Essay Instructions: Assignent:Read the article below and answer the following questions.
1. Do you think Lasik''s methods would be useful to other health care organizations?
2. How would you suggest a non-profit organization use Lasik''s targeting efforts?
Use 12 font and Times Roman. Have introduction and conclusion.
Article:
Please read the following article about Lasik eye surgery.
Developing a profile of LASIK surgery customers
Marketing Health Services; Chicago; Summer 2001; Frederick A Barber; Richard K Thomas; Mei-chih Huang;
Volume: 21
Issue: 2
Start Page: 32-35
ISSN: 10941304
Subject Terms: Health care industrySurgeryEyes & eyesightMarket strategyMarket segmentationStatistical data
Classification Codes: 9190: United States8320: Health care industry7000: Marketing9140: Statistical data
Geographic Names: United StatesUS
Abstract:
Customer profiling is standard practice in most industries, but has never really taken off in health care because the end user was historically not the one making the decision regarding choice of health service. However, with the resurgence in consumerism among baby boomers, the search for alternative therapies, and the emerging interest in defined contributions, the consumer may reign supreme in health care for the first time in decades. One area ripe for DTC marketing methods is elective surgery. Elective procedures make up a significant proportion of the procedures performed in health care. Customer profiling is used in an analysis to discover the best patients for LASIK eye surgery.
Full Text:
Copyright American Marketing Association Summer 2001
[Headnote]
To find the best candidates for elective surgery, start by envisioning your ideal customer.
Customer profiling is standard practice in most industries, but has never really taken off in health care. The lack of interest in customer profiling has rested on the same principles that discouraged direct-to-consumer (DTC) marketing and other targeted approaches to reaching health care consumers. Historically, the end user was not the one making the decision regarding choice of health service. Why spend resources on patients, the argument went, when their physicians or health plans were the ones ultimately determining which health services they used?
This argument, of course, makes less and less sense in today''s environment. With the resurgence in consumerism among baby boomers, the search for alternative therapies, and the emerging interest in defined contributions, the consumer may reign supreme in health care for the first time in decades. Pharmaceutical companies have spearheaded the growing interest in DTC marketing, but other health care organizations are quickly following suit.
One area ripe for DTC marketing methods is elective surgery. Elective procedures (i.e., those not considered medically necessary and thus not reimbursable under insurance) make up a significant proportion of the procedures performed in health care. These include everything from elective knee surgery to laser eye correction to face-lifts. These procedures are truly elective because it is usually the consumer and not the physician or health plan making the choice. Moreover, patients for many types of elective surgeries are characterized by a particular profile. Individuals obtaining face-lifts or tummy tucks do not represent a cross section of the population, and the same could be said for most elective surgeries.
Customer Profiling
Customer profiling has transformed into a science in other industries, with most companies maintaining massive customer marketing information files (CMIF). Not only do these files allow marketers to develop a profile of their customers, but they also can be used to dynamically track changes in customer composition or consumption habits. They can serve as a basis for cross-selling, follow-up promotions, and other marketing activities.
Companies in varied industries have used profiling to try to create a picture of their "best" customers. Best customers are typically defined as those who use the company''s services often, spend above-- average amounts, do not default on payments, or produce other desirable outcomes. A best customer for an elective medical procedure would be, for instance, someone who responded to a direct-mail solicitation, such as an invitation to a seminar, and then went through with the surgery or treatment. Private-pay customers can be considered best customers for many health care services, where negotiated rates are low and payment may be delayed. Identifying a profile of health care consumers most likely to fall into these desirable categories, therefore, can be a profitable move for a private health care practice.
One of the drawbacks of applying customer profiling in health care has to do with the state of data in the field. While health care probably generates more data than any other industry, the way it''s collected often limits its potential for customer profiling. Within an organization, the necessary data may be maintained in more than one data file, and certain useful types of data (e.g., income) may not be collected at all. Health plans are often good sources of such data, but are not likely to compile information on elective procedures because they are not covered under the insurance plans. Also, concerns about confidentiality limit access to patient data even within those organizations that collect it.
Marketers may get around this problem by developing generic profiles of patients for various services using utilization rates based on samples of patient records. For example, it might be possible to take the data sets generated by the National Center for Health Statistics (NCHS) and use the tens of thousands of records the NCHS compiles as a basis for creating procedure-specific patient profiles. Although there are limitations to the data available, researchers can profile users of services in terms of age, sex, payor category, region, and selected other characteristics. Although this level of detail is not perfect, in many cases it still offers significantly more than the information otherwise available for customer profiling.
