White et al.,
2006). Furthermore, these individuals appear to heavily access other social service
agencies, including emergency departments. As one example, over a recent 6-year
span in Austin, Texas, nine patients made 2,678 visits to Austin emergency depart-
ments at a cost of more than 3 million U.S. dollars. Eight of the nine patients were
substance abusers, seven of the nine were mentally ill, and three were homeless
(Associated Press, 2009). In sum, the frequent flier population appears to be relatively
small but very “high cost,” making policy recommendations needed and feasible.
The contribution of the present study is to propose a method of identifying frequent
fliers by calculating the amount of time elapsed between a PwMI’s multiple contacts
with police. Using more or less stringent thresholds, the analysis defines several
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

100 Criminal Justice Policy Review 27(1)
groups of frequent fliers, including rapid cyclers, those with very frequent contacts
with police. Once frequent fliers are identified, descriptive analyses will document the
size of the frequent flier population and its contact with police in one county in Oregon.
Method
Research Location: Benton County, Oregon
Benton County is in the central Willamette Valley region of western Oregon. The
county has approximately 86,000 residents, the majority of them live in the county
seat, Corvallis, which is the location of Oregon State University. In 2012, heads of
local law enforcement in Benton County, Oregon, requested a meeting with research-
ers at Oregon State University to discuss a collaborative investigation of the amount
of contact between local law enforcement and suspects displaying symptoms of men-
tal illness, prompting the work described in this study.
As in other places, police in Benton County have limited options when dealing with
PwMI. They may resolve the matter informally, arrest the person if they have commit-
ted a crime, or perform a peace officer custody (POC), which is a type of arrest that
occurs because an individual is believed to be a danger to self or others due to mental
illness. According to Oregon Revised Statute 426.228, the officer completing a POC
is directed to take the individual detained to the nearest hospital or non-hospital facil-
ity approved by the Oregon Health Authority.
Benton County generally and Corvallis in particular have a number of traits that
likely contribute to a larger than would be expected population of PwMI, particularly
those that are dual diagnosis and homeless. Corvallis is home to a major regional
medical center with an inpatient mental treatment facility. PwMI from a wide geo-
graphic area in Oregon are brought to this facility under the POC process described
above.4 Based on interviews with local officers and mental health officials, upon
release from the inpatient medical center, many individuals choose to remain in the
area, particularly those who have few or no ties in their place of origin (see Akins,
Burkhardt, Lanfear, Amorim, & Stevens, 2014). In addition, the city, being compara-
tively affluent, provides a relatively large range of services and housing for homeless
persons that may increase the mentally ill population (Akins et al., 2014).
Data
The analysis below relies on two distinct sets of data: arrests and incidents resolved
informally by police. The arrest data comprise all arrests (including charge informa-
tion) made by the Corvallis Police Department (CPD) or the Benton County Sheriff’s
Office (BCSO) in the 6 years between January 1, 2007, and December 31, 2012.5 Both
suspects and arresting officers were identified with random numbers to preserve ano-
nymity. The arrest data capture 13,650 unique suspects with 22,875 arrests and 33,064
charges.6 The analysis below examines a sub-sample of arrests that involve a suspect
perceived to have a mental illness. These individuals were identified in the data on the
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

Akins et al. 101
basis of having a POC charge in an arrest (described above). Within the arrest data,
there were 914 POC charges applied to 697 individuals. This data set allows examina-
tion of POCs charging in aggregate, as well as characteristics of individuals charged
with a POC. The current statutory authority for a POC arrest (ORS 426.228) was
instituted in 1994 and predates the beginning of our data by 13 years.
POC charging may be subject to net-widening, in which officers begin to use their
POC authorities in cases that previously would have produced no police action. As
such, it is important to know about incidents involving persons with mental illness that
do not result in a POC. As a complement to the POC data set, a second data set con-
tains all contacts that did not result in an arrest or case number (i.e., it omits POCs).
Despite not yielding an arrest, all informal contacts made by BCSO or CPD are
recorded in a database, which contains a wealth of information from the responding
officer and (where applicable) a 911 dispatcher. Informally resolved contacts involv-
ing a person suspected of having a mental illness were identified if they met one of two
criteria. First, responding officers included the word “mental” or the associated code,
“12-60,” in a free-text field of an incident report. Here, the “mental” designation is
based on officers’ subjective, non-clinical assessment of the situation. Second, a 911
dispatcher flagged the field “mental” in the computer-aided dispatch system.
Dispatchers for these agencies are trained to record information from the caller, and
thus the designation of a case as “mental” originates with subjective interpretation of
the situation by the caller. Using these criteria, the informal resolution data contain
1,388 informally resolved encounters with PwMI in the 6-year span. The informal
resolution data therefore complement the POC data. Combined, the two sets of data
should capture all known contacts—both formal and informal—in which the officer
and/or dispatcher records a mental health issue.
Results
Figure 1 depicts yearly counts of POCs, informal resolutions, and the ratio of POC to
non-POC arrests. Informally resolved contacts with persons with mental illness were
stable from 2007 to 2010, hovering around 200 per year. In 2011, informal resolutions
abruptly rose to over 300, a roughly 50% increase. Informal resolutions declined
slightly in 2012, but remained above the historical average. Formally resolved POC
arrests were also stable throughout much of the series, but they show a later rise.
Unlike informal resolutions, POCs increased dramatically in 2012, going from 144 to
245. The ratio of POC arrests to all other (non-POC) arrests rules out the possibility
that the rise in POC arrests was an artifact created by a rising overall arrest rate. The
rising ratio from 2011 to 2012 indicates that POC arrests were increasing faster than
non-POC arrests.
The increase in both POCs and informal resolutions translates to an increase in the
amount of time police spent on such interactions. The POC and informal resolution
data contain start and end times for each interaction, and these were used to calculate
the duration of each event. These durations were then aggregated to produce yearly
sums of hours spent responding to these incidents (Figure 2).7 Hours spent responding
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

