Whether for personal or professional purposes, managers and leaders have learned and applied the SMART framework to create objectives that are clear, actionable, and achievable. The SMART mnemonic represents goals that are Specific, Measurable, Achievable, Relevant, and Time-bound.
Physician organizations have evolved balanced incentives for physician compensation, providing financial incentives for both productivity and quality. Most organizations measure physician productivity using work relative value units (wRVUs), which reflect the work intensity of care delivered and are independent of payer reimbursement. When creating quality goals for physician incentive compensation, smarter SMART goals provide additional layers to the SMART framework to create goals that are fair, explainable, understandable, and engender physician engagement, alignment, and support. Smarter SMART goals help ensure physician understanding of why the goals are important to patients, themselves, and the organization. Smarter SMART goals better motivate and enable physicians’ clinical behavior changes to close gaps between actual performance and desired performance.
Making SMART Goals Smarter
Smarter SMART goals employ additional mnemonics to help create reasonable and appropriate physician quality incentive compensation goals. As we work through these mnemonics, we’ll use clinical examples from the inpatient and outpatient care settings, using hospitalists and office-based physicians. The principles can be applied to all specialties.
Specific Goals
Add the mnemonic POUFCOP.
Process and
Outcome
Under
Full
Control
Of
Physician
This mnemonic reinforces the principle that the physician and only the physician should be accountable for the process and outcome of achieving their quality incentive goal. Achieving the goal is under the full control of the physician. This is especially important with regard to compensation. When money is at stake, it is unfair if the actions or inactions of other individuals affect whether a physician achieves a quality incentive goal.
Some examples:
Hospitalists and 30-day readmission rates. The hospitalist is one of many players participating in the patient’s hospital discharge and transition to outpatient care. The measure is not fully under the hospitalist’s control; they could execute a perfect discharge and handoff to the next clinician, yet the patient could be readmitted for reasons beyond the hospitalist’s control. Options to satisfy POUFCOP:
Choose a process measure (see POM below) that drives the desired outcome. For example, measure the rate of the hospitalist communicating with the patient’s primary care physician (PCP) at the time of discharge.
Make the measure more specific. Many 30-day readmissions, particularly after the first two weeks, are related to a different diagnosis, are unforeseen, and are not preventable. A 3-day readmission for the same diagnosis is more reflective of a “failed discharge” and may more appropriately measure the hospitalist’s clinical effectiveness in the transition of care.
Outpatient patient experience. It is fairer to measure and reward physicians on how the patient perceives the physician communicates with them. How a physician interacts with the patient in the exam room is POUFCOP. On the other hand, a patient’s likelihood to recommend the practice (LTR) is not POUFCOP; it is influenced by many non-physician factors, such as friendliness of support staff, ease of telephone access, and convenience of parking. However, LTR could be POUFCOP for the practice manager and managing physician, the dyad leaders of the practice who have control over most of the factors that influence the patient’s total experience and likelihood to recommend the practice.
Measurable Goals
Add the mnemonics POM, R2D2INS, NARDD, and AROMAS.
POM
Process or
Outcomes
Measures
As we saw in the hospitalist example above, if the outcome measure is not POUFCOP, you can consider measuring and incentivizing physicians for completing processes that contribute to achieving the outcome measure. Process measures reporting was more difficult in the days of paper records, but in the age of the electronic health record (EHR), with discrete data points and time stamps, process measurement can be easy.
As an example, at a hospital where I was chief medical officer (CMO), we developed a goal for the hospitalists to reduce the time between the Emergency Department (ED) physician and hospitalist agreement to admit the patient and the patient leaving the ED, to improve ED throughput and reduce ED overcrowding. An information technology (IT) specialist created a report that captured the time of the bed request, reflecting the mutual decision of the ED physician and hospitalist to admit the patient, the time the patient left the ED, and the name of the admitting hospitalist. We shared the reporting of the mean, median, maximum, and minimum for this time interval for each hospitalist.
