Category: 3. Business

This article originally appeared on OncLive®.

Use of a sequential “sandwich” strategy using CD22/CD19 chimeric antigen receptor (CAR) T-cell therapy followed by autologous stem cell transplant (ASCT) resulted in deep and durable remissions in patients with newly diagnosed Philadelphia chromosome (Ph)–negative B-cell acute lymphoblastic leukemia (B-ALL) who were unable to undergo or declined allogeneic hematopoietic stem cell transplant (allo-HSCT), according to findings from a phase 2 single-center study (NCT05470777).¹

Findings published in Cancer demonstrated that at a median follow-up of 28 months (range, 10-50), evaluable patients (n = 37) achieved a median overall survival (OS) and leukemia-free survival (LFS) that had not yet been reached. The 2-year OS rate was 97% (95% CI, 90%-100%), and the 2-year LFS rate was 72% (95% CI, 58%-90%). MRD clearance deepened throughout treatment.

Following induction chemotherapy, 92% of patients experienced a complete remission (CR), 54% had a multiparameter flow cytometry minimal residual disease (MFC-MRD)–negative CR, and no patients achieved next-generation sequencing (NGS) MRD–negative CR. After consolidation chemotherapy, 80% of evaluable patients (n = 35) remained in CR, 71% had an MFC-MRD–negative CR, and 23% (n = 5 of 22) had an NGS-MRD–negative CR. After the first CD22/CD19 CAR T-cell therapy infusion, all patients achieved MFC-MRD–negative CR, and 68% achieved NGS-MRD–negative CR. Following ASCT and a second CAR T-cell infusion, all patients remained in MFC-MRD–negative CR, and 93% achieved NGS-MRD–negative CR (n = 25 of 27).

Among the 35 patients who completed the full treatment sequence, all remained alive at the last follow-up, and most sustained durable MRD-negative remissions beyond 1 and 2 years. Survival outcomes were comparable across standard- and high-risk genetic subgroups, and although patients with residual NGS-detectable disease prior to transplantation were associated with a trend toward shorter LFS, this difference was not statistically significant.

“The CD22/CD19 CAR T-cells and [ASCT] sandwich strategy is a promising approach for treating Ph-negative B-ALL in adolescent/young adult [AYA] and adult patients, offering high efficacy with a favorable safety profile,” lead study author Chong-Sheng Qian, MD, PhD, of the Research Center for Hematologic Diseases at the Jiangsu Institute of Hematology, The First Affiliated Hospital of Soochow University, in Suzhou, China, and colleagues wrote in the publication. “Future studies with larger sample sizes and longer follow-up are warranted to further validate these findings and further explore allo-HSCT–free strategies.”

What was the design of the study?

This investigation was a phase 2, single-arm, open-label study conducted at a single center and approved by the institutional review board of the First Affiliated Hospital of Soochow University.^1,2 The study enrolled newly diagnosed AYA and adult patients with Ph-negative B-ALL who had CD19 and CD22 expression by MFC. Eligible patients were either unable to undergo allo-HSCT or declined transplantation.

The primary end point was overall survival, measured from the first day of the initial CAR T-cell infusion. Secondary end points included LFS, MRD-negativity rate and duration, incidence of adverse effects (AEs) following first CAR T-cell therapy infusion, and non-relapse mortality.¹ Exploratory analyses assessed survival outcomes of the studied “sandwich” CAR T-based strategy in comparison with an external cohort of patients who received allo-HSCT.

According to the study protocol, patients received induction and consolidation therapy prior to CAR T-cell sequencing. After induction and lymphocyte recovery, peripheral blood lymphocytes were collected via leukapheresis to manufacture CAR T cells. CD22- and CD19-directed CAR T cells were infused sequentially at a dose of 5 × 10⁶ cells/kg as the first infusion. Autologous stem cell mobilization and collection occurred 6 to 8 weeks later, followed by conditioning with a modified BuCy26 regimen and ASCT. A second course of CD22/CD19 CAR T cells was administered 2 days after ASCT. No maintenance therapy was used following the second CAR T-cell therapy infusion, except tyrosine kinase inhibitors for patients with Ph-like ALL harboring ABL-class fusions.

Per protocol, patients with MRD progression after the first CAR T-cell therapy infusion discontinued the sandwich approach and were considered for allo-HSCT or individualized salvage therapy.

What were the baseline patient characteristics of those patients enrolled?

A total of 38 patients were screened for eligibility, of whom 37 were enrolled; one patient was excluded due to active hepatitis B. The median age at enrollment was 28 years (range, 15-60 years), and 35% of the cohort were older than 35 years of age. Elevated baseline disease burden was observed in a subset of patients, with 6 individuals (16%) presenting with a white blood cell count of more than 30 × 10⁹/L at diagnosis. Most patients (89%) had Ph-negative B-ALL, and 4 patients (11%) were classified as having Ph-like B-ALL, including 2 with ABL-class alterations and 2 with JAK-STAT pathway abnormalities. Based on National Comprehensive Cancer Network genetic risk stratification criteria, 21 patients (57%) were categorized as high-risk, including those with adverse genetic features such as TP53 mutations, complex karyotypes, or ZNF384 rearrangements.

Of the 37 enrolled patients, 35 successfully completed the full sandwich strategy. One patient with Ph-like B-ALL in the ABL class did not receive the protocol-specified TKI, representing a protocol deviation. Two patients experienced relapse following the first CAR T-cell therapy infusion and therefore did not proceed with the remainder of the sandwich approach; both subsequently underwent allo-HSCT.

What was the safety profile observed in the study?

The safety profile of the sequential CD22/CD19 CAR T-cell therapy and ASCT was consistent with expected toxicities of CAR T-cell therapy and high-intensity chemotherapy, with no unexpected signals. All patients experienced grade 3/4 hematologic toxicities, which reflected the treatment intensity; after the second CAR T-cell infusion, the median duration of neutropenia was 11 days, and thrombocytopenia lasted a median of 15 days.

Cytokine release syndrome (CRS) was generally mild: grade 1/2 CRS occurred in 22% of patients following the first CAR T-cell therapy infusion and in 34% of patients following the second infusion; no cases of grade 3 or higher CRS or immune effector cell–associated neurotoxicity syndrome were reported. No patients experienced severe organ toxicity. B-cell aplasia, a pharmacodynamic marker of CAR T-cell activity, was observed in all patients and persisted for a median of 170 days after the second infusion. Infectious complications included 2 cases of sepsis and 2 pulmonary infections, all of which were managed clinically. Importantly, no non-relapse mortality occurred.

