CLINICAL STUDY

Know Your Market: Use of Online Query Tools to Quantify Trends in Patient Information-seeking Behavior for Varicose Vein Treatment Asheesh K. Harsha, MD, J. Eric Schmitt, MD, PhD, and S. William Stavropoulos, MD

ABSTRACT Purpose: To analyze Internet search data to characterize the temporal and geographic interest of Internet users in the United States in varicose vein treatment. Materials and Methods: From January 1, 2004, to September 1, 2012, the Google Trends tool was used to analyze query data for “varicose vein treatment” to identify individuals seeking treatment information for varicose veins. The term “varicose vein treatment” returned a search volume index (SVI), representing the search frequency relative to the total search volume during a specific time interval and region. Linear regression analysis and Kruskal-Wallis one-way analysis of variance were employed to characterize search results. Results: Search traffic for varicose vein treatment increased by 520% over the 104-month study period. There was an annual mean increase of 28% (range, 18%–100%; standard deviation [SD], 35%), with a statistically significant linear increase in average yearly SVI over time (R2 ¼ 0.94, P o .0001). All years showed positive growth in mean SVI except for 2008 (18% decrease). There were statistically significant differences in SVI by month (Kruskal-Wallis, P o .0001) with significantly higher mean SVI compared with other months in May (190% increase; range, 26%–670%; SD, 15%) and June (209% increase; range, 35%–700%; SD, 20%). The southern United States showed significantly higher search traffic than all other regions (TukeyKramer, P o .00001). Conclusions: There have been significant increases in Internet search traffic related to varicose vein treatment in the past 8 years. Reflected in this trend is an annual peak in search traffic in the late spring months with an overall geographic bias toward southern states. Rigorous analysis of Internet search queries for medical procedures may prove useful to guide the efficient use of limited resources and marketing dollars.

ABBREVIATIONS ANOVA = analysis of variance, GIS = Google Insights for Search, SD = standard deviation, SVI = search volume index

The Internet has become an important primary source of medical information (1,2) for patients; 4 184 million From the Departments of Radiology (A.K.H., J.E.S.) and Interventional Radiology (S.W.S.), Hospital of the University of Pennsylvania, 3400 Spruce Street, Philadelphia, PA 19104. Received July 17, 2013; final revision received September 12, 2013; accepted September 13, 2013. Address correspondence to A.K.H.; E-mail: [email protected] From the SIR 2013 Annual Meeting. S.W.S. has research funded by B. Braun, Gore, and Bard PV, is a paid consultant for Cook, and is a paid research consultant for Bard PV. Neither of the other authors has identified a conflict of interest. & SIR, 2014 J Vasc Interv Radiol 2014; 25:53–57 http://dx.doi.org/10.1016/j.jvir.2013.09.015

U.S. residents are regular Internet users, and a large percentage (70%–80% of U.S. adults) use the Internet to research information on their health conditions (1,4,5). This process of health information seeking online is not only a form of self-reporting medical illness but also has been shown to represent an indirect measure of disease prevalence, as has been shown for kidney stone incidence (6–8) and influenza outbreaks (9,10). In the era of directto-consumer Internet advertising for services including elective medical treatments, patients are increasingly being exposed to a wide variety of treatment options and treatment providers while investigating their medical condition. Patients who use the Internet for health information are 60% more likely to contact a health provider than patients who do not use the Internet (4).

54



Patient Information-seeking for Varicose Vein Treatment

Treatment of symptomatic varicose veins is an area of elective interventional practice in which patients elect to proceed with treatment and choose their provider. Varicose veins affect 23% of adult Americans (22 million women and 11 million men). Since their emergence as efficacious and cost-effective therapies (11–14), endovenous laser therapy and radiofrequency ablation have increasingly become a component of practice of numerous physicians; endovenous procedures performed increased from 80,000 to 195,000 during the period 2004–2007 (11,15). Patients’ desire for less invasive treatments, improved reimbursements to practitioners, and lack of requirement for surgical skills have resulted in a larger pool of physicians offering endovenous therapy (11,16). In the era of declining reimbursements (17), increasing provider competition for patients, and limited financial and personnel resources (17,18), developing and maintaining a successful venous practice requires the ability to plan for changes in market trends and allocate resources strategically based on annual and seasonal changes in volume (19). Although traditional methods of market analysis may have a primary role in market research and practice management, Internet search data are a powerful repository of information that has the potential to reveal overall population interest and identify market trends in near real time with minimal expense and time to the investigator. The leading Internet search provider (20), Google (Google, Inc, Mountain View, California), has developed a publicly available Web-based tool, Google Insights for Search (GIS), which provides normalized Web traffic for user-defined search terms in a numerical and graphic format. Several studies have established the epidemiologic value of GIS for disease (6,7,10,21–23); we sought to examine the use of Internet search data to identify and characterize the temporal and geographic population interest in varicose vein treatment.

