Data quality is the foundation of any data warehouse. As the old saying goes “garbage in – garbage out.” If there are inconsistencies or irregularities in the data loaded into a data warehouse, all analyses based on those data are potentially flawed.
Healthcare data are complex, and even data from organizations well experienced in claims administration and data transmission contain errors from time to time. Having processes in place to detect and correct data problems is critical for any healthcare information platform, but even with the best of processes, errors from submitted data result in increased costs and often delay data warehouse updates.
Many MedInsight clients are dependent on multiple data supplier partners to provide source data for their healthcare data warehouse. Some are of these clients are self-insured employers requesting data from the organizations who administer their claims. Others are community or business coalitions and some are states building all payer claims databases. As the number of data sources increases, the complexity and potential for error increases.
Establishing strong partnerships with all your data suppliers is the one of the best ways support efficient data warehouse updates.
Different organizations have different relationships with their data suppliers, so the possible techniques for keeping data suppliers engaged in the success of a data warehouse often vary from one client to a next. Below are some ways to engage data suppliers in the project.
- Notify data suppliers as early as possible when considering the implementation of a data warehouse. Ask about any technical or legal constraints that may impact their ability to fully participate.
- Clearly define the data elements you will need to support the analyses you plan to produce, but don’t dictate a specific layout. If the data supplier can provide the required data elements in a layout they have already developed for other projects, this will greatly reduce the amount of time they need to produce their initial data submission and reduce the risk of error.
- If appropriate, and agreed to by all parties, offer to share information back to the data suppliers. Creating a deliverable of value back to the data suppliers can help keep them interested in the success of the project.
- If data are being purchased from the data supplier, or if the submission of data is included as part of a broader service contract, consider adding financial incentives (or penalties) into the contract for data being submitted on time and without error.
- Ask the data suppliers to describe the types of data audits they will run on the data prior to submitting files.
All organizations are busy. Claims administrators have to focus on many different priorities to meet the needs of their customers, but if data suppliers are fully engaged in a data warehouse initiative, and incorporate data quality processes with each data submission, everyone will save time and money by avoiding rework.
The use of risk adjustment in provider reimbursement arrangements has increased as alternative payment arrangements are becoming more widespread in health insurance. Risk adjustment has been used by Medicare Advantage and managed Medicaid plans to reimburse health plans for the unique risks and populations in their care. More recently, as carriers have transferred utilization risk to providers through alternative payment arrangements, such as global budgets and bundled payments, risk adjustment has been used to reflect a provider’s patient’s severity.
However, many risk adjustment methodologies were developed using a standard population representing a combination of adults and children. Adults comprise a larger proportion of the average population, and as a consequence, the disease states recognized in these methodologies were optimized with greater emphasis on adults. Since a chosen risk adjustment methodology should reflect the characteristics of the underlying patient population organizations such as children’s hospitals, pediatric provider groups, and health plans that enroll a large proportion of children, health organizations have begun to question these standard risk adjustment models. These groups argue that there are fundamental differences in clinical profiles, patient mix, treatment options, and patient management needs between the pediatric population and the general population.
In a recent Milliman research paper, we built a control model for a standard commercial population and a model optimized for a pediatric population from the Truven Health Analytics MarketScan®  database using the open source hierarchical condition categories (HCC)  system to compare results. We limited our focus to New England States  and developed a concurrent  risk adjustment model with 184 disease classifications based on the HCC system.
The R-Squared value  in our control model is 58% on the standard population. This is very similar to the reported R-Squared values for many commercially available concurrent risk adjusters . However, if we remove the adults from this population, our model’s R-Squared reduces significantly to 45%.
To improve upon the control model’s R-Squared of 45% for pediatric-only populations we developed the pediatric-only model through an iterative process using only the pediatric population included in our MarketScan database sample:
This pediatric risk adjustment model results in an R-Squared of 58% on pediatric populations, which is a significant improvement from the control model’s R-Squared of 45%. This increase in statistical fit will impact the financial results of organizations bearing financial risk for pediatric populations. For example, using the pediatric-only model on children in the data used to develop our model, results in a risk score that is approximately 1.5% higher than the control model developed for a standard population.
