The US Federal Reserve requires “strong, sound and steady leadership”, according to Donald Trump. The president found a man to lead the central bank who would “provide exactly that type of leadership”, he declared.“He’s strong, he’s committed and he’s smart.”
This is not how Trump described Kevin Warsh, the former Fed governor whom he unveiled as his new nominee to chair the central bank on Friday – but how he hailed Jerome Powell, the current Fed chair, when nominating him for the job about eight years ago.
“If he’s confirmed by the Senate, Jay will put his considerable talents and experience to work leading our nation’s independent central bank,” Trump said as the November sun shone down on him in the White House Rose Garden in 2017.
Trump has long stopped praising Powell. After refusing to bow to Trump’s demands for drastic interest rate cuts, he is now “a moron”, “stupid”, and “incompetent”, according to the president. “There’s something wrong with him,” Trump claimed in November. “I’ll be honest, I’d love to fire his ass.”
Donald Trump and Jerome Powell at the Federal Reserve building in Washington DC on 24 July 2025. Photograph: Kent Nishimura/Reuters
Powell, whose second term as Fed chair is due to expire in May, has refused to give Trump the loyalty he requires of every other leader in the executive branch. And while the president has been able to fire the vast majority of officials who he sees as disloyal, the Fed has so far proven out of his reach. Courts and Wall Street have moved to defend the central bank’s longstanding independence from political interference.
With Warsh, Trump believes he’s struck a balance. Global markets were broadly steady after the announcement, which was welcomed by figures in the central banking establishment, like Mark Carney, former governor of the banks of England and Canada – and the president seems confident he has tapped a Fed chair who will prove loyal.
“I have known Kevin for a long period of time, and have no doubt that he will go down as one of the GREAT Fed Chairmen, maybe the best,” Trump said in a social media post Friday. “On top of everything else, he is ‘central casting,’ and he will never let you down.”
Warsh has a reputation for being “hawkish”, meaning he takes a more conservative approach to monetary policy. Fed hawks typically are sensitive about inflation, encouraging high interest rates – which tame price increases – even when it could lead to more unemployment.
When Warsh was a Fed governor, he often focused on rising inflation, even as the labor market was in freefall. But over the last year, as Trump openly considered who should replace Powell at the top of the central bank, Warsh publicly made arguments that were more in line with the president’s demands for rate cuts.
In a Wall Street Journal op-ed in November, he praised Trump’s “pro-growth policies” and argued (just like the president has) that the Fed had held back the economy with high rates.
Some have voiced skepticism that Warsh will remain dove-ish on inflation, and maintain his belief that rates should be lower, once he walks back through the doors of the Fed. “His dovishness today stems from convenience,” researchers at Renaissance Macro Research said on Friday. “The president risks getting duped.”
Warsh’s nomination does not guarantee him the job. He still requires confirmation by the US Senate – and the support of key figures in Congress who have grown alarmed by the administration’s treatment of the current Fed chair.
Trump unveiled Warsh as his pick two weeks after the White House has faced widespread backlash when it emerged the Department of Justice had launched a criminal investigation into Powell. Republican senator Thom Tillis made clear on Friday that, though he supports Warsh as nominee, he will block his confirmation until the investigation is resolved.
And even if Warsh is confirmed, chairing the Fed is tantamount to walking a delicate tightrope.
The US central bank depends on the credibility of its independence. This credibility could be damaged if its leader even appears to prioritize pleasing the president over steady guidance of the world’s largest economy.
US interest rates since 2019
There are built-in protections to insulate the Fed from politics. Its chair doesn’t decide on interest rates alone: any change requires a consensus among the 12 voting members of the rate-setting Federal Market Open Committee (FOMC).
Seven of FOMC’s voting members are Fed governors who serve out 14-year terms. And while Trump has tried to kick out Lisa Cook, one of those governors, the US supreme court seems prepared to protect her from the president’s wrath.
Powell’s term as chair is set to end in May, but his term as a Fed governor is not up until 2028. While most chairs usually step down from the board after their term as chair concludes, Powell has so far refused to comment on whether he will stay on.
That staying on even remains a possibility for Powell points to the unspoken uncertainty surrounding the end of his term.
At a press conference earlier this week, after the Fed again declined to cut rates, Powell refused to answer any questions about Trump, the justice department’s investigation, or his possible successor. “I have nothing for you on that,” he repeatedly told assembled reporters.
But one thing he was willing to speak on, at length, was the importance of the Fed’s independence. Should it ever be used to sway elections, Powell cautioned: “It would be hard to restore the credibility of the institution.
“If people lose their faith that we’re making decisions only on the basis of our assessment of what’s best for everyone … it’s going to be hard to retain it,” he said. “We haven’t lost it, I don’t believe we will. I certainly hope we won’t.”
If you are considering buying a new car, now might be the time to act as new data shows manufacturers and dealers slashing prices by more than 10%, with the average discount close to £6,000.
The typical discount available across all petrol, diesel and electric cars sold in the UK is 11.4% of the on-the-road price – the equivalent of £5,911 – according to Insider Car Deals, which sells discount data to people looking to buy.
On-the-road prices include all of the extras required to drive a car, including road tax, registration fees, number plates, delivery and VAT.
For electric vehicles (EVs) the average price reduction is 12.9% – including the savings from government electric car grants that apply to certain vehicles. The absolute size of the discounts on battery cars is larger, at £7,091, mainly because manufacturers have been slower to produce electric versions of the smaller, cheaper cars that are the backbone of the petrol market.
While this is all good news for consumers, car industry executives complain that official EV sales targets are too tough, forcing them to offer “unsustainable” discounts on battery cars amid intense competition – particularly from new Chinese brands.
Also, despite the car industry’s focus on EVs, as they lobby for easier targets, for some models the discount on a petrol model is proportionally bigger. Take the popular British-made Nissan Qashqai: the price data suggest that a new model can be found for £36,000, or 17.9% off the recommended retail price.
Nissan has not begun producing its electric Qashqai, but its similar-sized electric vehicle, the Ariya, has a discount of only 13.1%, putting it at an average price of £45,264 (with some versions eligible for grants).
Kia’s Niro EV can be found at a 7.1% discount, at £38,460, without being eligible for a grant, whereas there is 13.2% off a Kia Niro Hybrid, which combines a small battery with a petrol engine, putting it at £33,225.
Whichever choice buyers make, it usually pays to haggle – particularly at the end of the month or the quarter when dealers may have sales targets they need to meet.
Even if the upfront price for a new electric car remains higher in many cases, the total cost of ownership may well be lower, particularly for people who can charge the car at home. Battery motors are much more efficient, meaning less energy is lost to heat and noise, which leads directly to lower energy costs (and much lower carbon emissions), while maintenance costs are also lower on average.
The Energy & Climate Intelligence Unit (ECIU), a campaign group, last month calculated that internal combustion engines cost drivers an annual £1,300 “petrol premium” at 2025 fuel prices.
Colin Walker, the ECIU’s head of transport, says: “With a majority of people able to charge their vehicles at home and take advantage of cheap, night-time charging tariffs, these vehicles can deliver savings worth hundreds, even thousands of pounds a year.”
The UK’s electric car grant of up to £3,750 has meant that price parity between electric and petrol has arrived for some new car models. Walker highlighted the Ford Puma, the bestselling car in the UK in 2025, which starts from £26,580. That compares with £26,245 for the electric Puma Gen-E, after the grant.
“It’s no surprise that EV sales increased so significantly in 2025,” says Walker. “The choice of models is constantly expanding and, crucially, the price of EVs is coming down as manufacturers compete with each other to hit their EV sales targets.”
Shortages of new cars during the coronavirus pandemic and the consequent chip shortage meant discounts were not needed to shift cars, but the market is shifting back – with the added impetus of competition from Chinese marques such as BYD, SAIC’s MG and Chery’s Omoda and Jaecoo offering keen prices. That has put bargains back on the table at car dealerships.
Pat Hoy, the founder of Insider Car Deals, says the level of discounting is “not unprecedented. The market as a whole is starting to look a lot like it did just before Covid, where the manufacturers are getting back to usual sale patterns, turning discounts on and off.”
Track your investments for FREE with Simply Wall St, the portfolio command center trusted by over 7 million individual investors worldwide.
