Research

Revealed preference for

community and mail service
pharmacy
Joshua N. Liberman, Yun Wang, David S. Hutchins, Julie Slezak,
and Will H. Shrank

Received October 2, 2009, and in revised

form January 16, 2010. Accepted for publication January 20, 2010.

Abstract
Objective: To determine revealed pharmacy preference and predictors among
patients enrolled in a pharmacy benefit that offered a 90-day supply of prescriptions
via mail service and community pharmacy channels, with no differences in out-ofpocket costs.
Design: Retrospective cohort study.
Setting: United States in 200809.
Patients: 324,968 commercially insured participants enrolled in plans that required use of mail service pharmacy for maintenance medications.
Intervention: Implementation of a pharmacy benefit design with optional use of
either mail service or community pharmacy for 90-day supply prescriptions.
Main outcome measures: Selection rates of mail service and community pharmacy and adjusted odds ratios for predicting community pharmacy for selected characteristics.
Results: In the first 4 months of the benefit design, 31.8% of participants previously mandated to use mail service pharmacy elected to fill 90-day prescriptions
at community pharmacies. Selection of community pharmacy ranged from a low of
23.7% (previous mail service pharmacy users) to 66.3% (previous community pharmacy users). Among those initiating therapy, 44.3% selected community pharmacy
for their new prescriptions, and among those with no previous mail use, 68% selected
community pharmacy for new prescriptions. Preference for community/mail service
pharmacy was dependent on numerous characteristics, including age, gender, household income, region, driving distance (time), and concomitant medication use.
Conclusion: Patient behavior indicates that certain patients prefer to access
prescription medications via mail service and others through community pharmacy
channels. Restrictive benefit designs that incentivize patients to use less preferable
pharmacy channels may adversely affect patient convenience, which could have the
unintended consequence of reducing medication use and adherence.
Keywords: Pharmacy benefit design, revealed preference, mail service pharmacy.
J Am Pharm Assoc. 2011;51:5057.
doi: 10.1331/JAPhA.2011.09161

50 JAPhA 51 : 1 J a n / F e b 2011

50

www.japha.org

Joshua N. Liberman, PhD, is Vice President,

Strategic Research, CVS Caremark, Hunt
Valley, MD. Yun Wang, PhD, was Senior
Research Analyst, Strategic Research, CVS
Caremark, Hunt Valley, MD, at the time this
study was conducted; she is currently Senior
Analyst, Fulcrum Analytics, Fairfield, CT.
David S. Hutchins, MHSA, MBA, is Senior
Research Advisor, Strategic Research, CVS
Caremark, Scottsdale, AZ. Julie Slezak, MS,
is Vice President, Enterprise Analytics, CVS
Caremark , Northbrook, IL. Will H. Shrank,
MD, is Assistant Professor, Division of Pharmacoepidemiology and Pharmacoeconomics, Brigham and Womens Hospital, Harvard Medical School, Boston.
Correspondence: Joshua N. Liberman, PhD,
CVS Caremark, 11311 McCormick Rd., Suite
230, Hunt Valley, MD 21031. Fax: 410-7858140. E-mail: josh.liberman@caremark.com
Disclosure: Dr. Liberman, Mr. Hutchins, and
Ms. Slezak are employees and shareholders
of CVS Caremark. Dr. Shrank was supported
by a research grant from CVS Caremark. Dr.
Wang declares no conflicts of interest or financial interests in any product or service
mentioned in this article, including grants,
employment, gifts, stock holdings, or honoraria.
Funding: CVS Caremark. Dr. Shrank is supported by a career development award
(K23HL090505) from the National Heart,
Lung, and Blood Institute.

Journal of the American Pharmacists Association

1/10/11 4:31 PM

revealed pharmacy preference Research

educing barriers to the appropriate use of and adherence to essential medications for chronic conditions is
of paramount importance to insurers and policy makers.
Patients may encounter numerous barriers to medication use,
including elevated out-of-pocket costs, inadequate medication
education, and complexity of therapy.1 One barrier that pharmacy benefits managers (PBMs) can influence is convenience
of access to medications.
For patients who are frail, have limited transportation
services, prefer receiving 90-day supplies, or prefer the convenience of home delivery, mail service pharmacies may simplify access to essential medications. For patients who prefer
the convenience of a community pharmacy, prefer face-to-face
interactions with pharmacists, or are intimidated by the enrollment or ordering process in a mail service pharmacy system,
community pharmacies may be a preferable distribution channel for medications.2
PBMs must weigh the potential value of increased convenience against the potential costs of providing access through
different distribution channels. Increasingly, mail service
pharmacy options are promoted as a way to improve efficiency
and reduce prescription drug costs for payers, and 96.7% of
employers now offer a mail service pharmacy option.3 A 2003
Government Accounting Office report highlighted the financial
value of mail service pharmacy to organizations underwriting

