Original Article
Improving Our Understanding of the Use of Traditional Complementary/Alternative Medicine in Children With Cancer Elena J. Ladas, PHD, RD1,2; Meiko Lin, MS3; Federico Antillion, MD4; Silvia Rivas, MD4; Guillermo Chantada, MD5; Walter Cacciavillano, MD5; Roberta Ortiz, MD6; Katja Stein, PHD, RD7; Luis Castillo, MD8; Valeria Rocha, PhD8; Ligia Fu, MD9; Hilze Rodriquez, MD10; and Kara M. Kelly, MD1
BACKGROUND: Children with cancer in high-income and low-income countries often use traditional complementary/alternative medicine (TCAM). With efforts by the World Health Organization and international twinning programs improving access to conventional care for patients with childhood cancer, understanding the global use of TCAM is important because reliance on TCAM may affect time to presentation, adherence, and abandonment of care. In the current study, the authors describe the process and validation of an international survey documenting the use of TCAM among children with cancer. METHODS: The survey was designed to collect information on TCAM use and associated factors through both open-ended and close-ended questions. During the period between June 2012 and December 2013, the survey was administered to 300 children and adolescents (or their parents) who were undergoing treatment for cancer at a collaborating institution located in Mexico, Uruguay, and Nicaragua. RESULTS: For the majority of constructs, the survey demonstrated strong test-retest reliability as evidenced by an intraclass correlation of at least 0.79 in each of the participating countries. The survey demonstrated good internal consistency and reliability across countries (a range from. 77 to. 85 for the belief scale; and an a range from. 60 to. 86 for the cause scale) and convergent validity between TCAM beliefs and behavior constructs (adjusted correlation range, 0.35-0.60). CONCLUSIONS: The results of the current study demonstrate the successful development of a cross-cultural survey that produced results that were reliable and valid. These findings will aid investigators in providing guidelines concerning TCAM, support the development of education and research priorities, and identify variables associated C 2014 American Cancer Society. with TCAM that are region-specific. Cancer 2014;000:000-000. V KEYWORDS: pediatric oncology, complementary/alternative medicine, supportive care, survey design, twinning programs.
INTRODUCTION The differential mortality outcomes in childhood and adolescent cancer between developed and developing countries have become increasingly apparent. Greater than 80% of children with cancer reside in developing countries, in which survival rates are estimated to be approximately 35%, which is in stark contrast to the overall survival rate of 70% observed for children residing in developed countries.1,2 Although improved access to drugs, political stability, and the availability of locally trained health care providers have been crucial in expanding the delivery of Western care, the role of traditional complementary/alternative medicine (TCAM) in regions with diverse medical systems is less understood. Understanding the application and role of TCAM in other pediatric conditions, including human immunodeficiency virus/acquired immunodeficiency syndrome and severe malaria, has helped to bridge the gap in communication among traditional healers and conventional health care providers and improve access to medical care in children with these conditions.3-5 To date, the impact of TCAM on the quality of care and treatment-related outcomes of childhood and adolescent cancer is not well-understood. TCAM are therapies that are outside the context of conventional medical care and include modalities such as acupuncture, massage, energy medicine, and herbal remedies. International surveys have reported that the use of TCAM is high among children with cancer, with the prevalence of use exceeding 60%.6 However, a validated survey that is crossCorresponding author: Elena J. Ladas, PhD, RD, Division of Pediatric Hematology, Oncology, and Stem Cell Transplant, Institute of Human Nutrition, Columbia University Medical Center, 7th Fl, 161 Fort Washington Ave, New York, NY 10032; Fax: (212) 305-5848; [email protected] 1 Division of Pediatric Hematology, Oncology, and Stem Cell Transplant, Columbia University Medical Center, New York, New York; 2Institute of Human Nutrition, Columbia University Medical Center, New York, New York; 3Interdisciplinary Studies Program, Teachers College, Columbia University, New York, New York; 4Hemato-Oncology Service, National Pediatric Oncology Unit, Guatemala City, Guatemala; 5JP Garrahan Hospital, Buenos Aires, Argentina; 6Manuel de Jesus Rivera La Mascota Children’s Hospital, Managua, Nicaragua; 7Dr. Juan I. Menchaca Civil Hospital of Guadalajara, Jalisco, Mexico; 8Pereira Rossell Medical Center, Montevideo, Uruguay; 9Escuela University Hospital, Tegucigalpa, Honduras; 10Panama City, Republic of Panama
We thank the Tamarind Foundation for its support of this international effort. The results of the current study demonstrate the successful development of a cross-cultural survey exploring the use of traditional complementary/alternative medicine (TCAM) in children with cancer. These findings will aid investigators in providing guidelines concerning TCAM, support the development of education and research priorities, and identify variables associated with TCAM that are region-specific. DOI: 10.1002/cancer.29212, Received: September 29, 2014; Revised: October 28, 2014; Accepted: October 29, 2014, Published online Month 00, 2014 in Wiley Online Library (wileyonlinelibrary.com)