- We began the modeling at the DxGroup level that underlies the HCCs. There are 784 DxGroups in the original HCC classification system.
- We modeled DxGroups with more than 30 patients separately and left those with less than 30 patients in their original HCCs.
- We created two-way and three-way disease interactions for inclusion in the model (e.g., diabetes and chronic obstructive pulmonary disorder (COPD) would be included as an additional explanatory variable, in addition to diabetes alone, and COPD alone). We calculated the sample size of each and retained only those that had at least 30 patients in a cell.
- We regrouped DxGroups and disease interaction terms with statistically insignificant coefficients (at a 5% significance threshold) with the other small-cell DxGroups in the same HCC and recalculated their coefficients (risk weights).
- We reset the coefficients of DxGroups and disease interaction terms with statistically significant but negative coefficients to zero. Negative coefficients often imply a confounding variable; if left in the model, they will produce spurious relationships among conditions. From a payment perspective, negative coefficients result in reduction in payment for diagnosing or treating a condition, which does not have face validity either.
- We repeated steps (4) and (5) until all variables left in the model had statistically significant and non-negative coefficients. This resulted in 570 DxGroups/HCC categories.
In alternative payment models which use risk adjustment to distribute payments to providers, models calibrated using a standard population could result in inequitable reimbursement to providers specializing in pediatric populations. As a result, these providers should carefully review the risk models used in any alternative payment arrangement before participation.
 Truven Health Analytics MarketScan® is a large and nationally representative commercial claims database. It is used to develop risk adjustment tools by many vendors of commercial risk adjustment tools.
 The HCCs are used in Medicare Advantage and Part D plans, in the federally administered risk adjustment model for commercial individual and small groups starting in 2014, and in several states’ Medicaid and subsidized insurance programs. The HCCs used in all of these systems have not been calibrated for a pediatric population.
 We only used claims in New England States – Maine, Massachusetts, Connecticut, New Hampshire, Rhode Island, and Vermont for model development.
 A concurrent model uses the current year’s data to risk adjust total cost of care within the year. We chose to develop a concurrent model because many recent global risk contracts retrospectively use risk adjustment at settlement.
 The R-Squared statistic measures the amount of variability a model is capable of explaining in a population and is often used to evaluate the effectiveness of a risk adjustment model. A more accurate model results in a higher R-Squared value.
Rationale for Developing the Initial 20 Measures
The concepts for the measures for the newest MedInsight product, the Health Waste Calculator, were identified from various publications about avoidable healthcare, medical waste, and Choosing Wisely. The team identified an initial list of 134 measures to develop. The prioritization of the initial 20 waste calculator measures, version 1 and 2 (see Appendix I below), was based on the following criteria:
- High prevalence rate or incidence of the wasteful events as reported in different publications
- High cost impact due to the wasteful events
- Representation of different specialties or clinical conditions
- Representation of different types of services – preventive screening tests, diagnostic tests and prescription of drugs
- Representation of relevant measures for different age groups (children, adults, elderly or all population) as well as gender specific measures
The following examples highlight a summary of evidence that supported the prioritization of these measures in version 1.
- Rhinosinusitis related three measures – The prevalence of both acute and chronic rhinosinusitis in a National Health Interview Survey conducted in 2004, found that 14% of people in the US population were diagnosed with sinusitis. Rhinosinusitis accounts for health-care expenditure of more than $3.0 billion per year in the United States. Despite the fact that CT imaging has moderate sensitivity (76%) and specificity (79%) in diagnosing acute bacterial sinusitis, it is still being ordered. In another survey data it was found that 81% of adults presenting with acute sinusitis were prescribed antibiotics. Rhinosinusitis is still the fifth leading indication for practitioners to prescribe antimicrobials despite the fact that 70% of patients improve without antibiotics.
- Imaging for low back pain – Approximately 15 percent of the U.S. population reports having frequent low back pain. Healthcare cost for back pain includes both direct medical expenses and indirect cost of time lost from work, disability payments and diminished productivity. Additionally, medical care for these individuals cost approximately $35 billion dollars, with imaging driving much of the cost. In the calendar year 2000, Medicare made approximately $55 million in combined technical and professional component payments for 1.36 million claims for conventional radiography of the lumbar spine (For CPT code 72100) and $339 million for 592,000 lumbar spine MR imaging claims.