102 Criminal Justice Policy Review 27(1)
closely track the number of POCs and informal resolutions seen in Figure 1. Durations
for both forms of response were relatively stable until 2011, when time spent on infor-
mal resolutions suddenly increased, followed by time spent on POCs the following
year. For the year 2012, the two major police agencies in Benton County spent nearly
2007 2008 2009 2010 2011 2012
Informal Hours 89.5 113.0147.7127.7 249.9184.0
POC Hours159.2174.1 171.6159.6 178.1305.0
0.0
100.0
200.0
300.0
400.0
500.0
600.0
Figure 2. Estimated total duration of POCs and informal resolutions.
Note. POCs = peace officer custodies.
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0
50
100
150
200
250
300
350
2007 2008 2009 2010 2011 2012
Rao of POCs to Other Arrests
POC Arrests and Informal Resoluons
Year
POC Arrests Informal Resoluons POC / Arrest Rao
Figure 1. Types of police contacts by year.
Note. POC = peace officer custody.
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

Akins et al. 103
500 hr responding to calls for service involving suspects perceived to have mental ill-
ness, twice the level from 2007.8
As noted above, frequent fliers are thought to be a relatively small subset of the
broader justice-involved PwMI population that prompts a high frequency of contacts
with law enforcement personnel. Analysis of the current data indicates that over the 6
years examined, 697 individuals received at least one POC. Of these, 117 individuals
received multiple POCs.9 These 117 individuals resulted in 334 POC arrests for an
average of 2.85 POCs per person over 6 years.
Previous work has not explicitly defined frequent fliers beyond saying that they are
justice-involved PwMI who have repeat contacts with law enforcement. Using data on
the timing of POC arrests, it is possible to precisely define the frequent flier popula-
tion. For each individual with multiple POC arrests in the data, an inter-POC span is
calculated as the difference between the current POC date and the prior POC date, if
one exists in the data. The distribution of these spans is depicted in Figure 3. It reveals
that many POC spans are very short. Nearly half (47.9%, or 104) of all repeat POC
arrests occurred within 60 days of the initial POC arrest. In fact, over a quarter (25.8%,
or 56) of repeat POC arrests occurred within just 14 days of the initial POC arrest.
The spans between POCs can be used as bandwidths for identifying frequent fli-
ers. The analyses below utilize three bandwidths of POC spans to identify frequent
010 20 30 40 50
Percent
0
60
120
180
240
300
360
420
480
540
600
660
720
780
840
900
960
1020
1080
1140
1200
1260
1320
1380
1440
1500
1560
1620
1680
1740
1800
1860
1920
1980
2040
2100
Days
Figure 3. Time elapsed since previous POC arrest.
Note. Initial POCs in the data are omitted due to inability to calculate time since last POC. POC = peace
officer custody.
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

