Category: 3. Business

Fears of a growing bubble around the artificial intelligence frenzy resurfaced on Thursday as leading US stock markets fell, less than 24 hours after strong results from chipmaker Nvidia sparked a rally.

Wall Street initially rose after Nvidia, the world’s largest public company, reassured investors of strong demand for its advanced data center chips. But the relief dissipated, and technology stocks at the heart of the AI boom came under pressure.

The benchmark S&P 500 closed down 1.6%, and the Dow Jones industrial average closed down 0.8% in New York. The tech-focused Nasdaq Composite closed down 2.2%.

Earlier in the day, the FTSE 100 had closed up 0.2% in London while the Dax had risen 0.5% in Frankfurt. The Nikkei 225 had climbed 2.65% in Tokyo.

Nvidia, now valued at some $4.4tn, has led an extraordinary surge in the valuations of AI-related firms in recent months. As firms splurge on chips and data centers in a bid to get a foothold in AI, fears of a bubble have mounted.

While Nvidia’s highly anticipated earnings exceeded expectations on Wednesday, as the chipmaker continues to enjoy robust demand, concerns persist around the firms using those chips to invest in AI, spending heavily and driving that demand.

“The people who are selling the semiconductors to help power AI doesn’t alleviate the concerns that some of these hyper-scalers are spending way too much money on building the AI infrastructure,” said Robert Pavlik, senior portfolio manager at Dakota Wealth. “You have the company that’s benefiting it, but the others are still spending too much money.”

A mixed jobs report on Thursday morning, which revealed healthy growth in the labor market in September but a slight rise in unemployment, also reinforced expectations that policymakers at the Federal Reserve will likely keep interest rates on hold at their next meeting, in December.

Shares in Nvidia sank 3.2%. The VIX, a measure of market volatility, also climbed 8%.

Reuters contributed reporting

Global Business Solutions Online Ecosystem Revenue Grew 21 percent; Consumer Revenue Grew 21 percent

MOUNTAIN VIEW, Calif.–(BUSINESS WIRE)–
Intuit Inc. (Nasdaq: INTU), the global financial technology platform that makes Intuit TurboTax, Credit Karma, QuickBooks, and Mailchimp, announced financial results for the first quarter of fiscal 2026, which ended October 31.

“We delivered an exceptional first quarter as we continue to execute on our AI-driven expert platform strategy. Intuit is creating a system of intelligence, leveraging data, data services, AI, and human intelligence to fuel the success of consumers, small and mid-market businesses, and accountants,” said Sasan Goodarzi, Intuit’s chief executive officer. “We launched significant done-for-you innovations across our platform that are helping businesses manage from lead to cash, and consumers manage credit building to wealth building, all in one place.”

Financial Highlights

For the first quarter, Intuit:

• Grew total revenue to $3.9 billion, up 18 percent.

• Increased Global Business Solutions revenue to $3.0 billion, up 18 percent; grew Online Ecosystem revenue to $2.4 billion, up 21 percent. Excluding Mailchimp, Global Business Solutions revenue grew 20 percent, and Online Ecosystem revenue grew 25 percent.

• Grew Consumer revenue to $894 million, up 21 percent.

• Increased GAAP operating income to $534 million, up 97 percent.

• Grew non-GAAP operating income to $1.3 billion, up 32 percent.

• Increased GAAP diluted earnings per share to $1.59, up 127 percent.

• Grew non-GAAP diluted earnings per share to $3.34, up 34 percent.

Unless otherwise noted, all growth rates refer to the current period versus the comparable prior-year period, and the business metrics and associated growth rates refer to worldwide business metrics.

Snapshot of First-quarter Results

	GAAP			Non-GAAP
	Q1 FY26	Q1 FY25	Change	Q1 FY26	Q1 FY25	Change
Revenue	$3,885	$3,283	18%	$3,885	$3,283	18%
Operating Income	$534	$271	97%	$1,258	$953	32%
Earnings Per Share	$1.59	$0.70	127%	$3.34	$2.50	34%
Dollars are in millions, except earnings per share. See “About Non-GAAP Financial Measures” below for more information regarding financial measures not prepared in accordance with Generally Accepted Accounting Principles (GAAP).

“We delivered a strong first quarter of fiscal 2026, driven by continued momentum across the company,” said Sandeep Aujla, Intuit’s chief financial officer. “We are confident in delivering double-digit revenue growth and expanding margin this year, and we are reiterating our full year guidance for fiscal 2026.”

Business Segment Results

Global Business Solutions

Global Business Solutions revenue grew to $3.0 billion, up 18 percent, and Online Ecosystem revenue increased to $2.4 billion, up 21 percent.

• QuickBooks Online Accounting revenue grew 25 percent in the quarter, driven by higher effective prices, customer growth, and mix-shift.

• Online Services revenue grew 17 percent, driven by growth in money and payroll offerings.

• Total international online revenue grew 9 percent on a constant currency basis.

Consumer

Consumer revenue of $894 million was up 21 percent in the quarter.

• Credit Karma revenue grew 27 percent to $651 million, driven by strength in personal loans, credit cards, and auto insurance.

• TurboTax revenue grew 6 percent to $198 million.

• ProTax revenue grew 15 percent to $45 million.

Capital Allocation Summary

The company:

• Reported a total cash and investments balance of approximately $3.7 billion and $6.1 billion in debt as of October 31, 2025.

• Repurchased $851 million of stock, and $4.4 billion remains on the company’s share repurchase authorization.

• Received Board approval for a quarterly dividend of $1.20 per share, payable January 16, 2026. This represents a 15 percent increase per share compared to the same period last year.

Forward-looking Guidance

Intuit reiterated guidance for the full fiscal year 2026. The company expects:

• Revenue of $20.997 billion to $21.186 billion, growth of approximately 12 to 13 percent.

• GAAP operating income of $5.782 billion to $5.859 billion, growth of approximately 17 to 19 percent.

• Non-GAAP operating income of $8.611 billion to $8.688 billion, growth of approximately 14 to 15 percent.

• GAAP diluted earnings per share of $15.49 to $15.69, growth of approximately 13 to 15 percent.

• Non-GAAP diluted earnings per share of $22.98 to $23.18, growth of approximately 14 to 15 percent.

The company also reiterated full fiscal year 2026 segment revenue guidance:

• Global Business Solutions: growth of 14 to 15 percent. Excluding Mailchimp, the company expects Global Business Solutions Group revenue growth of 15.5 percent to 16.5 percent.

• Consumer: growth of 8 to 9 percent. This includes TurboTax growth of 8 percent, Credit Karma growth of 10 to 13 percent, and ProTax growth of 2 to 3 percent.

Intuit announced guidance for the second quarter of fiscal year 2026, which ends January 31. The company expects:

• Revenue growth of approximately 14 to 15 percent.

• GAAP diluted earnings per share of $1.76 to $1.81.

• Non-GAAP diluted earnings per share of $3.63 to $3.68.

Conference Call Details

Intuit executives will discuss the financial results on a conference call at 1:30 p.m. Pacific time on November 20. The conference call can be heard live at https://investors.intuit.com/news-events/ir-calendar. Prepared remarks for the call will be available on Intuit’s website after the call ends.

Replay Information

A replay of the conference call will be available for one week by calling 800-934-4245, or 402-220-1173 from international locations. There is no passcode required. The audio call will remain available on Intuit’s website for one week after the conference call.

About Intuit

Intuit is the global financial technology platform that powers prosperity for the people and communities we serve. With approximately 100 million customers worldwide using products such as TurboTax, Credit Karma, QuickBooks, and Mailchimp, we believe that everyone should have the opportunity to prosper. We never stop working to find new, innovative ways to make that possible. Please visit us at Intuit.com and find us on social for the latest information about Intuit and our products and services.

About Non-GAAP Financial Measures

This press release and the accompanying tables include non-GAAP financial measures. For a description of these non-GAAP financial measures, including the reasons management uses each measure, and reconciliations of these non-GAAP financial measures to the most directly comparable financial measures prepared in accordance with Generally Accepted Accounting Principles, please see the section of the accompanying tables titled “About Non-GAAP Financial Measures” as well as the related Table B1, Table B2, and Table E. A copy of the press release issued by Intuit today can be found on the investor relations page of Intuit’s website.

Cautions About Forward-looking Statements

This press release contains forward-looking statements, including expectations regarding: forecasts and timing of growth and future financial results of Intuit and its reporting segments; Intuit’s prospects for the business in fiscal 2026 and beyond; timing and growth of revenue from current or future products and services; demand for our products; customer growth and retention; Intuit’s corporate tax rate; the amount and timing of any future dividends or share repurchases; and the impact of strategic decisions on our business; as well as all of the statements under the heading “Forward-looking Guidance.”

Because these forward-looking statements involve risks and uncertainties, there are important factors that could cause our actual results to differ materially from the expectations expressed in the forward-looking statements. These risks and uncertainties may be amplified by the effects of global developments and conditions or events, including macroeconomic uncertainty and geopolitical conditions, which have caused significant global economic instability and uncertainty. Given these risks and uncertainties, persons reading this communication are cautioned not to place any undue reliance on such forward-looking statements. These factors include, without limitation, the following: our ability to compete successfully; potential governmental encroachment in our tax business; our ability to develop, deploy, and use artificial intelligence in our platform and offerings; our ability to adapt to technological change and to successfully extend our platform; our ability to predict consumer behavior; our ability to anticipate and solve new and existing customer problems; our reliance on intellectual property; our ability to protect our intellectual property rights; any harm to our reputation; risks associated with our environmental, social, and governance efforts; risks associated with acquisition and divestiture activity; the issuance of equity or incurrence of debt to fund acquisitions or for general business purposes; cybersecurity incidents (including those affecting the third parties we rely on); customer or regulator concerns about privacy and cybersecurity incidents; fraudulent activities by third parties, including through the use of AI; our failure to process transactions effectively; interruption or failure of our information technology; our ability to develop and maintain critical third-party business relationships; our ability to attract and retain talent and the success of our hybrid work model; our ability to effectively develop and deploy AI in our offerings; any deficiency in the quality or accuracy of our offerings (including the advice given by experts on our platform); any delays in product launches; difficulties in processing or filing customer tax submissions; risks associated with international operations; risks associated with climate change; changes to, and evolving interpretations of public policy, laws, or regulations affecting our businesses; allegations of legal claims and legal proceedings in which we are involved; fluctuations in the results of our tax business due to seasonality and other factors beyond our control; changes in tax rates and tax reform legislation; global economic conditions (including, without limitation, inflation); exposure to credit, counterparty, and other risks in providing capital to businesses; amortization of acquired intangible assets and impairment charges; our ability to repay or otherwise comply with the terms of our outstanding debt; our ability to repurchase shares or distribute dividends; volatility of our stock price; and our ability to successfully market our offerings.

More details about these and other risks that may impact our business are included in our Form 10-K for fiscal 2025 and in our other SEC filings. You can locate these reports through our website at https://investors.intuit.com. Second-quarter and full-year fiscal 2026 guidance speaks only as of the date it was publicly issued by Intuit. Other forward-looking statements represent the judgment of the management of Intuit as of the date of this presentation. Except as required by law, we do not undertake any duty to update any forward-looking statement or other information in this presentation.

TABLE A

INTUIT INC.

GAAP CONSOLIDATED STATEMENTS OF OPERATIONS

(In millions, except per share amounts)

(Unaudited)

	Three Months Ended
	October 31, 2025				October 31, 2024
Net revenue:
Service	$	3,497			$	2,889
Product and other		388				394
Total net revenue		3,885				3,283
Costs and expenses:
Cost of revenue:
Cost of service revenue		824				772
Cost of product and other revenue		15				14
Amortization of acquired technology		44				37
Selling and marketing		1,082				962
Research and development		843				704
General and administrative		422				394
Amortization of other acquired intangible assets		121				120
Restructuring		—				9
Total costs and expenses [A]		3,351				3,012
Operating income		534				271
Interest expense		(58	)			(60	)
Interest and other income, net		85				2
Income before income taxes		561				213
Income tax provision [B]		115				16
Net income	$	446			$	197
Basic net income per share	$	1.60			$	0.70
Shares used in basic per share calculations		279				280
Diluted net income per share	$	1.59			$	0.70
Shares used in diluted per share calculations		281				283

See accompanying Notes.

INTUIT INC.

NOTES TO TABLE A

[A]	The following table summarizes the total share based compensation expense that we recorded in operating income for the periods shown.

	Three Months Ended
(In millions)	October 31, 2025			October 31, 2024
Cost of revenue	$	97		$	111
Selling and marketing		156			137
Research and development		185			161
General and administrative		105			102
Total share-based compensation expense	$	543		$	511

[B]

We compute our provision for or benefit from income taxes by applying the estimated annual effective tax rate to income or loss from recurring operations and adding the effects of any discrete income tax items specific to the period. We recognized excess tax benefits on share-based compensation of $30 million and $28 million in our provision for income taxes for the three months ended October 31, 2025 and 2024, respectively. Our effective tax rate for the three months ended October 31, 2025 was approximately 20%. Excluding discrete tax items primarily related to share-based compensation, our effective tax rate was approximately 24%. The difference from the federal statutory rate of 21% was primarily due to state income taxes and non-deductible share-based compensation, which were partially offset by the tax benefit we received from the federal research and experimentation credit. Our effective tax rate for the three months ended October 31, 2024 was approximately 8%. Excluding discrete tax items primarily related to share-based compensation, our effective tax rate was approximately 24%. The difference from the federal statutory rate of 21% was primarily due to state income taxes and non-deductible share-based compensation, which were partially offset by the tax benefit we received from the federal research and experimentation credit. In the current global tax policy environment, the U.S. and other domestic and foreign governments continue to consider, and in some cases enact, changes in corporate tax laws. As changes occur, we account for finalized legislation in the period of enactment.

