Health
Services Research. 51.3 (June 2016): p1273. From InfoTrac
Humanities Collection 2017.
Abstract:
Objective. To pilot public health interventions
at women potentially interested in maternity care via campaigns on social
media (Twitter), social networks (Facebook), and online search engines
(Google Search).
Data Sources/Study Setting. Primary data from
Twitter, Facebook, and Google Search on users of these platforms in Los Angeles
between March and July 2014.
Study Design. Observational study measuring the
responses of targeted users of Twitter, Facebook, and Google Search exposed to
our sponsored messages soliciting them to start an engagement process by
clicking through to a study website containing information on maternity care
quality information for the Los Angeles market.
Principal Findings. Campaigns reached a little
more than 140,000 consumers each day across the three platforms, with a little
more than 400 engagements each day. Facebook and Google search had broader
reach, better engagement rates, and lower costs than Twitter. Costs to reach
1,000 targeted users were approximately in the same range as less well-targeted
radio and TV advertisements, while initial engagements--a user clicking through
an advertisement--cost less than $1 each.
Conclusions. Our results suggest that
commercially available online advertising platforms in wide use by other
industries could play a role in targeted public health interventions.
Key Words. Social media, social networks,
maternity care quality
Full Text:
Widespread attempts have been made to publicize
clinical performance measures aimed at improving delivered quality (Agency for
Healthcare Research and Quality, 2011). Such "report cards" exist for
a number of reasons: to educate consumers and referrers, to illustrate and to
enhance potential choices of provider, to allow greater autonomy, and to
improve efficiency in decision making.
Today, the National Quality Forum currently
endorses 743 standards (National Quality Forum, 2011). Public report cards
administered by health insurance plans are similarly available nationwide,
with, for example, the National Committee for Quality Assurance listing 136
national plans reporting some measure of physician or hospital-specific
performance (National Committee for Quality Assurance, 2011), culminating in
the launch of the Hospital Compare public reporting website by CMS in 2005.
Given this generation of potentially accessible
valuable information, it is clear that stakeholders need to and wish to inform
the public of these resources. Traditional marketing campaigns may be less
potent in an increasingly digital, technology-enabled society with home
broadband and high bandwidth mobile Internet on ubiquitous smartphones that
allow real-time searches for online information and live interactions on social
media and social networks. Consumers wish to search for and create content
online and interact with like-minded others. The Pew Internet and American Life
Project finds that more than 50 percent of online adults between the ages of 18
and 55 years use social networking sites, while one in four Americans report
that the Internet has helped them deal with at least one major health-related
life decision (Pew Research Internet Project, 2014).
However, public health messaging that does
attempt to use these virtual channels is typically highly passive. Merely
tweeting broadly about the importance of seasonal flu vaccination is not likely
to be sufficient. Using static websites to offer comparative hospital quality
information-based messages will not work unless consumers become aware of and
have confidence in the information (Huesch, Currid-Halkett, and Doctor 2014).
It is reasonable to ask whether we have fully
exploited technology-based solutions to empower and better inform patients. In
other industries, digital advertising through social media, online social
networks, and Internet search engines is premised on the valuable data that
such online platforms build up on their users.
Objective
In this article, we describe a pilot study of
an intervention to provide information on maternity care quality to Los Angeles
consumers plausibly interested in such information using three commercial
campaigns on social media, an online social network, and an Internet
search engine. Our objective is to understand whether we can quickly and
cheaply reach out to prospective consumers of public health information using
new digital technology. This pilot explicitly did not seek to actually change
the health-related behavior of consumers nor did it ascertain whether consumers
understood and acted on such public health information.
Empirical Setting
We chose to focus on the maternity care setting
for a number of important reasons, aside from the obvious aspect that maternity
care is a highly "shoppable" condition in which consumers have
substantial time to acquire information and make decisions on health care
utilization.
Maternity care is the second most common reason
for hospitalization, the fourth most common reason for seeking ambulatory care
(Sakala and Corry 2008) includes the top three procedures billed to Medicaid or
private payors, and accounts for more than fourth of all Medicaid-billed
hospital charges, and nearly a sixth of all private insured-billed hospital
charges (Agency for Healthcare Research and Quality, 2008).
Yet on key objective evidence-based metrics
such as the appropriate use of Cesarean sections (Main et al. 2011), the
proportion of women who received antenatal care within the first trimester, low
birth weight infant deliveries, infant mortality, and maternal death rates,
progress has either been away from or incompletely toward federal targets (U.S.