Barclays (LSE:BARC) is back on many investors’ screens after its latest reported figures showed £26,016m in revenue and £5,945m in net income. This has prompted fresh interest in how the share price lines up with those fundamentals.
See our latest analysis for Barclays.
At a share price of £4.862, Barclays has seen a 19.46% 90 day share price return and a 1 year total shareholder return of 68.05%. This suggests momentum has been building around the story, both on recent results and the longer term turnaround.
If Barclays’ recent run has caught your eye, it could be worth broadening your search to other banks and financials or casting the net wider across the market. If you are curious about what else is moving, now is a good time to broaden your investing horizons and check out fast growing stocks with high insider ownership
With Barclays trading at £4.862, an implied intrinsic discount of around 45% and only a small gap to analyst targets, investors are left with a key question: is there still a buying opportunity here, or is the market already pricing in future growth?
Barclays’ most followed narrative points to a fair value of £4.92, just above the current £4.86 share price. This frames the recent rally in a tight valuation range.
Investments in digital banking, client relationship growth, and acquisitions are boosting efficiency and expanding revenue in high margin, structurally growing segments. Strategic cost control, technology upgrades, and business mix improvements are driving consistently higher returns and strengthening long term earnings quality.
Read the complete narrative.
Curious what sits behind that near match between price and fair value? The narrative leans on steady top line expansion, firm margins and a richer earnings multiple. The model also builds in buybacks and a specific required return. Want to see exactly how those moving parts combine into that £4.92 figure?
Result: Fair Value of £4.92 (UNDERVALUED)
Have a read of the narrative in full and understand what’s behind the forecasts.
However, this depends on deposits remaining stable and credit quality staying resilient, with tougher competition or weaker conditions potentially putting pressure on margins and earnings.
Find out about the key risks to this Barclays narrative.
If you look at the numbers and reach a different conclusion, or simply prefer to test your own view, you can build a personalised Barclays story in just a few minutes with Do it your way
A great starting point for your Barclays research is our analysis highlighting 3 key rewards and 3 important warning signs that could impact your investment decision.
If Barclays has sharpened your thinking, do not stop here. Broaden your opportunity set with focused stock ideas that match the kind of portfolio you want to build.
This article by Simply Wall St is general in nature. We provide commentary based on historical data and analyst forecasts only using an unbiased methodology and our articles are not intended to be financial advice. It does not constitute a recommendation to buy or sell any stock, and does not take account of your objectives, or your financial situation. We aim to bring you long-term focused analysis driven by fundamental data. Note that our analysis may not factor in the latest price-sensitive company announcements or qualitative material. Simply Wall St has no position in any stocks mentioned.
Companies discussed in this article include BARC.L.
Have feedback on this article? Concerned about the content? Get in touch with us directly. Alternatively, email editorial-team@simplywallst.com
Disclaimer: IMF Working Papers describe research in progress by the author(s) and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management.
Summary
The absence of reliable data on fundamental economic indicators (e.g. real GDP), combined with structural shifts in the economy, can severely constrain the ability to conduct accurate macroeconomic analysis and forecasting. This paper explores alternatives to address data limitations by integrating machine learning and satellite data to estimate real GDP. Specifically, it finds that incorporating satellite-based nightlight data into a random forest model significantly improves the accuracy of quarterly GDP growth estimates compared with models relying solely on traditional indicators. This empirical application contributes to the emerging nowcasting field to enhance economic forecasting in economies with significant data gaps.
Subject: COVID-19, Econometric analysis, Economic and financial statistics, Economic forecasting, Environment, Health
Keywords: COVID-19, GDP, Machine learning, Macroeconomic forecast, Nowcasting, Pacific Islands, Random Forest, Satellite data
One in seven food businesses on major delivery platforms, including Deliveroo and Just Eat, is now a “dark kitchen”, a university study shows.
The findings, which shine a light on the scale of the hidden takeaway industry, found that 15% of all online food retailers in England were dark kitchens.
Also known as “cloud”, “ghost” or “virtual” kitchens, they are delivery-only with no customer-facing storefront.
Despite rapid growth, they have – until now – lacked a clear and consistent definition, creating challenges for regulators, local authorities, food safety officers, industry stakeholders and consumers.
Dr Lucie Nield, co-lead investigator from the University of Sheffield, said: “People deserve greater transparency about the food they are ordering online, and these businesses must be held to the appropriate regulatory standards.
“Without this, dark kitchens risk falling through the gap, with potential consequences for public health, particularly by encouraging increased use of online takeaways, greater availability and therefore greater consumption of high fat, salt or sugar food.
“Dark kitchens have previously been poorly defined and under-researched, making their impacts difficult to fully understand.
“Adopting a shared definition is essential for clearer communication, more effective regulation and inspection and for driving public health agendas.”
A dark kitchen on an industrial estate in south London. The sites can help companies lower their operating costs. Photograph: Anna Watson/Alamy
The study, which was commissioned by the National Institute for Health and Care Research, brought together multiple university teams to establish the first industry-wide framework for defining and identifying dark kitchens.
Researchers worked with academics, public health professionals, local authorities, national governing bodies, industry workers and consumers to ensure the definition reflected how the businesses operated.
The final wording was: “Technology-enabled commercial kitchen(s) operating primarily for delivery, to fulfil remote, on-demand, consumer online orders of food for immediate consumption.”
The academics also examined the scale of dark kitchens on major delivery apps.
Using data-scraping methods across platforms such as Uber Eats and Deliveroo, they identified clusters of food brands operating from the same postcode, a key indicator of delivery-only hubs.
The study also cross-referenced locations using tools such as Google Maps, highlighting how difficult these businesses can be for consumers and regulators to spot.
Although dark kitchen models offer commercial advantages, including lower operating costs and flexible locations, the research highlighted potential implications for public health.
Unlike traditional takeaways, which can be regulated by local authorities using spatial planning policy, dark kitchens are far less visible.
Management zones around schools, which are designed to limit the density of takeaways and support healthier food environments, do not apply to dark kitchens and may be undermined by their activities.
Researchers also raised concerns around food safety and transparency, particularly for customers with allergies or dietary sensitivities. Because multiple businesses can operate from the same kitchen space, consumers may not always be aware of shared preparation environments or potential allergen cross-contamination.
A survey in 2023 found that 40% of participants bought a takeaway at least weekly, commonly via delivery apps and mainly as a treat or for convenience.
Awareness of dark kitchens was low, with only a quarter having heard of them and just 9% knowingly using one. However, after reading a working definition, more than half said they would consider buying from a dark kitchen, though most wanted this to be made explicit.
The new industry-wide definition is aimed at bringing delivery-only hubs under stricter planning and public health oversight with clearer regulation, inspection processes and consumer understanding.
The number of cancer survivors in the United States is increasing, with over 18 million currently living with a history of cancer diagnosis [,]. While cure rates are encouraging, cancer survivors represent a population with a high burden of comorbidity, including cardiovascular disease, diabetes, and obesity [,]. Furthermore, most cancer survivors do not meet the nutrition and physical activity guidelines recommended by the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) and the American Cancer Society (ACS), often reporting low intake of vegetables and fruits (V&Fs) and insufficient physical activity [-]. These poor lifestyle practices can further exacerbate existing comorbidities, compounding their impact on long-term survivorship. Despite these challenges, cancer survivors frequently express a strong interest in improving their lifestyle behaviors []. Thus, lifestyle interventions that target diet, physical activity, and weight management have been implemented to promote healthier behaviors, support weight loss, and enhance the quality of life of cancer survivors [-].
Over the past decade, the landscape of lifestyle interventions for cancer survivors has evolved, with an increasing shift toward accessible and scalable digital delivery methods [-]. Among these, web-based lifestyle interventions have emerged as a promising approach to disseminate diet and physical activity guidance to cancer survivors [,]. Websites offer several advantages, such as scalability, personalization, cost-effectiveness, and the flexibility for participants to engage with content at their own pace [,,]. To harness these advantages, the SurvivorSHINE pilot study, a 3-week, single-arm web-based lifestyle intervention, was implemented through an interactive website to promote diet and physical activity guidelines among 41 cancer survivors [,]. The website incorporated common strategies such as evidence-based diet and exercise knowledge, tools to facilitate behavior change, and resources for self-monitoring and goal setting []. The study reported that cancer survivors perceived the SurvivorSHINE website as a user-friendly platform that provided trustworthy information on diet and exercise, and reported improvements in knowledge related to diet and exercise [].
