|Home | About | Journals | Submit | Contact Us | Français|
The last fifteen years have seen a dramatic growth in US Direct-to-Consumer Advertising (DTCA) expenditures in healthcare. Despite this, DTCA remains an under-studied phenomenon of critical importance in pharmaceutical and healthcare marketing. This is emphasized in the “call for papers” of this special issue of the journal:
“Empirical research on the effect of DTCA is especially scarce. Much of the extant research on DTCA is exploratory in nature, mostly based on literature reviews and using content analysis as the methodology. Comprehensive and empirically validated models of consumer responses to DTCA are still rare in the literature and little is known about its effects on the attitudes and behaviour of concerned parties”.
To date, DTCA research has primarily been focused on the practice of prescribing drugs (DeLorme, Huh et al. 2010) and has less frequently examined the impact of DTCA on patient satisfaction or actual clinical outcomes (Gilbody, Wilson et al. 2005). In the past decade, healthcare spending on DTCA for non-pharmaceuticals has increased nearly two fold faster all other forms of promotion (Liang and Mackey 2011). For instance, the average US hospital now spends $1.3 million/yr on advertising and this is the fastest growing form of DTCA (Schenker, Arnold et al. 2014). Much of this investment is used to promote their high risk services that are profitable, such as surgical robotics (Jin, Ibrahim et al. 2011, Mirkin, Lowrance et al. 2012) or oncology (Vater, Donohue et al. 2014). Advertising claims made by not-for-profit hospitals about physician services are only subjected to the ethical guidelines of medical and surgical societies (AMA opinion 5.021; STS advertising policy2) and not regulation by either the FDA or Federal Trade Commission (Schenker, Arnold et al. 2014). As a result, this form of advertising is felt to live at the “margins of what is ethical”.
The purpose of this article is to examine the marketing efficacy of DTCA and its potential impact on the patient-physician relationship using real-world direct empirical evidence. In March 2008, our institution initiated the first robotic coronary artery bypass surgery (rCABG) program in Massachusetts. This was followed four months later by a mass media DTCA campaign using television, radio, newspaper and billboards. Our study provides novel insights into (1) the ability of DTCA to influence patient decisions about discrete therapies and to drive patient recruiting and (2) the incidental influence of DTCA on clinical outcomes and patient satisfaction. We also advance DTCA research by examining a different type of DTCA communication message (i.e., direct patient recruiting) (Nikki Lee-Wingate and Xie 2010) used in a novel medical context (i.e., cardiac surgery).
Coronary artery bypass grafting (CABG) is one of the most commonly performed and thoroughly studied surgical procedures in the medical field. Classically, this surgical procedure is performed via a highly morbid incision (i.e., a saw is used to divide the sternal bone in the midline of the chest). There are alternative interventions to treat coronary artery disease such as percutaneous coronary intervention (PCI) using balloon angioplasty and/or placement of a coronary artery stent that are less invasive. Because patients prefer less invasiveness (Pitter 2014), they are usually referred for CABG only if they meet specific indications and they are not candidates for PCI. Similar to PCI, a robotic operation enables coronary artery disease to be treated less invasively via small incisions in between the ribs that spare the sternal bone. R-CABG has been shown to reduce invasiveness and postoperative complications compared to the open chest option (Cavallaro, Rhee et al. 2015, Whellan, McCarey et al. 2016). In contrast to PCI and CABG via sternotomy, rCABG is technically complex and has not been widely adopted outside of a few expert centers. A recent analysis showed that <1% of annual CABG volume is done using the rCABG approach (Whellan, McCarey et al. 2016). In the context of a CABG market that is one of the most highly competitive in all of medicine, the failure to achieve safe and wide-spread adoption has plagued rCABG with controversy from its early beginning (King 1999) until today (Dhawan, Roberts et al. 2012).