Of course, the most effective approach involves obtaining actual patient data that will help create a reasonably accurate profile of the best prospects for that particular service. Aggregate data on patients obtaining face-lifts or elective knee surgery would provide the basis for identifying "look-alikes" within the population. Even better, however, are actual patient names and addresses. This information can help link the patients to a variety of consumer databases in order to develop a more in-depth profile of the typical candidate for a procedure.
Thus, patients for a particular service can be profiled in terms of their demographic characteristics, lifestyle clusters, and consumer behavior among other characteristics. Further, these detailed data can be subjected to sophisticated statistical analyses to empirically validate the profile. The following example demonstrates how this approach was used to identify prospects for laser eye surgery.
Discovering the Candidates
In many ways, laser eye surgery (LASIK) is the poster child for this type of analysis. This surgery has grown in popularity as new techniques have been perfected and outcomes have improved. LASIK surgery is almost always an elective procedure requiring out-of-pocket patient expenditure, it is relatively expensive, and it appeals disproportionately to certain segments of the population. The interest is particularly high in laser eye surgery today because of the potential represented by nearsighted baby boomers and the drying up of other types of surgical opportunities for ophthalmic surgeons.
The purpose i of this analysis was three-- fold. First, it aimed to provide a comparative profile of LASIK patients and prospects vs. the consumer population from which they were drawn. Second, it attempted to derive a predictive model to estimate the probability that a given consumer would self-select to have the surgery as a function of the available demographic data obtained from the Experian national consumer database. Finally, it sought to develop a methodology for extracting lists of consumers who appear to be look-alikes of past patients and, thus, candidates for LASIK surgery.
For this particular analysis, we obtained patient data from a successful laser surgery clinic. Although information could have been obtained on the characteristics of the patients, all we requested for this analysis were patient name and address. The clinic provided the names and addresses of three groups of patients: those on whom they had performed surgery; those who had presented themselves for surgery, but turned out to be clinically ineligible; and those who had indicated an interest in laser eye surgery at some point but, as far as was known, had never obtained the procedure.
The analysis involved approximately 700 LASIK patients (and medically ineligible prospective patients) and more than 1,200 prospective patients who did not have the surgery after an initial inquiry. In addition, a random sample of 1,000 consumers was drawn from the ZIP codes that predominated among the patients and prospects.
The random sample of consumers was extracted from the Experian Marketing Solutions national consumer database, based on a random selection of available households in the relevant ZIP codes. The Experian national consumer database contains name, address, and household demographic information on more than 180 million adult Americans in more than 105 million households. Experian, one of the three major consumer credit bureaus, compiles this database using credit "header" data (i.e., the name and address information of consumers supplied to those who offer credit), plus telephone listings, motor vehicle registrations, voter registrations, and other publicly available data sources. This database is widely used in the direct marketing industry and is considered one of the best compiled consumer files available. To ensure that the patient sample and the random sample represented the same underlying population, any consumer whose ZIP code fell outside of the primary area served by the LASIK clinic was removed from the research database.
The Experian consumer database provides three types of enhancement data for each consumer:
1. Individual-level data (e.g., age, marital status, and gender)
2. Household-level data (e.g., home ownership, estimated household income, and presence of children)
3. Geographic-area data, (e.g., median household income of the consumer''s Census Block Group, the racial and ethnic profile and median home value off the Block Group, as well as a lifestyle cluster code-the Experian Mosaic cluster-that classifies the ZIP+4 into one of 62 demographic and lifestyle groups)
A Glimpse at Results After comparing the characteristics of
the patients, prospects, and the matched general population, we found significant differences in the demographics and household characteristics of the three groups. Differences existed in terms of age, gender, race, marital status, income levels, and rates of home ownership. We found little distinction, however, between the patients and the other groups in terms of educational status or occupational characteristics. Comparing the groups in terms of Mosaic lifestyle clusters illustrated that four of these lifestyle clusters were negatively associated with a propensity for laser eye surgery. These clusters generally typified lower socioeconomic status neighborhoods.
Based on these comparative data, we derived a logistic regression model to predict the probability that a given consumer would fit the profile of previous LASIK patients. Prospects who failed to obtain LASIK surgery were eliminated from the analysis; the contrast groups were patients (including medically ineligible) and random consumers.
