Abstract
Background
Little is known about the specific information parents of children with cancer search for online. Understanding the content of parents’ searches over time could offer insight into what matters most to parents and identify knowledge gaps that could inform more comprehensive approaches to family education and support.
Methods
We describe parents’ health-related Google searches starting six months before cancer diagnosis and extending through the date of study enrollment which was at least one month after initiating cancer treatment. Searches were obtained retrospectively and grouped into health-related and non-health-related categories. The median time to parent enrollment from date of cancer diagnosis was 264 days.
Results
Parents searched for health-related topics more frequently than the general population (13% vs. 5%). Health-related searches increased in the months preceding the child’s cancer diagnosis and most commonly pertained to symptoms and logistics, “directions to hospital”. Health-related search volume peaked about a month after cancer diagnosis when general health-related searches were present in addition to cancer-specific searches. Eighteen percent of health-related searches were cancer-specific and of these cancer-specific searches, 54% pertained to support, for example “cancer quote for son.”
Conclusions
Google search content offers insight into what matters to parents of cancer patients. Understanding search content could inform more comprehensive approaches to family education and support initiatives.
Keywords: pediatric oncology, supportive care, digital health, Google, Internet, quantified-self research
Introduction
For a parent, learning their child has cancer can be overwhelming, stressful, and debilitating. After a new cancer diagnosis, parents seek social support and information from multiple sources including healthcare providers and the Internet.1–3 While it has been shown that cancer searches online correlate to cancer prevalence at a population level,4,5 little is known about the specific, granular information parents search for online. Understanding the content of parents’ searches over time could offer insight into what matters most to parents and identify knowledge gaps that could inform more comprehensive approaches to family education and support. Google search histories provide a unique opportunity to describe the temporal relationship between the information-seeking needs of parents and their child’s cancer diagnosis. We sought to describe the content of health-related Google searches performed by parents of pediatric oncology patients both prior to the child’s cancer diagnosis and during cancer treatment.
Methods
This retrospective pilot study used a convenience sample of parents of pediatric oncology patients with a clinic visit or hospitalization at a tertiary care center, the Children’s Hospital of Philadelphia (CHOP), between June-August 2017. It was approved by the CHOP Institutional Review Board, #17–013821. English-fluent parents of patients who had completed at least one month of treatment and were less than six months removed from therapy completion were eligible if they consented to share their Google search history. Parents of patients who had a cancer relapse were eligible for study inclusion.
The text of Google queries participants entered into the search bar were retrospectively extracted from their Google account originating six months prior to cancer diagnosis through the enrollment date which was at least one month into cancer treatment. If the patient had relapsed, the search extraction included all searches dating back to six months prior to the original diagnosis. The broad inclusion criteria resulted in some parents having much longer time periods of searches included. Therefore, during analysis of search frequency, time periods were grouped into six-month intervals and only time periods with at least one third of the participants represented were included. This corresponded to the first two years after diagnosis.
For analysis, the timing of searches was standardized relative to the date of original cancer diagnosis. One of two reviewers (AH, CP) independently coded Google searches into health-related and non-health-related categories (Table 1). These categories were derived from prior work in a non-oncologic adult population that classified Google searches into health-related and non-health-related categories.6 The authors revised these previous Google search categorizations a priori to include cancer-specific categories. For example, the category of “Drug [medication] use” was split into the categories of “Chemotherapy or immunotherapy medication” and “Non-chemotherapy medication.” Health-related searches that explicitly mentioned cancer such as “symptoms of leukemia,” were coded as cancer-specific. A purposefully broad cancer-specific category was “cancer support.” This category encompassed social and emotional supports for the parents themselves as well as for their children (Table 1). Thirteen percent of searches were coded by both reviewers. Kappa coefficient was calculated for agreement regarding a search being health-related or not (kappa= .93) as well as cancer-specific or not (kappa= .83). The range of kappa coefficients for health-related categories was .6-.9 while the range for non-health-related categories was .5 - .9. A third reviewer (SH) adjudicated coding disagreements within health-related searches.
Table 1.