- Pap smear – Pap smear is the most common screening test conducted in women over 21 years of age and in a survey conducted by the National Center for Health Statistics, 93% of American women report having had at least one Pap smear in their lifetime. Among women with no history of abnormal smears, 55% undergo Pap smear screening annually, 17% report a 2-year screening interval, 16% report being screened every 3 years, and 11% are not being screened regularly. Even the very elderly report frequent screening where 38% of women age 75 to 84 and 20% of women age 85 and older reported annual Pap smears. Ideally in women with no history of abnormal smears the U.S. Preventive Services Task Force (USPSTF) recommends screening for cervical cancer in women aged 21 to 65 years with cytology (Pap smear) every 3 years.
- Cardiac stress testing – Half of all patients in community practice had 1 or more stress tests within 24 months of coronary revascularization. Of those tested, only 5% required repeat revascularization. From 1993 to 2001, roughly 180,000 cardiac revascularizations, 360,000 catheterizations, and 1 million stress tests and nuclear imaging studies were performed on a 5% national sample of Medicare beneficiaries. As overall rates of stress testing in the US Medicare population doubled from 1993 to 2001, the proportion of tests performed with radionuclide imaging increased from 50% to 80%. The nearly 3-fold absolute rise in US rates of radionuclide imaging suggests that these imaging studies have become the standard of clinical practice without clear evidence to support their routine use in place of exercise tests without radionuclide imaging
Further an analysis was conducted on a set of claims data for a small Midwestern HMO health plan to find the average cost and the number of events that were wasteful out of a total of 10,074 enrollees, about 1% of their total costs were classified as wasteful using the first version of this product. A sample of this claim cost analysis is shown in the table below.
Over the last few years one day hospital stays have been a focal point for medical management efforts to convert these to observation stays. In addition the 2014 Medicare’s Inpatient Prospective Payment System’s final rule on inpatient admission defines an “appropriate” inpatient admission as one that in the judgment of the admitting physician requires a hospital stay of at least two midnights or in medical management terms, a two day length of stay (LOS). Patients not meeting this criteria but needing inpatient hospital care lasting past one midnight but less than two will be classified as observation cases. The combination of these and other factors is leading to an increase in utilization for observation. So is this a positive outcome from a cost management perspective or is it problematic? This is becoming an important question for population management and one that does not have a simple answer.
A not uncommon situation is one where the reimbursement levels for inpatient hospital admissions and observation stays are not rationale. By this I mean that observations stays are paid more than if the patient were admitted as a regular inpatient admission. Avoiding an admission through the use of observation may be the right thing to do in this situation from a resource efficiency perspective but may actually cost a payer more than the hospital admission that is being avoided. Regardless, the overall trend is to move patients to observation status when this is appropriate clinically.
So how should observation status utilization be measured and what benchmarks should be used to monitor results? If we see an increase in observation utilization is it a desired outcome or should we be concerned? The typical practice of measuring billed units may not be the best approach since observation stays are billed using different units depending on how payer hospital contracts are designed so some observation services may be billed in hourly increments in some settings, in 24 hour increments in others and as a generic per observation episode in others. Our recommendation is to use “observation episodes” as the unit for measuring utilization. An observation episode is measured using the same logic as an inpatient stay, each midnight occurring during an observation stay counts as 1 observation unit with a minimum value of 1 for observation stay not spanning midnight. As examples, an observation stay starting at 4 AM and ending at 8 PM would be 1 observation unit, a stay starting at 4 AM and ending at 1 AM would also be 1 observation unit and finally a stay starting at 4 AM and ending at 1 AM after two midnight would be 2 observation units. This allows us to directly compare observation unit utilization with inpatient hospital utilization and one goal for directing patients towards observation is to reduce inpatient utilization.
Now for the question of how do we determine if our rate for observation utilization is positive or a problem we need to address. Since observation is a substitute for short stay hospital admissions we need to look both at observation utilization and short stay hospital utilization. We recommend combining observation episode with 1 day LOS hospital admissions to produce a combined utilization rate and then comparing this to either a historical target or a benchmark from a source such as Milliman. An example taken from a Milliman analytic tool (MedInsight Guideline Analytics) is shown below.
