DocG DocG | AI-Powered Clinical Summaries
Sign In

Stay Current with Medical Literature

AI-powered clinical summaries delivered to your inbox 3 times per week. Evidence-based insights in 300 words, designed for busy physicians.

Sign Up

See What You Get

Subscribers receive email summaries and access to our interactive AI-powered article rankings.

Email Summary Example

Moderate caffeinated coffee or tea linked to less dementia
Higher long-term intake of caffeinated coffee and tea was associated with lower dementia risk and slightly better thinking test scores, while decaffeinated coffee was not.
*Prospective cohort study; Level 2b (OCEBM).

Citation

Zhang Y, Liu Y, Li Y, et al. Coffee and tea intake, dementia risk, and cognitive function. JAMA. 2026;335(11):961-974. doi:10.1001/jama.2025.27259.

Background

Prior studies on coffee and tea and brain health have been mixed, and many did not separate caffeinated from decaffeinated coffee. This study examined long-term beverage intake and later dementia and cognitive outcomes.

Patients

131,821 US health professionals (mostly White; 65.7% women) without cancer, Parkinson disease, or dementia at study entry; diet assessed repeatedly.

Intervention

Higher habitual intake of caffeinated coffee and tea (also evaluated decaffeinated coffee).

Control

Lower intake groups.

Outcome

Primary: incident dementia. Secondary: self-reported cognitive decline; telephone-based thinking tests (women’s cohort only).

Follow-up Period

Up to 43 years (median 36.8 years).

Results

Exposure (highest vs lowest) Dementia rate (per 100,000 person-years) Relative association with dementia Self-reported cognitive decline (%) Relative association with self-reported decline NNT for self-reported decline
Caffeinated coffee 141 vs 330 Hazard ratio 0.82 (95% CI, 0.76 to 0.89) (primary) 7.8% vs 9.5% Prevalence ratio 0.85 (95% CI, 0.78 to 0.93) ≈59
Tea 201 vs 321 Hazard ratio 0.86 (95% CI, 0.83 to 0.90) (primary) 8.1% vs 9.5% Prevalence ratio 0.86 (95% CI, 0.80 to 0.93) ≈71
NNT = number needed to treat, calculated from absolute percentage differences; these are observational associations, not proven treatment effects.
In the women’s cohort, higher caffeinated coffee intake was linked to a small increase in Telephone Interview for Cognitive Status score (mean difference 0.11; 95% CI, 0.01 to 0.21). Tea intake was linked to small improvements across several test scores. Decaffeinated coffee was not linked to lower dementia risk.
Most pronounced associations were at about 2–3 cups/day of caffeinated coffee or 1–2 cups/day of tea.

Limitations

Because this was an observational study, it cannot prove cause and effect; unmeasured lifestyle or health factors could partly explain results. Beverage intake was self-reported, and dementia identification relied partly on records and reported diagnoses. Changes in thinking test scores were small and may not be noticeable to individual patients. Participants were mostly health professionals, limiting generalizability.

Funding

US National Institutes of Health; no reported role in study conduct.

Clinical Application

For patients who already drink these beverages, moderate caffeinated coffee or tea intake may be reasonable; do not recommend starting solely to prevent dementia based on this study.

