Communication of anticancer drug benefits and related uncertainties to patients and clinicians: document analysis of regulated information on prescription drugs in Europe

Objective To evaluate the frequency with which relevant and accurate information about the benefits and related uncertainties of anticancer drugs are communicated to patients and clinicians in regulated information sources in Europe.

Design Document content analysis.

Setting European Medicines Agency.

Participants Anticancer drugs granted a first marketing authorisation by the European Medicines Agency, 2017-19.

Main outcome measures Whether written information on a product addressed patients’ commonly asked questions about: who and what the drug is used for; how the drug was studied; types of drug benefit expected; and the extent of weak, uncertain, or missing evidence for drug benefits. Information on drug benefits in written sources for clinicians (summaries of product characteristics), patients (patient information leaflets), and the public (public summaries) was compared with information reported in regulatory assessment documents (European public assessment reports).

Results 29 anticancer drugs that received a first marketing authorisation for 32 separate cancer indications in 2017-19 were included. General information about the drug (including information on approved indications and how the drug works) was frequently reported across regulated information sources aimed at both clinicians and patients. Nearly all summaries of product characteristics communicated full information to clinicians about the number and design of the main studies, the control arm (if any), study sample size, and primary measures of drug benefit. None of the patient information leaflets communicated information to patients about how drugs were studied. 31 (97%) summaries of product characteristics and 25 (78%) public summaries contained information about drug benefits that was accurate and consistent with information in regulatory assessment documents. The presence or absence of evidence that a drug extended survival was reported in 23 (72%) summaries of product characteristics and four (13%) public summaries. None of the patient information leaflets communicated information about the drug benefits that patients might expect based on study findings. Scientific concerns about the reliability of evidence on drug benefits, which were raised by European regulatory assessors for almost all drugs in the study sample, were rarely communicated to clinicians, patients, or the public.

Conclusions The findings of this study highlight the need to improve the communication of the benefits and related uncertainties of anticancer drugs in regulated information sources in Europe to support evidence informed decision making by patients and their clinicians.

Conceptual and analytical approach

Development of the coding schedule

To operationalise these broad categories of information and to develop an evidence informed framework (see fig 1) for data extraction and analysis, we developed predefined coding categories for data extraction that corresponded to patients’ commonly asked questions about drug benefits and related uncertainties. See the supplementary methods for an outline of our coding strategy.

Fig 1

Analytical framework for evaluating the quality of written information on benefits and related uncertainties of prescription drugs

Our coding scheme for evaluating the strength and applicability of the available evidence on drug benefits was informed by empirically documented threats to internal and external validity of clinical study findings. These related to specific aspects of the design, conduct, analysis, and reporting of studies that have been shown to undermine the reliability or interpretation of results for benefit,61626364 or that create uncertainty about the relevance of those results for clinical practice.65666768

In addition, we included a separate category—the need for additional or longer term data—to capture cases where EMA assessors explicitly noted the need for additional data to address evidence gaps and uncertainties related to a drug’s efficacy. For example, the need for comparative data for drugs approved on the basis of single arm trials, or the need for larger sample sizes or longer follow-up to confirm the magnitude or duration of effect when results were based on a small number of patients or interim analyses.

Analysis

Identification of the study sample and initial analysis of the EPARs and summaries of product characteristics were undertaken by one researcher (JL) and checked by three researchers (CD, HN, and AP). CD and JL extracted data and analysed the public summaries and patient information leaflets, verified by HN. Final analysis of the data was undertaken by CD and checked by HN. Disagreements over the coding of specific items were resolved in the first instance by discussion among JL, AP, CD, HN, and MS-K, then by discussion among all collaborators. Consensus was reached in all cases.

We searched the publicly available EMA database to identify all anticancer drugs with a first marketing authorisation in the EU between 1 January 2017 and 31 December 2019. Consistent with the approach used in a previous study,34 we used the Anatomical Therapeutic Chemical Classification codes L01-L04 to retrieve the list of antineoplastic and immunomodulating agents approved within the study period. We cross checked this list against the haematology and haemostaseology section and the cancer section in the Human Medicines Highlights reports, which EMA publishes annually to highlight new drugs recommended for approval. We excluded generic, biosimilar, and hybrid products, and treatments for supportive care.

Textual and numerical data pertinent to the analytical categories in our coding scheme were extracted from all documents and entered into a data analysis workbook. Data were analysed at the indication level.

Our initial coding scheme included the category for goal of treatment—a category to record whether the drug was intended to cure, prevent, or alleviate disease. As the goal of treatment was rarely explicitly stated in EPARs, we had to infer whether the treatment was intended to be curative or non-curative (palliative). These inferences (which could be based on the inclusion criteria for the pivotal studies, the approved indication and target population, or general statements in EPARs about the condition and stage of disease for which the treatment was intended) were sometimes difficult to make. Unclear inferences were flagged for discussion and resolved in consultation with an oncologist (AP).

