Arete Healthcare LLC

Independent Practices | Shared Infrastructure | Clinician-Led Governance

Key Findings

In Pictures

All figures appear here, on four pages: what changed (Figure 1), annual trends (Figure 2), chiropractic code mix (Figure 3), and physical therapy code mix (Figure 4).

Figure 1. Chiropractic and physical therapy in Original Medicare: what changed, 2013–2024. (1a) Percent change, 2024 compared with 2013, across seventeen national measures for chiropractic (dark blue) and physical therapy (light blue). Chiropractic declined or stayed nearly flat on providers (−13.1%), patients (−20.5%), services (−18.6%), share of beneficiaries receiving care (−11.9%), and allowed spending per eligible beneficiary per year (PMPY, +2.0%), while physical therapy roughly doubled on most of them (providers +98.5%, services +108.8%, access +112.9%, allowed PMPY +129.6%). (1b) Average allowed amount per unit, 2024 compared with 2017 (the first year of the current PT evaluation codes), in actual dollars (dark blue) and adjusted for inflation (light blue; CPI-U, 2024 dollars; prices rose 28.0% over the period). In actual dollars, allowed per unit fell slightly for each chiropractic manipulation code (98940 −0.7%, 98941 −0.3%, 98942 −0.8%) and fell for PT treatment codes 97110 (−15.3%), 97140 (−11.8%), and 97112 (−10.7%); 97530 was unchanged (−0.1%), and PT evaluation codes 97161–97164 rose 21.8% to 26.3%. On an inflation-adjusted basis, both professions lost material purchasing power from 2017 to 2024: about 22% for each manipulation code (−22.1% to −22.5%), 30% to 34% for 97110, 97140, and 97112, 21.9% for 97530, and 1.4% to 4.8% even for the evaluation codes. No chiropractic evaluation code appears because Medicare does not pay chiropractors for evaluations (Section 2.2). PerPt = per patient; PerSvc = per service unit. Charged amounts are submitted charges, not payments. Patient-based measures count a beneficiary once per provider (Section 2.4). Panel 1a is in nominal dollars. Source: Arete analysis of CMS Medicare Physician & Other Practitioners and Medicare Monthly Enrollment public-use data; BLS CPI-U (Section 2.7). Visualization was human generated.

Figure 2. Annual trends, 2013–2024: the divergence is about how many beneficiaries receive care, not how much care each receives. (2a) Share of eligible Original Medicare beneficiaries receiving care: chiropractic drifted down from 6.07% to 5.35%, while physical therapy more than doubled, from 7.81% to 16.64%. Both fell in 2020 (icon); PT passed its 2019 level (12.37%) by 2021, while chiropractic (5.66% in 2019) had not returned to it by 2024. (2b) Services per patient changed little: chiropractic 8.97 to 9.19 (+2.4%), PT 27.89 to 30.33 (+8.7%). Services are billing units, not visits; timed PT codes often produce several 15-minute units per visit. (2c) Allowed dollars per patient rose faster for chiropractic ($289.42 to $335.05, +15.8%) than for PT ($737.50 to $795.26, +7.8%), though PT's level stayed more than twice as high. (2d) Allowed dollars per eligible beneficiary per year were nearly flat for chiropractic ($17.58 to $17.93, +2.0%) and more than doubled for PT ($57.63 to $132.33, +129.6%), driven mainly by the access gains in 2a. Patient-based measures (2a–2c) count a beneficiary once per provider, so levels are somewhat overstated, more so for PT (Section 2.4); 2d uses enrollment and needs no patient count. Nominal dollars. Source: Arete analysis of CMS public-use data. Visualization was human generated.

Figure 3. Chiropractic: three covered codes, a shift toward 98941, and flat payment per unit. (3a) Share of chiropractic manipulation services by code. These three codes are the only chiropractic services Medicare covers; no evaluation is payable to a chiropractor, even though CMS requires the chiropractor's record to document a history, examination, and treatment plan (Section 2.2). CPT 98941 (three to four spinal regions) rose from 62.1% to 74.5% of services, while 98940 (one to two regions) fell from 31.0% to 20.2% and accounted for 78.4% of the net decline in manipulation services; 98942 (five regions) fell from 6.9% to 5.3%. (3b) Average allowed amount per unit: 98940 $25.15 to $26.75, 98941 $34.45 to $38.12, 98942 $44.38 to $49.84. Most of the increase occurred between 2013 and 2014; amounts peaked in 2020 and declined each year through 2024. Since 2017 each code's allowed amount per unit fell slightly, yet chiropractic's average per service rose 3.2% because billing shifted toward the higher-paying 98941 (Section 7). Nominal dollars. Source: Arete analysis of CMS public-use data. Visualization was human generated.