Using this example the target utilization is the “combined Obs + 1 Day LOS” or 27.7/1000. Our actual is 28.3 or a bit higher than we would like. In addition, our 1 Day LOS utilization is higher than the benchmark while our Obs utilization is less meaning we should have opportunity to shift more 1 day admits cases to Obs without causes excess utilization.
As we are all aware, the Affordable Care Act contained specific regulations that govern payers’ Medical Loss Ratio or MLR. These rules set minimums for the amount of the premium dollar that plans must spend on benefits. If we think of a premium as having three components: Benefits, Administrative cost, and profit or surplus, by setting a minimum for the size of the benefit component, the ACA essentially set maximums for administrative expense and profit.
These restrictions created new pressure for plans to manage their administrative expense as this is the primary opportunity for increasing profit or surplus. In reality, changing a payer’s administrative cost structure can be a challenge: It takes a disciplined approach; it may actually require increased spending through investments in technology and other efficiency improvements; and it doesn’t happen overnight.
Regardless of these challenges, organizations must find ways to manage their administrative expenses. To help organizations, we have identified five best practice approaches that organizations can use to support this work.
- Develop a defensible and accurate way to allocate administrative costs. Organizations must ensure that they are appropriately allocating administrative costs among lines of business. Not all product lines are subject to MLR reporting requirements, and thus it is important to ensure that costs are appropriately allocated to the right products based on cost-generating activities. Best practice organizations use a cost allocation model that uses quantifiable data to allocate costs and generate line of business financial reports.
- Employ an enterprise effort. Administrative cost management isn’t just finance’s problem—it requires an enterprise focus from managers and front-line staff throughout the organization. Efficiency improvements can come from anywhere within the organization. Any administrative cost management project requires leadership and stakeholder engagement, organizational understanding and buy-in, and transparency.
- Use benchmarking to set targets and understand what is possible. Benchmarking helps organizations understand how their own costs and performance stack up against the competition. Use of benchmarks can help identify opportunities for process or organizational renovation, estimate the potential savings from specific initiatives based on efficiency improvements, or even identify areas where additional investment is appropriate.
- Track and trend improvements over time. At the beginning of any cost management project, leadership should establish targets (benchmarks can give targets credibility). These targets should include both the overall goals (e.g., departmental administrative cost levels) as well as metrics related to drivers of cost reduction (e.g., efficiency, production, and quality). Over the course of the project, the organization should monitor changes and report progress so that participants can see progress being made. Best practice organizations use systems and tools such as dashboards, to report information throughout the organization.
- Ensure incentives are aligned to achieve desired outcomes. Many organizations recognize that what gets rewarded is what gets done. Incentive alignment strategies include: using benchmark data to set annual operating budgets; empowering cost center managers to negotiate trade-offs within benchmark budget targets; and tying budget performance into incentive compensation.
Administrative cost pressure is a reality for all payers. These five best practice approaches can provide the foundation on which organizations can more effectively manage their administrative performance and achieve long-term goals for organizational success.
To learn more about administrative cost and MLR, click the following link to access our most recent webinar: Managing the Other Side of the Medical Loss Ratio.
The Patient-Centered Medical Home (PCMH) model emerged in response to the recognized need for more systematic and evidence-based approaches to ensuring access to, and coordinating care for, an entire population. Within that framework an additional focus has developed that recognizes the specific needs of members with chronic and complex conditions. In order to respond to that population, additional models have emerged that focus on proactive identification of chronically ill members using claims-based predictive model scoring as well as clinically based referrals and information. These models are often referred to as Ambulatory Intensive Care Unit (AICU) or Intensive Outpatient Care Program (IOCP).
The AICU and IOCP models, now in pilot testing and operation in a number of sites and with different Milliman clients, have the following core design characteristics:
- Payer and medical group partnerships that emphasize clinician leadership and active involvement in design and implementation;
- Program designs that promote patient-centered process changes and result in enhanced roles for nursing and clinic administrative staff;
- Proactive identification of members with chronic or complex illnesses using claims-based risk scoring tools such as the Milliman Advanced Risk Adjuster (MARA) coupled with clinic and clinician-based identification of candidates for the program;
- Clinic-based outreach to members to encourage active engagement;
- Personalized care plans that address the specific needs and priorities of the members; and
- Recognition of psychosocial barriers to adequate medical care resulting in innovative approaches to removing these barriers.