EPARs were used to identify the main clinical studies supporting EMA decisions. We extracted data on drug benefits from EPARs according to the primary study outcomes, as well as information on survival or quality of life benefits when available. Our determination that a statistically significant survival or quality of life benefit had been shown was based on reported hazard ratios, confidence intervals, and P values.

We determined whether there was uncertainty about evidence supporting drug efficacy if EPARs contained an explicit or implicit acknowledgement by EMA assessors of unresolved shortcomings, concerns, or questions corresponding to the predefined list of uncertainty sources as set out in our analytical framework. This included instances where concerns were expressed by scientific advisors consulted during the assessment process. We concluded that EMA assessors had expressed concerns about the therapeutic value of a drug if they questioned the accuracy of estimates of drug efficacy or whether demonstrated benefits were clinically relevant for patients.

For each drug indication in our sample, we systematically compared data extracted from summaries of product characteristics, public summaries, and patient information leaflets with data extracted from EPARs. We judged the information in summaries of product characteristics, patient information leaflets, and public summaries to be accurate if it consistently and comprehensively reflected EMA’s scientific assessment of the clinical evidence. We judged written product information to be inaccurate if it appeared to omit, or only partially report, pertinent information in a way that might lead patients and clinicians to overestimate or underestimate the benefits of a drug or to have unwarranted confidence in the evidence.

We calculated the proportion of summaries of product characteristics, patient information leaflets, and public summaries that communicated relevant and accurate information on drug benefits and uncertainties about benefit.

Patient and public involvement

Input from individuals with lived experience of cancer treatment has been pivotal in setting out the design and analytical strategy of this study. Our analytical strategy was informed by a literature review of empirical studies of patients’ treatment information needs and commonly asked questions. One member of the author team (HS) has substantial experience in patient advocacy and public involvement roles. We also discussed the relevance and usefulness of our taxonomy of key benefit information with a patient leader from Just Treatment, a UK based patient advocacy organisation.

Anticancer drug approvals and sample characteristics

The study sample included 29 anticancer drugs that had received a first marketing authorisation between 2017 and 2019 (fig 2). Three were approved for more than one indication, resulting in 29 drugs approved for 32 cancer indications and supported by 37 main studies.

Fig 2

Identification and selection of anticancer drugs receiving a first marketing authorisation during 2017-19. ATC=Anatomical Therapeutic Chemical Classification

Of the 32 approved indications, 22 (69%) were for the treatment of solid tumours, one was for solid tumours that were histology agnostic, 10 (31%) were for the treatment of haematological malignancies, four (13%) were for early stage disease, and 28 (88%) were for the treatment of advanced or metastatic disease (see supplementary table 1). Based on available information in EPARs, we determined there were four (12%) indications where the goal of treatment appeared to be curative. For the other 28 (88%) indications the goal of treatment was non-curative.

Supplementary table 1 provides further information on the regulatory and treatment characteristics of the sample. Supplementary table 2 provides information on the characteristics of the main studies supporting approval.

Drug benefits and related uncertainties according to EPARs

Availability of relevant and accurate information in documents

Figure 3 shows the proportion of summaries of product characteristics, public summaries, and patient information leaflets that reported accurate and complete information about a drug and approved indication, how the drug was studied, and the benefits shown, compared with information reported in EPARs. Compared with other sources of written prescription drug information, patient information leaflets had the lowest rate of reporting across all categories of information. These leaflets did not contain any information on how the drug was studied and its benefits.

Fig 3

Communication of information about a drug, how it was studied, and evidence of benefit. EPAR=European public assessment report

Communicating concerns and uncertainties about drug benefits

Figure 4 shows the percentage of EPARs in our sample where a concern was raised about each of the uncertainty domains included in our analytical framework, and the proportion of summaries of product characteristics, public summaries, and patient information leaflets that reported those concerns. Regulators’ concerns about the reliability or clinical relevance of evidence on drug benefits were rarely communicated to clinicians or patients—with the partial exception of concerns around the non-generalisability of findings—even when EMA’s scientific assessors were unable to reach a consensus on whether the drug should be approved. For example, EMA assessors could not agree on the approval of atezolizumab for the treatment of locally advanced or metastatic urothelial carcinoma owing to “substantial uncertainties regarding the efficacy of [the drug]” (see supplementary table 5). Approval of first line use of atezolizumab in this indication was based on one single arm trial in which response rates were acknowledged to be inferior when indirectly compared with an historical control. Approval in the second line setting was based on a randomised controlled trial in which atezolizumab failed to meet the primary endpoint of improved overall survival compared with chemotherapy. Regulatory concerns over the clinical efficacy of atezolizumab were not communicated in the written information for clinicians or patients. Instead, the public information leaflet stated: “Tecentriq helps your immune system to fight your cancer.”

Fig 4

Communication of EMA assessors’ concerns about study methods and findings. EMA=European Medicines Agency; EPAR=European public assessment report

Supplementary table 5 provides further illustrative examples of drugs where EMA assessors and scientific advisors raised multiple concerns in EPARs, and compares comments with the information communicated about drug benefits (or in the case of patient information leaflets, drug mechanisms of action) and related uncertainties in the corresponding public summaries and patient information leaflets.