References

1. Qian, C, Wang Z, Li Z, et al. A phase 2 trial of a “sandwich” strategy: sequential CD22/CD19 chimeric antigen receptor T‐cells therapy combined with autologous hematopoietic stem cell transplantation in patients with Philadelphia chromosome–negative B‐cell acute lymphoblastic leukemia. Cancer. 2025;131(22):e70168-e70168. doi10.1002/cncr.70168
2. CD22/CD19 CAR-T and auto-HSCT sandwich strategy as consolidation therapy for B-ALL. ClinicalTrials.gov Updated June 5, 2025. Accessed November 11, 2025. https://clinicaltrials.gov/study/NCT05470777

What’s going on here?

China and Hong Kong stocks are on track to end the week higher, powered by a sharp rebound in artificial intelligence (AI) and tech names, even as property developers and some consumer brands weigh on the rally.

What does this mean?

China’s CSI 300 and Shanghai Composite indexes both edged higher on Friday, extending their weekly gains, and Hong Kong’s Hang Seng Index is still up for the week despite a small daily dip. The real action has been in AI and tech: onshore AI-related stocks have surged about 6.5% this week after four straight weeks of losses, and Hong Kong–listed tech heavyweights have climbed nearly 4%. That suggests investors are leaning back toward growth and innovation plays as major indexes approach multi-year highs. Local broker Huaxi Securities expects that tilt to continue, projecting that by 2026 China’s market will be dominated by technology and high-dividend stocks – but also that higher indexes will come with sharper swings, making entry timing and technical signals more important. The rally still has weak spots, though: sportswear makers Anta Sports and Li Ning slipped after a Reuters report said they were among firms exploring a potential takeover of struggling German brand Puma, and state-backed developer Vanke’s Hong Kong shares fell nearly 2% to a record low on renewed debt-restructuring worries, echoed by softer bond prices.

Why should I care?

The bigger picture: Tech advances while traditional sectors struggle

Hong Kong’s market reflects a clear divide between digital winners and old-economy losers. The Hang Seng sits 33.34% higher than a year ago, though it’s slipped 1.46% over the past month. This week’s trading range—between 25,862 and 26,089—shows the market taking a breather after its big run. Tech and digital infrastructure companies attract buyer interest, while property and traditional sectors face headwinds. Long Forecast models suggest the index will trade sideways through 2028 before any meaningful breakout. Investors are voting with their wallets, backing companies with strong business models and avoiding those tied to yesterday’s economic playbook.

Zooming in: Markets wait for Beijing’s next move

The Hang Seng added just 18 points on November 27, trading in a tight range between 25,862 and 26,089 this week. Most investors are sitting on their hands until China’s Central Economic Work Conference in December. Beijing did announce plans to boost consumption—including rural consumer goods upgrades and support for pet-related sectors (yes, really)—but these moves barely rippled the market. Until the conference delivers concrete policy signals, expect more of the same quiet trading. The market’s essentially in wait-and-see mode, with any big moves likely on hold until Beijing shows its cards.

During 19th century gold rushes, few prospectors struck it rich, but merchants made real fortunes by selling essential tools for the job—giving rise to the term “pick and shovel businesses.”

Fast-forward, while today’s news headlines heralding the rise of artificial intelligence (AI) focus on chip innovation or software advances, an equally significant story unfolds in the background—massive infrastructure development that enables this technological revolution to even happen.

According to McKinsey research, a staggering $5.2 trillion in data center capacity expansion is required by 2030 to handle AI processing loads, far outstripping the additional $1.5 trillion needed for traditional IT applications.

While it may be too soon to pick long-term winners in the AI gold rush now, an immense investment opportunity lies with the modern ‘pick and shovel’ companies—design firms, contractors and others responsible for creating and powering the vast data centers that underpin AI and make up a group that is steady, essential and often overlooked.

Complex requirements

Data center construction is not new, but facilities now driven by the scope of AI demands are on an entirely different scale. These are often multiple times larger than legacy sites and consume exponentially more energy—in some cases enough to match the needs of a small city. The scale and technical sophistication of data centers now and the speed at which developers and users want them built is creating demand for companies that train and employ skilled engineers, craft workers. technologists and managers who understand and can execute complex requirements of these projects.

Cooling is a particularly critical challenge. AI processors generate immense heat, prompting the industry to move beyond traditional air cooling and toward liquid and immersion systems. Firms with expertise in providing these advanced solutions are seeing demand accelerate as new data centers come online.  

Hyperscalers also seek experts who can find large enough sites to accommodate these facilities, along with access to affordable and abundant power and water supply.  This has reshaped the geographic landscape and need for local construction resources and expertise. While Virginia was an early hub for data center development, the buildout has shifted to Southeast and Midwest states such as Texas, Georgia, Louisiana, Ohio and the Dakotas where project siting and permitting have fewer potential hurdles.

 

Tapping into the tailwind

Technology cycles can be volatile, and AI has experienced bouts of exuberance and uncertainty. But data centers’ significant need for up-front investment, land acquisition, long lead time equipment orders and multiple approvals have created a multi-year backlog of demand—which insulates contractors and suppliers from any short-term swings in sentiment.

This reality has not gone unnoticed by strategic and financial buyers who are increasingly drawn to construction sector firms with significant data center exposure. Stephens recently worked on the sale of a multi-generation, family-owned mechanical contractor. Before the current AI phenomenon, potential acquirers may have viewed the company’s exposure to new construction as a risk. But seeing a meaningful portion of its pipeline in data center work, private equity pounced with bids to purchase the company at a premium multiple to capitalize on this durable, multi-year tailwind.

Public markets are also rewarding firms making the pivot. One civil contractor also recognized the opportunity and acquired a private electrical contractor specializing in semiconductor and data center work. The buyer’s repositioning to gain exposure to one of the fastest-growing construction markets has been a driver of its share-price boom in recent years.

No one knows precisely how the AI ecosystem will evolve over the next decade. There will be winners, losers and inevitable hype cycles. But if the infrastructure to support this transformation is to be built, construction services businesses will remain in high demand to do the building. According to ConstructConnect, US data center starts—both AI and traditional—reached $26.9 billion over the first seven months of 2025. The figure is almost triple the amount for the same period in 2024 and nearly ten times greater than 2023.