Department of Health and Human Services, 2006).
Reflecting these quality imperative, a number
of easily accessible state government, federal government, and commercial
entities' websites provide substantial data on local, regional, and national
maternity care quality by named hospitals. Our study then provided links to
these online report cards on the study website; the first author will provide
by request detailed additional information on the quality metrics listed on
each of these online report cards.
CONCEPTUAL FRAMEWORK
We have employed three complementary frameworks
in this study, each of which represents a different perspective on what we seek
to do. At the highest level, we see this pilot study as being a classical
public health intervention. The theoretical underpinnings of this intervention
are in social cognitive theory, while the practical bases lie in standard
commercial marketing management and sales management.
Public Health Intervention Perspective
Using a classic public health model (Keller et
al. 1998), our study is a population-based, individual-focused, primary
prevention, public health intervention which combines elements of Outreach,
Social Marketing, and Health Teaching.
As an Outreach intervention, we seek to locate
individuals at risk of receiving less than optimal maternity care and ensure
their access to information that can improve their maternity and delivery care
by choosing a hospital of higher quality as reported by a public hospital
reporting website. As a Social Marketing intervention, we seek to utilize
commercial marketing tools and techniques to influence these individuals and
their beliefs and decisions. As a Health Teaching intervention, we intend to
communicate facts and ideas to change the beliefs and behaviors of those
individuals.
Social Cognitive Theoretical Perspective
The theoretical grounding of our intervention
is in social cognitive theory. The principles and processes underlying a
target's susceptibility to outside influences are grounded in light of three
goals fundamental to rewarding human functioning (Cialdini and Goldstein 2004).
Specifically, targets are motivated to form accurate perceptions of reality and
react accordingly, to develop and preserve meaningful social relationships, and
to maintain a favorable self-concept. Of particular importance are social
norms, behavior expectations within a particular group that can influence
behavior of group members due to a desire to conform with actual behavior (the
descriptive norm) or sanctioned behavior (the injunctive norm).
Previous research has demonstrated that social
norms of appropriate behavior can exert a stronger effect on behavior than
modest economic incentives or self-interest (Heyman and Ariely 2004;
Griskevicius, Cialdini, and Goldstein 2008; Nolan et al. 2008). We are
especially likely to follow the lead of others whom we perceive to be similar
to ourselves (Hornstein, Fisch, and Holmes 1968; Murray et al. 1984; White,
Hogg, and Terry 2002).
From an economic perspective, social norms may
convey information concerning appropriate behavior or social consequences of
failing to conform; however, behavioral studies find that these effects persist
even when behavior is unobservable (e.g., littering when nobody is around) and
social information is not particularly informative to one's own preferences
(e.g., towel recycling).
We expect that engaging with consumers to
provide accurate perceptions of the reality of differences in hospital quality,
and providing information regarding the actions of geographically and
psychosocially similar consumers in choosing high-quality hospitals for
maternity care (a descriptive social norm), as well as providing information
that there exist nationally recommended guidelines for avoiding unnecessary
Cesarean sections (an injunctive norm) will lead consumers to choose to visit
the recommended website (i.e., the study website) and click-through to existing
sources of publicly reported data on hospital quality.
Marketing Perspective
Well-known terms of sales management can be
applied to electronic patient education and information provision. We recognize
that patient education is an actual sales process in which information is being
sold even when it's being given away. In line with this sales process,
potential customers are initially prospects, then become qualified prospects, then
are converted to actual customers (see Figure 1).
We compete for some share of the customers'
mind in an environment where a patient's fixed attention span is increasingly
divided among many screens (i.e., TV, PC, and mobile) and by many attention-seekers.
Online consumers in the United States are already estimated (eMarketer, 2012)
to be exposed to nearly $22 billion worth of small advertisements that appear
on paid search engine results (e.g., Google, Bing, Yahoo), and another $21
billion in display advertisements (e.g., on most any websites and on
online social media and social networks).
Although we may offer our information for free,
the potential customer or prospect faces noncash costs in acquiring this
information. The prospect must invest time and effort in accessing and
understanding this information. Conversely, there are most often opportunity
costs for that time and effort. Accordingly, we must sell the patient on the
worth of these investments. Partly this is through attractive and eye-catching
advertisements, partly through the intrinsic value of the information that we
seek to make available.