Most cases of coronary artery disease involve preference-sensitive choices between viable options. Typically, a patient is first referred for CABG, discusses available options with the surgeon and finally an elective operation is scheduled over a timeframe ranging between 1–3 weeks. This typically provides an adequate window for the patients themselves to participate in the decision about where care is received. However, the vast majority accept the advice of their provider about where to have CABG and rarely investigate options on their own (Wilson, Woloshin et al. 2007). Legally, informed consent must “entail an opportunity to evaluate knowledgeably the options available” (Cantebury v. Spense, 1972). Consistently, patients have shown strong preferences towards less invasive options (Blue Cross Blue Shield of Massachucetts, 2007, Boden, 2008). Yet, it has been uncommon for centers that only offer traditional surgery to disclose robotic surgery as an option (Scherr, Fagerlin et al. 2016). This lack of disclosure about less invasive options may result because its proposed advantages are less meaningful to physicians than their patients (Cooper, Hutfless et al. 2014). An explosive growth of health information has created a shift in consumer sophistication and a platform for hospitals to articulate their most attractive innovations directly to potential patients (Larson, Schwartz et al. 2005). DTCA is one strategy that seems favorably positioned to leverage the latent patients’ desire for information about less invasive procedures like rCABG, raise patient awareness for this surgical procedure, and recruit new patients for the medical centers where it is performed.
Previous research on DTCA of hospital services has shown that most physicians believe DTCA motivates patients to seek specific treatments and also changes their expectations of their doctors (Robinson, Hohmann et al. 2004). However, the same research also demonstrates that few patients believe DTCA motivates them to seek a specific type of care or changes their expectations vis-a-vis their doctors. Although there appears to be a wide perceptual chasm on how the two sides of the patient-physician dyad perceive the impact of DTCA, both sides agree that DTCA does not provide enough information on treatment options. Notwithstanding the general merits of these types of self-reported opinion surveys, it is also important to consider direct evidence based on actual clinical data in order to truly understand DTCA’s effectiveness and its impact on patient satisfaction and expectations.
The intent of the DTCA campaign concerning rCABG was to highlight the advantages of this novel procedure to potential patient candidates and health care providers, while maintaining realistic expectations regarding outcomes related to the procedure. In this context, our research objectives were to evaluate the overall DTCA campaign efficacy based on the number of referrals caused by the campaign relative to the estimated number of potentially appropriate candidates in the market. In addition, we investigated how well the DTCA campaign communicated claims that were relevant and accurate as evidenced by a variety of outcomes such as patient satisfaction.
Based on previous research, we expected that DTCA carried with it a risk of distorting patient expectations (Gilbody, Wilson et al. 2005) and that those expectations would play a critical role in determining patient satisfaction with a treatment (Zeithaml 1990). Drivers of satisfaction and consumers’ perceptions of service quality have been well described in the SERVQUAL model (Parasuraman, Zeithaml et al. 1985), which has been successfully applied to studying service quality in many service settings, including healthcare delivery. Using the assumption that “quality” from the standpoint of the consumer is generally driven by the difference between his/her service expectations and his/her perceptual assessments of the service outcomes, SERQUAL identifies five discrete gaps in the service delivery chain. These gaps arise from differences in: 1) consumer expectations and the provider's perception of those expectations (empathy), 2) provider perceptions of consumer expectations and actual service quality specifications (reliability), 3) service quality specifications and delivery (technical quality), 4) external communications (e.g. advertisements) about the service and actual service delivery (honesty), and 5) expectations and actual perceptions of the service (functional quality) (Parasuraman, Zeithaml et al. 1985). In this field-experiment setting, we were able to compare outcomes and unmet expectations among patients who responded to the ads (DTCA responders) to those of a control group referred via routes that are traditional for a cardiac surgical practice (e.g. cardiologist referral, word of mouth, etc.). Because we could expect (and/or statistically control) that DTCA responders and controls would have similar clinical experiences (same surgeon and hospital), we theorized that any satisfaction difference between the two patient groups should be explained by differences in their expectations and/or motivation to be critical of their outcome. The hypothesis of this study was that the process of responding to DTCA creates higher expectations and greater tendency be dissatisfied with rCABG than controls.