Top Journal Rankings - May 2026

17 abstracts scored across 7 criteria. Click any article to expand criterion scores.
1. 6.8
E-cigarette switching, smoking cessation, and the risk of hepatocellular carcinoma in patients with chronic hepatitis B: A nationwide cohort study in South Korea.
Overall: A very large national cohort links quitting and e-cigarette switching with lower HCC incidence in CHB, but residual confounding and limited data on harms/absolute benefit mean it mainly supports standard cessation counseling rather than changing practice.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 7.5
Smoking cessation counseling is a common primary-care task, and chronic hepatitis B patients are frequently co-managed with primary care; however, HCC risk management in CHB often involves specialty hepatology workflows and surveillance.
Validity, Bias Control & Precision 6.5
Large nationwide retrospective cohort with multivariable adjustment and fairly tight confidence intervals, but nonrandomized exposure groups (quitting vs switching vs continued smoking) are highly susceptible to residual confounding and exposure misclassification; limited detail on how behaviors were measured and controlled over time.
Patient-Oriented Outcomes 8.5
Incident hepatocellular carcinoma is a clearly patient-important clinical outcome (hard endpoint), not merely a surrogate marker.
Magnitude of Net Benefit 6.0
Associations suggest moderate relative risk reductions (HR ~0.78 for quitting or switching; stronger for persistent quitting HR ~0.64), but absolute risk differences, adverse effects (including potential harms of e-cigarettes), and patient burden are not quantified in the abstract.
Implementability & Practicality 7.0
Counseling for complete cessation is readily actionable, while recommending e-cigarette switching is more complex given product variability and the need to address dual use and follow-through; the abstract notes low transition from switching to quitting.
Practice-Changing Potential 5.5
Findings generally reinforce existing cessation-first messaging rather than clearly establishing a superior new strategy; observational design and absence of absolute effects/harms limit immediate practice change regarding endorsing e-cigarette switching.
2. 6.8
Prospective Associations Between Early Adolescent Problematic Screen Use, Mental Health, Sleep, and Substance Use.
Overall: A large prospective cohort links problematic screen use at ages 11–12 with multiple clinically important outcomes one year later, making it highly relevant to primary care counseling and assessment. However, observational design, self-reported measures, and lack of effect sizes/precision metrics in the abstract limit certainty and prevent estimating clinical magnitude or net benefit. It is practical to incorporate screening questions and a family media plan, but evidence that such actions improv
View 6 Criterion Scores
Primary-Care Relevance & Applicability 8.0
Addresses common primary-care concerns in early adolescents (mental health, sleep, substance initiation) and suggests clinician actions (assessing problematic screen use, family media plan), though it is observational rather than a tested primary-care intervention.
Validity, Bias Control & Precision 6.0
Large, prospective national cohort with confounder adjustment supports credibility, but as an observational study it remains vulnerable to residual confounding and self-report bias; no effect sizes or confidence intervals are provided in the abstract to judge precision.
Patient-Oriented Outcomes 8.0
Outcomes include depressive and behavioral symptoms, suicidal behaviors, sleep disturbance, and substance use initiation, which are patient-important and clinically meaningful in adolescent care.
Magnitude of Net Benefit 3.0
The abstract reports associations without absolute risks, effect sizes, or any intervention; therefore the clinical magnitude and net benefit cannot be estimated from the abstract.
Implementability & Practicality 7.0
Screening for problematic use via questionnaires and counseling on a family media plan are generally feasible in outpatient settings, but the abstract does not specify workflow, time burden, or evidence that acting on these assessments improves outcomes.
Practice-Changing Potential 9.0
Moves beyond screen time to 'problematic' (addiction-like) use and links it prospectively to multiple important outcomes in a large cohort, providing a clear, actionable framing that could change how clinicians assess and counsel families, despite lacking interventional proof.
3. 6.8
Polypharmacy as a Simple Measure for Assessing the Risk of Fall-Related Hospitalization in Older Adults.
Overall: A very large cohort links simple medication-based measures to subsequent fall-related hospitalization with moderate predictive performance, but as observational prognostic work (not an intervention) it offers limited evidence of downstream outcome improvement despite being highly practical for primary-care risk flagging.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 8.0
Falls risk and medication review in adults >66 are common primary-care concerns, and the predictors (polypharmacy, FRIDs, PIMs) map to decisions generalists routinely make.