Qualified prospects are those who have
expressed an interest in the product or service offered. We estimated that our
total qualified prospects would be approximately 500,000-750,000 women and
close friends, relatives, and partners at any point in time in Los Angeles
county. We based this estimate on the 300,000 annual births in the county and
the simplifying assumption of persisting interest throughout a typical 9-month
pregnancy. This estimate was in line with recent findings of more than 1
million searches per month on Google in Los Angeles regarding pregnancy, more
than 50,000 monthly searches for maternity care providers, and around 20,000
monthly searches for hospital quality information (Huesch, Currid Halkett, and
Doctor 2014).
In our setting, the process of qualification is
predominantly owned by the advertising platforms we access, who serve us
prospects that they have deemed to be interested in our information on
maternity care quality. Each of these platforms has different strengths and
weaknesses. For example, a young lady in Los Angeles who uses Google to
privately search the web regarding early pregnancy care options may have left a
digital footprint with Google which could include every search she has ever
made (if she has signed up for gmail or if she uses Google's social network,
Google+, or if she uses the same home computer to perform all her searches). If
she mentions pregnancy concerns in private to her Facebook friends, publicly
follows a maternity care account on Twitter, or privately clicks-through Tweets
to maternity care websites, then Facebook and Twitter can similarly infer
relevant interests.
Finally, qualified prospects are incented to
undertake some behavior that closes the deal. By navigating to a website,
liking or following a Facebook page, following a Twitter account, or retweeting
a Tweet, our qualified prospects have been converted to actual customers in
this conceptual framework. Clearly, the ultimate objective of public health
interventions that seek to educate patients on health care quality is in
effecting actual change in decision making and health care utilization.
However, in this pilot study, we sought merely to achieve a proof of concept of
being able to reach and provide such information to a large number of
consumers.
METHODS
We used three online platforms, each a leading
example of a type of value-added Internet service. We contracted with Facebook,
a platform that allows users to form social networks online, with Twitter, a
platform that allows users to post brief 140-character messages online, and
with Google Search, a platform that allows users to search the Internet.
Our study team has produced detailed guides
(see Appendix S2 and S3), including step-by-step platform website recording, as
to how these commercial arrangements are set up, maintained and adjusted, and
ultimately wound down.
Our terms of trade with all three platforms
were generally similar, although the individual technical details and terms
varied. We sought to purchase access to qualified prospects for information
provision on maternity care quality. This qualification is important to the
platform: too many poorly targeted advertisements can affect user loyalty. It
was therefore in each platform's interest--as well as ours--to present our
advertisements only to those users who were likely to be interested.
We provided overall requested demographics
(women, Los Angeles city + 25 miles or Los Angeles county, aged 18-49 if
available, including Spanish speaking if available and broken out separately)
and customer interest information (e.g., keywords such as pregnant, dar a luz)
to the platforms to facilitate their qualification. Platforms drew from their
pool of users but did not make detailed lists of exposed users available to us
for privacy reasons. We thus relied completely on the integrity of the
platform's respective user databases. Especially with regard to Twitter, where
users provide limited account information on location and demographics and
interests, the pool of users may have been selected partially based on
Twitter's inferences of user behavior and interests.
We wished to limit our financial risk and only
pay per converted customer, where this study intends "conversion" to
mean an initial engagement by the customer in clicking through an advertisement
on Facebook or Google Search and arriving at our test website. For Twitter,
conversion was measured in two ways. The first is analogous to the narrow
definition of conversion of a qualified prospect on Facebook or Google Search.
The second is a broader and looser measure of engagement tailored to
Twitter's social media business model, which includes click-throughs
but also comprises additionally the following actions: following our account,
expanding an advertisement to read the full copy, favoriting an account,
retweeting, or replying. Twitter insists on payment on the basis of this
broader measure of engagement, so when we report comparable costs per
click-through for Twitter users, these will include the costs paid for
customers who did not click-through to our website, but otherwise engaged with
our message in Twitter's definition.
For each platform, we negotiated prices per
customer and/or set bids to reach such customers, and were able to coarsely
tune such prices and/or bids throughout the campaign to test whether reach or
conversion increased. However, it is important to understand that we are
bidding in a "sealed bid" auction for position in search results and
in Facebook feeds so that reach and conversion are jointly determined by the
competitive actions of very many marketers, each seeking to reach similar sets
of customers.