In July 2008, our institution initiated a DTCA campaign regarding its rCABG program, the first and only one in Massachusetts at that time, using a variety of media outlets including television, radio, newspaper and roadside billboards. Each ad provided the address of our website with further program details and contact information (i.e. the personal email address of the lead surgeon and office phone number). An example of the TV ad is available online3. Healthcare marketing campaigns were common in Boston (Rowland 2006), particularly for profitable services such as cardiac surgery (Resnick, Corrigan et al. 2005). Most ads aimed to build an image and reputation and not drive patients towards specific procedures. The campaign by our institution was unique in its clear goal to influence patient decisions about a discrete therapy.
Between July 2008 and July 2009, all patients contacting the cardiac surgery office to request an appointment to discuss rCABG were logged into a prospectively collected database that included information about whether patients learned about our program through DTCA. An established relationship existed between the cardiac surgeon and referring cardiologist in virtually every case prior to our DTCA campaign and for all control patients in the study. This made it straightforward to screen for those patients that contacted our office and whose cardiologist had no prior referral relationship to our group. We confirmed in 103 patients that the marketing campaign triggered their consultation (“DTCA responders”) after contacting their primary care providers and/or cardiologists. A subgroup of 54 DTCA responders was appropriate for rCABG and underwent this procedure at our institution. All other rCABG cases during the study period were confirmed to have been referred via traditional means and not due to DTCA (Controls, n=77). Complete clinical data was available for all the controls but only 43 DTCA Responders, 11 were from outside Massachusetts and did not return for 30 day follow-up and were excluded.
The inclusion criteria were all patients presenting to Boston Medical Center during the study timeframe that had been diagnosed by their cardiologist with severe coronary artery disease that met standard indications for surgical revascularization (Hillis, Smith et al. 2011) and their coronary targets were deemed suitable for grafting via a robotic approach. Patients were excluded from the study if they did not have an indication for CABG, their coronary artery blockages were unable to be completely revascularized or they were contraindicated for less invasive surgery (e.g. unstable blood pressure, severe pulmonary disease, or decompensated heart failure). Patients with no valid contact information or did not return for postoperative follow-up were also excluded. Study patients were assigned to the DTCA group if they were “self-referred” and their cardiologists did not refer them for the robotic option. The remainder were told about robotics by their cardiologist and stated that DTCA did not lead them to the option of rCABG. These patients were assigned as controls.
Written informed consent for anonymous data collection and analysis of all patients was obtained before the operation. The local institutional review board (IRB) approved the study and waived the need for additional patient consent.
Patients in the DTCA and control groups were all treated with the identical surgical approach - rCABG – as described per the protocol used by our group provided elsewhere (Poston, Tran et al. 2008) and illustrated in a video posted online4. Any differences in therapy were due to individual differences in specific coronary anatomy, operative risk factors, and co-morbid disease. In appropriate patients, a hybrid strategy was used with PCI performed in a separate setting using drug-eluting stents where aspirin and clopidogrel were given prior to PCI (300 mg followed by 75 mg daily thereafter). Long term success of this new procedure was determined by phone interview with each patient at 4 years after surgery to determine freedom from the need for repeat coronary revascularization or recurrent angina (chest pain related to the heart).
Background demographic and clinical information was gathered from the clinical records. Demographic information included age, sex, race, marital status and level of formal education. Clinical information included relevant medical history, comorbid conditions, laboratory information, medication profile. Patients without clinical follow-up were excluded from the study, leaving 43 patients in the DTCA group. Patient satisfaction surveys (i.e. HCAHPS) were returned at 3 weeks from 61% of the DTCA cohort and 63% of controls. The collection of these data by the hospital was restricted by protocols designed to satisfy federal requirements for the uniform measurement of patient satisfaction which constrains attempts to increase the response rate5.
Outcomes of interest during the study included the following:
Each group was assessed for factors expected to influence patient satisfaction: (1) demographical information including level education and marital status, (2) the incidence of major complications at 30 days, (3) perceptions of pain control during hospitalization and (4) recovery of physical function at 3 weeks. The level of education and marital status was obtained from each rCABG patient by phone interview. The incidence of major complications during the first 30 days post-surgery (i.e. mortality, permanent stroke, deep sternal infection, reoperation for bleeding, renal failure, and intubation for longer than 24 hours) was obtained from the Society of Thoracic Surgeons (STS) National Database. Pain control was defined by patient response to question #13 on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey mailed approximately 3 weeks after discharge6. Specifically, we monitored the percentage of patients answering “always” to the question: “how often pain was well controlled?” Recovery of physical functioning was assessed at 3 weeks after discharge using the Duke Activity Status Index (DASI) questionnaire. The DASI is a brief, standardized, self-administered questionnaire administered via mail that gauges the patient’s ability to perform common activities and provides a weighted score (range: 0–58.2) that assesses overall functional capacity (Hlatky, Boineau et al.).
