Crowdfunding for medical expenses has grown exponentially during the past decade.1 Through several websites, patients create public campaign profiles to which donors can contribute. These pages are typically publicized via digital or local social networks. Successful campaigns have raised as much as $2 million.2 One website, YouCaring.com, currently lists 52,787 campaigns for medical expenses; the highest-earning campaign has raised approximately $902,000 from 816 donors. Another popular website reported a recent marked increase in funding for medical campaigns, from approximately $1.6 million in 2011 to approximately $150 million in 2014.1
Many authors have opined on the ethical issues posed by medical crowdfunding.2–6 In a recent viewpoint published in the Journal of the American Medicine Association, Young and Scheinberg2 expressed concern that medical crowdfunding resources may be disproportionately available to patients with social media savvy, particularly photogenic patients, and patients with the most emotionally appealing causes, rather than patients with the greatest need. An editorial in Lancet Oncology raised concern about the unregulated nature of medical crowdfunding and the potential for fraudulent use.6 Snyder et al5 discussed a variety of ethical concerns, including loss of patient privacy, skewed distribution of crowdfunding resources, and restrictions on the use of funds.
Despite these issues, research on medical crowdfunding has been limited and largely qualitative. Snyder et al4 reviewed campaigns to identify thematic elements, and Berliner and Kenworthy7 described a cohort of 200 campaigns, analyzing bivariate correlation between the characteristics of the cases and the amounts raised. Thus far, no research has been published on medical crowdfunding for orthopedic diagnoses and procedures. As crowdfunding grows in popularity, orthopedic surgeons should expect a greater proportion of patients to seek financial assistance through online platforms. The current study analyzed the characteristics and outcomes of crowdfunding campaigns for orthopedic trauma, total joint arthroplasty, and spine surgery.
Materials and Methods
Data Source and Campaign Selection
Active medical crowdfunding campaigns were queried from YouCaring.com—a popular medical crowdfunding website—with the search terms “spine surgery” (queried May 11, 2017), “hip replacement” (queried May 12, 2017), “knee replacement” (queried May 12, 2017), “shoulder replacement” (queried May 13, 2017), and “fracture” (queried June 12, 2017). Campaigns were included if donations were being requested for expenses related to the procedures of interest, as determined through the title and description of the campaign. Veterinary campaigns, duplicate campaigns, and those with indeterminate patient age or sex or those missing a description were excluded. Campaign inclusion and exclusion are summarized in Figure 1. This study assumed that the duration of active campaigns did not differ across the characteristics of campaigns used for multivariate models. Because this study considered only publicly available data, it did not involve human subjects and therefore did not require institutional review board approval.
Campaign inclusion and exclusion. Abbreviation: TJA, total joint arthroplasty.
For this study, the primary dependent variables were receipt of any donation (defined as amount raised >$0) and the amount raised for campaigns that received at least 1 donation.
The following independent variables were collected for each campaign: number of social media shares, goal amount, patient sex, pediatric vs adult age category, location, and detailed description of the campaign. Patient sex and age group were determined based on campaign descriptions (eg, personal pronoun use), titles, and images provided. Location was coded into regions based on US Census Bureau designations, including “international” and “unspecified” locations. The character length for each detailed description was calculated. Campaigns were categorized as trauma (fracture query, plus spine cases identified as related to trauma), total joint arthroplasty (knee, hip, and shoulder replacement queries), or spine (spine surgery query, minus cases related to trauma).
Descriptive statistics for campaign characteristics and outcomes were generated. For bivariate and multivariate analyses, all continuous variables were winsorized at the 5th and 95th percentiles to reduce the influence of outliers. The raised amount was log-transformed for all linear regression analyses. Bivariate analyses were conducted with chi-square tests, Kruskal–Wallis tests, Wilcoxon–Mann–Whitney tests, logistic regression, and linear regression, as appropriate. Multivariate binomial logistic and linear regression were used for adjusted analyses, reporting odds ratios and percent change, respectively. Social media shares were log-transformed for use as the independent variable in linear regresion analysis, and the results of these analyses were reported as the percent outcome change given a +10% change in social media shares. For logistic regression, model fit was assessed with the Hosmer–Lemeshow goodness-of-fit test, and discrimination was assessed with the C-statistic. For linear regression, model performance was assessed with R2. All statistical analyses were conducted with SAS version 9.4 software (SAS Institute, Cary, North Carolina). Statistical significance was defined as P<.005.
For this study, 444 campaigns were analyzed, raising a total of $1,443,528. The largest proportion of campaigns were from the South (31.3%, n=139), followed by the West (21.9%, n=97). The distribution of campaigns by state is shown in Figure 2. Trauma accounted for most of the campaigns (59.0%, n=262), followed by total joint arthroplasty (23.7%, n=105) and spine (17.3%, n=77). Most campaigns were for male patients (55.2%, n=245) and adults (87.8%, n=390). Mean goal amount was $18,185 (SE, $1662), and mean description length was 1021 characters (SE, 37). Most campaigns received at least 1 donation (73.65%, n=327), and among those that did, mean amount raised was $4414 (SE, $611) (Table 1).