104 Criminal Justice Policy Review 27(1)
fliers: (a) 365 days, (b) 60 days, and (c) 14 days. If an individual has two POC
arrests within the given bandwidth, he is classified as a frequent flyer for the entire
6-year period covered by the data. For example, an individual with one POC in 2008
and another 364 days later in 2009 would be counted as a 365-day frequent flyer for
the entire 2007-2012 period. Similarly, an individual with two POCs within a 14-day
span in 2007 would be counted as a 14-day frequent flyer for the entire period. (This
person would also qualify as a 60-day frequent flyer and a 365-day frequent flyer.)
Shorter bandwidths offer a stringent definition of frequent fliers and will only cap-
ture rapid cyclers, here defined as frequent fliers with two or more POCs in a 14-day
period.
For each bandwidth considered here, Table 1 depicts the number of frequent fliers,
number of POC arrests from frequent fliers, and the mean number of POCs in the data
for frequent fliers and non-frequent fliers. Narrowing the bandwidth that determines
frequent flier status reduces the count of frequent fliers and POCs but simultaneously
increases the rate of POC arrests. For example, while 365-day frequent fliers averaged
3.06 POC arrests in the data, 14-day frequent fliers averaged 3.68. Frequent fliers (of
all bandwidths) have a disproportionate effect on the total number of POC arrests. The
365-day frequent fliers represent 13.3% of all POC’ed individuals but 31.2% of all
POC arrests that occurred in the 6-year period under study. Similarly, the 14-day fre-
quent fliers (“rapid cyclers”) represent 5.5% of all POC’ed individuals but 15.3% of all
POC arrests.
The outsized contribution of frequent fliers to POC counts can be seen over time in
Figure 4, which graphs the annual number of POC individuals and arrests by frequent
flier status using various bandwidths. For all bandwidths, the numbers of POC arrests
and POC individuals track each other closely among non-frequent fliers. This is not
surprising, as a non-frequent flier will either have a single POC or, at most, multiple
POCs spread over long time. Among frequent fliers, however, there is a large and
growing divergence between the number of POC individuals and POC arrests. For
each bandwidth, the number of frequent flier-related POC arrests grew faster than the
number of frequent flier individuals. Consider the 14-day bandwidth in 2012: 19 fre-
quent fliers accounted for 65 POC arrests. Looking at the 365-day bandwidth in 2012,
47 frequent fliers contributed 108 POC arrests, nearly as many as contributed by the
Table 1. Counts of Individuals and POC Arrests by Frequent Flier Bandwidth.
POC span
bandwidth
(days)
Individuals Arrests
FFs % FFsaFF POCs % from FFsbFF mean POCs Non-FF mean POCs
365 93 13.34 285 31.18 3.06 1.04
60 65 9.33 216 23.63 3.32 1.15
14 38 5.45 140 15.32 3.68 1.17
Note. POC = peace officer custody; FFs = frequent fliers.
aPercentage of all individuals with a POC who are frequent fliers.
bPercentage of all POC arrests contributed by frequent fliers.
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

Akins et al. 105
137 non-frequent fliers (137 POC arrests). Thus, while the populations of both fre-
quent fliers and non-frequent flier individuals have grown, the nature of frequent fli-
ers—repeated POCs, often in rapid succession—means that they contribute
disproportionately to the total number of POCs.
It is likely that the POC figures shown here understate the true impact of frequent
fliers on law enforcement, as they omit informal resolutions. The informal resolution
data did not contain information on the contacted citizen, and thus could not be used
to identify (or match to) frequent fliers. Two plausible assumptions can be made about
the frequent fliers identified on the basis of repeated POCs: (a) they also have infor-
mally resolved contacts with law enforcement and (b) such contacts are more frequent
than individuals without repeated POCs. If these assumptions are correct, then fre-
quent fliers engage more law enforcement resources than what is suggested in the
analyses here.
Discussion
Despite the significant attention being directed to mentally ill and dual diagnosis jus-
tice-involved individuals, little explicit attention has been directed to so-called “fre-
quent fliers,” those justice-involved PwMI who have repeat, often high-frequency
contacts with law enforcement. This article proposed a simple method of identifying
frequent fliers using varying time bandwidths based on the difference between the cur-
rent and prior police contact (if one existed). Using this method of defining frequent
fliers, the article proceeded to document the disproportionate contribution that fre-
quent fliers made to the aggregate amount of contact between law enforcement and
PwMIs over a 6-year span in Benton County, Oregon.
The analyses revealed that for many individuals, the elapsed time between mental-
health-related police contacts is very short. Nearly half of all repeat POC arrests (an
indicator of police contact with PwMI) occurred within 60 days of the initial POC
0
50
100
150
200
2007
2008
2009
2010
2011
2012
2007
2008
2009
2010
2011
2012
2007
2008
2009
2010
2011
2012
14-day bandwidth60-day bandwidth 365-day bandwidth
POCs from non-FF POCs from FF Non-FF persons FF persons
Figure 4. POC arrests of individuals, by FF status and bandwidth.
Note. POC = peace officer custody; FF = frequent flier.
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