Many construction sector companies capitalizing on the current surge in data center construction are privately held, multi-generational businesses with strong regional ties that have learned how to have staying power. Those legacies can make them especially attractive to acquirers and capital partners as the steady ‘pick and shovel’ providers in the new era of AI—but should also help them survive if AI’s current hyper-growth market of hyperscaling falls short of the current rosy predictions.

Nicholas V. Beare is a managing director in the investment banking division of New York City-based financial services firm Stephens. He heads its building and infrastructure group and co-heads its industrials group, also overseeing mergers & acquisitions, management buyouts and exclusive sales and divestitures, among other areas.

 

Introduction

The ethical foundation of the physician–patient relationship rests on the principles of privacy and confidentiality. These are not merely procedural concepts but essential conditions for building trust, which enables patients to disclose sensitive information vital for accurate diagnosis and treatment. Historically, this obligation has been central to medical professionalism, exemplified in the Hippocratic Oath, which mandates the safeguarding of patient secrets.^1–4

Although often used interchangeably, privacy refers to a patient’s right to control access to personal health information, while confidentiality concerns the provider’s ethical and legal duty to protect that information from unauthorized disclosure.^5,6 Adhering to both principles is imperative for ethical practice; violations can undermine patient trust, damage care quality, and result in legal and reputational consequences.

Modern healthcare environments, shaped by digital records and interconnected systems, amplify the risks of confidentiality breaches. While regulations such as HIPAA in the United States and GDPR in Europe provide legal protections, cultural and systemic variations influence implementation.^7–10 In Jordan and other Middle Eastern societies, family-centered cultural norms and collective decision-making traditions may influence how privacy and confidentiality are interpreted in medical encounters.^11–17 A regional survey also revealed insufficient physician knowledge of privacy standards, despite citing undergraduate training as their main source.¹⁸ Physicians and trainees sometimes face ethical tension between maintaining patient autonomy and responding to family expectations for information sharing, particularly when relatives are deeply involved in treatment decisions. Recognizing these sociocultural influences is essential to understanding how medical students in Jordan perceive and apply confidentiality principles within their clinical and educational environments.

Medical education plays a pivotal role in shaping ethical competence. However, studies from Saudi Arabia and Pakistan reveal that medical students often lack formal ethics training and struggle to apply confidentiality principles in complex settings.^19–21

Western studies, while showing higher knowledge levels, also report behavioral lapses, for example, chart access without consent in Spain and policy violations among US students.^22,23 These disparities underscore the influence of curriculum design and cultural context.

In Jordan, limited research has specifically examined medical students’ understanding and practices related to patient confidentiality and privacy. As the healthcare landscape evolves, there is a pressing need to assess how future physicians are prepared to uphold patient privacy. Although this study was conducted at a single institution, the participating medical school follows the national medical curriculum and accreditation standards, admits students from across Jordan, and is the second largest medical school in the country, making it broadly representative of undergraduate medical education in Jordan. Students in the 4th to 6th years were selected because they are engaged in clinical rotations and patient interactions where confidentiality issues are encountered directly, making them best positioned to provide informed perspectives on ethical knowledge and practice. This study addresses the knowledge-practice gap by evaluating the knowledge and self-reported practices of Jordanian medical students regarding patient confidentiality and privacy, with attention to demographic variables that may inform targeted ethics education.

Materials and Methods

A cross-sectional study was conducted among undergraduate medical students from a Jordanian medical school, aiming to assess their knowledge and self-reported practices regarding patient confidentiality and privacy, and to examine demographic and academic predictors of these practices. The study population comprised undergraduate medical students from one Jordanian medical school, selected to reflect typical training experiences in the region.

Sample and Participants

A convenience sampling approach was used, as all accessible medical students from the 4th–6th years of study were invited to participate through class WhatsApp groups. Participation was voluntary and anonymous, and informed consent was obtained electronically before accessing the questionnaire. Only currently enrolled medical students in the specified academic years were eligible to participate.

The required sample size was estimated using the formula for estimating a mean with a specified precision:

where Z is the standard normal value corresponding to a 95% confidence level (1.96), σ is the estimated population standard deviation of practice scores (assumed = 0.6 based on similar prior studies), and d is the acceptable margin of error (0.07). The calculated minimum sample size was approximately 282 participants, which was exceeded by the final number of valid responses (n = 297).

Study Tools

Data were collected using a structured, self-administered questionnaire that was divided into two main sections. The first section collected demographic information, including the participant’s age group (categorized as 20–24 years and 25–29 years), sex (male/female), and current academic year level (4th, 5th, or 6th year).

The second section consisted of 17 questions designed to evaluate students’ knowledge and self-reported practices concerning patient confidentiality and privacy. Responses to these questions were measured using a 5-point Likert scale, ranging from 1 (Strongly Disagree) to 5 (Strongly Agree).

Knowledge Domain

This domain included 10 questions (1, 2, 3, 4, 5, 6, 7, 8, 9, 10) assessing students’ understanding of the principles of confidentiality, legal implications of breaches, specific scenarios requiring consent (eg, disclosure to third parties, parents of adult patients, employers), and the importance of confidentiality for patient disclosure.

Practice Domain

This domain consisted of 7 questions (11, 12, 13, 14, 15, 16, 17) evaluating students’ self-reported behaviors related to maintaining confidentiality in clinical settings, for example, documenting information, discussing cases in public areas, handling phone calls, allowing non-medical personnel access, and dealing cautiously with sensitive patient information.

Reverse Scoring

To ensure that higher scores consistently reflect better knowledge or better practices, specific negatively worded questions were reverse-scored (original score subtracted from 6, assuming a 1–5 scale). The questions reverse-scored were: Q7 (knowledge domain) and Q12, Q13, Q14, Q16, Q17 (practice domain).

Domain Scores

For each participant, mean scores were calculated for the knowledge and practice domains by averaging the responses (on the 1–5 Likert scale, after reverse-scoring where applicable) to the questions within each respective domain. The domain scores were considered as the primary outcome variables for bivariate and multivariate analyses. Higher scores indicate better knowledge or better self-reported practices.