TABLE B1

INTUIT INC.

RECONCILIATION OF NON-GAAP FINANCIAL MEASURES

TO MOST DIRECTLY COMPARABLE GAAP FINANCIAL MEASURES

(In millions, except per share amounts)

(Unaudited)

	Fiscal 2026
	Q1				Q2			Q3			Q4			Year to Date
GAAP operating income (loss)	$	534			$	—		$	—		$	—		$	534
Amortization of acquired technology		44				—			—			—			44
Amortization of other acquired intangible assets		121				—			—			—			121
Net (gain) loss on executive deferred compensation plan liabilities		16				—			—			—			16
Share-based compensation expense		543				—			—			—			543
Non-GAAP operating income (loss)	$	1,258			$	—		$	—		$	—		$	1,258
GAAP net income (loss)	$	446			$	—		$	—		$	—		$	446
Amortization of acquired technology		44				—			—			—			44
Amortization of other acquired intangible assets		121				—			—			—			121
Net (gain) loss on executive deferred compensation plan liabilities		16				—			—			—			16
Share-based compensation expense		543				—			—			—			543
Net (gain) loss on debt securities and other investments [A]		(34	)			—			—			—			(34	)
Net (gain) loss on executive deferred compensation plan assets		(15	)			—			—			—			(15	)
Income tax effects and adjustments [B]		(182	)			—			—			—			(182	)
Non-GAAP net income (loss)	$	939			$	—		$	—		$	—		$	939
GAAP diluted net income (loss) per share	$	1.59			$	—		$	—		$	—		$	1.59
Amortization of acquired technology		0.16				—			—			—			0.16
Amortization of other acquired intangible assets		0.43				—			—			—			0.43
Net (gain) loss on executive deferred compensation plan liabilities		0.05				—			—			—			0.05
Share-based compensation expense		1.93				—			—			—			1.93
Net (gain) loss on debt securities and other investments [A]		(0.12	)			—			—			—			(0.12	)
Net (gain) loss on executive deferred compensation plan assets		(0.05	)			—			—			—			(0.05	)
Income tax effects and adjustments [B]		(0.65	)			—			—			—			(0.65	)
Non-GAAP diluted net income (loss) per share	$	3.34			$	—		$	—		$	—		$	3.34
Shares used in GAAP diluted per share calculations		281				—			—			—			281
Shares used in non-GAAP diluted per share calculations		281				—			—			—			281

[A]	During the three months ended October 31, 2025, we recognized $34 million in net gains on other long-term investments.
[B]	As discussed in “About Non-GAAP Financial Measures – Income Tax Effects and Adjustments” following Table E, our long-term non-GAAP tax rate eliminates the effects of non-recurring and period-specific items. Income tax adjustments consist primarily of the tax impact of the non-GAAP pre-tax adjustments and tax benefits related to share-based compensation.
See “About Non-GAAP Financial Measures” immediately following Table E for information on these measures, the items excluded from the most directly comparable GAAP measures in arriving at non-GAAP financial measures, and the reasons management uses each measure and excludes the specified amounts in arriving at each non-GAAP financial measure.

TABLE B2

INTUIT INC.

RECONCILIATION OF NON-GAAP FINANCIAL MEASURES

TO MOST DIRECTLY COMPARABLE GAAP FINANCIAL MEASURES

(In millions, except per share amounts)

(Unaudited)

	Fiscal 2025
	Q1				Q2				Q3				Q4				Full Year
GAAP operating income (loss)	$	271			$	593			$	3,720			$	339			$	4,923
Amortization of acquired technology		37				37				38				44				156
Amortization of other acquired intangible assets		120				120				120				121				481
Restructuring		9				4				1				1				15
Professional fees for business combinations		—				—				2				—				2
Net (gain) loss on executive deferred compensation plan liabilities		5				8				(7	)			21				27
Share-based compensation expense		511				498				469				490				1,968
Non-GAAP operating income (loss)	$	953			$	1,260			$	4,343			$	1,016			$	7,572
GAAP net income (loss)	$	197			$	471			$	2,820			$	381			$	3,869
Amortization of acquired technology		37				37				38				44				156
Amortization of other acquired intangible assets		120				120				120				121				481
Restructuring		9				4				1				1				15
Professional fees for business combinations		—				—				2				—				2
Net (gain) loss on executive deferred compensation plan liabilities		5				8				(7	)			21				27
Share-based compensation expense		511				498				469				490				1,968
Net (gain) loss on debt securities and other investments [A]		42				3				2				(2	)			45
Net (gain) loss on executive deferred compensation plan assets		(4	)			(7	)			7				(20	)			(24	)
Income tax effects and adjustments [B]		(208	)			(196	)			(172	)			(260	)			(836	)
Non-GAAP net income (loss)	$	709			$	938			$	3,280			$	776			$	5,703
GAAP diluted net income (loss) per share	$	0.70			$	1.67			$	10.02			$	1.35			$	13.67
Amortization of acquired technology		0.13				0.13				0.13				0.16				0.55
Amortization of other acquired intangible assets		0.42				0.42				0.43				0.43				1.70
Restructuring		0.03				0.01				—				—				0.05
Professional fees for business combinations		—				—				0.01				—				0.01
Net (gain) loss on executive deferred compensation plan liabilities		0.02				0.03				(0.02	)			0.07				0.10
Share-based compensation expense		1.80				1.76				1.66				1.74				6.95
Net (gain) loss on debt securities and other investments [A]		0.15				0.01				0.01				(0.01	)			0.16
Net (gain) loss on executive deferred compensation plan assets		(0.02	)			(0.02	)			0.02				(0.07	)			(0.09	)
Income tax effects and adjustments [B]		(0.73	)			(0.69	)			(0.61	)			(0.92	)			(2.95	)
Non-GAAP diluted net income (loss) per share	$	2.50			$	3.32			$	11.65			$	2.75			$	20.15
Shares used in GAAP diluted per share calculations		283				283				282				282				283
Shares used in non-GAAP diluted per share calculations		283				283				282				282				283

[A]	During the three months ended October 31, 2024, we recognized a $42 million net loss on other long-term investments.
[B]	As discussed in “About Non-GAAP Financial Measures – Income Tax Effects and Adjustments” following Table E, our long-term non-GAAP tax rate eliminates the effects of non-recurring and period-specific items. Income tax adjustments consist primarily of the tax impact of the non-GAAP pre-tax adjustments and tax benefits related to share-based compensation.
See “About Non-GAAP Financial Measures” immediately following Table E for information on these measures, the items excluded from the most directly comparable GAAP measures in arriving at non-GAAP financial measures, and the reasons management uses each measure and excludes the specified amounts in arriving at each non-GAAP financial measure.

TABLE C

INTUIT INC.

CONDENSED CONSOLIDATED BALANCE SHEETS

(In millions)

(Unaudited)

	October 31, 2025			July 31, 2025
ASSETS
Current assets:
Cash and cash equivalents	$	3,506		$	2,884
Investments		190			1,668
Accounts receivable, net		579			530
Notes receivable held for investment		1,519			1,403
Notes receivable held for sale		48			—
Income taxes receivable		31			50
Prepaid expenses and other current assets		630			496
Current assets before funds receivable and amounts held for customers		6,503			7,031
Funds receivable and amounts held for customers		3,918			7,076
Total current assets		10,421			14,107
Long-term investments		92			94
Property and equipment, net		965			961
Operating lease right-of-use assets		596			541
Goodwill		13,980			13,980
Acquired intangible assets, net		5,136			5,302
Long-term deferred income tax assets		1,173			1,222
Other assets		828			751
Total assets	$	33,191		$	36,958
LIABILITIES AND STOCKHOLDERS’ EQUITY
Current liabilities:
Short-term debt	$	749		$	—
Accounts payable		670			792
Accrued compensation and related liabilities		479			858
Deferred revenue		1,045			1,019
Other current liabilities		658			625
Current liabilities before funds payable and amounts due to customers		3,601			3,294
Funds payable and amounts due to customers		3,918			7,076
Total current liabilities		7,519			10,370
Long-term debt		5,391			5,973
Operating lease liabilities		643			597
Other long-term obligations		316			308
Total liabilities		13,869			17,248
Stockholders’ equity		19,322			19,710
Total liabilities and stockholders’ equity	$	33,191		$	36,958

TABLE D

INTUIT INC.

CONDENSED CONSOLIDATED STATEMENTS OF CASH FLOWS

(In millions)

(Unaudited)

	Three Months Ended
	October 31, 2025				October 31, 2024
Cash flows from operating activities:
Net income	$	446			$	197
Adjustments to reconcile net income to net cash provided by operating activities:
Depreciation		44				44
Amortization of acquired intangible assets		165				157
Non-cash operating lease cost		23				19
Share-based compensation expense		543				511
Deferred income taxes		58				(91	)
Other		(6	)			63
Total adjustments		827				703
Changes in operating assets and liabilities:
Accounts receivable		(49	)			31
Income taxes receivable		19				51
Prepaid expenses and other assets		(119	)			(27	)
Accounts payable		(135	)			(75	)
Accrued compensation and related liabilities		(378	)			(507	)
Deferred revenue		25				19
Operating lease liabilities		(23	)			(22	)
Other liabilities		24				(8	)
Total changes in operating assets and liabilities		(636	)			(538	)
Net cash provided by operating activities		637				362
Cash flows from investing activities:
Purchases of corporate and customer fund investments		(101	)			(306	)
Sales of corporate and customer fund investments		115				55
Maturities of corporate and customer fund investments		1,473				235
Purchases of property and equipment		(38	)			(33	)
Originations and purchases of notes receivable held for investment		(1,297	)			(666	)
Sales of notes receivable originally classified as held for investment		213				110
Principal repayments of notes receivable held for investment		876				420
Other		(43	)			(3	)
Net cash provided by (used in) investing activities		1,198				(188	)
Cash flows from financing activities:
Proceeds from borrowings under secured revolving credit facilities		166				85
Proceeds from issuance of stock under employee stock plans		62				96
Payments for employee taxes withheld upon vesting of restricted stock units		(244	)			(239	)
Cash paid for purchases of treasury stock		(854	)			(557	)
Dividends and dividend rights paid		(341	)			(296	)
Net change in funds receivable and funds payable and amounts due to customers		(3,160	)			1,672
Other		(1	)			—
Net cash provided by (used in) financing activities		(4,372	)			761
Effect of exchange rates on cash, cash equivalents, restricted cash, and restricted cash equivalents		(1	)			—
Net increase (decrease) in cash, cash equivalents, restricted cash, and restricted cash equivalents		(2,538	)			935
Cash, cash equivalents, restricted cash, and restricted cash equivalents at beginning of period		9,481				7,099
Cash, cash equivalents, restricted cash, and restricted cash equivalents at end of period	$	6,943			$	8,034
Reconciliation of cash, cash equivalents, restricted cash, and restricted cash equivalents reported within the condensed consolidated balance sheets to the total amounts reported on the condensed consolidated statements of cash flows
Cash and cash equivalents	$	3,506			$	2,872
Restricted cash and restricted cash equivalents included in funds receivable and amounts held for customers		3,437				5,162
Total cash, cash equivalents, restricted cash, and restricted cash equivalents at end of period	$	6,943			$	8,034
Supplemental schedule of non-cash investing activities:
Transfers of notes receivable originated or purchased as held for investment to held for sale	$	253			$	113

TABLE E

INTUIT INC.

RECONCILIATION OF FORWARD-LOOKING GUIDANCE FOR NON-GAAP FINANCIAL MEASURES TO PROJECTED GAAP REVENUE, OPERATING INCOME, AND EPS

(In millions, except per share amounts)

(Unaudited)

	Forward-Looking Guidance
	GAAP Range of Estimate									Non-GAAP Range of Estimate
	From			To			Adjmts			From			To
Three Months Ending January 31, 2026
Revenue	$	4,519		$	4,549		$	—		$	4,519		$	4,549
Operating income	$	676		$	691		$	695	[a]	$	1,371		$	1,386
Diluted net income per share	$	1.76		$	1.81		$	1.87	[b]	$	3.63		$	3.68
Twelve Months Ending July 31, 2026
Revenue	$	20,997		$	21,186		$	—		$	20,997		$	21,186
Operating income	$	5,782		$	5,859		$	2,829	[c]	$	8,611		$	8,688
Diluted net income per share	$	15.49		$	15.69		$	7.49	[d]	$	22.98		$	23.18

See “About Non-GAAP Financial Measures” immediately following Table E for information on these measures, the items excluded from the most directly comparable GAAP measures in arriving at non-GAAP financial measures, and the reasons management uses each measure and excludes the specified amounts in arriving at each non-GAAP financial measure.
[a]	Reflects estimated adjustments for share-based compensation expense of approximately $530 million; amortization of other acquired intangible assets of approximately $121 million; and amortization of acquired technology of approximately $44 million.
[b]	Reflects estimated adjustments in item [a], income taxes related to these adjustments, and other income tax effects related to the use of the non-GAAP tax rate.
[c]	Reflects estimated adjustments for share-based compensation expense of approximately $2.2 billion; amortization of other acquired intangible assets of approximately $483 million; amortization of acquired technology of approximately $176 million; and net losses on executive deferred compensation plan liabilities of approximately $16 million.
[d]	Reflects estimated adjustments in item [c], income taxes related to these adjustments, other income tax effects related to the use of the non-GAAP tax rate, and adjustments for a net loss on other long-term investments.

INTUIT INC.

ABOUT NON-GAAP FINANCIAL MEASURES

The accompanying press release dated November 20, 2025 contains non-GAAP financial measures. Table B1, Table B2, and Table E reconcile the non-GAAP financial measures in that press release to the most directly comparable financial measures prepared in accordance with Generally Accepted Accounting Principles (GAAP). These non-GAAP financial measures include non-GAAP operating income (loss), non-GAAP net income (loss), and non-GAAP diluted net income (loss) per share.