Building upon the frameworks and strategies of the SurvivorSHINE intervention, a more refined website was developed that included 24 weekly serialized, interactive sessions, as well as additional tools to support behavior change not only among cancer survivors interested in cancer control, but also among their family members and friends for the purposes of cancer prevention. This newly developed website was then evaluated for feasibility and its impact on various health outcomes in a randomized controlled trial, known as DUET (Daughters, Dudes, Mothers, and Others Together; NCT04132219) that targeted cancer survivors and their chosen partners [,].
The DUET trial was 6 months in duration and evaluated the web-based weight loss intervention against a waitlist control among 112 participants (56 dyads, each consisting of a cancer survivor and a chosen partner) [,]. Results showed that dyads in the intervention arm achieved significant weight loss, along with improvements in diet quality and physical activity levels compared to the control arm []. A mediation analysis further revealed that reductions in perceived dietary barriers significantly contributed to the observed weight loss, highlighting the effectiveness of the intervention’s strategies, which were based on social cognitive theory [-]. These findings underscore the potential of minimal-touch, web-based lifestyle interventions to facilitate meaningful behavior change and weight loss among cancer survivors and their partners. However, to optimize delivery and behavioral outcomes, it is important to understand how cancer survivors and their partners used the DUET website. While many web-based interventions focus on evaluating behavioral outcomes, relatively few have described website use or examined the association between website use and changes in diet, physical activity, and weight loss. Exploring patterns of use within the DUET website may provide insight into participant interaction with web-based platforms and inform refinements for future web-based lifestyle interventions for cancer survivors and their partners.
Aim and Objective
This secondary analysis aims to describe website use in the DUET trial, and also examine the associations between website use (as measured by number of weeks of website use, average time spent, total page views, and session page views) and changes in diet quality, moderate to vigorous physical activity (MVPA), and weight.
Methods
Study Design
The DUET study was a 2-arm single-blinded randomized controlled trial that enrolled 112 cancer survivors and their chosen partners (56 dyads). Dyads were randomized to either the 6-month web-based weight loss intervention or a waitlist control arm. This secondary analysis focuses on the subset of 56 participants (28 dyads) assigned to the DUET intervention arm, as website usability data were only collected for this group. Full details of the trial methods, primary outcomes, and mediation analysis have been published elsewhere [-].
Study Participants
Cancer survivors were identified through multiple recruitment strategies, including cancer registries, self-referrals, and curated lists of individuals who had previously expressed interest in lifestyle interventions. Recruitment efforts focused on cancer survivors who had completed treatment for obesity-related cancers with a 5-year survival rate of ≥70% (eg, localized renal, locoregional ovarian, colorectal, prostate, endometrial, or female breast cancer). Potential cancer survivors were contacted by mail with telephone follow-up, and, if interested, were screened by study staff for the following inclusion criteria: (1) BMI ≥25 kg/m²; (2) V&F intake <2.5 cups/day; (3) engaging in <150 minutes/week of MVPA; and (4) routine access to the internet via a computer, tablet, or mobile phone. Eligible cancer survivors were asked to identify a partner who lived nearby (within 10 minutes by car) and interacted with them at least biweekly; partners were screened using the same criteria, excluding a cancer diagnosis. However, partners with a history of cancer were eligible.
Study Protocol
Eligible dyads were provided with an overview of the study protocol, and written informed consent was obtained electronically using Adobe Sign []. After completion of baseline assessments, dyads were randomized to either the 6-month web-based weight loss intervention or a waitlist control arm. Those assigned to the intervention arm were granted access to the DUET website via an electronic link and instructed to create a secure profile using a username and password unique for each survivor and partner []. Dyads were encouraged to log in weekly via text messages and engage with the key features of the DUET website throughout the 6-month intervention period. Follow-up assessments were conducted at 6 months, after which waitlist control dyads were provided access to the DUET intervention. Additional details on the study procedures have been published previously [].
DUET Intervention Website
The DUET intervention was theoretically grounded in social cognitive theory and incorporated elements from interdependence theory and the theory of communal coping to support both individual and dyadic behavior change by targeting diet and exercise barriers, enhancing social support, and building self-efficacy [,,,]. DUET was adapted from 2 prior evidence-based lifestyle interventions, Daughters and Mothers Against Breast Cancer and SurvivorSHINE, and delivered via a secure, interactive website that served as the primary web-based platform for intervention delivery [,,]. Details on the intervention development have been published previously [].
The DUET website included a total of 9 key features: Home Page, My Profile, Sessions, Healthy Weight, Healthy Eating, Exercise, Tools, News You Can Use, and Team Support. Upon account creation, participants accessed the My Profile feature to enter demographic and lifestyle data, including their current weight, height, and responses to 1-item questions that assessed their dietary intake (eg, consumption of V&Fs, whole grains, red and processed meats, added sugars, and alcohol), and frequency of snacking and physical activity (both endurance and resistance exercise). Cancer survivors also provided details on diagnosis and treatment, which informed tailored feedback on overcoming treatment-related challenges and generating dietary, physical activity, and weight management goals, as described previously [-]. The Home Page displayed a “Tip of the Day” designed to encourage ongoing engagement with the website, along with visual indicators showing participants’ completed weekly sessions and the next upcoming session. It also included direct links to other key features of the website (ie, Healthy Weight, Healthy Eating, Exercise, Tools, News You Can Use, and Team Support).
The Sessions feature included 24 weekly interactive e-learning modules (~15 minutes each), designed using Articulate Storyline software []. Sessions were released sequentially each week over the 24 weeks and introduced via a Monday SMS “push” text message, with additional text messages sent on Wednesdays and Fridays to reinforce continued engagement. These interactive e-learning sessions were designed to provide information on the WCRF/AICR and ACS diet and physical activity guidelines and equip participants with practical and actionable strategies to support behavior change and weight management [,]. Dietary recommendations were supported by sessions focused on promoting the consumption of V&Fs, whole grains, and legumes, and limiting red and processed meat, added sugar, alcohol, and snacking, with additional sessions focused on portion control, grocery shopping, and food preparation to support healthy eating habits. Physical activity recommendations were supported by sessions focused on aerobic, resistance, balance, and flexibility exercises, with emphasis on goal setting and problem solving to help participants gradually achieve the goal of 150 minutes of MVPA per week. Details on weekly topics have been published previously [].
The Healthy Weight feature was designed to support self-monitoring of weight by providing an interactive bar graph that tracked participants’ current weight and healthy weight. Accompanying educational materials helped contextualize these values by explaining the concept of a healthy weight and its relevance to cancer prevention and survivorship. The website delivered tailored guidance to help participants progress toward a healthy weight by supporting caloric restriction and strategies to promote gradual weight loss of approximately 0.5 kg/week [].
The Healthy Eating feature supported self-monitoring of dietary goals aligned with the WCRF/AICR and ACS guidelines. Participants could record their current intake of key dietary components, that is, V&Fs (≥5 servings/day), whole grains (≥50% of total grain intake), added sugars (≤6 teaspoons/day), avoid snacks, red and processed meats (≤18 ounces/week), and alcohol (≤1 drink/day), through an interactive bar graph that visually displayed their reported intake alongside goal targets. To further support dietary changes, the feature provided educational resources on each dietary component.
The Exercise feature supported self-monitoring of physical activity through tailored recommendations based on participants’ self-reported activity levels. Physical activity goals were aligned with WCRF/AICR and ACS guidelines, encouraging participants to achieve at least 150 minutes of MVPA and engage in strength training 2-3 times per week.
The Tools feature provided a centralized hub for participants to access downloadable materials that supported both dietary and physical activity behaviors aimed at achieving a healthy weight. This feature included 11 distinct resources: (1) BMI calculator, (2) calorie calculator, (3) sample meal plans, (4) food exchange lists, (5) SMART goal templates, (6) serving size guides, (7) fast food guide, (8) grocery lists and shopping tips, (9) calorie-burning guide, (10) exercise logs, and (11) tools for tracking Fitbit data (note: all DUET participants regardless of randomization status, received a Fitbit Inspire and were encouraged to use it during the study period; however, Fitbit data were not integrated into the DUET website) []. These resources were designed to offer participants additional support, practical strategies, and accessible tools to enhance self-efficacy throughout the intervention.