Figure 4. Physical therapy: a wide code set and a shift toward higher-paying codes. (4a) Share of PT services by billing code. Therapeutic exercise (97110, green) fell from about 46% to 33% of units and manual therapy (97140, blue) from about 21% to 15%, while therapeutic activities (97530) and neuromuscular re-education (97112) (purples) rose from a combined 14.5% to 40.0% and accounted for 63.3% of net PT growth. Evaluation codes (red, bottom) are about 2%–3% of units. (4b) Average allowed amount per unit. Evaluation codes (red) are the highest-paid: 97001 averaged about $74–75 through 2016, and its 2017 replacements 97161–97163 rose from about $81 to about $99 by 2024. From 2013 to 2024, allowed per unit fell for 97110 ($27.06 to $22.77), 97140 ($24.01 to $20.59), and 97112 ($30.33 to $27.09), while 97530 ended near its 2013 level ($31.70). Because billing moved toward the higher-paying codes, PT's average allowed per unit fell only 0.8%; at the 2013 code mix it would have fallen 13.3% (Section 7). The 2016–2017 gap reflects the evaluation code replacement. Both charts encode more than six series; color groups related code families and gray de-emphasizes low-volume codes (Style Guide §3.6, §13.3c). Nominal dollars. Source: Arete analysis of CMS public-use data. Visualization was human generated.

Arete's Take

At a Glance

Table 1. Chiropractic (DC) and physical therapy (PT) in Original Medicare, 2013 → 2024

Source: Arete analysis of CMS public-use enrollment and provider-service data. Dollar amounts are nominal. *Rows marked with an asterisk use CMS's beneficiary-count field summed at the provider level (Section 2.4). That count includes a beneficiary once per provider, so a beneficiary who saw two physical therapists in a year is counted twice; these levels are therefore somewhat overstated, more so for PT, whose provider base is larger (Section 2.6).

Abstract

This report describes national Original Medicare (fee-for-service) use of, and spending on, services billed by chiropractors (DCs) and outpatient physical therapists (PTs) from 2013 through 2024. It uses CMS public-use files that report, for each provider and billing code, the number of services, beneficiaries, and dollars, together with CMS's annual Original Medicare enrollment.

Over the period, Original Medicare enrollment fell 9.8%. Chiropractic billing providers fell 13.1%, services fell 18.6%, and allowed spending fell 8.0%. Physical therapy billing providers rose 98.5%, services rose 108.8%, and allowed spending rose 107.0%. The share of beneficiaries receiving chiropractic care fell from 6.07% to 5.35%, while the share receiving PT rose from 7.81% to 16.64%. Allowed spending per eligible beneficiary per year was nearly flat for chiropractic (+2.0%) and more than doubled for PT (+129.6%).

The divergence was not driven by payment rates. The blended allowed amount per service changed less than 1% for either profession from 2014 to 2024. Underneath that average, the two professions bill very different sets of services. Medicare covers only spinal manipulation (CPT 98940–98942) when furnished by a chiropractor and does not pay a chiropractor for an evaluation; PT billing in this data spans 30 itemized codes, including evaluations. Since 2017, allowed amounts per unit fell slightly for each manipulation code and fell 10.7% to 15.3% for three major PT treatment codes, while PT evaluation codes rose 21.8% to 26.3%. Over the same period consumer prices rose 28.0%, so on an inflation-adjusted basis the manipulation codes lost about 22% of their purchasing power, the major PT treatment codes 22% to 34%, and even the PT evaluation codes 1% to 5%. Both professions' billing shifted toward higher-paying codes: chiropractic toward 98941, and PT toward 97530 and 97112. Holding the 2013 code mix fixed, PT's average allowed per unit would have fallen 13.3% rather than the 0.8% observed.

These are descriptive national trends from aggregated public data. They do not establish why the trends occurred, compare clinical outcomes, or show that one profession's services substituted for the other's.