106 Criminal Justice Policy Review 27(1)
arrest and over a quarter of repeat POC arrests occurred within just 14 days of the
initial POC arrest. Results further showed that the 93 individuals with multiple POCs
in a year (365-day bandwidth) accounted for 285 POCs. Narrowing the bandwidth
used to define frequent fliers, the 38 individuals with multiple POCs in a 2-week
period (14-day bandwidth) accounted for 140 POC arrests. As noted above, these fig-
ures omit informal resolutions and therefore understate the true impact of frequent
fliers on law enforcement. This is consistent with existing research indicating that
PwMI who are regularly contacted by police are also significantly more likely to be
handled with “no action” (Green, 1997), as handling the situation in this way mini-
mizes paperwork and unwanted “down time” (Teplin, 1984, 2000). Thus, these results
confirm that a small subset of justice-involved PwMI disproportionately affect the
justice system.
Research indicates that contacts between justice-involved PwMI and police are
typically prompted by non-criminal behaviors or minor misdemeanors (Borum et al.,
1997), and limited research on the frequent flier subset of this population suggests this
behavior is primarily motivated by chronic, co-occurring mental health and substance
use disorders (Green, 1997; Houston Police Department, 2010; White et al., 2006).
Although, by default, law enforcement is typically the primary initial responder to
these individuals, failure to address the underlying conditions that led to their interac-
tions with law enforcement will waste limited justice system resources and will likely
exacerbate the mental health problems of the individual in the process (White et al.,
2006).
In the absence of a significant shift in policy addressing the non-institutionalized
mentally ill, significant numbers of individuals with mental health disorders and co-
occurring substance abuse will continue to be encountered by law enforcement (White
et al., 2006). Preventing PwMI from penetrating further into the criminal justice sys-
tem is a major challenge, but research suggests a number of steps that may help accom-
plish this. Most broadly, intensive collaboration between law enforcement agencies
and mental health agencies is a foundational step in thoroughly addressing the rise in
law enforcement contacts with PwMIs (e.g., Almquist & Dodd, 2009; Council of State
Governments, 2002; Deane, Steadman, Borum, Veysey, & Morrissey, 1999). Inter-
agency collaboration is not so much a discrete policy intervention, but rather an over-
arching philosophy that informs and facilitates various possible interventions.
Specifics may vary substantially by locale but this may involve regularly scheduled
meetings between agencies, mental health agents providing trainings on crisis inter-
vention, a shared case manager (or liaison) specializing in justice-involved mental
health cases, and formalized information sharing between mental health and law
enforcement on persons of high need.
Frequent fliers are likely to be citizens with high (and possibly unmet) needs. The
analytic methods for identifying frequent fliers could prove valuable for facilitating
knowledge exchange and cooperation between law enforcement and mental health
agencies with shared clients. Exchange of personal health information between agen-
cies is complicated by federal privacy regulations. Although the Health Insurance
Portability and Accountability Act (HIPAA) does place real restrictions on private
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

Akins et al. 107
health information sharing, it also offers allowances for disclosure of such information
to law enforcement in some instances (Petrila, 2007; Petrila & Fader-Towe, 2010).
Recently, Leon Rodriguez, Director of the Office for Civil Rights at the Department of
Health and Human Services, stated,
Privacy Rule [in HIPAA] does not prevent your ability to disclose necessary information
about a patient to law enforcement, family members of the patient, or other persons, when
you believe the patient presents a serious danger to himself or other people. (U.S.
Department of Health and Human Services, 2013, p. 1)
By using the frequent flier identification method described above to prioritize
those in the community with the highest need, law enforcement and mental health
can collaboratively determine what approaches are most promising for ensuring
future mental health and minimizing contact with police (e.g., Houston Police
Department, 2010).
For persons suffering from mental illness who have been charged criminally, men-
tal health courts provided a specialized venue to address treatment. Mental health
courts generally have a specialized docket of cases involving PwMI. They feature a
collaborative and non-adversarial team, comprising a judge, prosecutor, defense attor-
neys, representatives from parole and probation, and representatives from a mental
health agency (Almquist & Dodd, 2009; Sirotich, 2009). These parties can tailor a
response plan to fit the needs of the defendant, which may involve a referral to the
local mental health and substance abuse resources and may include compliance moni-
toring (Wolff, 2002).
Reviews of research on mental health courts provide reason for optimism.
Although the body of work on mental health courts is limited in terms of the number
of studies and their scope, some studies have found that participation in mental
health courts reduces recidivism or re-incarceration (see Almquist & Dodd, 2009;
DeMatteo, LaDuke, Locklair, & Heilbrun, 2013; Sarteschi, Vaughn, & Kim, 2011).
There is also evidence that mental health courts have positive mental health conse-
quences for participants (see Almquist & Dodd, 2009; DeMatteo et al., 2013),
although the evidence here is not definitive (Sarteschi et al., 2011; Sirotich, 2009).
And while mental health courts may require new expenses (e.g., court staff, addi-
tional treatment expenses), there is some evidence that these costs would be offset
by savings to the traditional criminal justice system, particularly in the form of
reduced frequency of jail stays for those with mental illness (Almquist & Dodd,
2009; Ridgely et al., 2007).10
While this study presents a straightforward methodology for identifying PwMI with
high-frequency contacts with law enforcement for intervention, there are notable limi-
tations and opportunities for future research. First, the analysis was conducted on a
single county in Oregon. Replication in other locales should be conducted to examine
variation in the size and impact of the frequent flier population in other areas. Identifying
and quantifying the impact of the frequent flier sub-population among all arrestees can
allow communities and law enforcement agencies to develop effective mitigation
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