R2D2INS
Rates
Require
Defining
Denominator
Inclusion
Numerator
Satisfaction
Rates are important for quality measures, as they reflect the number of times a process or outcome occurs — the numerator — as it relates to the number of opportunities for that process or outcome to occur — the denominator. If the process or outcome is desired, a higher rate reflects the effectiveness of the physician in completing the process or achieving the outcome.
Reporting a rate is preferable to reporting the total number of processes or outcomes, which depends on a physician’s panel size, which is influenced by their time in practice or full-time equivalent (FTE) status. As an example, physician A completes 210 annual wellness visits (AWVs) per year. Physician B completes 500. Is physician B more effective than physician A? By examining the opportunities, or the denominator, we observe that physician A, with two years of practice, completes 210 AWVs out of 300 eligible patients, for a rate of 70%. Physician B, with 10 years of practice, completes 500 AWVs out of 1,000 eligible patients, for an AWV rate of 50%. Physician A has a higher performing AWV rate and is more effective at ensuring their patient panel receives AWVs.
Also determine the improvement impact — the total number of patients affected by an improved rate. Consider a desired AWV rate of 80%; Dr. X’s 60% rate is 60 out of 100 patients; Dr. Y’s 60% rate is 600 out of 1,000. Applying resources to improve to 80% results in Dr. A completing AWVs on an additional 20 patients, and Dr. Y completing AWVs on an additional 200 patients. The underperforming physician with a larger denominator has a greater improvement impact, so applying improvement resources to that physician will be more impactful and generate a greater return for the investment, a greater bang for the buck.
The rate must clearly define how an opportunity is included in the denominator, denominator inclusion, and how that opportunity is satisfied, numerator satisfaction.
One aspect of denominator inclusion is patient attribution. For hospitalists, this can be a conundrum to assign the patient to a particular hospitalist for measuring and reporting quality. Larger hospitalist programs often separate admitters from rounders, so a patient should not be attributed to an admitter for a measure that reflects care rendered during their entire admission.
Admitters can have their own set of admission quality measures; for example, the hospitalist ED throughput measure discussed above, rate of order set usage rate, or rate of code status being addressed. A rounder discharging a patient on the first day they assume care is less impactful on the outcomes of that patient’s hospital care, so many programs assign the patient to the rounder with the plurality of visits.
Patient attribution in the office setting can be determined by the PCP or specialist physician designated in the patient’s EHR, the physician with the plurality of the patient’s office visits, or the physician who provides comprehensive services to the patient, such as an AWV or transitional care management (TCM) service.
Once patient attribution is determined, it must be established by what criteria the patient is included in the denominator. For a diabetes quality measure, how do we determine if a patient is diabetic? For a defined look-back period, is the patient included if they have diabetes on the problem list or a visit diagnosis of diabetes? Remember that with ICD-10, that can be a long visit diagnosis list. Or an A1C greater than 6.5, a fasting blood sugar greater than 125, a non-fasting blood sugar greater than 199? Or all the above?
Leaders can also use currently available and accepted inclusion and satisfaction criteria, such as those used in HEDIS or CMS quality measures.
Numerator satisfaction should be determined by measurable data and reports, not patient reporting. The data source can impact the fairness of the measure. Data can be obtained from EHR clinical data, practice billing data, or payer claims data. Each source has its pros and cons.
It is important to recognize that the EHR data reflect the care rendered to the patient, which may not be consistent with billing or payer data. As an example, as our medical group AWV rates were increasing, our compliance department delayed filing AWV claims for several months to allow time for auditing the documentation and coding of every AWV. AWV rates were reported on a quarterly basis; consequently, AWVs performed during the quarter with claims held and submitted after the end of the quarter led to lower payer data source AWV rates than EHR data source AWV rates. Billing functions and payer decisions were not POUFCOP. The EHR data source accurately reflected the services provided by the physician and received by the patient during a defined time interval, which were POUFCOP and fairer to physicians.
