http://orcid.org/0000-0002-3763-5494
Figures
Abstract
Background
We aimed to provide health practitioners greater insight into the public perception of traditional and complementary medicine (T&CM) use. Our objectives were to identify news media reports of children abandoning conventional treatment for traditional and complementary medicine, analyze the thematic content of these news articles and estimate the tonality portrayed.
Methods
LexisNexis and Factiva were searched for terms related to cancer, children and T&CM. Inclusion criteria were children less than 18 years, in curative phase of treatment who attempted to abandon conventional therapy for any traditional and complementary medicine use. A secondary search was performed in LexisNexis, Factiva and Google News Archive with the names of children in identified cases. Qualitative analysis of news media reports was completed using a grounded theory approach. Quantitative analysis of article sentiment was performed using a linear support vector machine.
Results
Seventeen cases occurring between 2002 and 2016 were included. Five main themes were identified: treatment as torture, power imbalances, rights of parents, evidence versus beliefs and the rights of Indigenous Peoples. Sentiment analysis revealed an overall negative tone, as demonstrated by 73% of the articles.
Citation: Diorio C, Afanasiev M, Salena K, Marjerrison S (2018) ‘A world of competing sorrows’: A mixed methods analysis of media reports of children with cancer abandoning conventional treatment. PLoS ONE 13(12): e0209738. https://doi.org/10.1371/journal.pone.0209738
Editor: Marta M. Alonso, Universidad de Navarra, SPAIN
Received: December 29, 2017; Accepted: December 11, 2018; Published: December 21, 2018
Copyright: © 2018 Diorio et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: The Hamilton Health Sciences New Investigator Fund (grant 16392), held by SM, supported this work. URL: http://www.hhsresearchadmin.ca/researcher-support/internal-funding/new-investigator-fund/. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Traditional and complementary medicine (T&CM) use among children with cancer is common worldwide, with use often motivated by intention to cure.[1, 2] These T&CM therapies include any health practices outside of the dominant, Western or allopathic paradigm, hereafter referred to as conventional therapy (CT). With regard to cancer treatments, CT typically includes chemotherapy, radiation and/or surgery. Treatment abandonment, defined as failure to begin or complete curative CT,[3] may be related to a desire for exclusive use of T&CM.[4] Treatment abandonment is rare in high-income countries (HIC), with estimates of less than 5% of families attempting to abandon treatment.[5] A recent study in adult cancer patients showed that exclusive use of T&CM strategies without CT is associated with a greater risk of death.[6]
There is a paucity of medical literature examining why some families prefer T&CM over CT. However in HIC, cases where families leave or attempt to leave CT may be published in the news media. These cases may have a disproportionate impact on public perception of the interplay between T&CM and CT. Previous studies have examined legal cases where families have refused CT in HIC. [7, 8] To our knowledge, no studies have examined media reports of abandonment of CT for T&CM in HIC.
We reviewed media portrayals of families leaving CT for T&CM in pediatric cancer in order to help practitioners understand what pre-conceived notions exist about the role of T&CM and CT, with the hope of improving collaboration between healthcare providers, T&CM practitioners and families.
Methods
Data sources and searches
We searched the news databases LexisNexis and Factiva for terms related to cancer, children and traditional and complementary medicine (T&CM) (see S1 Appendix) between May 24, 2006 and May 24, 2016. LexisNexis is an online database of over 350 full text newspapers from Europe and Norther America.[9] Factiva is an online database from Dow Jones containing over 32, 000 sources. Both databases include local, national and international news sources.[10] For feasibility, we reviewed only news articles in English and over a ten-year period; however, included any cases that had occurred since 2002, the year when the first iteration of the World Health Organization (WHO) T&CM strategy was published.[11, 12] Articles were screened by two reviewers (CD and SM) and cases were included based on the following criteria: 1) Child (<18) with a cancer diagnosis, 2) Attempt was made to abandon or successfully abandoned cancer therapy that was deemed lifesaving or necessary by the treating team, 3) Case occurred between January 1, 2002 and May 24, 2016 and 4) Treatment was left to seek out any type of T&CM as per National Center for Complementary and Integrative Health (NCCIH) definition, including prayer.[13] Exclusion criteria included: 1) Adult (>18) patient, 2) Non-cancer diagnosis, 3) Child in palliative phase of treatment and 4) Treatment sought not T&CM.