There are a number of good articles written about these initiatives including “Enhancing Quality of Primary Care Using an Ambulatory ICU to Achieve a Patient-Centered Medical Home” Lewis, Hoyt and Kakoza, Journal of Primary Care & Community Health May 27, 2011 and “American Medical Home Runs” Arnold Milstein and Elizabeth Gilbertson, Health Affairs, 28, no. 5 (2009) pages 1317-1326”. These and other articles provide good information on the results of these programs.
Most of these models involve hiring clinic-based case managers or imbedding health plan case managers in clinical sites. Working with the clinical teams and health plans involved in these projects has provided a number of insights into using risk-scoring information in supporting the needs of case managers in these new models.
These observations include the following.
1. Risk scoring data can be interesting but overwhelming to clinical team members - it is best to design reports and portals with early input from the people who will ultimately use it. As fascinating as claims data may be to some of us, these clinicians have an important perspective on what they are likely to use when and how.
2. Clinical user training can help them interpret and use risk scoring data more effectively–focus on the basics and priorities (eg. the difference between prospective and concurrent scores) and include case studies that lead participants through practice sessions and use of the information.
3. Clinical insight and patient knowledge is a valuable addition to the use of data – encourage care managers to use risk scoring data to find potential candidates and as a tool for adding a broader clinical, utilization and cost perspective to the information in a medical record. Clinicians and clinical teams can then provide a unique and highly useful perspective on what the information may indicate about a population and what can be done about it.
4. Managers can use summarized risk scoring data analysis to focus on a reasonable number of cases for review – risk scoring analysis can end up identifying large numbers of candidates.
Additional analysis of risk scores can be used to identify a more manageable list of candidates for care management. In the example below a client distribution analysis showed the number of members by risk score level both concurrently and prospectively for members with a score greater than one. This analysis helped identify patterns and clusters of members by risk score range. Further analysis isolated a smaller population (approximately 7% of the total) with concurrent and prospective scores that remained level or increased. This allowed the management team to focus on new candidates more quickly.
Client Example - Using Population Risk Score Distributions to Identify Care Management Candidates
The AICU and IOCP models, viewed as specialized programs designed to support the needs of chronically ill members, may in fact be good blueprints for all care management programs. The concept of integrating payer and provider data and expertise more directly can be rewarding for care management teams and beneficial to members.Claims analysis and risk scoring tools can be key tools in identifying opportunities and targeting interventions in these models.
Up until now, the integration of claims data and EHR data has been an often sought, but seldom realized goal in the healthcare analytics world. It’s a business intelligence idea with so much potential benefit, and it seems like it just wouldn’t be that hard to do. In reality, unifying these two types of data is enormously complex. To date, there have been too many barriers to sourcing this data, and a clear vision of the true ROI has not really emerged. So why are we poised for a breakthrough now?
The Convergence of Need and Availability
For the first time there are entities that have both access to the data and the interest to leverage it. Accountable Care Organizations (ACOs) need better support in understanding and managing the risk of their populations. Provider-Owned Health Plans (POHPs) must improve the efficiency of medical management in order to survive in the new healthcare market. With the Accountable Care Act and meaningful use, the penetration of EHR systems is growing, creating electronic data that is accessible and increasingly coded rather than free-form.
Rather than admit that integration of this data is a Mount Everest-level task beyond its worth, ACOs and POHPs are now examining the data and data sources to locate the first foot-hold. They’re finding strengths, weakness and characteristics in the two data types that are complementary. Indeed, the combination of claims data and EHR data will permit advanced analytics at the individual patient and patient population levels that could drive significant improvements in health outcomes and financial performance.
How We Can Get There
Let’s take an “If You Build It, They Will Come” approach, in two phases. Phase 1 starts with a focus on a single EHR system, perhaps one that is chosen as a pilot. The data is loaded from the EHR system and unified with the claims data in a data warehouse. It is transformed, meaning the coding is synchronized so distinct data, duplication and gaps are identified. Then it is presented in a unified, logical way using dashboards, member profiles and other analytic tools. At this stage, the two data sets are not fully integrated, but value can still be derived from the aggregation and normalization work that have taken place.
