Direct-to-consumer campaigns sponsored by the pharmaceutical industry, including direct-to-consumer advertising (DTCA) and direct-to-consumer information (DTCI) campaigns, have been a recent focus of prescription drug-use promotion in developed countries [1–4]. DTCA campaigns directly promote specific brand name medications, whereas DTCI campaigns are aimed at informing people about seeking healthcare services but should not directly promote specific drugs [4–7].

In the US and New Zealand, pharmaceutical companies have widely conducted DTCA campaigns, which have increased the use of prescription drugs in those countries [1, 3, 4]. Spending for DTCA campaigns has increased from USD 1.2 billion in 1998 to USD 4.5 billion in 2009 in the US [1, 4]. In many countries including Japan, where direct brand name drug promotion has been prohibited, DTCI campaigns (also known as disease awareness campaigns) have rapidly increased [5–7]. The current Japanese government regulations prohibit a DTCA campaign due to its direct promotion of a specific drug, but do not prohibit a DTCI campaign without such direct promotion of a specific drug. As a result, pharmaceutical industries are practically able to conduct DTCI campaigns in Japan. Although DTCI campaigns, including public awareness campaigns for specific diseases, are distinguished from the advertisement of specific drugs by law and permitted in many settings, some recent reports raised concerns about their potential advertising effects [6, 7]. However, little is known about how DTCI campaigns affect the patterns of prescription of treatment drugs.

In 2011, Astellas Pharma Inc. conducted a 5-week DTCI campaign for overactive bladder, a condition with complex urinary symptoms that affects quality of life in Japan [8]. Prior to the start of this DTCI campaign, a large survey conducted by the Japanese Continence Society in 2003 suggested that awareness about treatment for overactive bladder treatment was low [9]. The informational content of the DTCI campaign sponsored by Astellas was that overactive bladder could be cured with pharmaceutical treatment [8]. We investigated the potential change in the prescription patterns for overactive bladder treatment drugs as a result of the DTCI campaign in Japan.

We analyzed a large health claims data set to test our hypothesis that the 2011 DTCI campaign increased the prescription rate for the treatment drugs among patients with overactive bladder who had not previously used the treatment drugs.

Of 795,370 enrollees (458,198 male and 337,172 female) covered by the JMDC from May 1, 2010 to November 18, 2010, a total of 2812 (0.3%) patients with overactive bladder met the inclusion criteria. Excluding 1480 patients who were prescribed the treatment drugs during the run-in period, the final sample size for analysis was 1332 participants.

The Kaplan–Meier estimate (Fig. 1) indicated that compared with the other periods, the prescription rate clearly increased from the end of Period 4 to Period 6 in Year 1, 10–15 weeks subsequent to the end of the DTCI campaign.

The crude prescription rate ratios during Periods 4–6 in Year 1 compared with Period 10 in the Pre-Campaign Year were 7.13 (47.8 [per 100,000 person-days]/6.69 [per 100,000 person-days]), 13.7 (91.8/6.69), and 4.28 (28.7/6.69), respectively, which are similar to the magnitude of the estimated HRs during the same periods.

The ITSA did not show a one-time increase in the regression-intercept during Period 1 to Period 3, as well as Period 5, in Year 1. However, as expected, the DTCI campaign significantly increased the regression-intercept at Period 4 in Year 1, which coincide with the result of our Cox model (Fig. 2 and Table 3): β₂ = 1128.1 (95% CI, 181.7 to 2074.4; p-value <.01). This result suggests that the DTCI campaign raised the level of prescription rate by 2.4 times (β₂/β₀) with a 15-week delayed time lag. This substantial one-time increase of prescription rate was followed by the decreasing time trend (long-term effect, p-value <.05).