Cancer
Month 00, 2014
1
Original Article
culturally adaptable has not been described for children with cancer.6 The current study describes the development and testing of a survey documenting the use, reasons, and predictors of TCAM among children with cancer. We describe several phases of the process and provide the reader with an iterative methodology for the design and validation of an international survey exploring TCAM in a conventional medical setting. MATERIALS AND METHODS During the period of June 2012 to December 2013, the described survey was administered to 300 children and adolescents (or their parents) who were undergoing treatment for cancer at a collaborating institution located in Guadalajara, Mexico; Montevideo, Uruguay; and Managua, Nicaragua. Approval was obtained from the Institutional Review Boards at Columbia University Medical Center and those of each participating institution. Eligible participants were identified by a computer-generated block randomization procedure and approached about participation during a routine visit to the outpatient center. After consent and assent (when applicable) was obtained, the survey was then administered face-to-face by a trained interviewer who was from the local region and was not involved in the care of the child. Instrument Development
The survey tool was expanded and modified from a previous survey that was administered to children and adolescents undergoing treatment for cancer at Columbia University Medical Center, the lead institution.7 Participating institutions involved in the development and distribution of the survey are centers with a pediatric oncology division and are located in North, Central, or South America. A select group of centers representing the United States, Guatemala, and Argentina were involved in the development of the survey tool and represent pediatric oncologists and clinical investigators with expertise in TCAM, survey design, and validation. Individuals with expertise in medicine, nutrition, public health, CAM, psychology, and international outreach programs provided additional consultation throughout the development process. This process was divided into 5 phases that included: 1) purpose and domain specification; 2) development of survey specifications; 3) identification of culturally appropriate TCAM classification; 4) content validation; and 5) empirical validation. Figure 1 provides an overview of the development process.8 2
Theoretical framework
The survey was designed to identify the perceived benefits (health promotion, cure, etc), susceptibility (development or management of a side effect or late effect), strength of beliefs in TCAM theories, and expected outcomes (risk of death or disease recurrence) of using TCAM as a component of cancer care. Constructs of the Health Belief Model and the Theory of Reasoned Action and the Theory of Planned Behavior provided the theoretical foundation of the survey design and development. Construct definition
The survey instrument was designed to examine 3 constructs related to TCAM use: 1) beliefs in TCAM theories and its perceived efficacy; 2) behaviors related to TCAM; and 3) causes of cancer or symptoms related to TCAM theories. A total of 24 items were developed according to the domain specifications. The items were screened for match with the domain indicators, difficult vocabulary, double-barreled language, or ambiguous content. The breakdown of the items by construct is as follows: 1) Belief scale: 9 items collecting information regarding belief in TCAM theories and perceived efficacy. 2) Behavior scale: 7 items related to patients’ TCAM therapy-seeking and use behaviors. 3) Cause due to TCAM scale (cancer or symptoms due to TCAM-related theories): 8 items related to patients’ perceived causes of symptoms and cancer. Both the belief and cause scales are scored on a 5point Likert scale with responses ranging from 1 (strongly agree) to 5 (strongly disagree). The behavior scale is scored on a dichotomous scale (0 indicates no and 1 indicates yes). For the belief scale and the behavior scale, 4 questions addressed parental beliefs and practices in TCAM for their personal medical care. Survey format
The survey was classified into 5 sections: 1) health behaviors; 2) health beliefs; 3) TCAM modalities; 4) demographic data; and 5) medical record abstraction. The first 3 sections were separated by construct with the sequencing structure of the questions designed to initiate trust and rapport between the interviewer and respondent before discussing the construct of interest. The survey was designed with both open-ended and closed-ended questions. The closed-ended questions allow for the data to be collapsed or expanded into categories, to evaluate variations in the frequency of TCAM by region, and to allow for administration to a large sample. The open-ended questions provide more descriptive details regarding the Cancer
Month 00, 2014
Global TCAM in Children with Cancer/Ladas et al
Figure 1. An iterative process for designing and validating traditional complementary/alternative medicine (TCAM) measures is shown.
