Should We Judge a Pill by Its Cover?


Written By

Roshni S. Patel, PharmD, BCPS
Stephanie M. Seaton, PharmD

Reviewed By

P. Tim Rocafort, PharmD, BCACP
KyAnn Wisse, PharmD
Dawn Fuke, PharmD, BCPS


Kesselheim AS, Bykov K, Avorn J, et al. Burden of changes in pill appearance for patients receiving generic cardiovascular medications after myocardial infarction. Ann Intern Med 2014;161:96-103.

Have you ever had difficulty convincing some of your patients that generic medications work just as well as their brand-name counterparts?  While there is little evidence to suggest that generic medications aren’t therapeutically equivalent, a more subtle and perhaps more vexing problem is now emerging.  From a health policy standpoint, generic medications are critically important because they are the driving force that makes the pharmaceutical marketplace more competitive.  Generic drug use results in very significant financial savings for patients and payers.  And data has shown that making medications more affordable improves medication adherence.1

While the Food and Drug Administration (FDA) mandates that a generic drug be bioequivalent to the brand-name product, it need not be similar in color, shape, or size.2  This results in numerous versions of the product  – each with a different physical appearance.  Understandably, this can cause confusion for patients, especially for those who identify their medications as “the little yellow pill” and “the purple one that’s shaped like a football.”   Confusion can then lead to interruptions in medication use or, perhaps worse, therapeutic duplication.  Although medication adherence is multifactorial, a few studies have found an association between poor medication adherence and changes in physical appearance due to generic substitutions.3-5 

The most recent study to explore this problem, published in the Annals of Internal Medicine, examines adherence with cardiovascular (CV) medications following medication switches.6  The investigators conducted two separate analyses.  The first analysis was a cohort study conducted in a group of patients who had been hospitalized for a myocardial infarction (MI).  This analysis was intended to identify the incidence of pill appearance changes among dispensed CV medications following hospital discharge.  The cohort was identified from a commercial insurance research database.  Patients were at least 18 years of age and newly started on treatment with a generic beta blocker (BB), angiotensin-converting enzyme inhibitor (ACE-I), angiotensin receptor blocker (ARB), or HMG-CoA reductase inhibitor (statin) within 90 days of hospital discharge.  Only tablet or capsule formulations were included in the analysis. Each drug included in the analysis needed to have at least two generic versions available on the market.  Patients were required to have at least 6 months of continuous insurance enrollment prior to and 12 months after hospital discharge.  Refill patterns were followed for one year after the index hospitalization.  Within the post-discharge year, patients were censored if they were rehospitalized for a subsequent MI, switched to a different drug in the same therapeutic class, or discontinued treatment (defined as a lapse of greater than 31 days after the end of the ‘days supply’ of the last fill).

Using the NDC numbers for each dispensed medication, the investigators were able to determine the dose, color, and shape for each product.  A change was defined as a “discordance in color or shape” at the time of medication refill and the rate of change was calculated by dividing the number of changes by the total number of filled prescriptions.  Since patients could have been initiated on therapy from more than one drug class, each treatment course was evaluated separately, with up to three medication analyses per patient.  Additionally, the investigators identified whether changes in color or shape were related to changes in dose.

In the second analysis, a nested-case control study was conducted to determine the association between medication persistence and changes in pill appearance.  Nonpersistence was defined as a gap of greater than 31 days after the end of the ‘days supply’ of the last fill.  Episodes of medication nonpersistence amongst case patients were matched to control patients based on sex, age, and the number of post-MI medications filled.  Odds ratios were used to determine the association between medication nonpersistence and discordance in pill appearance.

In the cohort study, a total of 11,513 patients met criteria and 19,379 treatment episodes were analyzed.  The cohort included mostly men with a mean age of 56 years.  During the 1-year post-hospitalization period, 28.5% (n = 3,286) of patients and 20.1% (n = 3,891) treatment courses had at least one change in either medication shape or color unrelated to a dose change.

In the case-control study, both groups were similar in terms of age and prior use of non-index study medications; however, the control group had a lower mean comorbidity score (0.2 vs. 0.3; p<0.05) and a higher incidence of revascularization procedure during the index hospitalization (82.2% vs. 78.0%; p<0.05), compared to the case group.  Additionally, there were differences in terms of mean healthcare utilization between the two groups.  Control group patients were prescribed a larger number of drugs at baseline (9.4 vs. 8.4 medications), had fewer prior hospital admissions (0.06 vs. 0.09), fewer hospital days (0.4 vs. 0.6), and fewer physician visits (3.0 vs. 3.3) [all comparisons p <0.05].  Lastly, the case group used mail-order pharmacies more often than the control group (9.6% vs. 5.1%; p<0.05) and were more likely to have changed pharmacies during the last 2 refill periods preceding nonpersistence (10.3% vs. 6.7%; p<0.05).