Non-GAAP financial measures should not be considered as a substitute for, or superior to, measures of financial performance prepared in accordance with GAAP. These non-GAAP financial measures do not reflect a comprehensive system of accounting, differ from GAAP measures with the same names, and may differ from non-GAAP financial measures with the same or similar names that are used by other companies.

We compute non-GAAP financial measures using the same consistent method from quarter to quarter and year to year. We may consider whether other significant items that arise in the future should be excluded from our non-GAAP financial measures.

We exclude the following items from all of our non-GAAP financial measures:

• Amortization of acquired technology

• Amortization of other acquired intangible assets

• Restructuring charges

• Share-based compensation expense

• Gains and losses on executive deferred compensation plan liabilities

• Goodwill and intangible asset impairment charges

• Gains and losses on disposals of businesses and long-lived assets

• Professional fees and transaction costs for business combinations

We also exclude the following items from non-GAAP net income (loss) and diluted net income (loss) per share:

• Gains and losses on debt securities and other investments

• Gains and losses on executive deferred compensation plan assets

• Income tax effects and adjustments

• Discontinued operations

We believe these non-GAAP financial measures provide meaningful supplemental information regarding Intuit’s operating results primarily because they exclude amounts that we do not consider part of ongoing operating results when planning and forecasting and when assessing the performance of the organization, our individual operating segments, or our senior management. Segment managers are not held accountable for share-based compensation expense, amortization, restructuring, or the other excluded items and, accordingly, we exclude these amounts from our measures of segment performance. We believe our non-GAAP financial measures also facilitate the comparison by management and investors of results for current periods and guidance for future periods with results for past periods.

The following are descriptions of the items we exclude from our non-GAAP financial measures.

Amortization of acquired technology and amortization of other acquired intangible assets. When we acquire a business in a business combination, we are required by GAAP to record the fair values of the intangible assets of the business and amortize them over their useful lives. Amortization of acquired technology in cost of revenue includes amortization of software and other technology assets of acquired businesses. Amortization of other acquired intangible assets in operating expenses includes amortization of assets such as customer lists and trade names.

Restructuring charges. This consists of costs incurred as a direct result of discrete strategic restructuring actions, including, but not limited to severance and other one-time termination benefits, and other costs, which are different in terms of size, strategic nature, and frequency than ongoing productivity and business improvements.

Share-based compensation expense. This consists of non-cash expenses for stock options, restricted stock units, and our Employee Stock Purchase Plan. When considering the impact of equity awards, we place greater emphasis on overall shareholder dilution rather than the accounting charges associated with those awards.

Gains and losses on executive deferred compensation plan liabilities. We exclude from our non-GAAP financial measures gains and losses on the revaluation of our executive deferred compensation plan liabilities.

Goodwill and intangible asset impairment charges. We exclude from our non-GAAP financial measures non-cash charges to adjust the carrying values of goodwill and other acquired intangible assets to their estimated fair values.

Gains and losses on disposals of businesses and long-lived assets. We exclude from our non-GAAP financial measures gains and losses on disposals of businesses and long-lived assets because they are unrelated to our ongoing business operating results.

Professional fees and transaction costs for business combinations. We exclude from our non-GAAP financial measures the professional fees we incur to complete business combinations. These include investment banking, legal, and accounting fees.

Gains and losses on debt securities and other investments. We exclude from our non-GAAP financial measures credit losses on available-for-sale debt securities and gains and losses on other investments.

Gains and losses on executive deferred compensation plan assets. We exclude from our non-GAAP financial measures gains and losses on the revaluation of our executive deferred compensation plan assets.

Income tax effects and adjustments. We use a long-term non-GAAP tax rate for evaluating operating results and for planning, forecasting, and analyzing future periods. This long-term non-GAAP tax rate excludes the income tax effects of the non-GAAP pre-tax adjustments described above, and eliminates the effects of non-recurring and period specific items which can vary in size and frequency. Based on our long-term projections, we are using a long-term non-GAAP tax rate of 24% for fiscal 2025 and fiscal 2026. This long-term non-GAAP tax rate could be subject to change for various reasons including significant acquisitions, changes in our geographic earnings mix, or fundamental tax law changes in major jurisdictions in which we operate. We will evaluate this long-term non-GAAP tax rate on an annual basis and whenever any significant events occur which may materially affect this rate.

Operating results and gains and losses on the sale of discontinued operations. From time to time, we sell or otherwise dispose of selected operations as we adjust our portfolio of businesses to meet our strategic goals. In accordance with GAAP, we segregate the operating results of discontinued operations as well as gains and losses on the sale of these discontinued operations from continuing operations on our GAAP statements of operations but continue to include them in GAAP net income or loss and net income or loss per share. We exclude these amounts from our non-GAAP financial measures.

The reconciliations of the forward-looking non-GAAP financial measures to the most directly comparable GAAP financial measures in Table E include all information reasonably available to Intuit at the date of this press release. These tables include adjustments that we can reasonably predict. Events that could cause the reconciliation to change include acquisitions and divestitures of businesses, goodwill and other asset impairments, sales of available-for-sale debt securities and other investments, and disposals of businesses and long-lived assets.

View source version on businesswire.com: https://www.businesswire.com/news/home/20251120052286/en/

Investors

Kim Watkins

Intuit Inc.

650-944-3324

kim_watkins@intuit.com

Media

Kali Fry

Intuit Inc.

650-944-3036

kali_fry@intuit.com

Source: Intuit Inc.

Released November 20, 2025

Introduction

Background

The digital transformation of health care industries worldwide has been dedicated to promoting initiatives that aim to improve human health and quality of life []. In South Korea, digital health care is increasingly integrated into various health care services, encompassing personal health records, mobile health, health information technology, wearable devices, telehealth and telemedicine, personalized medicine, and digital therapeutics. These comprehensive approaches empower consumers by enabling them to independently manage and control health and well-being []. Particularly, electronic personal health records (e-PHRs) refer to systems where individuals centrally manage and integrate lifelong health information and selectively share it with chosen recipients []. Such systems have gained attention for the utmost potential to provide valuable data for personalized health care services, thereby addressing societal health challenges []. Furthermore, the usage of personal health records within the health care sector has expanded into various domains, including personal medical device–linked health management services and public health care systems. With advancing technology, this usage is increasingly extended to wearable devices [].

Globally, e-PHRs have become increasingly important as the Fourth Industrial Revolution accelerated the expansion of remote medical services, prompting the development of consumer-centered, effective health management solutions []. In particular, global IT corporations are playing pivotal roles in establishing and advancing the digital health care ecosystem through devices and platform technologies. Prominent international examples include the “Blue Button” service introduced by the United States Department of Veterans Affairs in 2010, Australia’s “My Health Record” managed by the Australian Digital Health Agency, and Canada’s establishment of regional health integration networks to facilitate coordination among health care providers and care centers. Additionally, countries such as Sweden, Norway, the Netherlands, France, Germany, Australia, and Singapore have undertaken significant efforts to develop platforms that enable individuals to access and use personal health data easily [].

However, as remote interactions become commonplace and living environments rapidly transition to an online-centered context, people with disabilities face significant challenges in adapting to these changes due to limitations in daily activities and the accelerating pace of digital transformation []. If smart health care environments are designed without taking into account the different types of disabilities and physical limitations, people with disabilities may face significant challenges in using these devices effectively [,]. Furthermore, when offering telemedicine services to individuals with disabilities, issues related to language, cognitive abilities, and sensory limitations can hinder effective communication between users and health care providers, thus highlighting and worsening inequalities in digital accessibility []. Therefore, addressing health rights issues among people with disabilities necessitates heightened social attention and proactive integration of digital health care services. Furthermore, comprehensive health information management, starting from hospital-based care and extending through rehabilitation services, must be linked systematically with community-based support networks to effectively meet the health care needs of people with disabilities [].

Recent studies indicate that numerous developed welfare states, including those in Europe, the United States, and Japan, are actively implementing e-PHRs to promote healthier individuals and societies [,]. With the paradigm shift from hospital- and health care provider-centered care to patient-centered approaches, driven by the integration of medical information and scientific technologies, comprehensive data, including lifestyle habits, medical information, and patient emergency conditions, are increasingly used for both treatment and preventive health care []. Currently, e-PHR services are primarily implemented in standalone formats through health care institutions, using mobile apps that provide individuals with basic personal health information, including appointment scheduling, health screenings, medication details, and laboratory results [].

However, unlike people without disabilities, those with disabilities frequently interact with multiple institutions, such as hospitals, public health centers, disability welfare centers, and rehabilitation facilities, from the onset of disability onward for treatment, rehabilitation, and health management []. Given that information provided solely through medical institutions is limited for comprehensive health management, there is an urgent need to emphasize community-based rehabilitation approaches and develop integrated e-PHR services to facilitate holistic health care management for people with disabilities [].

To address these challenges, it is necessary to investigate the causal relationships among external variables influencing the intention to use e-PHRs for health management among people with disabilities, using the technology acceptance model (TAM) framework []. Additionally, comparative analyses of multiple structural models can further provide empirical insights into the practical implementation of e-PHR services for people with disabilities []. Understanding perceptions toward e-PHR services among people with disabilities is fundamental not only to improving the quality of life but also to enhancing broader social metrics, such as preventing secondary disabilities, managing chronic illnesses, and reducing medical costs, thereby contributing to the establishment of a healthier society. Emphasizing social values and fostering sustained resource development is crucial for collectively addressing social, psychological, and physical challenges encountered by people with disabilities. Therefore, this study aims to conduct a comprehensive survey among people with disabilities in South Korea, focusing on the usage of e-PHR services among various digital health care platforms to manage health effectively. Using structural equation modeling (SEM) based on the TAM, this study aims to identify and analyze service-related factors influencing the intention to use e-PHRs among people with disabilities, thereby predicting perceptions toward e-PHR services prior to the actual implementation and providing foundational data for future service development.

Although the TAM has been widely used in health care adoption research, previous studies focusing on people with disabilities typically have several shortcomings. These studies often (1) concentrate on single disability groups or small clinical cohorts [], (2) overlook considerations of consent, security, and content quality that are essential for people with disabilities navigating fragmented care [], and (3) treat digital skills as universally enabling rather than recognizing them as potentially critical or obstructive []. To address these gaps, this study expands on the foundational TAM framework (perceived ease of use [PEU] → perceived usefulness [PU] → usage intention [UI]) by incorporating 6 external factors that are relevant for people with disabilities. These factors include health information consent (HIC; willingness to share data amidst privacy concerns), information security (IS; trust in protective measures), content characteristics (CC; structure, clarity, and cognitive load), effectiveness (EF; perceived assistance and facilitating conditions across different institutions), health consciousness (HC), and eHealth literacy (eHL). Using a large, nationally representative sample of people with disabilities from various health institutions, this study tests several pathways and reveals some unexpected effects. This study expands the core TAM by incorporating 6 important external factors related to disability. By testing this model in a large, nationally representative sample, the research clarifies how design features, consent and security considerations, and supportive conditions influence acceptance among individuals with disabilities. Therefore, this study lays the foundation for developing more responsive e-PHR services that cater to the needs of people with disabilities.

Research Model

This study develops a research model based on the TAM, introduced by Davis [], and integrates findings from prior research related to the intention to use digital health care services. The model incorporates 6 external factors, such as HC, consent to use health information, CC, IS, eHL, and EF. The primary objective is to investigate how these external factors influence UI through the mediating roles of PEU and PU. The proposed research model is illustrated in .

‎

Methods

Participants and Data Collection

This study conducted a nationwide survey as a foundational investigation into e-PHRs within the digital health landscape. The sampling methods used were proportionate stratified sampling and systematic stratified cluster sampling. The selection criteria included individuals aged 19 years or older with physical disabilities diagnosed at least 3 years prior, who had used or were currently using community-based institutions, such as public health centers or disability welfare centers. Participants were excluded from the study for the following reasons: (1) those with Mini-Mental State Examination (MMSE) scores below 24 were excluded to ensure cognitive ability sufficient to comprehend [] and voluntarily participate in the survey, thereby enhancing the validity and reliability of the collected data; (2) individuals who were hospitalized for acute care, undergoing active surgery or treatment, or unable to complete the survey, for example, due to incomplete responses or withdrawal during participation, were also excluded to reduce bias related to acute clinical status and to ensure representativeness of stable, community-dwelling individuals with disabilities; and (3) surveys with missing data on key variables necessary for analysis were excluded according to established guidelines for missing data based on the “Statistical Analysis” section to ensure data integrity. These criteria were adopted to ensure that responses accurately reflected the experiences, intentions, and health conditions of the target population capable of meaningful engagement with digital health technologies for health management.

Recruitment took place across 3 community-based settings in South Korea, namely rehabilitation hospitals, disability welfare centers, and public health centers, from August 30 to November 30, 2023. Each participating site designated a gatekeeper who undertook the following tasks: (1) posted Institutional Review Board (IRB)–approved flyers in public areas and program rooms; (2) distributed large-print and plain-language information sheets during group sessions; and (3) provided assisted, standardized screening. To implement proportionate stratified and systematic cluster sampling, target quotas were allocated to strata defined by region (metropolitan vs medium or small cities) and type of institution. Within each stratum, sites (clusters) were selected, and on preselected recruitment days, staff approached every k-th eligible visitor (with k determined by expected daily traffic, typically ranging from 3 to 5) to minimize selection bias. A total of 1217 participants completed the survey; however, only 800 participants’ survey data (a response rate of 65.7%) were available for analysis.