Methods

2.1 Study Design and Scope

This is a national, longitudinal, descriptive analysis of Original Medicare (fee-for-service, FFS) activity from 2013 through 2024. It compares services billed by chiropractors with services billed by physical therapists in independent outpatient settings, using CMS public-use provider-service files. Both are paid under Medicare Part B, the part of Original Medicare that covers outpatient and professional services. Beneficiaries enrolled in Medicare Advantage (Part C) are excluded; the Original Medicare population is the denominator for every access-rate and per-beneficiary measure.

The unit of observation in the source data is a provider-by-billing-code combination, not a patient or an episode of care. The analysis describes changes in provider participation, service volume, code mix, and allowed amounts. It does not evaluate clinical effectiveness, episode cost, diagnoses, outcomes, referral patterns, or substitution between professions.

The two-stage AI-assisted drafting and verification process, and the reconciliation record, are documented in the internal AI Assistance and Reproducibility Record maintained under the Arete Research Library Governance Framework.

2.2 Why Chiropractic and Physical Therapy Bill Different Numbers of Codes

The two Medicare benefits are designed differently, which is why this analysis finds three chiropractic codes but dozens of PT codes. The difference reflects benefit design, not what chiropractors and physical therapists are licensed to do.

Under the Medicare Benefit Policy Manual (Pub. 100-02), Chapter 15, Section 240, Medicare's chiropractic benefit is limited to manual manipulation of the spine to correct a subluxation, billed as CPT 98940 (one to two spinal regions), 98941 (three to four regions), or 98942 (five regions), and only when care is active or corrective rather than maintenance. Other services a chiropractor furnishes or orders — including evaluation and management visits, X-rays, extraspinal manipulation (CPT 98943), and therapeutic exercise or other therapies — are not covered when furnished by a chiropractor. At the same time, CMS requires the chiropractor's record for an initial visit to document a history, an examination, a subluxation diagnosis, and a treatment plan to support manipulation. Medicare therefore requires the content of an evaluation in the chiropractic record but does not pay the chiropractor for it.

Medicare's outpatient physical therapy benefit covers medically necessary therapy services furnished under a plan of care, without limiting coverage to a named set of procedures. PT billing therefore spans evaluations, re-evaluations, active therapies, manual therapy, and modalities. Instead of a per-code limit, PT is subject to an annual per-beneficiary spending threshold above which therapists attest to medical necessity with the KX modifier; the Bipartisan Budget Act of 2018 (§50202) replaced the earlier hard annual therapy cap with this threshold, effective 2018. Effective January 1, 2017, CMS replaced PT evaluation codes 97001 and 97002 with three complexity-based evaluation codes (97161–97163) and a re-evaluation code (97164).

2.3 Data Sources

Table 2. Data sources

Data were accessed through Mimilabs Databricks, a third-party mirror of CMS public-use sources, not CMS's own systems. Source URLs appear in References; exact table vintages are retained in the internal reproducibility file (Governance Framework §9). The methods memorandum also listed the Physician/Supplier Procedure Summary (PSPS) as a source of national totals, but the supplied query takes all-provider totals from the provider-service file instead (internal reproducibility file); this report labels those totals accordingly (Section 3).

2.4 Provider Identification and Aggregation

• Chiropractic: rendering provider type “Chiropractic” (Medicare specialty code 35). Unique providers are distinct rendering NPIs billing at least one service in the year.

• Physical therapy: rendering provider types “Physical Therapist” and “Physical Therapist in Private Practice” (both are included; the methods memorandum maps them to specialty code 65). Unique providers are distinct rendering NPIs in the year.

• Dollars: submitted charges, allowed amounts, and Medicare payments are reported in the source as averages per service; each was multiplied by the number of services on its provider-code row and summed nationally.

• Services: the reported number of services (tot_srvcs), summed by specialty and year. For timed PT codes one visit often produces several 15-minute units.

• Patients: beneficiary counts (tot_benes) summed across providers from the provider-level file. A beneficiary is counted once per provider, not once overall: someone who saw two chiropractors in a year is counted twice. CMS public files carry no beneficiary identifier, so a fully unduplicated count is not possible here.

• Year: the performance year of each source file. Enrollment: the CMS-designated annual national record.

• Code-level detail: the code-level extract itemizes the three manipulation codes for chiropractic (more than 99.9% of chiropractic services in every year) and the 26 to 28 highest-volume codes per year for PT (99.3% to 99.9% of PT services). The few services not itemized are included in specialty totals but not in code-level tables.

2.5 Derived Measures

Table 3. Derived measures

*Uses the provider-level beneficiary count described in Section 2.4.

