Promotion and Prescribing of Hormone Therapy After Report of Harm by the Women’s Health Initiative FREE

Some (but by no means all^{1 - 3} ) studies have shown that publication of reports associating a medical treatment with harm or adverse events can lead to a prompt relinquishment of the treatment in question.^{4 - 7} For example, greater than 20% to 30% decreases in use were noted within months of reports of adverse events related to calcium channel blockers in the setting of myocardial infarction⁴ and use of α-blockers for hypertension.⁵ These studies are descriptive in nature and postulate that the decreases in use of the implicated medications were a result of the dissemination of evidence of harm through scientific channels to physicians (eg, presentations at meetings and publication in journals) and through lay media channels to patients (eg, newspaper and television reports).^{4 - 7} To our knowledge, the response of the pharmaceutical industry to evidence of harm has not been studied. This response may be particularly important to understand because many studies have documented that patterns of physician practice are often more aligned with promotional messages than with scientific evidence.^{7 - 11}

In terms of promotion, 1 of 3 potential strategies might be expected of the pharmaceutical industry in response to widely publicized randomized trial evidence of harm associated with one of its products. First, the affected companies might increase promotional activities, in an effort to “countermarket” the adverse results. This would be a particularly effective strategy if the evidence itself lacked validity or was ambiguous. Second, affected companies might maintain current levels of promotional activity in an effort to facilitate and expedite the dissemination of this new evidence in a fair and balanced fashion to patients and physicians.^{12 - 13} Third, there might be a substantial decrease in promotional activity in an effort to reduce financial losses and minimize the ill will that might be engendered by companies marketing a product associated with harm. While some initial news reporting suggested increased promotional efforts immediately before the publication of the Women’s Health Initiative Estrogen Plus Progestin Trial (WHI E+P) results,¹⁴ it is not known whether this strategy was sustained.

We took advantage of a wide-scale natural experiment concerning the use of hormone therapy in the United States to study these questions. By the mid 1990s, almost half of all postmenopausal women in the United States were being prescribed long-term hormone therapy.⁶ Despite the lack of evidence from randomized trials to support this practice,¹⁵ in 1998 hormone therapies were the 10th most heavily promoted class of medications and accounted for promotional expenditures of more than $300 million per year.¹⁶ Hersh et al⁶ recently reported a 38% decrease in the prescription of hormone therapy within months of the prerelease and publication of results from the WHI E+P, a randomized trial demonstrating that combination estrogen-progestin therapy was associated with significant harms and no net clinical benefits.¹⁷ We undertook the present study to examine the affected pharmaceutical companies’ responses, in terms of their promotional efforts, to these negative findings.

We used nationally representative databases published by IMS Health (Plymouth Meeting, Pa), an independent pharmaceutical research company, to describe national trends in hormone therapy prescription and promotion. Information on prescriptions was obtained from the National Prescription Audit Plus (NPA), which we have described in detail elsewhere.^{5 - 6 ,11} In summary, the NPA consists of a national random computerized sample of approximately 20 000 retail pharmacies, independent pharmacies, mass merchandise and discount houses, and mail order pharmacies. These stores account for more than half of all retail pharmacies in the United States and constitute a nationally representative sample. The NPA provides monthly data on the total number of dispensed prescriptions provided to consumers (new plus refill) for all medications. We classified hormone therapies as (1) Premarin (Wyeth, Philadelphia, Pa) brand of conjugated equine estrogen (≥0.625 mg); (2) Prempro (Wyeth) brand of combined conjugated equine estrogen (original standard dose of 0.625 mg) and medroxyprogesterone acetate; (3) lower-dose formulations of Prempro and Premarin, with either 0.3 or 0.45 mg of conjugated equine estrogen; and (4) other formulations besides Premarin and Prempro that include estrogen alone or in combination with either a progestin or an androgen. A total of 20 products used for postmenopausal hormone therapy were included in the category of other estrogen products. Transdermal, injection, and vaginal formulations, while uncommon, were included in the appropriate categories. We accessed dispensing data for January 2001 through December 2003.