Sample Size Calculation

To determine an appropriate sample size, guidelines provided in prior literature were followed. According to Hair et al [] and Kline [], SEM requires at least 10-20 respondents per observed variable (questionnaire item) to ensure reliable parameter estimates and model fit. In this research, the measurement instruments comprised 73 observed variables across 9 latent variables. Based on these criteria, the required sample size was calculated as follows: the required minimum sample size was calculated by multiplying the number of observed variables (73) by the recommended minimum number of respondents per variable (10), resulting in a total of 730 respondents. Considering the need to secure sufficient statistical power and to address possible missing or invalid responses, the final sample size was set at 800 respondents. This figure comfortably exceeds the recommended sample size based on the number of observed variables, ensuring suitability for SEM analysis using AMOS (Analysis of Moment Structures).

Measurement Instruments

The survey instrument was developed by modifying and refining existing measurement tools from previous literature and theoretical frameworks related to technology acceptance, ensuring alignment with the study’s objectives and context.

Since the sample included people with disabilities, all instruments were adapted in advance according to universal design principles to ensure accessibility while preserving the intended meaning. Accommodation was made for various needs, including (1) visual accommodation, such as large print materials and read-aloud options; (2) hearing accommodation, such as sign language interpretation or captioning; (3) motor accommodation that involved assistance with pointing or dictation and extended time for completion; and (4) cognitive or communication support through plain-language summaries and brief examples. Researchers also supported augmentative and alternative communication methods, such as tablet typing or pointing, and ensured that surveys could be completed privately, with adequate time provided for all participants. Administration followed a standardized protocol for completing assessments through self-administration or with the assistance of an interviewer, with assistants receiving brief training in disability etiquette, neutrality, and confidentiality. For each case, the mode of administration, type of assistance provided, and time taken to complete the assessment were recorded. Content validity was ensured through a pilot study focused on individuals with disabilities, during which a multidisciplinary panel—including a rehabilitation physician, a public health expert, and representatives from the community of persons with disabilities—reviewed the clarity and relevance of the assessment items, making necessary revisions. Cognitive debriefing, along with a small pilot study involving 10 individuals with physical disabilities, confirmed the assessment’s feasibility and understanding, resulting in minor changes to wording and response options. Following the above content validity testing and a pilot survey, the final questionnaire was established.

The TAM was used to explore intentions regarding the use of e-PHRs for health management among people with disabilities. The finalized survey comprised a total of 73 items organized into 9 categories, namely HC; 8 items, HIC (8 items), CC (6 items), IS (4 items), eHL (14 items), EF (16 items), PU (7 items), PEU (5 items), and intention to use (5 items). Detailed descriptions of the measurement instruments are presented in .

HC generally refers to the level of active engagement and efforts directed toward health promotion and disease prevention, particularly using e-PHRs []. The measurement instrument was developed by modifying and refining items from the HC questionnaire originally used by Belloc and Breslow []. Responses were measured on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with higher scores indicating greater engagement in preventive health behaviors, such as exercise and dietary management. Previous research reported an internal consistency reliability (Cronbach α) of 0.852. In this study, Cronbach α was 0.843, confirming reliability. Exploratory factor analysis (EFA) yielded a KMO (Kaiser-Meyer-Olkin) value of 0.787, indicating adequate sampling adequacy.

Given the exponential increase in personal health data in the digital age, HIC, specifically consent for data sharing, raises significant privacy concerns, necessitating broad social consensus []. The measurement instrument for HIC was developed by adapting and refining items based on the health and psychological theories used by Bowman et al [], specifically addressing consent related to the disclosure of personal health information. The questionnaire used a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicated a greater willingness to disclose personal health information, such as health status, medical examinations, treatment details, and exercise information, to relevant stakeholders. The internal consistency reliability (Cronbach α) was 0.886 in previous research, while in this study, Cronbach α was 0.945, demonstrating strong reliability. EFA yielded a KMO value of 0.892, indicating excellent sampling adequacy.

CC in the context of digital technologies refers to the ability of users to freely access and use information through various platforms, including computers, mobile devices, and the internet []. The measurement instrument for CC was developed by modifying and refining items based on previous studies [,]. Responses were recorded using a 5-point Likert scale ranging from 1 (not helpful at all) to 5 (extremely helpful). Higher scores indicated greater perceived benefits derived from content attributes, such as mobility and personalization. Internal consistency reliability (Cronbach α) from previous research was 0.938, while in this study, Cronbach α was 0.949, demonstrating excellent reliability. EFA yielded a KMO value of 0.921, indicating strong sampling adequacy.

IS refers to the technical measures implemented to prevent corruption, alteration, or unauthorized disclosure of information. Issues related to personal data protection and security represent significant social concerns and act as barriers to the broader adoption of ITs []. In this study, IS pertains specifically to trust and beliefs regarding the protection and security of personal data within e-PHR services. The measurement instrument for IS was developed by modifying and refining items derived from prior research by van Houwelingen et al [], Shareef et al [], and Lee and Ham []. Responses were recorded on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicated greater trust and confidence in IS, reflecting perceptions of higher safety, confidentiality, and reliability. The internal consistency reliability (Cronbach α) was 0.856 in previous research, and Cronbach α in this study was 0.931, indicating excellent reliability. EFA yielded a KMO value of 0.850, signifying strong sampling adequacy.

eHL refers to a comprehensive set of skills required for acquiring and effectively using basic health information within health care contexts. Additionally, it is recognized as a critical determinant of health outcomes, shaping health-related decisions and enabling predictive approaches to everyday health management []. The measurement instrument for eHL was developed by modifying and refining items from the eHL scale originally proposed by Nutbeam [] and subsequently adapted for the Korean context by Chung et al. []. Responses were recorded on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree), with higher scores indicating greater proficiency in eHL, a stronger understanding of health concepts, and a higher level of preventive health behaviors. Internal consistency reliability (Cronbach α) was 0.961 in previous research, and Cronbach α obtained in this study was 0.842, indicating satisfactory reliability. EFA yielded a KMO value of 0.894, signifying excellent sampling adequacy.

The EF provided by e-PHRs to people with disabilities measures the perceived helpfulness of applying this technology to disability-related health management []. In this study, an assistance degree specifically refers to perceptions regarding the utility and beneficial impact of e-PHR services on personal health management. The measurement instrument was developed by modifying and refining items based on the PHR System Functional Model R1 report by the Healthcare Information and Management System Society (HIMSS []). Responses were recorded using a 5-point Likert scale ranging from 1 (not helpful at all) to 5 (extremely helpful). Higher scores indicated a greater perceived level of utility and assistance from e-PHR services. Internal consistency reliability (Cronbach α) reported in previous research was 0.931, while the reliability for this study was confirmed with a Cronbach α of 0.972, indicating excellent reliability. EFA yielded a KMO value of 0.943, demonstrating robust sampling adequacy.

PU refers to the degree to which individuals believe that using health management services can effectively enhance health outcomes and improve management efficiency []. The measurement instrument was developed by modifying and refining items based on the original constructs proposed by Venkatesh and Davis [] and further adapted by Choi et al. []. Responses were recorded using a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicated stronger perceptions that e-PHR services would offer efficient and beneficial assistance. Previous research reported internal consistency reliability (Cronbach α) of 0.899, and this study confirmed excellent reliability with a Cronbach α of 0.955. EFA yielded a KMO value of 0.910, indicating robust sampling adequacy.

The PEU refers to the extent to which individuals perceive e-PHR services as easy and effortless to use, reflecting perceptions of ease in accessing and accepting the technology []. The measurement instrument was developed through the modification and refinement of items derived from the original constructs proposed by Venkatesh and Davis []. Responses were recorded using a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicated stronger perceptions of ease and convenience in using e-PHR services for health management. Previous research reported an internal consistency reliability (Cronbach α) of 0.902, and the reliability confirmed in this study was Cronbach α of 0.914, indicating excellent reliability. EFA yielded a KMO value of 0.836, suggesting strong sampling adequacy.

UI refers to an individual’s intention or commitment to accept and continuously use e-PHR services []. In this study, UI specifically addresses planned or intended future adoption and continued usage of e-PHR services. The measurement instrument was developed by adapting and refining items from the original constructs developed by Venkatesh et al []. Responses were measured on a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). Higher scores indicated a greater intention to use e-PHR services. Previous research reported an internal consistency reliability (Cronbach α) of 0.936. The reliability confirmed in this study was also Cronbach α of 0.936, indicating excellent reliability. EFA yielded a KMO value of 0.874, suggesting strong sampling adequacy.

Statistical Analysis

Descriptive statistics were used to summarize the participants’ general characteristics, while differences based on these characteristics were analyzed using 2-tailed independent t tests and ANOVA. The validity and reliability of the survey instrument were assessed through EFA and Cronbach alpha tests. Correlation analysis was conducted to investigate the relationships among the variables. Subsequently, the proposed research model was evaluated using SEM, and the model fit was assessed by examining indices sensitive to sample size and model parsimony, specifically the incremental fit index (IFI), Tucker–Lewis Index (TLI), comparative fit index (CFI), and root mean square error of approximation (RMSEA), and standardized root mean square residual. Mediation effects were tested using bootstrapping with 2000 resamples and bias-corrected 95% CIs, which allowed for the estimation of indirect, direct, and total effects associations among key constructs. Moderation effects were tested via multigroup SEM by comparing path coefficients between the mild and severe disability groups (severity defined by self-reported category). Statistical significance of between-group differences was evaluated using critical ratios for parameter differences. Prior to comparisons, configural and metric measurement invariance were examined to ensure meaningful cross-group tests.

Questionnaires that were clearly invalid, either blank pages or patterned straight-lining, were excluded prior to analysis. For item-level missing responses, descriptive statistics, correlations, and EFA used pairwise deletion; scale scores were computed when ≥50% of items in a scale were present, otherwise treated as missing. For SEM in AMOS, listwise deletion was applied to ensure stable ML estimation, consistent with common AMOS practice. Univariate outliers were screened using standardized z scores (|z|>3.29) and boxplots, and multivariate outliers using Mahalanobis distance based on all observed indicators (threshold: chi-square with P<.001 for the relevant df). Potential influence was examined in auxiliary ordinary least squares regressions for the main structural paths (leverage and Cook’s distance; rule-of-thumb>4/n). To assess common method bias, we estimated (1) a common latent factor (CLF; variance fixed to 1; equal method loadings) and (2) a marker-based unmeasured latent method construct (ULMC) model. Neither approach yielded material improvement in fit (CLF vs baseline: ΔCFI=0.002, ΔRMSEA=0.001; ULMC vs baseline: ΔCFI=0.003, ΔRMSEA=0.000), and standardized loadings and paths changed by less than 0.20. Thus, CMB is unlikely to bias the substantive conclusions.

Data processing and statistical analyses were performed using SPSS 23.0 (IBM Corp) and AMOS 23.0 (IBM Corp) software.

Ethical Considerations

This study received ethics approval from the Korea University Institutional Review Board (KUIRB-2023-0286-01). Participants received information regarding the purpose, potential benefits, and risks of the study and were assured that all data would remain confidential. Each individual had the option to decline participation or withdraw from the study at any time. All participants signed an informed consent form. To protect anonymity and privacy, the data were encoded. To encourage participation, each respondent received a bathroom towel valued at US $5.

Results

General Characteristics

This study conducted a survey involving a total of 800 people with disabilities in Korea. The general characteristics of participants are summarized in . Regarding city size, 307 (38.4%) respondents resided in large metropolitan cities, while 493 (61.6%) were from small-to-medium-sized towns. The gender distribution indicated that males comprised a higher proportion (n=470, 58.8%) compared to females (n=330, 41.2%). The largest age group was 40‐49 years old (n=188, 23.5%), followed by the age groups of 50‐59 years and 60‐69 years, each at 20.3% (n=162). Regarding employment status, unemployed respondents (n=495, 61.9%) outnumbered those with employment (n=305, 38.1%). The educational background of the respondents showed that most had completed high school (n=383, 47.9%), followed by those with college degrees or higher (n=222, 27.8%). The duration since disability onset was highest for the group of 5 to less than 10 years (n=346, 43.3%), followed by less than 5 years (n=161, 20.1%). Disability severity showed a higher proportion of mild cases (n=432, 54%) compared to severe cases (n=368, 46%). Regarding marital status, unmarried respondents, including those who were widowed and divorced, accounted for a higher proportion (n=445, 55.6%) compared to married respondents (n=355, 44.4%). Institutions used by participants were hospitals (n=377, 30.3%), disability welfare centers (n=319, 25.7%), and local public health centers (n=278, 22.4%).

Correlation Matrix and Measurement Model

A correlation analysis of the measurement model was conducted to verify the causal relationships among key variables, and the presence of multicollinearity was evaluated by examining variance inflation factors (VIFs) and tolerance values. As shown in , the correlation matrix between the dependent variable, UI, and other study variables revealed statistically significant positive correlations with PU (r=0.780; P=.004), PEU (r=0.649; P=.005), HC (r=0.538; P=.005), IS (r=0.420; P=.002), EF (r=0.651; P=.005), CC (r=0.591; P=.003), HIC (r=0.616; P=.008), and eHL (r=0.323; P=.007).

Correlation values among variables ranged from 0.344 to 0.772, all below the threshold of 0.90, while VIF values ranged from 1.296 to 2.921, all below the critical threshold of 10. Additionally, no tolerance values fell below 0.1, confirming the absence of multicollinearity. Criteria indicating the absence of multicollinearity include correlation coefficients among variables below 0.90, VIF values below 10, and tolerance levels above 0.1. The results of the structural model fit assessment are as follows: the chi-square value was calculated to be χ²₆₇₂=2998.6, and model fit indices were IFI=0.929, TLI=0.922, CFI=0.929, and RMSEA=0.06. These values collectively indicate that the measurement model satisfactorily meets standard acceptance criteria, demonstrating good model fit.