At a Glance

Synopsis: Commercially insured patients previously

enrolled in plans requiring use of mail service pharmacy
for maintenance medications were given the option of using either mail service or community pharmacy channels
for 90-day prescriptions. In the initial 4 months of the
benefit design, 31.8% of patients previously required to
use mail service pharmacy elected to fill 90-day prescriptions at community pharmacies. Selecting the community
pharmacy option ranged from 23.7% (for previous mail
service pharmacy users) to 66.3% (for previous community pharmacy users). Age, gender, household income,
region, driving distance (time in minutes), and concomitant medication use were among the factors affecting patient selection of mail service or community pharmacy.
Analysis: The authors determined that recent use of
mail service pharmacy for another therapy was the most
important predictor of selecting the mail option for a new
prescription, suggesting that initial adoption may be the
greatest challenge to mail service pharmacy use. Driving distance to the nearest community pharmacy (in this
case, CVS/pharmacy) location was the strongest predictor of selecting community pharmacy for a new prescription, highlighting the perceived value of convenience.
This work illustrates that optimizing access to essential
medications should be a key function of pharmacy insurance benefits.

Journal of the American Pharmacists Association

51

pharmacy insurance.4 To encourage mail service pharmacy

use, many PBMs offer mandatory mail programs that limit the
use of community pharmacies.5 However, pharmacy benefits
that restrict choice or limit access, such as step therapy or elevated member share of the drug cost, may negatively affect
adherence to essential medications.6,7
Understanding patient preferences for the distribution
channel for prescription drugs is critical for PBMs to strike a
thoughtful balance between fostering convenience and controlling costs. A 2003 patient survey sponsored by the National Association of Chain Drug Stores reported that 72% of patients
opposed a mandatory mail service pharmacy benefit primarily
on the grounds that the design limits personal choice and access to local pharmacists, while 48% agreed that mail service
pharmacy was more convenient than community pharmacy access.8
Although reports of patients stated preferences for distribution channel are limited, no studies exist (to our knowledge)
of revealed preferencethe decisions that patients actually
make when given the choice between mail service and community pharmacy.

Objectives
We sought to determine revealed pharmacy preference and factors associated with pharmacy selection. The patients studied
were newly enrolled in a plan that allowed them to purchase
90-day supplies of chronic medications at either mail service
or community pharmacies, with no differences in out-of-pocket
costs for either channel. This provided us with an opportunity
to study, in isolation, patient preference for distribution channel. We studied actual behavior in cohorts of patients who had
previously used or not used mail service pharmacy. We also
evaluated predictors of behavior such as sociodemographic
characteristics and driving distance to the nearest participating pharmacy. A better understanding of patient preferences
and characteristics associated with those preferences will
have important implications for employers and insurers that
create benefit designs to support appropriate use and adherence.

Methods
This retrospective cohort study examined the pharmacy selection patterns of patients whose pharmacy benefit plan was
expanded from a mandatory or incentivized mail service pharmacy design in 2008 to a plan, beginning on January 1, 2009,
that required 90-day supply prescriptions for maintenance
medications but allowed the participant to select either a local community pharmacy (i.e., CVS/pharmacy) or mail service
pharmacy with no difference in copayment.
Mandatory and incentivized mail service pharmacy designs
maximize mail service pharmacy use by restricting access to
community pharmacies. Under mandatory mail service pharmacy design, participants can fill a prespecified number of
30-day prescriptions (typically two) before being required to
transfer the prescriptions to a 90-day supply dispensed by a
mail service pharmacy. Under incentivized mail service pharwww.japha.org