After cases were identified through the primary search, a secondary search was initiated using the child’s name. Where the child’s name was not available, specific terms related to their case were searched. Databases searched included LexisNexis Academic, Factiva and Google News Archive. All articles available through LexisNexis Academic and Factiva were retrieved for sentiment mining but only cases with greater than 15 articles were included in the sentiment analysis. A sample of up to 20 articles for each identified case were retrieved for qualitative analysis.
Outcomes
For every case identified, we sought to identify the following outcomes where available: name, age, gender, location, conventional therapy received, T&CM proposed or used, and clinical outcome.
Qualitative analysis
Two study team members (KS and CD) analyzed 20 articles for each case, or all articles that were available if less than 20 articles were published, from a grounded theory perspective.[14, 15] Documents were coded and concepts were identified based on coded elements. Concepts were compared between study team members and organized into themes and sub-themes using a consensus-based approach.
Sentiment analysis
Articles related to 10 patients were analyzed (Abraham Cherrix, Cassandra Fortin, Daniel Hauser, Jessica Crank, JJ, Katie Wernecke, Makayla Sault, Neon Roberts, Oshin Strachan and Sarah Hershberger) using automated sentiment analysis. 100 articles from across the case spectrum were analyzed by two study team members (CD and KS), and each article’s sentiment was rated as positive or negative. Designation of sentiment was appended based on the overall tone of the article. We did not distinguish between articles that were negative or positive toward a specific party, i.e. the patient or the physician, but instead evaluated the overall tone of the article.
These articles, which composed the training dataset, were then loaded into a computer program and ‘vectorized’. In machine learning, ‘vectorization’ refers to the process whereby real data are transformed into a numerical representation more easily interpretable by computational algorithms. The vectorization of the training set followed the ‘bag of words’ approach. Each word was mapped to an index in a vector, with the numerical value of that vector component encoding the number of occurrences.[16] In order to reduce the number of unique features, each word was lemmatized and transformed into its base dictionary equivalent. Stop words, punctuation, digits, and proper nouns were removed. The lemmatization process was assisted by the use of the Natural Language Toolkit (NLTK) package for Python.[17]
Once lemmatized, the bag-of-words vectors were used as training data for a linear support vector machine (SVM). A SVM was chosen over other competing machine learning algorithms (i.e. neural networks) due to its relative stability in the presence of fewer samples.[16] The goal of a linear SVM is to find the maximum margin hyperplane between two classes—positive and negative articles—based on the components of each of the bag-of-words feature vectors and each vector’s associated class.
Several parameters were tuned in order to train the SVM. These included loss and penalty functions, weight of regularization parameters, and number of iterations required for stable convergence of a stochastic gradient descent (SGD) classifier. A 5-fold cross validation settled upon a modified huber loss function with an L2 penalty, a regularization weight of 0.001, and 10 iterations. When applied to the training dataset, the trained model classified approximately 70% of the documents correctly. All training and classification was done within the scikit-learn ecosystem, a Python package for machine learning and data analysis.[18]
Results
Outcomes
7722 articles were retrieved in the primary searches. After duplicates were removed, 6172 articles were screened by two reviewers (CD and SM). A total of 17 cases were identified that met inclusion criteria; features of the cases are outlined in Table 1.
Qualitative analysis
Five main themes were identified. See Table 2 for a complete list of themes and sub-themes with supporting quotations.