participant’s experience with TCAM therapy and allow for responses to be thematically categorized and analyzed. A protocol was developed to guide transcription and coding procedures using a generic qualitative approach. Language
Participating institutions include many countries in North, Central, and South America in which Spanish is the official language. Different local dialects and words among localities often indicate the same concept; therefore, several steps were taken to ensure the survey was culturally and linguistically relevant. The survey was first translated from English into Spanish from the original version administered in North America. The Spanish version was reviewed by individuals and professionals at the local hospital to ensure clarity and participant comprehension. Modifications were indicated in Spanish, and surveys were then back-translated into English. CompariCancer
Month 00, 2014
son of the 2 English versions was evaluated for agreeability and discrepancies. All identified variations were reviewed by an independent translator familiar with the dialect of the intended country and modified if indicated. Survey comprehension was ensured by administering the translated survey in the local language to a small number of patients and members of the staff. Administration of the survey
The survey was designed to be administered by an interviewer who was not involved in the respondent’s cancer care to ensure respondents felt comfortable discussing TCAM use without feeling that their medical care would be compromised. This format also allowed for literate and illiterate families to participate and did not limit respondents to specified topics. Administration of the survey was conducted during a routine clinic visit and required 20 to 60 minutes to complete. 3
Original Article TABLE 1. Patient Demographic Characteristics by Country Nicaragua (n=100)d Characteristics Age at diagnosis, y Mean 6 SD Median Range Sex Female Male TCAM use Yes No Primary respondent Self Mother Father Othera Race/ethnicity Mestizo Black White Otherb Diagnoses Leukemia Brain tumor Solid tumor Lymphoma Other Phase of treatmentc Recently diagnosedf Receiving treatment Recurred Palliative care Off treatment
Uruguay (n=100)
Mexico (n=100)e
Frequency % Frequency % Frequency %
7.3 6 4.7 6.3 0.1-21.4
7.4 6 5.4 6.4 0.0-21.8
7.0 6 4.8 4.8 0.5-17.4
56 42
56% 42%
42 58
42% 58%
53 47
53% 47%
79 21
79% 21%
76 24
76% 24%
77 23
77% 23%
2 70 10 12
2% 70% 10% 12%
16 65 16 3
16% 65% 16% 3%
7 78 6 8
7% 78% 6% 8%
97 2 1 0
97% 2% 1% 0%
41 2 50 7
41% 2% 50% 7%
99 0 0 1
99% 0% 0% 1%
63 4 16 8 9
63% 4% 16% 8% 9%
48 11 25 12 4
48% 11% 25% 12% 4%
65 9 19 6 1
65% 9% 19% 6% 1%
3 47 19 2 28
3% 48% 19% 2% 28%
12 38 20 0 30
12% 38% 20% 0% 30%
24 55 8 2 10
24% 55% 8% 2% 10%
Abbreviations: SD, standard deviation; TCAM, traditional complementary/alternative medicine. a Other respondent includes grandfather, grandmother, and guardian. b Other race includes Asian, >1 race, or unknown. c Phase of treatment at the time of survey administration. d Two cases were missing information regarding sex and 6 cases were missing information regarding the primary respondent. e One case was missing information regarding phase of treatment. f Within 3 months of diagnosis.