We used IMS Health’s Integrated Promotion Services reports for promotional expenditures related to hormone therapies.^{11 ,16} We accessed promotional spending data from January 2001 through December 2003. This data specifically consisted of estimated promotional spending on office visits to physicians (“detailing”), visits to hospital physicians and pharmacists, medical journal advertising, and the retail value of free samples provided to physicians. To estimate office-based detailing expenditures, IMS Health uses a stratified random sample of representative office-based physicians and asks them to record the amount of time spent with pharmaceutical industry representatives each day for 1 month. Standardized estimates of cost per unit of time were used to generate the dollar value of detailing activity per drug. In 2001, this sample consisted of 3862 of 813 800 practicing physicians in the United States.¹¹ Using similar methods, expenditures on hospital visits were estimated by surveying 2067 hospital-based physicians and 514 hospital pharmacy directors.

Medical journal advertising was also estimated by IMS Health by examining the pages of all journals received by physicians in all disciplines of medicine. This represents a universal census figure (rather than a sampling) for journal advertisements.¹⁶ Costs of each advertisement were estimated by incorporating prespecified characteristics and applying rates and charges according to established advertising industry standards.

To estimate the retail value of samples provided, IMS Health recruited 1265 frontline office staff (from the 3862 aforementioned office-based physicians) to record all drug samples supplied by a pharmaceutical representative or by mail over a 1-month period. The retail value of these samples was computed using standardized retail pricing information independently collected by IMS Health. The estimate of the value of free samples likely exceeds the true economic costs borne by the pharmaceutical companies in providing these samples.¹⁶

Finally, we used the Consumer Ad$pender Trend Report created by Consumer Media Reports (New York, NY) to track expenditures on advertising directed toward consumers via television, radio, magazines, and newspapers.¹⁶ Similar to the methods described above, advertising expenditures were estimated according to standardized and prespecified criteria that reflect medium-specific advertising rates, durations, and frequencies of direct-to-consumer advertisements.¹⁶

Our analyses are descriptive in nature. We present aggregated quarterly data for prescriptions and for promotional spending. The WHI E+P results were first prereleased on July 9, 2002, and were published on July 17, 2002.¹⁷ A time series of prescription data for hormone therapies was first constructed. Time series also were used to display overall promotional spending for hormone therapies. These data were stratified by promotional mode and grouped according to the categories of hormone therapy described herein. We focused on comparing patterns of prescribing and promotion in the period immediately prior to the WHI E+P release (quarter 2 [Q2] of 2002) with patterns noted 9 months afterward and then 18 months after the WHI E+P results were published (Q4 of 2003).

In aggregate, prescription rates for hormone therapy were stable from 2001 through the first half of 2002, then declined precipitously after the WHI E+P results were reported in July 2002 (Figure 1). The total number of prescriptions dispensed for hormone therapy was constant between Q1 of 2001 (22.8 million) and Q2 of 2002 (22.4 million). Following the WHI E+P report in July 2002, hormone therapy prescribing quickly began to decline, with 19.0 million prescriptions dispensed in Q3 of 2002. Continued declines in prescribing were noted, with 15.2 million prescriptions in Q1 of 2003 (a 32% decline from Q2 of 2002). Eighteen months after the WHI E+P report, in Q4 of 2003, there were 12.7 million prescriptions made, a 43% decline in prescribing.

The decline in postmenopausal hormone therapy use was most marked for standard-dose Prempro, the agent used in the WHI E+P study, whose use decreased 64% between Q2 of 2002 (4.4 million prescriptions) and Q1 of 2003 (1.6 million prescriptions) (Figure 2). By Q4 of 2003, the use of Prempro had declined by 80% (to 0.9 million prescriptions). Premarin (≥0.625 mg of conjugated equine estrogen) use decreased by 48%, from 8.3 million to 4.3 million prescriptions in Q4 of 2003. A less substantial 26% decline was noted for other estrogen-containing products (9.0 million to 6.7 million). Contrary to these trends, however, we noted that prescriptions for lower-dose formulations of Premarin and Prempro increased modestly, from 0.7 million in Q2 of 2002 (exclusively Premarin, 0.3 mg) to 0.9 million in Q4 of 2003 (including new lower doses of Premarin and Prempro; Figure 2).