Distribution of campaigns by state.
Campaign Characteristics and Outcomes
Bivariate analyses showed that campaign location, procedure category, and description length were significantly associated with differences in the likelihood of donation receipt. Goal amount was significantly associated with the amount raised (Table 2).
Results of the Bivariate Analyses
Multivariate Analysis—Donation Receipt
Multivariate analysis showed that unspecified region (odds ratio, 0.26; P=.0008 vs West) and total joint arthroplasty campaigns (odds ratio, 0.35; P=.0003 vs trauma) had significantly lower odds of donation receipt. An incremental 100-character increase in description length was associated with higher odds of donation receipt (odds ratio, 1.13; P<.0001). Overall, the model discriminated campaigns with moderate efficacy (C-statistic=0.761) (Table 3; Figure 3).
Association of Campaign Characteristics With Outcomes on Multivariate Analysis
Receiver operating characteristic (ROC) curve for the donation receipt model.
Multivariate Analysis—Raised Amount
Among campaigns that received at least 1 donation, those with Southern location (−65.5%, P<.0001), international location (−68.5%, P=.0028), and unspecified location (−63.5%, P=.0039) raised lower amounts compared with campaigns located in the West. An incremental +$1000 increase in goal amount was associated with higher amount raised (+3.2%, P<.0001). The multivariate model accounted for a relatively low proportion of outcome variance (R2=0.188) (Table 3).
Number of Social Media Shares
Among campaigns that received at least 1 donation, an incremental 10% increase in the number of social media shares was associated with higher amount raised (+7.9%, P<.0001). Notably, social media shares alone accounted for a substantial proportion of outcome variance (R2=0.405) (Figure 4).
Raised amount by shares.
This analysis identified characteristics of crowdfunding campaigns associated with donation receipt and amount raised. Ultimately, campaign region, procedure category, goal amount, description length, and social media shares were significantly associated with campaign outcomes.
Longer description length was associated with higher odds of donation receipt, but was not significantly associated with amount raised. It appears that a brief description generally discourages potential donors, and attributes other than description length drive incremental campaign success. Toward that end, higher goal amount was associated with higher amount raised. This finding may be the result of a ceiling effect whereby campaigns that achieve their goal amount are less likely to receive further donations; according to this model, setting a higher goal amount avoids such negative feedback. Alternatively, goal amount may serve as a proxy for need. Snyder et al8 found that medical crowdfunding campaigns draw from 3 general themes: preexisting personal connections to donors, extent of patient need, and the idea of “giving back” to a deserving patient. Further research is needed to better understand these dynamics.
Although models of donation receipt showed reasonable campaign discrimination, multivariate models for amount raised showed poorer performance. However, the number of social media shares was markedly associated with amount raised, and a bivariate model with this variable alone accounted for a substantial proportion of variance. This finding is similar to that of Berliner and Kenworthy7 in a random sample of 200 GoFundMe campaigns. Shares are an intermediate outcome and therefore cannot be considered “predictive” of campaign success. Rather, these results strongly link campaign outcomes to digital social network access and use. Campaigns for trauma patients showed markedly greater odds of donation receipt compared with campaigns for total joint arthroplasty. This difference may be attributed to the greater incidence of orthopedic trauma compared with total joint arthroplasty and spine procedures among younger patients.9–11 These patients are more likely to use digital social networks, which potentially may account for increased campaign success.12 Alternatively, trauma diagnoses may be more emotionally appealing to potential donors. Research is needed to clarify these issues.
Campaigns with unspecified location had lower odds of donation receipt and lower amount raised compared with Western campaigns. Snyder et al8 conducted qualitative research on medical crowdfunding that showed that campaigns often appeal personally to local community members. Campaigns that do not specify a location may lack a connection to the local community and have poorer outcomes as a result. Southern campaigns also raised significantly lower amounts compared with Western campaigns. A 2013 survey found that the West had the highest proportion of households with a computer and Internet access and that the South had the lowest proportion.13 Therefore, communities with lower online engagement may be less attractive for medical crowdfunding. International campaigns similarly were associated with lower amount raised, although this finding is likely biased by differences in the relative value of currency. Disparities in access to crowdfunding resources by geography should be further examined.
This study was limited by data availability. Although various characteristics of campaigns can be determined from information found on campaign pages, many potentially important characteristics were not present (eg, campaign start and end dates, precise patient age, household income). Therefore, the ability to control for confounding factors was limited. Further investigation should be conducted as more information on medical crowdfunding becomes publicly available. Additionally, this study used data from a single crowdfunding website. Although selection of a single crowdfunding platform may introduce bias, the authors believe that their approach of considering all campaigns matching their search criteria markedly decreased the likelihood of bias compared with past qualitative studies. Future investigations should compare variations in the characteristics and outcomes of campaigns between websites. The authors hope that data on medical crowdfunding will be more widely available in the future, facilitating further research.