The questionnaire was adapted from previously validated instruments assessing medical students’ knowledge and practices regarding patient confidentiality and privacy.^18,24 To ensure contextual relevance for Jordan, the items were reviewed by three faculty members with expertise in medical ethics and medical education, who evaluated content clarity, cultural appropriateness, and face validity. Internal consistency was reassessed in the current sample using Cronbach’s alpha. The knowledge domain showed moderate reliability (α = 0.624; 95% CI 0.557–0.685), while the practice domain demonstrated good reliability (α = 0.767; 95% CI 0.725–0.806), indicating acceptable to good internal consistency for the scales used in subsequent analyses.

Data Collection

Data was collected using an electronic survey hosted on a secure Google Forms platform. The survey link was distributed via WhatsApp to accessible groups comprising 570 students. A total of 300 responses were received; three were excluded due to incomplete data, resulting in 297 valid responses for analysis. Completing the survey required approximately 8–10 minutes, and the data collection process was conducted over a four-week period, between November 15, 2024 and December 12, 2024.

Statistical Analysis

All statistical analyses were performed using SPSS version 23. The internal consistency reliability of the questions within each revised domain (knowledge and practice, using reverse-scored questions where applicable) was assessed using Cronbach’s alpha coefficient. Confidence intervals (95%) for the alpha values were also calculated. Categorical variables (age group, sex, year level) were presented as frequencies and percentages. For individual survey questions and the calculated domain scores, descriptive statistics including means and standard deviations (SD) were computed. Frequency distributions (percentages) for each response option were also recorded for the 17 questions.

Independent samples t-tests were used to compare the mean knowledge and practice scores between male and female students. One-way Analysis of Variance (ANOVA) was used to compare the mean domain scores across the different academic year levels and age groups. Post-hoc pairwise comparisons (Tukey Honestly Significant Difference, HSD) test was used to identify which specific groups differed significantly when a significant difference was found with ANOVA (p < 0.05). Multiple linear regression analyses were conducted to examine the simultaneous influence of demographic factors and knowledge scores on practice scores. Two separate regression models were built: 1. knowledge score as the dependent variable, with age group, sex, and year level as categorical predictor variables, and 2. practice score as the dependent variable, with knowledge score, age group, sex, and year level included as predictor variables. This model assessed the contribution of knowledge and demographics to self-reported practices. A p-value of less than 0.05 was considered statistically significant for all inferential tests.

The study was conducted in accordance with the Declaration of Helsinki, and the protocol received ethical approval from the Institutional Review Board at the Jordan University of Science and Technology (no. 25/176/2024). All participants consented to participate anonymously and voluntarily by ticking a statement before filling the questionnaire.

Results

Demographic Characteristics

A total of 297 medical students completed the survey. The vast majority of participants (97.3%, n=289) were in the 20–24 age group, and more than half (53.9%, n=160) were females. Regarding academic progression, half of the respondents (50.2%, n=149) were in their 5th year of medical school, while 29.6% (n=88) were in their 4th year, and 20.2% (n=60) were in their 6th year. Table 1 shows the demographic characteristics of participants.

Table 1 Demographic Characteristics of Participants. (n=297)

Descriptive Statistics for Survey Questions

Tables 2 and 3 present the questionnaire items separately for the two domains. Table 2 shows the frequency distribution and descriptive statistics of knowledge-domain items, while Table 3 shows the corresponding results for self-reported practice-domain items. Reverse-scoring was applied where appropriate in both domains. For the 10-item knowledge domain, the highest mean scores (indicating better knowledge) were observed for questions related to fundamental principles, such as “When reporting a case in a medical journal, do not disclose patients’ identity” (Mean=4.60, SD=0.58) and “Medical privacy is an implied term of contract between doctor and patient” (Mean=4.53, SD=0.53). The reverse-scored item Q7 (“In case of minor patients, disclosing the findings to parents or guardians constitutes a violation of confidentiality”) had a mean of 3.87 (SD=0.98), indicating that students generally disagreed with this incorrect statement, reflecting correct knowledge.

Table 2 Frequency Distribution and Descriptive Statistics of Knowledge-Domain Items (n = 297)

Table 3 Frequency Distribution and Descriptive Statistics of Practice-Domain Items (n = 297)

In the 7-item practice domain, the highest mean scores were for “I handle the information of patients with sensitive diseases (mental illnesses, sexually transmitted diseases, etc.) with more caution” (Mean=4.61, SD=0.56) and “Make sure that information taken from the patient is documented with complete confidentiality” (Mean=4.48, SD=0.54). The reverse-scored questions reflecting poor practices generally showed high means after reversal, indicating disagreement with those poor practices. For example, reverse-scored Q12 (“I discuss a patients’ medical condition with them in front of other patients to save time”) had a mean of 4.27 (SD=0.93), and reverse-scored Q13 (“I allow non-medical personnel (eg, cleaning staff) to enter the examination room while providing care to patients”) had a mean of 4.15 (SD=1.05). The lowest mean score in this domain was for the reverse-scored Q14 (“I discuss my patients’ conditions with my colleagues during work breaks”), with a mean of 2.84 (SD=0.97), suggesting this practice might be perceived as less problematic or more common compared to other breaches.

Domain Scores

Both domains showed relatively high mean scores. The knowledge domain had a mean score of 4.2 (SD=0.36), suggesting a good overall level of understanding regarding confidentiality principles. The practice domain also had a high mean score (Mean=4.060, SD=0.558), indicating that students generally reported engaging in practices consistent with maintaining patient confidentiality.

Bivariate Analysis

The results of the bivariate analysis comparing the domain scores across demographic characteristics are presented in Table 4. Female students reported significantly better confidentiality practices compared to male students, (4.166 vs 3.935, p=0.001, t-test). No significant sex difference was found in the knowledge domain (p=0.094), although females had a slightly higher mean score.

Table 4 Bivariate Analysis of Knowledge and Practice Scores Across Demographic Characteristics. (n=297)

Regarding year level, one-way ANOVA showed a statistically significant difference across year levels only in the practice domain (p<0.001). Post-hoc tests revealed that 5th-year students had significantly higher practice scores (indicating better practices) compared to 4th-year students (mean difference=0.326, p<0.001). No other significant pairwise differences were found between year levels (4th vs 6th, p=0.107; 5th vs 6th, p=0.201). There were no significant differences across year levels in the knowledge domain (p=0.139). No statistically significant differences were found between the two age groups (20–24 and 25–29 years) in either knowledge domain (p=0.358) or practice domain (p=0.902). Figure 1 shows the mean knowledge and practice scores regarding patient confidentiality and privacy by sex and academic year.