Validity, Bias Control & Precision 7.0
Very large population-based cohort with time-to-event modeling and reported CIs supports precision, but this is observational claims-based work with potential misclassification and residual confounding; calibration/decision-utility details are not provided.
Patient-Oriented Outcomes 8.0
Fall-related hospitalization is a clearly patient-important outcome (hard clinical endpoint), not just a surrogate marker.
Magnitude of Net Benefit 4.5
The study evaluates prediction rather than testing an intervention that reduces falls; discrimination is only moderate (c-statistics ~0.72–0.74) and differences between measures are small, so the practical incremental benefit is unclear from the abstract.
Implementability & Practicality 9.0
Polypharmacy counts and identification of fall-risk increasing drugs can be done easily from medication lists/claims with low added burden and no special equipment.
Practice-Changing Potential 4.5
Findings largely reinforce existing clinical intuition that more medications/FRIDs relate to falls; without demonstrated impact on changing prescribing or reducing hospitalizations, it is more supportive for surveillance/risk flagging than immediately practice-changing.
4. 6.4
Effectiveness and cost-effectiveness of a parenting programme to improve family wellbeing in England (TOGETHER): a multicentre, single-blind, randomised controlled trial.
Overall: A well-conducted multicentre RCT suggests a group parenting programme modestly improves parent mental wellbeing in disadvantaged, diverse communities, with limited safety concerns but notable attrition and substantial delivery burden that may temper immediate practice impact.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 7.0
Addresses parental mental wellbeing and family functioning in disadvantaged communities—highly relevant to primary care and social prescribing/referral pathways, though delivery is community-group based rather than a typical clinic intervention.
Validity, Bias Control & Precision 7.5
Multicentre randomized trial with stratified allocation, masked outcome collection/analysis, intention-to-treat analysis, and prospective registration; however, participants were unblinded and 30% attrition at 6 months raises risk of bias.
Patient-Oriented Outcomes 8.0
Primary outcome is parent-reported mental wellbeing (Warwick-Edinburgh scale), which is directly patient-important, measured at post-intervention and 6-month follow-up.
Magnitude of Net Benefit 5.5
Shows statistically significant but modest mean improvements (≈1.7–1.9 points) with no intervention-related adverse events reported; however, the clinical importance of the score change and the net value given the time/cost burden are not clearly established in the abstract.
Implementability & Practicality 5.0
Requires weekly 3-hour group sessions over 13 weeks and cost about £1081 per attendee, implying staffing, space, and participant time commitments that may limit routine uptake despite feasibility at scale in community services.
Practice-Changing Potential 5.5
Provides RCT evidence supporting referral/commissioning of a structured parenting programme in deprived, diverse populations, but the modest effect size and substantial delivery burden make immediate widespread practice change less certain from the abstract alone.
5. 5.9
Associations of pregnancy complications with paternal cardiovascular risk: a retrospective cohort study.
Overall: A large, precise observational study links cumulative maternal pregnancy complications with increased paternal cardiometabolic risk, relevant for primary-care risk stratification but limited by confounding and lack of evidence that acting on these associations improves patient outcomes.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 7.5
Findings could inform primary-care risk assessment/screening for common conditions (diabetes, hypertension, CVD) in men using partner pregnancy history, though integration into routine workflows is not fully specified.
Validity, Bias Control & Precision 6.5
Very large, population-based retrospective cohort with time-to-event modeling and tight confidence intervals, but observational design using administrative data leaves substantial risk of residual confounding and misclassification.
Patient-Oriented Outcomes 7.5
Outcomes include incident diabetes, hypertension, and cardiovascular disease events, which are clinically meaningful to patients (though diagnosis-based endpoints may vary in patient impact compared with hard outcomes).
Magnitude of Net Benefit 3.5
Effect sizes are modest to moderate (e.g., HRs ~1.11–1.84) and this is not an intervention study, so there is no direct evidence of improved outcomes, harms, or burden from acting on the information.
Implementability & Practicality 6.0
Asking about partner pregnancy complications is low-tech, but may be difficult to capture accurately (especially years later) and the abstract does not specify a clear, implementable management pathway once risk is identified.
Practice-Changing Potential 4.5