2.6 External Checks and Data Provenance

Enrollment. An earlier version of this report checked the enrollment figures directly against CMS's published Medicare Monthly Enrollment annual national record and found agreement to within about 100 beneficiaries per year out of roughly 34–39 million. That check was not repeated for this version. KFF reports a smaller “traditional Medicare” count because it is limited to beneficiaries with both Part A and Part B, for the purpose of measuring Medicare Advantage penetration; the same enrollment denominator is used for both professions in every year, so the definitional difference does not affect any DC-versus-PT comparison.

Chiropractic access benchmark. Whedon, Song, and Davis (2013), using Medicare claims, estimated that 6.9% of Medicare Part B beneficiaries used chiropractic services in 2008. This report's 2013 figure, 6.07%, is directionally consistent with that estimate given the different years and methods.

Size of the cross-provider overcount. The overcount described in Section 2.4 cannot be measured with public files. A directional check is possible: PT patients per PT provider were roughly stable (75.5 in 2013 and 73.0 in 2024). If the number of physical therapists each patient saw had risen sharply, that ratio would be expected to move more than it did. This does not bound the overcount's size in any year, but it suggests the overcount did not grow much over time. The trends in this report are therefore more robust than the absolute levels of the access-rate and per-patient measures.

Figures. All ten charts were prepared by the Author in Microsoft Excel. Panels 1a and 2a–4b use the same source workbook used for every statistic in this report; panel 1b uses the Author's inflation workbook described in Section 2.7. Consistent with prior Original Research reports, all charts appear in In Pictures, arranged as four figure pages (Figures 1–4, panels a–d); the panels were placed side by side without altering chart content, and Results refers to them by panel. The values plotted in panels 1a and 1b were checked against independently recomputed values; panels 2a–4b were checked visually against the recomputed annual series (internal reproducibility record).

2.7 Inflation Adjustment

Dollar amounts in this report are nominal (not adjusted for inflation) unless labeled otherwise. For the 2017-to-2024 comparison of allowed amounts per unit (panel 1b, Section 7, and Key Findings), amounts were also converted to constant 2024 dollars using the Consumer Price Index for All Urban Consumers (CPI-U; U.S. city average, all items, annual average, 1982–84 = 100) published by the U.S. Bureau of Labor Statistics. Each year's allowed amount per unit was multiplied by the ratio of the 2024 annual CPI-U (313.689) to that year's annual CPI-U (245.120 for 2017), so prices rose 28.0% from 2017 to 2024. The inflation-adjusted percent change is the 2024 allowed amount per unit divided by the 2017 amount in 2024 dollars, minus one — equivalently, (1 + nominal change) ÷ 1.280 − 1. 2017 was chosen as the base year because it is the first year of the current PT evaluation codes, allowing like-for-like code comparisons.

CPI-U measures the general purchasing power of a dollar. It is not a measure of practice costs, clinician wages, or Medicare's own Medicare Economic Index, so the inflation-adjusted figures describe what an allowed payment would buy in general terms, not how practice margins changed. Allowed amounts per unit are observed averages across claims, not published fee schedule rates (Section 2.4).

Results

3. The Original Medicare Population Shrank

Original Medicare enrollment rose to a peak of 38.7 million in 2017, then fell to 33.9 million in 2024 — a 9.8% decline from 2013 (37.6 million). Over the same period, allowed spending for all physician and other-practitioner services in the provider-service file rose from $122.85 billion to $165.99 billion (+35.1%), or from $3,266 to $4,894 per Original Medicare beneficiary (+49.9%). Because the denominator shrank, raw counts understate per-beneficiary change, so this report presents both.

4. Access and Spending per Beneficiary Moved in Opposite Directions

The share of Original Medicare beneficiaries who received chiropractic care drifted down from 6.07% in 2013 to 5.35% in 2024 (−11.9%). The share who received outpatient PT more than doubled, from 7.81% to 16.64% (+112.9%) (Figure 2a). Both fell in 2020; PT exceeded its 2019 level (12.37%) by 2021, while chiropractic (5.66% in 2019) had not returned to it by 2024.