NARDD
Never
Average
Rates with
Different
Denominators
A simple example illustrates this mnemonic. Dr. Slick, managing physician of Swell Primary Care, claims that his practice met the AWV rate goal of 80% and demands that his practice receive its AWV quality incentive payment. Swell Primary Care has three office locations: Northside, Eastside, and Southside. Their AWV rates:
Northside: 90%
Eastside: 80%
Southside: 70%
Dr. Slick averages the rates (90% + 80% + 70%) and divides by 3, which is 80%. However, when we drill down into the numerator and denominator of each location, we observe the following:
Northside: AWVs performed for nine out of 10 eligible patients.
Eastside: AWVs performed for 80 out of 100 patients.
Southside: AWVs performed for 700 out of 1,000 patients.
The true aggregate rate, which is the sum of the numerator (9 + 80 + 700) divided by the sum of the denominator (10 + 100 + 1000), is 71%.
When aggregating rates, always calculate a weighted average, or divide the total of all numerator data by the total of all denominator data, which achieves the same result.
AROMAS
Automated
Reporting
Over
Manual
Abstraction and
Sampling
As discussed in the examples above, the EHR provides opportunities to capture and report discrete data. Chart review and manual data extraction are highly dependent on the accuracy and completeness of notes created by clinicians, which often vary.
Manual abstraction is time-consuming and expensive, and subject to human error, including transcription mistakes and misinterpretation of clinical language. Sampling increases the potential for bias if the sample is not randomly selected, is too small, or is nonrepresentative of the entire population.
To be accurate and fair, use automated EHR reporting. This requires IT experts who are adept at creating and validating EHR reports. Validation before implementation is critical. The report must undergo testing, validation, and a trial period of measurement and reporting to confirm its accuracy and reliability. Physician compensation and trust are at stake.
Achievable Goals
Add the mnemonics APIB, DULLE, ASPIRE, BRITE, and CPI.
APIB
Absolute
Performance or
Improvement from
Baseline
APIB becomes a concern when the quality incentive measure represents an opportunity to improve from current performance to desired performance, and that performance difference is large. There is only so much improvement that can be achieved in a defined interval of time, and to be fair, when physician compensation is at risk, the goal should be achievable in that time interval.
As an example, if a practice’s AWV rate is 20% and the desired rate is 80%, the practice is unlikely to achieve the desired rate in one year. Incremental improvement targets would be fairer, such as 40% for year one, 60% for year two, and 80% for years three and thereafter. As another example, patient experience survey vendors can use their advanced analytics to provide reasonable and achievable improvement targets for a defined time interval.
DULLE
Do
Upper or
Lower
Limits
Exist above or below which further improvement is not expected to occur?
Returning to the AWV rate example, where a higher rate is better, the AWV rate is unlikely to ever be 100%. With the older Medicare patient population, some patients alive at the beginning of the measurement period may pass away before receiving an AWV. Patients relocate and change practices. Snowbirds may have their AWV completed in their sunny winter locale. A reasonable upper limit may be 80%.
Going to the inpatient side and deep venous thrombosis (DVT) rates, where a lower rate is better, inpatient DVT rates are unlikely to be 0%. Published studies of anticoagulant prophylaxis show a reduction in DVT rates, but not elimination. DVTs still occur despite adequate prophylaxis.
Sometimes, explaining these concepts to non-physician leaders can be challenging. They may label these explanations as cop-outs or excuses. Using analogies can enhance understanding. No one expects a baseball player to hit for a .500 batting average, nor a track and field athlete to run a two-minute mile.
ASPIRE
Always
Share
Performance
Improvement
Resources with
Everyone
Always provide resources to improve physician performance. Improvement resources can be obtained from external sources, such as colleagues, organizations, or published literature. You can also learn internally from higher performers and share with lower performers.
After the quality incentive goal measurement is created and validated, determine the internal distribution of performance across the physicians being measured. Identify the high performers and talk with them to learn their “secret sauce.”
At the beginning of our medical group’s efforts to improve AWV rates, we identified a PCP practice with remarkably high AWV rates. We met with the practice managing physician, who related that he had ascertained, and shared with his practice team, that AWVs were a win-win — a beneficial wellness and preventive service for patients, with no copays or deductibles, and beneficial for the practice, providing generous wRVUs and reimbursement.