MATERIALS AND METHODS GIS (http://google.com/insights/search) is a publicly available online tool that provides aggregated search volume for search terms that receive a significant amount of traffic for user-designated geographic regions and time restrictions. GIS also provides regional variations in search activity by country, state, or metropolitan area as well as a forecast function that is purported to predict future trends based on analysis of previous data. GIS search data are available only beginning January 1, 2004. In the present study, GIS was queried from January 1, 2004, to September 1, 2012, for the United States using lay terminology related to varicose vein treatment in English. With the objective of identifying potential patients interested in varicose vein treatment, the initial phrase employed was “varicose vein treatment.” Additional search terms that GIS identified as related

Harsha et al



JVIR

terms that received significant search volume included “varicose veins treatment” and “vein removal.” All three terms were queried in combination. This query returned search volume results in the form of a search volume index (SVI). SVI represents the relative search frequency of a particular search term normalized to the overall search traffic from the reference population and time period. This analysis indicates the likelihood of a random user to search for a particular search term from a certain location at a certain time (24). GIS enforces a threshold of traffic for search terms, and terms with low search volume are not displayed. GIS also eliminates repeated queries from a single Internet Protocol address to eliminate the possibility that results are artificially impacted by multiple queries by a single user. Results are displayed on a scale of 0– 100 with 100 representing the maximum relative search volume during the queried time period. Data points consisted of repeated measurements of search activity with a value provided for each week in the year over the 8-year study period. Each week of the study was grouped into its corresponding month, and the SVI was averaged to create an average SVI for each month. For example, the average June SVI represented the SVI for all weeks occurring in June from January 2004 to September 2012. In addition, GIS provides regional search volume for each state using Internet Protocol addresses from server logs to establish the origin of web queries. A single SVI value was provided from data aggregated from each state over all 8 years. To analyze geographic differences in search volume, states were grouped into the following regions: South, West, Midwest, and East according to U.S. Census Bureau regional identification (25). The data were imported into the R statistical programming environment for analysis (The R Project for Statistical Computing, 2012). Linear regression analysis and Kruskal-Wallis one-way analysis of variance (ANOVA) were employed to characterize annual search trends and peak search activity. One-way ANOVA also was employed to characterize differences in SVI from different geographic regions. Specific regional group differences were identified via post hoc Tukey-Kramer analysis. An α of .05 was considered statistically significant.

RESULTS Normalized search traffic for varicose vein treatment and related key words increased by 520% over the 104-month study period with SVI increasing from an average of 10 in 2004 to 62 in 2012. There was a year-over-year mean increase of 28% (range, 18%–100%; standard deviation [SD], 35%), with a statistically significant linear increase in average yearly SVI over time (R2 ¼ .94, P o .0001) (Fig 1). All years showed positive growth in mean SVI except for 2008, which showed an 18% decrease (Fig 1).

Volume 25



Number 1



January



2014

55

Figure 1. GIS query results for “varicose vein treatment” and related search terms during 2004–2012. (a) GIS graphic display of SVI over time. There is an overall positive trend in annual search traffic with seasonal peaks corresponding to the spring months. (b) Scatterplot of annual SVI shows a statistically significant linear increase in average yearly SVI over time (R2 ¼ 0.94, P o .0001). A yearover-year decrease in average SVI was present in the 2007–2008 interval with remainder years showing annual mean increase in search traffic. (Available in color online at www.jvir.org.)