Suggests a potentially useful risk marker, but without evidence that using this history changes management or outcomes, it is more hypothesis-generating than practice-changing.
6. 5.4
Associations of Adolescent BMI and Physical Fitness With Cardiovascular Health in Middle Age: A Population-Based Prospective Study of Swedish Men.
Overall: A large prospective cohort suggests adolescent obesity and low cardiorespiratory fitness predict worse midlife cardiovascular health scores, but observational design and non-hard endpoints limit actionable, practice-changing conclusions.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 6.5
Addresses long-term cardiovascular prevention and modifiable factors (adolescent BMI and fitness), which are conceptually relevant to primary care, but the cohort is limited to Swedish men and the exposure measurement setting (military conscription testing) is not typical of primary-care workflows.
Validity, Bias Control & Precision 6.0
Large, population-based prospective cohort with decades of follow-up and objective baseline fitness measures strengthens credibility, but it remains observational with potential residual confounding; the abstract provides limited quantitative effect sizes and no confidence intervals, reducing assessable precision.
Patient-Oriented Outcomes 4.5
Primary outcome is the Life's Essential 8 cardiovascular health score (a composite health metric) rather than hard clinical outcomes such as MI, stroke, mortality, or functional status.
Magnitude of Net Benefit 3.5
The abstract indicates associations (J-shaped/linear) but does not clearly report absolute differences, risk reductions, or clinically interpretable effect sizes; harms and burdens are not directly evaluated because this is not an intervention study.
Implementability & Practicality 7.0
Promoting healthy weight and aerobic fitness is broadly feasible in primary care and public health counseling, though implementing effective, sustained behavior change at scale can be resource-intensive and is not tested here.
Practice-Changing Potential 5.0
Supports existing preventive messaging linking healthy weight/fitness with later cardiovascular health, but observational design and use of a composite score (not clinical events) limit the likelihood of immediate practice change.
7. 5.3
Mortality by income in the elderly population in Italy: new evidence from an innovative microdata integration.
Overall: A massive, precise observational linkage study shows a strong income–mortality gradient in Italy’s elderly population, but because it is non-interventional and observational, it offers limited direct guidance on implementable clinical actions or demonstrable net benefit in primary care.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 6.5
Addresses a major determinant of health (income) and its association with mortality in older adults, which is broadly relevant to primary care risk context and equity-focused care, though it does not test a primary-care clinical decision or intervention.
Validity, Bias Control & Precision 7.0
Very large nationwide retrospective cohort with individual-level administrative income linkage and tight confidence intervals, but observational design with limited covariate adjustment in the abstract leaves residual confounding and bias concerns.
Patient-Oriented Outcomes 9.0
Uses all-cause mortality, a clearly patient-important hard outcome, with sex-stratified estimates.
Magnitude of Net Benefit 2.0
No intervention is evaluated, so net benefit versus harms/burden cannot be assessed; the abstract only quantifies an association (MRR gradient) rather than a change in outcomes from an actionable strategy.
Implementability & Practicality 4.0
The findings are conceptually actionable for policy and targeting, but implementing individual-income-informed care requires data integration/access that is not routine in many clinical primary-care workflows and no specific practical intervention is presented.
Practice-Changing Potential 3.5
Strong descriptive evidence of inequality may support prioritization of social risk screening or resource allocation, but without testing a clinical action it is unlikely to change day-to-day primary-care management on its own.
8. 5.1
Transmissibility and Disease Progression of Asymptomatic Mycobacterium tuberculosis Infection, Lima, Peru.
Overall: A large prospective household-contact cohort provides clinically relevant, patient-important TB infection/disease outcomes, but as an observational study with imprecise estimates and no tested intervention, it offers limited direct, actionable practice change despite useful epidemiologic insight.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 6.0
Tuberculosis exposure evaluation and household-contact risk are relevant to outpatient care and public health, but the study question is more epidemiologic than a direct primary-care decision or intervention.
Validity, Bias Control & Precision 6.0
Prospective cohort with a large sample and adjusted analyses, but it remains nonrandomized with potential residual confounding; effect estimates have wide confidence intervals that frequently include no difference.