- Treatment as torture: the idea of cancer treatment being tantamount to torture was raised in discussion of many of the identified cases. For example, when describing the experience of their child Oshin receiving chemotherapy his parents stated, “It almost feels like Nazi Germany and I am honestly sickened by the treatment of all these children… nothing short of toxic hell”. [19]
- Power imbalances: several reports portrayed the cases as a power struggle between conflicting decision makers, all trying to represent the best interest of the child. Parents were portrayed as feeling powerless in the face of medical systems: “Mr Strachan posted another comment about the treatment, saying: ‘Just when you think you know what pain is something comes along to show you some more. In hospital watching young Oshin getting sick and there is not a f***ing thing I can do about it.’”[20]
- Rights of parents: much of the controversy centered on the challenging issue of parental rights versus the rights of the state. For example, in an interview with the Wernecke family, the interviewer Kerry Sanders stated, “Her father and mother argue this is a case of parental rights, but Texas state officials say those rights do not include the right to gamble with Katie [Wernicke]’s life.” [21] Regional differences exist in how each case was arbitrated, however, common among all cases was the theme of whether or not parents had the right to make a decision that they felt was in the best interest of the child if that decision involved abandoning a potentially curable treatment for an unproven one.
- Evidence versus belief: also controversial was the right of parents to reject empirically based medical therapies for treatments they believed to be less harmful or more effective. Patients and parents sought a more personal and individual treatment then what they felt was available from the medical system. Cassandra Fortin, a 17-year old teenager diagnosed with Hodgkin lymphoma stated, “I am so sick of being treated like a number and how everything is based off of statistics. I am a patient not a number.” [22]
- Rights of Indigenous Peoples: two cases, J. J. and Makayla Sault, both occurring in the same region of Canada, had the additional complexity of the Indigenous identity of the children involved. These controversial cases revealed the wide breadth of opinions held by Canadians with regard to the experience of Indigenous patients. For example, an editorial published in the Calgary Herald declared: “Traditions and community are wonderful things, but they are not cures for cancer. And if any indigenous treatments are potential cures, they should be used only after rigorous clinical trials have proven their efficacy, as happens with all other treatments.” [23] Conversely, an unattributed quotation from an article in the National Post noted, "Aboriginal healers, medicines pulled from the soil and even seers are fixtures of the community, located barely an hour southwest of Toronto, residents say. To many of the reserve's people, Western style medicine is alternative health-care and native remedies the mainstream, said one community leader.” [24]
Sentiment analysis
The SVM applied to the entire collected corpus of articles reported a 27% / 73% split between positive and negative articles. Most cases had an overall negative tone (See Table 3). Additionally, we looked at the positive or negative weight given to each word. For a given word, the weight corresponds to how important that feature is in generating a positive or negative prediction.
The top 5 most positive and negative features from the training dataset are reported in Table 4. Precision reported from training is approximately 70%, compared with the precision reported by multiple humans classifying the same dataset has been reported as approximately 80%. [25] A review of a subset of the corpus resulted in a precision of 72%, which was expected from the training data.
Discussion
Media reports of children with cancer leaving CT for T&CM offer insight for health practitioners into public perception. From the 17 cases we identified, 5 major themes emerged: treatment as torture, power imbalances, rights of parents, evidence versus beliefs and rights of Indigenous Peoples. Tonality analysis of reports was largely negative, with the exception of reports about Abraham Cherrix (64% positive). It is unclear why this particular case had more positive coverage than other cases from the same time period (e.g. Katie Wernecke). This may be related to regional differences in reporting style and in acceptability of abandonment of therapy.
Standardized care, fundamentally responsible for the survival gains in pediatric cancer over the past half-century, was identified in several cases as a negative aspect of conventional care.[26] Families stated that they wanted their child to be treated as an individual as opposed to a “statistic”. Physicians as representatives of an industry that were “experimenting” on children was also a recurring theme. This sentiment is anathema to the vast majority of physicians; however, some families may enter the therapeutic relationship with profound mistrust. Working to include trusted T&CM practitioners in the care team may provide the approach to care that families require to build trust with the medical system.