Allowed spending per eligible beneficiary per year followed access (Figure 2d). Chiropractic was nearly flat, from $17.58 to $17.93 (+2.0%); PT rose from $57.63 to $132.33 (+129.6%). Spending per eligible beneficiary equals access rate × services per patient × allowed amount per service. For chiropractic, a lower access rate (−11.9%) was offset by more services per patient (+2.4%) and a higher allowed amount per service (+13.0%). For PT, the access rate (+112.9%) accounts for most of the growth, with a smaller contribution from more services per patient (+8.7%) and none from the allowed amount per service (−0.8%).

5. Per-Patient Intensity Changed Modestly

Among beneficiaries who received care, services per patient rose slightly for both professions: chiropractic from 8.97 to 9.19 (+2.4%) and PT from 27.89 to 30.33 (+8.7%) (Figure 2b). The difference in level partly reflects how services are counted. A chiropractic manipulation is generally one unit per visit, while timed PT codes are billed in 15-minute units, so one PT visit often produces several units; the ratio of services is not a ratio of visits.

Allowed dollars per patient rose faster for chiropractic ($289.42 to $335.05, +15.8%) than for PT ($737.50 to $795.26, +7.8%), although PT's level remained more than twice as high (Figure 2c). Most of the chiropractic increase came from a higher allowed amount per service between 2013 and 2014 (Section 7). PT's per-patient amount peaked at $817.05 in 2021.

6. Provider Participation and Service Volume Diverged

Chiropractic billing providers declined from 36,187 to 31,446 (−13.1%); PT billing providers nearly doubled, from 38,942 to 77,296 (+98.5%). Chiropractic providers declined somewhat faster than enrollment, so density fell 3.6% (from 96 to 93 billing providers per 100,000 beneficiaries); PT density rose 120.1% (from 104 to 228).

Chiropractic services declined from 20.5 million to 16.7 million (−18.6%), or from 545 to 492 per 1,000 beneficiaries (−9.7%). PT services rose from 82.0 million to 171.1 million (+108.8%), or from 2,179 to 5,046 per 1,000 beneficiaries (+131.5%). Chiropractic allowed spending fell from $661.2 million to $608.2 million (−8.0%) and Medicare payments from $481.2 million to $430.3 million (−10.6%); PT allowed spending rose from $2.17 billion to $4.49 billion (+107.0%) and payments from $1.70 billion to $3.45 billion (+103.1%).

As a result, the relative scale of the two professions changed. In 2013, PT had 1.1 times as many billing providers as chiropractic, 4.0 times as many services, 3.3 times the allowed spending, and 3.5 times the Medicare payments; by 2024 those ratios were 2.5, 10.3, 7.4, and 8.0. Chiropractic's share of allowed spending for all physician and other-practitioner services fell from 0.54% to 0.37%; PT's rose from 1.76% to 2.70%. These shifts show a redistribution in relative scale; they do not show that PT services replaced chiropractic services.

7. Payment per Service Was Flat on Average, but Not Code by Code

The blended allowed amount per service — total allowed dollars divided by total services across all codes — rose 13.0% for chiropractic ($32.25 to $36.45) and fell 0.8% for PT ($26.44 to $26.22) from 2013 to 2024. Nearly all of the chiropractic increase occurred between 2013 and 2014 (+12.3%), when the allowed amount per unit rose for each manipulation code (Figure 3b); this analysis does not establish the reason. From 2014 to 2024 the blended amount changed only +0.6% for chiropractic and +0.5% for PT. These are nominal dollars; the inflation-adjusted comparison below shows how much purchasing power that flat payment represents.

Code-level detail shows what the blended average hides (Figure 1b). From 2017 to 2024, the allowed amount per unit fell for each chiropractic manipulation code (98940 −0.7%, 98941 −0.3%, 98942 −0.8%), yet chiropractic's blended amount rose 3.2% — entirely because billing shifted toward the higher-paying 98941 (Section 8). For PT, the allowed amount per unit fell for therapeutic exercise (97110, −15.3%), manual therapy (97140, −11.8%), and neuromuscular re-education (97112, −10.7%), held flat for therapeutic activities (97530, −0.1%), and rose 21.8% to 26.3% for the evaluation and re-evaluation codes; the blended PT amount fell 2.0%.

Adjusted for inflation, both professions lost material purchasing power per unit from 2017 to 2024 (Figure 1b). Consumer prices rose 28.0% over the period (CPI-U; Section 2.7). In 2024 dollars, the allowed amount per unit fell 22.4% for 98940 ($34.46 to $26.75), 22.1% for 98941 ($48.93 to $38.12), and 22.5% for 98942 ($64.32 to $49.84). For PT it fell 33.8% for 97110, 31.0% for 97140, 30.3% for 97112, and 21.9% for 97530; even the evaluation codes, which rose 21.8% to 26.3% in nominal dollars, fell 4.0% to 4.8% (97161–97163) and 1.4% (97164). On a blended basis, the inflation-adjusted allowed amount per service fell 19.4% for chiropractic and 23.4% for PT from 2017 to 2024.

