Who Pays Higher Prices for Prescription Drugs?
Who Pays the Lowest Prices?
Using data from the prescribed medicine event file of the Medicare Current Beneficiary Survey (MCBS) 1998 Cost and Use File, we examined nearly 70,000 drug purchases by seniors. The data included information on the total price paid and the share of the cost paid by the patient or a third party - private insurers or such government health programs as veterans' health. We performed statistical tests to determine the prices paid when patients or insurers pay most of the cost and whether patients or insurers were able to obtain the lowest price. What follows is a discussion of the results. The details of the testing procedures are described in Appendix B. Because the data is based on survey responses, any rebates an insurer received from a pharmacy or pharmaceutical company that are not included in the price of the prescriptions at the time of purchase are not included in the analysis.
"We analyzed nearly 70,000 drug purchase by seniors on medicare, some of which were paid for by insurers and some by seniors out of pocket."
The 39 Most Commonly Prescribed Drugs.7 In the first test, we examined the 39 most commonly prescribed drugs for Medicare patients to determine in each case whether the patient or insurer obtained the lowest cost. In categorizing buyers, we used the following method.8 If 75 percent or more of a prescription was paid out of pocket, we designated the purchase as an out-of-pocket payment by patients. If a private insurer (HMO, employer-sponsored insurance, individually purchased insurance or some combination thereof) paid 75 percent or more of a prescription's cost, we designated it as an insurance purchase. We put all other combinations of third-party payments and out-of-pocket payments in the "other" category. In Figure I, the drugs are grouped according to which buyer paid the lowest price. With the above definitions in mind, note that:9
- Private insurance paid the lowest price for 33.3 percent of drugs.
- Patients paid the lowest price for 38.5 percent of the drugs.
- Other payer combinations paid the lowest price for 28.2 percent of the drugs.
Figure II takes the same set of 39 drugs and identifies whether the average price paid by the lowest-price payer was significantly lower than the next-lowest average price paid. As the figure shows:
- For almost 60 percent of the drugs, the price paid by the lowest cost payer was not significantly less than the next lowest cost payer.
- Patients and private insurers paid the lowest price for 23.1 percent and 10.3 percent of the drugs, respectively.
- Other combinations paid the lowest price for the remaining 7.7 percent of the drugs.
The 229 Most Commonly Prescribed Drugs. In the second test, we repeated the process for an expanded set of 229 drugs. To eliminate small sample effects, calculations were limited to drugs for which there were at least 100 prescriptions in the sample, before other restrictions. This left a total of 229 drugs, for which the 75 percent thresholds were valid, and a total of 69,576 prescriptions. The 39 most commonly prescribed drugs, summarized in the first two figures, accounted for 36,509 prescriptions, or 52.5, percent of the prescriptions in the larger sample. Figure III shows that:
- Patients paying out of pocket paid the lowest price as often as private insurers did - 35.8 percent of the time.
- Other combinations paid the lowest average price 28.4 percent of the time.
This indicates that for 147 drugs the average unit price paid by those with private insurance or other coverage that does not meet the 75 percent threshold is lower than the price paid by those who primarily pay out of pocket. This may suggest that the dominant discount effect is the power of large purchasers. However, that patients paying out of pocket pay the lowest price as often as private insurers is evidence of the price incentive effect.
Most of the differences are not significant and may be due to purely random factors. Consistent with Figure II, Figure IV shows that:
- For close to three-quarters of the drugs, the unit price paid by the lowest cost group was not significantly less than the price paid by the next lowest cost group.
- Among the 62 drugs for which price differences were significant, individuals paid lower prices twice as often as did third parties.
"For three-fourths of the 229 drugs, there was no significant price difference, regardless of who paid."
Regression Analysis. We performed additional statistical tests to determine the relationship between the share of a prescription's total cost that is paid out of pocket and the unit costs of the drugs. We first separated 247 diverse drugs into two groups: the 39 drugs that were each prescribed more than 1,000 times (see Appendix B, Tables III and IV) and the remaining 208 drugs that were prescribed no more than 1,000 times.10 We found that:
- Across all of the 39 individual drugs prescribed more than 1,000 times, the relationship between the percentage of a prescription a patient paid and the unit price is negative and significant, meaning that the unit price fell as the patient's share of the cost rose.