In most countries, parents have a legal right to make decisions for their child; limitations exist on these rights to protect children from harm. This includes harm from failing to provide potentially curative medical therapy, known as medical neglect.[27] Controversy arises when the duty of the state to protect the child is pitted against opposing parental or individual rights. The case of Jessica Crank demonstrated conflict between the right to religious freedom and the State’s duty to protect children from medical neglect. Previous analyses have identified potential inadequacies in the laws in the United States protecting children from religiously motivated medical neglect.[28] The cases of J.J. and Makayla Sault set the rights of individuals as Indigenous Peoples to pursue Traditional medicine against the fiduciary duty of healthcare providers to protect children from harm.[29] The interplay between child protection rules as they pertain to medical responsibility and the rights of Indigenous families to choose Indigenous practices was described by ethicist Margaret Somerville as "a world of competing sorrows, because no matter what you do somebody is going to be hurt or harmed or upset".[30] In Canada, a long history of colonialism and post-colonial institutionalized racism has spawned significant mistrust between Indigenous patients and the healthcare and legal systems.[31, 32] Disturbingly, we found that the word “Aboriginal” was associated with a negative sentiment in news articles. As outlined in the Calls to Action of the Truth and Reconciliation Commission of Canada, it is imperative that healthcare practitioners recognize the importance of Traditional healing and incorporate these practices in the treatment of Indigenous patients where requested.[33]
Our study presents a mixed-methods approach to analyzing news reports in pediatric oncology. We used news media as primary source material to interrogate the data source most accessible to patients and families. Our approach allowed insight in to a larger corpus of published articles than would have been achievable with solely qualitative methods. Our study has several limitations. As is the case with all machine learning algorithms, the addition of more training data would improve the results of the automated sentiment analysis. Our study is also limited by the quality of source material available. Many important cases are likely kept out of the news media. Due to privacy laws in most countries, cases that are in the news media are brought forward by parents and families, and not hospitals or physicians. This is an important source of bias in any analysis of news media reports of a health topic. Further, LexisNexis and Factiva present only a sampling of the news available. Further methodological strategies are necessary with the advent of online news, and non-traditional news platforms such as social media.
We were unable to include information from low and low-middle income countries, where treatment abandonment is a more significant contributor to mortality, because of language barriers, availability of newspapers in online databases, and in settings where treatment abandonment is more common it is less likely to be “newsworthy”. News reports of abandonment of conventional cancer treatment for traditional and complementary medicine in low and low-middle income countries deserve careful scrutiny, however, it was outside the scope of this project. Further study should be focused on this area.
We present general assessments of overall tonality. Future analyses of media reports should attempt to quantify the positive or negative sentiment for specific elements, i.e. for or against providers, for or against parents, etc. This level of analysis was beyond this scope of this project. We noted that the term “doctor” was associated with a more negative sentiment, implying that there may be negative sentiment in news articles towards conventional health providers. Alternatively, “family” was associated with a positive sentiment. Future studies should consider interviewing families who have abandoned or attempted to abandon CT as a primary source. This task represents a daunting feasibility challenge, as the willingness of these families to participate in research may be lower than average.
We have analyzed news reports of families abandoning or attempting to abandon CT for T&CM. We hope a better understanding of what is portrayed to the public can help healthcare providers collaborate with families and T&CM providers to continue to improve outcomes for all children with cancer.
Supporting information
S1 Appendix. Search strategy.
Search strategy for primary searches.
https://doi.org/10.1371/journal.pone.0209738.s001
(DOCX)
Acknowledgments
We would like to acknowledge the SIOP PODC T&CM Task Force for their support of this work through feedback and collaboration.
References
- 1. Diorio C, Salena K, Ladas EJ, Lam CG, Afungcwhi GM, Njuguna F, et al. Traditional and complementary medicine used with curative intent in childhood cancer: A systematic review. Pediatric blood & cancer. 2017;64(9):e26501.
- 2. Diorio C, Lam CG, Ladas EJ, Njuguna F, Afungchwi GM, Taromina K, et al. Global Use of Traditional and Complementary Medicine in Childhood Cancer: A Systematic Review. Journal of Global Oncology. 2016: ePub ahead of print Nov 9.
