Catastrophic expenditure rates and barriers for treatment adherence in patients with colorectal cancer in India: The CROCODILE study protocol

Little is known about the delivery of colorectal cancer treatment in India and its associated costs. The aim of this study is to identify financial and nonfinancial barriers to adherence to colorectal cancer treatment in India.


INTRODUC TI ON
Colorectal cancer is the sixth most common cancer in India, with a recent increase among young men under 50, who are typically active and have a long life expectancy [1][2][3]. To support this burden, access to effective colorectal cancer care is essential to improve individual and societal health. The ratio of oncologists per cancer patient in India is 1:2000, compared with a ratio of 1:100 in Western countries [4]. There is also a major shortage of radiotherapy services, with only 545 machines existing in the country in 2019 [5]. This is not enough to meet even half of patients' needs, according to World Health Organization (WHO) guidance. Nearly 80% of cancer surgery in India is provided by general surgeons in district hospitals [6]. Although the National Cancer Control Programme in India has streamlined the delivery of oncological treatment, inefficiencies in referral systems and centralization of treatment for complex cases have resulted in fragmentation of care [6,7]. Chronic underfunding of the public sector in India has led to the general perception that the care provided in public facilities is of low quality and this has allowed the emergence of an expensive and underregulated private sector [6].
In combination, these factors have created heterogeneity in the availability and accessibility of cancer care. Little is known about the factors influencing patients' options and decisions regarding colorectal cancer treatment once a diagnosis is made. Improving cancer care and its outcomes has been prioritized in the WHO sustainable development goals (goal 3.4) and by international research networks [8,9].
The aim of this study is to identify financial and nonfinancial barriers for adherence to colorectal cancer treatment in India, using prospectively collected frontline patient-level data. Data on patients' expenditure and income is typically hard to collect, as it is time-consuming and susceptible to recall and social desirability bias.
To overcome these common limitations, both now and in the future, this study will also assess the feasibility of new data collection methods for assessing patient income.

Study setting and design
This study will be run in five hospitals in India where cancer treatment is available and routinely delivered. This will be a prospective, multicentre, mixed-methods study with two main workstreams: 1. quantitative workstream: a prospective cohort feasibility study to determine treatment adherence and catastrophic expenditure rates; 2. qualitative workstream: to explore patient and professional views and experiences of barriers to and facilitators for adherence to colorectal cancer treatment.
Our research involves identification of barriers for treatment adherence and quantification of their impact on patients' ability to receive and pay for treatment. A mixed-methods design is therefore appropriate for CROCODILE.

QUANTITATIVE WORK S TRE AM
This protocol follows the STROBE guidelines for cohort studies [10].

Objectives
The primary objective of this workstream is to assess treatment adherence in colorectal cancer patients in India at 6 months after treatment decision.
The secondary objectives are: 1. to determine the treatment adherence rate at 6 weeks and 3 months after treatment decision; 2. to assess the catastrophic expenditure rate at 6 weeks, 3 and 6 months after treatment decision; 3. to identify the main drivers of treatment costs in colorectal cancer treatment; 4. to assess the feasibility of data collection on treatment costs and patient income; 5. to compare patient-reported income with income assessed through the International Wealth Index and the Gapminder Foundation household pictures pool [11].

Inclusion criteria
This study will include consecutive newly diagnosed adult patients (18 years of age or older) with a histopathologically proven malignant colorectal cancer.

Exclusion criteria
Patients will be excluded if histopathology is unavailable or reveals a premalignant colorectal tumour (e.g. an adenoma). Patients for whom a treatment plan is made at the recruiting centre but who undergo the full course of treatment in another hospital will be excluded, as well as patients for whom telephone follow-up is not possible.

Patient identification and consent
Patients with colorectal cancer for inclusion in the study will be identified from multidisciplinary team meetings. In centres where a multidisciplinary team meeting is not in place or where not all the patient cases are routinely discussed in those, both medical oncology, surgical oncology and radiation oncology outpatient lists will be reviewed to identify consecutive newly diagnosed patients.
All patients will provide written informed consent to enter this study. After identification, patients will be invited to participate in the study and will be provided with a patient information sheet (Appendix 2). This has been translated into six languages for use in the different Indian states (Hindi, Punjabi, Bengali, Malayalam, Tamil and Telugu).

Outcomes
The primary outcome of CROCODILE will be the treatment adherence rate at 6 months after treatment decision.
The secondary outcomes will be: 1. treatment adherence rate at 6 weeks and 3 months; 2. catastrophic expenditure rate at 6 weeks, 3 and 6 months after treatment decision; 3. the five most expensive items comprising colorectal cancer treatment cost; 4. data completeness rates for treatment costs and patient income assessment data points; 5. correlation between reported patient income and Gapminder predicted income.
A schedule of outcome assessment timepoints is shown in Table 1.

