In each edition, Manatt will feature a “deep dive” topic that shares new cross-cutting benchmarking program developments as states seek to evolve and advance their cost growth benchmarking programs to meet new regulatory and landscape needs. In this issue, Manatt examines opportunities for states to leverage All Payer Claims Databases (APCDs) and other key data assets to supplement state benchmarking programs.
Leveraging APCDs and Other Data Assets
The takeaway. State benchmarking programs may leverage other data resources - including All Payer Claims Databases (APCDs), private claims databases, and federal and state survey data - to provide policy-makers, regulators, consumer advocates, and researchers with important context for findings (e.g., who bears burden of cost growth) and allow results to be as actionable as possible (e.g., specific providers or drugs contributing to cost growth).
What it is. State cost growth benchmarking programs are data-driven, transparency-focused cost-containment initiatives that measure resident health care spending growth in relation to established targets; payers and providers that exceed targets may be subject to public inquiry or penalty. States collect benchmarking data directly from public and private payers operating in their states, monitoring health care spending across all lines of business. Payers may be asked to segment spending data by service category, key populations or product types, attribute spending to providers who may influence patient service utilization, or supplement “core” reporting with contextual information such as premium cost growth, Alternative Payment Methodology (APM) adoption rates, and member cost-sharing growth to help states better understand cost drives across payers and populations. Payer submissions, typically delivered in a set of summative tables with aggregate data (i.e., not person-level information), are sourced from a combination of their administrative (claims/encounter) data and financial data (non-claims-based payments), to present a complete, timely, and verifiable accounting of health care spend. Payers may be required to have an accountable person at their organization (e.g., CEO, chief actuary) certify that the data presented is valid to the best of their knowledge.
States may also use All Payer Claims Databases (APCD) to better understand health care market cost trends. APCDs are large-scale databases that collect health care claims and encounter data from public and private payers across most lines-of-business, with the notable exceptions of the private self-insured (unless voluntarily reported) and Medicare fee-for-service (unless manually integrated by the state from CMS files).1 Claims/encounter data can be a rich source of information, including person-level detail on patient diagnosis, the service delivered, the provider delivering the service, and the amount paid for delivery, by both the payer and patient. APCDs also collect other administrative information from payers to supplement and contextualize claims data, including enrollee demographic characteristics (e.g., age, zip code), and characteristics of enrollees’ coverage types and details (e.g., network characteristics, plan premiums). APCDs can provide health services researcher with large sample sizes, person-, provider-, and service-level detail, and the ability to following patient populations/panels over time (i.e., longitudinal information), making them valuable and powerful – if at times unwieldy – data assets.2
Unfortunately, APCD data cannot replace payer benchmarking data reporting, a common myth in health data circles, for reasons including that APCD data:
- Does not include non-claims payment information, which is of increasing importance as more payers and providers are paid under APMs;
- Does not include the vast majority of ERISA-preempted self-insured claims data (self-insured lives typically comprise over 60% of the employer-sponsored insurance market);
- Is not as timely as benchmarking data files, with calendar year benchmarking data received as soon as five months after year-end with payer-provided incurred but not reported (IBNR) estimates; and
- Does not have payer verification of results – or the methods used to derive them.
However, benchmarking and APCD data analyses can be paired to great effect: with benchmarking data uniquely capable of identifying cross-market concerns, while APCD data can be used to add context and detail to findings, making them that much more actionable for policy-makers, regulators, advocates, and researchers.