Study subjects were again contacted by phone at 4 years after surgery and asked the following questions: 1) have the doctors performed any other intervention on your heart since your original surgery, including repeat surgery or another angioplasty/stent procedure?, 2) have you had any recurrent angina (chest pain related to the heart) since the original surgery? To minimize response bias, the surveys were confidential, with only a non-surgeon reviewer having access to the identifying patient information.
DTCA penetration was defined as the number of inquiries about the ad to our office divided by the estimated target population for the ad. The typical target population for rCABG is patients referred for surgical revascularization that require only 1 or 2 bypass grafts to completely address all the blocked coronary arteries. Patients requiring bypass of a more extensive number of blockages require a conventional (i.e. sternotomy) approach and are not usually suitable for rCABG. The number of patients who underwent CABG at any Massachusetts hospital using either 1 or 2 coronary bypass grafts during the study period (n=934) was available from the state’s publically reported CABG database: Massachusetts Data Analysis Center-Cardiac Surgery and Percutaneous Coronary Intervention (MASSDAC)7. This information defined our target population. Due to the scope of the study, carry-over effects of DTCA (e.g. r-CABG cases done after the campaign) were not considered or measured for this analysis.
The primary endpoint of this study was patient satisfaction as defined by a comparison of perceptions vs. expectations of rCABG. A custom survey (see below), performed by phone interview, was obtained at 6 months. The survey instrument consisted of questions that assessed the three claims about rCABG made in each of the mass media ads (other than billboards): 1) quick recovery, 2) small incisions and 3) the expertise of the surgeon (RP). At the 6-month interview, each patient was asked about their actual perceptions compared to what they expected prior to surgery. The results were categorized using a Likert (1 to 5) scale. A response of 2 (“less than expected”) or 1 (“far less than expected”) was categorized as “unmet expectation” according to prior criteria (Freilich, Penna et al. 2010). In order to evaluate for potential characteristics associated with unmet expectations, post-hoc subgroup analysis comparing baseline characteristics and clinical outcomes of patients with unmet expectations to the rest of the cohort was also performed.
As a control for unmet expectations, more generalized patient satisfaction with the hospital and staff was determined using the HCAHPS survey at approximately 3 weeks after discharge. This was performed by mail, with surveys returned from 61% of the study cohort and 63% of controls. A raw composite satisfaction score was generated on a Likert scale from answers to the first 25 questions of the survey and then converted to an interval (1–100) scale to obtain an overall score as per HCAHPS protocol. A sample of the survey is available online8.
The primary outcome of comparison between the two groups was length of hospital stay as a surrogate for recovery time and self-efficacy (Hall and Dornan 1990). Univariate linear regression was used to adjust for potential confounders (independent variables) on length of hospital stay (dependent variable). Potential confounders determined a priori included age, sex, and baseline clinical comorbidities found to be significantly different between groups (e.g. pre-existing chronic lung disease). We also performed a post hoc comparison of patients with all expectations satisfied compared to those that reported unmet expectations. These analyses were performed using the Fisher’s exact and Student’s t-test for categorical and continuous variables, respectively. Continuous variables were subjected to a Komogorov-Smirov test for normality. Non-normally distributed data were subjected to the Mann-Whitney U test for continuous variables. Statistical significance was set a p-level of 0.05. All analyses were performed using IBM SPSS Statistics 21.
During the 12-month study period (July 2008 and July 2009), a total of 103 potential patients, or their representatives, contacted our office as a consequence of the DTCA campaign for a “second opinion” consultation. After this consultation, 71 were found to be appropriate candidates for rCABG (i.e. DTCA responders). Out of these candidates, 54 (76%) went on to have rCABG surgery, 2 (3%) underwent nonrobotic procedures and 15 (21%) chose to not have surgery at our institution. During the study period, an additional 77 patients were referred by traditional means and underwent rCABG at our institution. These patients were enrolled into the control group.