In general, bids were a function of our
advertisement's popularity with users, our desired placement within a user's
visibility, and our desired reach. We paid monthly on electronic invoicing,
using credit cards as payment mechanism.
We provided each platform with advertising
materials consisting of pictures and text copy embedding a URL (a hyperlink) to
click-through (Figures 2A, B, and C). We chose standard commercial images of
neutral images of a baby, or a mother and baby, and within the context of
dramatically limited word or character counts used neutral, easy to understand
language to "pitch" our messages. The lead author took overall
responsibility for approving team pictures and text copy.
The platforms supplied us with detailed data on
a daily frequency of how many users had been exposed to the advertisement (the
number of impressions or the reach), and what behavior had resulted (whether
the user had clicked through to our test website, as well as for Twitter only,
the other measures of engagement listed above). With some restrictions and
prior permission by the platforms, we were able to modify and fine-tune
advertising copy to test different responses.
Finally, we also used a fluent Latin American
Spanish speaker to translate our promotional materials into Spanish for the
large number of Hispanic prospects in our geographical area. We used
ethnicity-appropriate images and culturally appropriate text, including Central
and Southern American slang terms where appropriate. We ran this part of the
intervention separately from the English-language campaigns.
We created a visually attractive study website
that explained the study objective and that encouraged arriving users to
consult the collated sources of local, state, or federal data on maternity care
quality. For the Spanish-language campaign, we similarly translated this
website into Spanish. We enabled Google Analytics on this website to analyze
the origin of incoming web traffic to ensure that the platforms' reports of
outgoing traffic for which we were billed matched our analysis of incoming
traffic. This match proved to be almost exact, with some additional organic
traffic (<1 percent) coming to our website from repeat visitors who could
have bookmarked our site and subsequently returned to consult it.
We designed the campaign to use masters'
student research assistants, one acting as "channel manager" for each
of the three platforms and one to administer the study website. The lead author
supervised and managed the channel managers and approved all commercial
negotiations, text copy, bid prices, and changes in website design. All
statistics reported are purely descriptive and no statistical comparisons or
extrapolations to the population were attempted. This study was approved by the
Institutional Review Board of the lead author's home institution.
RESULTS
The summary performance measures across the
three channels are described in terms of daily metrics and overall study
performance in Table 1.
The individual platform performance measures
are further described below at a summary level, with detailed platform
performance data, website designs, advertising copy, and other details
available from the lead author on request.
[FIGURE 2 OMITTED]
Facebook
Our campaign on Facebook ran on consecutive
days between March 20 and July 30, 2014. We spent a total of $13,689 to reach a
nonunique total of 4,480,119 Facebook users. Of these, our total number of
click-through engagements achieved was 19,923 nonunique clicks and 17,764
unique user clicks.
Our overall unique engagement rate was thus 0.4
percent of those reached. On average, we reached 33,685 unique users each day,
soliciting 134 unique clicks each day and spending $0.77 per unique user to
achieve that engagement. Overall, to reach 1,000 users, or the CPM metric, cost
us $3.06.
In the subset of our results in which we ran
our Spanish-language pilot, results were similar. Over 28 days we reached
1,496,818 nonunique Facebook users in Los Angeles and solicited 5,752 unique
engagements at an average cost of $0.75 per unique user. The Spanish-language
unique engagement rate of 0.38 percent was similar to the overall campaign
results.
Google
Our campaign on Google search ran on the
consecutive days between April 1 and July 28, 2014. We spent a total of $25,177
to reach a nonunique total of 10,959,961 Google search users. Of these, our
total number of click-through engagements achieved was 27,676 without data on
how many were unique individuals.
Our overall engagement rate was thus 0.25
percent of those reached. On average, we reached 92,100 nonunique searchers
each day, soliciting 232 clicks each day and spending $0.91 per user to achieve
that engagement. Overall, to reach 1,000 users cost us the least at $2.30.
In the subset of our results in which we ran
our Spanish-language pilot, activity was different. Over 21 days we reached
only 203,463 nonunique Google search users in Los Angeles and solicited 1,054
click-through engagements at an average cost of $4.65 per unique user. The
Spanish-language engagement rate of 0.52 percent was higher than the overall
campaign results.
Twitter
Our campaign on Twitter ran between March 26
and July 31,2014. We spent a total of $20,542 to reach a nonunique total of
2,223,493 Twitter users. Based on Twitter's broad measure of engagement, we
achieved a total of 30,858 engagements for an engagement rate of 1.4 percent.