108 Criminal Justice Policy Review 27(1)
strategies tailored to the local context, both in terms of resources available and scale of
the problem. Communities with very limited treatment resources, for instance, would
be best served by diverting the highest risk individuals with very narrow spans between
contacts with police. Large communities with more substantial resources may prefer to
use wider bandwidths to divert more individuals into community or residential treat-
ment programs. Second, as noted above, data limitations do not allow for a complete
analysis of repeated PwMI contacts that are resolved informally. The inability to calcu-
late repeated informal contacts means that the estimates above understate the true
amount of contact between police and PwMI. Third, the data analyzed here do not
capture individuals’ experience with other parts of the criminal justice system. Notably,
the data do not contain information on jail or prison spells. Long spans between con-
tacts with law enforcement may appear positive on paper (at least compared with con-
tacts in rapid succession). However, these long spells may simply be due to incarceration,
during which time a person cannot experience a police contact. Future work should
therefore be attentive not just to PwMI contacts with law enforcement but also to con-
tacts with carceral agencies. Finally, the costs of responding to individuals with mental
illnesses are often hidden in overall law enforcement budgets, obscuring the severity of
impacts of untreated mental illnesses on communities. Future cost analyses of police
contacts with these individuals, derived from service call duration data or similar met-
rics, may allow law enforcement agencies to better justify expansion of diversion pro-
grams or adjustment of police budgets to address these issues.
As frequent fliers, by definition, experience multiple contacts with law enforce-
ment, they are a critical sub-population in efforts to address the overall amount of
police contacts with PwMI. The method described above is a simple yet effective
means for police agencies or researchers to estimate the size of this population and to
target interventions, perhaps in collaboration with mental health service providers or
agencies. As the worlds of mental illness and criminal justice increasingly intersect,
addressing the frequent flier population proves to be a way of targeting limited
resources for the most impact.
Appendix
The article identifies frequent fliers on the basis of peace officer custody (POC) spans
(i.e., time elapsed between multiple POC arrests), such that individuals with a subse-
quent POC arrest within X number of days of a prior POC arrest were deemed frequent
fliers. We used several intuitive, a priori cutoff “bandwidths” to distinguish frequent
fliers from other POC arrestees: 14, 60, and 365 days. Using various bandwidths,
analysts can identify larger or smaller groups of frequent fliers, with smaller groups
having higher rates of POC arrest (see Figure 4). The value of this approach is its sim-
plicity: It is easy to understand, describe, and implement (e.g., in law enforcement or
mental health agencies).
More sophisticated analytic methods can also be used to identify frequent fliers.
We present one such approach here. Group-based trajectory modeling is a statistical
method for analyzing heterogeneity in growth trajectories in longitudinal data. It is
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

Akins et al. 109
based on an assumption that the growth trajectories within a population are hetero-
geneous and that they correspond to latent, unobserved groups within the population
(Nagin, 2005). The heterogeneity in trajectories is therefore modeled as a function of
group membership. (Group-based trajectory models are also referred to as “finite mix-
ture models” because the overall population trajectory is assumed to be a mixture of
a finite number of distinct trajectories—for example, Deb, 2008; Land, 2001.) Maxi-
mum likelihood algorithms use variation in developmental trajectories to estimate the
size of groups and the properties of each group-specific trajectory. The number of
groups and the general form of the trajectories are specified by the analyst. Multiple
models are compared, and an optimal model is selected, typically on the basis of the
Bayesian information criterion and theoretical plausibility (Nagin, 2005).
Because our data on POC arrests are longitudinal, with individuals observed over time,
they are amenable to group-based analysis of POC trajectories. The intuition behind such
an analysis is that within the overall population of POC arrestees, there are discrete sub-
groups that manifest distinct trajectories of POCs over time. One such group is likely to
consist of frequent fliers, who would be expected to have higher rates of POCs overall
and, perhaps, more rapid increases in POCs. We conducted a group-based analysis of
POC trajectories for all individuals with at least one POC in the 6 years of data available
(N = 697 individuals). Within this sample, we analyzed monthly frequencies of POCs for
each individual. Monthly POC frequencies for individuals ranged from 0 to 5, with the
vast majority of individuals having zero POC in a month. We used the Stata plug-in traj
to conduct the group-based analysis (Jones & Nagin, 2013; StataCorp LP, 2014). The
traj plug-in produces maximum likelihood estimates of group membership and group-
specific trajectories on the basis of user-specified link function, polynomial order, and
number of groups. The analysis below uses the zero-inflated Poisson (ZIP) link because
the outcome of interest is the count of POCs in a given month, which is skewed to the
right with a disproportionate number of zero outcomes. We present results for a model
with two groups with a cubic polynomial order.
Figure A1 summarizes the results of the model graphically, with monthly predicted
POC arrests (scatterplot) and over-time trajectories (lines) by group. The model splits
the sample into two groups, with 73.6% of POC arrestees in Group 1 and 26.4% of
POC arrestees in Group 2. Group 1 arrestees show a very slightly declining, but rela-
tively flat, trajectory over time. Group 2 arrestees show a rapidly rising trajectory, with
nearly zero POC arrest from 2007 through 2009, followed by rapidly rising numbers
of POC arrests through the end of 2012. By 2012, the predicted number of POCs for a
person in Group 2 approached (and eventually exceeded) 0.1 per month. Extrapolating
over a year, a Group 2 person is predicted to have 1.2 POCs per year (~0.1 per month
× 12 months). A person in Group 1, however, is predicted to have approximately 0.12
POCs per year (~0.01 per month × 12 months), roughly the same number of POCs that
a Group 2 person should have per month.
Group 2, generated through group-based trajectory modeling, resembles the fre-
quent fliers identified in the analysis based on POC spans. In both analyses, a small
subset of POC arrestees have frequent POC events and account for a large share of the
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