Clinical follow-up at 30 days was complete in controls (77 out of 77 patients) but incomplete in the DTCA cohort (43 out of 54 patients). A lower proportion of DTCA patients lived within the primary service area of the hospital (13 vs. 74%), which likely increased the difficulty of returning to our office for clinical follow-up (Table 1). Surveys at 6 months and 4 years were obtained from all but one of the eligible study patients (1 control patient lost to follow-up).
DTCA-responders who underwent this procedure were younger, less likely to have lung disease and more likely to be scheduled as an elective case when compared to controls (table 1).
Comparing the early clinical results for the DTCA-responder and control groups, there were no differences in the lengths of time on the ventilator or in the intensive care unit (ICU) (Table 2). Mean total hospital stay (dependent variable) was less in the DTCA group (5.9 vs. 7.9 days, p=0.011, F=-2.321) but did not remain significant after adjusting for age, sex, and preoperative comorbidities (independent variables) in a linear regression model (6.3 vs. 7.8 days, P=0.072, R=0.244). At three weeks, there were no significant differences in the degree of activity tolerance between groups when expressed in absolute value (DASI scores: 21.8 vs. 21.2 out of 58.2, Table 2). At 4-year follow-up, there were two patients in each group that had required repeat revascularization due to failure of a bypass graft (96.4 vs. 97.5% freedom from revascularization, p=1.00).
The DTCA campaign generated 71 qualified leads out of a potential market of 934 CABG cases involving 1 or 2 bypass grafts that were performed in the local region during the study period (i.e. 7.6% potential share of the market). There were 54 leads that were converted to a surgical case at our center (i.e. 5.1% actual share of the potential market). Carry-over effects (i.e. self-referrals after the campaign) were not measured.
Short-term satisfaction scores, obtained from HCAHPS surveys returned from 61% of the study cohort and 63% of controls, showed no differences between the DTCA and control group (Table 2). Both groups were very satisfied with the surgery and the immediate outcomes. However, when patients were surveyed six months after the surgery and asked to reflect on their rCABG experiences, a significantly smaller percentage of DTCA patients reported that their expectations had been fully met (i.e. response to all 3 questions with ≥3 rating on a 1 to 5 scale) compared to controls (82.1% vs. 95.7% of patients/group, p=0.033, Table 2). This difference was driven by unmet expectations (i.e. responses ≤2) regarding the “size of the incision” and “speed of recovery” but not the “expertise of your surgeon” (Table 2).
A post hoc subgroup analysis comparing patients with all their expectations satisfied to patients with unmet expectations (n=12) demonstrated that both groups had did not have statistically significant differences marital status (78% vs. 83% married), level of education (56% vs. 67% college or higher) and equivalent 30 day rates of complications (6.6 vs. 7.4%) and satisfaction scores at 3 weeks (85.9 vs. 85.1).
Our study suggests that DTCA was effective at communicating to prospective patients within the lay public about a complex surgical procedure (rCABG). During the campaign, new patients arrived from outside of primary service area of the hospital to seek a second opinion from the sole program in the region that offered rCABG. This suggests that DTCA did not just cannibalize existing referrals but actually stimulated primary demand for rCABG. Within the context of a media market already saturated with healthcare advertising (Rowland 2006), our one year DTCA campaign generated qualified leads representing almost 8% of the potential market segment (i.e. subset of patients referred for CABG that were likely suitable for rCABG). In contrast, the typical DTCA campaign from other healthcare fields are expected to prompt action in <2% of their target population (Liebman 2001). While many factors contributed, we speculate that the primary determinant of this response was the strong valence of rCABG as an advertising message. Prior evidence has shown that patients often have unmet information needs about less invasive surgical options (Larson, Nelson et al. 1996, Horwitz 1999), and view DTCA as a helpful and legitimate source for information about complex medical topics (Bozic, Smith et al. 2007, Abel, Burstein et al. 2009).