However, to properly compare our results on
Twitter with Facebook and Google, we focus on the comparable and far narrower
measure of nonunique click-through engagements. Here, our total was far less at
3,798 nonunique engagements. Our overall engagement rate was thus 0.17 percent
of those reached. On average, we reached 17,507 nonunique users each day,
soliciting 29.9 nonunique clicks each day and spending $5.41 per user to
achieve that engagement. Overall, to reach 1,000 users cost us the most at
$9.24.
In the subset of our results in which we ran
our Spanish-language pilot, results were similar. Over 17 days, we reached
336,439 nonunique Twitter users in Los Angeles and solicited 4,030 broad
engagements and 430 nonunique click-throughs at an average cost of $7.12 per
unique user. The Spanish-language unique click-through rate of 0.13 percent,
slightly lower than the overall campaign results.
DISCUSSION
This study used Twitter, Facebook, and Google
Search to reach out to consumers with potential interest in maternity care
quality information. Contracting on commercial terms, we spent a little more
than $500 a day across these three platforms to obtain engagements from a
little more than 400 consumers each day to our study website containing
relevant information.
As an initial proof of concept, we believe that
this pilot has shown that it is possible to drive consumer interest toward a
static website for a little more than $1 per qualified consumer on average.
This intervention was also relatively simple to design and launch, and was
greatly facilitated by the professional counterparties at each of the three
platforms.
It is important to put our results into
perspective and understand other options for reaching consumers. These options
depend on whether the patient is identified or not, and whether in-person or
remote channels are used.
On one hand, if such patients are not yet
identified, the total cost of in-person outreach to unidentified high-risk,
low-income pregnant women to enroll them in high-risk antenatal care using case
workers has previously been estimated in one study as $850 per enrollee. That
study sought to enroll women at welfare offices, clinics, and in high-potential
residential areas in an urban environment (Brooks-Gunn et al. 1989).
On the other hand, if such consumers are
already identified by name, location, or phone number, then direct outreach to
enroll them (i.e., to achieve a conversion) into a counseling or educational
program is possible. Historically, the direct costs of in-person outreach to
female low-income patients is around $50 per patient (typically 2-3 labor hours
at $20/hour staff time) down to $3/patient for labor costs related to a phone
call, and as low as $1/patient for labor and postage costs for a letter to a
known patient (Wagner et al. 2007).
Beyond in-person outreach, cheaper health
education and social marketing using traditional media channels has different
cost structures. Here, almost all cost data are on reach to prospects or qualified
prospects, not on actual conversions. The relevant cost metric is CPM or cost
per thousand impressions, where one impression is synonymous with reaching one
prospect. It costs more than $30 to reach a thousand prospects using newspaper
advertisements, around $20 for magazines, about $7 for radio, and $5 for TV
advertisements (Flannagan 2015). Other less well-targeted modalities are still
lower such as billboard marketing with an estimated CPM of $3-$5 (Grunert
2015).
In our results, our CPM was comparable to these
metrics, lying between $2 and $10, although the quality of our impressions is
likely to have been better than the less targeted advertising media listed
above due to the platforms' better information on users. Reassuringly,
our social media and social network achieved CPM is also very similar
to cited results for CPM achieved in these digital channels by other relatively
unsophisticated small business marketers who tend to pay in the range of around
$4-$20 on average (Grunert 2015).
Limitations
We readily acknowledge several well-known and
important limitations of such campaigns. We do not know whether such
information provision did or even can affect users' decision making and thus
lead to a health benefit, and this was explicitly not a study objective or
ascertainable in this design.
Future studies should track user behavior,
identifying follow-on behavior (e.g., which websites were contacted
subsequently), satisfaction (e.g., with depth and breadth of information
obtained), and actions taken (e.g., whether choice of providers was changed).
Relatively simple trackers added to the website can track online behavior;
these include outbound link trackers such as, for example, www.[insert your
site URL here]/linktracker.php?link=[insert the URL of a subsequent site you
wish to track outward traffic to] or simply by exploiting Google Analytics
(https://www.google.com/analytics/) functionality on your site, and these allow
researchers to understand how effective a marketing campaign is at driving qualified
consumers to destination sites.