J a n /F e b 2011 51:1

JAPhA 51

1/10/11 4:31 PM

Research

revealed pharmacy preference

macy design, participants who reach the community pharmacy

prescription limit may only continue receiving 30-day supply
prescriptions at community pharmacy at an escalated copayment.
The 2009 pharmacy benefit provided plan participants with
the choice of receiving a 90-day supply of maintenance medications through either mail service pharmacy or their local CVS/
pharmacy for the same copayment. This benefit also was implemented via a mandatory or incentivized design. In the mandatory program, patients were allowed to fill a maximum number
of 30-day supply prescriptions (typically two) at a community
pharmacy and then were required to select either mail service
or a CVS/pharmacy location to receive subsequent 90-day supply prescriptions. In the incentivized program, patients who
reached the maximum allowable 30-day supply prescriptions
were invited to transition prescriptions to a 90-day supply (as
above) but were allowed the option of continuing with 30-day
supply prescriptions dispensed at the community pharmacy of
their choice for an additional copayment.
Study population
Three study cohorts (previous mail service pharmacy users,
previous community pharmacy users, and new to therapy) were
constructed for this analysis to test the hypothesis that substantially different pharmacy selection behavior could be expected between patients who did and did not previously receive
prescriptions via mail service pharmacy. To be included in a
cohort, patients had to have (1) either a mandatory or incentivized mail service pharmacy benefit design in 2008; (2) the
new benefit design implemented on January 1, 2009; (3) been
continuously eligible for pharmacy benefits from July 1, 2008,
through April 30, 2009; (4) a valid date of birth, home address,
and gender listed in the administrative records; and (5) filled a
90-day prescription for a drug within a selected maintenance
class (Table 1) between January 1, 2009, and April 30, 2009
(i.e., the evaluation period). Selected maintenance classes
were defined by Medi-Spans Generic Product Identifier (GPI).
Eligible study participants were assigned to a cohort based
on prescriptions filled during the baseline period (i.e., July 1,
2008, through December 31, 2008). Participants with at least
one prescription dispensed via mail service pharmacy were
included in the previous mail service pharmacy user cohort.
Participants with community pharmacydispensed prescriptions only were included in the previous community pharmacy
user cohort. Participants with no maintenance medication
prescription(s) in the baseline period were included in the
therapy initiator cohort.
Data analysis
Demographic characteristics of interest included age (in years,
as of January 1, 2009), gender, U.S. Census region of residence,
median household income (defined by ZIP Code, in calendar
year 2000 dollars, from the 2000 U.S. Census), and proximity to the nearest participating community pharmacy (defined
as driving time in minutes from the center of the residential
ZIP Code to the community pharmacy). Proximity was calculated using ArcGIS 9.3 Network Analyst (Esri, Redlands, CA).
52 JAPhA 51 : 1 J a n / F e b 2011

52

www.japha.org

Table 1. Prescription drug classes used in study of revealed

preference for community and mail service pharmacy
GPI4
2720
2725
2810
3320
3400
3610
3615
3699
3720
3760
3940
4420
4927
5685
5816
6610
7260
7970
8320
8515

Class name
Sulfonylureas
Biguanides
Thyroid hormones
Cardioselective beta blockers
Calcium channel blockers
Angiotensin-converting enzyme inhibitors
Angiotensin II receptor blockers
Antihypertensive combinations
Loop diuretics
Thiazide diuretics
HMG CoA reductase inhibitors (statins)
Sympathomimetics
Proton pump inhibitors
Prostatic hypertrophy agents
Selective serotonin reuptake inhibitors
Nonsteroidal anti-inflammatory agents
Anticonvulsants
Potassium
Coumarin anticoagulants
Platelet aggregation

Abbreviations used: GPI, Generic Product Identifier; HMG-CoA, 3-hydroxy-3-methylglutaryl coenzyme A.

Pharmacy benefit design was categorized in one of three ways,

representing the transition of former to current plan: incentivized to incentivized, incentivized to mandatory, and mandatory
to mandatory. Additional descriptive characteristics included
concurrent medication use (count of active maintenance medication prescriptions, defined by GPI4, with days supply ending no more than 30 days before the fill date of the 2009 index
prescription), previous mail service pharmacy use (at least one
paid pharmacy claim for a mail-dispensed prescription in the
baseline period), selection of a community pharmacy (yes, if
a patient filled the last 90-day prescription in the evaluation
period at a community pharmacy), and selection of mail service
pharmacy (yes, if the patient filled the last 90-day prescription
in 2009 at a mail service pharmacy).
Separate logistic regression models were constructed for
each cohort. The dependent variable was whether patients selected a community pharmacy for the final 90-day prescription
in 2009. Categories for the following independent variables
were included in the models: proximity to the nearest preferred
community pharmacy, age, gender, median household income,
region of residence, concomitant medication use (count of active GPI4 classes), pharmacy benefit design, and previous mail
use (therapy initiator cohort only).

Results
The study included a total of 324,968 patients. Previous mail
users (n = 239,668) were 53.4% women, had a mean age of 65
years, and had a mean of 3.9 active prescriptions for mainteJournal of the American Pharmacists Association

1/10/11 4:31 PM

revealed pharmacy preference Research

Table 2. Characteristics of study population in study assessing revealed preference for community and mail service pharmacy
Characteristic
n
Age (years)
<55
5564
6574
75
Gender
Women
Men
Region
Midwest
Northeast
South
West
Median household income ($)
<35,000
35,00040,000
40,00155,000
>55,000
Plan design transition
Incentivized to incentivized
Incentivized to mandatory
Mandatory to mandatory
No. active maintenance GPI4
0
12
34
5
Baseline mail service pharmacy use
No
Yes
CVS/pharmacy proximity (min)
5
610
1115
1620
2125
2630
>30

Therapy initiator
No. (%)
45,288

Previous mail service pharmacy user

No. (%)
239,668

Previous community pharmacy user

No. (%)
40,012

11,142 (24.6)
10,629 (23.5)
9,723 (21.5)
13,794 (30.5)

