AUTH/3185/4/19 - Complainant v Mundibiopharma

Promotion of Invokana

  • Received
    26 April 2019
  • Case number
    AUTH/3185/4/19
  • Applicable Code year
    2016
  • Completed
    27 November 2019
  • No breach Clause(s)
  • Breach Clause(s)
  • Sanctions applied
    Undertaking received
  • Additional sanctions
  • Appeal
    No appeal
  • Review
    To be published in the February 2020 Review

Case Summary


A complainant who described him/herself as a concerned UK health professional, complained about the online promotion of Invokana (canagliflozin) and Vokanamet (canagliflozin and metformin) by Mundibiopharma. Invokana and Vokanamet were used as an adjunct to diet and exercise in certain adults with type 2 diabetes.

The complainant had received an email titled ‘ISN-WCN [International Society of Nephrology – World Conference of Nephrology] Invokana (R) (canagliflozin) virtual booth: Get the latest data on SGLT2 [sodium-glucose transport protein 2]’. The ISN-WCN 2019 had taken place in Melbourne, Australia, 12-15 April.

The complainant alleged that the email linked to the summary of product characteristics (SPC) rather than prescribing information. The ‘microsite’ accessed from the email mentioned Vokanamet but there was no generic name or prescribing information.

Within the microsite the complainant referred to a presentation about renal outcomes in the CANVAS (canagliflozin cardiovascular assessment study) programme and alleged that the outcomes on a slide looked good because the y-axis was suppressed. The complainant added that prescribing information on the material was extremely difficult to read.

With regard to another presentation on the CANVAS programme the complainant alleged that three slides made the outcomes look very good because the y-axis was not correctly numbered which exaggerated the effect. Similarly, the y-axis was suppressed on another slide.

The complainant alleged that a claim on a slide that canagliflozin reduced the risk of cardiovascular death was misleading as the confidence interval crossed 1 and the information below the graph stated that this was a non-statistical, potentially random outcome. A similar allegation was made in relation to another slide which claimed that canagliflozin reduced the risk of mortality in both the intention to treat and left-truncated analyses. This was presented in the same way as information where the confidence intervals did not cross 1. The presentation included prescribing information but the complainant alleged that it was extremely difficult to read.

The complainant stated that he/she did not have time to look through all 114 slides, but he/she thought that the above illustrated that the material had not received the rigour required before being widely sent to health professionals.

The complainant alleged that links to two documents (Perkovic et al (2018) and Neal et al (2017)) should have been certified for use and should have included prescribing information.

The complainant alleged that high standards had not been maintained and, for the sake of completeness, alleged a breach of Clause 2.

The detailed response from Mundibiopharma is given below.

The Panel considered that the link to the Invokana SPC in the email did not fulfil all the requirements of the Code and the Panel therefore ruled a breach. The Panel also considered that the omission of the non-proprietary name for Vokanamet immediately adjacent to the brand name at its first appearance meant that Mundibiopharma had failed to maintain high standards and a breach was ruled.

In the Panel’s view, it would not be sufficiently clear to a user of the virtual booth that he/she would have to download a slide-deck in order to view the Vokanamet prescribing information. However, the Panel noted the complainant’s allegation that there was no prescribing information for Vokanamet on the microsite and that was not so. Based on the very narrow allegation, the Panel ruled no breach in this regard.

The Panel noted the complainant’s allegation that slide 33 of a deck titled ‘Time to protect the diabetic kidney: renal outcomes in the CANVAS program’ had ‘made the outcomes look good by suppressing the y axis’.

The Panel noted that the slide in question was headed ‘Canagliflozin reduced the decline in eGFR compared with placebo’ below which was the graph in question. The y-axis of the graph, adjusted mean eGFR, had a suppressed zero and went from 68 to 80 mL/min/1.73m2. To the right of the graph was a prominent red box which stated that the canagliflozin vs placebo eGFR slope difference was 1.2 mL/min/1.73m2 per year 95% CI [confidence interval] 1.0-1.4. The Panel noted Mundibiopharma’s submission that the graph was a faithful copy from Perkovic et al 2018 and that Mundibiopharma had highlighted the suppressed zero on the y-axis which was not done in the publication. Irrespective of whether the graph was a faithful copy of Perkovic et al its use in promotional material had to comply with the Code. The Panel noted that the graph in Perkovic et al had a y-axis which went from 68 to 80 mL/min/1.73m2 and that all plotted data in the two treatment arms were within this range. The Panel considered that the scale used in Perkovic et al and in the slide in question might aid the reader to analyse each data point on the graph. The Panel noted its comments above, including that the slope difference for canagliflozin vs placebo was prominently stated next to the graph within the slide in question and that the y-axis was clearly labelled, and considered that the complainant had not established that health professionals would be misled with regard to the difference between the treatment arms by virtue of a suppressed zero. No breach of the Code was ruled.

The Panel considered that the prescribing information for Invokana and Vokanamet on the slide-deck in question was legible and no breach was ruled.

The Panel noted that slides 14 to 16 featured 6 graphs which showed differences between canagliflozin and placebo in relation to various measurements such as HbA1c, body weight, blood pressure, cholesterol etc. The graph on Slide 30 showed differences between canagliflozin and placebo in change in urine albumin-to-creatinine ratio (UACR). The Panel noted that the use of a suppressed zero was not limited to those graphs where the difference in treatment favoured canagliflozin; for example, the LDL cholesterol graph on slide 16 favoured placebo and the y-axis was from 82 to 100 mg/dl. The Panel considered that the use of a suppressed zero was not necessarily unacceptable. In each of the 7 graphs in question, the y-axis was clearly labelled, all data points were within the y-axis range and the mean difference between treatment arms and the 95% confidence interval was stated clearly within each slide. In the Panel’s view, the scales used in the slides in question might aid the reader to analyse each data point on the graph. The Panel did not consider that the complainant had established, on the balance of probabilities, that the use of a suppressed zero in the graphs in question would mislead a health professional with regard to the difference between the treatment groups. The Panel therefore ruled no breach of the Code.