Figure 1 Mean knowledge and practice scores regarding patient confidentiality and privacy among medical students by sex and academic year (n = 297). Notes: *(p = 0.001), **(p < 0.001).

Multivariate Analysis

Table 5 shows the results of the multivariate linear regression analysis predicting practice score. After controlling for other factors, knowledge score was a significant positive predictor of practice Score (β=0.828, p<0.001), indicating that students with higher knowledge scores also tended to report better confidentiality practices.

Table 5 Multivariate Linear Regression Analysis Predicting Practice Score. (n=297)

Sex remained a significant predictor (β=−0.167 for male vs female, p=0.001), with male students having significantly lower practice scores (indicating poorer practices) compared to females, even after adjusting for knowledge and year level.

The year level also remained significant. Compared to 4th-year students (the reference group), both 5th-year students (β=0.250, p<0.001) and 6th-year students (β=0.158, p=0.037) had significantly higher practice scores, indicating better reported practices in later years of study.

Age group was not a significant predictor in the multivariate model (p=0.28).

Discussion

This study assessed medical students’ knowledge and self-reported practices regarding patient privacy and confidentiality in Jordan and interprets these findings in light of local educational and sociocultural factors. The generally good theoretical understanding and positive self-reported behavior indicate that students are exposed to sound ethical principles during training; however, the variations observed by sex and academic year reveal persistent challenges in applying these principles in real clinical contexts.

The high mean knowledge scores suggest that Jordanian medical students possess a solid conceptual grasp of confidentiality, implying that classroom instruction effectively conveys key ethical content. Nevertheless, the differences in practice scores highlight that theoretical understanding alone does not guarantee ethical conduct. This reinforces the idea that ethical competence develops through experiential learning, mentoring, and consistent exposure to professional role models rather than through didactic instruction alone.

The observed differences across sex and year level further enrich interpretation of these results. Female students’ higher practice scores may reflect greater empathic awareness and sensitivity to patient perspectives. While previous studies reported that female students exhibit higher knowledge and more positive attitudes toward patient privacy and confidentiality,²⁵ and better knowledge about confidentiality,^26,27 others suggest that male students may have better understanding of legal aspects of patient privacy.^28–30 These disparities are likely shaped by broader cultural and educational factors, rather than institutional factors, and may reflect gender-related communication styles and sociocultural expectations that emphasize empathy and relational awareness among female students. Addressing the sex gap in ethics requires targeted, sex-sensitive strategies. Such strategies could include structured mixed-group role-plays, mentorship that highlights diverse ethical perspectives, mentorship by role models who exemplify ethical behavior, and faculty workshops that address implicit bias and communication dynamics. Creating supportive environments where both male and female students can openly discuss ethical challenges may foster empathy and mutual understanding. Monitoring learning outcomes by sex and promoting female representation in ethics education and leadership can further enhance equity and inclusiveness in ethics education. Finally, further exploration of how sex influences ethical knowledge and behavior in medical training is warranted.

Differences in practice scores across academic years and improvement with academic progression likely stems from increased clinical exposure and socialization into professional norms, and aligns with previous research indicating that ethical competencies tend to develop with academic progression and increased clinical exposure.²⁸ These findings point to the value of integrating structured reflection and case-based learning throughout the curriculum, and support the value of continuous reinforcement and positive role modeling during clinical training. They also suggest areas for further investigation, including whether these differences persist after graduation, how specific teaching methods influence long-term ethical behavior, and which educational interventions most effectively foster lasting ethical behavior among students, particularly within the specific cultural context of Jordan.

Multivariate analysis further showed that higher knowledge scores significantly predicted better practice scores, reinforcing the importance of a strong theoretical foundation for ethical behavior. Sex and year level remained independent predictors, confirming their unique contributions to practice outcomes.

Cultural context plays a substantial role in shaping confidentiality norms. In Jordan’s collectivist society, families are deeply involved in healthcare decisions, often in ways that challenge individual privacy. A recent study found that only 48% of Jordanian physicians always obtain patient consent before sharing medical information.³¹ In contrast, over 80% routinely disclose patient details to family members when they are involved in care or when the patient is incapacitated. Such practices reflect social values that prioritize family solidarity and collective decision-making over strict individual autonomy. For students, these experiences create ambiguity about what constitutes an ethical or expected disclosure, particularly when family involvement is perceived as compassionate or obligatory. These norms may influence students to view such disclosures as acceptable or even expected professional behavior, especially when modeled by senior physicians or reinforced by family-centered expectations and familial involvement in treatment. While these actions may be well-intentioned, they can inadvertently compromise patient autonomy and confidentiality, underscoring the need for explicit ethical instruction that distinguishes between empathy and disclosure. To help students navigate this complexity, ethics instruction should include culturally relevant case discussions that contrast Western confidentiality standards with local expectations, allowing students to critically reflect on the moral reasoning behind each approach. This contrasts with Western frameworks that place greater emphasis on individual privacy. Cultural and religious values such as respect for elders and communal support further reinforce these tendencies. Consequently, a strictly Western ethics model may appear misaligned with local expectations. Ethics instruction in Jordan should explicitly address these cultural tensions, for example through role-play scenarios that simulate family pressure, and guided discussions on balancing confidentiality with family involvement. Emphasizing patient autonomy and trust can help students understand the ethical significance of confidentiality, even amid strong cultural expectations.

The study offers valuable insights into ethics education reform. The strong link between knowledge and practice underscores the need for robust, well-integrated ethics training. Addressing the observed knowledge and practice gaps requires curricular revisions that include small-group discussions of real-life privacy dilemmas, direct assessment of ethical behaviors, and mentorship by faculty and peers who visibly model ethical conduct in clinical settings, thereby cultivating a culture of ethics across the learning environment.^32–35 Ethical standards are most effectively transmitted when students observe them practiced routinely. When breaches of confidentiality occur without comment, students may perceive such behavior as tolerated or trivial. Conversely, when educators explicitly uphold patient privacy by lowering their voice in shared spaces, ensuring screens are not visible, or correcting improper disclosures, they demonstrate that ethics is an active component of professionalism. Institutional policies, such as mandatory confidentiality agreements and clear digital data-use protocols, can further reinforce these efforts.