Characteristics	Coefficient (95% Confidence Interval)b
Main analysisa
Interrupted time period at Period 4 in Year 1
β0: starting level of first-time prescription rate	471.9 (161.4 to 782.4)**
β1: time	− 10.4 (− 51.3 to 30.5)
β2: delayed effect of DTCI campaign (since Period 4 in Year 1)	1128.1 (181.7 to 2074.4)*
β3: time × delayed effect of DTCI campaign	−77.6 (− 175.0 to 19.7)
β1 + β3: post-intervention linear trend	− 88.1 (− 170.5 to −5.60)*
Sensitivity analysesa
Interrupted time period at Period 1 in Year 1
β0: starting level of first-time prescription rate	506.5 (102.0 to 911.1)*
β1: time	−18.4 (−94.8 to 58.0)
β2: effect of DTCI campaign (since Period 1 in Year 1)	757.3 (− 149.4 to 1664.0)
β3: time × DTCI campaign	−21.4 (− 129.1 to 86.3)
β1 + β3: post-intervention linear trend	−39.8 (−99.0 to 19.4)
Interrupted time period at Period 2 in Year 1
β0: starting level of first-time prescription rate	412.7 (−8.80 to 834.2)
β1: time	10.7 (−62.4 to 83.8)
β2: effect of DTCI campaign (since Period 1 in Year 1)	552.9 (−584.2 to 1690.0)
β3: time × DTCI campaign	−54.4 (− 162.3 to 53.5)
β1 + β3: post-intervention linear trend	−43.7 (− 120.3 to 32.9)
Interrupted time period at Period 3 in Year 1
β0: starting level of first-time prescription rate	439.3 (114.0 to 764.5)*
β1: time	−1.63 (−51.6 to 48.3)
β2: effect of DTCI campaign (since Period 1 in Year 1)	846.6 (− 234.8 to 1927.9)
β3: time × DTCI campaign	−62.2 (− 158.2 to 33.7)
β1 + β3: post-intervention linear trend	−63.9 (− 145.9 to 18.2)
Interrupted time period at Period 5 in Year 1
β0: starting level of first-time prescription rate	393.0 (−198.7 to 984.7)
β1: time	12.6 (− 101.9 to 127.1)
β2: delayed effect of DTCI campaign (since Period 4 in Year 1)	874.4 (− 1046.5 to 2795.4)
β3: time × delayed effect of DTCI campaign	−106.2 (− 272.2 to 59.9)
β1 + β3: post-intervention linear trend	−93.6 (− 227.2 to 40.0)

*p < .05 **p < .01 ***p < .001

^aSample size, N₁ of analyzed time periods = 30. The aggregated data samples were extracted from 1332 patients who were diagnosed with overactive bladder before May 2010 and who had not been prescribed a treatment drug during May 2010 to November 2010

^bCoefficients and 95% CIs were estimated for the first-time prescription rate per 5 weeks among a standardized 100,000 patients with overactive bladder who had not used the treatment drugs previously

Likewise, during November 2010 to November 2012, a significant one-time increase in the regression-intercept was seen in other outcomes, i.e., the number of new diagnosis and the number of newly diagnosed patients treated with medication in the whole JMDC cohort (Additional file 1: Table S1). In addition, for the number of newly diagnosed patients treated with medication, the significant inverse time trend of the slope before the DTCI campaign changed to null (Additional file 1: Figure S1 and Table S1), suggesting that patients newly diagnosed with overactive bladder were consistently prescribed medication regardless of time trends only after the DTCI campaign. In the Cox model and ITSA, the release of mirabegron was not associated with the prescription rate (data not shown).

This study is the first to find the actual change in prescription patterns as the result of a DTCI campaign in a cohort, by analysis of a large health claims data set at an individual level. As hypothesized, the DTCI campaign for overactive bladder increased the prescription rate for patients with overactive bladder who had not previously used the treatment drugs. The duration of this increase was 15 weeks with a 10- to 15-week lag after the end of the DTCI campaign. The magnitude of the increase was 4–13 times that of the pre-DTCI prescription rate. The aggregated ITSA confirmed that the DTCI-campaign was associated with the one-time increase of the level of prescription rate with the magnitude of 2.4 times that of the pre-DTCI campaign period, holding a similar time lag of 15 weeks. The post-DTCI decreasing slope (long-term effect) might be partly caused by this one-time increase in the prescription rate. Additional aggregated ITSA implied the DTCI-campaign’s impacts on other outcomes such as (a) positive immediate impact on the numbers of new diagnosis and newly diagnosed patients treated with medication, and (b) the positive long-term impact on the number of newly diagnosed patients treated with medication.