Study definitions
Treatment adherence will be defined as patients' ability to follow the treatment plan which was decided for them. Patients will be classified as compliant if they have completed or are receiving the decided treatment plan at the time of follow-up. Otherwise, patients will be classified as noncompliant.
Catastrophic expenditure will be defined as out-of-pocket (OOP) payments for colorectal cancer treatment that are greater than 40% of nonsubsistence household income per year (capacity to pay) ( Figure 1). This definition is recommended by the WHO [12].
The catastrophic expenditure rate will be defined as the proportion of patients who experience catastrophic expenditure out of all patients included in the study.
OOP payments will be defined as the total amount of money that the patient is charged at the time of service use (excluding insurance coverage, funding schemes, concessions or discounts). This will be done at three timepoints (6 weeks, 3 and 6 months after treatment decision).
Capacity to pay will be defined as nonsubsistence household income over 1 year, i.e. household income after paying for food and household expenses (including rent, loans, mortgage, water and electricity supply). Patient income data will be collected in a faceto-face assessment at the point of entry in the study through three distinct methods: reported patient income, the International Wealth Index and the Gapminder tool (see Appendix 4 for further details on methods for assessing patient income).
The main drivers of treatment cost will be the five most expensive items or services within colorectal cancer treatment, identified from the hospital bills. Categories will be predefined (e.g. theatre fees, ward fees, etc) and further cost categories will be added as data are collected. Funding schemes (e.g. Ayushman Barat, the Prime Minister Relief Fund or others), loans or donations used to support treatment cost for each patient will be recorded.

Sample size
From the previous literature on health and cancer care-related catastrophic expenditure in low and middle income countries (LMICs), the anticipated proportion of cancer patients suffering catastrophic expenditure is around 45% [13,14]. For a prespecified absolute precision of 10% and 5% (error margins recommended by the United Nations for household surveys), a respective sample size of 95-380 patients would be required at a confidence interval of 95% [15].
A pragmatic trade-off is needed for health economic analysis within interventional and observational studies. Longer assessments with fewer patients are likely to mitigate against recall bias and provide more reliable data, rather than performing a larger number of surveys [16]. Therefore, a final sample size of 200 patients is feasible for the CROCODILE study.

Data analysis
Statistical analysis will be undertaken using the R v.3.6.1 (R Project for Statistical Computing). An initial analysis will describe the overall treatment adherence and catastrophic expenditure rates in our sample of 200 patients. For all outcome measures, the appropriate summary statistics will be presented with the adequate effect measures, 95% confidence intervals and two-sided test p-values. Data completeness rates on costs and income will be reported as a secondary outcome. Completeness rates will be described separately for medical, nonmedical and indirect costs, and for the three methods for income assessment.

Subgroup analysis
A subgroup analysis will be performed to explore the impact on treatment adherence and expenditure of distance from home to the hospital. Catastrophic expenditure and treatment adherence rates will be compared between patients being treated at a local versus a distant hospital.

Sensitivity analysis
A sensitivity analyses will be performed to describe catastrophic expenditure rates using two different thresholds: 20% versus 40% of nonsubsistence household income. Different thresholds have been used to define catastrophic expenditure and there is no consensus to date [17]. The correlation between catastrophic expenditure rates at these two cut-offs and patient wealth level can inform future studies on the advantages and disadvantages of both methods.

QUALITATIVE WORK S TRE AM Aim
The aim of the qualitative workstream will be to explore the views and experiences of colorectal cancer patients and professionals regarding barriers to and facilitators for treatment adherence.

Participants
The participants in this workstream will be adults (18 years of age or older) who are diagnosed with colorectal cancer (histopathologically proven) or cancer care professionals who usually work with colorectal patients. The participants need to be able to provide written informed consent (signature or a fingerprint).

Participant identification and consent
Patients and professionals will be identified from surgical oncology, medical oncology or radiation oncology departments. Patients included in the qualitative and quantitative workstreams will overlap.
A participant information sheet (Appendix 5) will be given to each participant before consent is be taken by a researcher trained in qualitative methods.

Sample size and sampling method
For this workstream, three to five patients and three to five professionals will be included from each participating site. Purposive sampling will aim for diversity around patients' demographic and social features, wealth levels and types of treatment. A range of professionals will be invited to participate, enabling exploration of specialty-specific challenges that patients might face while being treated in the different oncology disciplines (medical, surgical and radiation oncology). At the time of inclusion in the quantitative workstream, demographic, economic and disease data will be collected, enabling identification of patients suitable for inclusion in the qualitative part of the study.