This study has several limitations. Its cross-sectional design prevents establishing causal relationships. Reliance on self-reported practices may introduce social desirability bias and self-report biases, as responses may not always reflect actual clinical behavior; future research using objective observational methods could mitigate this. Additionally, the use of a sample from a single institution and recruitment through WhatsApp groups may have introduced sampling bias, as not all eligible students were reached, and self-selection bias, as participation was voluntary. These factors may limit the generalizability of findings; however, the results are consistent with patterns observed in other regional and international studies, which strengthens their validity.

Conclusion

This study demonstrates how knowledge, clinical exposure, sex, and cultural context shape medical students’ ethical behavior regarding patient confidentiality. The findings provide context-specific evidence from Jordan, underscoring the need for longitudinal, practice-based ethics education and faculty role-modeling. Enhancing ethical competence in confidentiality can strengthen patient trust, reduce breaches, and promote safer, more patient-centered healthcare delivery.

Ethics Approval and Informed Consent

This study was approved by the Institutional Review Board of the Jordan University of Science and Technology (no. 25/176/2024). All participants provided informed consent prior to participation. Participation was voluntary, and responses were anonymized to protect confidentiality.

Acknowledgments

The authors would like to thank the participating medical students for their time and valuable input.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This research received no funding.

Disclosure

The authors report no conflicts of interest in this work.

References

1. Patient-physician relationships | AMA-Code. Available from: https://code-medical-ethics.ama-assn.org/ethics-opinions/patient-physician-relationships. Accessed June 5, 2025.

2. Lerch SP, Hänggi R, Bussmann Y, et al. A model of contributors to a trusting patient-physician relationship: a critical review using a systematic search strategy. BMC Prim Care. 2024;25(1):194. doi:10.1186/s12875-024-02435-z

3. Hall M, Dugan E, Zheng B, Mishra A. Trust in physicians and medical institutions: what is it, can it be measured, and does it matter? Milbank Q. 2001;79(4):613–639. doi:10.1111/1468-0009.00223

4. Edelstein L. The Hippocratic Oath: Text, Translation and Interpretation. Johns Hopkins Press; 1943.

5. AMA Code of Medical Ethics. Privacy, confidentiality & medical records. Available from: https://code-medical-ethics.ama-assn.org/chapters/privacy-confidentiality-medical-records. Accessed November 26, 2025.

6. Beauchamp T, Childress J. Principles of Biomedical Ethics. Oxford University Press; 2019.

7. Cohen I, Mello M. HIPAA and protecting health information in the 21st century. JAMA. 2018;320(3):231–232. doi:10.1001/jama.2018.5630

8. Shachar C, Cadario R, Cohen IG, Morewedge CK. HIPAA is a misunderstood and inadequate tool for protecting medical data. Nat Med. 2023;29(8):1900–1902. doi:10.1038/s41591-023-02355-y

9. Shojaei P, Vlahu-Gjorgievska E, Chow YW. Security and privacy of technologies in health information systems: a systematic literature review. Computers. 2024;13(2):41. doi:10.3390/computers13020041

10. Siraj S, Hens K, Ali Y. Disclosure of true medical information: the case of Bangladesh. BMC Med Ethics. 2024;25(1):1–14. doi:10.1186/s12910-024-01115-y

11. Borgan S, Amarin J, Othman A, Suradi H, Qwaider Y. Truth disclosure practices of physicians in Jordan. J Bioeth Inq. 2018;15(1):81–87. doi:10.1007/s11673-018-9837-x

12. Al Kindi R, Al-Mamari H, Al Salmani A, et al. Sharing unpleasant health information with patients: a baseline study exploring physician attitudes and adherence to the SPIKES protocol. Sultan Qaboos Univ Med J. 2024;24(3):343–351.

13. Al-Mohaimeed A, Sharaf F. Breaking bad news issues: a survey among physicians. Oman Med J. 2013;28(1):20–25. doi:10.5001/omj.2013.05

14. Alahmad G, Hifnawy T, Abbasi B, Dierickx K. Attitudes toward medical and genetic confidentiality in the Saudi research biobank: an exploratory survey. Int J Med Inform. 2016;87:84–90. doi:10.1016/j.ijmedinf.2015.12.015

15. Borgan S, Amarin J, Suradi A, Haya QY, Qwaider YZ. Attitudes of physicians in Jordan toward non-disclosure of health information. Sultan Qaboos Univ Med J. 2021;21(3):423–427. doi:10.18295/squmj.4.2021.005

16. Al Zomia AS, AlHefdhi HA, Alqarni AM, et al. Examining Saudi physicians’ approaches to communicate bad news and bridging generational gaps. Healthcare. 2023;11(18):2528. doi:10.3390/healthcare11182528

17. Ibrahim ME, Adarmouch L, Elgamri A, et al. Researchers’ perspectives regarding ethical issues of biobank research in the Arab Region. Biopreserv Biobanking. 2024;22(2):98–109. doi:10.1089/bio.2022.0112

18. Karasneh R, Al-Mistarehi A, Al-Azzam S, et al. Physicians’ knowledge, perceptions, and attitudes related to patient confidentiality and data sharing. Int J Gen Med. 2021;14:721–731. doi:10.2147/IJGM.S301800

19. Darwish M, Sabra A. Medical ethics: knowledge about confidentiality among medical university students, Eastern Province Saudi Arabia. J Am Sci. 2012;8(6):299–304.

20. Murtaza M, Ali S, Shahbaz W, Elger B. Doctor-patient confidentiality: a cross-sectional study of opinions among medical students and doctors. Discoveries Reports. 2021;4:e16. doi:10.15190/drep.2021.1

21. Woodman A, Waheed KB, Rasheed M, Ahmad S. Current state of ethical challenges reported in Saudi Arabia: a systematic review & bibliometric analysis from 2010 to 2021. BMC Med Ethics. 2022;23(1):1–36. doi:10.1186/s12910-022-00816-6

22. Beltran-Aroca CM, Ruiz-Montero R, Labella F, Girela-Lopez E. The role of undergraduate medical students training in respect for patient confidentiality. BMC Med Educ. 2021;21(1):273. doi:10.1186/s12909-021-02689-6

23. Davis L, Domm J, Konikoff M, Miller R. Attitudes of first-year medical students toward the confidentiality of computerized patient records. J Am Med Inform Assoc. 1999;6(1):53–60. doi:10.1136/jamia.1999.0060053