52,039 (21.7)
56,880 (23.7)
53,167 (22.2)
77,582 (32.4)

15,724 (39.3)
9,111 (22.8)
6,293 (15.7)
8,884 (22.2)

23,824 (52.6)
21,464 (47.4)

127,921 (53.4)
111,747 (46.6)

21,177 (52.9)
18,835 (47.1)

10,564 (23.3)
8,767 (19.4)
21,268 (47.0)
4,689 (10.4)

64,160 (26.8)
47,431 (19.8)
101,894 (42.5)
26,183 (10.9)

8,048 (20.1)
8,334 (20.8)
20,040 (50.1)
3,590 (9.0)

12,105 (26.7)
7,270 (16.1)
14,565 (32.2)
11,348 (25.1)

59,214 (24.7)
38,059 (15.9)
78,175 (32.6)
64,220 (26.8)

10,081 (25.2)
5,924 (14.8)
12,723 (31.8)
11,284 (28.2)

13,179 (29.1)
166 (0.4)
31,943 (70.5)

79,328 (33.1)
924 (0.4)
159,409 (66.5)

14,205 (35.5)
159 (0.4)
25,646 (64.1)

7,894 (17.4)
12,765 (28.2)
10,526 (23.2)
14,013 (30.9)

11,633 (4.9)
71,227 (29.7)
71,205 (29.7)
85,603 (35.7)

4,038 (10.1)
14,604 (36.5)
9,866 (24.7)
11,504 (28.8)

10,614 (23.4)
34,674 (76.6)

0
239,668 (100)

21,626 (54.0)
18,386 (46.0)

18,799 (41.5)
9,404 (20.8)
3,472 (7.7)
2,419 (5.3)
1,673 (3.7)
1,444 (3.2)
8,077 (17.8)

94,829 (39.6)
49,382 (20.6)
18,158 (7.6)
12,997 (5.4)
9,094 (3.8)
7,827 (3.3)
47,381 (19.8)

18,579 (46.4)
8,926 (22.3)
2,848 (7.1)
1,928 (4.8)
1,231 (3.1)
1,102 (2.8)
5,398 (13.5)

Abbreviation used: GPI, Generic Product Identifier.

All comparisons are statistically significant (P < 0.01).

nance medications during the baseline period. The most common drug classes were statins (filled by 37.1% of participants),
proton pump inhibitors (20.1%), angiotensin-converting enzyme inhibitors (17.7%), and thyroid hormones (15%). Previous community pharmacy users were substantially younger
and had fewer concomitant medications but were similar regarding other demographics. The therapy initiator cohort was
slightly younger (63 years of age) but demographically similar
to the previous mail service pharmacy cohort. However, the
Journal of the American Pharmacists Association

53

therapy initiator cohort used fewer concomitant medications

(3.3), and 23.4% had no prescriptions dispensed by mail service pharmacy at baseline. Approximately 40% of each population resided within 5 minutes of the nearest eligible community
pharmacy, and 20% resided between 6 and 10 minutes away;
the remainder of the population lived farther away (Table 2).
During the 4 months after the benefit design change, a total of 55,714 previous mail users (23.2%) elected to transfer
maintenance prescriptions from mail service to community
www.japha.org

J a n /F e b 2011 51:1

JAPhA 53

1/10/11 4:31 PM

54

54 JAPhA 51 : 1 J a n / F e b 2011

www.japha.org

10,408 (55.4)
5,053 (53.7)
1,629 (46.9)
1,013 (41.9)
619 (37.0)
479 (33.2)
881 (10.9)

8,391 (44.6)
4,351 (46.3)
1,843 (53.1)
1,406 (58.1)
1,054 (63.0)
965 (66.8)
7,196 (89.1)

5,431 (41.2)
81 (48.8)
14,570 (45.6)

7,748 (58.8)
85 (51.2)
17,373 (54.4)

7,210 (68.0)
12,780 (36.9)

4,919 (40.6)
2,980 (41.0)
6,441 (44.2)
5,742 (50.6)

7,186 (59.4)
4,290 (59.0)
8,124 (55.8)
5,606 (49.4)

3,389 (32.0)
21,817 (63.1)

3,694 (35.0)
4,415 (50.4)
10,538 (49.5)
1,435 (30.6)

6,870 (65.0)
4,352 (49.6)
10,730 (50.5)
3,254 (69.4)

4,651 (58.3)
6,160 (48.3)
4,324 (41.1)
4,947 (35.3)

10,193 (42.8)
9,889 (46.1)

13,631 (57.2)
11,575 (53.9)

3,333 (41.7)
6,605 (51.7)
6,202 (58.9)
9,066 (64.7)

6,149 (55.2)
5,118 (48.2)
3,978 (40.9)
4,837 (35.1)