Other studies have found it difficult to quantify the effectiveness of hospital ads at generating patient referrals. Several unique circumstances in the New England cardiac surgical market in 2008–2009 made an estimate possible for our study. First, the ad campaign stood out to the public because of its clear purpose – to alter the speed with which a new patient-centered innovation was recognized and popularized. Among the world-class CABG programs in Boston at the time of this DTCA campaign, our center was the only that would offer rCABG to patients that expressed interest in this idea. Second, the referral channel for CABG is well established. Every patient that requires CABG has a cardiologist. Prior to DTCA, a cardiologist’s referral was the sole means of a patient coming to our center for surgery. Self-referral for rCABG was a conspicuous reason to assign a patient to the DTCA cohort that could be corroborated through follow-up interviews with patients and their providers. Mandatory participation in a cardiac surgery database for all programs in the state provided details about the CABG market within the region. This enabled an accurate estimate of the number of potential candidates for rCABG during timeframe of the ads.
Control patients were provided information about rCABG by their providers and then referred to the surgeon, which is standard practice. The DTCA group found out about rCABG from the ad and had to “self-refer”, which may have created a bias in this group for patients with more self-efficacy, greater health literacy and potentially an economic motivation for faster recovery (Mackert and Love 2011, Junewicz 2014). A variety of studies have shown that these characteristics predict improved recovery in the setting of cardiovascular disease (King, Humen et al. 2001, Jackson, Leclerc et al. 2005). At the same time, they can set the stage for a disturbed patient-physician relationship. Many DTCA patients requested a referral for rCABG and were denied. Prior evidence suggests that such denials provoke dissatisfaction, particularly in males facing a life threatening situation (e.g. severe coronary artery disease), (Blose and Mack 2009). Patients that “self-refer” are often left feeling they took sole responsibility for this high-stakes decision, which creates a risk for decisional regret in the future (Nordgren, van der Pligt et al. 2007). These associations are likely generalizable to our study and therefore require careful consideration as our findings are interpreted.
Our study design does not allow the impact of DTCA to be teased out amid other information sources (e.g. opinions of friends, family, medical providers) and cognitive biases triggered within the minds of patients making decisions under conditions of uncertainty. It is possible that some factor other than the ads might have prompted the decisions of patients in the DTCA cohort to choose rCABG. However, our analysis suggests that the impact of such unmeasured factors was likely to have been modest. The DTCA cohort was derived from a primary service area that included a higher proportion of working age males, with more education, less chronic lung disease and a less urgent cardiac problem than controls. Prior to the DTCA campaign, Boston Medical Center had a reputation as an “inner city” surgical program that recruited patients from a much different market segment. The temporal relationship of the arrival of a new market segment to the timing of the ad campaign strengthens our conclusion that DTCA was responsible.
One plausible explanation for the strong response to our ad campaign is that standard practice in Boston at that time did not sufficiently expose patients that needed CABG to the option of rCABG. Many of the patients enrolled in DTCA group learned about rCABG at relatively late stages in their work-ups for surgery, imposing a time constraint that could exacerbate the emotional intensity of their decision. If patients referred for CABG were routinely provided information about all available alternatives like rCABG from their providers or by their own independent research, one would have expected less response to our mass media ads. Conversely, their response to information provided in a mass media ad implies that they would have preferred this information to be disclosed during informed consent by their providers. The providers either misunderstood or were unable to meet this expectation, creating a service gap (i.e. Gap 1, empathy). Prior evidence suggests that intermediaries in the surgical referral channel have been either unwilling or unable to share information about previously undisclosed and underutilized services such as less invasive surgery (Cooper, Hutfless et al. 2014). While the exact reasons for this channel deficiency are unclear, one possible contributor may be that many of these patients’ healthcare providers were unable to provide confident advice about a novel procedure for which they had no personal experience. Variation in the use of “preference sensitive” therapies like the traditional open sternum CABG often reflects preferences of the surgeon rather than those of the patient (Wilson, Woloshin et al. 2007). Decision aids help patients by slowing down the decision process (Rosenbaum 2015). A Cochrane review of 86 studies on shared decision making showed that patients provided with information through decision aids are more likely to choose surgical options that are less invasive (Stacey, Bennett et al. 2011). DTCA acted as a decision aid in context of this study because it led interested patients that were often already referred for standard CABG to pursue a second opinion about rCABG, thereby closing this service gap for the desired information.
