Marketing of the Life Sciences: A New Framework and Research Agenda for a Nascent Field

Therapy pipeline optimization

In life sciences firms, therapy pipelines contain all innovation projects along the following temporal stages: During discovery, therapy candidates are screened for maximum activity on the biological target. Preclinical development and clinical development entail further development, using in vitro or animal experiments and human experiments, respectively.

Prior research on therapy pipelines aimed to determine the optimal number and sequencing of innovation projects that a firm's resource base could support and that served its goal to maximize the number of commercially launched innovations (see Blau et al. 2004; Chandy et al. 2006; Ding and Eliashberg 2002; Loch and Kavadias 2002). This research found that there is an inverted U-shaped relationship between the number of innovation projects undertaken and the number of innovations commercially launched. However, scholars in this literature stream did not discern the different temporal stages in the therapy pipeline. Although companies’ ability to convert innovation projects in commercially launched products may suffer from taking on too many projects in development, this may not be the case in discovery, in which more exploration leads to more effective knowledge on biological targets, resulting in more new therapy opportunities. Thus:

The optimal number of innovation development projects a firm should undertake may also be contingent on the type of innovation project. Targeted (specific for certain patient types) therapy innovation projects require fewer resources in development and feature higher probabilities of ultimate regulatory approval (Vernon and Hughen 2005). Thus:

Scholars might also study other types of innovation projects as contingency factors beyond targeted or nontargeted projects, such as radical versus incremental projects. Studying the therapy pipeline in the context of patent expiry might also be fruitful. Firms may anticipate expiry in multiple ways, such as the development of combination drugs, more convenient administration and dosage methods, and reengineered variants with higher effectiveness or less serious side effects. To develop and test such a contingency framework, scholars could analyze databases, such as the Pharmaprojects database, the R&D Focus Database that IMS Health maintains, and the Food and Drug Administration's (FDA's) Orange Book, all of which contain detailed pipeline information. As outcome variables, scholars could gather information on the number of approved new patents (U.S. Patent and Trademark Office) and new therapies (the FDA's Orange Book). Beyond direct innovation measures, they could also examine the impact of therapy pipeline decisions on sales, profits, or stock market returns.

Innovation alliance formation

As we noted previously, practitioners consider decisions on innovation alliances ancillary decisions. At the same time, this decision area has provided an ideal and often-used testing ground for theory development on interfirm cooperation. The reason is that the life sciences industry provides possibly the richest documentation on such alliances (e.g., Recap's database on interfirm agreements) and their outcomes (e.g., patents, new products, profits, sales, share price).

Similarity between parties in an alliance is probably most often studied. Dissimilarity between partners yields greater learning opportunity because there is less knowledge redundancy, while similarity between partners makes it easier to understand each other and share information. The tension between both arguments has led many researchers (Cloodt, Hagedoorn, and Van Kranenburg 2006; Prabhu, Chandy, and Ellis 2005; Wuyts et al. 2005) to find a curvilinear relationship between knowledge similarity between alliance partners and the innovative outcome that the alliance yields. This leads us to the following preliminary generalization:

Scholars have also studied the differential effect of alliances on radical versus incremental innovation (Wuyts, Dutta, and Stremersch 2004). For radical innovation, it is instrumental that alliance partners repeatedly cooperate to stimulate knowledge transfer through the development of relationship-specific heuristics and the sharing of mental models, among other things (Madhaven and Grover 1998; Uzzi 1997). Genentech and Roche provide a successful example of such repeated collaboration. For incremental innovation, large portfolios may be beneficial because of scale effects in development (Ahuja 2000; Wuyts, Dutta, and Stremersch 2004). We offer the following preliminary generalizations:

Further research on interfirm cooperation will likely continue to use the life sciences industry as a testing ground for theory development, with continued use of databases (e.g., Pharmaprojects, Recap), newspapers and magazines, and surveys. Novel breakthroughs are likely to be in the areas of social networks and the balance between internal and external innovation.

Therapy positioning

Therapy positioning refers to research-and-development (R&D) decisions on the envisioned therapy toward specific indications. The practitioners we surveyed considered therapy positioning an ancillary decision area, while academics did not foresee a strong need for further research. Therefore, we do not derive theoretical generalizations or propositions. Decision makers need to balance three key dimensions: (1) the likelihood that the therapy will be approved for the respective indication, (2) the price they will obtain from the therapy, and (3) the market size for the respective indication over time.