This study provided a first step toward a quantitative understanding of the dynamics of medical crowdfunding. Resources obtained through crowdfunding may be disproportionately available to patients with specific diagnoses, those from specific regions, and those who are able to craft a lengthy descriptive narrative. Leverage of both local and digital social networks may be instrumental to the success of these campaigns. It may be possible to manipulate the use of crowdfunding by selecting certain campaign parameters, such as specifying a high goal amount to avoid negative feedback on donation receipt. As crowdfunding gains popularity, clinicians may see more patients pursuing this option, and patients may ask physicians for information about crowdfunding or ask them to support these campaigns by providing medical testimonials. Orthopedic surgeons should consider their response to such requests individually. The current findings increase understanding of the dynamics of medical crowdfunding and can therefore support more robust personal and professional deliberation.
- Chandler R. GoFundMe sees boom in medically-related fundraising campaigns. http://whotv.com/2015/03/14/gofundme-sees-boom-in-medically-related-fundraising-campaigns. Accessed August 1, 2017.
- Young MJ, Scheinberg E. The rise of crowdfunding for medical care: promises and perils. JAMA. 2017; 317(16):1623–1624. doi:10.1001/jama.2017.3078 [CrossRef]
- Sisler J. Crowdfunding for medical expenses. CMAJ. 2012; 184(2):E123–E124. doi:10.1503/cmaj.109-4084 [CrossRef]
- Snyder J, Chow-White P, Crooks VA, Mathers A. Widening the gap: additional concerns with crowdfunding in health care. Lancet Oncol. 2017; 18(5):e240. doi:10.1016/S1470-2045(17)30259-0 [CrossRef]
- Snyder J, Mathers A, Crooks VA. Fund my treatment! A call for ethics-focused social science research into the use of crowdfunding for medical care. Soc Sci Med. 2016; 169:27–30. doi:10.1016/j.socscimed.2016.09.024 [CrossRef]
- The Lancet Oncology. Mind the gap: charity and crowdfunding in health care. Lancet Oncol. 2017; 18(3):269. doi:10.1016/S1470-2045(17)30117-1 [CrossRef]
- Berliner LS, Kenworthy NJ. Producing a worthy illness: personal crowdfunding amidst financial crisis. Soc Sci Med. 2017; 187:233–242. doi:10.1016/j.socscimed.2017.02.008 [CrossRef]
- Snyder J, Crooks VA, Mathers A, Chow-White P. Appealing to the crowd: ethical justifications in Canadian medical crowdfunding campaigns. J Med Ethics. 2017; 43(6):364–367. doi:10.1136/medethics-2016-103933 [CrossRef]
- Kurtz S, Mowat F, Ong K, Chan N, Lau E, Halpern M. Prevalence of primary and revision total hip and knee arthroplasty in the United States from 1990 through 2002. J Bone Joint Surg Am. 2005; 87(7):1487–1497.
- Court-Brown CM, Caesar B. Epidemiology of adult fractures: a review. Injury. 2006; 37(8):691–697. doi:10.1016/j.injury.2006.04.130 [CrossRef]
- Rajaee SS, Bae HW, Kanim LE, Delamarter RB. Spinal fusion in the United States: analysis of trends from 1998 to 2008. Spine (Phila Pa 1976). 2012; 37(1):67–76. doi:10.1097/BRS.0b013e31820cccfb [CrossRef]
- Haight M, Quan-Haase A, Corbett BA. Revisiting the digital divide in Canada: the impact of demographic factors on access to the internet, level of online activity, and social networking site usage. Inf Commun Soc. 2014; 17(4):503–519. doi:10.1080/1369118X.2014.891633 [CrossRef]
- File T, Ryan C. Computer and Internet use in the United States: 2013. Washington, DC: US Census Bureau; 2014.
Campaign Characteristics and Outcomes
|Region, No. (%)|
| International||43 (9.7)|
| Midwest||68 (15.3)|
| Northeast||28 (6.3)|
| South||139 (31.3)|
| Unspecified||69 (15.5)|
| West||97 (21.9)|
|Procedure category, No. (%)|
| Spine||77 (17.3)|
| Total joint arthroplasty||105 (23.7)|
| Trauma||262 (59.0)|
|Sex, No. (%)|
| Male||245 (55.2)|
| Female||199 (44.8)|
|Age group, No. (%)|
| Adult||390 (87.8)|
| Pediatric||54 (12.2)|
| Standard error||$1662|
|Description length (characters)|
| Standard error||37|
|Donation received, No. (%)|
| No||117 (26.35)|
| Yes||327 (73.65)|
| Standard error||$611|
Results of the Bivariate Analysesa
| Total joint arthroplasty||58.1||$1846|
|Goal amount (per $1000)||1.00||.8651||+3.2%||<.0001|
|Description length (per 100 characters)||1.14||<.0001||+4.0%||.0184|
Association of Campaign Characteristics With Outcomes on Multivariate Analysisa
|Procedure category (reference=trauma)|
| Total joint arthroplasty||0.35||.0003||−45.7%||.0218|
|Age group (reference=adult)|
|Goal amount (per $1000)||1.00||.4028||+3.2%||<.0001|
|Description length (per 100 characters)||1.13||<.0001||+2.0%||.2212|