Model refinement followed a prespecified, theory-first parsimony strategy. In AMOS, modification indices (MIs) were screened with MI≥10 and EPC≥0.10. Suggested residual covariances were freed only with a defensible common cause, otherwise left fixed. Indicators were considered for removal if λ<0.50 (implying <25% shared variance with the factor), |SR|>4, MI indicated cross-loading, or any Heywood case. Indicators producing a Heywood case, such as negative error variance or standardized loading greater than 1.0, were treated as inadmissible solutions and addressed through respecification; item parceling was not used. Alternative models were also estimated (the trimmed model dropping P>.10 paths, single-factor blocks, and higher-order variants). Model choice prioritized parsimony and information criteria alongside fit, with Bollen–Stine bootstrap (2000 resamples) to guard against overfitting. The final model retained a theory-congruent structure; MI-driven residual correlations without rationale were not adopted.

Hypothesis Testing

The results of hypothesis testing are presented in . First, standardized path coefficients to PEU from HC (β=0.233; P<.001), CC (β=0.163; P<.001), HIC (β=0.167; P<.001), IS (β=0.089; P=.005), and EF (β=0.276; P<.001) were statistically significant. Conversely, the path from eHL (β=0.025; P=0.406) to PEU was not statistically significant, leading to the rejection of this hypothesis. Second, standardized path coefficients to PU from CC (β=–0.121; P<.001), HIC (β=0.243; P<.001), eHL (β=–0.068; P=.003), and EF (β=0.368; P<.001) were statistically significant. In contrast, paths from HC (β=0.049; P=0.135) and IS (β=–0.009; P=.77) to PU were not statistically significant and thus rejected. Third, regarding hypotheses involving PEU, PU, and UI, the standardized path coefficient from PEU to PU (β=0.452; P<.001) was statistically significant. Additionally, the standardized path coefficients from PU (β=0.662; P<.001) and PEU (β=0.203; P<.001) to UI were statistically significant, thus supporting all related hypotheses.

The structural model was validated, and causal relationships among latent variables were examined using SEM (). HC had a significant positive effect on PEU (β=0.233; P<.001), which suggests that individuals with greater health concerns tend to find ePHRs more user-friendly. Additionally, CC has a notable impact on associations with PEU (β=0.163; P<.001), indicating that well-structured and easily understandable information enhances the user experience. Another important factor is the willingness to share personal health information (β=0.16; P<.001), which suggests that individuals who are open to sharing health data are more likely to find the system easier to accept. Although the standardized coefficient for IS awareness is relatively small (β=0.089; P=.005), it remains statistically significant. Among all predictors, the level of service support indicated the strongest associations with on PEU (β=0.276; P<.001). Users tend to find the system easier to navigate when comprehensive support and guidance are available. In contrast, eHL does not show a statistically significant relationship with PEU (β=0.025; P=0.406), leading to the rejection of the corresponding hypothesis.

‎

When examining the factors that influence PU, CC was found to have a statistically significant negative impact (β=−0.121; P<.001), which suggests that overly abundant or complex content may lead users to view the system as less useful. In contrast, agreement with health information usage (β=0.243; P<.001) and the level of service support (β=0.368; P<.001) demonstrated significant positive associations, which indicates that users who perceive greater assistance from the ePHR system are more likely to consider it useful. Interestingly, eHL also had a statistically significant negative association with PU (β=−0.068; P=.003). This implies that users who are more familiar with digital health information may adopt a more critical perspective toward the system. On the other hand, HC (β=0.049; P=0.135) and IS awareness (β=−0.009; P=0.772) did not significantly influence PU, leading to the rejection of these 2 hypotheses.

Regarding the hypotheses addressing PEU, PU, and UI, the analysis revealed a statistically significant standardized path coefficient from PEU to PU (β=0.452; P<.001). Furthermore, the standardized path coefficients from PU (β=0.662; P<.001) and PEU (β=0.203; P<.001) to UI were also statistically significant. These findings suggest that particularly for special populations, such as individuals with disabilities, environmental factors like usability, reliability, and social support are more crucial for actual technology adoption than informational literacy. Therefore, future models of technology acceptance should integrate these multidimensional environmental factors into a comprehensive approach.

Mediation Effect Analysis

Bootstrapping (2000 resamples; bias-corrected 95% CI) revealed the following mediation results (). First, HIC had significant positive direct effects on associations with PEU (β=0.167; P<.001) and PU (β=0.243; P<.001), and exerted a significant indirect effect on UI (total effect β=0.245; P<.001; ). Second, CC showed a significant positive association with PEU (β=0.163; P=.04) but a negative direct association with PU (β=−0.121, p = 0.345). When indirect effects were considered, the total effect on UI was negligible (β=0.002; P=.03). Third, IS exhibited a small positive association with PEU (β=0.089; P=.02) and, through PU, a limited indirect effect on UI (total effect β=0.039; P=.02). Fourth, EF had the strongest associations across all paths, with direct effects on PEU (β=0.276; P<.001) and PU (β=0.368, P<.001), yielding the largest total effect on intention to use when indirect effects were included (β=0.382; P<.001). Fifth, HC positively predicted PEU (β=0.233; P<.001) and contributed indirectly to intention to use via PU (total effect (β=0.150; P<.001). Finally, eHL did not show significant associations with PEU or PU and, if anything, trended negatively for PU (β=−0.068) and UI (β=−0.032).

Moderation Effect Analysis

Multigroup SEM (mild n=432; severe n=368) showed that HIC and EF were positively associated with PEU and PU in both groups (all P<.001; ). CC predicted PEU only in the mild group (β=0.201; P<.001), while IS predicted PEU only in the severe group (β=0.119; P=0.003). For PU, HIC and PEU were positively associated with both groups, whereas eHL was negatively associated with the mild group (β= −0.074; P=.006) and CC was also negatively associated with the mild group (β= −0.215; P<.001). UI was driven primarily by PU in both groups (mild β=0.727; severe β=0.511; both P<.001), with an additional contribution from PEU (severe β=0.272; mild β=0.171; P<.001).

Discussion

Determining Effect of PU and Ease of Use

In this study, PU emerged as the strongest predictor of the intention to use digital health technologies. The findings suggest that the adoption of digital health technologies among people with disabilities is more likely when there is a strong belief in the genuine benefits of these technologies for health management activities []. Furthermore, the study by Holden and Karsh [] also emphasized the applicability and robustness of the TAM within health care contexts. For instance, Harrison et al [] reported that 69.8% of patients with chronic kidney disease expressed a willingness to use digital technologies because perceived benefits include increased involvement in treatment and easier access to laboratory results. Similarly, Khor et al [] noted that PU significantly affects user attitudes toward e-PHR, positively impacting subsequent intentions to use these services. These findings suggest that the adoption of digital health technologies is driven not only by functional capabilities but also by perceptions of tangible, practical benefits, and that establishing trust in the effectiveness and practical value of digital health technologies is essential for this population.

PEU also significantly influenced users’ intention to adopt digital health technologies. This finding aligns with Pai and Huang [], who emphasized that minimizing technological complexity is crucial in forming UIs, particularly during the initial stages of implementing a hospital information system. Similarly, Tavares and Oliveira [] identified PEU as a direct determinant of adoption intentions in their study of electronic medical record portals. Additionally, a systematic literature review by Rahimi et al [] consistently confirmed that PEU significantly impacts the intention to use health informatics systems, acting as a key factor in reducing psychological resistance associated with technological complexity, especially during the early stages of system adoption. Akritidi et al [] further highlighted that intentions to use digital health care services are substantially influenced by PU, PEU, user satisfaction, privacy and security considerations, user age, and familiarity with electronic services.

Additionally, PEU was found to have a positive influence on PU. This finding suggests that as technology is perceived as intuitive and easy to use, the expectation that it will substantially benefit health management activities increases. These results align with empirical evidence provided by Yarbrough and Smith [], who demonstrated that medical professionals are more likely to perceive systems as applicable when those systems are uncomplicated to use. Similarly, Aggelidis and Chatzoglou [] empirically confirmed the significant influence of ease of use on PU in studies concerning hospital information systems and the acceptance of health IT among physicians. Chau and Hu [] further emphasized that PEU serves as a crucial mediating factor affecting PU and behavioral intentions in telemedicine and nursing information systems. A meta-analysis also highlighted this relationship as one of the most consistently validated paths across various TAM-based studies []. Consequently, ease of use emerges as an especially critical factor for user groups, such as people with disabilities, who often face substantial barriers to information access. Thus, when designing digital health technologies, strategies that prioritize user-friendly interfaces, such as clear visual displays, simple navigation, and step-by-step instructions, should be emphasized. For technologies targeting users with disabilities specifically, incorporating features, such as assistive device compatibility, voice-guided instructions, and adherence to accessibility standards is essential [-].

Factors Influencing Ease of Use

Factors significantly influencing the PEU included HC, CC, HIC, IS, and EF. Notably, the EF exerted the strongest influence, empirically demonstrating the crucial role of social support and facilitating conditions in the acceptance process of digital health technologies. According to Venkatesh et al [], social influence and facilitating conditions were key determinants in technology acceptance, with expectations and support from others significantly increasing an individual’s intention to adopt technology. Similarly, Heart and Kalderon [] emphasized that ongoing support from family members, professionals, or caregivers substantially facilitates technology acceptance among vulnerable populations, such as people with disabilities and older adults.

Furthermore, this study illustrated that the positive impact of HC on health management significantly impacts actual technology usage behavior, which aligns with findings by Or and Karsh [], who highlighted personal health interest and motivation as critical antecedents of eHealth technology acceptance, demonstrating that individuals with greater health motivation are more inclined to adopt such technologies readily. Additionally, research by Cocosila and Archer [] confirmed that individuals who perceive health improvement as a primary goal tend to accept and positively evaluate mobile health technologies more easily.

CC also significantly influenced PEU, indicating that the composition of content, presentation style, and degree of information structuring impact user intuitiveness and satisfaction with technology use. This finding aligns with Zhang and von Dran [], who demonstrated that users commonly assess the quality of web-based technologies based on the visual arrangement of information, ease of navigation, and the interconnected structure of content. Additionally, Liu and Shrum [] reported that how the presentation of this information affects cognitive load and user engagement, emphasizing that visual and interactive content can enhance ease of understanding compared to text-centric approaches. Crutzen et al [] similarly emphasized that visual design and information delivery methods in web-based health information systems substantially affect not only user usability but also learning outcomes.

It is also noteworthy that consent to share health information significantly influences the PEU. When users voluntarily consent to provide health information, it fosters trust in technology, which can potentially impact the overall PEU. This result aligns with the findings of Bansal et al [], who emphasized that individuals were comfortable using digital platforms to share sensitive health data only when the users trusted the platforms’ privacy protection mechanisms. Similarly, Li [] reported that users who depend on a system’s security features experience reduced psychological resistance, which in turn leads to more positive evaluations of the technology’s ease of use.

Finally, IS was also found to have a significant influence on PEU, which suggests that when personal data are perceived as securely protected, psychological anxiety about technology usage decreases, allowing for more intuitive system usage. These findings align with those of Bansal et al [], who emphasized that trust in privacy protection is a fundamental prerequisite for accepting technologies that involve sharing sensitive health information. Similarly, Angst and Agarwal [] showed that in the context of electronic medical record system adoption, building trust in data protection reduces resistance to technology and enhances intuitive and secure system usage. Klaver et al [] further reported that trust in security directly affects users’ psychological comfort and perceived ease of using mobile health technologies, highlighting that it is especially critical for older adults and vulnerable populations.

Factors Influencing PU

Factors identified as influencing PU included the EF, HIC, CC, and eHL. Most importantly, the EF had the strongest influence, suggesting that for users with physical disabilities, technology’s effectiveness is evaluated not merely based on its functional attributes but also significantly through social interactions and support experienced within the given environment. This result aligns with findings from Holden and Karsh [], who reported that users’ perceptions of health care technology acceptance are significantly influenced by external factors, such as facilitators or environmental conditions, rather than solely by individual judgments. Similarly, Chen and Chan [] demonstrated that support from family members or caregivers plays a crucial role in older adults’ recognition of the practical utility of technology.

Additionally, the finding that consent to share health information positively influences PU suggests that voluntarily agreeing to provide personal health data fosters trust in technology and enhances users’ sense of autonomy. These psychological factors, in turn, contribute significantly to the overall evaluation of the technology’s usefulness. This interpretation aligns with the findings of Bansal et al [], who reported that users’ perceptions of technology usefulness improve when trust in privacy protection and a sense of personal information control are established, particularly when dealing with sensitive health data. Likewise, Li [] noted that the voluntary provision of information enhances psychological comfort, positively influencing overall user evaluations of technology. Angst and Agarwal [] also empirically demonstrated that maintaining autonomy in information disclosure during the adoption of an electronic health record system critically influences user trust in the technology and subsequent PU.

The negative associations between content richness and eHL with PU may be consistent with cognitive-load and information-overload accounts. When information volume or complexity exceeds users’ processing capacity, it may become harder to extract actionable value, potentially lowering PU []. For higher-literacy users, expectation–disconfirmation processes may also contribute []. When these expectations are not met, evaluations of usefulness decline, even if users understand the content. This trend aligns with findings indicating that higher eHL can lead to more critical assessment and lower perceived utility, particularly when quality indicators are weak [-]. Conversely, several studies suggest a positive relationship between literacy and PU when the quality of content is demonstrably strong [,]. To address these issues, designers should avoid a one-size-fits-all approach to content richness and instead implement layered content and highlight quality indicators, such as sources, levels of evidence, and personalization logic, especially for users with high literacy. These strategies can help reconcile the negative associations observed in research with theoretical frameworks and suggest actionable steps to enhance PU across different literacy levels.