If positioned for a mild indication, a therapy may reach a large market, but at relatively low prices and with possible denial of approval. Consider Elidel (pimecrolimus), a therapy for eczema by Novartis. Novartis introduced Elidel for a mild to moderate indication of eczema—that is, for first-line use. Competitor Fujisawa introduced a variant of this molecule, Prograf (tacrolimus), which was targeted at moderate to severe indications of eczema—that is, for second-line use. Although both products showed scientific evidence, only tacrolimus was endorsed by the U.K. government, because the former could not show that it represented a good value for the money (Gregson et al. 2005) for the moderate indication. It was subsequently endorsed after resubmission, but then also for the severe indication. If positioned for a severe indication, a therapy may have a higher likelihood of being approved at a high price, but it may pertain to a relatively small market. For example, Symbicort by AstraZeneca was first approved for severe asthma, after which AstraZeneca enlarged the market for Symbicort to chronic obstructive pulmonary disease (COPD).

Because there are many possible indications, all with different levels of uncertainty for the respective therapy to be approved and varying price expectations, further research should aim to specify decision support models that simulate market size using patient flow dynamics (first use, reuse, switching from competition) at various price expectations and approval likelihoods.

Global market entry timing

Previous research has shown that pioneers do not have long-lasting market advantages (Golder and Tellis 1993; Shankar, Carpenter, and Krishnamurthi 1999). In the life sciences industry, an important moderator of the market return on a pioneering therapy may be whether it pertains to generic or branded therapies. In the case of branded therapies, pioneers are the first entrants in a therapeutic category (e.g., Mevacor [1987] for statins). In the case of generic therapies, pioneers are the first generic available for a specific therapy (e.g., the first generic Simvastatin, the statin introduced by Merck as Zocor).

There are many cases of late branded entrants that took over pioneers through increased effectiveness, higher convenience, or weaker side effects. Examples include Zocor and Lipitor in statins (increased effectiveness), Symbicort in asthma/COPD (higher convenience), and Xyzal in antihistamines (weaker side effects).

Contrary to common wisdom in other industries and contrary to branded variants in life sciences, generics may yield strong pioneering advantages. The first generic variant for a specific therapy (“the pioneer”) may attract and maintain a disproportionately large market share. The reasons for this are multifold. It takes substantial effort from physicians and pharmacists to explain bioequivalence between different variants (Gupta, Yu, and Guha 2006). At the same time, only the pioneering generic therapy benefits from the large price differential with the alternative (the branded variant). Generics that subsequently enter do not show as large of a price differential anymore, and when they do, the generic pioneer may readily match the lower price, with market shares remaining stable (Hollis 2002). The first generic entrant also typically makes supranormal profits before the entry of a second generic because it provides the only (cheap) alternative for an expensive branded variant (Gupta, Yu, and Guha 2006). Thus:

The life sciences industry lends itself well to the examination of order-of-entry effects because entry is well documented (e.g., with the FDA Drugs@FDA for the United States). These entry dates can be complemented with IMS Health's dollar sales estimates. Moderators that could be considered in such research effort are clinical profile of the treatment (e.g., from National Institute for Health and Clinical Excellence or published meta-analyses in scientific journals) and marketing support (commonly available from firms such as IMS Health, Kluwer, or Verispan).

Firms typically do not launch a new treatment simultaneously across the globe. Rather, they use specific launch sequences, often driven by a country's regulatory system, economic wealth, and size (Danzon, Wang, and Wang 2005; Kyle 2007; Verniers, Stremersch, and Croux 2008). Differential launch timing across countries has been shown not to affect unit sales (Stremersch and Lemmens 2009), though it has been shown to affect launch price (Verniers, Stremersch, and Croux 2008). In the life sciences industry, launch price is rarely a market price; rather, it is often an agreed-on price between the supplier and the government or insurance firm, which acts as a (co)payer. In such negotiations, entry timing may be used by both the payer and the firm as an instrument to affect the agreed-on price.

An important contingency factor that has not received any attention is the role of cross-country influence in launch sequencing. Often, this cross-country influence is institutionalized because payers will use the price of a therapy in a defined set of other countries (the “referent” countries), if available, as a reference price for the negotiations in their own country (the “referencing” country). Such regulation incentivizes companies to avoid spillover effects (Hunter 2005). Thus:

To test this proposition, regulatory data can be gathered from Urch Publishing and the Organisation for Economic Co-operation and Development, both of which track international regulatory health systems (including identification of the set of referent countries for each referencing country), and integrated with IMS Health data on international prices and introduction dates. It is also possible to include firm effects (firms may have differential policies, depending on their home market or size) or therapy effects (payers across countries may have differential price and market access policies for different therapy classes). In addition, diffusion studies can deliver valuable and complementary insights into launch decisions. Examples of such valuable inquiries that may inform launch decisions are improved models of physician learning and international diffusion studies.