110 Criminal Justice Policy Review 27(1)
total POC load, especially in later years. The two methods are complementary and lead
to the same general conclusions: POC arrestees are not a homogeneous group, and
frequent fliers contribute disproportionately to law enforcement contacts with people
with mental illnesses (PwMI).
Acknowledgments
Chief Jon Sassaman (Corvallis Police Department), Sheriff Scott Jackson (Benton County
Sheriff’s Office), and Chief Ken Elwer (Philomath Police Department) provided helpful insights
into the local law enforcement context. Chief Sassaman and Jennifer Hendricks (Corvallis
Police Department) assisted in acquiring the data used in the analyses. Mariana Amorim and
Katelyn Stevens made valuable contributions to the larger project of which this article is a part.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship,
and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of
this article.
Notes
1. The term people with mental illnesses (PwMI) is used to refer to people who are perceived
by law enforcement agents as displaying symptoms of mental illness.
0.05.1 .15
POC arrests
2007m1
2007m7
2008m1
2008m7
2009m1
2009m7
2010m1
2010m7
2011m1
2011m7
2012m1
2012m7
2013m1
Calendar months
Group 1: 73.6%Group 2: 26.4%
Benton County, OR , 2007-2012
Predicted monthly POC arrests per person, by group
Figure A1. Predicted monthly POC arrests per person using a group-based trajectory
modeling approach.
Note. Finite mixture model produced with “traj” and “trajplot” in Stata 13. Model specified as two groups
with third-order polynomials and zero-inflated Poisson link function. Sample consists of 697 individuals
arrested for a POC in Benton County, 2007-2012. N = 50,184 person months. POC = peace officer
custody.
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

Akins et al. 111
2. The term “frequent flier” is used in public accounts (Santos & Goode, 2014) and anec-
dotally by police to refer to PwMI who have frequent contacts with law enforcement.
In addition, the term “frequent fliers” has been used in academic research (and anecdot-
ally) to refer to habitual offenders and/or those who commonly cycle through correctional
institutions, regardless of mental health status (Ford, 2005; Johnson & Willman, 2012).
Following convention, and for purposes of clear communication, we refer to “frequent
fliers” in referencing individuals who have repeated contact with law enforcement due to
a real or perceived mental illness. We use the phrase to simplify a complex concept, not to
trivialize persons suffering from mental illnesses or their heightened likelihood of coming
to the attention of law enforcement. This population has also been referred to as “chronic
consumers” by some (Houston Police Department, 2010).
3. An additional study by Biebel and Cordner (2003) noted the geographic concentration of
calls for service in response to a situation with a PwMI. Of the 507 such calls in Lexington,
Kentucky, in a 1-year span, 20% were attributed to just 17 locations, and each of these
locations received a minimum of three visits from law enforcement. Because institutional
residences (i.e., hospitals, shelters, group homes) were included in these 17 locations, it is
unclear to what extent repeat visits were caused by the same or multiple individuals.
4. This is particularly the case since a state-run mental hospital in Salem, a neighboring city,
was recently closed. As of this, writing facilities designed to accommodate some of those
displaced by this closure remain under construction.
5. This omits arrests performed by the Philomath Police Department (PPD) or Oregon State
Police (OSP), particularly on the Oregon State University campus where OSP have sole
jurisdiction.
6. In the original data set of 34,629 charges, 182 (0.5%) charges had an invalid suspect ID and
1,383 (4.0%) had data entry errors. These were removed from the analytic sample.
7. Due to data limitations, durations could only be established for 197 peace officer custodies
(POCs). For missing cases, yearly mean POC durations were imputed. Informal resolu-
tion durations were reported completely, but may be underestimates due to exclusion of
informal encounters that were not flagged as “mental” by either dispatchers or responding
officers.
8. The duration estimates assume a response by a single officer (the only measure available in
our data) and as such are conservative estimates of the total consumption of officer hours.
9. This is necessarily a conservative count, as it does not capture individuals with additional
POCs prior to 2007 or after 2012.
10. One might also consider the extensive literature on other problem-solving courts, most
notably drug courts, which supports their efficacy both in terms of reduced recidivism
(Mitchell, Wilson, Eggers, & MacKenzie, 2012) and cost savings (Downey & Roman,
2010).
References
Akins, S., Burkhardt, B., Lanfear, C., Amorim, M., & Stevens, K. (2014). Law enforcement
response to people with mental illnesses in Benton County. Retrieved from http://www.
co.benton.or.us/da/wcjc/documents.php
Almquist, L., & Dodd, E. (2009). Mental health courts: A guide to research-informed policy
and practice. New York, NY: Council of State Governments Justice Center.
Associated Press. (2009, April 1). 9 patients account for nearly 2,700 emergency room visits,
study finds. NBCNEWS.com. Retrieved from http://www.nbcnews.com/id/29998460/ns/
health-health_care/t/patients-made-nearly-er-visits-texas/
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