Phase 2 involves building the data set, then developing/applying advanced analytics such as clinical opportunity identification, development of hybrid risk scores and stratification of better episodes of care. The business intelligence that results could be used in many ways, such as:
- to provide greater levels of disease monitoring to ACOs and POHPs, leading to more informed decision-making at the enterprise level
- to support clinicians and patients in more informed decision-making during visits, and limiting unnecessary tests or treatments
- to guide medical research on disease prevention and treatment
Yes, the integration of claims data and EHR data will be messy at first. However, the potential gains for all parties to the healthcare system in our country are enormous. It’s time we pursue them.
As discussed in our previous blogs, ACOs are evolving in their ability to measure the effectiveness of their efforts. Today’s blog continues our focus on ACO analytics and reviews some insights from Modern Healthcare’s 3rd Annual ACO Survey and MedInsight analytic support of ACO effectiveness in care coordination.
Discussing the results of Modern Healthcare’s 3rd Annual ACO Survey, the article titled Still seeking best practices: Annual ACO survey shows care coordination remains a work in progress for many providers, highlights both the progress and challenges facing ACOs at all levels of the ACO payment model, from providers to patients. The survey captured responses from over 35 ACOs across a variety of organizations including ACOs with and without hospitals, ranging in size from those that manage care for 553,000 covered lives to those with 5,600 covered lives. As noted in the article, as accountable care continues to expand and evolve, organizations seeking to embrace the payment model are experimenting with how best to measure the impact of critical strategies to improve patient care, including better care coordination.
To gain a better understanding of the care coordination measurements currently in place, Modern Healthcare added two new care coordination questions to this survey for the first time this year. They were 1) is care coordination measured? and 2) if so, name the ACO's top five measures. The responses varied but it is clear that effective care coordination measures are still evolving.
While there are limited consistent measures across the spectrum of care coordination, there was consistent feedback from the respondents about what is currently used. Measures mentioned in the article and available in Milliman tools include:
- Readmission Rates
- High cost members
- ED frequent visits
- Medication management
- Where patients received care, such as the emergency room, including the level of care
- Health promotion and education to monitor care coordination
- In-network coordination of care and
- Patient engagement in care management
The patient engagement capabilities are available through the use of Milliman’s Enrollment Assessment and Survey tool. This is a web based tool that includes three online survey tools that measure the level of engagement a patient has in their own care, as well as the ability to track individuals and the related outreach made to members on a periodic basis, including action plans.
Below is an example of population level reporting for In Network vs Out of Network Emergency Department (ED) visits. MedInsight uses the claim detail and the Milliman Heath Cost Guidelines analytic engine to efficiently bucket and identify the ED visits. Using this same engine and the MedInsight Analytic Cube, Figure 1 delivers a report that provides a wealth of information about ED Visits for a given population. In this example, the utilization of ED visits are split between in and out of network visits, including the distribution of visits across each day of the week. Additionally, MedInsight provides additional metrics that provide further insight into the relationship between ED visits and inpatient admissions. Subsequent analysis may include:
- The review of admit hour to evaluate whether these visits were after hours or during normal office hours, highlighting education opportunities and access issues.
- Facility location of these ER visits
- Evaluate the distribution of visits across the PCPs as well as members, to identify frequent visitors to the ED or specific patterns
- Compare ER utilization to benchmarks – the % of admits from ED visits nationally is approximately 13%, according to CDC Fast Stats. A more specific benchmark, that is integrated into MedInsight reporting and analytics, are the Milliman Benchmarks, as shown in figure 2.
Figure 1 – ED Cases = In and out of ED, ED Visits Admitted = ED visit turned into IP Admit, ED Patient Visits= Total of ED Cases and ED Visits Admitted
Figure 2 – Compares the ED Visits to 50% Degree of Health Care Management and Well Managed degree of health care management (Degree of Health Care Management is a measure of the level of management processes in place, e.g. case management, UM, Disease Management)
While real time data is important for immediate care transition management, retrospective measures are also critical to evaluating success. MedInsight’s strength is to efficiently report cost and utilization detail at a population level as well as the ability to report on detailed member data. This provides access to several levels of analysis to more effectively manage and measure care coordination trends and analytics.