A fixed-mix index separates price from mix. Holding each profession's 2013 code mix constant and pricing it at 2024 allowed amounts, chiropractic's average allowed per unit would have risen 9.8% (observed: +13.0%), and PT's would have fallen 13.3% (observed: −0.8%). In other words, shifts toward higher-paying codes offset most of the per-unit decline in PT's treatment codes, and added about three percentage points for chiropractic. The PT index uses the 23 codes billed in both 2013 and 2024 (97.0% of 2013 itemized PT units); it excludes the evaluation codes because those changed in 2017.

8. Chiropractic: Three Covered Codes and a Shift Toward 98941

Effectively all chiropractic services in Original Medicare were billed under the three covered manipulation codes — more than 99.9% in every year. The only other code itemized in the extract, extraspinal manipulation (98943), appeared in 2013 alone (201 services, average allowed $0.99), consistent with its non-coverage.

Within the three codes, billing shifted steadily toward 98941 (three to four spinal regions). Its share of manipulation services rose from 62.1% in 2013 to 74.5% in 2024, while 98940 (one to two regions) fell from 31.0% to 20.2% and 98942 (five regions) from 6.9% to 5.3% (Figure 3a). In volume, 98940 declined 47.0% and accounted for 78.4% of the net decline in manipulation services; 98941 declined only 2.4%. Providers billing 98940 fell from 18,917 to 11,901, while those billing 98941 declined only modestly, from 27,397 to 26,588. The data cannot say whether this reflects changes in patient presentation, documentation, coding, or payer review.

Table 4. Chiropractic manipulation codes, 2013 and 2024

Source: Arete analysis of CMS provider-service public-use data. CMT = chiropractic manipulative treatment. Units are reported services. Dollar amounts are nominal average allowed amounts per unit.

9. Physical Therapy: A Wide Code Set and a Shift Toward Higher-Paying Codes

PT billing spans a much wider set of services. The code-level extract itemizes 30 distinct codes across the period (26 to 28 in any year), covering evaluations and re-evaluations, active therapies, manual therapy, and modalities. Four treatment codes — therapeutic exercise (97110), therapeutic activities (97530), neuromuscular re-education (97112), and manual therapy (97140) — made up 88.1% of all PT services in 2024 and 94.8% of the net growth since 2013.

The mix shifted toward codes with higher allowed amounts per unit. Therapeutic activities (97530; $31.70 per unit in 2024) and neuromuscular re-education (97112; $27.09) grew from a combined 14.5% to 40.0% of PT services and accounted for 63.3% of net PT growth. Therapeutic exercise (97110; $22.77) remained the largest code, but its share fell from 45.6% to 33.0%; manual therapy (97140; $20.59) fell from 20.6% to 15.2%. Lower-paying modalities declined in volume: unattended electrical stimulation (G0283; $8.75) −15.4%, ultrasound (97035) −71.8%, and attended electrical stimulation (97032) −61.4% (Figure 4a). Providers billing 97530 rose from 14,384 to 60,432, and those billing 97112 from 18,368 to 61,867. The data cannot determine whether these shifts reflect clinical practice, documentation and coding, payment policy, case mix, or some combination.

Table 5. Selected high-volume physical therapy codes, 2013 and 2024

Source: Arete analysis of CMS provider-service public-use data. Share is of all PT services reported for the specialty. Timed PT codes may produce several units per visit. Dollar amounts are nominal average allowed amounts per unit and may reflect multiple-procedure payment reductions and other claim-level rules.

Evaluation codes are a small share of PT units — 3.0% in 2013 and 2.0% in 2024 — but carry the highest allowed amounts per unit (Figure 4b). Before 2017, the PT evaluation code (97001) averaged about $74–75. The complexity-based evaluation codes that replaced it (97161–97163) averaged $81.0 to $81.4 in 2017 and $99.1 to $99.5 in 2024, with a step increase of about 16% between 2020 and 2021; the re-evaluation code (97164) rose from $54.29 to $68.54. Because the codes changed in 2017, evaluation amounts before and after 2017 are not directly comparable.