- Looking at the relationship for each individual drug, we found that the relationship was negative and significant for 27 of the 39 drugs, positive and significant for six drugs and insignificant for six.
In general, then, higher cost sharing is associated with lower unit cost. We found no pattern with respect to the type of drug, either in terms of treatment use or whether it was generic or brand name.
"Additional statistical tests showed patient spending their own money were more likely to pay less for costlier drugs and repeatedly used drugs."
Patients with Repeated Drug Usage. To see if seniors have a greater incentive to shop for lower prices when they use a particular drug repeatedly, we divided the sample into the 197 drugs that are, on average, prescribed more than six times per patient (all the aforementioned 39 drugs fall into this group) and a second group including the 50 drugs with an average prescription frequency of fewer than six. Drug shopping affects the price paid for both groups of drugs, but it seems to have a greater price effect for drugs used repeatedly.
Expensive Drugs versus Inexpensive Drugs. The relative importance of the discount effect of concentrated purchasing power and the price-shopping effect may differ according to the price of the drug. Thus we divided the sample into "expensive" and "inexpensive" drugs to examine whether the impact of cost-sharing ratios on the price paid differed for these two groups. [See Appendix B, Table III.]
We sorted the drugs from high to low in terms of their mean unit costs. The first half (123) drugs form the "expensive" group, and the second half (124) fall into the "inexpensive" group. As expected, the average unit costs of the two groups differ significantly - the average cost of some drugs exceeds one dollar per milligram, while some others cost less than one-tenth of a cent.
On average, patients share only slightly more costs for the expensive drugs (76.5 percent) than for the inexpensive drugs (75.0 percent). The percentage change in unit cost that results from a change in out of pocket spending is similar for the two categories of drugs. However given that average unit costs are much higher for the expensive drugs, the savings associated with increasing the share of out of pocket costs are greater for them. This is consistent with the shopping hypothesis, since patients have an incentive to shop more - that is, they should be willing to incur higher search costs - because of the greater potential cost saving for more expensive drugs.11
Pill Splitting and Bulk Buying. In addition to searching for lower unit prices, patients can divide some pills into two doses for a lower unit cost because prices tend not to be proportional to pill size. For example, Atenolol comes in 25, 50 and 100 milligrams, and the average prices per milligram are $0.018, $0.009, and $0.007, respectively. If patients are splitting pills as a cost-reduction strategy, the average size of the pill purchased should rise with the share of the cost paid out of pocket. To test whether out-of-pocket share is related to the size of the pill purchased - and hence the price per unit - we identified all pill sizes for each of the 39 top drugs that accounted for at least 10 percent or more of the prescriptions written for that drug. Since pill splitting is not practical for encapsulated drugs, we examined prices for various sizes of the 24 drugs which are available as uncoated or scored pills. We found that:
- In four cases, the relationship between out-of-pocket share and the size of the pills patients purchased was positive and significant, consistent with the pill-splitting hypothesis.
- However, for six drugs, the relationship between out-of-pocket share and the size of the pill was negative and significant, meaning that patients tended to buy smaller doses of the pills as their out-of-pocket share rose.
- For the remainder, there was no significant difference.
Thus higher out-of-pocket shares are not necessarily related to the practice of buying larger pill sizes.12
"Shopping for drugs can reduce the costs because prices vary widely."
Reasons for the Results. For several reasons, price shopping can significantly reduce costs. First, retail drug prices vary greatly among geographically distinct markets and across retail pharmacies within the same market. For example, one study documented substantial price differences for cash customers among 20 pharmacies in two cities of upstate New York; average prices at high-price pharmacies (those at the 90th percentile) were more than five times as high as those at the lowest-cost pharmacies (those at the 10th percentile).13
Buying prescription drugs in the national marketplace via the Internet or mail order is another way beneficiaries can obtain lower prices.14 However, this practice was not as widespread in 1998 as it is today. Pill splitting and quantity buying are additional ways to lower the unit cost of a drug; however, these methods do not appear to be the primary means by which unit costs are lowered. Frequent prescription drug users have more motivation to employ these techniques because of their high drug bills. In 1998, one-third of Medicare beneficiaries accounted for three-fourths of prescription drug spending.