The News You Can Use feature provided brief, evidence-based summaries on current research related to cancer survivorship, diet, and exercise. The Team Support section provided strategies to strengthen dyadic communication, foster mutual goal setting, and enhance social support; it also allowed participants to directly connect with study staff for additional guidance and support.
Measures
Demographics
Cancer type and time since diagnosis were obtained from cancer registries or verified by treating physicians for self-referred participants. Demographic information, including age, sex, race, residence, educational status, employment, and income, was self-reported via electronic surveys completed at baseline. Cohabitation was assessed by comparing mailing addresses; dyads with the same address (0 miles) were classified as cohabitating, and those with different addresses (>0 miles) as noncohabitating.
Diet Quality, MVPA, and Weight
Dietary intake was assessed at baseline and 6 months via two 24-hour dietary recalls (1 weekday and 1 weekend day) conducted by a registered dietitian over the telephone. The Automated Self-Administered 24-hour Dietary Assessment Tool was used to capture dietary intake data. Diet quality was evaluated using the Healthy Eating Index 2015 [,].
MVPA was assessed both objectively and subjectively at baseline and 6 months. Participants wore ActiGraph accelerometers for 7 consecutive days. Data were then processed using ActiLife software following standardized procedures to calculate average weekly minutes of MVPA [,]. Self-reported MVPA was captured using the Godin Leisure-Time Exercise Questionnaire, a validated tool frequently used in cancer survivorship research [].
Weight was measured remotely at baseline and 6 months. Each survivor-partner dyad completed the virtual assessment together via Zoom with study staff []. Participants used a digital bathroom scale to report their weight; a scale was provided with the assessment materials for those who did not own one. During the virtual assessment, trained staff instructed participants to wear light clothing and remove their shoes, and partners assisted in holding the camera and angling it so that study staff could verify the weight displayed on the digital scale. Additional details on the remote assessment protocol and its validity have been described previously [].
Website Use Metrics
Website use was assessed using tracking data logged by the DUET website platform. Each participant was assigned a unique website username and password, which was linked to their website ID and further connected to their study ID, allowing for the tracking of individual-level website activity over the 24-week intervention period. Time-stamped data recorded the days, times, and pages accessed by participants (eg, Home Page, My Profile, Sessions, Healthy Weight, Healthy Eating, Exercise, Tools, News You Can Use, and Team Support). From these data, key website use metrics were derived, including (1) the total number of weeks participants accessed the website, defined as the number of distinct weeks during the 24-week intervention period in which any website activity was recorded; (2) the average time spent on the website per day of use, calculated using time-stamped activity logs that captured first and last activity on a given day; (3) the total number of page views, defined as the cumulative number of times participants navigated to individual pages within each of the website’s key features; and (4) the total number of session page views, defined as the number of times participants accessed the e-learning Sessions feature.
Statistical Analysis
Website usability was analyzed using descriptive statistics, including means, SDs, and ranges for continuous variables (eg, number of weeks accessed, time spent on the website, and page views). Website usability metrics were described for the total sample and stratified by dyad member (cancer survivors or partners), clinical (ie, cancer type and time since diagnosis) and sociodemographic factors (ie, age, sex, race, residence, educational status, employment, and income) and cohabitation status (cohabitate or did not cohabitate), with all stratification variables dichotomized for analysis. To examine differences in website use between dyad members, clinical and sociodemographic factors, and cohabitation status, assumptions of normality were assessed using Shapiro-Wilk tests and visual inspection of histograms and Q-Q plots. Given that usability metrics were not normally distributed, Wilcoxon rank-sum tests were conducted to compare median differences in website use between dyad members, clinical and sociodemographic factors, and cohabitation status. Assumptions of normality and linearity were assessed for the independent (website use metrics) and dependent variables (diet quality, MVPA, and weight). Given evidence of nonnormal distributions, likely influenced by the modest sample size and potential nonlinear relationships, nonparametric methods, such as bivariate Spearman partial rank correlation analyses, were conducted to examine associations between website use and diet quality, MVPA, and weight. Correlation coefficients were generated using 6-month outcome values as dependent variables, adjusting for baseline values of the respective outcome, as well as age, sex, and race to account for potential confounding and assess change over time. Given this was a secondary analysis and not prospectively powered for these aims, we conducted a post hoc power calculation to aid interpretation of the correlation analyses. With a total sample size of 56 participants, the study had ≥80% power to detect correlations of approximately r≥0.40. Missing data were handled using complete-case analysis. One participant had missing diet quality data at 6-month follow-up, and 13 participants had missing accelerometer-measured MVPA data at baseline (n=8) and 6-month follow-up (n=5); these cases were excluded from their respective models. No adjustments for multiple comparisons were made, given the exploratory nature of this secondary analysis. However, to address the increased type I error risk from multiple correlations, 95% CIs were reported with P values to assist in interpreting the precision of the coefficients. All analyses were conducted using SAS (version 9.4, SAS Institute Inc), and statistical significance was set at P<.05 [].
Ethical Considerations
The DUET study was approved by the Institutional Review Board at the University of Alabama at Birmingham (IRB# 300003882) and was registered with ClinicalTrials.gov (NCT04132219). All participants provided written informed consent, and study procedures were conducted in accordance with the ethical standards of the Declaration of Helsinki to maintain participant confidentiality. No compensation was provided to study participants.
Results
Sample Characteristics
The average age of the sample was 58 (SD 12.5) years, with survivors averaging 60 (SD 11.2) years and partners 56 (SD 13.7) years. The majority of cancer survivors (24/28, 85.7%) had a breast cancer diagnosis, with an average time since diagnosis of approximately 71 (SD 80.4) months (6 years). Most participants identified as female (44/56, 78.6%), non-Hispanic White (37/56, 66.1%), and residents of urban areas (53/56, 94.6%). Employment status was evenly split between employed (30/56, 53.6%) and retired (26/56, 46.4%), and most (45/56, 80.4%) reported an annual household income above US $50,000 per year ().
Table 1. Characteristics of 56 cancer survivors and their chosen partners randomized to the intervention arm, stratified by dyad status.
Characteristics
Total sample
Survivor
Partner
P valuea
Age (years), mean (SD; range)
58.1 (12.5; 23-78)
60.0 (11.2; 32-78)
56.3 (13.7; 23-74)
.28
Months from diagnosis, mean (SD; range)
71 (80.4; 10-303)
71 (80.4; 10-303)
—b
—b
BMI (kg/m2), mean (SD; range)
31.4 (4.9; 25-45)
32.0 (5.4; 25-44)
30.9 (4.4; 25-45)
.41
Diet quality (HEIc), mean (SD; range)
53.1 (12.8; 29-87)
53.9 (13.7; 30-87)
52.2 (12; 29-81)
.62
MVPAd (min/week), mean (SD; range)
43.8 (60.5; 0-280)
48.5 (67.8; 0-280)
39.1 (52.9; 0-225)
.57
Cancer typee, n (%)
<.001
Breast
25 (44.6)
24 (85.7)
1 (3.6)
Otherf
7 (12.5)
4 (14.3)
3 (10.7)
Sex, n (%)
.05
Male
12 (21.4)
3 (10.7)
9 (32.1)
Female
44 (78.6)
25 (89.3)
19 (67.9)
Race, n (%)
.78
Non-Hispanic White
37 (66.1)
19 (67.9)
18 (64.3)
Non-Hispanic Black or otherg
19 (33.9)
9 (32.1)
10 (35.7)
Residence, n (%)
.55
Urban
53 (94.6)
27 (96.4)
26 (92.9)
Rural
3 (5.4)
1 (3.6)
2 (7.1)
Educational status, n (%)
.13
High school or less
8 (14.3)
2 (7.1)
6 (21.4)
Some college or more
48 (85.7)
26 (92.9)
22 (78.6)
Employment, n (%)
>.99
Employed
30 (53.6)
15 (53.6)
15 (53.6)
Retired or otherh
26 (46.4)
13 (46.4)
13 (46.4)
Income, n (%)
.09
Less than US $50,000/year
11 (19.6)
8 (28.6)
3 (10.7)
More than US $50,000/year
45 (80.4)
20 (71.4)
25 (89.3)
aP values were calculated using independent samples t tests for continuous variables (based on equal or unequal variances as appropriate) and chi-square tests for categorical variables, comparing survivors vs partners, significance set at P<.05.
bData on months since diagnosis is unavailable for partners.
cHEI: Healthy Eating Index 2015.
dMVPA: moderate to vigorous physical activity.
eTotal percentage does not sum to 100% due to the inclusion of nonsurvivors who were not diagnosed with cancer.
fOther cancer diagnoses include prostate, colorectal, gynecologic, and renal.
gOther race includes Hispanic ethnicity, accounting for 1%.
hOther employment includes student and disabled.