Key opinion leader selection

Life sciences firms often stimulate reviews of their therapy by select key opinion leaders because such leaders may serve as product champions to their peers. The effect of such opinion leaders on other physicians’ prescriptions can be large when considerable uncertainty exists (e.g., a change in the regulation or the introduction of a new therapy) or when physicians experience normative pressures (e.g., there is strong formulary adherence) (Coleman, Katz, and Menzel 1966; Iyengar, Valente, and Van den Bulte 2008). For example, Nair, Manchanda, and Bhatia (2006) show that the effect of opinion leader prescriptions is 100 times larger than the detailing effect on regular physicians after the market underwent a change in National Institutes of Health guidelines.

However, we cannot take the positive role of opinion leaders for granted (e.g., Van den Bulte and Lilien 2001), and further research should inventory the contingencies that affect the role of opinion leaders. In such research, it is worthwhile to consider two types of key opinion leaders with potentially differential effectiveness: clinical and market leaders. Clinical leaders are experts within the respective disease and therapy class with a strong reputation, as evidenced by their publication records in top-ranked medical journals. They are typically involved in premarket product testing and have cooperated with the firm to reduce clinical uncertainty of the therapy. In contrast, market leaders are tightly connected to the local patient and physician communities. They are typically general practitioners with large practices who gain recognition by the satisfaction and loyalty of their patients. They deliver key experiential messages on the therapy to their peers.

For example, as a contingency factor, consider whether uncertainty manifests in terms of effectiveness or side effects of a life sciences therapy. The impact of uncertainty on effectiveness can be reduced through quantitative assessments without much detail on specific physician practices (i.e., large scale, study based). Conversely, the impact of side effects information is more qualitative and dependent on the specific composition of a practice (i.e., case based). Because clinical leaders support quantitative assessments of effectiveness and market leaders share case detail on side effects from practices similar to other physicians, we propose the following:

Another contingency factor to consider is the physician's institutional setting. Hospitals have formal ethical guidelines (Gallego, Taylor, and Brien 2007) to which an individual practitioner must adhere, which increases the return on legitimacy compared with general practitioners. Clinical leaders enhance legitimacy to a greater degree than market leaders, which fits with their high impact on formulary decisions. At the same time, market leaders achieve their influence through similarity of practice. In general, the practice of a market leader is more similar to a general practitioner practice than to a hospital-based practice. Thus:

Researching these propositions can include surveying all physicians of a certain area to inventory their opinion leaders, including Likert-type scales on each of the identified leaders regarding the extent to which they are clinical and/or market leaders.

Sales force management

A first decision area in therapy promotion is sales force management. Visits by the sales force of life sciences firms to physicians are referred to as “detailing.” Much academic research has emerged on the effectiveness (return on investment) of detailing (Azoulay 2002; Berndt et al. 1995; Leeflang, Wieringa, and Wittink 2004; Manchanda and Chintagunta 2004; Manchanda, Dong, and Chintagunta 2004; Manchanda and Honka 2005; Manchanda, Rossi, and Chintagunta 2004; Mantrala, Sinha, and Zoltners 1994; Mizik and Jacobson 2004; Narayanan, Desiraju, and Chintagunta 2004; Narayanan, Manchanda, and Chintagunta 2005; Parsons and Vanden Abeele 1981; Venkataraman and Stremersch 2007). We derive the following generalization from this literature:

“Mean” in G₄ refers to the mean across brands and physicians. Prior literature has shown high physician- and drug-level heterogeneity, including some brands and physicians showing a negative return on detailing (Leeflang, Wieringa, and Wittink 2004), and has investigated specific contingency factors, such as drug characteristics (e.g., side effects, effectiveness [Venkataraman and Stremersch 2007], and physician traits [e.g., Gönül et al. 2001]).

There is room for further study. A first opportunity is to increase the reliability of this preliminary generalization through meta-analysis. Kremer and colleagues (2008) offer a first attempt at such generalization, but they provide only a limited number of significant moderators and omit drugs’ clinical profile. A second opportunity lies in the development of models that allow for policy experiments. Although we have reliable estimates of the mean effect of detailing, all models are estimated on data that show relatively little policy variance, which inhibits any extrapolation to policy shifts in detailing, either by the manufacturer (many firms are now considering drastically reducing their detailing efforts) or by the regulator (several European countries are considering curtailing detailing). The third opportunity lies in developing physician targeting models based on volume, physician responsiveness to detailing, and competitive detailing patterns (for working papers in this tradition, see Dong, Manchanda, and Chintagunta 2008; Kappe, Stremersch, and Venkataraman 2008).

