The International Review Board of National Cheng Kung University Hospital approved this study (A-ER-100-079) before initiation. Because this study is a retrospective research design, the International Review Board of National Cheng Kung University Hospital waived the requirement for informed consent from subjects. The study method was conducted in accordance with the STROBE guidelines.
Lifetime survival data sources
In this study, we interconnected the databases of the Taiwan Cancer Registry (TCR) and the Taiwan National Death Registry provided by the Ministry of Health and Welfare’s Health and Welfare Data Science Center (HWDC) to establish survival functions for lung cancer patients. After interconnection, all personal identification information was encrypted at HWDC before data analysis to ensure confidentiality. We extracted anonymized demographic and clinical information of non-small cell lung cancer patients, including sex, age, baseline performance according to the Eastern Cooperative Oncology Group (ECOG) classification, date of diagnosis, survival status, histological subtype, and smoking status. To comply with the lung cancer screening policy, we only included NSCLC patients aged 50 to 89 years. All cases were classified according to the American Joint Committee on Cancer (AJCC) system, 7th edition, ref. 15, with pathological staging taking precedence over clinical staging. We also obtained national life tables from the Ministry of the Interior of Taiwan to simulate the survival functions of the reference population for comparison.
ADL and QOL hospital cohort
All NSCLC patients aged 20 years or older (index group) who visited National Cheng Kung University Hospital (NCKUH), a national medical center, were asked to complete a self-report questionnaire including EQ-5D and activities of daily living (ADL) at each clinical visit from May 2011 to December 2020. Patients admitted to the chest ward and other departmental outpatient clinics since 2017 were also included. Scores for each item were extracted along with each patient’s demographic and clinical data (e.g., sex, age, performance status, date of diagnosis, and staging). For this study, only patients with non-small cell lung cancer were selected and stratified by age (50–64 years vs. 65–89 years) and stage (stage I–IIIa vs. stage IIIb–IV).
Measuring Functional Impairment and Health Utility
To measure each patient’s functional disability, the Barthel Index (BI), administered by well-trained standardized interviewers or case managers with nursing background, was applied. The Barthel Index consists of 10 items. Each item is rated in terms of whether the patient is able to perform self-care tasks or whether they need assistance based on observation. Two items (bathing and grooming) are rated on a 2-point scale (0 and 5 points), six items (climbing stairs, eating, using the toilet, dressing, bowel control, and bladder control) are rated on a 3-point scale (0, 5, and 10 points), and two items (transferring from wheelchair to bed and horizontal walking) are rated on a 4-point scale (0, 5, 10, and 15 points). The total score ranges from 0 to 100, with higher total scores indicating greater independence. To compare our results with the proportion of people with functional disability in the general population, this study classified individuals with a BI score of more than 70 as not disabled, based on the criteria established by the National Research Project to Estimate the Demand for Long-Term Care Services issued by the Economic Planning and Development Commission of the Executive Yuan of Taiwan16.
To estimate quality of life utility values, the three-part European Quality of Life Five Dimensions (EQ-5D) questionnaire was applied, and each patient completed the form with partial assistance from the interviewer if necessary. The EQ-5D includes five dimensions: mobility, self-care, daily activities, pain or discomfort, and anxiety or depression.17 By using the scoring function from a Taiwanese study18, the results were converted into utility values ranging from 0 to 1, with 0 representing the worst state and 1 the best health state.
Statistical analysis
Briefly, we estimated the lifetime survival function to calculate life expectancy (LE), which was then multiplied by QOL utility to calculate QALE. We then compared them with age-, sex-, and calendar year-matched reference subjects simulated from vital statistics to obtain LE loss and QALE loss. Since the Taiwanese government considers ADL scores below 70 to be functionally disabled, we adopted the same criteria in this study and estimated disability-free LE (DFLE) by multiplying the proportion of people without disabilities by survival rate and adding it to lifetime, and DFLE loss was calculated by comparing with the corresponding reference subjects. The details are as follows:
Estimation of life expectancy (LE) and LE loss
We verified the survival status of all NSCLC patients by cross-linking the Taiwan Cancer Registry and the Taiwan National Death Registry from 2011 to 2018. The Kaplan-Meier method was first employed to estimate the survival function (denoted as \(S(t|index)\) of the cohort of NSCLC population from 2011 to 2018 to the follow-up limit. Then, a semi-parametric extrapolation method19 was applied to predict the lifetime survival function of the population. The overall extrapolation process mainly involved three steps. First, we established an age (at diagnosis), sex, and calendar year-matched reference population (denoted as \(S(t|ref)\) corresponding to the index cohort simulated from vital statistics and also estimated by the Kaplan-Meier method. Secondly, we constructed the survival ratio between the index cohort and the reference population, denoted as \(W\left(t\right)=S\left(t|index\right)/S\left(t|ref\right)\). Finally, the survival ratio \(W
(1)
$$\mathrm{DFLE}={\int }_{0}^{\infty }\widehat{\mathrm{S}}\left(\mathrm{t}\right)\times \widehat{{P}_{health}}\left(t\right)dt.$$
(2)
We then calculated the QALE loss and DFLE loss for each subgroup by subtracting them from the QALE and DFLE, respectively, of each age-, sex-, and calendar year-matched index subhort. We also calculated expected years lived with disability (EYLD) by subtracting DFLE from LE, which has also been proposed and validated as an estimate of long-term care need in previous studies.
Ethical issues
The study protocol was approved by the International Review Board of National Cheng Kung University Hospital (A-ER-100-079). Because the study design was retrospective, the requirement for informed consent from study subjects was waived by the International Review Board of National Cheng Kung University Hospital. The study methods were conducted in accordance with the STROBE guidelines.