On the other hand, CC and eHL were each found to have a negative influence on PU. This suggests that users with higher information interpretation skills may react more sensitively to perceived qualitative limitations or technological shortcomings in content. Similar findings were reported by Chung and Nahm [], who indicated that users with higher eHL tend to critically assess the reliability, accuracy, and personalization of content, which can negatively affect the perceptions of its usefulness. Also, studies by Diviani et al [] and Neter and Brainin [] emphasized that higher literacy increases user expectations and criteria for information quality, potentially resulting in diminished perceptions of the technology if these expectations are not met. Conversely, research by Norman and Skinner [] and Mackert et al [] has demonstrated that increased literacy enhances information-seeking abilities, thereby improving the PU and acceptance of digital technologies.

Mediation and Moderation Effects of Technology Acceptance in People With Disabilities

Applying a TAM-based model, the study examined the determinants of digital health service acceptance among individuals with disabilities and analyzed the mediation effects as well as the moderation by disability severity. First, mediation analyses identified PEU as a central mediator within the model. The HIC significantly increased both PEU and PU and, via PU, exerted a positive indirect effect on the intention to use. This suggests that transparency and trust in the use of personal information are foundational factors for health care technology acceptance [,] and should be emphasized in services designed for individuals with disabilities. In addition, the EF showed the strongest explanatory power across all paths, confirming that social and professional support are decisive drivers of technology adoption—a finding consistent with prior reports underscoring the importance of support systems for vulnerable populations [,].

Second, not all predictors exerted uniformly positive effects. CC had a positive effect on PEU but showed a nonsignificant negative association with PU, resulting in a minimal total effect on intention to use. This pattern suggests that informational richness does not automatically translate into usefulness, indicating that groups with extensive digital experience may evaluate system quality more critically []. IS positively influenced the PEU yet had a limited direct effect on the intention to use, implying that security functions as a necessary but not sufficient condition for adoption []. eHL did not exhibit a positive association with either PEU or PU, instead trending negatively. A plausible interpretation is that higher-literacy users were more sensitive to simplicity or the lack of personalization in the service. This is consistent with Norman and Skinner’s [] eHealth Literacy framework, which suggests that elevated literacy prioritizes information quality and sophistication, potentially heightening dissatisfaction with comparatively simple systems.

Third, moderation analyses by disability severity revealed between-group differences among several key paths. HC and EF were consistently significant in both the mild and severe groups, whereas CC was significant only in the mild group, and IS was significant only in the severe group. In addition, eHL had a negative association with PU in the mild group. The above patterns suggest that individuals with milder disabilities, who generally possess higher digital capability, tend to evaluate systems more critically, while individuals with more severe disabilities place greater emphasis on accessibility and security []. In both groups, PEU emerged as the strongest determinant of intention to use, consistent with prior TAM evidence []. The association with PU was larger in the severe group, indicating that the practical benefits of technology carry greater weight in acceptance decisions among those with more severe disabilities [].

The finding that eHL had a negative association with PEU in the mild group can be interpreted in several ways. First, differences in evaluative standards may play a role. Individuals in the mild group typically have greater digital access and more extensive online information experience, which fosters more critical appraisal of system quality, accuracy, and convenience. In contrast, individuals in the severe group often face constrained digital options; even with higher literacy, accessibility and availability of external support may be prioritized over ease of use, attenuating any literacy–ease relationship []. Second, expectation–disconfirmation theory offers a complementary explanation. When a system is relatively simple or lacks personalization, unmet expectations may lead to depressed ease-of-use judgments. Conversely, for the severe group, basic accessibility itself is paramount, and unmet expectations may have been less salient in this domain [].

Limited Role of eHL

In this study, eHL did not significantly influence PEU, but it was negatively associated with PU. Similar results were reported by Chung and Nahm [], who indicated that among older adults, even high levels of eHL could not compensate for inadequate system accessibility and ease of use, consequently limiting technology acceptance. Moreover, van der Vaart et al [] emphasized that adequate technological infrastructure and ongoing support conditions must accompany information interpretation abilities to translate literacy into actual technology use. Czaja et al [] further demonstrated that environmental factors, tool accessibility, and social support are more critical determinants of technology acceptance than cognitive capabilities alone.

The aforementioned findings suggest that, particularly among groups such as people with disabilities, factors such as physical accessibility, ease of device use, and the presence of social support are more critical determinants of actual technology acceptance than literacy alone. Having said that, there is a vital need for integrated model designs that comprehensively incorporate environmental, social, and policy-related factors surrounding the adoption of technology [].

Limitations

While the study yielded promising results, several important limitations should be acknowledged. First, the study was conducted among individuals accessing specific institutions, such as rehabilitation hospitals, welfare centers for people with disabilities, and public health centers. This focus limits the generalizability of the findings to all people with disabilities, as the sample does not include those living in diverse environments. Future research should aim to include broader samples that consider various factors, such as residential settings, levels of community participation, and access to information resources. Second, the study used a cross-sectional design, which limits the ability to draw clear causal relationships between variables. Therefore, conducting longitudinal studies that track the same group of participants over time is recommended to clarify temporal causality among variables []. Third, data collection relied solely on self-report questionnaires. It could introduce social desirability bias, as participants may respond in ways they believe to be socially acceptable rather than reflecting actual behaviors. For the study, to address this issue, on-site paper surveys were conducted in private rooms, allowing unlimited time for responses. Emphasis was placed on anonymity and aggregate reporting, with neutral instructions provided. When assistance was needed, staff read the survey items verbatim and recorded responses without offering evaluative feedback. Any remaining bias is expected to elevate response levels rather than alter observed associations, indicating that the negative or null relationships noted are unlikely to be artifacts. Future studies should incorporate a brief social desirability scale or use indirect questions, enhance the use of private self-administered methods, and adjust models to account for the mode of administration. Fourth, this study analyzed people with disabilities as a pooled group, which improves statistical stability but limits granularity with respect to disability type and severity. Heterogeneity across these dimensions may shape both baseline levels and structural relations. Small subgroup counts and the absence of established measurement invariance precluded reliable multigroup estimation in the present data. Conducting such analyses without adequate power risks overfitting and spurious differences. Future research should use stratified sampling to ensure sufficient cases and consider hierarchical models to estimate between-group variance components []. Practically, intervention design should anticipate heterogeneity by offering layered content and adaptable interfaces and by segmenting onboarding/support according to severity. These steps will enable more actionable, subgroup-specific recommendations while maintaining psychometric rigor.

Practical Implications for Policy, System Design, and Clinical Implementation

Beyond the theoretical validation of TAM constructs, the present findings offer concrete implications for disability-specific digital health strategies. At the policy level, governments should establish national accessibility standards for e-PHR platforms, expand digital literacy programs for people with disabilities, and allocate funding to community-based support systems to reduce disparities. From a system design perspective, developers should incorporate accessibility features such as screen readers, voice navigation, and simplified interfaces tailored to cognitive or sensory impairments, ideally through participatory co-design with people with disabilities. Clinically, healthcare providers should integrate e-PHR services into rehabilitation and chronic care workflows, deliver clinician-guided onboarding sessions, and ensure interoperability with assistive devices to maximize usability and trust. To operationalize service support, it may help to establish assisted onboarding services within hospitals and welfare centers, integrate caregiver-managed accounts for those with cognitive or motor impairments, and simplify content presentation to accommodate users with low health or digital literacy. For individuals with severe disabilities, priority should be given to caregiver-managed accounts, voice-navigation features, and continuous assisted support. For those with mild disabilities, simplified user interfaces, standardized accessibility tools, and short-term digital literacy training may be sufficient.

Conclusion

This study examined the factors influencing the intention to use digital health technologies among people with disabilities through the lens of the TAM and analyzed the structural relationships among relevant variables. The findings revealed that both PU and PEU were significantly associated with the intention to use these technologies. Several external factors, including EF, HIC, and CC, also influenced these mediating variables. Notably, the EF demonstrated the strongest associations with both PEU and PU, emphasizing the crucial role of social support in the technology acceptance process for people with disabilities. Future research should explore longitudinal studies and incorporate mixed methods approaches to further validate these findings and gain deeper insights into long-term acceptance and real-world usage behaviors among diverse people with disabilities. Overall, this study provides valuable insights into developing digital health services and informs policy-making efforts specifically tailored for people with disabilities.

Introduction

Electronic cigarette (e-cigarette) use has been prevalent among youth and young adults in recent years. Using National Youth Tobacco Survey data (2013‐2022), the cross-sectional analysis found that the prevalence of current e-cigarette use among US youth rose from 3.10% in 2013 to a peak of 20.18% in 2019, then declined and remained relatively stable in 2021 and 2022 (7.50% and 9.44%, respectively) []. Although the prevalence of e-cigarette use among middle school and high school students has decreased from 7.7% in 2023 to 5.9% in 2024, about 1.63 million youth still reported e-cigarette use in 2024 []. Results from the 2020 National Youth Tobacco Survey (NYTS) data analyses showed that the e-cigarette product characteristics (eg, e-cigarette flavors, concealability, and vape tricks), family and peer influence (family or friend use), curiosity, and replacing cigarettes were the main reasons for e-cigarette use []. While e-cigarettes are often cited for their potential as smoking-cessation aids [], their long-term health effects remain uncertain [,], and therefore, there is broad consensus that youth access should be prevented [,]. Chemical analyses of e-cigarette aerosol have detected carcinogens like formaldehyde, acetaldehyde, acrolein, diacetyl, and toxic, carcinogenic metals such as chromium, nickel, and lead [-]. Human studies have shown that e-cigarette use was associated with several respiratory disorders (such as wheezing, asthma, and chronic obstructive pulmonary disease) [-] and mental health problems (such as depression) [,]. E-cigarette use costs the United States approximately US $15 billion annually in 2018 due to the US $2024 excess health care expenditures per e-cigarette user per year compared to never tobacco product users [].

With over 1 billion monthly active users, TikTok is one of the most popular social media platforms among youth and young adults in the United States []. TikTok’s short video format, engaging content, and personalized recommendation system have attracted millions of youths and young adult users. The TikTok platform allows users to create, view, and share videos with posts leading to likes, shares, comments, and followers []. Recognizing the popularity of social media, the vaping industry and vaping proponents have prioritized social media to promote vaping. E-cigarette companies and vape shops promote e-cigarettes on TikTok by posting professionally designed videos using popular hashtags [], creating fake user accounts to disseminate spam and favorable views [,], and providing celebrity sponsorship []. Although TikTok’s Community Guidelines prohibit content that facilitates the purchase, sale, trade, or solicitation of e-cigarettes, e-cigarette promotional posts remain prevalent on the platform [].

In contrast to vaping promotion messages on social media, the number of social media posts educating or warning the public about the adverse health effects of vaping is minimal on all social media platforms [,]. To reduce the prevalence of vaping among youth, in September 2018, the US Food and Drug Administration (FDA) launched a vaping prevention campaign – “The Real Cost” on social media and other platforms to educate youth about the adverse health effects and risks of vaping []. However, the impact of the FDA’s “The Real Cost” vaping prevention campaign on Instagram (Meta) is limited partially due to low engagement []. For example, the FDA-sponsored TheRealCost account has a much lower median number of likes per Instagram post than the vaping promotion accounts []. Therefore, it becomes essential to understand and (especially) design vaping prevention messages with high user engagement for effective health communication in the community.

High user engagement, such as a high number of likes, might significantly impact user behavior []. Improving social media user engagement of vaping prevention messages can prevent vaping initiation and motivate cessation []. Vaping promotion posts have high engagement due to well-designed content and popular hashtags [,]. Some features (such as hashtags) in e-cigarette promotion posts with high user engagement could be integrated into the vaping prevention messages to increase user engagement. In addition, identifying the features of vaping promotion social media messages targeting vulnerable populations (such as youth) will facilitate future regulation of such social media posts. Thus, our study will include both vaping promotion and prevention posts. Current social media studies on e-cigarette-related TikTok videos focus on characterizing content rather than identifying high user engagement post features [,]. More recently, several studies have started to describe the features of social media posts associated with high user engagement [,]. Yet, none of the previous studies identified video features associated with high user engagement on TikTok. Video is popular on social media platforms and is more attractive than images and texts due to its combination of visuals, sounds, and motions to create an emotional connection with viewers through a storytelling mode, including facial expressions and different tones of voice []. Thus, identifying video features associated with high user engagement is important to attract TikTok users’ attention to vaping prevention and cessation content. This study aims to fill this knowledge gap by identifying video features from e-cigarette-related TikTok videos associated with high user engagement. Findings from this study are valuable in designing effective vaping prevention video messages for tobacco control and vaping cessation interventions.

Methods

Study Design

Our study is a cross-sectional TikTok post content analysis to identify important features associated with high user engagement in TikTok posts. All data collected and analyzed in our study are publicly available from the TikTok platform, and all results were reported in aggregate form to ensure user anonymity.

Ethical Considerations

Our study has been reviewed by the University of Rochester Research Subjects Review Board (RSRB) office and was determined to meet federal and university criteria for exemption under Exempt Category 4 (Study ID: STUDY00009334). The RSRB determined that this study qualifies as secondary research involving existing data and, therefore, does not require informed consent. All publicly available TikTok videos included in the analysis were deidentified before use, and no compensation was provided to the TikTok users. Our study report followed the Strengthening and Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines [].

Data Collection

TikTok short videos (less than one minute) in English related to e-cigarettes from the United States were searched and downloaded using the TikTok research application programming interface (API) using keywords related to e-cigarettes, including “e-cig,” “vaping,” “e-cigarettes,” “e-cigs,” “ecig,” “ecigs,” “electroniccigarette,” “ecigarette,” “ecigarettes,” “vape,” “vapers,” “vapes,” “e-liquid,” “ejuice,” “eliquid,” “e-juice,” “vapercon,” “vapeon,” “vapefam,” “vapenation,” “juul.” []. The TikTok API has a limitation of around 100 on the number of videos we can collect each month. In February 2024, through the TikTok API, a total of 1487 TikTok videos (from 1064 unique creators) related to e-cigarettes posted between January 2023 and January 2024 were collected and confirmed manually in this study.