We also know that much reach is repetitive:
identical users are repeatedly exposed to the same advertisement, so that
overall reach must be interpreted as being nonunique. Except for Facebook,
where unique user engagements are tracked by Facebook, we have to assume that
engagements achieved from Twitter and Google include nonunique user behavior.
CONCLUSIONS
In our results, Google and Facebook appeared to
be of greater potential for other researchers and public health stakeholders
undertaking similar interventions. Overall reach was greater, click-throughs
higher and cheaper to obtain, and CPM and click-through rates higher than for
Twitter. We assume that the greater knowledge built up by Facebook and Google
over its members and users could better be leveraged to serve up more qualified
customers.
Similarly, the more private nature of
information exchange of those two properties (i.e., completely privately in
Google searches, and semiprivately among friends in Facebook) compared to the
more public nature of Twitter probably led to more genuine interest. Our
results are directionally consistent with the relative amounts of per user
revenue earned each year by these three platforms. Google is estimated to earn
$45, Facebook $7.24, and Twitter $3.55 in advertising revenues per user per
year (Meeker 2014).
In interesting but only introductory results,
it appeared that Spanish-language users of Google search were substantially
less likely to be exposed to our campaign, compared to English-language users.
Daily reach was only a tenth, although engagement rates were double. It is not
clear whether this indicates that this platform is used less often by Spanish
speakers, perhaps due to insufficient access to Internet or desktop computers,
or whether Spanish speakers use English-language searches. It was, however, the
case that in Twitter and Facebook campaigns, there were essentially no
differences between the English- and Spanish-language campaign responses.
Public health interventions can reach qualified
prospects through many possible communication channels. Yet overall,
traditional channels remain overrepresented in marketing communications: 45
percent of American's media time last year was spent on the Internet and on
mobiles, yet only 26 percent of media spend went to those new channels (Meeker
2014). Specifically in health care marketing communications, the entire health
care industry accounts for just 2.5 percent of total online digital advertising
(eMarketer, 2013), and much of that is direct to consumer advertising by the
pharmaceutical industry.
This disproportionately small share of these
increasingly important communication channels highlights an opportunity for
better, more tailored, closer, and more cost-effective outreach. While such
campaigns require often substantial budgets, there are cost less options too.
For example, Google offers in-kind advertising space (through
https://www.google. com/grants/) similar to the offering of free public service
announcements on traditional media.
This study showed that it was possible to reach
out to qualified consumers and at least initiate an engagement with these
customers at marketing costs that were comparable to traditional media. Still
lacking are data on whether such initial engagement or the overall impressions
actually predisposes users to obtain, understand, and internalize such
health-related information and act appropriately on it by choosing high-quality
health care providers.
Some leading examples for similar outreach
exist. Directly targeting smokers through carefully placed online advertising
has proved to be an inexpensive way of attracting traffic to the California
Department of Public Health's TobaccoFreeCA website (personal communication,
Valerie Quinn, June 12, 2014). Similar campaigns could target vaccine skeptics
or users of complementary and alternative medicine. But more broadly, we are
not aware of large-scale uses of such platforms except in relatively passive
modes by hospitals with, for example, a Facebook page or a Twitter feed, or an
advertisement on Google Search for their care delivery business.
It is our hope that this study conveys a sense
of the relative ease and simplicity, and the relatively low costs and
circumscribed financial risk of such campaigns to market public report cards.
For our health system to overcome the serious challenges that threaten our
entire nation fiscally, we need to have better informed patients taking charge
and participating in their own health care and wellness. Understanding the role
that Internet and social media-based custom education approaches could
play to inform patient decision making and possibly incent patient behavioral
change appears to be an important next step for public health stakeholders.
DOI: 10.1111/1475-6773.12496
ACKNOWLEDGMENTS
Joint Acknowledgment/Disclosure Statement The
authors gratefully acknowledge financial support by the Agency for Healthcare
Research and Quality (AHRQ) under grant R21 HS21868, and helpful project advice
and ongoing support by Brent Sandmeyer and Galen Gregor at AHRQ. The authors
also acknowledge the excellent research assistance of the following graduate
students at USC over the period 09/30/2012-12/31/2014: Brent Costa, Shakeh
Missian, Natalie Abidjian, Don Marshall, and Shyamala Shastri.
Disclosures: None.
Disclaimers: None.
REFERENCES
Agency for Healthcare Research and Quality.
2008. HCUPnet, Healthcare Cost and Utilization Project. Rockville, MD: AHRQ
[accessed on March 30, 2016]. Available at http://hcupnet.ahrq.gov/
Agency for Healthcare Research and Quality.