Interest in varicose vein treatment as measured by SVI doubled from the decrease in 2008 to 2012. There was a great deal of seasonal variation in SVI with the lowest search traffic in the winter months (December and January) and the highest in the late spring months (May and June) (Fig 2). These differences were statistically significant (Kruskal-Wallis, P o .0001) with post hoc analysis showing significantly higher mean SVI compared with all other months in May (190% increase; range, 26%–670%; SD, 15%) and June (209% increase; range, 35%–700%; SD, 20%) after correction for multiple testing. Qualitatively, regional differences in interest in varicose vein treatment were evident in the heat map generated by GIS with apparent overall higher search activity in the South (Fig 3). To facilitate analysis, all states were grouped into their geographic region as defined by the U.S. Census Bureau. The average SVI for the southern region was 61.2 (range, 55–71; SD, 5.0), followed by 48.3 (range, 41–55; SD, 4.4) in the Eastern states and 47.3 in the Midwest (range 40-60, SD 6.1) and Western states (range, 27–60; SD, 8.6) (Fig 4). There were statistically significant regional differences (ANOVA, P o .0001)

Figure 2. Significant differences in varicose vein treatment interest by month. Scatterplot shows monthly SVI plotted for all years during the study interval. Interest in varicose vein treatment was significantly higher in May and June compared with all other months (Kruskal-Wallis, P o .0001). (Available in color online at www.jvir.org.)

with post hoc analysis showing significantly higher search traffic in the Southern states (Tukey-Kramer, P o .00001) (Table).

56



Patient Information-seeking for Varicose Vein Treatment

Harsha et al



JVIR

Table . Regional Comparisons of Search Volume for Varicose Vein Treatment Region

P Value

Midwest-East

.9820749

South-East

.0000151

West-East South-Midwest

.9801654 .0000061

West-Midwest

.9999967

West-South

.0000082

Regional comparisons of search volume for varicose vein treatment show statistically significant differences in search traffic between the Southern region compared with all other regions. Statistically significant P values (P o .05) are italicized. All other regions showed no significant differences. Figure 3. Regional variations in varicose vein treatment interest. Heat map produced by GIS shows regional variations in search activity by state for the period 2004–2012. States showing darker shades of blue color represent higher relative search traffic compared with states showing lighter blue shading. States shaded white showed relative search traffic that did not reach the minimum threshold for inclusion and can be interpreted to have much lower relative search traffic compared with the remainder states. (Available in color online at www.jvir.org.)

Figure 4. Southern states show statistically significant relative search traffic compared with other regions (Tukey-Kramer, P o .00001).

DISCUSSION Our results indicate that there has been significant growth in population interest for varicose vein treatment in the past 8 years as indicated by Google search queries, suggesting a growing market (Fig 1). Several factors may be responsible for this trend, including increased patient awareness, increasing use of the Internet for medical information (1), establishment of noninvasive treatment alternatives to surgical vein stripping, and increased exposure of patients to online advertising (4,26). All years during the period 2004–2012 except for 2008 showed positive growth in relative search traffic. We postulate that the decrease in search traffic in 2008 may be related to the poor economic climate at the time given the fact that most venous treatments are elective

procedures and may incur a large out-of-pocket expense for patients. Similar findings have been reported for other elective procedures, such as laser-assisted in situ keratomileusis, breast augmentation, eye surgery, and liposuction; all experienced weakening population interest in the time period beginning in 2007 (27,28). Internet users may also be affected by online media reports and Internet-based patient education materials (1–3,5,29), and although GIS provides a feature that allows correlation in search trends with popular media publications, no such correlation could be made with any particular media publication during this time period. The relatively larger number of queries in May and June indicates that interest in varicose vein treatment is affected by climate and seasonality (Fig 2). This finding is not surprising given that other researchers have shown distinct seasonal variations in Internet queries for mental illness (22,30), kidney stones (6,7), and breast cancer (5) and correlated this with known sociologic, pathophysiologic, and epidemiologic factors. Venous disorders may become more symptomatic in warmer temperatures (11,31), and increasing ambient temperatures across the United States during the late spring months may suggest a potential cause for this trend. In addition, many patients seek treatment not only because varicose veins may be functionally symptomatic but also because they are often seen as unattractive and result in dermatologic changes (11,12). It is conceivable that this phenomenon drives patients to seek treatment during the time of year when they may wear clothing that exposes these unsightly skin changes. The geographic bias in varicose vein treatment interest in the southern United States may further support the notion of cosmetics becoming a primary driver of interest in warmer climates. However, these trends are likely multifactorial in nature with differences in disease incidence, patient demographics (31,32), and provider density and advertising in certain regions playing a major role in regional variation in search activity. A great deal of time and financial resources are often devoted to qualitative market research, and given the increasingly pervasive use of the Internet in modern society, tools such as GIS represent a cost-effective and