4,993 (44.8)
5,511 (51.8)
5,745 (59.1)
8,957 (64.9)

65,398 (69.0)
34,532 (69.9)
13,514 (74.4)
10,086 (77.6)
7,474 (82.2)
6,595 (84.3)
45,370 (95.8)

0
182,969 (76.7)

8,027 (69.0)
54,964 (77.2)
54,454 (76.5)
65,524 (76.5)

62,404 (78.7)
627 (67.9)
119,931 (75.2)

46,555 (78.6)
29,531 (77.6)
59,456 (76.1)
47,427 (73.9)

52,438 (81.7)
34,853 (73.5)
73,244 (71.9)
22,434 (85.7)

98,684 (77.1)
84,285 (75.4)

38,765 (74.5)
42,879 (75.4)
40,188 (75.6)
61,137 (78.8)

29,431 (31.0)
14,850 (30.1)
4,644 (25.6)
2,911 (22.4)
1,620 (17.8)
1,232 (15.7)
2,011 (4.2)

0
55,714 (23.3)

3,606 (31.0)
16,263 (22.8)
16,751 (23.5)
20,079 (23.5)

16,924 (21.3)
297 (32.1)
39,478 (24.8)

12,659 (21.4)
8,528 (22.4)
18,719 (23.9)
16,793 (26.1)

11,722 (18.3)
12,578 (26.5)
28,650 (28.1)
3,749 (14.3)

29,237 (22.9)
27,462 (24.6)

13,274 (25.5)
14,001 (24.6)
12,979 (24.4)
16,445 (21.2)

Previous mail service pharmacy user

(n = 239,668)
Mail service pharmacy Community pharmacy
(n = 182,969 [76.3%])
(n = 56,669 [23.7%])
No. (%)
No. (%)

Abbreviation used: GPI, Generic Product Identifier.

All comparisons between mail service and community pharmacy selection were statistically significant (P < 0.01).

Characteristic
Age (years)
<55
5564
6574
75
Gender
Women
Men
Region
Midwest
Northeast
South
West
Median household income ($)
<35,000
35,00040,000
40,00155,000
>55,000
Plan design transition
Incentivized to incentivized
Incentivized to mandatory
Mandatory to mandatory
No. active maintenance GPI4
0
12
34
5
Baseline mail service pharmacy use
No
Yes
CVS/pharmacy proximity (min)
5
610
1115
1620
2125
2630
>30

Therapy initiator (n = 45,288)

Mail service pharmacy Community pharmacy
(n = 25,206 [55.7%])
(n = 20,082 [44.3%])
No. (%)
No. (%)

4,331 (23.3)
2,308 (25.9)
875 (30.7)
702 (36.4)
516 (41.9)
495 (44.9)
4,256 (78.8)

4,556 (21.1)
8,927 (48.6)

943 (23.4)
4,139 (28.3)
3,396 (34.4)
5,005 (43.5)

5,231 (36.8)
42 (26.4)
8,209 (32.0)

3,716 (36.9)
2,221 (37.5)
4,316 (33.9)
3,230 (28.6)

3,492 (43.4)
2,269 (27.2)
5,718 (28.5)
2,004 (55.8)

7,502 (35.4)
5,981 (31.8)

4,367 (27.8)
2,903 (31.9)
2,336 (37.1)
3,877 (43.6)

14,248 (76.7)
6,618 (74.1)
1,973 (69.3)
1,226 (63.6)
715 (58.1)
607 (55.1)
1,142 (21.2)

17,070 (78.9)
9,459 (51.4)

3,095 (76.6)
10,467 (71.7)
6,470 (65.6)
6,499 (56.5)

8,974 (63.2)
117 (73.6)
17,437 (68.0)

6,365 (63.1)
3,703 (62.5)
8,407 (66.1)
8,054 (71.4)

4,556 (56.6)
6,065 (72.8)
14,322 (71.5)
1,586 (44.2)

13,675 (64.6)
12,854 (68.2)

11,357 (72.2)
6,208 (68.1)
3,957 (62.9)
5,007 (56.4)

Previous community pharmacy user

(n = 40,012)
Mail service pharmacy
Community pharmacy
(n = 13,483 [33.7%])
(n = 26,529 [66.3%])
No. (%)
No. (%)

Table 3. Distribution of demographic, patient, and plan design characteristics by selection of mail service or community pharmacy for 90-day supply prescriptions, by population cohort