DUET Website Use
On average, participants accessed the DUET website for 11.2 (SD 7.4; range 0-24) weeks and spent a total of 312.9 (SD 255.7; range 0-1119) minutes on the platform, or about 13 (SD 10.7; range 0-46.6) minutes per week over the 24-week intervention period. A total of 5885 page views were recorded, with an average of 105.1 (SD 105.7; range 0-545) page views per participant. Cancer survivors used the website more frequently than their partners, accessing it for a longer duration (mean 13.0, SD 7.2 weeks vs mean 9.5, SD 7.4 weeks; P=.08), spending more time per week (mean 15.6, SD 11.4 minutes vs mean 10.5, SD 9.4 minutes; P=.07), and recording a higher average number of page views (mean 124.2, SD 114.2 vs mean 86.0, SD 94.5; P=.07); however, these differences were not statistically significant ().
Table 2. Daughters, Dudes, Mothers, and Others Together website usability over the 24-week intervention period for the total sample (n=56) and stratified by dyad status.
Engagement metrics
Total sample, mean (SD; range)
Survivor, mean (SD; range)
Partner, mean (SD; range)
P valuea
Weeks participants accessed website
11.2 (7.4; 0-24)
13.0 (7.2; 0-24)
9.5 (7.4; 0-23)
.08
Total time spent on website (min)
312.9 (255.7; 0-1119)
373.8 (273.3; 0-1119)
252.0 (225.5; 0-670)
.07
Time spent on website per week (min)
13.0 (10.7; 0-46.6)
15.6 (11.4; 0-46.6)
10.5 (9.4; 0-27.9)
.07
Total page views per user (n=5885)
105.1 (105.7; 0-545)
124.2 (114.2; 0-545)
86.0 (94.5; 0-352)
.07
aP values represent comparisons between survivors and partners using Wilcoxon rank-sum tests due to the nonnormal distribution of website usability metrics, significance set at P<.05.
Website use differed significantly by age, time since diagnosis, and sex. Older adults aged ≥65 years demonstrated higher website use compared to younger participants (<65 years), accessing it for a longer duration (mean 14.4, SD 7.4 weeks vs 9.2, SD 6.8 weeks; P=.009), spending more time per week (mean 17.0, SD 9.7 minutes vs mean 10.5, SD 10.6 minutes; P=.01), and recording a higher average number of total page views (mean 135.7, SD 90 vs mean 85.3, SD 111.9; P=.008). Survivors who were 5 or more years post diagnosis also accessed the website for longer duration than those more recently diagnosed (<5 years; mean 18.4, SD 6.3 weeks vs mean 10.8, SD 6.4 weeks; P=.009). Additionally, females used the website significantly more than males, accessing it for a longer duration (mean 12.8, SD 7.1 weeks vs mean 5.6, SD 5.9 weeks; P=.003), spending more time per week (mean 14.6, SD 10.3 minutes vs mean 7.4, SD 10.3 minutes; P=.02), and recording a higher average number of total page views (mean 120, SD 110.2 vs mean 50.3, SD 64.4; P=.01; ).
Table 3. Website use stratified by clinical characteristics, sociodemographic factors, and cohabitation status.
Measures
Weeks
Average time
Total page views
Session page views
Mean (SD)
Test statistica
P value
Mean (SD)
Test statistic
P value
Mean (SD)
Test statistic
P value
Mean (SD)
Test statistic
P value
Age
6.83
.009
6.55
.01
7.03
.008
6.34
.01
<65 years
9.17 (6.8)
10.5 (10.6)
85.3 (111.9)
38.4 (35.2)
≥65 years
14.4 (7.4)
17 (9.7)
135.7 (90)
65 (38.5)
Time since diagnosis
6.88
.009
1.62
.20
2.57
.11
0.622
.43
<5 years
10.8 (6.4)
14 (11.9)
108.4 (117.8)
52.5 (35.1)
≥5 years
18.4 (6.3)
19.4 (9.6)
163.6 (100.9)
65.3 (32.3)
Cancer type
0.64
.42
0.03
.87
0.53
.47
1.15
.28
Breast
13.6 (7.2)
15.3 (11.3)
122.9 (119.5)
54.8 (37.1)
Other
11.6 (4.9)
15.9 (9.7)
129.6 (81)
27 (36.7)
Sex
8.82
.003
5.70
.02
6.69
.01
6.39
.01
Female
12.8 (7.1)
14.6 (10.3)
120 (110.2)
54.8 (37.1)
Male
5.6 (5.9)
7.4 (10.3)
50.3 (64.4)
27 (36.7)
Race
0.02
.89
0.08
.78
0.11
.75
0.01
.91
Non-Hispanic White
11.4 (6.8)
12.9 (9.6)
100.2 (86.2)
46.4 (33.2)
Non-Hispanic Black
10.9 (8.6)
13.3 (12.5)
113.9 (136)
53.4 (47.1)
Residence
0.01
.91
0.41
.52
0.28
.60
0.90
.34
Urban
11.2 (7.3)
12.7 (10.3)
104 (106.8)
47.2 (36.3)
Rural
12.3 (11)
18.4 (17.1)
123.7 (98.1)
77.7 (70.3)
Education
0.95
.33
0.01
.92
0.08
.78
0.00
.98
High school or less
8.9 (6.3)
12 (10.5)
116 (131.7)
45.9 (37.8)
Some college or above
11.6 (7.6)
13.2 (10.8)
103.3 (102.2)
49.4 (38.9)
Employment
1.29
.26
0.14
.71
0.35
.55
0.14
.71
Employed
10.4 (6.3)
12.9 (11.7)
101.8 (117.1)
47.6 (39.7)
Retired or other
12.2 (8.5)
13.3 (9.5)
108.9 (92.8)
50.3 (37.7)
Income
0.08
.78
0.45
.50
0.18
.67
0.81
.37
Less than US $50k/year
10.6 (5.9)
15.9 (13.9)
132 (152.8)
59.7 (44.9)
More than US $50k/year
11.4 (7.8)
12.3 (9.8)
98.5 (91.7)
46.2 (36.8)
Cohabitation
2.23
.14
1.54
.21
0.40
.53
2.01
.16
Cohabitate
9.5 (8.1)
11.1 (10.3)
93.1 (91)
42.2 (39.8)
Do not cohabitate
12.5 (6.7)
14.5 (10.8)
114.1 (116)
53.8 (37.2)
aDifferences between groups were assessed using the Mann-Whitney U test, implemented via the Wilcoxon rank-sum approach. A chi-square approximation was used to derive the test statistics, with statistical significance defined as P<.05.
Website use was generally highest during the initial 9 weeks of the intervention, although some fluctuations were observed, particularly in weeks 4, 6, and 8. Website use gradually declined over time, with a slight increase observed in the final week of the intervention. Cancer survivors spent more time on the website each week compared to their partners, with notable peaks observed among cancer survivors at weeks 1, 2, 6, 8, 13, 16, 19, and 24 ().
Figure 1. Average weekly time spent on the Daughters, Dudes, Mothers, and Others Together (DUET) website during the 24-week intervention period by the total sample, highlighting high and low engagement weeks stratified by dyad status. All weekly DUET sessions were designed to take 15 minutes or less, except for the initial onboarding session.
