The Epidemiology of Prescriptions Abandoned at the Pharmacy

advertisement
The Epidemiology of Prescriptions
Abandoned at the Pharmacy
William H. Shrank, MD, MSHS1,2; Niteesh K. Choudhry MD,PhD1;
Michael A Fischer MD MPH1; Jerry Avorn MD1, Mark Powell MA
MEd2; Sebastian Schneeweiss MD ScD1; Joshua N. Liberman
PhD3; Timothy Dollear MS3; Troyen A. Brennan MD, JD3; M. Alan
Brookhart PhD1
1. Division of Pharmacoepidemiology and Pharmacoeconomics,
Brigham and Women’s Hospital and Harvard Medical School
2. Center for American Political Studies, Harvard University, Faculty of
Arts and Sciences
3. CVS Caremark
Funding/Disclosures
• Funded by CVS Caremark
• Dr. Shrank supported by career
development award from the NHLBI
• 3 authors are employees of CVS
Caremark
• Dr.s Shrank and Choudhry have received
research funding from Aetna and Express
Scripts
Background
• Under-use of effective prescription medication is
common and a significant public health problem
• Most studies are conditional on a patient filling a
prescription and evaluate refill patterns
• Identifying the specific locus on the medication
prescribing/use pathway where patients fail to
receive essential medications is important in
order to create effective interventions
• No previous studies have evaluated one specific
point in the process – the rate that prescriptions
are bottled at the pharmacy and then never
purchased, or abandoned
Objective
• An evaluation of prescriptions abandoned at the
pharmacy can help us to better understand
medication underuse:
– The scope of prescription abandonment and the
contribution to non-adherence (i.e. the clinical
implications)
– The pharmacy inefficiency associated with
abandonment
– The patient, prescription and insurance-level factors
associated with abandonment
• This data can be used to drive intervention
development to improve the quality of patient
care and pharmacy efficiency
Methods
• True abandonment must be measured at the
retail pharmacy level (CVS data)
• To better understand previous use, subsequent
use, and use at other pharmacies, we linked this
to insurance data (Caremark)
• We included only patients who filled
prescriptions at CVS and who were insured by
Caremark
Methods
• Identification period: All Rx bottled and either purchased or
returned to stock (RTS – or abandoned) – using CVS claims
– For each patient, the first prescription within a class during the
identification was considered – the Index Prescription
• Matched to Caremark claims data
• Baseline period – 6 months to determine previous use
(Caremark data)
• Follow-up – 3 months of subsequent use at any pharmacy
(Caremark data)
Methods: Outcomes
• Each index prescription assigned to one of the following:
– 1) filled prescription - indicating that the Rx was purchased
– 2) RTS - indicating that the patient abandoned the prescription
– 3) RTS with fill - indicating that the patient abandoned the
prescription, it was returned to stock, but the patient purchased a
prescription for any medication in the same medication class at
the same or another pharmacy in the subsequent 30 days.
• Main outcome variable: (dichotomous) RTS vs. either a
filled prescription or an RTS prescription with fill
Methods:
Correlates:
• Prescription: New or prevalent use (6 month washout for “new”)
• Patient: Age, gender, comorbidity (number of unique meds), income
(income in zip code of residence), geography, urban vs rural
• Insurance: Copayment, type of healthcare insurance
Analysis Plan:
• Descriptives, bivariates
• Logistic controlling for clustering at the patient level with GEE
• Prediction rule – created by running logistic model on half the
population (randomly selected)
– included only the strongest predictors  model used to develop scores
based on effect sizes
– applied to the other half of the population to evaluate whether the score
can discriminate risk of abandonment
Results: study population 5,249,380 patients
who filled 10,349,139 index prescriptions
Mean (Standard Deviation) or Percent (Frequency) Characteristics Patient Age Gender 47.3 (21.3)
Female
Male
3.3% of Index Rx
were returned to stock
-1.8% abandoned (RTS)
-1.5% subsequently
filled (RTS with Fill)
60.12 (3134854)
39.88 (2079784)
Urban vs. Rural Rural (<1000 persons per sq mi)
Urban (≥ 1000 persons per sq mi)
31.93 (1435886)
68.07 (3061167)
Insurance Type Cash-Card/Other
Health Plan
Employer
Medicare
Medicaid
Geographic Region Northeast West South Midwest Other Territories Median Family Income In Zip Code Total Unique Prescriptions 4.85 (254336) 24.86 (1304744) 59.04 (3099450) 6.71 (352018) 4.55 (238832) 35.10 (1830011) 6.92 (360925) 42.16 (2198134) 15.81 (824603) 0.01 (730) 61762.1 (25349.9) 1.97 (1.6) Abandonment by Class
Filled
Drug Class Opiates
Anti‐Hypertensives
Anti‐Depressants
Statins
PPIs
Diabetes: Orals
Diabetes: Insulin
Percent (Freq)
RTS
95% C.I.