Research
revealed pharmacy preference

Journal of the American Pharmacists Association

1/10/11 4:31 PM

revealed pharmacy preference Research

pharmacy (Table 3). Of those with multiple prescriptions, only

17,182 (6.9%) elected to split prescriptions between distribution channels. Although statistical differences emerged,
the rate of community pharmacy selection was largely consistent across demographic characteristics, with community
pharmacy selection slightly more likely among men, individuals younger than 75 years, and individuals living in ZIP Codes
with higher household incomes. Transferring prescriptions to
community pharmacy was significantly less likely among those
with at least one concomitant chronic medication and as driving distance to the nearest community pharmacy increased. In
fact, nearly 30% of patients who lived within a 10-minute drive
of the nearest community pharmacy transferred prescriptions
to community pharmacy compared with only 4% of those with
a drive of 30 minutes or more. We also noted only modest differences in transfer rates across therapeutic classes, ranging
from a low of 18.2% (loop diuretics, thiazide diuretics, and
coumarin anticoagulants) to a high of 22.7% (anticonvulsants). After adjustment, all characteristics remained significant predictors of pharmacy selection, with driving distance
emerging as the most predictive of transferring (Table 4).
In contrast with the 23.2% of previous mail service pharmacy users who transferred prescriptions to community pharmacies, 66.3% of previous community pharmacy users elected
to have their 90-day supply prescriptions dispensed at community pharmacies (Tables 3 and 4). Patients whose benefit
design changed from incentivized mail to a mandatory 90-day
design had lower odds of selecting community pharmacy.
Younger adults, those with fewer concomitant medications,
and those residing in ZIP Codes with higher incomes were substantially more likely to select community pharmacy. In this
cohort, driving distance had an even greater effect on pharmacy selection, with individuals within a 5-minute drive selecting
community pharmacy 76.7% of the time.
Among the therapy initiators, 44.1% elected the community pharmacy channel for their new prescriptions, and among
those with no previous mail service pharmacy use, 67.9%
chose community pharmacy for new prescriptions (Table 3).
Patients who were men, of younger age, and resided in higherincome ZIP Codes were more likely to select community pharmacy. The strongest predictors of pharmacy selection were
recent use of mail service pharmacy and driving distance to the
nearest eligible community pharmacy, both of which remained
highly predictive after adjustment (Table 4). Patients who used
mail service pharmacy previously were nearly twice as likely to
choose mail distribution (63.1% vs. 36.9%, P < 0.0001) compared with those who had not (Table 3). Only 10.8% of patients
who lived 30 miles or more from a participating pharmacy
chose to use community pharmacy, compared with more than
55.1% of those who lived 5 miles or less from a participating
pharmacy (P < 0.0001) (Table 3).

Discussion
This study highlights that when pharmacy benefit design does
not preferentially support one pharmacy distribution channel,
both community pharmacy and mail service pharmacy options
Journal of the American Pharmacists Association

55

appeal to patients. Overall, within the first 4 months of the

new benefit design, approximately 32% of patients elected
community pharmacy for 90-day maintenance medications.
Among those who initiated therapy under the new benefit design, nearly equal proportions elected mail service and community pharmacy channels, while among those who previously
used community pharmacy, nearly 79% elected community
pharmacy if they had not recently used mail service pharmacy. We found that the most important predictor of selecting
mail service pharmacy for a new prescription was recent use
of mail for another therapy, suggesting that initial adoption
may be the greatest challenge to mail service use. The strongest predictor of selecting community pharmacy for a new
prescription was driving distance to the nearest community
pharmacy, highlighting the perceived value of convenience.
Moreover, all characteristics studied, including age, gender,
U.S. Census region of residence, and household income in ZIP
Code of residence, were significant predictors of pharmacy
selection, suggesting that restricting distribution channel affects different patient populations in different ways.
The decision to use mail service or community pharmacy is
influenced by both personal preferences and pharmacy benefit
design incentives. Patient surveys of pharmacy satisfaction
consistently indicated high levels of overall satisfaction with
both community and mail service pharmacy services.2,9 Community pharmacy patients cite satisfaction from face-to-face
pharmacist interactions, and both community and mail service
pharmacy users report satisfaction derived from financial savings (lowest pharmacy costs), confidence in prescription accuracy, trust, and convenience. However, the concept of convenience clearly differs between community and mail service
pharmacy patients. Community pharmacy patients may view
face-to-face conversations with pharmacists as an attribute
that imparts convenience, whereas mail service pharmacy patients may view an efficient online ordering process and home
delivery as convenient attributes. The perceived value of different pharmacy services provided by mail service and community pharmacies underscores the importance of pharmacy
choice.
Pharmacy choice, however, is also influenced by barriers
to access and financial (dis)incentives embedded in the pharmacy benefit design. Mail service pharmacy use is encouraged by direct financial incentives (e.g., discounted member
copays),3 incremental increases to community pharmacy copays (i.e., incentivized mail service pharmacy), or limiting the
allowable prescriptions dispensed at community pharmacies
(i.e., mandatory mail service pharmacy). Most employers include a financial incentive to use mail service pharmacy, usually in the form of a lower copayment for the additional days
supply3,10; however, 17.4% of employers mandate use of mail
service pharmacy for maintenance medications. Lower copayments increase the voluntary selection of mail service pharmacy moderately (and in proportion to the discount), while
mandating mail service pharmacy dramatically increases
mail service pharmacy selection.5 Because financial incentives alone fail to motivate the majority of patients to transwww.japha.org