Characterizing Google search queries
| Query Category | Query Group | Example Google Search Query |
|---|---|---|
| 1 - Entertainment | General | “concerts,” “sports,” “movies,” “celebrities” |
| 2 - Directions to non-clinical locations | General | “current Location -> non-clinical location”, “directions to grocery store” |
| 3 - Retail (online shopping, physical store location) | General | “Amazon,” “Walmart,” “Macy’s in Memphis” |
| 4 - Social Media / apps | General | “Facebook,” “aol mail,” “myverizon,” “app game” |
| 5 - Food / drink | General | “restaurant,” “how long to cook pizza,” “recipe” |
| 6 - Financial and insurance excluding health insurance | General | “how to reach IRS,” “auto insurance,” “bank”. |
| 7 - Legal questions | General | “what are the traffic laws when no stop sign,” “how to apply for child support” |
| 8 - Unrelated / unintelligible | General | “askldfhj,” “22234” |
| 9 - Other | General | “how to clean windows,” “sandals,” “John Smith,” “weed,” “porn” |
| 10 - Information about hospital/ care sites /pharmacy | Health-related | “dentist,” “directions to children’s hospital,” “CVS hours,” “Dr. John Smith”, “hospital scheduling department” |
| 11 - Health insurance | Health-related | “healthcare.gov” “does my insurance cover surgery” |
| 11.1 - Prescription coverage | Health-related | “express scripts,” “how many days do you have to wait with insurance to get prescription refilled,” “sleep aid without prescription” |
| 12 - Symptoms, disease and medical information | Health-related | “cramping 12 weeks agony,” “what is a platelet,” “vitamin d deficiency,” “define ataxia” |
| 12.1 - Cancer-specific symptoms, disease and medical information | Cancer-specific | “symptoms of leukemia,” “can cancer cause a rash,” “headache in neuroblastoma patient,” “tongue cancer” |
| 13 - Treatment and disease management | Health-related | “baby constipation relief,” “nosebleed treatment,” “hangover remedy” |
| 13.1 - Cancer-specific treatment and disease management | Cancer-specific | “how to treat leukemia,” “when does a cancer patient need radiation,” “home remedies for sore throat for cancer patient” |
| 14 - Non-chemotherapy medication | Health-related | “how many mgs of Advil to take,” “side effects,” “zantac” |
| 15 - Chemotherapy or immunotherapy medication | Cancer-specific | “vincristine” “side effects of CART,” “methotrexate toxicity testing” |
| 16 - Medical devices and accessories | Health-related | “Syringe,” “casts,” “wheelchair” |
| 17 - Healthy lifestyle | Health-related | “healthy weight loss,” “crossfit gym,” “healthy recipe” |
| 17.1 - Cancer-specific healthy lifestyle | Cancer-specific | “good foods to eat when you have cancer” |
| 18 - Cancer support | Cancer-specific | “Cancer charities,” “Make-A-Wish,” “leukemia ribbon color,” “cancer poem,” “cancer quotes for son,” “beads of courage for siblings” |
Google search queries were divided into one of eighteen categories that were developed by the research team. Categories were grouped and designated as pertaining to general, health-related, or cancer-specific searches. Example searches for each of the categories are shown in the right-hand column.
Results
One hundred three parents were approached with 52 having Google accounts and 21 consenting to share their search history. For the 21 participants, the mean age was 41 years (Table 2). Seventeen of the participants (81%) were the patient’s mother, fifteen (71%) were Caucasian, and twelve (57%) had a household income less than $50,000 annually while three (14%) had a household income greater than $200,000 annually. The oncologic diagnoses for their children included nine hematologic malignancies (43%), eight solid organ malignancies (38%), and four neurologic malignancies (19%). The median time to enrollment from date of cancer diagnosis was 264 days. These twenty-one parents had a total of 81,725 searches (range 15–23980 per person, interquartile range (IQR) 875–4629). Health-related searches comprised 13% (10,905/81,725) of the sample. Per parent range of health-related searches was 0–3662 (IQR 95–636).
Table 2.
Demographic characteristics of participating parents
| Demographic Characteristic | Number (%) | |
|---|---|---|
| Age (mean) | 41 | -- |
| Relation to patient | ||
| Mother | 17 (81) | |
| Father | 4 (19) | |
| Race | ||
| White | 15 (71) | |
| Black | 2 (10) | |
| Asian | 1 (5) | |
| Other | 1 (5) | |
| Did not disclose | 2 (10) | |
| Annual household income | ||
| <$50,000 | 12 (57) | |
| $50,000–200,000 | 6 (29) | |
| >$200,000 | 3 (14) | |
| Cancer diagnosis of child | ||
| Hematologic | 9 (43) | |
| Solid organ | 8 (38) | |
| Neurologic | 4 (19) |
Demographic information was self-reported by the participating parent.