24. Parmar P, Patond S, Rathod G, Ninave S. Awareness among intern doctors regarding privacy and confidentiality in medical practice. Indian J Med Forensic Med Toxicol. 2020;14(3):534–539. doi:10.37506/ijfmt.v14i3.10420

25. Hosseini-Ghavam-Abad L, Asghari F, Bandehagh A, Najafipour S, Bigdeli S. Patient privacy: awareness and attitudes of Iran university of medical sciences medical students. Med J Islam Repub Iran. 2019;33:12. doi:10.34171/mjiri.33.12

26. Padmanaban H, Simiyon M, Thilakan P, Ravichandran K. Knowledge about confidentiality among medical students from a tertiary care teaching hospital in South India. Indian J Psychiatr. 2020;62(6):749–750. doi:10.4103/psychiatry.IndianJPsychiatry_230_19

27. Dos Lima SMF, da S Silva SMM, Neves NMBC, Crisostomo LML. Assessment of medical students’ knowledge about medical confidentiality. Rev Bioét. 2020;28:98–110. doi:10.1590/1983-80422020281372

28. Terán AÁ, Palazuelos C, Dierssen-Sotos T, Alonso-Molero J, Llorca J, Gómez-Acebo I. Evolution of medical students’ perception of the patient’s right to privacy. Int J Environ Res Public Health. 2022;19(17):11067. doi:10.3390/ijerph191711067

29. Rani V. Knowledge and attitude about medical ethics among undergraduate medical students – A cross sectional comparative study. Indian J Health Sci Biomed Res Kleu. 2022;15(2):131. doi:10.4103/kleuhsj.kleuhsj_13_21

30. Soares AAS, Batista CB. Confidentiality in physician-patient relation: the knowledge of medical students. Rev Bioét. 2023;31:e3492PT. doi:10.1590/1983-803420233492EN

31. Karasneh R, Al-Azzam S, Nusair M, Al-Mistarehi A, Aldeyab M, Massad I. Confidentiality and family involvement in healthcare: a mixed-method approach of physicians’ perspectives in Jordan. BMC Med Ethics. 2025;26(1):52. doi:10.1186/s12910-025-01213-5

32. AlMahmoud T, Hashim M, Elzubeir M, Branicki F. Ethics teaching in a medical education environment: preferences for diversity of learning and assessment methods. Med Educ Online. 2017;22(1):1328257. doi:10.1080/10872981.2017.1328257

33. Ali A, Abodunrin A, Al Khayyat S, Novakovic D, O’Connor N, Hussein G. Medical school curriculum relating to clinical ethical decision making during a pandemic: a scoping review. J Med Educ Curric Develop. 2024;11:23821205241240610. doi:10.1177/23821205241240610

34. Jaramillo Villegas C, Hernandez Rincon EH, Gallego Beltran JC, Guzman Sabogal YR. Assessment of professionalism in undergraduate and graduate medical education: a scoping review. BMC Med Educ. 2025;25(1):1–16. doi:10.1186/s12909-025-07756-w

35. Bashir A, McTaggart IJ. Importance of faculty role modelling for teaching professionalism to medical students: individual versus institutional responsibility. J Taibah Univ Sci. 2022;17(1):112–119. doi:10.1016/j.jtumed.2021.06.009

By Philip van Doorn and Emily Bary

The iPhone maker leads the way for the sector in terms of dollars spent on stock buybacks. But other tech companies have reduced their share counts significantly, as well.

Under the leadership of Tim Cook, Apple has spent more than $708 billion buying back its own stock, which has reduced the company’s share count by 34.6% and provided a considerable boost to earnings per share.

Just 20 companies in the S&P 500 accounted for more than half of all the dollars spent on stock buybacks in the second quarter, according to research from DataTrek co-founder Nicholas Colas. But that doesn’t mean big companies are all getting the same bang for their buck, or that they are delivering for their investors.

Companies often brag about “returning capital” to their shareholders when they repurchase stock. But this isn’t really the case. If a company buys back shares on the open market, it is spending its owners’ money to purchase stock from people who no longer wish to be shareholders.

But buybacks can benefit long-term stockholders if the share count is reduced. A company’s diluted common-share count is the number used to calculate its earnings per share.

The share count will increase when newly created stock is shoveled to executives as part of their compensation or if shares are issued to help pay for acquisitions. The share count will decline if the company spends cash to repurchase stock.

So what really matters to investors is net buybacks – those that lower the share count and increase earnings per share. And there is a compounding effect. Here is an example of how the math works:

— A company’s profit is $1,000.

— There are 100 shares.

— That makes for $10 in earnings per share.

What if the share count had been reduced by 10%?

— The company’s profit would still be $1,000.

— There would be 90 shares.

— EPS would be $11.11.

— EPS would have increased 11%.

Using similar math, the 34.6% reduction in Apple Inc.’s (AAPL) share count over the past 10 years has led to a 53% increase in earnings per share, all other things being equal.

Screening technology stocks for net buybacks

These 15 tech companies have reduced their diluted common-share counts by at least 20% over the past 40 reported fiscal quarters.

   Company                                Ticker   10-year change in share count  5-year change in share count  Year-over-year change in share count  Total spent on buybacks over the past 10 years ($mil) 
   HP Inc.                               HPQ                              -47.9%                        -30.1%                                 -2.7%                                                $24,197 
   Jabil Inc.                            JBL                              -44.5%                        -34.0%                                 -6.4%                                                 $6,874 
   Apple Inc.                            AAPL                             -34.6%                        -13.9%                                 -2.5%                                               $708,713 
   Applied Materials Inc.                AMAT                             -32.9%                        -13.3%                                 -4.0%                                                $30,426 
   NetApp Inc.                           NTAP                             -31.8%                         -9.8%                                 -3.8%                                                 $8,437 
   Qualcomm Inc.                         QCOM                             -31.5%                         -5.9%                                 -4.6%                                                $50,033 
   VeriSign Inc.                         VRSN                             -28.9%                        -18.5%                                 -3.8%                                                 $8,105 
   Lam Research Corp.                    LRCX                             -27.2%                        -13.8%                                 -2.7%                                                $23,736 
   TE Connectivity PLC                   TEL                              -26.9%                        -10.0%                                 -3.3%                                                $12,491 
   Fair Isaac Corp.                      FICO                             -25.7%                        -19.0%                                 -3.1%                                                 $5,528 
   Teradyne Inc.                         TER                              -24.9%                        -13.7%                                 -3.1%                                                 $4,010 
   Skyworks Solutions Inc.               SWKS                             -23.2%                        -11.3%                                 -7.5%                                                 $5,440 
   CDW Corp.                             CDW                              -22.9%                         -9.0%                                 -2.3%                                                 $5,448 
   Cisco Systems Inc.                    CSCO                             -21.9%                         -5.9%                                 -0.5%                                                $72,299 
   Cognizant Technology Solutions Corp.  CTSH                             -20.5%                        -10.3%                                 -1.8%                                                $11,233 
                                                                                                                                                                                               Source: LSEG 