According to the 2008 U.S. Preventive Services Task Force (USPSTF) Screening for Colorectal Cancer Recommendation Statement, Colorectal is the second leading cause of cancer death in the United States and appropriate screening could save thousands of lives a year.
The USPSTF recommends colorectal screening for everyone between the age of 50 and 75 years of age. There as several screening tests currently available and modeling conducted by the USPSTF suggested that any of 3 screening programs would be “equally effective in life-years gained, assuming 100% adherence to the same regimen for that period.”
- Annual high-sensitivity fecal occult blood testing
- Sigmoidoscopy every 5 years combined with high-sensitivity fecal blood testing every 3 years
- Screening colonoscopy at intervals of 10 years.
Although other screening programs are less expensive and less invasive, and effectiveness is dependent upon the experience and expertise of those performing the procedure, well performed colonoscopies were assessed to have higher sensitivity and specificity for detecting colon cancer. This finding, along with Medicare and the Affordable Care Act mandating no cost sharing for colonoscopies, and increased public awareness has greatly increased the number of colonoscopies performed each year.
As the number of colonoscopies performed has increased, so has the variance in total cost for the procedure. Allowed charges can vary by thousands of dollars depending on the provider, place of service and other variables. Monitoring utilization and evaluating the charges for these procedures has become increasingly important for health plans striving to improve health while managing costs.
The 2013 version of the Milliman HCG (Health Cost Guidelines) grouper includes separate detail lines to track utilization and cost of facility, as well as professional costs associated with preventive colonoscopy.
Using illustrative data from 3 health plans, allowable charges and utilization counts for facility and professional services associated with preventive colonoscopy are shown in Figure 1 below. These data include claims for patients between the ages of 50 and 75. HCG 051b represents outpatient facility services and HCG P40b represents professional services for preventive colonoscopy. Note there could be related services submitted on separate claims that are not captured in these totals.
Utilization units are counted separately for the facility and professional services. The utilization count associated with the professional services represents the total number of preventive colonoscopies because some procedures will be provided in an office setting and not have a separate facility record.
To compare utilization rates and average allowable charges for preventive colonoscopies across the 3 plans, sum the allowable charges for both HCG detail lines but use only the professional unit counts to avoid double counting of procedures as shown in figure 2 below.
This simple analysis compares the cost and utilization of preventive colonoscopies across 3 plans, but additional analyses comparing costs across places of service (office, ambulatory surgery center and outpatient hospital) provide further insights into cost drivers associated with these procedures.
In previous blog posts we have discussed the analytics needs for population health management. One significant population that has been historically underserved by analytics is the pediatrics population. Recently, there has been renewed interest in managing the care and cost of this population. Some of the interest has been generated by CMMI grants in managing pediatric care and the rapid expansion of children in the Children Health Insurance program (CHIP) and Medicaid expansion.
Defining the Pediatric Population
This is a relatively simple population to define, typically using birth date to define the population. Many define the population from birth to 18 years of age. Often the population is further sub-divided into infants (0 – 1 ages), children (2-12 years old) and adolescents (13-18 years old). Another typical method of sub-dividing the population is by payer and program, with Medicaid programs such as CHIP and TANF/AFDC and commercial payers. Other important ways of segmenting this population are by socio-economic status, geography and by disease state. Milliman has developed a tool that hierarchically assigns patients to a single pre-dominant disease (chronic condition hierarchical groups) and is currently developing a pediatric version of this analytic.
Pediatric Specific Analytics
Effective population health management requires a balanced set of metrics to profile and target improvements for the population. It requires healthcare quality, care delivery efficiency, patient safety, cost, and utilization metrics. The ability to benchmark the specific pediatrics population is also critical to be able to target, plan and measure the impact of interventions. Below is a sample of the metrics that should be available to manage pediatrics populations.
These metrics are examples of the tools that need to be available within analytic systems to be able to effectively manage pediatrics populations. It will become increasingly important to have specific measures that are relevant to the unique characteristics of populations that are being managed.