In relation to slide 22 the Panel noted that the primary outcome measure in the CANVAS program was a composite of death from cardiovascular causes, non-fatal myocardial infarction, or non-fatal stroke. Secondary outcomes planned for sequential conditional hypothesis testing included death from any cause and death from cardiovascular causes. If sequential testing was not significant for all the outcomes specified, the remaining outcomes were scheduled for assessment as exploratory variables in the integrated data set. The Panel noted that the study authors stated that significantly fewer participants in the canagliflozin group than in the placebo group had a primary outcome event (the composite of death from cardiovascular causes, non-fatal myocardial infarction, or non-fatal stroke): HR 0.86; 95% CI, 0.75 to 0.97; p<0.001 for non-inferiority, p=0.02 for superiority. The study authors stated that superiority was not shown for the first secondary outcome in the testing sequence (death from any cause; p=0.24) and hypothesis testing was discontinued. Therefore, estimates for the fatal secondary outcomes, including death from any cause (HR 0.87, 95% CI, 0.74 to 1.01) and death from cardiovascular causes (HR 0.87, 95% CI, 0.72 to 1.06) were not considered to be significant.

The Panel noted the headline claim on slide 22 stated ‘Canagliflozin reduced risk of CV death’. The hazard ratio (0.87) and confidence interval (0.72 to 1.06) were stated. The footnote to the graph, in small font, stated, ‘Intention-to-treat analysis, exploratory outcome, no p-value is reported due to hierarchical testing strategy, only nominal effect estimate is given’. The Panel noted Mundibiopharma’s submission that this slide indicated that there was a non-significant reduction in cardiovascular death as the confidence interval was shown crossing unity. The Panel further noted that the authors of the paper stated, ‘All three components of the primary outcome – death from cardiovascular causes, nonfatal myocardial infarction, and non-fatal stroke – showed point estimates of effect that suggested benefit, although the individual effects did not reach significance’.

The Panel noted the hazard ratio and Mundibiopharma’s submission that the point estimate was to the left of 1, favouring canagliflozin as illustrated in figure 2 of the Invokana SPC. The Panel considered that presenting data which did not reach statistical significance was not necessarily unacceptable, however, the presentation of such data, including claims, must not be misleading in this regard.

The Panel considered the immediate and overall impression to a health professional. In the Panel’s view, the headline claim ‘Canagliflozin reduced risk of CV death’, which was in prominent bold red font above the graph, implied that the difference between canagliflozin and placebo reached statistical significance which was not so. The 95% confidence interval stated within the slide and the footnote in small font below the graph did not negate the immediate misleading impression given by the headline claim. The Panel noted its comments above and considered that the claim was misleading and was a misleading comparison of canagliflozin compared with placebo in this regard. The Panel ruled breaches of the Code.

In the Panel’s view, the headline claim to slide 23 ‘Canagliflozin reduced risk of mortality in both the intention-to-treat and left-truncated analyses’ implied that the difference between canagliflozin and placebo reached statistical significance which was not so, and the inclusion of the 95% confidence intervals and p-value did not negate the immediate misleading impression given by the headline claim. The Panel therefore considered that the claim was misleading and was a misleading comparison of canagliflozin compared with placebo in this regard. The Panel ruled breaches of the Code.

The Panel considered that the prescribing information for Invokana and Vokanamet on the slide-deck in question was legible and no breach of the Code was ruled.

With regard to the allegation that there were two links to items that did not appear to be part of the microsite and that both items required prescribing information and certification, the Panel noted Mundibiopharma’s submission that the links in question were accessed by clicking once on pillars 2 and 4 within the microsite which provided a copy of Perkovic et al and Neal et al, respectively. The Panel noted Mundibiopharma’s submission that pillars 1 and 2 were linked, as were pillars 3 and 4, and that the articles accessed from pillars 2 and 4 were references to support the content within pillars 1 and 3, respectively. The Panel further noted that pillars 1 and 3 were certified and that the signatory had read Perkovic et al and Neal et al to check that the articles complied with the Code and were not inconsistent with the SPC.

The Panel disagreed with Mundibiopharma’s submission that certification of the articles was not required as ‘each possible combination does not need to be certified’ in the context of digital material as per the supplementary information to Clause 14.1. The Panel noted that the supplementary information stated, inter alia, that as the final form of digital material might not be static, consideration needs to be given to the context in which it appears, but each possible combination does not need to be certified. In the Panel’s view, all promotional material required certification and the supplementary information cited by Mundibiopharma pertained to navigation within digital material; it did not preclude any of the content from being certified. Pillars 2 and 4 were sections of the virtual booth designed to provide health professionals with access to Perkovic et al and Neal et al, respectively, and therefore the material should have been certified for such use. In the Panel’s view, this was no different to a physical exhibition stand providing a reprint. The Panel ruled a breach of the Code.

In the Panel’s view, the exemption for cost in the limited scenarios described in the Code did not apply to the microsite in question. The Panel considered that the link to the Invokana SPC did not fulfil all the requirements for prescribing information and therefore ruled a breach of the Code.

The Panel noted its rulings of breaches of the Code and considered that Mundibiopharma had failed to maintain high standards and ruled a breach of the Code.

In the particular circumstances of this case and noting that there was a link to the Invokana SPC on the microsite and that Invokana and Vokanamet prescribing information was available within two slide-decks on the microsite, the Panel ruled no breach of Clause 2.