The most common health-related search category was “Symptoms, disease and medical information” (31%, 3,418/10,905) followed by the logistical category, “Information about hospital/ care sites/ pharmacy” (29%, 3,210/10,905) (summary for general and cancer-specific searches is shown in Table 3). No other category accounted for more than 10% of the health-related searches. Other relatively common health-related search categories included “Medications” (8%, 920/10,905) and “Treatment and disease management” (6%, 608/10,905). “Healthy lifestyle” searches were also relatively common (9%, 942/10,905) and very few of these, 24, were cancer-specific.
Table 3.
Health-related and cancer-specific searches before and after cancer diagnosis
| Query Category | Before Diagnosis | After Diagnosis | Total |
|---|---|---|---|
| General Health-related searches | n, (%) | n, (%) | n, (%) |
| Information for hospital/care sites/pharmacy | 404 (31%) | 2,806 (37%) | 3,210 (36%) |
| Symptoms, disease & medical information | 485 (37%) | 2,477 (32%) | 2,962 (33%) |
| Healthy lifestyle | 147 (11%) | 772 (10%) | 919 (10%) |
| Medications | 101 (8%) | 594 (8%) | 695 (8%) |
| Medical devices and accessories | 50 (4%) | 538 (7%) | 588 (7%) |
| Treatment and disease management | 87 (7%) | 349 (5%) | 436 (5%) |
| Health insurance and prescription coverage | 27 (2%) | 147 (2%) | 174 (2%) |
| Subtotal | 1,301 (100%) | 7,683 (100%) | 8,984 (100%) |
| Cancer-specific searches | n, (%) | n, (%) | n, (%) |
| Cancer support | 23 (20%) | 1021 (57%) | 1,044 (54%) |
| Cancer-specific symptoms, disease and medical information | 84 (72%) | 372 (21%) | 456 (24%) |
| Chemo or immunotherapy medication | 1 (1%) | 224 (12%) | 225 (12%) |
| Cancer-specific treatment and disease management | 8 (7%) | 164 (9%) | 172 (9%) |
| Cancer-specific healthy lifestyle | 0 (0%) | 24 (1%) | 24 (1%) |
| Subtotal | 116 (100%) | 1,805 (100%) | 1,921 (100%) |
| Combined Total | 1,417 | 9,488 | 10,905 |
Cancer-specific searches comprised 18% (1,921/10,905) of health searches. “Cancer support” was the most common cancer-specific category (54%, 1,044/1,921) (Table 3). Other common cancer-specific categories were “Cancer-specific symptoms, disease and medical information” (24%, 456/1,921), and “Chemotherapy or immunotherapy medication” (12%, 225/1,921).
The date range for health-related searches was 182 days (six months) prior to diagnosis through 4371 days post-diagnosis while the range of cancer-specific searches was 166 days before diagnosis through 4364 days post diagnosis. Only three persons performed searches >2 years post-diagnosis and these searches (4410) were excluded from search rate analysis. Starting six months prior to diagnosis, the typical parent searched for about 0.2 health-related searches per day (Figure 1). Health-related searches increased prior to diagnosis (Figure 1) and consisted primarily of symptom, logistical, and treatment searches (health-related searches with cancer-specific searches excluded is shown Figure 2). Insurance, medication and healthy lifestyle searches were also noted during the pre-diagnosis period. Search volumes peaked at approximately 1.1 searches per person per day about one-month post-diagnosis when general health searches remained common and new, cancer-specific queries were also present. Parents typically searched for health-related terms sporadically bursts of search activity on a given day followed by multiple days with no health-related searches.
Figure 1.
Health-related Google searches over time relative to the date of cancer diagnosis Health-related Google searches performed by parents of children with cancer are shown relative to the date of diagnosis (Day 0) as determined by chart review. The number of searches is standardized to mean number of searches per day per person. Only six persons performed searches >2 years post-diagnosis (4415 searches) were excluded from search rate analysis. Starting six months prior to diagnosis (Day −183), the typical parent searched for about 0.2 health-related searches per day. The mean number of searches per parent per day increased steadily through the date of diagnosis and peaked at approximately 1.2 searches per person per day shortly after diagnosis. The number of health-related searches decreased with time and was continuing to slowly downtrend at two years after diagnosis.