For this tech-stock screen, we began with the information-technology sector of the S&P 500 SPX. Then we added the 12 stocks in the Nasdaq-100 Technology Index XX:NDXT that aren’t in the S&P 500 IT sector, including Meta Platforms Inc. (META) and Alphabet Inc. (GOOGL), for an initial list of 82 stocks.

Then we screened the list as follows:

— IPO date had to be at least 10 years ago. This brought the list down to 70 companies.

— Quarterly average diluted share counts used to calculate EPS had to have been reduced for the most recent one-year, five-year and 10-year periods. This cut the list to 32 companies.

More from MarketWatch: Is the ‘Magnificent Seven’ over? Focus on these three stocks in particular.

Exploring Apple

No company has spent as much as Apple to buy back stock. The company’s diluted share count has declined by 34.6% over the past 10 years, as the company has spent $708.7 billion on buybacks.

During its most recent reported fiscal quarter, which ended Sept. 27, Apple spent $20.1 billion on stock buybacks. For the iPhone maker’s past four reported fiscal quarters, it has spent $90.7 billion to repurchase shares.

That $90.7 billion number is notable because it comes as peers have allocated similar amounts toward capital expenditures for their artificial-intelligence buildouts, all while Apple has been measured in its AI spending.

Apple’s divergent path has become somewhat controversial on Wall Street, with some investors worried the company has fallen behind on AI because of underinvestment.

“I’m old enough to remember a year and a half ago when I was reading all these glowing stories about Apple buying back all these shares and what it was doing for its share count over time,” Seaport Research analyst Jay Goldberg told MarketWatch. “I don’t think people thought there was much to invest in.”

But now times are different, and, until recently, rivals got rewarded almost uniformly for boosting their spending forecasts in a race to compete.

“Should Apple be doing more in AI?” Goldberg asked. “Yes. Should they be cutting back their share buyback to fund that? I’m OK with them not doing that until they actually know what they want to do with AI.”

Don’t miss: These two ‘Magnificent Seven’ stocks could be the strongest survivors of an AI apocalypse

But Alexander Laskin, a Quinnipiac University professor who focuses on public and investor relations, took a more negative view of Apple’s buyback strategy. “Steve Jobs was famously opposed to paying dividends or buying back stock, arguing that the money was better spent on making the next big thing,” he told MarketWatch.

He thinks a heavy focus on buybacks signals that a company “simply doesn’t have great, high-return ideas for how to spend that money.” Over the long run, that “likely caps the firm’s potential for truly massive growth.”

Apple finds itself in a difficult spot from a messaging perspective, he noted. The company faces “an important communication challenge because its AI progress trails its competitors and not investing in AI now may in fact jeopardize the long-term future of Apple.”

That said, investors have softened their views in recent weeks. While Apple’s stock lagged many of its “Magnificent Seven” peers for much of the year, it has outperformed all but Alphabet Inc.’s stock over the past month, reflecting increased scrutiny of capital expenditures elsewhere in the technology world.

It remains to be seen whether Wall Street will opt to reward discipline or flip back to AI-at-all-costs sentiment.

See more: Why Apple’s stock is beating the market even as the tech sector sells off

-Philip van Doorn -Emily Bary

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

  (END) Dow Jones Newswires

  11-28-25 0934ET
Copyright (c) 2025 Dow Jones & Company, Inc.

Asda has criticised the government for “killing confidence” among consumers but blamed “self-inflicted” problems that left gaps on shelves for a big reverse in sales.

Total sales at the UK’s third-largest supermarket fell 3.8% to £5.1bn in the three months to the end of September compared with the same period a year before – diving back from 0.2% growth in the previous quarter. Comparable store sales fell 2.8%.

The company said it had struggled with technology problems from a lengthy effort to separate its IT systems from those of its former owner, Walmart.

Allan Leighton, the chair of Asda, returned to the business last year to try to revive the business for a second time. He said the supermarket’s fall in sales and market share was “totally self-inflicted” and had put his turnaround plans “back six months”.

However, Leighton also hit out at the government for hindering growth and depressing consumer sentiment.

“The country is stuck in reverse,” he said. “They have got to encourage business to invest and they keep hampering that with costs.”

He said in order to “sell more stuff” retailers needed “a positive consumer” and the government was “killing consumer confidence because of the fact there is no growth and nobody is investing”.

“The government isn’t doing anything to stimulate growth,” he added. Leighton said the impact of feared business rates changes for large retailers in Wednesday’s budget would be “neutral” for Asda.

Official figures last week showed that sales at UK retailers slumped unexpectedly last month as shoppers waited for Black Friday deals, and uncertainty over the budget dampened consumer confidence.

Asda’s sales and profits have dived since a debt-fuelled £6.8bn takeover in early 2021 by Blackburn’s billionaire Issa brothers and the private equity company TDR Capital. Aldi is poised to overtake it as the UK’s third-largest supermarket, according to analysts at Worldpanel by Numerator, formerly known as Kantar.

Leighton said problems with its new IT systems had caused availability problems across the business, with clothing and homewares supplied to more than a quarter of its stores hit by problems at a specialist distribution centre. A new grocery home shopping app had proved “more clunky than what we had before”, putting customers off, he admitted.

skip past newsletter promotion

after newsletter promotion

Leighton said the retailer had “made good progress” on fixing the IT issues, and availability was back up to appropriate levels.

“It’s all behind us now,” he said. “We are pretty confident in our strategy.”

Asda plans to continue its strategy of price cuts. Leighton said “the competition didn’t double down on price” in response to its well-publicised investments in keeping prices down for shoppers.

“Our price position is [improving],” he said, claiming Asda was now between 4% and 7% cheaper than its “major competitive set”, such as Tesco, Asda and Morrisons.