tive team. The survey was reviewed and analyzed to ensure that relevant TCAM indications were covered for the target region. Recommendations concerning modifications to the survey were collated and reviewed and then revised according to the consensus among the investigative team. Once the survey was modified, it was reviewed for readability and then field-tested by the members at the lead institution (data not shown). Internal consistency reliability was measured with the Cronbach a; a coefficient with a cutoff value of .70 was considered desirable. Convergent validity was assessed using a Spearman rank order correlation coefficient. Disattenuated convergent validity coefficients were also estimated to adjust for various levels of reliability in the construct measures used to establish convergence. The disattenuated correlation is obtained by dividing the observed correlation between 2 constructs by the square root of the product of the reliability estimates of the 2 constructs. Finally, test-retest reliability was measured in 20% of individuals who were retested by the same interviewer within 7 days of administration of the first survey. Agreeability among responses was measured with the intraclass correlation coefficient (ICC). An ICC value of 0.75 was considered excellent agreement. All analyses were performed using SPSS statistical software (version 20.0; IBM Corporation, Armonk, NY). A P value .05 was considered to be statistically significant. RESULTS Demographics
Data were collected from 300 participants at 3 pediatric oncology centers located in Nicaragua, Uruguay, and Mexico. Table 1 presents the demographic characteristics of the participants. Internal Consistency Reliability
Statistical Analysis
Demographic information was summarized using descriptive statistics. Frequencies and proportions were reported for categorical demographic variables. The mean, standard deviation, and range were reported for age, a continuous variable. Instrument Validation
The survey was designed so that its validity and reliability could be evaluated at each participating institution. Specifically, the survey evaluated content validity, internal consistency reliability, convergent validity, and test-retest reliability. Content validity was assessed by the investiga4
The internal consistency reliability of the scales is presented in Table 2. Items comprising the belief scale and the scale regarding cause due to TCAM demonstrated good internal consistency and reliability across the 3 countries (a range from. 77 to. 85 for the belief scale and a range from. 60 to. 86 for the cause scale). Although still within the acceptable range, the internal consistency reliability of the behavior scale was slightly lower than the other 2 scales in all 3 countries (a =. 69 in Nicaragua, a =. 60 in Uruguay, and a =. 60 in Mexico), which was largely due to the lack of variation in respondents’ answers to at least 2 behavioral items. For Nicaragua and Uruguay, internal consistency reliability was estimated based on Cancer
Month 00, 2014
Global TCAM in Children with Cancer/Ladas et al
TABLE 2. Internal Consistency Reliability of the TCAM Constructsa Nicaragua
Uruguay
Mexico; disattenuated correlation of 0.20 for Nicaragua, 0.16 for Uruguay, and 20.08 for Mexico).
Mexico
Test-Retest Reliability Construct
No. Reliability No. Reliability No. Reliability
TCAM beliefs Cause due to TCAM TCAM behaviors
95 95
0.80 0.86
83b 98
0.85 0.83
97 97
0.77 0.60
98
0.69
84b
0.60
98
0.60
Abbreviation: TCAM, traditional complementary/alternative medicine. a Internal consistency reliability was estimated using Cronbach a. b Uruguay had 17 patients as the primary survey respondent. Because of the adaptive design of the survey, these 17 patients skipped questions regarding parental beliefs and practices in TCAM and were not included in the internal consistency reliability calculation.