He and his team developed practice workflows to enable high AWV rates, such as offering patients a “free” AWV at the time of an evaluation and management (E/M) visit, developed documentation templates for AWVs and combined AWV and E/M visits, and obtained a favorable opinion letter from Medicare for these templates. Had we not determined the internal distribution of AWV rate performance, identified this high-performing PCP practice, learned their secret sauce, and shared it with all PCPs, our medical group’s AWV rate performance would not have improved as rapidly as it did.
BRITE
Blinded
Reporting
Initially
Transparent
Eventually
Initially present performance data to physicians that is comparative but blinded except for the recipient, who only sees their own data and their practice’s data; all other physicians and practices are blinded. As physicians’ comfort level with data transparency improves, you can transition to sharing data comparatively and transparently, identifying all physicians and practices by name.
Given the competitive nature of physicians, sharing comparative and transparent data can drive performance improvement in the absence of any other intervention.
CPI
Celebrate
Performance and
Improvement
Once physicians are comfortable receiving performance reports that are comparative and transparent, you can begin to celebrate both high performers and improvers. When communicating interval and final performance (see DIOMAR below), call out not only high-performing physicians and practices, but also the improving physicians and practices. This encourages continued high performance and continued improvement toward the goal.
Relevant Goals
Add the mnemonic RPPO.
Relevant to
Patient
Physician and
Organization
The ideal quality incentive goal is one that is relevant to all parties: the patient, the physician, and the organization. Using these criteria avoids the mistake of reporting what can be measured, regardless of relevance.
Patient experience is an example of RPPO being satisfied. Improved patient experience is relevant to patients, as it is associated with improved compliance, improved health outcomes, and reduced likelihood of filing malpractice suits.
These associations are relevant to physicians as well. Improved patient experience is associated with improved physician resilience, improved patient acquisition and retention, and greater practice growth.
Improved patient experience is relevant to the organization for all the reasons stated above, which are associated with organizational quality outcomes, growth of new patients, retention of existing patients, and reduced liability risk.
It is also important to ensure that the physician quality measure is relevant to the physician’s specialty. Patient experience applies to all specialties. AWV rates apply to PCPs. For office-based specialists, quality incentive measures may need to be developed in collaboration with the specialist to better achieve relevance and are often process measures.
Choosing hospital-based specialist physician quality incentive measures is an opportunity to collaborate with the specialty physicians and hospital leadership, and utilize hospital quality measures relevant to the specialist and the hospital.
Many specialty organizations have published recommendations for clinical testing or treatments to perform or not perform based on evidence supporting necessity, lack of duplication, and freedom from harm. Although tempting to use one of these recommendations as a quality improvement measure, keep in mind that data reporting may be challenging, and the recommendations may be so widely promulgated and accepted that the unrecommended services are rarely, if ever, performed, and the recommended services are almost always performed, with the few exceptions based on patient circumstances and preferences.
Time-Bound Goals
Add the mnemonic DIOMAR.
Define
Intervals
Of
Measurement
And
Reporting
Returning to AWV rates as an example, if the AWV quality incentive is determined by the AWV rate over an interval of measurement of a calendar year, measurement and reporting should occur more frequently to allow physicians to monitor progress toward the goal, apply improvement tools, and celebrate ongoing improvement with their office team. Quarterly measurement and reporting are reasonable. The number of measurements, or “n,” can help determine the frequency of measurement and reporting; a smaller “n” requires a larger interval of measurement and reporting.
As an example, if your patient experience survey vendor informs you that thirty survey responses are required for validity, the interval of measurement and reporting must be a time interval that includes at least thirty survey responses.
Conclusion
In conclusion, using smarter SMART goals for physician quality incentive compensation adds additional depth to the SMART goal framework, helping to ensure that physicians are treated fairly, understand and support their quality incentive goals, and are more likely to achieve these goals, to the benefit of patients, physicians, and the organization.