Methodological approach
A generic qualitative approach will be used in this study with a main focus on participants' experience of challenges regarding treatment adherence [18]. A combination of inductive and deductive approaches will be used to build the framework for analysis. A literature search of barriers to access and receipt of cancer treatment in LMICs will be performed to preselect codes and themes to be included in the framework (deductive approach). Further codes and categories will be generated from the collected data through unrestricted coding (inductive approach).
Planning of the interviews will be aligned with this, ensuring that preidentified topics are covered with the participants but also allowing the space for new unexpected aspects to be raised during the interviews.

Data collection and management
One-to-one interviews will be conducted with all the participants. A semistructured approach will be used, with a few prompt questions but allowing the participants to express their views (topic guide for interviews in Appendix 6 and 7). The interview will be planned to last between 30 and 60 min and will take place in the most convenient location possible for participants. Due to the travel restrictions related to COVID-19, the interviews might need to be done remotely via video call. For participants who cannot communicate fluently in English, an onsite interpreter will be used to allow translation of the questions into the local language and answers back to English.
Interpreters will be trained about the need for accurate translation and confidentiality prior to undertaking interviews.
The interviews will be digitally recorded after the participants' consent and written notes will be taken. The interviews will then be transcribed clean verbatim and anonymized before analysis. NVivo v.12 will be used to assist data management. The transcript will be coded by a researcher before the categories and themes emerging from the data are identified. A second researcher will code the first set of three or four interviews, maximizing interpretation of the data and refining the labelling and coding strategy for the subsequent transcripts.

Data analysis
From the available methodologies for qualitative data analysis, framework analysis is recognized as the most useful for health systems and policy informing research [19]. An analytical framework will be used for the transcripts to be indexed to the different codes and categories within the framework matrix. While charting the data into the framework, the researchers will keep a record of illustrative quotes to enrich the final conclusions of this workstream. Data interpretation will be developed from the identified categories and themes, and the connections between them. The main domains of the results of the study will be the barriers to and the solutions for treatment adherence. The consolidated criteria for reporting qualitative research (COREQ) will be used to report the results. Finally, the results of this workstream will be interpreted together with the results of the quantitative workstream, to provide a deeper understanding of patients' options and decisions.

DISCUSS ION
The CROCODILE study will identify a vulnerable group of patients in India who struggle to adhere to a treatment plan for colorectal cancer either because they cannot pay for it or due to other reasons.
The Lancet Oncology Commission recommends that the scaling up of cancer services should be based on a country's cancer burden and state of cancer care, highlighting the importance of including both patients and cancer care providers in this process [20]. CROCODILE will provide real-world data on the rates of catastrophic expenditure, barriers that patients face to treatment adherence and how to overcome them.
Financial protection is one of the global health priorities defined by the WHO (sustainable development goal 3.8) and mitigating against catastrophic expenditure has also been prioritised by the Lancet Commission on Global Surgery [21]. An evaluation of 74 000 Indian household surveys showed that when family members were affected by cancer households more likely to make OOP healthcare payments and had higher rates of asset sales than households not affected by cancer [22]. CROCODILE will validate novel methods for patient-level data collection on catastrophic expenditure. The results will be used to inform health economic evaluations alongside global interventional and observational studies.
We acknowledge the exploratory nature of this study, and we will draw careful conclusions from our results. The number and type of hospitals taking part in CROCODILE (four charity hospitals and one government hospital) might influence the type of patients included in the study and their views.
The COVID-19 pandemic has had a huge impact on non-COVID research. Although the study design has not been changed as a result of the pandemic, its delivery has been affected. The research teams are being trained through online platforms. Onsite training with direct contact with patients, patient files and hospital bills has not been possible, which makes data collection more challenging.
The results will be interpreted in light of the effects of the pandemic on cancer care.

ACK N OWLED G EM ENTS
India Institute Fellowship scheme for supporting this research, under the Strategic Global Engagement programme of the University of Birmingham.

CO N FLI C T O F I NTE R E S T
There are no conflicts of interest to declare.

E TH I C A L S TATEM ENT
Indian Council of Medical Research, University of Birmingham and hospital level ethical approvals were obtained for the study.

AUTH O R CO NTR I B UTI O N
All the listed co-authors have contributed to the design and draft of this protocol, critical revision, final approval of the version to be published and agreed to be accountable for the accuracy and integrity of the produced work.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data can be requested from all the participating centres and will be made available for further analysis.

R E FE R E N C E S
*Joint senior authors. †Joint first authors.

A PPEN D I X 2
Patient information sheet (CROCODILE quantitative workstream)

Previous studies show that colorectal cancer (CRC) patients in
India present with more advanced disease compared to western patients. The main goals of this project are to explore barriers in access to cancer treatment in India, particularly affordability of care.