Volume 25



Number 1



January



2014

time-efficient strategy to identify trends and modify practice patterns and marketing strategies rapidly. One could investigate similar search trends in one’s state or metropolitan area and employ Web-based marketing during times of peak online search activity and subsequently apportion resources appropriately as volumes increase. Future investigation using search engine query data is warranted and may provide opportunities to identify and elucidate complex relationships between patients’ priorities in medical treatment and provider supply. This study has several limitations. First, GIS is unable to provide absolute search traffic numbers and instead provides an indexed or normalized value (SVI). According to Google, trends in SVI do correspond to actual trends in absolute search traffic and are unlikely to be spuriously affected by small changes in the background search traffic of the population (24). Second, although the language of our search terms was aimed to capture interest from patients seeking varicose vein treatment, the actual motivations for users’ queries are unknown making it impossible to identify which users represent potential new patients, established patients, or the casual curious Internet user. In addition, it is unknown in what geographic markets Google represents the dominant search engine, especially for the search terms applied in this study related to varicose veins. Finally, national and statewide search trends are not directly applicable to local practice settings where greater inhomogeneities in the patient population may exist. In conclusion, interest in varicose vein treatment among the U.S. population as indicated by query volume from the popular Google search engine has increased severalfold over the past 8 years. There is predictable peak search traffic in the late spring months with an overall geographic bias in interest toward the Southern states. Given that the Internet is becoming more central to patients’ health information–seeking behavior, rigorous analysis of Internet searches offers unique real-time insight into seasonal and geographic trends in population interest for medical procedures and with further investigation may prove useful to guide the efficient use of limited resources and marketing dollars.

57

6.

7.

8.

9. 10. 11. 12. 13.

14. 15.

16.

17.

18. 19. 20.

21. 22.

23.

24.

25.

26.

REFERENCES 1. Fox S, Duggan M. Health Online 2013. Pew Internet & American Life Project 2013. Available at: http://www.pewinternet.org/  /media//Files/ Reports/PIP_HealthOnline.pdf. Accessed August 28, 2013. 2. Shukla P, Sanghvi SP, Lelkes VM, Kumar A, Contractor S. Readability assessment of internet-based patient education materials related to uterine artery embolization. J Vasc Interv Radiol 2013; 24:469–474. 3. Hay J, Coups EJ, Ford J, DiBonaventura M. Exposure to mass media health information, skin cancer beliefs, and sun protection behaviors in a United States probability sample. J Am Acad Dermatol 2009; 61: 783–792. 4. Ybarra ML, Suman M. Help seeking behavior and the Internet: a national survey. Int J Med Inform 2006; 75:29–41. 5. Glynn RW, Kelly JC, Coffey N, Sweeney KJ, Kerin MJ. The effect of breast cancer awareness month on internet search activity—a

27.

28.

29. 30.

31.

32.