Figure 2.
General health-related Google searches relative to the date of cancer diagnosis General health-related Google searches are shown by subcategory (x-axis) from six months prior to diagnosis through two years after diagnosis (y-axis). Parents primarily search for symptoms and logistical information prior to a cancer diagnosis. Searches for health insurance cluster around the time of cancer diagnosis.
The frequency of health-related searches during the first two years after diagnosis was assessed based upon the type of tumor of the patient. Parents of children with solid organ malignancies averaged 1.1 health-related search per day while parents of children with neurologic and hematologic malignancies averaged 0.8 and 0.9 health-related searches per day, respectively.
Regarding cancer-specific searches from six months prior to diagnosis through two years after diagnosis (Figure 3), only 27 (2%) were performed more than 6 days prior to the child’s cancer diagnosis. Cancer-specific searches increased in the week leading up to diagnosis, with 7% (89/1,241) cancer-specific searches occurring during this time period. A large number of cancer-specific searches occurred shortly after diagnosis with 8% (100/1,241) occurring in the following seven days and 31% (382/1,241) occurring in the following 60 days. The number of health-related and cancer-specific searches decreased relative to their peak after diagnosis and subsequently remained relatively stable over time during the two years after diagnosis (Figure 1).
Figure 3.
Cancer-specific Google searches relative to the date of cancer diagnosis Cancer-specific Google searches are shown by subcategory (x-axis) from six months prior to diagnosis through two years after diagnosis (y-axis). Cancer-specific searches are most common shortly after diagnosis. The most common cancer-specific search category was “cancer support.”
Discussion
Overall, health-related searches comprised 13% of the total Google-search sample. This is higher than Google’s self-reported baseline rate in the general population, 5%,7 as well as the baseline rate reported in a study of adults presenting to the emergency room, 6%.6 Parents and caregivers have long reported using the internet and a source of cancer information.1,3,8–11 This study adds to this literature by exploring search data at a much more granular level and links search patterns relative to the cancer diagnosis.
At the present time, our results are most applicable for family support and education. Given the peak of internet use near the time of diagnosis, educational interventions to improve parents’ ability to navigate the internet for cancer information should be considered. Supportive care and logistic-related health searches were extremely common and health insurance searches also present, although to a lesser extent. Collectively, these searches suggest parents utilize the internet to help with the day to day fiscal and logistical issues that arise from having a sick child. Parents’ desire for logistical information has not previously been reported. It represents a measurable and potentially modifiable domain through interventions like website design and educational materials. This study is descriptive and cannot establish causality for the high prevalence of supportive care and logistical searches. It may be that they are searched because parents do not feel their providers discuss these issues in appropriate depth. Alternatively, because these types of searches require less medical sophistication to interpret, parents may feel more comfortable searching for them online compared to medical information, or parents may simply forget to ask questions about these topics when medical providers are present and subsequently search for the supportive care information at a later time.
The increase in symptom searches pre-diagnosis suggests a signal that may be relevant for early cancer detection, a concept previously demonstrated in pancreatic cancer.12 Other potential applications for symptom data include adverse event monitoring which has been shown to be feasible in social media data including Twitter.13 However, caution must be used when considering Google data for this purpose as substantial scientific,14 ethical and privacy concerns must be considered.
Limitations for this study include a small sample size from a single institution and the nature of Google search data as searches may not pertain to the patient or could have been performed by a different person using the parent’s account. Searches from multiple devices including computers, phones, and tablets were captured if the user was logged into their Google account on that device. However, searches performed when logged out of Google accounts were not captured and theoretically searches from different devices could have disproportionate representation. Unless parents purposefully logged out of their account to perform searches that were sensitive in nature, our data should be a representative sample of their Google searches. Due to the broad timeframe allowed under the inclusion criteria, the timeframe varied between participants depending upon how long the child was in treatment prior to study enrollment. This limitation was addressed during search rate analysis by eliminating the time period greater than two years after diagnosis when fewer parents had eligible searches.
Future directions of this research need to include larger sample size as it is possible different cancer types will have unique information seeking needs. Additionally, this study did not address adolescent or young adult online information seeking and future studies should include this group as they may have different information needs compared to their parents.