We evaluated a wide range of possible models. All available household-level and geographic area-level variables were regressed on the outcome variable (patient or medically ineligible) to determine those variables significantly related to the outcome. Variables with no significant multivariate correlation to the outcome were discarded at the final model. In addition, we ran tests of linearity and colinearity to determine whether it was necessary to transform variables to reflect nonlinear associations.
We ultimately settled upon a model that demonstrated reasonable predictive validity. Overall, 63% of the sample was correctly classified as either a patient or non-patient, based on the consumer data readily available from the Experian database. While a logistic regression model will yield both false positives and false negatives, the overall result, with 60% of the positives correctly classified, and 66% of the negatives correctly classified, indicated that using this model could significantly increase the probability of identifying prospects for laser eye surgery in the market served by the ophthalmic surgery practice.
After constructing a gains table to illustrate the distribution of propensities toward laser eye surgery for various deciles, we could determine the percent of each group (patients, candidates, and random consumers) present at each decile of the logistic model probability score. The top-three deciles contain 40% of patients, and only 12.5% of consumers. This represents a more than 300% gain in the probability of being a patient from the top-three deciles of the model vs. a random selection of consumers. This indicates that using the model should increase the probability of targeting likely patients in a DTC program by at least 300%. (See Exhibit 1.)
The predictive model eliminated some of the variables that showed promise on the basis of the frequency distribution. In the final analysis, the characteristics with the most predictive power (in no particular order) were race, age category, gender, marital status, and income. While these factors all contributed to the likelihood of laser eye surgery, certain occupational statuses and lifestyle categories reduced the likelihood of becoming
LASIK patients.
Sizing It Up
The initial profile of patients revealed characteristics that could potentially predict a propensity for LASIK surgery. We compared these characteristics to a comparable population from relevant ZIP codes and tested the relationships using a number of different predictive algorithms. Based on the "best fit" model, LASIK patients were found to differ from prospects and the general population along a number of dimensions, including age, race, sex, marital status, and income. Other characteristics considered salient during the initial analysis were eliminated when other factors were controlled for.
This approach made it much easier to predict classification of an individual as a patient vs. a non-patient. Using this model as a basis for identifying targets for direct mail would increase the "hit rate" significantly, and careful use of the model (i.e., targeting those that fall into the top-three deciles in terms of propensity) should increase marketing effectiveness by 300%.
EXHIBIT
Enlarge 200%
Enlarge 400%
Obviously, the bottom-line impact of these discoveries on an ophthalmic surgery practice can be substantial. A typical cost to print and distribute a seminar invitation for LASIK surgery would be in the range of $600 to $900 per 1,000 pieces mailed, including postage costs. Typical direct-mail response rates average around 1%. Therefore, a mailing of 5,000 invitations could be expected to yield approximately 50 prospective patients. The conversion rate of prospects to patients is typically 20% to 30%, so of those 50 prospects, only around 12 would be expected to go through with the surgery. Thus, the marketing program may have cost upward of $4,500, or $375 per acquired patient. A best fit model, on the other hand, may be expected to yield three times the response, with a higher conversion rate. If it generates 150 prospects, with a 40% conversion rate, that equates to 60 patients, or $75 per acquired patient.
These results are encouraging, but should be examined with some caution. The data were obtained from a particular service area characterized by a population profile that may differ from other localities. While it may be possible to generalize these findings to a certain extent, a broader sample of patients would pose fewer limitations. However, the influx of patient data from additional sites around the country will help resolve this issue. In addition, while these findings allow the analyst to identify populations similar to existing patients, they cannot predict the responsiveness of these look-alike patients to direct marketing campaigns.
The next step in this process will involve additional validation of the model, which will be carried out in several ways. Prospects in the defined market area will be scored using the model, and the top decile and the bottom decile will be targeted for follow-up. A telephone or mail survey will then be conducted with a random sample of these two groups to measure the gap in awareness of and receptiveness to the idea of LASIK surgery, as well as the presence of corrective lenses and/or contact lenses within the two groups.
A second approach would involve mailing a LASIK seminar invitation to a sample of top-scoring consumers, along with a comparison random sample of consumers in the area. The incremental response rate from the top-scoring group would validate the model''s usefulness for direct marketing applications.
Although this research is in its early stages, initial findings suggest it is indeed possible to profile high-propensity prospects for laser eye surgery and use this information as a basis for developing targeted direct mail campaigns. For more information on this research methodology or for access to information on LASIK prospect mailing lists, contact Conclusive Strategies at or Health Analytics at .
