Defining a standard set of patient-centred outcomes for lung cancer

Survival

Duration of survival is crucial for patients with lung cancer, a disease with high mortality rates due to both tumour burden and morbid therapies [9]. Our working group reached consensus that overall survival, cause of death and treatment-related mortality were essential measures of survival for the Lung Cancer Standard Set (table 2 and online supplementary material, appendix 4). Although progression-free survival is an intermediate end-point commonly collected in the clinical trial setting as a measure of disease control, this outcome was excluded from the set as it was deemed potentially unreliable in routine practice due to ascertainment bias [10], and ultimately less important than overall survival. Ideally, survival outcomes would be routinely collected throughout the course of care, but annual querying of national death indices where available is recommended to validate local statistics (figure 1).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Sample timelines illustrating when case-mix variables and key outcomes should be collected for patients treated with different modalities, including a) surgery or b) multiple treatments with definitive chemoradiation followed by chemotherapy after disease progression. These timelines are intended to represent the outcome data collection points for possible treatment paths and not to advocate any particular treatment approach. PROMs: patient-reported outcome measures.

Complications

Treatment complications are a significant cause of morbidity and mortality in lung cancer, and can profoundly impact patient preferences, outcomes and costs [11]. Our working group reviewed treatment-related toxicity scales in common practice today. For patients treated with surgery, the consensus was to use the Clavien–Dindo classification [12]. We selected a simplified version of the Common Terminology Criteria for Adverse Events version 4 for patients receiving systemic therapy and/or radiotherapy [13]. This platform was designed as a comprehensive grading system for identifying adverse effects of cancer treatment and indicating their severity, and is commonly used in clinical trials. We propose a modified version to simplify and expedite data collection, which assigns the event to a general category (constitutional, skin, bone marrow suppression, infection, cardiovascular, lung, gastrointestinal, hepatic, renal, neurological or other). We recommend collecting all grade 3 or higher toxicities that occur during or within the first 6 months following the initiation of each treatment course for lung cancer (figure 1).

Degree of health

Performance status is a strong independent predictor of survival in lung cancer [14]. Our working group recommends measuring performance status per the Eastern Cooperative Oncology Group (ECOG) and World Health Organization (WHO) scoring method, ranging from 0 to 5, as it provides clinician-reported data to correlate with patient-reported outcomes. Baseline, pretreatment performance status will also be captured as a case-mix variable, allowing detection of change in performance status over time.

Given the importance of health-related quality of life (HRQoL) in patients with lung cancer [15] and its prognostic impact [16], PROMs have been increasingly developed and implemented [17]. Lung cancer is associated with burdensome symptoms, and often requires treatments associated with significant toxicity. General HRQoL domains that we advocate collecting include global health status/quality of life, pain, fatigue and physical, social, emotional and cognitive function (table 2 and online supplementary material, appendix 4). Lung cancer-specific HRQoL domains deemed essential included dyspnoea and cough.

Common existing PROMs that cover these domains were reviewed by the group (online supplementary material, appendix 5). PROM selection was thoroughly deliberated, as was our goal to provide clear advice to newly developing initiatives, with the hope of ultimately aligning existing measurement efforts. Our consensus recommendation included one general and one lung cancer-specific instrument to adequately cover the prioritised domains. The heterogeneous nature of lung cancer presentations challenged us to pick from a variety of well-validated PROMs and outcome definitions, many of which captured the most important domains. Our working group ultimately selected the European Organisation for the Research and Treatment of Cancer (EORTC) core quality of life questionnaire (QLQ-C30) [18] and its corresponding lung cancer-specific module (QLQ-LC13) [19], since these best covered the domains we considered most important. These validated instruments are internationally recognised, widely used in a variety of languages, feasible to implement and well studied, with scores that can be clinically interpreted [20, 21]. While the EORTC has a separate questionnaire assessing the HRQoL of patients in the palliative care setting (QLQ-C15-PAL) [22], the same domains are captured in the EORTC QLQ-C30 and therefore it was felt to be sufficient.

For institutions with compelling reasons to use other validated PROMs, our recommendation is to collect data on the equivalent domains at the same time points. We eagerly anticipate the development of algorithms enabling conversion of commonly used lung cancer PROMs to a standard scoring system, as has been achieved in other diseases [23].