Google search content offers insight into what matters to parents of cancer patients. More research is needed to explore use of Google to obtain health-related information and utilize this to inform future education, quality, and research initiatives and better understand how internet use influences healthcare decision making.
Acknowledgements
The authors would like to acknowledge Sara Handley, MD for her time and effort while serving as a reviewer for discrepant Google searches.
Raina Merchant received funding from the Robert Wood Johnson Foundation Pioneer Award number 72695 and Charles Phillips was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under Award Number T32HD060550.
Abbreviations Table Key:
- CHOP
Children’s Hospital of Philadelphia
- IQR
Interquartile range
Footnotes
Conflict of Interest
The authors have no conflicts of interest.
Data availability
The data that support the findings of this study are not publicly available due to privacy or ethical restrictions.
References
- 1.Dominguez M, Sapina L. Pediatric cancer and the internet: exploring the gap in doctor-parents communication. Journal of cancer education : the official journal of the American Association for Cancer Education 2015;30(1):145–151. [DOI] [PubMed] [Google Scholar]
- 2.Kilicarslan-Toruner E, Akgun-Citak E. Information-seeking behaviours and decision-making process of parents of children with cancer. European journal of oncology nursing : the official journal of European Oncology Nursing Society 2013;17(2):176–183. [DOI] [PubMed] [Google Scholar]
- 3.Gage EA, Panagakis C. The devil you know: parents seeking information online for paediatric cancer. Sociology of health & illness 2012;34(3):444–458. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Phillips CA, Barz Leahy A, Li Y, Schapira MM, Bailey LC, Merchant RM. Relationship Between State-Level Google Online Search Volume and Cancer Incidence in the United States: Retrospective Study. Journal of medical Internet research 2018;20(1):e6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wehner MR, Nead KT, Linos E. Correlation Among Cancer Incidence and Mortality Rates and Internet Searches in the United States. JAMA dermatology 2017. [DOI] [PMC free article] [PubMed]
- 6.Asch JM, Asch DA, Klinger EV, Marks J, Sadek N, Merchant RM. Google search histories of patients presenting to an emergency department: an observational study. BMJ open 2019;9(2):e024791. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Ramaswami P A remedy for your health-related questions: health info in the Knowledge Graph Google: official blog 2015; https://googleblog.blogspot.com/2015/02/health-info-knowledge-graph.html. Accessed September 12, 2018.
- 8.Arora NK, Hesse BW, Rimer BK, Viswanath K, Clayman ML, Croyle RT. Frustrated and confused: the American public rates its cancer-related information-seeking experiences. Journal of general internal medicine 2008;23(3):223–228. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Cooper CP, Mallon KP, Leadbetter S, Pollack LA, Peipins LA. Cancer Internet Search Activity on a Major Search Engine, United States 2001–2003. Journal of medical Internet research 2005;7(3):e36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Diaz JA, Griffith RA, Ng JJ, Reinert SE, Friedmann PD, Moulton AW. Patients’ use of the Internet for medical information. Journal of general internal medicine 2002;17(3):180–185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Foroughi F, Lam AK-Y, Lim MSC, Saremi N, Ahmadvand A. “Googling” for Cancer: An Infodemiological Assessment of Online Search Interests in Australia, Canada, New Zealand, the United Kingdom, and the United States. JMIR Cancer 2016;2(1):e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Paparrizos J, White RW, Horvitz E. Screening for Pancreatic Adenocarcinoma Using Signals From Web Search Logs: Feasibility Study and Results. Journal of oncology practice 2016;12(8):737–744. [DOI] [PubMed] [Google Scholar]
- 13.Smith K, Golder S, Sarker A, Loke Y, O’Connor K, Gonzalez-Hernandez G. Methods to Compare Adverse Events in Twitter to FAERS, Drug Information Databases, and Systematic Reviews: Proof of Concept with Adalimumab. Drug safety 2018;41(12):1397–1410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Caster O, Dietrich J, Kurzinger ML, et al. Assessment of the Utility of Social Media for Broad-Ranging Statistical Signal Detection in Pharmacovigilance: Results from the WEB-RADR Project. Drug safety 2018;41(12):1355–1369. [DOI] [PMC free article] [PubMed] [Google Scholar]