Well-informed physicians improve prescribing trends
A quality and cost success story
By Dale Brown, Director
Product & Proposal Strategy
Caremark Inc., Northbrook, IL
Information-based technology continues to revolutionize the delivery and quality of health care. The ongoing challenge for providers of health care services is to ensure that physicians, patients, and clients are receiving, understanding, and utilizing information to improve patient health outcomes. In a physician-focused effort to increase the utilization of drug prescribing information, Caremark has completed the first 12 months of a physician-profiling initiative involving the prescribing analysis of 117,904 physicians. These physicians potentially care for 503,470 beneficiaries of payers with prescription drug expenditures totaling more than $164,773,000.
Because of the initiative’s magnitude the large number of doctors, patients and prescriptions involved it has impacted physician prescribing trends. The objective of this paper is to outline the initiative, its findings, and its clinical and financial implications. The importance of this initiative stems from its scope and from providing physicians pertinent information they cannot get elsewhere. The information is specific to their prescribing, is statistically valid, and is communicated to the physician as a partner in the initiative.
There are four major goals associated with the initiative. First, the initiative combines prescription data sets of four large payers based in Chicago. Second, it analyzes physician prescribing trends through a physician-by-physician measure. Third, the physician-specific information is communicated to each physician who wrote an adequate number of prescriptions. Last, the change in physician prescribing behavior is measured and communicated.
After 12 months, the findings reveal a number of clinical and financial implications related to several core variables, including: the generic substitution rate and the physicians’ acceptance of clinical recommendations made by the payer's prescription benefit manager’s clinical pharmacists.
The first stage of the initiative involved aggregating data representing 2,740,998 prescriptions dispensed for the 503,470 employees, retirees, and their dependents of four large payers in the Chicago area. A proprietary system algorithm produced a physician-by-physician profile of the prescribing trends for each of the 117,904 physicians providing care for the patients covered by the payer groups. This initial analysis established the baseline for future comparison and measurement.
The core variables measured were:
• generic substitution rate;
• the physicians’ acceptance of clinical recommendations made by the payer's prescription benefit manager Caremark.
Exhibit 1 illustrates the high concentration of patients and prescribing within a small population of physicians, and it illustrates the influence that a small percentage of physicians have on the four payers’ prescription drug expenditures. Of the 117,904 profiled physicians, 8,253, just 7% (see Figure 1), prescribed 50% of the 2,740,998 prescriptions (Figure 2) and accounted for 51%, of $164,773,160 in expenditures (Figure 3).
A summary prescribing profile was sent to the 8,253 physicians having the largest impact, with the purpose of sharing aggregated prescribing data. The profile provided the physicians with the following items:
• their specific prescribing data for the payers and patient set;
• comparable prescribing data of other physicians in the same specialty;
• information regarding best demonstrated cost practices in that specialty.
Included with the profile was a letter that explained the initiative, its clinical relevance, and the critical role the physician plays in the initiative’s success. Pursuant to the initial communication and establishment of the baseline, this process was repeated over two consecutive six-month periods to establish a record of longitudinal changes and trends.
Core Study Findings
Our hypothesis was that physicians would act on the information provided according to its relevance and specificity. This is supported by the fact that aggregate and normative data are otherwise impossible for physicians to obtain due to the prohibitive information systems requirements and disparate systems used to collect and maintain data across the health care industry. Physician response to this information was positive.
Significant changes in prescribing were documented (see Exhibit 2, p. 186):
• In acceptance of clinical recommendations made by the prescription benefit manager (PBM), the rate increased from 64.83% to 77.5%. This represents a realization of 36% of the upside potential, driving both clinical enhancements to the therapy prescribed and cost savings.
The PBM’s clinical recommendations aid physicians in delivering high-quality patient care. This is achieved through providing the physicians with information that would otherwise be inaccessible. This information includes compliance patterns, patient drug histories, and the drugs prescribed by other physicians for particular patients.
• In generic substitution, the rate increased from 86.89% to 92.34%. This increase represents a realization of 42% of the upside potential available in cost savings.
In addition to the three core variables measured, multiple disease and drug-specific clinical studies have been launched following the same format. Here is an overview of the first two of these studies that were set forth, as well as their preliminary results.
This initiative was implemented to address the high utilization rate of selective serotonin reuptake inhibitor (SSRI) antidepressant drugs (such as Prozac and Zoloft). After establishing a baseline for SSRI prescribing, the relevant facts impacting prescribing in this drug category, as well as physician and patient specific data, were communicated to the physicians. Issues relevant to prescribing these drugs include appropriate indications for use, fraud, multiple prescriptions, and cost.
The pursuant changes in SSRI drug prescribing were recorded:
• The prescribing of SSRI drugs for the patients profiled in this initiative fell by 9.3%.
• Prescribing for SSRI drugs for a random sample of payers not involved in physician profiling increased on average by 12.9% in 1996.
The resultant cost saving is illustrated in Exhibit 3 (see p. 186).
This initiative was implemented to study the use and dosing of early- and later-generation antibiotics for a variety of infections and to communicate the dramatic cost differences between them (about $9 vs. $90 per prescription). Unneces-sary antibiotic use and the clinical costs of bacterial resistance have been estimated at $7.5 billion annually. This initiative was derived from prudent antibiotic use being consistent with high-quality care and cost-effectiveness.
The initial measurements established that 27% of the volume of all antibiotic prescribing fell into the targeted categories, principally fluoroquino-lones (i.e. Cipro, Floxin) and the newer macrolides (i.e. Biaxin, Zithromax). However, this same volume represents more than 60% of the expenditures for antibiotics.
This initiative provides information to physicians to assist them in optimal treatment, which results in a shift in initial prescribing of later-generation antibiotics to more clinically appropriate and cost-effective early-generation antibiotics. We predict that by educating physicians on the over-zealous prescribing of antibiotics, the prescribing volume of these drugs will indeed decrease.
The physician profiling initiative is expanding with more payers becoming involved both regionally and nationally. Caremark is in the process of gauging physician satisfaction within the initiative. The initiative will continue to expand clinically in the number of disease-specific clinical variables being measured, analyzed, and communicated to the physicians. Exhibit 4 (see p. 187) shows the clinical initiatives under way for 1997.
Clinical and Financial Implications
The study resulted in the following observations about physicians’ prescribing practices:
1. Improving clinical recommendation acceptance and generic substitution has an immediate impact as well as a long-term impact. Once a behavior pattern is changed, physicians will continue to prescribe the more appropriate drug therapy and the cost-effective generic drug on an ongoing basis. This leads to improved clinical outcomes and immediate and long-term cost reductions.
2. Changing prescribing behavior not only impacts the plan’s specific cost but also flows into the general population, reducing unnecessary health care costs for patients outside of these payers’ patients. This supports community and national health care reform led by information.
3. The physician profiling initiative confirms the value of information-driven change in health care, thus leading the way for other information-led improvements and efficiencies.
In summary, the study’s findings support the hypothesis that physicians are willing to change prescribing based on clinical facts and patient-specific information. Such information-driven programs provide doctors with tools they can then utilize to positively impact the efficiency and quality of patient care.
(Editor’s note: Sarah Smith, product strategy analyst for Caremark Inc. in Oakbrook, IL, contributed to this report.)