J a n /F e b 2011 51:1

JAPhA 55

1/10/11 4:31 PM

Research

revealed pharmacy preference

Table 4. Adjusted odds for selecting community pharmacy for the distribution of 90-day supply maintenance medications, by
cohort
Characteristic
Baseline mail service pharmacy use
Yes (referent)
No
Gender
Men (referent)
Women
Age (years)
<55 (referent)
5564
6574
75
Median household income ($)
<40,000 (referent)
40,00055,000
>55,000
Concurrent maintenance drugs
0 (referent)
12
34
5
Region
South (referent)
Midwest
Northeast
West
Driving distance (min)
>30 (referent)
2630
2125
1620
1115
610
5
Plan design
Incentivized to incentivized
(referent)
Incentivized to mandatory
Mandatory to mandatory

Therapy initiators Previous mail service pharmacy users Previous community pharmacy users
OR (95% CI)
OR (95% CI)
OR (95% CI)

1.00
3.77 (3.544.01)

NA
NA

1.00
3.77 (3.573.98)

1.00
0.92 (0.880.96)

1.00
0.92 (0.900.94)

1.00
0.89 (0.850.93)

1.00
0.98 (0.931.05)
0.89 (0.830.94)
0.71 (0.670.76)

1.00
1.01 (0.981.04)
1.03 (1.001.06)
0.85 (0.830.88)

1.00
1.11 (1.041.19)
1.07 (0.991.15)
0.87 (0.810.93)

1.00
1.02 (0.971.07)
1.10 (1.041.16)

1.00
1.03 (1.001.05)
1.01 (0.981.03)

1.00
1.07 (1.011.13)
1.18 (1.111.26)

1.00
1.36 (1.271.46)
1.33 (1.231.43)
1.15 (1.061.24)

1.00
0.68 (0.650.71)
0.73 (0.700.77)
0.75 (0.720.79)

1.00
1.07 (0.981.17)
1.23 (1.121.35)
1.05 (0.951.16)

1.00
0.64 (0.600.67)
0.80 (0.750.84)
0.48 (0.450.52)

1.00
0.65 (0.630.66)
0.75 (0.730.77)
0.48 (0.460.50)

1.00
0.58 (0.550.62)
0.75 (0.700.80)
0.33 (0.300.36)

1.00
1.50 (1.311.73)
1.83 (1.602.08)
2.36 (2.122.62)
2.83 (2.603.08)
3.62 (3.383.87)
4.05 (3.824.29)

1.00
1.37 (1.271.48)
1.59 (1.481.70)
2.21 (2.092.33)
2.63 (2.522.74)
3.31 (3.203.43)
3.59 (3.483.69)

1.00
1.66 (1.431.94)
1.92 (1.662.22)
2.62 (2.322.96)
3.40 (3.083.74)
4.19 (3.894.52)
4.88 (4.595.19)

1.00
1.26 (0.911.76)
1.27 (1.201.35)

1.00
1.62 (1.411.88)
1.22 (1.181.26)

1.00
1.57 (1.062.31)
1.36 (1.291.43)

Abbreviations used: NA, not applicable; OR, odds ratio.

fer maintenance medications from community pharmacy to

mail service pharmacy,11,12 many patients elect to pay higher
out-of-pocket costs for prescriptions in order to receive them
at their community pharmacy. This may explain, in part, why
participants previously enrolled in mandatory mail service
pharmacy had higher rates of selecting (or returning) to community pharmacy after the benefit design changed. Optimizing
access to essential medications while managing pharmacy ex56 JAPhA 51 : 1 J a n / F e b 2011

56

www.japha.org

penditures is a key function of pharmacy insurance benefits.13

Pharmacy benefits that restrict access, such as dispense-aswritten penalties, step therapy, coinsurance, and increased
member share/copayment levels, can lead to lower rates of
medication use and adherence.57 Restricting pharmacy options, another potential barrier to access, has been identified
as an important predictor of pharmacy benefit choice14 and of
pharmacy benefit dissatisfaction.15
Journal of the American Pharmacists Association

1/10/11 4:31 PM

revealed pharmacy preference Research

To our knowledge, this is the first study to measure revealed

pharmacy preference (mail service versus community pharmacy), especially with a financially neutral benefit design in which
90-day prescriptions incurred the same cost via both channels.
Without a survey, we have little information on patient satisfaction, which is a potentially important predictor. However, to
the extent that driving distance is a surrogate for convenience,
these results align with previous research indicating that convenience is an important part of pharmacy preference and satisfaction.10,11 Although not explicitly studied, the demand for
community pharmacy services is a function of the time and associated costs (e.g., driving distance/time, travel costs, in-store
interactions/wait time). In our case, driving distance is a reasonable surrogate because the time component of mail service
pharmacy use (e.g., completing and submitting forms) is assumed to be fixed, without variability across patients. If driving
distance is a reasonable surrogate for the monetary value of
time, the results can be compared with time elasticity models
documenting the relationship between time and demand for
medical services.16,17