variances observed in only 5 items because no variations in response were observed in 2 behavioral items that asked whether a TCAM provider was consulted the last time the patient was not feeling well and whether a TCAM provider was consulted the last time the patient’s parent was not feeling well. For Mexico, one additional question, which asked about respondents’ regular visits to TCAM providers, also lacked variation. As a result, only 4 behavioral items were available to estimate internal consistency reliability in Mexico. Finally, we examined whether variations in reliability coefficients were observed based on the respondent (self, parental, or other family member). No significant differences among the correlation coefficients were observed (data not shown). Convergent Validity
Table 3 presents the observed and the disattenuated convergent validity coefficients between the 3 constructs. A moderate positive correlation between the belief and the behavior constructs was observed across 3 countries: Nicaragua (P =. 26 and P
Improving Our Understanding of the Use of Traditional Complementary/Alternative Medicine in Children With Cancer Elena J. Ladas, PHD, RD1,2; Meiko Lin, MS3; Federico Antillion, MD4; Silvia Rivas, MD4; Guillermo Chantada, MD5; Walter Cacciavillano, MD5; Roberta Ortiz, MD6; Katja Stein, PHD, RD7; Luis Castillo, MD8; Valeria Rocha, PhD8; Ligia Fu, MD9; Hilze Rodriquez, MD10; and Kara M. Kelly, MD1
BACKGROUND: Children with cancer in high-income and low-income countries often use traditional complementary/alternative medicine (TCAM). With efforts by the World Health Organization and international twinning programs improving access to conventional care for patients with childhood cancer, understanding the global use of TCAM is important because reliance on TCAM may affect time to presentation, adherence, and abandonment of care. In the current study, the authors describe the process and validation of an international survey documenting the use of TCAM among children with cancer. METHODS: The survey was designed to collect information on TCAM use and associated factors through both open-ended and close-ended questions. During the period between June 2012 and December 2013, the survey was administered to 300 children and adolescents (or their parents) who were undergoing treatment for cancer at a collaborating institution located in Mexico, Uruguay, and Nicaragua. RESULTS: For the majority of constructs, the survey demonstrated strong test-retest reliability as evidenced by an intraclass correlation of at least 0.79 in each of the participating countries. The survey demonstrated good internal consistency and reliability across countries (a range from. 77 to. 85 for the belief scale; and an a range from. 60 to. 86 for the cause scale) and convergent validity between TCAM beliefs and behavior constructs (adjusted correlation range, 0.35-0.60). CONCLUSIONS: The results of the current study demonstrate the successful development of a cross-cultural survey that produced results that were reliable and valid. These findings will aid investigators in providing guidelines concerning TCAM, support the development of education and research priorities, and identify variables associated C 2014 American Cancer Society. with TCAM that are region-specific. Cancer 2014;000:000-000. V KEYWORDS: pediatric oncology, complementary/alternative medicine, supportive care, survey design, twinning programs.
INTRODUCTION The differential mortality outcomes in childhood and adolescent cancer between developed and developing countries have become increasingly apparent. Greater than 80% of children with cancer reside in developing countries, in which survival rates are estimated to be approximately 35%, which is in stark contrast to the overall survival rate of 70% observed for children residing in developed countries.1,2 Although improved access to drugs, political stability, and the availability of locally trained health care providers have been crucial in expanding the delivery of Western care, the role of traditional complementary/alternative medicine (TCAM) in regions with diverse medical systems is less understood. Understanding the application and role of TCAM in other pediatric conditions, including human immunodeficiency virus/acquired immunodeficiency syndrome and severe malaria, has helped to bridge the gap in communication among traditional healers and conventional health care providers and improve access to medical care in children with these conditions.3-5 To date, the impact of TCAM on the quality of care and treatment-related outcomes of childhood and adolescent cancer is not well-understood. TCAM are therapies that are outside the context of conventional medical care and include modalities such as acupuncture, massage, energy medicine, and herbal remedies. International surveys have reported that the use of TCAM is high among children with cancer, with the prevalence of use exceeding 60%.6 However, a validated survey that is crossCorresponding author: Elena J. Ladas, PhD, RD, Division of Pediatric Hematology, Oncology, and Stem Cell Transplant, Institute of Human Nutrition, Columbia University Medical Center, 7th Fl, 161 Fort Washington Ave, New York, NY 10032; Fax: (212) 305-5848; [email protected] 1 Division of Pediatric Hematology, Oncology, and Stem Cell Transplant, Columbia University Medical Center, New York, New York; 2Institute of Human Nutrition, Columbia University Medical Center, New York, New York; 3Interdisciplinary Studies Program, Teachers College, Columbia University, New York, New York; 4Hemato-Oncology Service, National Pediatric Oncology Unit, Guatemala City, Guatemala; 5JP Garrahan Hospital, Buenos Aires, Argentina; 6Manuel de Jesus Rivera La Mascota Children’s Hospital, Managua, Nicaragua; 7Dr. Juan I. Menchaca Civil Hospital of Guadalajara, Jalisco, Mexico; 8Pereira Rossell Medical Center, Montevideo, Uruguay; 9Escuela University Hospital, Tegucigalpa, Honduras; 10Panama City, Republic of Panama