Figure 1. Benchmarking Data vs. APCD Data
What it means. While benchmarking data can provide important insights into aggregate, year-over-year cost growth trends by state, payer, provider (often), service category, and population group, states can derive additional insights by pairing benchmarking data and findings with analyses of other data assets, such as APCDs.6 Six of the eight states that have a benchmarking program, or have one actively in development, also have an APCD - including Washington, Oregon, Massachusetts, Delaware, Connecticut and Rhode Island – though analytic coordination across the two data assets varies considerably.7
Figure 2. Current State of APCD Implementation as of February 2022
Massachusetts has used its APCD, as well as other data assets it stewards, such as the Massachusetts Health Insurance Survey (MHIS), to support the state’s broader cost trends evaluations and hearings. For example, in 2021, the Massachusetts Health Policy Commission (HPC) used state benchmarking data to identify hospital outpatient spending as the fastest-growing service category for the commercial market, followed by spending on physician services and other professional services (Figure 3).8, 9
The HPC then leveraged other available data to better understand the equity implications of cost growth: pairing service category trends with information on who bears the costs.
Using commercial claims data from the Massachusetts APCD, the HPC found that individuals in the highest-income quintile had a higher proportion of their overall health care spending concentrated in hospital outpatient and professional services, whereas individuals in the lowest-income quintile had a higher proportion of their spending concentrated on prescription drugs, inpatient services, and emergency department (ED) services (Figure 4). These findings – made possibly only by linking benchmarking data with other available data resources - highlighted an important disparity in individuals’ access and utilization of health care services by income for policymakers and regulators consideration.
Figure 3. Percentage annual growth in spending per commercial enrollee, 2016-2019.10
Figure 4. Percent of health care spending by category and income for commercially insured adults by lowest- and highest-income quintile, 2018. 11
State APCD Use Cases
APCDs have been used to support numerous use cases across the 18 states that presently host them. They may be used by policy-makers, regulators, consumer advocates, researchers, and other stakeholders to: better understand health care spending and utilization by payer, provider, and population group; answer specific policy and research questions (e.g., potential impact of Medicaid expansion); support public health monitoring, population health assessments, and cross-payer quality measurement initiatives; increase health system and cost transparency; and guide purchasers in decision-making.12
Massachusetts’ HPC uses the state’s APCD to expand upon benchmarking-related topics and issues, including to:13
- Analyze out-of-pocket costs for commercial insured populations,14 including copayments, co-insurance, and deductibles for both medical and prescription spending among the commercially insured using APCD data. The HPC found that from 2015-2017, average annual OOP spending for the commercially insured grew about 20%, from $601 to $721. Within that average annual OOP spend, APCD data provided additional clarity on who faced the cost of increasingly high OOP spending by examining the distribution of the annual OOP spend: while half of all members spent $345 or less OOP annually, individuals at or above the 90th percentile of OOP spending in all three years spent nearly ten times more, or nearly $3,499 on average, in comparison.
- Analyzing telehealth visits among commercially insured residents in 2015-2017,15 which examined telehealth visits using commercial claims data from the Massachusetts APCD. This analysis found that the rate of telehealth utilization among commercially-insured patients in Massachusetts almost doubled between 2015 and 2017, from 2.0 visits per 1,000 members in 2015 to 4.0 visits per 1,000 members in 2017 – even prior to the pandemic - a finding that mirrors trends observed in a similar national commercially-insured population. The investigation only added context to future benchmarking discussions around patient access, service utilization, and cost.
While having an APCD allows for more robust analyses of heath care trends, states may also use private claims data assets – such as data from the Health Care Cost Institute (HCCI) or FAIR Health - to better understand their health care spending trends.
State-level survey data has been widely used by states to understand health care cost growth and consumer affordability, including Massachusetts, Oregon, and Connecticut. For example, Connecticut paired American Community Survey (ACS) data with APCD data to create the Connecticut Healthcare Affordability Index (CHAI), a tool for policymakers and consumers to better understand the growing burden of rising health care costs for Connecticut families. The CHAI used data from the Connecticut APCD to calculate out-of-pocket costs for families with employer-sponsored and individual marketplace insurance by town, county, age group, gender, and health risk score.16
States may also use the Medical Expenditure Panel Survey (MEPS) Insurance Component (IC), which fields questionnaires to private and public sector employers to collect data on the number and types of private health insurance plans offered, benefits associated with these plans, annual premiums, annual contributions by employers and employees, eligibility requirements, and employer characteristics.17 For many states that do not have health insurance premium reporting as part of their benchmarking data collection process, the MEPS-IC provides similar premium cost growth data – though with a lag of at least two years – that provides insight into how employees and employers are directly confronting cost growth in local markets.