Our key empirical findings regarding the timing of the DTCI campaign’s impact (i.e., with a time lag of 3 months or 3 Periods in Year 1) aligned with the expected timeline of the standard clinical practices for patients with overactive bladder in Japan (e.g., the average time to prescribe the first treatment drug being 3-month and the average visit interval being 3-month) [11, 15]. For instance, for the patients who did not visit clinics routinely but was motivated to return to a clinic by the DTCI campaign (implemented in Period 1 in Year 1), their earliest clinic visit timing was Period 1 (weeks 1–5) in Year 1 that allows the earliest timing of the first-time treatment drug prescriptions between late Period 3 and Period 4 (around weeks 14–18) in Year 1 because of the clinical guideline [11]. For the patients who visited clinics every 3 months regularly, the first visits after DTCI exposure should range between Period 1 and Period 3 (around week 13) in Year 1. Then, the second visits, or the first prescriptions after three steps recommended in the guidelines [11], should occur between the latter half of Period 3 (around week 13) and early Period 6 (around week 26) in Year 1, indicating the majority of the first prescriptions should be observed in Period 4 and Period 5 in Year 1. These estimated timelines were consistent with the larger impact of the DTCI campaign during Periods 4 and 5 in Year 1 (in terms of the hazard ratio magnitude) in our main individual-level Cox model analysis, compared to that during Period 3 and Period 6 in Year 1. Moreover, our secondary aggregated-level ITSA also support these delayed effects. Namely, the ITSA showed that the first-time prescription rate was interrupted at Period 4 in Year 1 (i.e., intercept was significantly shifted upward in a regression model [p-value < .05]), and that the time periods representing Periods 4 through 5 in Year 1 were the two highest outlier periods (and that representing Period 6 in Year 1 was the third highest point).

The internal validity of our estimation is discussed below. The magnitude of the increase in the current study (an approximate 140%–600% increase) was much larger than those suggested in previous studies of DTCA campaigns for antidepressant drugs (5–16% increase) or influenza vaccines (5–17% increase) [3, 17]. This gap partly stems from a very low baseline rate for our study, e.g., the crude prescription rate during Period 4 in Year 1 increased from 6.69 to 47.8 [per 100,000 person-days], which is much lower than the commonly seen baseline antidepressant drugs prescription rate (60%) and influenza vaccination rate (50%). Therefore, these very low crude prescription rates and their crude ratios imply a reasonable face validity of our HR or β estimates. Although P5 in Year 1 (during March to April) seems to be an outlier, we did not observe a significant level change at P5 in Year 1 in ITSA as shown in the Results section and Table 3. Therefore, including P5 in Year 1 in the analysis does not influence the conclusion.

In the JMDC data set, 0.3% of enrollees (mean age 53 years) had a diagnosis of overactive bladder before the DTCI campaign. In a large population-based survey in 2003, the rate of physician visits for symptoms of overactive bladder was estimated at about 0.4% when restricted to respondents in their 50s, which concurs with our findings [9]. Therefore, examination of the change in prescription patterns among patients with overactive bladder who are registered in the JMDC database is warranted.

A potential confounder in our analysis is the synergistic effect between pharmaceutical representatives (who could appeal to physicians with the “novelty” of mirabegron in the sales promotion of their company’s drugs) and the release of mirabegron before the DTCI campaign. This synergy may have overestimated the effect of the DTCI campaign on increased prescription rates. Pharmaceutical representatives have played an important role in the increase of drug prescriptions in the US [18]. In the present study’s Cox model analysis, the HR of Period 9 in the Pre-Campaign Year, which corresponds to the release period of mirabegron, was not elevated (Table 2). Likewise, the ITSA did not show a one-time increase in the level of prescription rate or a long-term change by the release of mirabegron (data not shown).

Regarding the external validity of our estimation, it should be noted that the effect of the DTCI campaign is likely to be underestimated in the current study. This is because our study population was younger than most individuals with overactive bladder in Japan [9]. Compared with older or retired patients, our study subjects were current employees (and their family) and hence were more likely to face time constraints when attending medical facilities, owing to their work schedules. Therefore, the magnitudes of the effects of the DTCI campaign among all target patients may be larger than those reported earlier in this study.

Our study had some limitations. First, the absence of the relevant dataset did not enable us to measure other factors at the individual patient level (including the launch of mirabegron and potential pharmaceutical’s in-person marketing toward physicians) that might affect the prescription pattern. However, during the period when the prescription rate was increased, there was no major change that may have affected prescriptions. Second, we could not evaluate the behavior of newly diagnosed patients because we did not have data on healthy people who did not have symptoms of overactive bladder. Third, we could not evaluate the severity of overactive bladder because such information was unavailable in our data set. In addition, the limited data availability did not allow us to analyze the DTCI impacts on other patterns including total volume of prescriptions or adherence to treatment [5–7]. Therefore, further studies are needed to evaluate the impact of the DTCI campaign by incorporating these limitations.

The examined DTCI campaign appeared to change the pattern of prescriptions among patients with overactive bladder, increasing the prescription rate of treatment drugs for 15 weeks and the level of prescription rate for the post-DTCI campaign period. Future studies are expected to examine whether the increased prescription rate leads to improved health outcomes.