A total of 5885 page views were recorded throughout the intervention period across the 9 key website features. The highest number of page views was observed for the weekly interactive e-learning Sessions feature (n=2736), followed by the Home Page (n=975) and Tools feature (n=967). The remaining features, My Profile (n=400), Healthy Weight (n=248), Healthy Eating (n=234), Exercise (n=162), News You Can Use (n=104), and Team Support (n=59), had comparatively fewer page views ().
Figure 2. Page view distribution across key features of the Daughters, Dudes, Mothers, and Others Together website stratified by dyad status (n=5885 total page views).
further illustrates the distribution of 2736 page views across the weekly e-learning Sessions feature over the 24-week intervention period. Page views for the weekly sessions were highest during the early weeks of the intervention, particularly sessions 1 through 8, and gradually declined over time. The onboarding session received the greatest number of views, and the three most frequently viewed sessions were (1) Get on Track for Success (session 2), (2) Moving Towards Better Health (session 4), and (3) Been Resisting “Resistance” Exercises (session 8). Conversely, the three least viewed sessions were (1) Want to Join the Party Without Blowing Your Diet? (session 20), (2) Why Am I Hungry All the Time? (session 22), (3) Are Supplements Really Good for You? (session 23).
Figure 3. Page view distribution of weekly released diet and exercise sessions accessed by cancer survivors and partners (n=2736 total session page views). Top 3 most viewed sessions for the total sample: Get on Track for Success (session 2), Moving Towards Better Health (session 4), and Been Resisting “Resistance” Exercises (session 8). Bottom 3 least viewed sessions for the total sample: Want to Join the Party Without Blowing Your Diet (session 20), Why Am I Hungry All the Time (session 22), Are Supplements Really Good for You (session 23).
The Tools feature accounted for the third highest number of page views (n=967), following the Home page and Sessions, with participants most frequently viewing tools such as Sample Meal Plans (n=121 page views), Tracking with Fitbit (n=119), Exercise Logs (n=65), BMI Calculator (n=65), Fast Food Menu Maven (n=64), Calorie Calculator (n=62) and Calorie Burning Guide (n=60; ). The Healthy Weight, Healthy Eating, and Exercise features were also accessed throughout the intervention period. Within these features, the most commonly viewed content included “Common Questions About Weight Management” (n=20 page views) under the Healthy Weight category, “Increasing V&F Intake” (n=10 page views) under Healthy Eating, and “Leg Strengthening Exercises” (n=19 page views) under the Exercise category ()
Website Use and Behavioral Associations
Diet quality was positively associated with website use, weeks (r=0.50; P<.001), time (r=0.45; P<.001), total page views (r=0.46; P<.001), and sessions page views (r=0.39; P=.005). Self-reported MVPA was also positively associated with website use, weeks (r=0.37; P=.007), time (r=0.36; P=.009), total page views (r=0.36; P=.01), and sessions page views (r=0.35; P=.01). However, no statistically significant associations were detected for accelerometry-measured MVPA or weight (P>.05; ).
Table 4. Bivariate associations between website usability and Healthy Eating Index 2015 diet quality, moderate to vigorous physical activity (MVPA), and weight at 6-months. Bivariate associations were performed using Spearman partial correlations rank analysis for nonnormally distributed data. Adjusted for age, sex, race, and baseline outcome variables.
Measures
Weeks
Average time
Total page views
Session page views
r (95% CI)
P value
r (95% CI)
P value
r (95% CI)
P value
r (95% CI)
P value
Diet quality
0.50 (0.25 to 0.68)
<.001
0.45 (0.20 to 0.65)
<.001
0.46 (0.21 to 0.65)
<.001
0.39 (0.12 to 0.60)
.005
Self-reported MVPA
0.37 (0.10 to 0.58)
.007
0.36 (0.09 to 0.57)
.009
0.36 (0.09 to 0.57)
.01
0.35 (0.08 to 0.57)
.01
Accelerometer MVPA
0.15 (–0.16 to 0.44)
.33
0.06 (–0.25 to 0.37)
.68
0.03 (–0.28 to 0.34)
.82
0.04 (–0.27 to 0.35)
.79
Weight
–0.17 (–0.42 to 0.11)
.24
–0.06 (–0.32 to 0.22)
.67
–0.13 (–0.39 to 0.15)
.35
–0.11 (–0.38 to 0.16)
.41
Discussion
Primary Findings
There are very few web-based lifestyle interventions for cancer survivors and their support partners that use evidence-based theoretical constructs to promote healthful diet and physical activity behaviors []. As a result, no studies to date have described patterns of website use to inform the design and delivery of future dyadic, web-based lifestyle interventions. This study is among the first to address this gap by providing a detailed analysis of website use patterns among cancer survivors and their chosen partners randomized to the DUET intervention. Our findings suggest that cancer survivors and their partners used the DUET website to learn about healthy lifestyle guidelines, as reflected by website use metrics: on average, participants logged into roughly half of the 24-week content, interacting for a total of 313 minutes with the program, for which the highest page view activity was observed for Sessions (n=2736). Older adults and females engaged with the website to a significantly greater degree compared to younger participants and males, as reflected by weekly website activity (frequency, user time, and page views). Moreover, higher levels of website use were significantly associated with improvements in diet quality and self-reported MVPA. However, no significant associations were observed between website use and accelerometer-measured MVPA and weight.
Comparison With Previous Literature
Due to variability in how website usability metrics are reported, direct comparisons of website use across studies are challenging. However, findings from 3 pilot web-based lifestyle interventions implemented among cancer survivors generally align with our data that cancer survivors engage with a lifestyle website approximately once per week, particularly during the initial phase of the study. For example, the SurvivorSHINE study reported an average of 1.5 log ins per week over a 2-week period []. Similarly, in the A Lifestyle Intervention Via Email study, breast cancer survivors visited the website for an average of 9.6 out of 12 weeks in the physical activity arm and 10.7 out of 12 weeks in the diet arm []. A study by Blarigan and colleagues [] reported that colorectal cancer survivors in the intervention arm accessed the website on a median of 13 out of 84 days. Despite the 24-week DUET intervention being roughly twice as long as web-based interventions used in previous studies, we observed comparable weekly website use during the initial 12 weeks of the program, suggesting sustained and moderate website use among dyads during the initial phases of the intervention. Our findings also showed that dyads spent a total of 313 minutes on the website, which differs from the SurvivorSHINE study (94 minutes) but aligns closely with the Breast Cancer eHealth Self-Management study (337 minutes) [,]. The difference in website use observed in our study compared to SurvivorSHINE is likely due to the longer intervention duration and the inclusion of weekly serialized e-learning sessions and videos, which were not part of the SurvivorSHINE’s 2-week intervention.
This study also examined page view activity to assess participant usability with various DUET website features. The most frequently accessed feature was the interactive e-learning “Sessions,” followed by the “Home Page” feature. Prior systematic reviews have emphasized the importance of interactive educational modules to promote real-time engagement with behavior change content [-]. The DUET “Sessions” were developed with this framework in mind, incorporating brief, skill-building activities designed to help both survivors and their support partners apply evidence-based knowledge to everyday challenges. These sessions encouraged teamwork and joint problem-solving to support healthier behaviors, a strategy supported by the theory of communal coping and prior studies [,,]. Participants also received 3 weekly text messages reminding them to visit the website and engage with the weekly content, which likely contributed to the high usability observed with the “Sessions” feature. The brief, focused, and interactive nature of the sessions may have further enhanced their appeal. Similar to findings from the SurvivorSHINE study, the “Home Page” was also commonly viewed, likely because it automatically loaded each time participants accessed the website, allowing them to engage with core behavior change constructs such as self-monitoring, goal setting, and motivation [].
Consistent with prior research conducted in samples without a history of cancer, our findings indicate that website use was significantly higher among older adults (aged ≥65 years) and female participants. For example, Graham et al [] reported that older adults using the commercially available Lark Health digital platform logged more meals (174 vs 89) and used more self-monitoring devices (39 vs 28) compared to younger adults. Relatedly, a scoping review found that among adults aged 50 years and older, structured, tailored digital programs were well-accepted []. Similarly, a growing body of literature has consistently shown that women are more likely to seek health-related information, engage with online health platforms, and participate in digital lifestyle interventions compared to men [,,]. In our study, this trend may be amplified by the DUET program’s structured, self-paced design and tailored content, which likely resonated with older and female participants who may prefer individualized guidance and flexibility with web-based programs [,].