comparison with awareness campaigns for lung and prostate cancer. BMC Cancer 2011; 11:442. Breyer BN, Sen S, Aaronson DS, Stoller ML, Erickson BA, Eisenberg ML. Use of Google Insights for Search to track seasonal and geographic kidney stone incidence in the United States. Urology 2011; 78:267–271. Willard SD, Nguyen MM. Internet search trends analysis tools can provide real-time data on kidney stone disease in the United States. Urology 2013; 81:37–42. Zeiger R, Mohebbi MH. Re: Breyer et al.: Use of Google insights for search to track seasonal and geographic kidney stone incidence in the United States. Urology 2012; 79:486. Dugas AF, Jalalpour M, Gel Y, et al. Influenza forecasting with Google Flu Trends. PloS One 2013; 8:e56176. Kang M, Zhong H, He J, Rutherford S, Yang F. Using Google Trends for influenza surveillance in South China. PloS One 2013; 8:e55205. Hamdan A. Management of varicose veins and venous insufficiency. JAMA 2012; 308:2612–2621. Hamdan A, Livingston EH, Lynm C. JAMA patient page. Treatment of varicose veins. JAMA 2013; 309:1306. Health Quality Ontario. Endovascular radiofrequency ablation for varicose veins: an evidence-based analysis. Ont Health Technol Assess Ser 2011; 11:1–93. Gohel M. Which treatments are cost-effective in the management of varicose veins? Phlebology 2013; 28:153–157. Kanwar A, Hansrani M, Lees T, Stansby G. Trends in varicose vein therapy in England: radical changes in the last decade. Ann R Coll of Surg Engl 2010; 92:341–346. Hamel-Desnos C, Gerard JL, Desnos P. Endovenous laser procedure in a clinic room: feasibility and side effects study of 1,700 cases. Phlebology 2009; 24:125–130. Talenfeld AD, Soni SJ, Moser JW, Kassing PJ. Reimbursement trends for outpatient interventional radiology procedures: comparison of hospital and freestanding physician office sites of service. J Am Coll Radiol 2009; 6:417–427. Muroff LR. Culture shift: an imperative for future survival. J Am Coll Radiol 2013; 10:93–98. Soares GM. Varicose vein treatment: considerations for a radiology group practice. J Am Coll Radiol 2005; 2:626–633. Comscore.com. May 2013 U.S. Search Engine Rankings. Comscore.com 2013. Available at: http://www.comscore.com/Insights/Press_Releases/ 2013/6/comScore_Releases_May_2013_U.S._Search_Engine_Rankings. Accessed May 31, 2013. Althouse BM, Ng YY, Cummings DA. Prediction of dengue incidence using search query surveillance. PLoS Negl Trop Dis 2011; 5:e1258. Ayers JW, Althouse BM, Allem JP, Rosenquist JN, Ford DE. Seasonality in seeking mental health information on google. Am J Prev Med 2013; 44:520–525. Noll-Hussong M, Lahmann C. Whiplash and Werther effect: the potential of Google Insights for Search for medical research and public health. Fortschr Neurol Psychiatr 2011; 79:340–344. Google Incorporated. Google Trends Help Analyzing Data. Google Incorporated 2013. Available at: https://support.google.com/trends/topic/ 13975?hl=en&ref_topic=13762. Accessed September 1, 2013. United States Census Bureau. Census Bureau Regions and Divisions with State FIPS Codes 2013. www.census.gov/geo/www/us_regdiv.pdf. Accessed March 25, 2013. Ogbogu U, Rachul C, Caulfield T. Reassessing direct-to-consumer portrayals of unproven stem cell therapies: is it getting better? Regen Med 2013; 8:361–369. Stein JD, Childers DM, Nan B, Mian SI. Gauging interest of the general public in laser-assisted in situ keratomileusis eye surgery. Cornea 2013; 32:1015–1018. American Society of Plastic Surgeons. 2010 Plastic Surgery Procedural Statistics. American Society of Plastic Surgeons 2013. www.plasticsur gery.org/media/statistics.html. Accessed June 13, 2013. Metcalfe D, Price C, Powell J. Media coverage and public reaction to a celebrity cancer diagnosis. J Public Health (Oxf) 2011; 33:80–85. Yang AC, Huang NE, Peng CK, Tsai SJ. Do seasons have an influence on the incidence of depression? The use of an internet search engine query data as a proxy of human affect. PloS One 2010; 5:e13728. Langer RD, Ho E, Denenberg JO, Fronek A, Allison M, Criqui MH. Relationships between symptoms and venous disease: the San Diego population study. Arch Intern Med 2005; 165:1420–1424. Brand FN, Dannenberg AL, Abbott RD, Kannel WB. The epidemiology of varicose veins: the Framingham Study. Am J Prev Med 1988; 4:96–101.

Know your market: use of online query tools to quantify trends in patient information-seeking behavior for varicose vein treatment.

To analyze Internet search data to characterize the temporal and geographic interest of Internet users in the United States in varicose vein treatment...
607KB Sizes 0 Downloads 0 Views