By far, the most room for novel research seems to be in the content of detailing visits. Past and, for most companies, present detailing calls present only favorable information using positively biased information sets—that is, only studies favorable to the brand are presented, or side effects are omitted. This sales model seems increasingly dysfunctional, with hospitals and physicians reacting adversely to detailing, even rejecting it altogether, which is symptomatic for the conflicting logics between life sciences firms and the rest of the health care value chain (Singh, Jayanti, and Gannon 2008).

We propose that life sciences firms can gain substantial returns from communicating unfavorable information in their detailing calls, for two main reasons (Leffler 1981). First, in view of their ethical, gatekeeping function to patients, physicians prefer more complete information, even if unfavorable, over ambiguity. Second, communicating unfavorable information may enhance the legitimacy of the sales representative and the firm (Singh, Jayanti, and Gannon 2008). In turn, this enhanced legitimacy may deliver sustained physician access and increased trust in the firm's messages. Both will strengthen long-term return on investment from detailing. Thus:

In P_7b and P_7c, we conjecture that the effect of disclosure of complete information may be contingent on whether the disease is life threatening and on the physician's institutional setting. Agents confronted with a decision of high importance attach a greater value to information (Celsi and Olson 1988). Therefore, physicians’ preference for more complete information, even if unfavorable, over ambiguity will be higher in the case of life-threatening diseases than in the case of non-life-threatening diseases. For example, there is more value in reducing ambiguity about the side effects of chemotherapy, even if it concerns an increased probability of pneumonia versus an increased probability of insomnia caused by antihistamines. As we argued previously, practitioners in hospitals may have a higher return on legitimacy than general practitioners in the outpatient environment. Revealing unfavorable information together with favorable information enhances a sales representative's legitimacy.

There are several possible tests of P_7a–P_7c. IMS Health's U.S. panel data include data on which attributes of a drug were discussed in a sales call. Adding information on how drugs in a category compare on each of these attributes may reveal whether favorable rather than unfavorable attributes were discussed. Several individual firms have records on which studies were covered in sales calls, which can reveal whether unfavorable studies were covered. The return on investment from long-term detailing can be regressed on both types of data to test the propositions. Longitudinal experiments can also be conducted to test the propositions, in which physicians or medical school students are detailed within a simulation.

Communication management

Although communication efforts of life sciences firms may target both consumers and physicians, the budgets dedicated to the former group are more than ten times larger than the budgets dedicated to the latter (Kremer et al. 2008), and from the interviews we held with practitioners, direct-to-consumer advertising (DTCA) is also the most challenging. The academic literature on DTCA (Berndt et al. 1995; Bowman, Heilman, and Seetharaman 2004; Iizuka and Jin 2005; Narayanan, Desiraju, and Chintagunta 2004; Wosinka 2005) mostly examines overall effectiveness of DTCA and yields the following preliminary generalization:

Further research on other potential outcomes of DTCA, such as its effect on brand choice, would be fruitful because it is fraught with debate. Iizuka and Jin (2005) and Wosinska (2005) find that DTCA does not affect drug brand choice, while Berndt and colleagues (1995) and Narayanan, Desiraju, and Chintagunta (2004) find a positive effect of DTCA on drug brand choice. Such research could involve meta-analysis or the analysis of contingency frameworks.

An example of a contingency factor is the degree to which DTCA messages include favorable and unfavorable information. Although unfavorable information (e.g., information on serious side effects of therapy) may arouse consumers (Moorman 1990), it may also yield negative emotions that hinder information processing (Agrawal, Menon, and Aaker 2007; Keller 1999). Thus:

At the same time, no study develops a process view on the effects of DTCA on the demand for a specific therapy. The process involves DTCA triggering a patient's request for a therapy at the physician's office, which the physician can accommodate or not. The role of patient requests and the factors that affect the degree to which the physician accommodates them are not addressed in the academic literature at this point (for an exception, see Venkataraman and Stremersch 2007). Developing such a process view may yield relevant insights for managers (e.g., on audience targeting). As an example, consider audience gender. Prior research has shown that women are more concerned about their health (Verbrugge 1985) and interact more assertively in health care settings (Kaplan et al. 1995) than men and that physicians are more empathic to female than male patients (Hooper et al. 1982). Consequently, DTCA may more easily trigger requests among women, and female requests may be more easily accommodated by physicians than male requests. Thus:

Many other boundary conditions can be formulated on aspects such as the type of disease and patient-physician relationships, all of which may inform ad content and target audience decisions of firms. Data availability on DTCA is high. Secondary data sources include ACNielsen and TNS Media. Both data types can be connected with aggregate-level sales data (e.g., from IMS Health) or panel-level data (e.g., from IMS Health or Verispan). In addition, experimental studies may have high potential because they may reveal underlying psychological processes.