Patient-Oriented Outcomes 7.0
Incident TB infection and progression to TB disease are clinically meaningful outcomes rather than purely physiologic surrogates.
Magnitude of Net Benefit 2.0
No intervention is tested, so net benefit cannot be directly established; reported associations suggest lower risk with asymptomatic index cases but are imprecise and not clearly different from no effect.
Implementability & Practicality 4.0
Findings may inform risk counseling and contact-tracing intensity, but the abstract does not provide a concrete, implementable change in clinical workflow or a specific actionable threshold.
Practice-Changing Potential 5.5
Suggests asymptomatic TB may be less transmissible yet common, which could influence policy discussions, but uncertainty in estimates and lack of a tested strategy limit immediate practice change.
9. 5.0
Efficacy of front-of-package nutrient labels designed for mandatory implementation in the USA: an online randomised controlled trial.
Overall: A large online RCT suggests certain front-of-package label formats modestly improve consumers’ nutrient understanding and selection, but outcomes are not directly patient-health oriented and real-world clinical/practice impact is indirect.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 6.0
Diet quality is highly relevant to primary care counseling, but this is a consumer-labeling policy intervention tested outside clinical settings and does not involve clinician actions directly.
Validity, Bias Control & Precision 8.0
Large randomized design with masking to objectives and clear between-group comparisons with confidence intervals; however, it is an online experiment with exclusions and likely limited real-world behavioral fidelity.
Patient-Oriented Outcomes 3.0
Outcomes are understanding, perceptions, and hypothetical/experimental product selection rather than clinical events, symptoms, function, or quality of life.
Magnitude of Net Benefit 4.0
Effects on correct identification are small (e.g., ~1–5 percentage points), and reduced selection of high-in foods is modest (~5–8 percentage points) with no direct health outcome data or clear harms assessed.
Implementability & Practicality 5.0
If adopted, front-of-package labels are scalable and low-burden for consumers, but implementation depends on regulatory action and industry compliance rather than routine clinical workflows.
Practice-Changing Potential 4.0
Findings may inform labeling policy choices, but the abstract does not show downstream health impact and has limited immediate implications for changing typical primary-care practice.
10. 4.8
Subpopulation Differences in Connecting to Resources to Address Social Needs.
Overall: A randomized design supports credibility, but because the reported results are largely ad hoc subgroup/as-treated findings on a process outcome with limited precision and unclear downstream patient benefit, the evidence is only moderately applicable and unlikely to change practice by itself.
View 6 Criterion Scores
Primary-Care Relevance & Applicability 7.0
Addressing social needs and improving linkage to community/VHA resources is relevant to outpatient generalist care, but the Veteran Health Administration setting and veteran-specific eligibility factors limit direct applicability to broader primary-care populations.
Validity, Bias Control & Precision 5.0
The parent study is a 3-arm intent-to-treat RCT with site-level random effects, but the abstract reports ad hoc subgroup and as-treated analyses with no sample sizes and wide confidence intervals, increasing risk of chance findings and imprecision.
Patient-Oriented Outcomes 4.0
The primary outcome (connection to at least one resource) is a process/intermediate outcome; the abstract does not report downstream patient health, quality of life, housing/food security, utilization, or other patient-important outcomes.
Magnitude of Net Benefit 4.0
Some subgroups had higher odds of resource connection (e.g., OR 6.47 with very wide CI), but absolute effects are not provided and potential burdens/harms (time, staffing, opportunity cost) are not quantified, making net benefit uncertain.
Implementability & Practicality 5.0
Tailored sheets are simple, but navigation assistance from a social worker adds staffing and workflow requirements that may be feasible in integrated systems yet harder to scale in typical resource-constrained primary-care settings.
Practice-Changing Potential 4.0
Findings are hypothesis-generating (ad hoc subgroup signals) rather than definitive guidance for routine adoption or targeting, and the outcome is linkage rather than demonstrated improvements in patient health or functioning.
Score Guide: 9-10 Exceptional 7-8 Strong 5-6 Moderate 3-4 Weak 1-2 Poor
Showing top 10 of 17

How It Works

1

Sign Up with Google

Create your account in seconds using your Google account. No passwords to remember.

2

Receive Email Summaries

Get AI-curated clinical summaries delivered to your inbox three times per week (Monday, Wednesday, Friday).

3

Find the Best New Articles

Our interactive article rankings allow you to find the best articles, tailored to you.

Ready to Stay Current?

Sign Up