10. The COVID-19 Shock Did Not Create the Trend

Services fell sharply for both professions in 2020: chiropractic by 17.5% and PT by 20.2% from 2019. The divergence was already under way before the pandemic, and the recovery widened it. By 2024, PT services were 28.9% above their 2019 level, while chiropractic services remained 13.5% below theirs.

Summary

From 2013 to 2024, physical therapy in Original Medicare roughly doubled on providers, patients, services, and spending, while chiropractic contracted. The divergence came from participation and volume, not from a widening gap in average payment per service. Underneath, the benefit designs differ: chiropractors can be paid for three manipulation codes and no evaluation, while physical therapists bill a broad set of services, including evaluations whose allowed amounts rose after 2017. Both professions' billing shifted toward higher-paying codes. These data do not establish why these changes occurred, compare outcomes, or show substitution between the professions.

Key Limitations

• Patient counts are not fully unduplicated. Beneficiary counts are summed across providers, so a beneficiary who saw more than one provider of the same specialty in a year is counted more than once. Access-rate and per-patient figures are therefore somewhat overstated, likely more so for PT. Trends are more reliable than levels (Section 2.6).

• Original Medicare only. Medicare Advantage enrollees and their care are excluded; by 2024, roughly half of Medicare beneficiaries were in Medicare Advantage (KFF). Findings may not describe the whole Medicare population.

• Practitioner claims only. The provider-service files cover services billed by individual practitioners. They exclude institutional claims, including PT furnished in hospital outpatient departments.

• Privacy suppression. CMS suppresses provider-code rows with fewer than 11 beneficiaries, which can understate low-volume providers and services.

• Provider classification. Specialty identification relies on CMS rendering-provider type. Misclassification or changes in enrollment designation may affect counts.

• Code-level coverage. The code-level extract itemizes the highest-volume codes (more than 99.9% of chiropractic and 99.3% to 99.9% of PT services). The remainder is included in specialty totals only.

• Units are not visits. Services are reported billing units. Timed PT codes may produce several units per visit, while a manipulation is generally one unit; cross-profession service ratios are not visit ratios.

• Code changes and mix effects. The 2017 PT evaluation code replacement prevents direct comparison of evaluation codes across the full period, and blended allowed amounts per service are sensitive to code mix, modifiers, and payment rules such as multiple-procedure reductions.

• Nominal dollars and a general price index. Most amounts are nominal. Inflation adjustment is applied only to the 2017–2024 per-unit comparisons and uses CPI-U, a general consumer price measure rather than a practice-cost index; a practice-cost index such as the Medicare Economic Index could give different results.

• No risk adjustment or causal inference. The analysis does not adjust for age, diagnosis, severity, comorbidity, geography, or socioeconomic factors, and it cannot attribute changes to benefit policy, payment, workforce supply, referral patterns, patient preference, or clinical effectiveness.

Recommended Next Analyses

• Beneficiary-level claims analysis that counts each patient once across providers, links diagnoses, and builds episodes of care.

• State and Medicare-locality analysis, normalized to enrollment and provider supply.

• Inflation-adjusted payment trends for the full 2013–2024 period and against a practice-cost index (for example, the Medicare Economic Index).

• A Medicare Advantage comparison, subject to data availability.

• Functional-outcome and total-cost comparisons through the Arete PBRN or payer collaboration.

Disclosures and Publication Status

This work is classified as a public-data reimbursement and utilization analysis under the Arete Research Library Governance Framework (§2); it is not PBRN-sponsored human-subjects research. This analysis was produced by Arete Healthcare without external funding; Arete supports independent outpatient chiropractic and physical therapy practices and has a commercial and professional interest in Medicare coverage, network strategy, and payer contracting affecting those practices. AI tools assisted with data review, calculation, and drafting. Every factual assertion and calculation requires responsible human verification before publication, and AI reconciliation does not substitute for the named Independent Data/Methodology Reviewer or DAWG approval.

No protected health information or individual-level claims were used. No data were supplied by an interested third party.

AI-Assisted Research and Reconciliation

Earlier drafts were produced with ChatGPT from the CMS extraction and methodology memorandum, then independently recomputed and reconciled by Claude. For this version, Claude rebuilt the report through the shared Arete builder and again recomputed every statistic directly from the source workbook, without relying on prior drafts' figures. Discrepancies are recorded in the internal AI Assistance and Reproducibility Record maintained under the Arete Research Library Governance Framework. A second AI system's agreement is a disclosed first-pass check, not a substitute for human review: it confirms arithmetic, not the underlying extraction method, which both systems received as a shared input. The internal AI Assistance and Reproducibility Record is maintained under the Arete Research Library Governance Framework.