We thank the Tamarind Foundation for its support of this international effort. The results of the current study demonstrate the successful development of a cross-cultural survey exploring the use of traditional complementary/alternative medicine (TCAM) in children with cancer. These findings will aid investigators in providing guidelines concerning TCAM, support the development of education and research priorities, and identify variables associated with TCAM that are region-specific. DOI: 10.1002/cancer.29212, Received: September 29, 2014; Revised: October 28, 2014; Accepted: October 29, 2014, Published online Month 00, 2014 in Wiley Online Library (wileyonlinelibrary.com)
Cancer
Month 00, 2014
1
Original Article
culturally adaptable has not been described for children with cancer.6 The current study describes the development and testing of a survey documenting the use, reasons, and predictors of TCAM among children with cancer. We describe several phases of the process and provide the reader with an iterative methodology for the design and validation of an international survey exploring TCAM in a conventional medical setting. MATERIALS AND METHODS During the period of June 2012 to December 2013, the described survey was administered to 300 children and adolescents (or their parents) who were undergoing treatment for cancer at a collaborating institution located in Guadalajara, Mexico; Montevideo, Uruguay; and Managua, Nicaragua. Approval was obtained from the Institutional Review Boards at Columbia University Medical Center and those of each participating institution. Eligible participants were identified by a computer-generated block randomization procedure and approached about participation during a routine visit to the outpatient center. After consent and assent (when applicable) was obtained, the survey was then administered face-to-face by a trained interviewer who was from the local region and was not involved in the care of the child. Instrument Development
The survey tool was expanded and modified from a previous survey that was administered to children and adolescents undergoing treatment for cancer at Columbia University Medical Center, the lead institution.7 Participating institutions involved in the development and distribution of the survey are centers with a pediatric oncology division and are located in North, Central, or South America. A select group of centers representing the United States, Guatemala, and Argentina were involved in the development of the survey tool and represent pediatric oncologists and clinical investigators with expertise in TCAM, survey design, and validation. Individuals with expertise in medicine, nutrition, public health, CAM, psychology, and international outreach programs provided additional consultation throughout the development process. This process was divided into 5 phases that included: 1) purpose and domain specification; 2) development of survey specifications; 3) identification of culturally appropriate TCAM classification; 4) content validation; and 5) empirical validation. Figure 1 provides an overview of the development process.8 2
Theoretical framework
The survey was designed to identify the perceived benefits (health promotion, cure, etc), susceptibility (development or management of a side effect or late effect), strength of beliefs in TCAM theories, and expected outcomes (risk of death or disease recurrence) of using TCAM as a component of cancer care. Constructs of the Health Belief Model and the Theory of Reasoned Action and the Theory of Planned Behavior provided the theoretical foundation of the survey design and development. Construct definition
The survey instrument was designed to examine 3 constructs related to TCAM use: 1) beliefs in TCAM theories and its perceived efficacy; 2) behaviors related to TCAM; and 3) causes of cancer or symptoms related to TCAM theories. A total of 24 items were developed according to the domain specifications. The items were screened for match with the domain indicators, difficult vocabulary, double-barreled language, or ambiguous content. The breakdown of the items by construct is as follows: 1) Belief scale: 9 items collecting information regarding belief in TCAM theories and perceived efficacy. 2) Behavior scale: 7 items related to patients’ TCAM therapy-seeking and use behaviors. 3) Cause due to TCAM scale (cancer or symptoms due to TCAM-related theories): 8 items related to patients’ perceived causes of symptoms and cancer. Both the belief and cause scales are scored on a 5point Likert scale with responses ranging from 1 (strongly agree) to 5 (strongly disagree). The behavior scale is scored on a dichotomous scale (0 indicates no and 1 indicates yes). For the belief scale and the behavior scale, 4 questions addressed parental beliefs and practices in TCAM for their personal medical care. Survey format
The survey was classified into 5 sections: 1) health behaviors; 2) health beliefs; 3) TCAM modalities; 4) demographic data; and 5) medical record abstraction. The first 3 sections were separated by construct with the sequencing structure of the questions designed to initiate trust and rapport between the interviewer and respondent before discussing the construct of interest. The survey was designed with both open-ended and closed-ended questions. The closed-ended questions allow for the data to be collapsed or expanded into categories, to evaluate variations in the frequency of TCAM by region, and to allow for administration to a large sample. The open-ended questions provide more descriptive details regarding the Cancer
Month 00, 2014
Global TCAM in Children with Cancer/Ladas et al
Figure 1. An iterative process for designing and validating traditional complementary/alternative medicine (TCAM) measures is shown.