Policy implications

Our study highlights the need to improve the communication of drug benefits and related uncertainties in regulated prescription drug information in Europe. The provision of high quality information on drug benefits and related uncertainties is especially important for patients with time limiting conditions such as advanced cancer. Regulatory approval of an apparently effective new treatment is an important source of hope for patients with cancer. A major body of research shows that many patients with cancer, particularly those with advanced disease, misunderstand the purpose of drug treatment and overestimate the benefits.383940417071727374757677 When patients have unrealistic expectations of the benefits from treatment and misplaced confidence in the strength of study findings underlying approvals, informed decision making is undermined.

Despite the commitment of medicines regulators to shared decision making and person centred care,78 current regulated sources of prescription information in Europe do not allow patients to distinguish between new anticancer drugs that offer clinically meaningful benefits compared with those with considerable uncertainty about effects. In fact, by failing to communicate evidence gaps and uncertainties, current product information risks exacerbating and perpetuating common misconceptions about new drugs. Recent research suggests that in the absence of explicit information about the strength of the evidence underpinning recommended treatments and interventions, people assume the evidence is of high quality.79 And without a clear statement in the product information for new anticancer drugs that surrogate endpoints such as progression-free survival and tumour response do not reliably predict either patient survival or improved quality of life,80818283 it is unclear how patients will interpret standard descriptions of these endpoints in EMA’s public summaries. The small body of research investigating this topic suggests the terminology typically used in public summaries (“increasing the time patients live without their disease getting worse,” “the cancer shrank or was eliminated,” “the disease responded to treatment”) will lead to confusion about the medicine’s purpose and the types of benefits patients can expect from a drug.3137428485

Patients may have access to regulated information about prescription drugs only after a treatment decision has already been made. Even in such cases, better information on medicines can address patients’ evolving information needs during treatment, and help inform decisions about if and when to discontinue treatment.19 Patients may also seek information not only on their prescribed medicines but also on other available treatment options. Better sources of regulated prescription drug information can have spill-over effects for broader information about medicines; many stakeholders in the health system (including patient organisations, charities, and health technology assessment bodies) rely on regulated information sources when communicating information about new drugs.

We recognise that regulators cannot compel companies to include information on drug benefits and evidence about uncertainties of benefits in patient package leaflets, because current UK and EU legislation does not mandate this information. However, mechanism of action statements may be misunderstood as statements of effectiveness, especially in the absence of qualifiers, reporting on uncertainty, or quantitative benefits information. This may be at odds with legal requirements for information in patient leaflets to be accurate, non-misleading, and non-promotional. In addition, regulators could concentrate efforts on ensuring that relevant, accurate, and useful content are available in online lay summaries for the public. Although public summaries are produced in line with regulated templates, these templates could be amended without the need for legislative change. Information on benefit could be presented as part of an anticancer drug specific “drug facts box” and made available online to inform shared decision making in clinical practice.8687 Further research may be needed to determine how best to communicate information on evidence uncertainties and how to strike the right balance between too much and too little information.

Limitations of this study

Our study has limitations. Firstly, our initial taxonomy of key benefit information, although based on an extensive review of the academic and grey literature, may not have captured all information about drug benefits and uncertainties that might be relevant and useful for patients.

Secondly, our coding scheme could not include every category of information that we identified from the literature. For example, we did not include information on the average time before patients could expect to experience benefits, because this information was not always available in EPARs.

Thirdly, we have not captured every aspect of a trial’s design, conduct, or analysis that could affect the strength, reliability, or relevance of the evidence. For example, we did not examine the extent to which the population studied in the main trials was the same as the population for which the drug was ultimately approved (although discrepancies may sometimes have been captured in regulatory concerns about generalisability or concerns about the therapeutic value of the drug in certain patient populations).

Fourthly, in relying on EMA assessments for our determination of evidence uncertainties rather than some objective measure or instrument (such as the Cochrane risk of bias tool), we may have underestimated the extent to which problematic study methods undermined results. For example, EMA assessors flagged concerns about the effects of missing outcome data and the unblinded measurement of study outcomes on the reliability of findings in a smaller proportion of EPARs relative to the frequency with which EMA assessors expressed concerns in relation to the other uncertainty domains. In contrast, a previous study of EMA anticancer drug approvals using the Cochrane risk of bias tool found that incomplete outcome data and measurement of the outcome were the primary domains responsible for high risk of bias judgments.34

Finally, our analysis includes only new anticancer drugs approved between 2017 and 2019, and it is not clear whether our findings can be generalised to other disease areas. However, anticancer drugs now comprise the single largest category of new drug approvals in Europe.

Conclusion

Regulated information sources for anticancer drugs in Europe fail to address the information needs of patients. This study identified important shortcomings in the communication of information on drug benefits and related uncertainties in regulated sources. If patients lack access to such information, clinical decisions may not align with their preferences and needs.