Conversely, there are also concerns about DTCA. Robotic surgery as a marketing message is known to trigger emotional responses in patients (Dixon, Grant et al. 2014) and these emotions can drive patient decision making. Ad campaigns aim to provide information vivid enough to create a patient response and generate leads (Rook 1986) which in our case involved patient testimonials and compelling audiovisual images. Although rather mild by general advertising standards, any approach that aims to be persuasive might be seen as too “aggressive” in a healthcare context. There is a risk of overriding autonomy if patients establish a favorable decision about rCABG for no other reason other than they first heard about it within the controlled framework of our ads (McKneally 2002, FTC 2009). Emotionally charged information about new medical technology can take advantage of patient vulnerability caused by their overall poor health, their search for a miraculous cure and their inability to easily verify claims (Urbach 2016). Expensive ad campaigns can exacerbate the perception of the profit motive by the surgeon and hospital, further hindering the therapeutic relationship.
Ad claims might have “oversold” rCABG and caused unrealistic expectations from overt deception (i.e. Gap 4, honesty). Our study design included three safeguards against “overselling”. First, we obtained prospective feedback about the ads from a multidisciplinary committee of healthcare professionals at our institution. Second, we obtained retrospective feedback from the Better Business Bureau in Boston, the agency responsible for reviewing healthcare ads by nonprofit hospitals. Both groups compared the ads against published evidence about rCABG (Poston, Tran et al. 2008) and confirmed that the ad claims were understandable, not likely to be misinterpreted and not deceptive. Finally, we noted that the rate of unmet expectations in our DTCA cohort was similar to patients undergoing other types of high impact surgery that were not exposed to any ad claims (Sheehan, Sherman et al. 2007, Poston, Tran et al. 2008, Schroeck, Krupski et al. 2008). The clinical outcomes of those with unmet expectations was not found to be different in a post hoc analysis.
Even if the ads did not exceed standards for deception based on regulatory and professional perspectives, it is still possible that they implicitly instilled unrealistic expectations among the DTCA cohort. To evaluate this possibility, we obtained feedback from all study patients 6 months after rCABG. This demonstrated a rate of negative disconfirmation that was significantly higher in the DCTA cohort than controls (18 vs. 2.6%, p=0.007). Most patients find it difficult to rate their experience about a highly technical service like rCABG (Dixon, Grant et al. 2014). Psychological theory suggests that patient’s perceptions and memory recall for a complex service are often influenced by the positive framing effect of DTCA, which serves as a favorable cue that can substitute for independent judgment (Hoch and Ha 1986, Braun-LaTour and LaTour 2005). The DTCA cohort was exposed to ads prior to surgery, but our campaign had such a high degree of penetration that both groups were likely to be exposed at some point between surgery and their 6 month survey. This caused a systematic difference in the timing of ad exposure relative to the performance of rCABG that may help explain why unmet expectations at 6 months varied between groups. Controls exposed to ads after surgery may have responded to the positive framing effects of the ad, while only the DTCA cohort was exposed prior to surgery, which may have had the main effect of setting high expectations. The DTCA group had more education than controls. Prior research suggests that education, as an indicator of cognitive ability, moderates the effects of framed advertising (Sicilia, Ruiz et al. 2005). Therefore, the DTCA group may have been more capable than controls at discriminating the ad claims from their perceptions and considered this ability as being more diagnostic than an ad. Consistent with this idea and the findings of others, the DTCA group had gaps in their expectation vs. perceived quality (i.e. Gap 5) around two central and relatively unambiguous messages of the DTCA campaign that were relatively easy to judge (“size of the incision”, “speed of recovery”) but not about more technical issues that are difficult to judge (e.g. “expertise of the surgeon”). Our protocol did not control for exposure to the mass media ads after surgery and did not directly assess its psychologic influence, so these ideas remain highly speculative.