- 3. Mostert S, Arora RS, Arreola M, Bagai P, Friedrich P, Gupta S, et al. Abandonment of treatment for childhood cancer: position statement of a SIOP PODC Working Group. The Lancet Oncology. 2011;12(8):719–20. pmid:21719348
- 4. Friedrich P, Lam CG, Kaur G, Itriago E, Ribeiro RC, Arora RS. Determinants of Treatment Abandonment in Childhood Cancer: Results from a Global Survey. PLoS ONE. 2016;11(10):e0163090. pmid:27736871
- 5. Friedrich P, Lam CG, Itriago E, Perez R, Ribeiro RC, Arora RS. Magnitude of Treatment Abandonment in Childhood Cancer. PLoS ONE. 2015;10(9):e0135230. pmid:26422208
- 6. Johnson SB, Park HS, Gross CP, Yu JB. Use of Alternative Medicine for Cancer and Its Impact on Survival. JNCI: Journal of the National Cancer Institute. 2018;110(1):djx145–djx.
- 7. Hord JD, Rehman W, Hannon P, Anderson-Shaw L, Schmidt ML. Do parents have the right to refuse standard treatment for their child with favorable-prognosis cancer? Ethical and legal concerns. Journal of clinical oncology: official journal of the American Society of Clinical Oncology. 2006;24(34):5454–6.
- 8. Burns C, Rassekh S, Corbett R, MacFarlane S, Sullivan M. When Parents Refuse Potentially Curative Treatment for their Child's Cancer-Does Legal Intervention Improve Outcome? Pediatric blood & cancer. 2016;63:S87-S.
- 9.
LexisNexis. 2018 [Available from: https://www.lexisnexis.com/communities/academic/default.aspx.
- 10.
Proquest. 2018.
- 11.
WHO traditional medicine strategy 2002–2005. Geneva, Switzerland; 2002.
- 12.
WHO Traditional Medicine Strategy 2014–2023. Geneva, Switzerland; 2013.
- 13.
Complementary, Alternative, or Integrative Health: What’s In a Name?: NCCIH Clearinghouse; 2015 [Available from: https://nccih.nih.gov/sites/nccam.nih.gov/files/Whats_In_A_Name_08-11-2015.pdf.
- 14.
Strauss A, Corbin JM. Grounded theory in practice: SAGE; 1997.
- 15. Strauss A, Corbin J. Grounded theory methodology. Handbook of qualitative research. 1994;17:273–85.
- 16.
Ullman JD. Mining of massive datasets: Cambridge University Press; 2011.
- 17.
Bird S, Klein E, Loper E. Natural language processing with Python: analyzing text with the natural language toolkit: O'Reilly Media, Inc.; 2009.
- 18. Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, et al. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research. 2011;12(Oct):2825–30.
- 19. Noble F, Devlin P. "I don't want my son's brain fried': How this mother refuses doctors' advice to have her son undertake radiotherapy because she 'doesn't understand it'. Daily Mail Australia 2016 April 10.
- 20. Devlin P. 'I want to offer him peace, love and fun while we still can’: Heartbroken mother explains why she and her husband decided to deny their dying son debilitating chemotherapy. Daily Mail Australia 2016 April 8
- 21.
Wernecke K, Wernecke E, Wernecke M, Corrona L. Profile: Michele and Edward Wernecke and attorney Luis Corrona discuss involvement of court in Katie Wernecke's cancer treatment. In: Saunders K, editor. NBC News: TodayJune 9 2005.
- 22. Baker KC. Connecticut Teen Forced by Courts to Undergo Chemo Says New Mass Found: 'This Is Why I Fought So Hard Against Chemotherapy'. People. 2016 April 19.
- 23.
The Calgary Herald. A Needless Tragedy The Calgary Herald. 2015 January 21
- 24. Blackwell T. 'We don't have anything to prove'; No regrets for parents of cancer-stricken girl who opted for holistic treatment. National Post. 2014 June 7.
- 25.