Stimulating patient compliance

As our survey results show, life sciences firms undervalue the importance of stimulating patient compliance, from both a patient welfare and a profit perspective. Our interviews with managers revealed that they consider their impact on patient compliance minimal, though they believe that it is mostly affected by the provider in his or her interaction with the patient. In contrast, our survey among providers and payers shows that they believe that life sciences firms’ efforts to stimulate patient compliance may have important effects on patient welfare.

Despite its high relevance, academic research has not studied the role of the life sciences firm in patient compliance in depth. Prior research has found that provider expertise (Dellande, Gilly, and Graham 2004), the attitudinal homophily between patient and provider (Dellande, Gilly, and Graham 2004), the frequency of contact between patient and provider (Bowman, Heilman, and Seetharaman 2004), reminder messages (Becker and Rosenstock 1984; Rosenstock 1985), and the burden of therapy (Kahn et al. 1997; Kahn and Luce 2003, 2006) all affect patient compliance. The only research that exists on how life sciences firms may affect patient compliance examines warning labels. For example, Ferguson, Discenza, and Miller (1987) find that warning labels that include information on the consequences of poor compliance are effective.

Today, life sciences firms sporadically institute new types of compliance programs, the effectiveness of which remains void of academic scrutiny. We categorize such compliance programs in technology-enabled and customer relationship management– (CRM-) enabled programs.

Such CRM-enabled programs typically used in practice are direct mail or call campaigns. Pfizer has developed a “Staying on Track” CRM program for its statin drug Lipitor (Arnold 2004). Such programs monitor patients’ disease and refill status, motivate patients to stay on therapy regimen, and provide patients with therapy risk-related information tailored to the stage of therapy with their specific symptoms and motivations (Hopfield, Linden, and Tevelow 2006).

Technology-enabled programs include a technological device to remind patients to take their pills. Bang & Olufsen Medicon's blister card-based “The Helping Hand” gives a visual indication of therapy compliance through red or green LEDs (light-emitting diodes) as soon as a blister is inserted into the device. Another example is “SIMPill,” a smart pill bottle that reminds patients through SMS (short message service) that they have forgotten to take their medicine.

Both types of programs connect to different behavioral rationales for poor compliance: a patient's belief in self-efficacy and mindfulness. A patient's belief in self-efficacy refers to the belief of being capable of carrying through the prescribed therapy, and mindfulness refers to awareness of actions to be taken (Keller 2006). Customer relationship management-enabled programs promote a patient's belief in self-efficacy, and technology-enabled programs promote mindfulness. The potential of CRM programs to promote mindfulness is limited because the reminder frequency within a CRM program is unable to match therapy frequency (one or multiple therapy occurrences a day). Conversely, technology programs cannot offer the patient interpersonal coaching (e.g., Bandura 1982) to stay on therapy.

Given their differential behavioral rationales, the effectiveness of both programs is likely to depend on factors such as disease complexity and symptom salience. First, the more complex a disease, the higher is the likelihood that poor compliance is driven by disbelief in self-efficacy. As such, CRM-enabled programs can effectively reduce such uncertainty, but technology-enabled programs cannot. Second, the less salient the symptoms of a disease (e.g., the flu is a disease with salient symptoms and high cholesterol is a disease with low salience), the more compliance will be driven by mindfulness. When salience is low, technology-enabled programs will be more effective in stimulating compliance than CRM-enabled programs.

Further research might consider a broader array of contingency factors than those developed in these propositions. Such research promises to be impactful for both academia and practice, but at the same time, it is challenging to execute. Relatively few firms have instituted a compliance program, patient-level data are difficult to obtain, and patients self-select into a program (which may cause sample selection issues). One method may be to conduct a conjoint experiment using physicians as informants on patient behavior. In such a conjoint experiment, program design factors could be manipulated, and their effect on patient compliance (as informed by the physician) could be estimated. Test-retest reliability and comparison with actual cases could further support the validity of such an approach. A more demanding alternative is cooperation with a life sciences firm that is open to a field experiment, including a longitudinal survey to the compliance program participants. More generally, the field of compliance would benefit from extensive survey research across patient-physician relationships because compliance is intrinsically embedded in this relationship.