Percent
RTS with Fill
95% C.I.
Percent
95% C.I.
98.15 (671488)
98.12
98.18
1.00 0.98
1.02
0.85
0.83
0.87
97.55 (626631)
97.51
97.59
1.11 1.09
1.14
1.34 1.31
1.36
96.99 (443230)
96.94
97.04
1.44
1.41
1.48
1.57
1.53
1.61
97.27 (394908)
97.22
97.32
1.39
1.36
1.43
1.34
1.30
1.37
95.60 (250969)
95.49
95.65
2.60
2.54
2.66
1.83
1.78
1.89
96.97 (198272)
96.89
97.04
1.28
1.23
1.33
1.75
1.70
1.81
94.91 (62814)
94.75
95.08
2.24
2.13
2.35
2.85 2.72
2.97
Opiates least likely to be abandoned
Insulin and PPIs are most likely
Cost is an important correlate
Other correlates of abandonment
Predictor
Income Quintile
(reference
$0 - $41,094)
Copayment
(reference $0)
Electronic Rx
New User
Maintenance
Drug
Value
Odds Ratio
P value
$41,095 - $51,393
0.932
<.0001
$51,394 - $63,972
0.895
<.0001
$63,973 - $80,330
0.856
<.0001
$80,331 - $200,001
0.78
<.0001
$0.01 - $10.00
1.213
<.0001
$10.01 - $20.00
1.603
<.0001
$20.01 - $30.00
2.071
<.0001
$30.01 - $40.00
2.644
<.0001
$40.01 - $50.00
3.487
<.0001
$50.01 and Up
4.926
<.0001
Electronic prescription
1.722
<.0001
New User
2.831
<.0001
Maintenance Drug
0.964
<.0001
Abandonment Prediction Rule
Variables
# Points
New User
2
Electronic Rx
2
Age 18-34
1
Copay $30-50
2
Copay > $50
3
% Abandonment
12%
11.1%
10%
RTS
8%
7.6%
RTS with Fill
6%
4.8%
4%
2%
7.4%
3.2%
2.0%
1.8%
0.7%
0%
0
1-3
4-5
Score Group
6-8
Conclusions
• 3.3% of prescriptions bottled at the pharmacy
were abandoned, and in more than half the
patient did not fill another prescription in the class
• Cost and e-prescribing are important factors
• If similar rates are extended elsewhere –
approximately 110 million abandoned Rx annually
in the U.S.
• A large source of inefficiency for pharmacies –
conservative estimates suggest increased
pharmacy costs of over $.5 billion annually
Limitations
• Generalizability - uninsured under-represented
• No patient level data on income level - census
level data imprecise
• We did not require continuous enrollment in
Caremark for inclusion and may not have
complete claims
– Since this analysis took place in one calendar year,
we do not think this likely led to significant
misclassification
• Unable to assign causality to any associations
between the covariates and outcome, and
recognize that these covariates may be markers
of other patient or prescription characteristics.
Implications
• Every abandoned prescription for a chronic
medication could have clinical implications
• Meaningful source of waste and inefficiency
• “Sticker shock” is common – physicians
must be aware of out-of-pocket costs
• With increasing E-Rx-ing, this problem
could grow
• An opportunity to target intervention at highrisk patients
Download