From 2001 through Q2 of 2002, promotional spending was relatively stable (Figure 1). A sudden decrease in promotional spending immediately followed publication of the WHI E+P results. In Q2 of 2002, $71 million was spent promoting hormone therapy; in annual terms, $284 million, or about $350 per practicing physician in the United States. In Q3 of 2002, promotional expenditures were $59 million. By Q1 of 2003, $45 million was being spent (in annual terms, $180 million or $220 per physician), representing an overall 37% decline in promotional spending (Figure 1). After this nadir, total promotional expenditures began to increase again and had reached a level of $55 million by Q4 of 2003 (Figure 1 and Figure 2).

Promotional spending declined for all categories of hormone therapy except lower-dose formulations of Premarin and Prempro (Figure 2). Promotional expenditures for standard-dose Prempro declined from $19.4 million in Q2 of 2002 to $7.6 million in Q1 of 2003 to $3.5 million by Q4 of 2003, representing an overall 82% decline. Conversely, in the latter part of 2003, the initiation of first-time promotional spending on lower-dose estrogen formulations corresponded with the first-time availability of low-dose Prempro and had reached a level of $13.5 million by Q4 of 2003.

To better understand changes in different aspects of promotional spending over time, we tracked promotional spending by drug class and promotional mode (Table). Promotional expenditures prior to the WHI E+P report were dominated by free samples and office-based detailing. The $71 million spent on hormone therapy promotion in Q2 of 2002 included $32 million as the retail value of free samples and $28 million for office-based detailing. Much less was spent on hospital-based promotion ($3 million), medical journal advertising ($2 million), and direct-to-consumer advertising ($7 million).

After the WHI E+P results were released, expenditures for all modes of promotional activity decreased for nearly all types of hormone therapy (Table). The largest absolute decreases observed were for samples, which rapidly decreased 36% by Q1 of 2003 and then continued to decline slightly to 39% below earlier levels through Q4 of 2003. Samples for standard-dose Prempro decreased 83% from $5.8 million in Q2 of 2002 to $1.0 million in Q4 of 2003. Office-based detailing decreased for most products, with an especially marked 66% decline for standard-dose Prempro through Q4 of 2003. Most noteworthy, perhaps, was the complete abandonment of direct-to-consumer advertising for standard-dose Prempro. Contrary to the trends seen with older and established hormone therapies, we noted for the first time substantial promotional expenditures ($19 million in Q3 and $14 million in Q4 of 2003) directed toward the lower-dose formulations of Prempro and Premarin. Of note, these lower-dose formulations were also the only hormone therapies that had increased prescription rates following the WHI E+P study publication.

By the end of the 1990s, almost half of postmenopausal women in the United States were being treated with long-term hormone therapy.⁶ The findings of the WHI E+P—increased risk of harms related to cardiovascular disease, thromboembolic events, cognitive function, and breast cancer with no net clinical benefits—were widely and rapidly disseminated through both scientific and lay media communication channels.^{7 ,18} A year and a half after these results were published, there had been a 43% decrease in the prescription of hormone therapy in the United States. At its nadir, 9 months after the WHI E+P publication, promotional spending had decreased by 37%, followed by some resurgence over the next 9 months.

Decreases in promotional spending were pervasive and occurred for most types of hormone therapy. The only exceptions were increases in promotional activity for lower-dose formulations of Premarin and Prempro. While these lower-dose agents may fill a niche suggested by current recommendations for tapering estrogen to the lowest possible dosage for short-term use to relieve vasomotor symptoms,¹⁹ it is still unknown whether simple dose reductions are sufficient to alter the imbalance of harms and benefits noted in the WHI E+P.