WH AT TH I S S TU DY E NTA I L S
To do this, we need to collect data about your disease and the treatment you received, including how much it cost for you. You were enrolled in this study because you have a CRC diagnosis and you will undergo treatment. Participation in this study will have no impact on your treatment. No changes to your treatment will be made whether you decide to participate or not.

TI M E CO M M ITM ENT
The time commitment for you is very low (about 20 minutes). Whilst you are in hospital we will invite you to answer the survey and we will support you in any questions you have about it. At 6 weeks, 3 months and 6 months after this in-person contact, we will make a telephone call to confirm the total cost of cancer treatment with you. After this, your participation will be over and nothing further will need to be done.

I N FO R M ATI O N TO B E CO LLEC TED
We will collect data about your disease and the treatment you were or will be given. Data about your treatment cost will be collected from hospital records. You will be asked to fill a survey with some details about your income and the assets you own. Besides that, you will be asked to identify 5 pictures of house goods that look like your own. The questionnaire also covers transport, food and accommodation expenses. If any questions make you uncomfortable, you do not have to answer them.

CO N FI D ENTI A LIT Y
Information about you will be kept confidential. Your telephone contact will be used only for the study purpose and will be kept in A PPE N D I X 3

Details of patient income assessment
The patient income assessment will be performed by a trained researcher during the face-to-face assessment, at the point of study entry, through three different methods: reported household income and capacity to pay, the International Wealth Index (IWI) and the Gapminder assessment tool.

PAY
This is made up of three parts: least possible information about you that is needed for the research will be sent to the University of Birmingham which is coordinating this study. It will be stored for 05 years but will then be destroyed.
We will keep the data as safely and less detailed as possible; no records of your name e-mail or telephone will be kept in the study central files.

CO N S E NT
It is up to you to decide to join the study. If you agree to take part, we will ask you to sign (or fingerprint) a consent form. You are free to leave the study, without giving a reason. If requested, we can remove your information from the study. Withdrawing from the study will not affect the care you receive.

G A PM I N D ER A SS E SS M ENT TO O L
The Gapminder Foundation created the Dollar Street project, which aims to show how people live their daily lives through pictures taken from all over the world (2). Household income is collected from each household and each picture is matched to the income of the household where it belongs.
Given that each picture is matched to a specific income, this can be a user friendly way to assess patients' income, which is typically hard to collect. Patients will be asked to select the household items that look most like their own from the pool of pictures available on the Gapminder website for that particular item. By including all the pictures available in the Dollar Street project for that particular item we will ensure that a discriminatory range of matching incomes is available. Only pictures from India will be presented to patients.
Patient income will be calculated as a mean average of the incomes matched to the pictures selected by the patient. An example is given below: Income (per day, in dollars) = 1.489056 × exp (0.02918 × IWI score)

I NTER N ATI O N A L WE A LTH I N D E X
The IWI (see questionnaire below) is a validated tool for wealth assessment with 12 questions about household assets that individuals might or might not own. Each asset has a relative weight and contribution to the IWI score. The final score ranges from 0 to 100 and reflects the household wealth.

BACKG RO U N D
Colorectal cancer is the 4th most incident cancer in India according to World Health Organization and its incidence is increasing.
Previous studies show that colorectal cancer patients in India are usually younger than in the West and have more advanced disease.
This study aims to explore patient and healthcare professional's perspectives on how patients access treatment care in India.

WH AT TH I S S TU DY E NTA I L S
Participation in this study will have no impact on patient treatment. You were enrolled in this study because you are either a patient with a CRC diagnosis or a healthcare professional in charge of CRC patients (diagnosis or treatment). A researcher will ask you about the steps involved in receiving cancer treatment, difficulties in that process and your views on what should be done differently to facilitate access to cancer care.

TI M E CO M M ITM ENT
The time commitment for you is about 1 h. Once this interview is completed, your participation will be over.

I N FO R M ATI O N TO B E CO LLEC TED
Only basic information will be collected. This will include information about you, your illness, how did you seek health care and your ideas about how to improve access to cancer care. During the interview the researcher may make some notes. The researcher will ask for permission to audio record their interview with you (this is optional). If any questions make you uncomfortable, you do not have to answer them.

CO N FI D ENTI A LIT Y
Information about you will be kept confidential. You will be given the The least possible information about you that is needed for the research will be sent to the University of Birmingham which is coordinating this study. This information will be stored for 05 years but will then be destroyed. We will keep the data as safely and less detailed as possible; no records of your name or e-mail will be kept in the study central files.

CO N S E NT
It is up to you to decide to join the study. If you agree to take part, we will ask you to sign (or fingerprint) a consent form. You are free to leave the study, without giving a reason. If requested, we can remove your information from the study up to 30 days after your interview.
Withdrawing from the study will not affect the care you receive.