Bivariate analyses from this study showed that website use was significantly associated with improvements in diet quality and self-reported MVPA at 6 months. The evidence on whether website use improves diet quality remains unclear, largely due to heterogeneity in how diet quality is measured across studies. For instance, our findings differ from those of a systematic review and meta-analysis of 29 studies in adults with chronic conditions, which found no evidence for website use to improve overall diet quality []. However, most studies in this review were short in duration (<3 months) and relied on self-reported measures; in contrast, our study used interviewer-administered 24-hour dietary recalls, the gold standard for diet quality assessment, which may be more sensitive to detecting change. Our findings are more consistent with systematic reviews and several studies that have reported moderate to strong associations between website use and increased physical activity using self-reported measures [,-]. One possible reason the DUET website led to improvements in diet quality and self-reported MVPA is its inclusion of a range of relevant content for cancer survivors and their partners, such as guidance on V&F intake, reducing processed foods, portion control, setting SMART goals, and providing tools for tracking and self-monitoring diet and physical activity behaviors. However, it is important to note that diet quality and self-reported MVPA models were modestly significant, and may not withstand correction for multiple testing and should be interpreted as hypothesis-generating.
Despite these positive associations, our analysis did not find a significant association between website use and accelerometer-measured MVPA and weight. Our findings for not detecting significant associations with subjectively measured physical activity aligns with findings from previous studies [,], but, they contrast with findings from the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Total Wellbeing Diet Online program and the Weight Loss Maintenance (WLM) trial, which found that website use was associated with greater weight loss and reduced weight regain, respectively [,]. However, the CSIRO program was commercially delivered and relied on participant self-reported weights within the platform, while the WLM trial focused on weight maintenance rather than weight loss []. Several methodological, individual, and behavioral level considerations may help explain the lack of association between website use and accelerometer-measured MVPA and weight []. Our power calculation confirmed that with 56 participants, the study was powered only to detect correlations of approximately r≥0.40. Given that the association between website use and accelerometer-measured MVPA and weight observed in our data was substantially smaller (r<0.20), the study was likely underpowered to detect this relationship. Additionally, differences in how outcomes were measured may further clarify the findings. Website use, diet quality, and self-reported MVPA rely on participant interaction and reporting, which may naturally relate to one another and explain shared method variance. In contrast, accelerometer-measured MVPA and weight are objective measures that do not rely on participant reporting, and therefore their associations with website use may be smaller and more difficult to detect. Beyond methodological explanations, individual and behavioral factors may also play a role. For instance, website use may support behavior change but not be sufficient on its own to produce weight loss, as some participants may have relied less on the website once new habits were established. Additionally, weight loss may be influenced more by theoretical and behavioral mechanisms (ie, reduced perceived barriers, self-monitoring, calorie restriction, increased accountability from participating in the study, or lifestyle changes occurring outside the platform). These possibilities suggest that website use alone may not fully reflect the processes that contributed to weight loss in the parent trial [].
Strengths and Limitations
This study had several notable strengths. It examined website use among cancer survivors and their partners, an area with limited prior research, as few lifestyle interventions have studied website use among dyads. Our analysis also provided detailed page view analytics across the entire DUET website, offering insight into which website features were most frequently used. Furthermore, the study used validated measures to assess both diet quality and MVPA, enhancing the accuracy of the findings. However, like all studies, there were limitations. Most importantly, website data were only available for the 56 participants in the intervention arm, resulting in a relatively small analytic sample. A post hoc power calculation indicated that with this sample size, the study was powered (≥80%) only to detect correlations of approximately r≥0.40, representing a moderate to large effect size. As a result, smaller associations, particularly for accelerometer-measured MVPA and weight change, may not have been detectable, and the null findings for these outcomes should be interpreted with caution. Future web-based trials should consider recruiting larger samples to ensure adequate power to detect smaller associations between website use and clinically relevant outcomes such as weight change. Additionally, accelerometer-measured MVPA had 20% data missingness. As a result, complete-case analysis may introduce bias if participants with missing MVPA data differed meaningfully from those with complete data. Importantly, because multiple correlations were examined, there is an increased risk of type I error, and some associations may reflect spurious findings; therefore, the results should be interpreted within the exploratory, hypothesis-generating context of this secondary analysis. Similarly, it was not possible to determine whether participants viewed the full content of the weekly sessions they accessed, as data only represented whether participants accessed the sessions. Thus, it is likely that participants who briefly clicked into a session could therefore be coded similarly to those who reviewed the entire content, and time-stamp data may overestimate engagement if browser windows were left open. To partially address this limitation, we examined multiple measures of website use rather than relying on a single metric. Nevertheless, this measurement constraint may have led to imprecise estimates of website use and may help explain the modest strength of associations observed. Additionally, given the dyadic nature of the study, there may have been instances where dyads shared a single account and viewed website content together (as was documented in at least 1 case), potentially accounting for the difference between survivor and partner usage and underestimating individual-level usability. Another limitation is the demographic homogeneity of the sample, which primarily included female, non-Hispanic White breast cancer survivors residing in urban areas with higher socioeconomic status. As a result, these findings may not generalize to male survivors, racial and ethnic minority groups, individuals with lower income levels, or those living in rural or medically underserved settings.
Conclusions
Minimal-touch lifestyle interventions delivered through web-based platforms are being implemented to promote diet and physical activity behaviors for weight management among cancer survivors. However, patterns of website use are often understudied. Findings from our study suggest that cancer survivors and their partners (especially older adults and females) actively used the DUET website, particularly the interactive e-learning sessions, and higher levels of website use were significantly associated with improvements in diet quality and self-reported MVPA. However, no statistically significant improvements were detected for accelerometry-measured MVPA or weight. These results highlight that web-based platforms may serve as a promising, scalable approach for delivering diet and physical activity guidelines and promoting healthy behaviors for long-term older cancer survivors and their partners. However, larger, more diverse dyadic web-based lifestyle interventions, including male survivors, racial and ethnic minority populations, individuals with lower income, and survivors in rural or underserved areas, are needed to confirm and expand upon these findings. Importantly, future web-based lifestyle interventions should also use more detailed engagement tracking (ie, content completion indicators, differentiation between brief access and full interaction with embedded activities, and minimum and maximum time thresholds) to distinguish brief access from full content consumption to strengthen the validity of engagement measures.
First and foremost, we sincerely thank the cancer survivors and their partners who participated in the DUET intervention. We also gratefully acknowledge our funding sources: the American Institute for Cancer Research (585363); the American Cancer Society (CRP-19–175-06-COUN); the National Cancer Institute (P01 CA229997); O’Neal Comprehensive Cancer Center (P30 CA013148); the Cancer Prevention and Control Training Program (T32 CA047888); the Training to Reduce Burden across the Cancer Control Continuum multidisciplinary program (T32 CA251064-5); and the University of Arizona’s Comprehensive Cancer Center (P30 CA023074). Lastly, we extend our appreciation to the clinical staff and students (Drs Teri Hoenemeyer and Amber Kinsey, J Ryan Buckman, Grey Freylersythe, Lauren King, Doctorre McDade, Christopher Reid, Abigail Sims, and Fariha Tariq) for their dedication and efforts in the design, implementation, and successful completion of the study. No generative artificial intelligence was used in any part of the manuscript creation.
This work was funded by the American Institute for Cancer Research (grant 585363); the American Cancer Society (grant CRP-19–175-06-COUN); the National Cancer Institute (grant P01 CA229997); O’Neal Comprehensive Cancer Center (grant P30 CA013148); the Cancer Prevention and Control Training Program (grant T32 CA047888); the Training to Reduce Burden across the Cancer Control Continuum multidisciplinary program (grant T32 CA251064-5); and the University of Arizona’s Comprehensive Cancer Center (grant P30 CA023074). The funder was not involved in the conduct of the research, the decision to publish, or the approval of the publication.
The data reported in this study are available on request from the corresponding author.
None declared.
Edited by A Mavragani; submitted 31.Oct.2025; peer-reviewed by O Ibikunle, E Madondo, A Famotire; comments to author 25.Nov.2025; revised version received 07.Dec.2025; accepted 23.Dec.2025; published 30.Jan.2026.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research (ISSN 1438-8871), is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.