Taken together, our analysis suggests that DTCA improved a gap related to a patient’s expectation for information that is important to them (i.e. Gap 1, empathy), but a potential tradeoff was the risk of a new gap (i.e. Gap 5, functional quality). Even if the ads themselves did not directly cause any service gap, a mere association alone is important. Dissatisfaction with any aspect of healthcare directly relates to poor clinical outcomes and future risk of a malpractice claim. This underscores the need for providers of patients exposed to DTCA to spend time during the initial consult, pre- and post-operative hospital stay checking, managing and correcting expectations so that they are aligned with correct interpretations of ad claims. In response to this study, we implemented changes to improve communication (e.g. 24 hr access to the surgeon’s cell phone and a secure email after each initial office visit). Although corrective action such as this is an extra burden for healthcare providers (Gilbody, Wilson et al. 2005, Bozic, Smith et al. 2007), it is an essential one in order to minimize “false-hopes” that can be created (albeit completely indirectly) on aspects not even addressed by the ad. Our experience illustrates that careful forethought is required to identify and manage potential gaps in service quality that may result from DTCA promotion of a high risk procedure like rCABG.
An alternative point of emphasis for future ads might be to highlight the intrinsic value of a second opinion. Growing evidence suggests that being provided choices, particularly those that depend on the context of a patient’s social circumstances and willingness to accept risks, improves satisfaction and outcomes (King, Norris et al. 2009). In reality, any surgical patient (whether robotics or DTCA is involved or not) is at risk for an outcome that is less than optimal. The fact that results are never guaranteed is often not reflected in the current messages provided in DTCA about robotic surgery (Dixon, Grant et al. 2014). External marketing that focuses on the value of a second opinion is more likely to remain consistent with the information provided as part of informed consent about rCABG that is more dispassionate and reflects the reality of risk.
An important limitation of our study was the lack of validated instruments to survey rCABG patients. We screened for the DTCA cohort by calling patient’s referring providers. Alternatively, a dedicated toll-free number or a website address assigned to this specific promotion could have been used. This more accurate method for confirming group assignment might have excluded some patients from the DTCA cohort, which would affect our conclusions about ad penetration but not the influence of ad exposure on unmet expectations. There are no available tools for assessing disconfirmation of CABG patients, so we adopted the standard SERVQUAL survey to our purposes. While not formally validated as a measure of negative disconfirmation, our survey was context and objective-specific. Measurement items were directed explicitly at those claims made by the ad that could cause negative disconfirmation, thereby avoiding the ceiling effect associated with generic satisfaction surveys after heart surgery. We chose the 6 month time-point to minimize the chance for a recall bias and because this is generally accepted to represent a final outcome for the aspects of care that were investigated. Our overall approach was adopted from prior studies of postsurgical disconfirmation (Schroeck, Krupski et al. 2012) and has been shown to improve response rates without affecting the reliability and validity of this instrument. Our assessment of unmet expectations required subjects to remember 6 months after surgery what their expectations were prior to surgery. This type of memory is subject to inaccuracies and bias. Future research should assess DTCA respondent expectations before surgery to provide a more objective comparison with their perceptions measured after the surgery. Finally, we were aware that carry-over effects (i.e. patients who sought rCABG after the completion of the campaign) may have existed, but measuring them was not within the scope of this study. The fact that this type of confounding is possible weakens any conclusion that the ads had a causal effect on expectations.
A mass media ad can only be expected to alter a serious and potentially life-threatening healthcare decision when it provides a compelling message. The penetration of our DTCA campaign estimated in our study suggested that rCABG was a topic that provided this type of compelling message. Caution should be taken in extracting strong implications from these results given the nature and limitations of the study. Nevertheless, the main finding of a difference in satisfaction at 6 months post-surgery between those responding because of the DTCA campaign and those referred the traditional way warrants further research. More unintended service gaps occurred in our DTCA cohort despite our efforts to manage expectations and build trust. As marketing efforts continue to expand at US hospitals, our findings suggest the need for further research into DTCA used to promote high risk surgical procedures. Such research could be used to help refine the messages of the ads and discover better ways to manage expectations in order to improve the overall effectiveness of these campaigns.
5Federal Register / Vol. 78, No. 160; § 412.140 (f.1-2)