Limitations
Interpretation of our results is limited by several issues. We assessed patient selection of community or mail service pharmacy
from administrative pharmacy claims and did not directly survey patients. Community pharmacy convenience was assessed
by driving distance from a patients home address to the nearest
CVS/pharmacy location. However, the nearest and most convenient pharmacy may have been proximal to a workplace location (data unavailable in the administrative records) or to another community pharmacy. Because the community pharmacy
option was limited to CVS/pharmacy only, our results likely
underestimate the actual preference for community pharmacy.
We did not assess pharmacy channel switching behavior, which
would require substantially more follow-up time to profile thoroughly. Finally, our analysis was conducted in a population of
commercially insured beneficiaries, and we cannot generalize
our findings to uninsured patients or patients who are enrolled
in Medicaid or other government-sponsored insurance plans.

Conclusion
A pharmacy benefit design that allows patients the choice of
community or mail service pharmacy without any direct financial incentive (via copay differentials) resulted in selection of
both mail service and community pharmacy channels. These
results highlight the importance of constructing pharmacy
benefit designs that provide access to both channels in order
to support convenient, patient-centered access to medications.
Patient preference for access to medications via mail service
or community pharmacy depends on numerous characteristics,
including age, gender, household income, geographic region of
residence, driving distance (time) to community pharmacy location, and concomitant medication use. An important implication

Journal of the American Pharmacists Association

57

to monitor is the potential for benefit designs that mandate use

of a single pharmacy channel to affect prescription drug access
and adherence.
References
1. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med.
2005;353:48797.
2. Johnson JA, Coons SJ, Hays RD, et al. A comparison of satisfaction with mail versus traditional pharmacy services. J Manag
Care Pharm. 1997;3:32737.
3. Pharmacy Benefit Management Institute. Prescription drug benefit cost and plan design report: 2009-2010 edition. Scottsdale,
AZ: Pharmacy Benefit Management Institute; 2009:8.
4. U.S. Government Accountability Office. Effects of using pharmacy benefit managers on health plans, enrollees, and pharmacies. Accessed at www.gao.gov/new.items/d03196.pdf, August 5,
2009.
5. Roebuck MC, Liberman JN. Impact of pharmacy benefit design
on prescription drug utilization: a fixed effects analysis of plan
sponsor data. Health Serv Res. 2009;44:9881009.
6. Mark TL, Goldman TB, McGuigan KA. The effects of antihypertensive step therapy protocols on pharmaceutical and medical utilization and expenditures. Am J Manag Care. 2009;15:12331.
7. Goldman DP, Joyce GF, Zheng Y. Prescription drug cost sharing:
associations with medication and medical utilization and spending and health. JAMA. 2007;298:619.
8. National Association of Chain Drug Stores. Attitudes about mandatory mail order pharmacy. Reston, VA: Wirthlin Worldwide;
2003.
9. Birtcher KK, Shepherd MD. Users perceptions of mail-service
pharmacy. Am Pharm. 1992;NS31(12):3541.
10. Johnsrud M, Lawson KA, Shepherd MD. Comparison of mail-order with community pharmacy in plan sponsor cost and member
cost in two large pharmacy benefit plans. J Manag Care Pharm.
2007;13:12234.
11. Linton A, Garber M, Fagan NK, Peterson M. Factors associated
with choice of pharmacy setting among DoD health care beneficiaries aged 65 years or older. J Manag Care Pharm. 2007;13:677
86.
12. OMalley AJ, Frank RG, Kaddis A, et al. Impact of alternative interventions on changes in generic dispensing rates. Health Serv
Res. 2006;41:187694.
13. Shrank WH, Porter ME, Jain SH, Choudhry NK. A blueprint for
pharmacy benefit managers to increase value. Am J Manag Care.
2009;15:8793.
14. Cline RR, Mott DA. Demand for a Medicare prescription drug benefit: exploring consumer preferences under a managed competition framework. Inquiry. 2003;40:16983.
15. Desselle SP. Determinants of satisfaction with prescription drug
plans. Am J Health Syst Pharm. 2001;58:11109.
16. Acton JP. Nonmonetary factors in the demand for medical services: some empirical evidence. J Pol Econ. 1975;83:595614.
17. Coffey RM. The effect of time price on the demand for medicalcare services. J Hum Res. 1983;18:40724.

www.japha.org

J a n /F e b 2011 51:1

JAPhA 57

1/10/11 4:31 PM