Wilson T, Wiebe J, Hoffmann P, editors. Recognizing contextual polarity in phrase-level sentiment analysis. Proceedings of the Conference on Human Language Technology and Empirical Methods in Natural Language Processing; 2005: Association for Computational Linguistics.
- 26. Pearson HA. History of pediatric hematology oncology. Pediatric research. 2002;52(6):979–92. pmid:12438679
- 27. Jenny C. Recognizing and responding to medical neglect. Pediatrics. 2007;120(6):1385–9. pmid:18055690
- 28. Asser SM, Swan R. Child fatalities from religion-motivated medical neglect. Pediatrics. 1998;101(4 Pt 1):625–9.
- 29. Mitchell I, Guichon JR, Wong S. Caring for children, focusing on children. Paediatr Child Health. 2015;20(6):293–5. pmid:26435666
- 30. Hasham A, Alamencizk T. Aboriginal medicine ruling ignites debate on children's rights; Legal experts see flaws in judge's decision to allow girl's family to refuse treatment for leukemia. The Toronto Star 2014 November 20.
- 31.
Hill DM. Traditional medicine in contemporary contexts. National Aboriginal Health Organization; 2003.
- 32. Richardson L, Stanbrook MB. Caring for Aboriginal patients requires trust and respect, not courtrooms. CMAJ. 2015;187(3):162. pmid:25602005
- 33.
Truth and Reconciliation Commission of Canada: Calls to Action Winnipeg; 2015.
- 34.
Harris E. Connecticut Teenager With Cancer Loses Court Fight to Refuse Chemotherapy. The New York Times. 2015 January 10;Sect. A (col. 0).
- 35. Boyle D. Teen, 17, in remission from cancer after being forced to undergo chemo fights to be reunited with her mother after she was put into care. Daily Mail 2015 March 17.
- 36. Young E. For son's own good: parents share angst over forced treatment; Exclusive—Health—Doctors could not estimate success. Sydney Morning Herald. 2016 April 9;Sect. News.
- 37.
Linehan D. Hausers agree to chemo. Mankato Free Press. 2009 May 27.
- 38. Verville J. The right not to feel pain. The Toronto Star. 2008 May 14.
- 39. Ellicott C, Bentley P. Neon's beaming again: Boy, 8, snatched by mother who thought cancer doctors would 'fry his brain' is on the mend after radiotherapy. Daily Mail. 2013 August 9.
- 40. Simpson E. For now, Abraham Cherrix is in remission. The Virginian-Pilot. 2011 September 25.
- 41. Butler D. Father loses access to his son with leukemia in fight over chemo. Ottawa Citizen 2014 September 13;Sect. CITY.
- 42. Smith R. Mum loses chemo battle for cancer boy. Geelong Advertiser. 2016 April 9.
- 43.
Toobin J. Faith's Failing. In: Hemmer B, editor. CNN: American Morning with Paula Zahn. Federal Document Clearing House: Cable News Network 2002.
- 44. Branley A. Decision to decline chemotherapy for son parent's hardest call. Newcastle Herald. 2013 June 13.
- 45. Germanotta T. Court, teen with cancer come to terms on treatment plan The Virginian-Pilot. 2006 17 August
- 46.
Michele Wernecke EWKW. Profile: Michele and Edward Wernecke and attorney Luis Corrona discuss involvement of court in Katie Wernecke's cancer treatment In: Saunders K, editor. NBC News: Today NBC Universal Inc.; 2005.
- 47. Chapman R. Parents Refuse Chemo For Daughter, Court Intervenes. CBS Minnesota. 2012 6 August.
- 48. McLaren L. Whose rights are served when a girl dies? The Globe and Mail. 2015 23 January.
- 49. Hershberger A. Amish Girl to Get Cancer Treatment Court Overrules Parents Objections to Chemo. In: Perez A, editor. ABC News: Good Morning America. United States of America 2013.
- 50.
Linehan D. Chemo fugitives return on own. Mankato Free Press. 2009 May 25.
- 51. Gallant J. Ontario may appeal ruling allowing girl, 11, to be taken off chemo. Toronto Star. 2015 6 March.