Good morning, and thank you for the invitation to join you today.1 It is a pleasure to be with you for the Southwestern Graduate School of Banking’s 161st Assembly for Bank Directors. Before we get started on the fireside chat, since the Federal Open Market Committee (FOMC) concluded its January meeting earlier this week, I think it would be helpful to summarize my views on the recent policy decision. I will then offer some remarks on the economy and also share my perspective on the outlook for monetary policy.
As we enter 2026, the economy has continued to grow, and I see inflation moving closer to our goal. But beneath the surface, the labor market is fragile. I will provide some perspective on why I think that fragility poses the greater risk and what that means for the path of policy.
Update on the Most Recent FOMC Meeting
At our FOMC meeting this week, my colleagues and I voted to hold the federal funds rate target range at 3-1/2 to 3-3/4 percent. Let me explain why I agreed to support this decision. I continue to see policy as moderately restrictive, and, looking ahead to 2026, my Summary of Economic Projections includes three cuts for this year. In my mind, the question at this meeting was about the timeline for implementing these cuts, essentially choosing between continuing to remove policy restraint and arriving at my estimate of neutral by the April meeting, or moving policy to neutral at a more measured pace throughout this year.
I do not consider downside risks to the employment side of our mandate to have diminished, and I see several indications that the labor market remains vulnerable. I could have voted in favor of continuing to remove policy restraint in order to hedge more against the risk of further labor market deterioration. But we have seen some signs of stabilization, and, after lowering the policy rate by a total of 75 basis points in the latter part of last year, in my view, we can afford to take time and “keep policy powder dry” for a little while in order to carefully assess how the lower degree of policy restraint is flowing through to broader financial conditions and strengthening the labor market. I am also reluctant to take meaningful signal from the latest data releases given the statistical noise introduced by the government shutdown. And, given that by the time of our March meeting we will have received two additional inflation and employment reports, I saw merit in waiting to take action.
It was not a straightforward decision. Ultimately, also considering that inflation remains somewhat elevated, at this meeting I decided to lean in favor of waiting for the upcoming sequence of data releases in order to gain more certainty about how the economy is likely to evolve in the coming months.
Current Economic Conditions
Since I presented a detailed assessment of economic conditions in a speech two weeks ago, today I will mention a few highlights and some new data points.2 As the flow of official economic reports has been normalizing, my views on the economy have not changed appreciably, in part because I am not taking much signal from the employment and prices data given increased measurement challenges in the wake of the government shutdown. The U.S. economy has been resilient and has continued to expand at a solid pace, but I remain concerned about fragility in the labor market. I am also confident that inflation will come down toward 2 percent as tariff effects on goods inflation continue to wane in coming months.
GDP growth strengthened in the third quarter of last year as consumer spending accelerated. However, growth likely slowed in the fourth quarter, reflecting the government shutdown and softer momentum in consumer spending, consistent with recent weakness in personal income. Disappointingly, residential investment seems on track to decline again in the fourth quarter.
Labor Market Conditions
Turning to the labor market, we have seen conditions gradually weaken over the past year, as unemployment rose and payroll employment flattened out. Private payroll employment growth slowed further to about 30,000 per month in the fourth quarter, and weekly ADP data show job gains remaining at a similarly subdued pace through early January, well below earlier last year and below the level necessary to keep unemployment stable.
Although the unemployment rate edged down to 4.4 percent in December and has moved sideways in recent months, it has increased by 1/4 percentage point since the middle of last year. Moreover, the Conference Board job availability index dropped sharply in January to its lowest value since early 2021, suggesting that the unemployment rate could move back up in the first quarter.
The labor market has become increasingly fragile over the past year and could continue to deteriorate in the near term. Despite some tentative signs of the unemployment rate leveling off, it seems too early to say that the labor market has stabilized, especially with the added statistical noise from the government shutdown and the sharp drop in the CPS survey response rate to below the pandemic lows. Job gains have been concentrated in just a few nonbusiness service industries that are less cyclically sensitive, with health care accounting for all private job gains last quarter.
With a less dynamic low-hiring, low-firing labor market, which some have said is giving rise to a “jobless expansion,” we could see layoffs rise quickly if firms begin to reassess their staffing needs in response to weaker activity. Although initial claims for unemployment insurance have remained low, private job cut announcements increased considerably last year, and there has been news of significant additional layoffs in January, as we heard this week from two large employers.
Inflation Developments
On inflation, we have seen considerable progress in lowering the underlying trend, considering that still-elevated inflation mostly reflects tariff effects on goods prices that I expect will fade this year. After removing these effects, core PCE inflation would have hovered close to 2 percent in recent months. The underlying trend in core PCE inflation appears to be moving much closer to our 2 percent target than is currently showing in the data.
Based on the latest consumer and producer price reports, 12‑month core PCE inflation likely stood at or a little below 3 percent in December, up somewhat since September. However, the Dallas and Cleveland Fed trimmed-mean measures of PCE and CPI price indexes suggest that 12-month core inflation has continued to decline. The discrepancy between these alternative measures of core inflation seems to reflect increased volatility in the recent data, with unusually large price increases in small categories, like software and video streaming, largely explaining the pickup in the core PCE inflation measure since September.
Outlook for the Economy
Looking ahead, my baseline expectation is that economic activity will continue to expand at a solid pace and the labor market will stabilize near full employment as monetary policy moves closer to a neutral setting. Less restrictive regulations, lower business taxes, and a more favorable business environment will continue to boost supply—largely due to higher productivity—and more than offset any negative effects on economic activity and inflation from other policies. I expect that the supply-side policies that I just mentioned, along with strength in AI-related investment, will continue to boost productivity gains and help ensure that inflation remains on a downward path.
The Path Forward
On the outlook for monetary policy, with inflation close to 2 percent, after excluding one-off tariff effects, and with unemployment near estimates of its natural rate but at risk of deteriorating, I continue to see policy as moderately restrictive. Downside risks to the labor market have not diminished, and we should not overemphasize the latest reading on the unemployment rate.
I appreciated and supported the language in the post-meeting statement about the recent data, which reflects an appropriate characterization of the unemployment data as showing “some signs of stabilization.”3 It will take time to get a clear signal about stability in the labor market. My view is that we should continue to focus on downside risks to our employment mandate, and the description of the labor market is helpful to communicate that we are not overly confident. History tells us that the labor market can appear to be stable right up until it isn’t.
As I think about the upcoming data, I am aware that first-quarter data tend to be more volatile. Therefore, in my view, we should not rely on these data as a reason to delay policy action if we see a sudden and significant deterioration in labor market conditions. We should also not immediately react if we see inflation go up in January, which has been common in recent years and could reflect residual seasonality or additional statistical noise from the government shutdown and ongoing measurement challenges.
I recognize and appreciate that other FOMC members may be concerned that inflation remains somewhat elevated and that we have not achieved our inflation goal for some time. However, absent a clear and sustained improvement in labor market conditions, we should be ready to adjust policy to bring it closer to neutral. We should also not imply that we expect to maintain the current stance of policy for an extended period of time because it would signal that we are not attentive to the risk that labor market conditions could deteriorate.
At the same time, it is important to remember that monetary policy is not on a preset course. At each FOMC meeting, my colleagues and I will evaluate incoming data, the evolving outlook, and the balance of risks to our dual-mandated goals of maximum employment and price stability. I will also continue to meet with a broad range of contacts to inform my assessment of economic conditions and the appropriate stance of policy.
Closing Thoughts
As the economy continues to evolve, policy must evolve with it. My focus will remain on acting early enough to preserve both price stability and a strong labor market. Thank you again for the invitation to share my views with you today. It is a pleasure to join you.
1. The views expressed here are my own and are not necessarily those of my colleagues on the Federal Reserve Board or the Federal Open Market Committee. Return to text
2. See Michelle W. Bowman (2026), “Outlook for the Economy and Monetary Policy (PDF),” remarks delivered at “Outlook 26: The New England Economic Forum,” Foxborough, Massachusetts, January 16. Return to text
3. See the January 2026 FOMC statement, which is available on the Board’s website at https://www.federalreserve.gov/monetarypolicy/fomccalendars.htm. Return to text