Results of the analysis revealed that changes in pill color, shape, or both were associated with medication nonpersistence (Table 1).  Relative to control patients, case patients were 34% more likely to be nonpersistence following a change in pill color and 66% more likely after a change in pill shape.  Changes in both shape and color increased the risk of nonpersistence even more.  This association remained true after adjusting for some covariates.  A sensitivity analyses was also conducted.  Shortening the period of nonpersistence to 7 days and extending it to 60 days did not change the association.

Table 1. Association between nonpersistence and color/shape discordance in medications after MI6


Discordance Among Case Group

[n = 4573, n (%)]

Discordance Among Control Group [n = 19881, n (%)]

OR (95% CI)

Adjusted OR

(95% CI)*


177 (3.9)

587 (3.0)

1.34 (1.13 – 1.59)

1.34 (1.12 – 1.59)


242 (5.3)

644 (3.2)

1.67 (1.43 – 1.95)

1.66 (1.43 – 1.94)

Color or shape

309 (6.8)

922 (4.6)

1.50 (1.31 – 1.71)

1.49 (1.30 – 1.71)

Color and shape

110 (2.4)

309 (1.6)

1.58 (1.27 – 1.98)

1.58 (1.27 – 1.98)

*Adjusted for age, year, comorbidity score, revascularization procedure, prior use of nonindex study drugs, and # of drugs used at baseline

Interestingly, when the two pharmacy confounders (mail-order pharmacy use and change in pharmacy) were included in the model, the risk of nonpersistence was diminished and persistence was no longer affected by changes in color alone (Table 2).  Regardless, the authors of the study concluded that variations in pill appearance were associated with nonpersistent medication use of CV drugs following MI. 

Table 2. Association between nonpersistence and color/shape discordance in medications after MI adjusted for pharmacy change and mail-order pharmacy use6


Adjusted OR for Pharmacy Change

(95% CI)*

Adjusted OR for Use of a Mail-Order Pharmacy (95% CI)


1.10 (0.91 – 1.32)

1.16 (0.97 – 1.39)


1.41 (1.19 – 1.66)

1.38 (1.18 – 1.62)

Color or shape

1.25 (1.08 – 1.45)

1.25 (1.09 – 1.44)

Color and shape

1.32 (1.05 – 1.66)

1.37 (1.09 – 1.72)

* Adjusted for all covariates in the primary adjusted OR model and an additional covariate for change in pharmacy 
Adjusted for all covariates in the primary adjusted OR model and an additional covariate for mail-order pharmacy

While this study introduces new findings, there are some limitations that should be noted.  First, because only four classes of CV medications were examined, the results may not be generalizable to other therapeutic classes.  Second, factors such as socioeconomic status were not accounted for, and this is an important factor that can impact medication adherence.  Third, information was taken from third party claims data and discrepancies in reporting may not have been captured in the analysis.  Only commercially insured patients were included and therefore, the results may not be generalizable to other patient populations.  Moreover, pharmacy claim data does not capture discontinuation due to adverse drug events and the reasons why patients stopped treatment are not known.  On the flipside, programs intended to improved medication adherence, such as the auto-refills, may underestimate nonpersistence.  An additional adherence assessment would have been beneficial to confirm medication use.  Fourth, no clinical surrogate data, such as blood pressure or low-density lipoprotein cholesterol, or clinical events, such as re-hospitalizations, recurrent CV events, or death, were captured.  And lastly, this study did not technically measure medication adherence, but rather persistence, defined as the amount of time from initiation to discontinuation of therapy.

Admittedly, changes in pill appearance are a potential problem because they could reduce a patient’s confidence that the dispensed medications are safe or effective.  Pharmacists and prescribers need to take a more proactive role to educate patients about changes in pill appearance to reassure them.  Some pharmacies place auxiliary labels on pill bottles when changes in medication occur and apps on smart phones and tablet computers can help patients identify their pills.  But that's not enough.  Some pharmacies change generic products, in order to stock the cheapest option, so frequently that it probably leads to mistrust.  This may explain why a recent pharmacy change or use of a mail order pharmacy diminished the association between nonpersistence and changes in pill appearance.  Are the economic advantages of the generic products potentially erased by these findings?  Probably not.  Indeed, a recent comparative effectiveness study found that patients who were dispensed a generic statin were more likely to adhere to therapy when compared to those patients who received a brand-name product.  And more importantly, patients who used a generic statin were less likely to be hospitalized for an acute coronary event, have a stroke, or die.7  That’s powerful!

So how should changes in pill appearance be addressed?  FDA mandates to require similar pill appearance would seem like an easy solution.  Historically, pharmaceutical companies have defended product appearance using “trade dress” laws, but the US Supreme Court has stated such protection cannot be given.8  Unfortunately, there are no industry standards that promote uniformity in product appearance.  The political reality, however, is that any new FDA mandate will be aggressively opposed.  In the meantime, how can ambulatory care pharmacists ensure that product appearance changes don’t result in treatment discontinuation?  Or is this a real problem that even needs to be addressed?