participant’s experience with TCAM therapy and allow for responses to be thematically categorized and analyzed. A protocol was developed to guide transcription and coding procedures using a generic qualitative approach. Language
Participating institutions include many countries in North, Central, and South America in which Spanish is the official language. Different local dialects and words among localities often indicate the same concept; therefore, several steps were taken to ensure the survey was culturally and linguistically relevant. The survey was first translated from English into Spanish from the original version administered in North America. The Spanish version was reviewed by individuals and professionals at the local hospital to ensure clarity and participant comprehension. Modifications were indicated in Spanish, and surveys were then back-translated into English. CompariCancer
Month 00, 2014
son of the 2 English versions was evaluated for agreeability and discrepancies. All identified variations were reviewed by an independent translator familiar with the dialect of the intended country and modified if indicated. Survey comprehension was ensured by administering the translated survey in the local language to a small number of patients and members of the staff. Administration of the survey
The survey was designed to be administered by an interviewer who was not involved in the respondent’s cancer care to ensure respondents felt comfortable discussing TCAM use without feeling that their medical care would be compromised. This format also allowed for literate and illiterate families to participate and did not limit respondents to specified topics. Administration of the survey was conducted during a routine clinic visit and required 20 to 60 minutes to complete. 3
Original Article TABLE 1. Patient Demographic Characteristics by Country Nicaragua (n=100)d Characteristics Age at diagnosis, y Mean 6 SD Median Range Sex Female Male TCAM use Yes No Primary respondent Self Mother Father Othera Race/ethnicity Mestizo Black White Otherb Diagnoses Leukemia Brain tumor Solid tumor Lymphoma Other Phase of treatmentc Recently diagnosedf Receiving treatment Recurred Palliative care Off treatment
Uruguay (n=100)
Mexico (n=100)e
Frequency % Frequency % Frequency %
7.3 6 4.7 6.3 0.1-21.4
7.4 6 5.4 6.4 0.0-21.8
7.0 6 4.8 4.8 0.5-17.4
56 42
56% 42%
42 58
42% 58%
53 47
53% 47%
79 21
79% 21%
76 24
76% 24%
77 23
77% 23%
2 70 10 12
2% 70% 10% 12%
16 65 16 3
16% 65% 16% 3%
7 78 6 8
7% 78% 6% 8%
97 2 1 0
97% 2% 1% 0%
41 2 50 7
41% 2% 50% 7%
99 0 0 1
99% 0% 0% 1%
63 4 16 8 9
63% 4% 16% 8% 9%
48 11 25 12 4
48% 11% 25% 12% 4%
65 9 19 6 1
65% 9% 19% 6% 1%
3 47 19 2 28
3% 48% 19% 2% 28%
12 38 20 0 30
12% 38% 20% 0% 30%
24 55 8 2 10
24% 55% 8% 2% 10%
Abbreviations: SD, standard deviation; TCAM, traditional complementary/alternative medicine. a Other respondent includes grandfather, grandmother, and guardian. b Other race includes Asian, >1 race, or unknown. c Phase of treatment at the time of survey administration. d Two cases were missing information regarding sex and 6 cases were missing information regarding the primary respondent. e One case was missing information regarding phase of treatment. f Within 3 months of diagnosis.