Video Feature Extraction

The TikTok video feature list was derived primarily from our previous work—namely, our characterization of Instagram image features for e-cigarette content and our manual coding of e-cigarette videos on TikTok and YouTube [-,,]. We then reviewed sample TikTok videos to identify additional features that could affect engagement and, as a team, agreed on the final set of features for this study. The TikTok video features analyzed in this study included promotional content (including advertisements or promoting the e-cigarette use), celebrity endorsements, background setting, perceived sex (male or female, based on the observation), social events, young adult themes, lifestyle portrayals, e-cigarette devices (the presence of physical devices), smoking or vaping behaviors (the action of smoking or vaping regardless of the presence of physical e-cigarette devices), talking, singing, dancing, humorous or silly content, cartoons or animations, vaping tricks, and the use of emojis. We considered 2 cutting-edge large language models to extract video features from TikTok videos: GPT-4 [] and Video Large Language Model Meta AI (Video-LLaMA-7B) []. Both models can understand the visual and auditory content in the video. Based on a reported AI labeling accuracy of 94% [] and a prespecified 10% margin of error for the 95% CI, the single-proportion sample-size calculation indicated that 22 videos are required for our pilot evaluation of GPT-4 and Video-LLaMA-7B feature labeling. To examine which large language model performs better, we randomly selected 25 videos from the 1487 videos and manually labeled the proposed features by 2 human coders. Two human coders worked together side-by-side to watch these videos and label the defined video features. GPT-4 uses vision models to break down the visual component of a video frame by frame and detect objects, faces, text in the image, and other visual elements to help us identify features from the TikTok videos. In this study, to reduce data size and the demand for computational resources while still capturing meaningful information in the video, we adopted a frame sampling strategy, one image frame for every 15 frames, to represent the video []. In this study, the OpenAI API was used to interact with the GPT-4 model, and well-designed prompts were used to generate the desired and optimal output from the model. Table S1 in lists prompts used to automatically extract video features from the remaining 1462 TikTok videos using the GPT-4 API.

We used the GPT-4 API to categorize each video’s background into one of the following 6 groups based on the description of the background, including car, public space (club, studio, indoor and outdoor, medical building, restaurant, spacecraft, and other), private space (bathroom, bedroom, home, kitchen, room, and office), outside (beach, forest, garden, natural, space, campus, cave, urban, and natural), shop (shop and vape shop), and not detectable. Based on the description of the lifestyles from GPT-4, we classified the lifestyles into casual, leisure (fitness, gaming, and leisure), and no lifestyle (educational, health, industrial, vaping, working, or others) categories.

Statistical Analysis

A linear mixed effects model with random intercept to account for multiple videos posted by the same user was used to model the association of TikTok user engagement with extracted features. We calculated the Spearman correlations among likes, shares, comments, and views of TikTok videos and found significant moderate-to-strong correlations across these metrics, as expected (Figure S1 in ). Given this interdependence, a more appropriate approach is to use overall engagement as the outcome variable, combining likes, shares, comments, and views using the formula applied by TikTok. The TikTok user engagement measure was calculated based on the formula [(number of likes+number of comments+number of shares)/(number of views+0.5)] X 100 []. For each video, add up all user actions (likes, comments, and shares) and then divide that total by the number of views. This gives interactions per view, so it reflects how actively viewers responded to the content, independent of how many people saw it. The value of 0.5 is added to the formula to minimize the generation of infinity values when the number of views is 0. Wilcoxon rank sum tests were used to compare the TikTok user engagement on a log scale between different feature categories. A linear mixed effects model was used to identify significant features associated with TikTok user engagement. The purposeful variable selection method was used to select important features for the final model, and Tukey’s method was used to control the familywise error rate for pairwise comparisons. Statistical analysis software R 4.3.1 (R Core Team, 2007) was used for the data analysis, with a significance level of 5% for all the analyses.

Results

Feature Extraction From TikTok Videos Using Large Language Models

From the 1487 e-cigarette-related videos, we randomly selected 25 TikTok videos for hand-coding video features. We compared the accuracy in extracting features between 2 popular large language models, GPT-4 and Video-LLaMA-7B, using the hand-coding video feature as the golden standard. As shown in Table S2 in , for the video features we selected, the large language model GPT-4 showed much higher accuracy (83% to 100%) than the model Video-LLaMA-7B (24% to 88%). Therefore, we used GPT-4 to extract features from the remaining 1462 videos.

User Engagement for Video Features

Using the Wilcoxon sum rank test, we compared the distribution of TikTok user engagement measures on a log scale between different categories of those identified video features ().

As shown in , TikTok videos with a background in a car versus other background categories (5.48 for car vs 3.06‐4.65 for other background categories), featuring females versus males or females and males or no people (4.98 for females vs 3.63‐4.47 for males or females and males or no people), advertising social events (4.60 vs 4.13), including young adults (4.69 vs 3.92), showing casual lifestyles (4.70 vs 4.03-4.36), without the presence of physical e-cigarette devices (4.72 vs 3.93), having the action of people smoking or vaping (4.61 vs 4.35), including people talking (4.88 vs 3.80) or singing (5.31 vs 4.39), have significantly higher user engagement than TikTok videos without those features (all P<.05). TikTok videos without promotion content had substantially higher user engagement than TikTok videos with promotion content (4.66 vs 3.02). In addition, TikTok videos that are funny or silly (5.09 vs 4.16) or contain emojis (4.90 vs 4.26) had significantly higher user engagement than other TikTok videos (P<.001). TikTok videos with features like celebrity endorsement, people dancing, cartoons or animations, and vape tricks were not significantly different from TikTok videos without those features.

TikTok Video Features Associated With High Social Media User Engagement

In the multivariate linear mixed effects model, several video features, such as lifestyle, dancing, and singing, were not significantly associated with TikTok user engagement. According to the purposeful model selection method, these were not included in the final linear mixed effects model. Results from the linear mixed effects showed significant features associated with the TikTok user engagement after adjusting for other variables in the model (). As the exponential of the parameter estimates were the rate ratios (RR) from the linear mixed effects model, an estimated RR of 1 or higher was classified as high engagement, and an estimated RR smaller than 1 was classified as low engagement. TikTok videos featuring young people had significantly higher user engagement than those without (estimated RR=1.24, 95% CI 1.00‐1.53; P=.048), corresponding to 24% greater engagement. TikTok videos with people talking had significantly higher user engagement (RR=1.63, 95% CI 1.30‐2.05; P<.001), meaning user engagement is 1.63 times that of videos without people talking. TikTok videos that are funny or silly had significantly higher user engagement than TikTok videos that are not funny or silly (RR=1.61, 95% CI 1.29‐2.00; P<.001). TikTok videos that contain emojis had significantly higher user engagement than TikTok videos without emojis (RR=1.88, 95% CI: 1.48‐2.38; P<.001). There is a marginally significant association between TikTok videos with vaping tricks and TikTok user engagement (RR=1.62, 95% CI 0.97‐2.71; P=.07). On the contrary, TikTok videos containing promotional content had significantly lower user engagement than TikTok videos without promotional content (RR=0.60, 95% CI 0.45‐0.81; P=.001).

‎

The background setting showed a significant association with engagement. Accordingly, we estimated all pairwise contrasts between video backgrounds using Tukey’s method. Pairwise comparisons showed significant differences in TikTok user engagement among different video backgrounds (). TikTok videos with a background in the car had significantly higher user engagement than TikTok videos with a public space background (RR=3.91, 95% CI 1.25‐12.20; P=.009). Other video background comparisons showed no significant differences, as their 95% CIs included 1.

‎

Discussions

Principal Findings

Our study is the first to investigate video features in e-cigarette-related TikTok videos associated with high TikTok user engagement. Using artificial intelligence (large language models such as GPT-4) and generalized linear models, we extracted and identified e-cigarette-related TikTok video features associated with high user engagement, such as background settings, young adult presence, funny or silly content, and emojis. Some of these identified features (such as background setting, humorous content, and proper emojis) could be used to design future e-cigarette prevention and cessation videos to increase user engagement.

In this study, as a proof of concept, we have tried to extract a set of features from e-cigarette-related TikTok videos. Some are general video features like the background, young people, and fun. In contrast, others are more specific to e-cigarette-related videos, such as smoking or vaping and vaping tricks. While Video-LLaMA is designed to analyze videos, our results showed that GPT-4 performed better than Video-LLaMA in identifying these video features. One potential reason for this is that GPT-4 is a multimodal model trained on a much more extensive training dataset. In addition, while Video-LLaMA might be more potent in understanding video motion and temporal dynamics, GPT-4 might perform better in semantic feature extraction at the frame level, especially from still frames in videos, by leveraging cross-modal information.

The pairwise comparison results within the linear mixed effects model framework showed the differences of various video background settings in their associations with TikTok user engagement. The reduced number of significant findings for the car background in the pairwise comparisons resulted from the stricter Type I error control applied via Tukey’s method. TikTok videos with backgrounds in a car had significantly higher user engagement than TikTok videos with backgrounds in public spaces. These results showed TikTok users prefer background settings in a more closed or private space than public spaces for e-cigarette-related TikTok videos. The reason for this might be related to the vaping ban in public places, including all bars and restaurants, which was recommended by the World Health Organization (WHO) in August 2016 and implemented in most United States states, including Alabama, California, and New York []. The widely adopted vaping ban in public places in the United States might lead people to view vaping in public spaces, such as bars, restaurants, or medical buildings, as inappropriate. Thus, e-cigarette-related TikTok videos in public places had less user engagement than TikTok videos with car backgrounds.

E-cigarette-related TikTok videos with young adults had higher user engagement. This might be due to the significant population of e-cigarette users being youth and young adults, and most of the TikTok users being young people. However, TikTok videos that showed only the physical e-cigarette device were less attractive to users—likely because they lacked human cues, offered little narrative value, and looked like advertisements. TikTok videos that are funny or silly attract more attention. Humor seems to be an effective tool to connect with audiences on TikTok by capturing users’ interest, encouraging sharing, and fostering a sense of community. Humor could also help increase the loyalty of followers []. E-cigarette-related TikTok videos featuring vaping tricks appear very impressive to viewers [], sparking curiosity among many users about how those tricks are performed []. Our study found a positive association between vaping tricks and user engagement that was marginally significant. While vaping tricks are commonly used to promote vaping on social media, they should be appropriately regulated or controlled to reduce their influence on susceptible social media users. Emoji is another powerful tool that attracts TikTok users’ attention. In TikTok videos, emojis could serve as visual cues to express feelings, reactions, and attitudes to effectively communicate with viewers on an emotional level and attract their engagement [,].

Current policy is insufficient in preventing the spread of vaping promotion videos on TikTok []. Identifying video features linked to higher engagement in e-cigarette–related TikTok posts has clear implications for policy and practice. For public health communication, prevention content can ethically adopt engaging formats to improve reach—eg, featuring young adults, using light emoji overlays and humorous content, and filming in relatable settings such as cars—while avoiding elements that glamorize product use.

Limitations

Our current study has several limitations. First, the e-cigarette-related TikTok video features included in this study are only a subset of video features. Other video features, such as the audio features, should be further investigated in future studies. Our study is the first one to investigate e-cigarette-related TikTok video features associated with social media user engagement. Here, we only tested some features that could be identified using current artificial intelligence techniques; for example, the current GPT model could not detect the background settings from some of the TikTok videos. With the rapid development of artificial intelligence techniques, we believe more features could be accurately labeled by the GPT models or other more advanced models, which can be included in our future studies. Second, our list of TikTok video features may not be applicable to other social media platforms. Therefore, it is crucial for future studies to explore important features on different social media platforms, such as Instagram and YouTube. Third, the accuracy of some feature labeling by the current GPT models has not reached 100% yet. There might be mislabeling of some features in our data, which might lead to some bias in our results. The accuracy of video features could be further improved with the availability of more advanced large language or video models. In addition, because sex was inferred from perceived physical characteristics and presentation in the videos rather than self-reported identity, the measure reflects perceived sex and is subject to misclassification bias, with potential exclusion or misrepresentation of transgender, non-binary, and gender-diverse individuals. Fourth, due to the unavailability of data, we cannot determine the demographics of TikTok users who were engaged with these e-cigarette-related videos, which should be investigated using other approaches. Moreover, we do not have the information about account-level factors—such as the number of unique accounts reacting, whether an account is personal versus sponsored or brand, and whether a post was boosted or advertised—that may also influence engagement. Fifth, not all features identified in this study might be appropriate for vaping prevention videos, which should be further tested and validated in future studies. Sixth, the data collection was conducted in February 2024 and included TikTok videos posted from January 2023 to January 2024. The numbers of views, likes, shares, and comments may change over time. Although the moderate-to-strong correlations across views, likes, shares, and comments imply that the engagement ratio (likes+shares+comments)/views should remain broadly consistent, potential bias could be introduced with the changes in the engagement metrics over time. Seventh, because the TikTok API limits monthly retrievals, our 1-year sample included 1487 videos. Larger samples in future work would increase power to detect additional features associated with user engagement. Eighth, 2 coders annotated video features side-by-side rather than independently; consequently, we could not quantify inter-rater reliability (eg, Cohen Kappa). This consensus approach may inflate agreement and introduce shared-observer bias. Finally, with the rapid changes in e-cigarette marketing and social media, it is necessary to closely monitor and update new developments of social media posts with high user engagement.

Conclusions

In this study, we showed that specific TikTok video features, including background settings (eg, car), talking, funny or silly, and use of emojis, were significantly associated with higher TikTok user engagement. These findings offer practical guidance for designing more engaging vaping-prevention videos for a broader reach of social media users.