2011. "Health Care Report Card Compendium" [accessed on March 30,
2016], Available at https://cahps.ahrq.gov/ consumer-reporting/talkingquality/
Brooks-Gunn, J., M. C. McCormick, J.
Brooks-Gunn, T. Shorter, J. H. Holmes, C. Y. Wallace, and M. C. Heagarty. 1989.
"Outreach as Case Finding: The Process of Locating Low-Income Pregnant
Women." Medical Care 27 (2): 95-102.
Cialdini, R. B., and N. J. Goldstein. 2004.
"Social Influence: Compliance and Conformity." Annual Review of
Psychology 55: 591-621.
eMarketer. 2012. "US Digital Ad Spending
to Top $37 Billion in 2012 as Market Consolidates" [accessed on June 12,
2014], Available at http://www.emarketer.com/
newsroom/index.php/digital-ad-spending-top-37-billion-2012-market-consolidates/
eMarketer. 2013. "The US Healthcare &
Pharmaceutical Industry 2013: Digital Ad Spending Forecast and Key Trends"
[accessed on June 12, 2014], Available at
https://www.emarketer.com/Coverage/HealthcarePharma.aspx
Flannagan, R. 2015. The Cost to Reach 1,000
People. Nuanced Media Analysis, [accessed on March 30, 2016]. Available at
http://nuancedmedia.com/cost-reach-1000-people/
Griskevicius, V., R. B. Cialdini, and N. J.
Goldstein. 2008. "Peer Influence: An Underestimated and Underemployed
Lever for Change." Sloan Management Review 49: 84-8.
Grunert, J. 2015. "What Is a Typical
CPM?" Chronicle Small Business article [accessed on March 30, 2016].
Available at http://smallbusiness.chron.com/typical-cpm74763.html
Heyman, J., and D. Ariely. 2004. "Effort
for Payment. A Tale of Two Markets." Psychological Science 15 (11):
787-93.
Hornstein, H., E. Fisch, and M. Holmes. 1968.
"Influence of a Model's Feelings about his Behavior and his Relevance as a
Comparison Other on Observers' Helping Behavior." Journal of Personality
and Social Psychology 10: 220-6.
Huesch, M. D., E. Currid-Halkett, and J. N.
Doctor. 2014. "Public Hospital Quality Report Awareness, Popularity, and
Sentiment." BMJ Open 4: e004417 [accessed on March 30, 2016]. Available at
http://bmjopen.bmj.com/content/4/3/ e004417.full
Keller, A., F. Strohschein, B. Lia-Hoagberg,
and M. Schaffer. 1998. "Population-Based Public Health Nursing
Interventions: A Model from Practice." Public Health Nursing 15 (3):
207-15.
Main, E. K., C. H. Morton, D. Hopkins, G.
Giuliani, K. Melsop, and J. B. Gould. 2011. Cesarean Deliveries, Outcomes, and
Opportunities for Change in California: Toward a Public Agenda for Maternity
Care Safety and Quality. CMQCC White Paper 2011 [accessed on June 12, 2014].
Available at http://wwwv.cmqcc.org/ resources/2079
Meeker, M. 2014. Internet Trends 2014- Code
Conference. Kleiner Perkins Caufield Byers, [accessed on June 12, 2014].
Available at http://www.kpcb.com/intemet-trends
Murray, D. M., R. V. Luepker, C. A. Johnson,
and M. B. Mittelmark. 1984. "The Prevention of Cigarette Smoking in
Children: A Comparison of Four Strategies.^" Journal of Applied Social
Psychology 14:274-88.
National Committee for Quality Assurance. 2011.
"Physician and Hospital Quality Report Card" [accessed on March 30,
2016]. Available at http://reportcard. ncqa.org/phq/ external/phqsearch.aspx
National Quality Forum. 2011. "NQF
Endorsed Performance Measures" [accessed on March 30, 2016], Available at
http://www.qualityforum.org/Measures_List. aspx
Nolan, J. P., P. W. Schultz, R. B. Cialdini, N.
J. Goldstein, and V. Griskevicius. 2008. "Normative Social Influence Is
Underdetected." Personality and Social Psychology Bulletin 34: 913-23.
Pew Research Internet Project. 2014. "Pew
Research Center" [accessed on June 12, 2014]. Available at
http://www.pewinternet.org/
Sakala, C., and M. P. Corry. 2008.