Although the dominant initial response of the makers of hormone therapies was to reduce the magnitude of all promotional activities following the WHI E+P, over time we observed an evolving mixture of responses. For example, for standard-dose Prempro (the agent most directly implicated by the WHI E+P), there were substantial reductions in all types of promotional spending and a complete abandonment of direct-to-consumer advertising. This was, however, followed by relatively heavy new promotional spending by the same company on a lower-dose formulation of the same product. At least through 2003, promotion of Premarin has declined only modestly. This pattern suggests a corporate strategy that recognized the importance of the WHI E+P results but also envisioned a continuing and viable market for hormone therapy. That said, our data were collected prior to the release of the WHI estrogen-only trial, a study also terminated early because of an excess of risk over benefit.²⁰

In absolute terms, we estimated that pre-WHI E+P promotional spending amounted to $350 per physician per year in the United States compared with $220 per physician in the post-WHI E+P era. It is noteworthy that decreases in hormone therapy prescribing following the release of the WHI E+P results mirrored the declining expenditures on promotion. Simultaneously, other influences, such as widespread media attention and patient demand,^{7 ,18} also may have contributed to the decline in hormone therapy use. Acknowledging the observational nature of these relationships, a potential corollary to our findings is that sustained promotional efforts prior to the WHI E+P study may have fostered the practice of hormone therapy prescribing in the United States.^{13 ,21} In the case of proven efficacious treatments, we have previously demonstrated a “dose-response” relationship between pharmaceutical industry promotional efforts and the accelerated translation of published evidence beyond that which would be expected by diffusion through the usual scientific channels.¹¹

To our knowledge, patterns of promotional spending in response to randomized trial evidence of harm have never been previously reported. In examples of product-related adverse events within both health care and other industries, the usual response frequently appears to be an increase in promotional spending aimed at countering negative publicity. For example, advertising increased for the Tylenol brand of acetaminophen following adverse events later determined to be due to product tampering.²² A similar increase in promotion accompanied reports of Firestone-Bridgestone tire tread separation²³ and of bacterial contamination in Odwalla juices.²⁴

There are several limitations that need to be considered when interpreting our findings. First, although we have estimates of the numbers of dollars spent on promotion, we have not tracked the content, quality, or intensity of the promotional activities themselves. It is likely that changes in promotional content were as dramatic as the changes in promotional spending. Examples of promotional materials and messages, for example, might distinguish a particular product from that used in the WHI E+P trial, highlight the possibility that (unstudied) lower-dose formulations might be safer, or emphasize a narrower therapeutic role for estrogen in temporarily relieving vasomotor symptoms.

Second, we did not have data on changes in spending on additional activities that might be considered to be “promotional” in nature,^{21 ,25} such as industry-sponsored symposia, workshops, other forms of continuing medical education, or Web-based promotional efforts. Thus, our estimates are certainly underestimates of the total resources devoted to the promotion of postmenopausal hormone therapy and might fail to capture reallocation of resources to these specific forms of physician outreach. Third, we examined only US data, and whether our results are generalizable to other parts of the developed world is unknown. Nonetheless, similar rapid decreases in the prescription of hormone therapies have been reported in Canada²⁶ and New Zealand.²⁷ Finally, we examined only 1 example of harm and the responses of the companies that make hormone therapies; future research should attempt to replicate these findings with situations in which the evidence of harm is unequivocal (eg, doxazosin for hypertension⁵ ) and in which the evidence suggesting harm is more ambiguous (eg, selective cyclooxygenase 2 inhibitors and cardiovascular events²⁸ ).

Concordant with its widespread use, hormone therapy was among the most heavily promoted medications prior to WHI E+P. Following the evidence of harm from this trial, promotional spending declined, particularly for the agents most directly implicated in the trial. The declining hormone therapy use that resulted from the complex interplay of credible published evidence, declining patient demand, widespread media attention, and declining promotion suggests the value of better understanding strategies for disseminating information to improve the quality of prescribing.