tive team. The survey was reviewed and analyzed to ensure that relevant TCAM indications were covered for the target region. Recommendations concerning modifications to the survey were collated and reviewed and then revised according to the consensus among the investigative team. Once the survey was modified, it was reviewed for readability and then field-tested by the members at the lead institution (data not shown). Internal consistency reliability was measured with the Cronbach a; a coefficient with a cutoff value of .70 was considered desirable. Convergent validity was assessed using a Spearman rank order correlation coefficient. Disattenuated convergent validity coefficients were also estimated to adjust for various levels of reliability in the construct measures used to establish convergence. The disattenuated correlation is obtained by dividing the observed correlation between 2 constructs by the square root of the product of the reliability estimates of the 2 constructs. Finally, test-retest reliability was measured in 20% of individuals who were retested by the same interviewer within 7 days of administration of the first survey. Agreeability among responses was measured with the intraclass correlation coefficient (ICC). An ICC value of 0.75 was considered excellent agreement. All analyses were performed using SPSS statistical software (version 20.0; IBM Corporation, Armonk, NY). A P value .05 was considered to be statistically significant. RESULTS Demographics
Data were collected from 300 participants at 3 pediatric oncology centers located in Nicaragua, Uruguay, and Mexico. Table 1 presents the demographic characteristics of the participants. Internal Consistency Reliability
Statistical Analysis
Demographic information was summarized using descriptive statistics. Frequencies and proportions were reported for categorical demographic variables. The mean, standard deviation, and range were reported for age, a continuous variable. Instrument Validation
The survey was designed so that its validity and reliability could be evaluated at each participating institution. Specifically, the survey evaluated content validity, internal consistency reliability, convergent validity, and test-retest reliability. Content validity was assessed by the investiga4
The internal consistency reliability of the scales is presented in Table 2. Items comprising the belief scale and the scale regarding cause due to TCAM demonstrated good internal consistency and reliability across the 3 countries (a range from. 77 to. 85 for the belief scale and a range from. 60 to. 86 for the cause scale). Although still within the acceptable range, the internal consistency reliability of the behavior scale was slightly lower than the other 2 scales in all 3 countries (a =. 69 in Nicaragua, a =. 60 in Uruguay, and a =. 60 in Mexico), which was largely due to the lack of variation in respondents’ answers to at least 2 behavioral items. For Nicaragua and Uruguay, internal consistency reliability was estimated based on Cancer
Month 00, 2014
Global TCAM in Children with Cancer/Ladas et al
TABLE 2. Internal Consistency Reliability of the TCAM Constructsa Nicaragua
Uruguay
Mexico; disattenuated correlation of 0.20 for Nicaragua, 0.16 for Uruguay, and 20.08 for Mexico).
Mexico
Test-Retest Reliability Construct
No. Reliability No. Reliability No. Reliability
TCAM beliefs Cause due to TCAM TCAM behaviors
95 95
0.80 0.86
83b 98
0.85 0.83
97 97
0.77 0.60
98
0.69
84b
0.60
98
0.60
Abbreviation: TCAM, traditional complementary/alternative medicine. a Internal consistency reliability was estimated using Cronbach a. b Uruguay had 17 patients as the primary survey respondent. Because of the adaptive design of the survey, these 17 patients skipped questions regarding parental beliefs and practices in TCAM and were not included in the internal consistency reliability calculation.
variances observed in only 5 items because no variations in response were observed in 2 behavioral items that asked whether a TCAM provider was consulted the last time the patient was not feeling well and whether a TCAM provider was consulted the last time the patient’s parent was not feeling well. For Mexico, one additional question, which asked about respondents’ regular visits to TCAM providers, also lacked variation. As a result, only 4 behavioral items were available to estimate internal consistency reliability in Mexico. Finally, we examined whether variations in reliability coefficients were observed based on the respondent (self, parental, or other family member). No significant differences among the correlation coefficients were observed (data not shown). Convergent Validity
Table 3 presents the observed and the disattenuated convergent validity coefficients between the 3 constructs. A moderate positive correlation between the belief and the behavior constructs was observed across 3 countries: Nicaragua (P =. 26 and P