Review Status

This is version 15.1, approved for release. Two non-conflicted DAWG reviewers approved the report without modification on October 19, 2026; review records are maintained in the Arete Operations Platform. The Arete's Take section was written by the Author and approved by at least two of the three Arete Executive Reviewers, as Advocacy/Commentary content separate from the DAWG-reviewed Findings.

Recommended Citation: Elton, D. A Decade of Divergence. Arete Original Research ARETE-OR-2026-004, Version 15.1. Arete Healthcare LLC. Published October 19, 2026. https://arete.healthcare/research/a-decade-of-divergence https://doi.org/10.13140/RG.2.2.29102.88646

References

1. Centers for Medicare & Medicaid Services. Medicare Monthly Enrollment. https://data.cms.gov/summary-statistics-on-beneficiary-enrollment/medicare-and-medicaid-reports/medicare-monthly-enrollment

2. Centers for Medicare & Medicaid Services. Medicare Physician & Other Practitioners — by Provider and Service. https://data.cms.gov/provider-summary-by-type-of-service/medicare-physician-other-practitioners/medicare-physician-other-practitioners-by-provider-and-service

3. Centers for Medicare & Medicaid Services. Medicare Physician & Other Practitioners — by Provider. https://data.cms.gov/provider-summary-by-type-of-service/medicare-physician-other-practitioners/medicare-physician-other-practitioners-by-provider

4. Centers for Medicare & Medicaid Services. Medicare Benefit Policy Manual (Pub. 100-02), Chapter 15, Covered Medical and Other Health Services (Section 240, Chiropractic Services; outpatient therapy sections). https://www.cms.gov/regulations-and-guidance/guidance/manuals/downloads/bp102c15.pdf

5. Centers for Medicare & Medicaid Services, Medicare Learning Network. Medicare Provider Compliance Tips: Chiropractic Services. https://www.cms.gov/training-education/medicare-learning-networkr-mln/compliance/medicare-provider-compliance-tips/chiropractic-services

6. Centers for Medicare & Medicaid Services. 2017 Annual Update to the Therapy Code List. Pub. 100-04 Transmittal 3654, Change Request 9782; November 10, 2016. https://www.cms.gov/regulations-and-guidance/guidance/transmittals/downloads/r3654cp.pdf

7. Centers for Medicare & Medicaid Services. MLN Matters MM11055: Annual Update to the Per-Beneficiary Therapy Amounts (Bipartisan Budget Act of 2018, §50202); December 2018. https://www.cms.gov/outreach-and-education/medicare-learning-network-mln/mlnmattersarticles/downloads/mm11055.pdf

8. Whedon JM, Song Y, Davis MA. Trends in the use and cost of chiropractic spinal manipulation under Medicare Part B. Spine J. 2013;13(11):1449–1454.

9. KFF. Medicare Advantage in 2024: Enrollment Update and Key Trends. August 2024. https://www.kff.org/medicare/medicare-advantage-in-2024-enrollment-update-and-key-trends/

10. U.S. Bureau of Labor Statistics. Consumer Price Index for All Urban Consumers (CPI-U), U.S. city average, all items, annual averages. https://www.bls.gov/regions/mid-atlantic/data/consumerpriceindexhistorical_us_table.htm

11. Arete Healthcare Research & Analytics. Internal reproducibility file: 20260729_CMSData.xlsx; 20260728_MedicareAnalysis_Methods.docx; 20260729_DecadeOfDivergence_SQL.docx; Patient_Count_Reconciliation_SQL.docx; Figure1b_Arete_DecadeOfDivergence_Inflation_2017_2024.xlsx (Mimilabs Databricks, mimi_ws_1.datacmsgov tables).

Glossary

Terms and abbreviations used in this report.

Appendix A — Supplemental Tables: Annual National Summary

Annual values used throughout this report, recomputed from the source workbook. Patient counts use the provider-level beneficiary count (Section 2.4).

Table A1. Original Medicare enrollment and chiropractic

Table A2. Outpatient physical therapy

Related Arete research

• Arete Research & Analytics: all Original Research and Analytic Briefs