112 Criminal Justice Policy Review 27(1)
Biebel, E., & Cordner, G. (2003). Repeat calls for people with mental illness: An application of
hot-spots analysis. Police Forum, 13, 1-8.
Bittner, E. (1967). Police discretion in emergency apprehension of mentally ill persons. Social
Problems, 14, 278-292.
Borum, R., Swanson, J., Swartz, M., & Hiday, V. (1997). Substance abuse, violent behavior and
police encounters among persons with severe mental disorder. Journal of Contemporary
Criminal Justice, 13, 236-250.
Council of State Governments. (2002). The consensus project report. Retrieved from http://
csgjusticecenter.org/mental-health-projects/report-of-the-consensus-project/
Deane, M. W., Steadman, H. J., Borum, R., Veysey, B. M., & Morrissey, J. P. (1999).
Emerging partnerships between mental health and law enforcement. Psychiatric Services,
50, 99-101.
Deb, P. (2008). Finite mixture models. Retrieved from http://www.stata.com/meeting/snasug08/
deb_fmm_slides.pdf
DeMatteo, D., LaDuke, C., Locklair, B. R., & Heilbrun, K. (2013). Community-based alter-
natives for justice-involved individuals with severe mental illness: Diversion, problem-
solving courts, and reentry. Journal of Criminal Justice, 41, 64-71.
Downey, P. M., & Roman, J. K. (2010). A Bayesian meta-analysis of drug court cost-effective-
ness. Washington, DC: District of Columbia Crime Policy Institute.
Engel, R. S., & Silver, E. (2001). Policing mentally disordered suspects: A reexamination of the
criminalization hypothesis. Criminology, 39, 225-252.
Ford, M. C. (2005). Frequent fliers: The high demand user in local corrections. California
Journal of Health Promotion, 3, 61-71.
Franz, S., & Borum, R. (2011). Crisis intervention teams may prevent arrests of people with
mental illness. Police Practice & Research, 12, 265-272.
Green, T. M. (1997). Police as front line mental health workers. International Journal of Law
and Psychiatry, 20, 469-486.
Honberg, R., & Gruttadaro, D. (2005). Flawed mental health policies and the tragedy of crimi-
nalization. Corrections Today, 67, 22-27.
Hoover, L. T. (2007). Atypical situations-atypical responses. In T. J. Jurkanin, L. T. Hoover, &
V. A. Sergevnin (Eds.), Improving police response to persons with mental illness: A pro-
gressive approach (pp. 5-23). Springfield, IL: Charles C Thomas.
Houston Police Department. (2010). Chronic consumer stabilization initiative: A multi-agency
collaboration between the city of Houston health and human services department and the
mental health retardation authority of Harris County. Retrieved from http://www.popcen-
ter.org/library/awards/goldstein/2010/10-13(F).pdf
Johnson, O. & Willman, E. A. (2012). Frequent flyers: Potential hazards for law enforcement.
The Journal of Law Enforcement, 1, 1-7.
Jones, B. L., & Nagin, D. S. (2013). A note on a stata plugin for estimating group-based trajec-
tory models. Sociological Methods & Research, 42, 608-613.
Lamb, H. R. (1998). Deinstitutionalization at the beginning of the new millennium. Harvard
Review of Psychiatry, 6, 1-10.
Land, K. C. (2001). Introduction to the Special Issue on finite mixture models. Sociological
Methods & Research, 29, 275-281.
Lurigio, A. (2001). Effective services for parolees with mental illnesses. Crime & Delinquency,
47, 446-461.
Mitchell, O., Wilson, D. B., Eggers, A., & MacKenzie, D. L. (2012). Assessing the effective-
ness of drug courts on recidivism: A meta-analytic review of traditional and non-traditional
drug courts. Journal of Criminal Justice, 40, 60-71.
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