We recommend assessing patient-reported outcomes prior to treatment as a baseline (or at diagnosis if no treatment was pursued), at 3, 6 and 12 months following initiation of treatment, and then annually (figure 1). We recognise that the burden of answering 43 items from EORTC QLQ-C30 and EORTC QLQ-LC13 at each interval may be significant for some patients. However, both the clinical and patient representatives in the working group felt that all these domains were essential and if used in the process of clinical care, would be accepted by patients. This position is bolstered by research suggesting that respondent satisfaction is more directly linked to the salience of questions than their length [24]. We also anticipate that in time computer-adaptive PROMs will decrease respondent burden while providing similar domain coverage [25].

The working group recognised that patient wellbeing can change rapidly and profoundly in this population with frequently advanced cancer. Thus, where feasible, we encourage more routine measurement of PROMs in clinical practice as regularly as possible to complement the defined time points.

Other outcomes

Previous studies have demonstrated that treatment delays are prognostic in lung cancer [26, 27]. We therefore recommend collecting the time from diagnosis of lung cancer to initiation of treatment. For consistency, we recommend defining the date of diagnosis in accordance with the hierarchy established by the European Network of Cancer Registries, with date of hospital admission or outpatient consultation being used in cases where no histological or cytological confirmation of malignancy was obtained [28].

Quality measures at the end of life

Quality of end-of-life care is an important consideration in lung cancer. Outcomes such as frequent hospitalisations and intensive care unit (ICU) admissions near the end of life have been assessed, and suggest improper use of aggressive intervention [29]. It is less clear whether any treatment near the end of life should be considered a marker of poor quality, as some procedures or therapies may be effective in palliating distressful symptoms. Appropriate palliative care has been shown to be associated with improved quality of life and a reduced likelihood of aggressive care [30]. We reviewed existing measures used to assess the quality of the death and dying process. Although we desired a patient- or caregiver-reported measure, existing options were not found to be suitable for widespread use [31]. As a proxy, our group recommends capturing place of death, i.e. in a hospital, care home, hospice or the patient's own residence as well as the duration of time spent in hospital, including in the ICU, in the last 30 days of life.

Case-mix variables

Baseline demographic, clinical and tumour factors are associated with survival and other outcomes in lung cancer. As outlined in table 1, the working group identified the baseline case-mix variables thought to be essential for risk adjustment to enable meaningful comparisons between patients. Given the number of potential variables associated with clinical outcomes following treatment for lung cancer, the group focused on the most established essential demographic, clinical and tumour factors (online supplementary material, appendix 4). Where applicable, we recommend that these case-mix variables be collected prior to treatment initiation.

Specifically, we recommend collecting the following demographic factors: age, sex, ethnicity and education level. While a key determinant of health outcomes in the lung cancer population [32], socioeconomic status (SES) is difficult to capture and quantify for a variety of reasons, including patient reluctance to disclose financial details and the challenge of accurately encapsulating a complex, multifactorial determinant of health within a single metric [33]. We recommend collecting the patient-reported highest level of education as a surrogate measurement of SES [33], as patients generally feel comfortable reporting this information and it can be compared across countries using the International Standard Classification of Education [34].

The working group also recommends that ethnicity data be collected, deferring to local standards for its definition, given the variance in population characteristics globally.

We advocate the collection of the following clinical factors: baseline ECOG/WHO performance status, unintentional weight loss, smoking status (defined as never smoker (<100 cigarettes in lifetime), ex-smoker (stopped ≥1 year before diagnosis) or current smoker) and pulmonary function as measured by the absolute and predicted forced expiratory volume in 1 s, baseline patient-reported health status as measured by the EORTC QLQ-C30 and QLQ-LC13 and comorbidities (table 1). Our working group endorsed the use of the Modified Self-Administered Comorbidity Questionnaire (SCQ) [35] to collect a list of comorbid diseases. The SCQ has been shown to predict functional outcomes as well as the medical record based Charlson Comorbidity Index, and better predict quality of life [36].

Baseline tumour factors to be recorded are clinical and pathological stage, histology, activating epidermal growth factor receptor mutation status, anaplastic lymphoma kinase translocation status and clinical versus pathological (histological or cytological) basis of diagnosis.

For patients undergoing treatment, our working group recommends collecting additional data on treatment intent (curative versus palliative, with the latter including best supportive care) and whether treatment was completed. We recognise that therapies for oligometastatic disease may not clearly constitute curative versus palliative treatment, and recommend that this type of treatment be classified as palliative for consistency in data collection.