Evidence-Based Maternity Care. Millbank Report, [accessed on March 30, 2016].
Available at http://www.milbank.org/reports/
0809MatemityCare/0809MatemityCare.html#US
U.S. Department of Health and Human Services.
2006. Healthy People 2010 Midcourse Review. Washington, DC: U.S. Government
Printing Office [accessed on March 30, 2016]. Available at
http://www.cdc.gov/nchs/healthy_people/hp2010/ hp2 010_final_review.htm
Wagner, T. H., L. P. Engelstad, S. J. McPhee,
and R. J. Pasick. 2007. "The Costs of an Outreach Intervention for
Low-Income Women with Abnormal Pap Smears." Preventing Chronic Disease
[accessed on March 30, 2016]. Available at http:// www.cdc.gOv/pcd/issues/2007/jan/06_0058.htm
White, K. M., M. A. Hogg, and D.J. Terry. 2002.
"Improving Attitude-Behavior Correspondence through Exposure to Normative
Support from a Salient Ingroup." Basic and Applied Social Psychology 24:
91-103.
SUPPORTING INFORMATION
Additional supporting information may be found
in the online version of this article:
Appendix SA1: Author Matrix.
Appendix SA2: Building the Back-End--The Study
Website and Its Analytics Function.
Appendix SA3: Building the Front-End--The
Facebook Channel and Its Advertising Function.
Address correspondence to Marco D. Huesch,
MBBS, Ph.D., USC Leonard D. Schaeffer Center for Health Policy and Economics,
University of Southern California, Verna & Peter Dauterive Hall (VPD), 635
Downey Way, Los Angeles, CA 90089-3333; e-mail: huesch.marco@gmail.com. Aram
Galstyan, Ph.D., is with the Information Sciences Institute, University of
Southern California; Department of Computer Science, Viterbi School of
Engineering, University of Southern California, Marina del Rey, CA. Michael K.
Ong, M.D., Ph.D., is with the Division of General Internal Medicine and Health
Services Research, David Geffen School of Medicine at the University of
California; Veterans Administration Los Angeles, Los Angeles, CA. Jason N.
Doctor, Ph.D., is with the USC Leonard D. Schaeffer Center for Health Policy
and Economics, University of Southern California; School of Pharmacy,
University of Southern California, Los Angeles, CA.
Table 1: Daily and Overall Campaign
Performance Measures
Platform
Google
Search Facebook
Twitter Total
Daily (all nonunique)
Reach
92,100 33,685 17,507
143,292
Engagements 232 149
30 411
Campaign (all nonunique)
Reach
10,959,961 4,480,119 2,223,493
17,663,573
Engagements
27,676 19,923 3,798
51,397
Spending
$25,177 $13,689 $20,542
$59,408
Cost per 1,000 reached
$2.30 $3.06 $9.24
Cost per click-through
$0.91 $0.69 $5.41
Note. Reach or impressions
identifies number of users who saw or
potentially saw the advertisement
because they were exposed to it.
Engagements identifies number of
users who responded to the
advertisement by clicking through
the website link contained in the
advertisement.
Figure 1: Conceptual Framework
Marketing ... Safes
Qualified
Prospects Conversions
Prospects
* Has expressed
* Potentially a the desire to
consumer of consume
such * Has started the
maternal care info or
this engagement process with the
quality info has been info provider
inferred
* Has not * (May not yet
have
expressed a * Ready to consumed the information)
desire to start an
consume this,
engagement * (Has not yet
changed
or it has not process their behavior, or such
been inferred
behavior is not known by
study)
* Anyone who * Of all * Of all qualified
may potentially prospects,
only prospects, only those users
at some stage those users
of who have responded to study
be pregnant or social media ad on social media or
may know and
social social network by
'clicking
someone who may networks
whose through' to reach the study
be pregnant patterns
of website
use, shopping
and privately * ('Closing the deal' by
disclosed actually changing behavior
content and choosing higher quality
suggests an provider not ascertained in
interest in this pilot study)
pregnancy
* Targeted by
study ads
* Correspond to Health Teaching * Corresponds to an Outreach
and Social Marker ting types
public health pilot
of public health pilot
intervention
interventions
Huesch, Marco
D.^Galstyan, Aram^Ong, Michael K.^Doctor, Jason N.
No comments:
Post a Comment