Earlier this month at the 2025 China International Import Expo (CIIE) in Shanghai, Illumina introduced a suite of new technologies and signed several agreements to partner and manufacture locally, demonstrating its commitment to the country’s “In China, For China” strategy. 

Over 922,000 people attended the annual event, where life science Illumina showcased their products and hosted panel discussions. This year marked Illumina’s sixth time participating in CIIE, and the company’s theme for the 2025 expo was “Innovating for Twenty Years, Sequencing for the Future.”

Ever since it began doing business in China in 2005, Illumina has steadily invested in the country. From introducing groundbreaking products to codeveloping research and clinical applications, Illumina has driven local innovation. Today, over 70%* of China’s clinically approved tumor NGS in vitro diagnostic applications are built on Illumina platforms.

“Illumina’s commitment to China spans two decades, and our vision is to advance genomics to transform health care for patients everywhere,” says Illumina CEO Jacob Thaysen, who met with the Chinese vice minister of commerce after the expo. “China is not only a strategic market but a vital contributor to global progress in life sciences. We appreciate the opportunity to engage in constructive discussions with the Chinese government. CIIE has become an important platform for introducing innovation and fostering collaboration. This year, we are proud to debut our latest multiomics solutions in China, empowering researchers and clinicians to accelerate discovery and drive genomic breakthroughs. Together, we aim to unlock the full potential of precision medicine and deliver meaningful impact for patients worldwide.”

Illumina’s Shanghai manufacturing site, established in 2022, began delivering locally produced sequencing systems and reagents to domestic customers in 2023. At CIIE, Illumina signed an agreement with a Shanghai development company to expand the capacity and scale of this manufacturing site. Illumina also signed strategic supply chain partnerships. These collaborations will drive greater integration and increase local production capacity.

“Localization has always been central to Illumina’s mission to better serve customers in China,” said Jenny Zheng, Illumina’s head of region for Greater China, during the signing session. “Today’s agreements mark another milestone in strengthening supply chain resilience and deepening the ‘In China, For China’ industrial landscape. We will continue accelerating full localization of products and solutions—enhancing manufacturing, quality, and compliance—to meet local needs. By leveraging our NGS and multiomics portfolio and working closely with partners, we aim to drive original innovation and advance precision medicine and biopharmaceutical development in China.” 

During CIIE, two exciting new multiomic technologies made their China debut: First, the Illumina Protein Prep solution gives scientists a deeper understanding of proteomics and provides multidimensional insights across cancer and cardio metabolic and immunologic diseases. And second, with the 5-base solution, researchers can simultaneously detect genomic variants and DNA methylation in a single library preparation, sequencing, and analysis run, facilitating the discovery of novel biomarkers and advancing precision medicine.

The Illumina booth also showcased the locally manufactured MiSeq i100 Plus-CN (prototype) and NextSeq 2000-CN, and highlighted its reagent portfolio across oncology, single-cell sequencing, proteomics, infectious disease, and microbiology. Illumina also introduced products codeveloped with local partner Berry Genomics, including the NextSeq CN500, which supports applications in reproductive health, genetic disease testing, and scientific services, and the NovaSeq 6000Dx-CN-BG. This platform was just approved by China’s National Medical Products Administration (NMPA) in August and was introduced at CIIE.

To foster in-country relationships and collaborations, the company hosted several sessions and roundtable discussions on the future of precision medicine, multiomics data standardization, methylation-based clinical diagnosis, disease mechanism analysis, and more. Participants shared their insights to push research and partnerships and help build a more open, collaborative innovation ecosystem.

*As of August 2025, based on NMPA approval data.

 

Paramaribo: As Suriname celebrates 50 years of independence, it finds itself at a critical juncture. In recent years, it had commendably restored macroeconomic stability and significantly improved its institutional frameworks for macroeconomic policymaking. At the eve of a significant oil boom, the authorities’ task is to act now to lay the groundwork and build the institutions needed to fully harness the country’s newly found oil wealth. Doing so successfully will ensure these precious resources are used efficiently and productively to materially improve people’s livelihoods. As these resources are being developed in the coming years, it will be essential to maintain a prudent fiscal-monetary policy mix, improve governance, and strengthen institutional capacity. The new government, which took office in July 2025, recognizes that such a reform package is necessary to improve the country’s health, education, safety, infrastructure as well as diversification (for example through tourism and agriculture), entrepreneurship, and growth potential.

Growth has been decent and is expected to continue at around 2-3 percent in the next few years. During the course of this year, gold production has been disappointing but, going forward, economic activity is expected to be increasingly supported by the development of the Block 58 oil project. The field development is, though, import intensive, and a large current account deficit is expected in 2026-28, financed by FDI inflows. Foreign exchange reserves coverage remains adequate as insurance against external shocks. Block 58 oil is expected to start in 2028 leading to a doubling of real GDP by 2030.

Macroeconomic stability is being eroded. After primary surpluses in 2022-2024, the fiscal position has worsened and is expected to record a primary deficit (on a cash basis and excluding a necessary central bank recapitalization) of about 1 percent of GDP in 2025 but with a sizable increase in suppliers’ arrears. This pre-election fiscal expansion has caused a significant reduction in the government’s cash balances and the resulting injection of liquidity has put pressure on the exchange rate. These factors and the fiscal boost to demand have increased inflation (from around 6 percent earlier in the year to over 10 percent). Furthermore, monetary aggregates have been allowed to grow faster than the central bank’s reserve money targets since late 2024 and the central bank has been intervening to moderate the currency depreciation.

The authorities conducted a successful liability management operation. The transaction was centered around the issuance of US$ 1.575 billion in 5- and 10-year Eurobonds. The proceeds financed a cash tender offer for some existing 2033 Eurobonds and the remainder are being held in an overseas escrow account to be used to buy back outstanding 2033 Eurobonds and some or all of the oil-linked value recovery instruments. These resources could also be used to prepay bilateral debt and will finance some interest payments on the new Eurobonds. The operation shores up the financing needed to service debt until after Block 58 oil revenues begin to flow in.

There is an urgent need to improve the fiscal balance in 2026-7. Staff projects a primary balance of around 0 percent of GDP in 2026. A larger and more credible consolidation, underpinned by clear policy measures, would reduce depreciation and inflationary pressures and help the central bank to meet its monetary goals. In turn, this would preserve purchasing power and help businesses operate. Such improvements would also create buffers against future downside risks (for example, a 25 percent decline in gold prices, which could reduce fiscal revenues by 2 percent of GDP).

The government’s fiscal plan should be consistent with the recently legislated fiscal frameworks. A five-year fiscal plan should be submitted to the National Assembly, alongside the 2026 budget, with both annual spending ceilings and a target for debt (net of Savings and Stabilization Fund assets), this year. While there are pressing spending needs in education, health, roads, electricity, and water and sanitation, spending limits should be raised only gradually to allow for an improvement in the government’s capacity to effectively execute such spending. Suriname should strengthen its public investment management practices and implement its Public Financial Management Priority Action Plan. The Savings and Stabilization Fund Suriname needs to be operationalized.

Electricity subsidies should be removed to provide the resources needed to fund social assistance and growth-enhancing investments. In particular, the automatic link between tariffs and the costs of electricity production established in 2024 should be restored and electricity prices should continue rising towards cost-recovery. Even though social assistance outlays have doubled over the past few years there are significant leakages. To address this, the authorities are reviewing the existing social programs to free up resources to expand coverage and raise the adequacy of benefits. Moreover, consideration could be given to using the resources freed by the debt operation to reduce the stock of supplier arrears.

Revenue administration needs strengthening and there is scope to raise excise taxes. The authorities’ plan to transition to a Semi-Autonomous Revenue Authority will help improve revenue collection. The high international price of gold may have increased smuggling and there is scope to step up enforcement to ensure small-scale gold miners fully pay their tax obligations. Excise taxes are low by international standards and should be increased and applied to a broader range of products.

There is an urgent need to strengthen transparency and anticorruption controls ahead of the surge in hydrocarbon revenues. The new procurement law should be implemented immediately. It requires the publication of all tenders, procurement contracts, names of the awarded entities and their beneficial owners, and the names of the public officials awarding the contracts. It also requires ex-post validation of the delivery of the contracted service. The amendment to the anti-corruption law—to mandate the declaration of income and assets of politically exposed persons, to require verification and publication of these declarations, and to establish dissuasive sanctions for non-compliance—should be passed by the parliament and then promptly implemented.

State-owned enterprises need to be subject to stronger safeguards. The financial operations of SOEs are not transparent and there is an urgent need to establish a timely collection of the data necessary to assess the financial performance of these enterprises. Non-performing enterprises should be either closed or sold and, for the remainder, there should be a broader roadmap to improve service delivery and safeguard public resources.

Monetary and exchange rate policy needs to refocus on reserve money targets to preserve price stability. The central bank should establish clear reserve money targets for the coming year and should endeavor to meet these goals by undertaking open market operations, regardless of the interest costs of sterilization. The central bank should eliminate interest rates ceilings to allow rates to be determined by the market to meet targets and improve monetary transmission. Transmission will also be aided by the ongoing phase out of the issuance of central bank certificates. Foreign exchange intervention should be used only in response to disorderly market conditions, which should be more narrowly defined. Foreign exchange regulations including the role played by the Foreign Exchange Commission should also be reviewed.

Efforts to improve central bank operations and institutional capacity are welcome. A monetary policy committee should be formed to institutionalize monetary policy decision making. The central bank should publish a Monetary Policy Statement following policy decisions and a quarterly Monetary Policy Report to increase public understanding of their actions. The analytical framework should work to better integrate data (including the planned extension of existing surveys of expectations of inflation and other key macroeconomic variables) and improve forecasting. There is scope to improve high-level coordination between the central bank and the Ministry of Finance and Planning. There should also be a clear strategy to transition to an interest-based framework including through the introduction of a central bank deposit facility and the development of an interbank market.

Internal risk management practices of the banks need to be strengthened. Credit growth has been fast and there is a need to more closely monitor banks’ internal risk management systems to ensure their underwriting activities are prudent, limits on net open FX positions are respected, and banks are accurately classifying loans. A comprehensive credit registry would help track the financial position of borrowers and reduce data gaps. The framework for bank resolution should be quickly operationalized and contingency plans should be developed to better react to downside scenarios. The central bank should be ready to provide liquidity to banks but undercapitalized banks that lack viable financial plans should be quickly resolved.

The authorities can help improve the business environment. Institutional reforms to manage the oil boom are a critical precursor to addressing developmental challenges. However, a key constraint identified by exporters and investors is government and regulatory inefficiency, especially bureaucratic delays. More generally, international experience suggests structural reforms, such as improvements in human and physical capital and the regulatory environment, bring larger benefits than industrial policies such as special economic zones.

The IMF team is grateful to the Surinamese authorities and other counterparts for the productive discussions and hospitality during the mission.

 

Table 1. Suriname: Selected Economic Indicators
			Proj.
		2024	2025	2026
(Annual percentage change, unless otherwise indicated)
Real sector
Real GDP		1.7	1.3	3.9
o/w Non-Natural Resource Real GDP		5.4	5.4	5.0
Nominal GDP		14.8	18.1	15.8
Consumer prices (end of period)		10.1	12.1	8.6
Consumer prices (period average)		16.2	9.3	11.0
Money and credit
Broad money		9.3	13.9	11.6
Private sector credit		16.0	28.9	10.6
Reserve money		10.1	21.2	14.0
(In percent of GDP, unless otherwise indicated)
Central government
Revenue and grants		26.9	28.1	27.4
Of which: Mineral revenue		10.9	10.5	10.5
Total expenditure 1/		29.3	38.2	32.9
Of which: central bank recapitalization			5.4
Overall Balance  (Net lending/borrowing)		-2.4	-10.2	-5.6
Primary Balance 1/		0.3	-6.4	-0.2
Primary Balance (excl central bank recap)		0.3	-1.1	-0.2
Deposits at Central Bank		9.2	1.6	1.2
Central government debt		88.0	105.5	98.4
Domestic		14.4	16.4	16.9
External		73.6	89.1	81.5
External sector
Current account balance		0.2	-35.3	-51.4
Capital and financial account		9.3	-30.2	-52.9
Memorandum Items
Gross international reserves (US$ millions) 2/		1,373	1,208	1,279
In months of imports		6.4	3.3	2.8
Escrow Account (US$ millions)			851.9	680.9
Exchange rate (SRD per USD, period average)		33.05	…	…
Sources: Suriname authorities; and IMF staff estimates and projections.
1/ Expenditure includes central bank recapitalization of 9,381 Million SRD.
2/ Excludes banks’ ring-fenced reserves.

 

 

ROME — ROME (AP) — Italian prosecutors have placed luxury group Tod’s and three of its executives under investigation for suspected labor abuses and exploitation, judicial documents showed on Thursday.

According to the documents, obtained by The Associated Press, Milan prosecutor Paolo Storari has also requested a six-month ban on the company’s advertising, with a hearing on the case set for Dec. 3.

In the documents, prosecutors allege that Tod’s — known for its high-end loafers and bags — was fully aware of and complicit in labor exploitation of Chinese workers at subcontracted workshops in Milan and the Marche region.

In a statement issued on Thursday evening, Tod’s denied any wrongdoing and said it will respond to the allegations in the appropriate courts.

In the documents, Storari noted a sort of “intentional blindness” from Tod’s, which had also carried out third-party audits on the workshops, but failed to address the problems they had revealed.

According to prosecutors, the workers’ exploitative conditions included hours exceeding the legal limit, inadequate wages, violations of various workplace safety regulations and degrading housing.

The probe focusing on Tod’s is the latest in a string of Italian police operations and investigations revealing the abusive treatment of subcontracted workers by high-end brands.

In April, Italian police disclosed that Chinese workers employed by an unauthorized subcontractor, produced handbags and accessories for Giorgio Armani.