What happens next. As benchmarking programs continue to proliferate and mature, additional opportunities will emerge to examine how states are pairing their data with other assets to reinforce findings and other, novel use cases.
1 APCD Council, FAQs. Available here: https://www.apcdcouncil.org/frequently-asked-questions
2 “Overview of All-Payer Claims Databases,” Agency for Healthcare Research and Quality. Available here: https://www.ahrq.gov/data/apcd/index.html
3 “Annual Report on the Performance of the Massachusetts Health Care System (October 2019),” Massachusetts Center for Health Information and Analysis (CHIA). October 2019. Available here: https://www.chiamass.gov/assets/2019-annual-report/2019-Annual-Report.zip
4 Interactive Dashboard, Top 20 Drugs in 2015: Total Expenditures. Commercial Prescription Drug Use & Spending, 2015-2017, Massachusetts Center for Health Information and Analysis (CHIA). February 2020. Available here: https://www.chiamass.gov/prescription-drugs/#prescription-dashboard
5 CHIA data set used does not reflect the impact of prescription drug rebates. Also does not include spending for drugs or administration of drugs covered under a medical benefit.
6 Other informative data assets that states have used include hospital discharge data, payer expenditure reports, provider financial reports, and surveys of employers and households.
7 Nevada is planning for APCD development, pending the release of federal funds to support establishment.
8 “2021 Annual Health Care Cost Trends Report,” Massachusetts Health Policy Commission (HPC). September 2021. Available here: https://www.mass.gov/doc/2021-health-care-cost-trends-report/download
9 Other informative data assets that states have used include hospital discharge data, payer expenditure reports, provider financial reports, and surveys of employers and households.
10 Data source: Payer reported TME data to CHIA and other public sources; HPC analysis of data from Center
for Health Information and Analysis Annual Report, March 2021.
11 Data source: HPC analysis of Massachusetts APCD, 2018.
12 D. McCarthy, “STATE ALL-PAYER CLAIMS DATABASES: Tools for Improving Health Care Value. Part 2: The Uses and Benefits of State APCDs,” The Commonwealth Fund. December 2020. Available here: https://www.commonwealthfund.org/sites/default/files/2020-12/McCarthy_State_APCDs_Part2_v2.pdf
13 Health Policy Commission (HPC) DataPoints Series. Available here: https://www.mass.gov/service-details/health-policy-commission-hpc-datapoints-series
14 “DataPoints Issue 19: Persistently High Out-of-Pocket Costs Make Health Care Increasingly Unaffordable and Perpetuate Inequalities in Massachusetts,” Massachusetts Health Policy Commission (HPC). Jan. 13, 2021. Available here: https://www.mass.gov/info-details/hpc-datapoints-issue-19-persistently-high-out-of-pocket-costs-make-health-care.
15 “DataPoints Issue 16: The Doctor Will (Virtually) See You Now,” Massachusetts Health Policy Commission. March 12, 2020. Available here: https://www.mass.gov/info-details/hpc-datapoints-issue-16-the-doctor-will-virtually-see-you-now
16 L. Manzer and D. M. Pearce, “Connecticut Healthcare Affordability Index,” prepared for the Connecticut Office of Health Strategy and Connecticut Office of the State Comptroller. December 2020. Available here: https://portal.ct.gov/-/media/OHS/CT-Healthcare-Affordability-Index/CHAI/CT-Healthcare-Affordability-Index.pdf
17 The Medical Expenditure Panel Survey, Insurance/Employer Component. Agency for Healthcare Research and Quality. Accessed December 22, 2021. Available here: https://meps.ahrq.gov/survey_comp/Insurance.jsp