Akins et al. 113
Nagin, D. S. (2005). Group-based modeling of development. Cambridge, MA: Harvard
University Press.
Petrila, J. (2007). Dispelling the myths about information sharing between the mental health
and criminal justice systems. Retrieved from http://www.ncdhhs.gov/mhddsas/providers/
NCjaildiversion/ncjaildiv-infosharingmyths1-08.pdf
Petrila, J., & Fader-Towe, H. (2010). Information sharing in criminal justice-mental health col-
laborations: Working with HIPAA and other privacy laws. Retrieved from https://www.bja.
gov/Publications/CSG_CJMH_Info_Sharing.pdf
Police Executive Research Forum. (2012). An integrated approach to de-escalation and mini-
mizing use of force. Retrieved from http://www.policeforum.org/assets/docs/Critical_
Issues_Series/an integrated approach to de-escalation and minimizing use of force 2012.pdf
Reuland, M., Draper, L., & Norton, B. (2010). Improving responses to people with mental
illness: Tailoring law enforcement initiatives to individual jurisdictions. New York, NY:
Council of State Governments Justice Center.
Reuland, M., Schwarzefeld, M., & Draper L. (2009). Law enforcement responses to people
with mental illnesses: A guide to research-informed policy and practice. New York, NY:
Council of State Governments Justice Center.
Ridgely, M. S., Engberg, J., Greenberg, M. D., Turner, S., DeMartini, C., & Dembosky, J. W.
(2007). Justice, treatment, and cost: An evaluation of the fiscal impact of Allegheny County
mental health court. Santa Monica, CA: RAND.
Santos, F., & Goode, E. (2014, April 1). Police confront rising number of mentally ill suspects.
The New York Times. Retrieved from http://www.nytimes.com/2014/04/02/us/police-
shootings-of-mentally-ill-suspects-are-on-the-upswing.html?_r=0
Sarteschi, C. M., Vaughn, M. G., & Kim, K. (2011). Assessing the effectiveness of mental
health courts: A quantitative review. Journal of Criminal Justice, 39, 12-20.
Sirotich, F. (2009). The criminal justice outcomes of jail diversion programs for persons with
mental illness: A review of the evidence. Journal of the American Academy of Psychiatry
and the Law, 37, 461-472.
Slovenko, D. (2013). Deinstitutionalization of people with mental illness: Causes and conse-
quences. American Medical Association Journal of Ethnics, 15, 886-891.
StataCorp LP. (2014). Stata | Data Analysis and Statistical Software. Available from http://
www.stata.com/
Szabo, L. (2014, May 12). The cost of not caring, nowhere to go: The financial and human toll
for neglecting the mentally ill. The USA Today. Retrieved from http://www.usatoday.com/
longform/news/nation/2014/05/12/mental-health-system-crisis/7746535/
Teller, J., Munetz, M. R., Gil, K. M., & Ritter, C. (2006). Crisis intervention training for police
officers responding to mental disturbance calls. Psychiatric Services, 57, 232-237.
Teplin, L. A. (1984). Managing disorder: Police handling of the mentally ill. In L. A. Teplin
(Ed.), Mental health and criminal justice (pp. 157-175). Beverly Hills, CA: SAGE.
Teplin, L. A. (2000). Keeping the peace: Police discretion and mentally ill persons. National
Institute of Justice Journal, 244, 8-15.
Teplin, L. A., & Pruett, N. S. (1992). Police as streetcorner psychiatrist: Managing the mentally
ill. International Journal of Law and Psychiatry, 15, 139-156.
Torrey, E. F., Kennard, A. D., Eslinger, D., Lamb, R., & Pavle, J. (2010). More mentally ill per-
sons are in jails and prisons than hospitals: A survey of the states (A report issued jointly by
the National Sheriffs Association and Treatment Advocacy Center). Retrieved from http://
www.treatmentadvocacycenter.org/storage/documents/final_jails_v_hospitals_study.pdf
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from

114 Criminal Justice Policy Review 27(1)
U.S. Department of Health and Human Services. (2013). Message to our nation’s health care
providers. Retrieved from http://www.hhs.gov/ocr/office/lettertonationhcp.pdf
White, M. D., Goldkamp, J. S., & Campbell, S. P. (2006). Co-occurring mental illness and sub-
stance abuse in the criminal justice system: Some implications for local jurisdictions. The
Prison Journal, 86, 301-325.
Wolff, N. (2002). Courts as therapeutic agents: Thinking past the novelty of mental health
courts. Journal of the American Academy of Psychiatry and the Law, 30, 431-437.
Author Biographies
Scott Akins is an associate professor of Sociology in the School of Public Policy at Oregon
State University. His research interests include drug use and policy and structural criminology
including immigration, acculturation and deviance and the spatial effects of disadvantage, eth-
nicity and crime. His recent work has been published in Sociological Perspectives, Criminology
& Public Policy, Homicide Studies, Journal of Drug Issues and Justice Quarterly. With Clay
Mosher he is co-author of Drugs and Drug Policy: The Control of Consciousness Alteration.
Brett C. Burkhardt is an assistant professor of Sociology in the School of Public Policy at
Oregon State University. He is currently conducting research on the use of private prisons in the
United States and has previously written on topics including felon voting rights policies, labor
market consequences of felony convictions, and child support debt. His work has been pub-
lished in Race and Justice, Law & Social Inquiry, and the Journal of Policy Analysis and
Management.
Charles Lanfear is a PhD student in Sociology at the University of Washington. His interests
include substance use & abuse, mental illness, and disadvantage.
at OREGON STATE UNIV LIBRARY on April 28, 2016cjp.sagepub.comDownloaded from