Visualizing interaction effects: a proposal for presentation and interpretation

Adversities in childhood and young adulthood and incident cardiovascular diseases: a prospective cohort study

Background

Much remains unknown regarding the associations of adversities in childhood and adulthood with incident cardiovascular diseases (CVD). We aimed to examine the independent and cumulative relations of adversities in childhood and adulthood with incident CVD and whether these associations can be mitigated by adopting a healthy lifestyle later in life.

Methods

We included 136,073 men and women [38-72 years at baseline] free of diagnosed CVD at baseline who responded to surveys on adversities in childhood and adulthood in the United Kingdom Biobank prospective cohort. They were recruited between 2006 and 2010 and were followed-up until 28 January 2021. Adversities included physical abuse, emotional abuse, sexual abuse, emotional neglect, and physical neglect. Participants were categorised into four groups according to the exposure periods, which were no adversity, childhood adversity only, adulthood adversity only, and cumulative adversity (both childhood and adulthood). The primary outcomes included incident fatal and non-fatal CVD events. The modifiable lifestyle factors were smoking, physical activity, diet, sleeping, social or leisure activities, and friend or family visits.

Findings

We identified 16,415 (10.71/1000 person-year) incident CVD during a median follow-up of 11.8 years. Compared with participants with no adversity, CVD incidence increased by 11% in those with childhood adversity only (adjusted hazard ratio [HR]: 1.11 [95% CI 1.06-1.17], p < 0.001), 4% in those with adulthood adversity only (1.04 [1.00-1.09], p = 0.05), and 21% in those with cumulative adversity (1.21 [1.16-1.26], p < 0.001). Analysis of interactions showed that adulthood adversity amplified the childhood adversity-CVD association (p for interaction = 0.03). Compared with the participants with one or fewer ideal lifestyle factors, those with more than four ideal factors had a 25%-36% lower risk of CVD across the three adversity groups.

Interpretation

Our findings suggested that childhood adversities were associated with an increased risk of CVD which can be magnified by adulthood adversities and substantially mitigated by adopting a healthy lifestyle later in life.

Funding

The National Natural Science Foundation of China and Guangzhou Foundation for Basic and Applied Basic Research.

Variability, Mean, and Baseline Values of Metabolic Parameters in Predicting Risk of Type 2 Diabetes

CONTEXT

The effect of baseline (B) and alteration of metabolic parameters (MPs), including plasma glucose (PG) testing, insulin resistance surrogates, and lipid profile and their mutual interactions on the development of type 2 diabetes mellitus (T2DM), has not been investigated systematically.

OBJECTIVE

To access the association of the past variability (V), past mean (M), and B values of various MPs and their mutual interaction with the risk of T2DM.

METHODS

A community-based, longitudinal analysis was conducted using the Korean Genome and Epidemiology Study comprising 3829 nondiabetic participants with completed MPs measurements during 3 biannually visits who were followed over the next 10 years. Outcomes included the incidence of T2DM during follow-up.

RESULTS

Among predictors, PG concentrations measured during the oral glucose tolerance test were the most prominent T2DM determinants, in which the M of the average value of fasting PG (FPG), 1-hour, and 2-hour PGs had the strongest discriminative power (hazard ratios and 95% CI for an increment of SD: 3.00 (2.5-3.26), AUC: 0.82). The M values of MPs were superior to their B and V values in predicting T2DM, especially among postload PGs. Various mutual interactions between indices and among MPs were found. The most consistent interactants were the M values of high-density lipoprotein cholesterol and the M and V values of FPG. The findings were similar in normal glucose tolerance participants and were confirmed by sensitivity analyses.

CONCLUSION

Postload PG, past alteration of measurements, and mutual interactions among indices of MPs are important risk factors for T2DM development.

Androgen receptor DNA methylation is an independent determinant of glucose metabolic disorders in women; testosterone plays a moderating effect

BACKGROUND

We have previously shown that serum testosterone was associated with impaired fasting glucose (IFG) and type 2 diabetes (T2D). Testosterone can be acting through binding the androgen receptor (AR). Therefore, we aimed to explore the independent associations of AR DNA methylation (ARm) with IFG and T2D and the moderation effects of serum testosterone on the associations.

METHODS

A case-control study with 1065 participants including 461 men and 604 women was performed. ARm in peripheral blood sample and serum testosterone were measured using pyrosequeuncing and liquid chromatography-tandem mass, respectively. Multivariable logistic regression was performed to estimate the associations of ARm (including 2 cytosine-phosphoguanine [CpG] islands and average methylation levels) with different glucose status. Serum testosterone was used as a moderator to estimate the moderation effect.

RESULTS

After multivariate adjustment, CpG 1, 2 and CpG average methylation were all significantly associated with IFG (CpG 1: Odds ratio (OR) = 4.80, 95% confidence interval (CI): 2.24-10.27; CpG 2: OR = 4.35, 95% CI: 2.50-7.58; CpG average: OR = 11.73, 95% CI: 5.36-25.67) in women. In addition, testosterone played negative moderation effects in above associations. Moreover, no significant independent associations of methylation levels with T2D was observed both in men and women.

CONCLUSION

Our findings demonstrate that ARm was positively associated with IFG in women and the associations would be weakened by testosterone. The individuals experiencing low testosterone and ARm levels reported a lower state of IFG than those who experienced high levels of testosterone and ARm in women.

Improved adherence with Medicines Use Review service in Slovenia: a randomized controlled trial

Background

Based on several existing patient-oriented activities, Medicines Use Review (MUR) service was standardized and officially adopted in Slovenia in 2015. Service aims to provide adherence support and ensure safe and effective medicines use. Therefore, the aim of the study was to evaluate the benefits of MUR in Slovenia, primarily the impact on medication adherence.

Methods

A randomised controlled trial was performed in community pharmacies to compare MUR with standard care. Patients were randomised into either the test (patients received MUR by a certified MUR provider at visit 1), or control group. The study primary outcome was self-reported adherence to multiple medications, assessed by electronic ©Morisky Widget MMAS-8 Software at the first visit (V1) and after 12 weeks (V2). A sub-analysis of intentional and unintentional non-adherence was performed. MUR impact was defined as the relative difference in ©MMAS-8 score after 12 weeks between the test and control group. A multiple linear regression model was used to predict MUR impact based on baseline adherence (low versus medium and high). Several secondary outcomes (e.g. evaluation of drug-related problems (DRPs)) were also assessed.

Results

Data from 153 (V1) and 140 (V2) patients were analysed. Baseline adherence was low, moderate and high in 17.6, 48.4 and 34.0% patients, respectively. In the low adherence subpopulation, test group patients showed a 1.20 point (95% CI = 0.16–2.25) increase in total ©MMAS-8 score (p = 0.025) compared to control group patients. A 0.84 point (95% CI = 0.05–1.63) increase was due to intentional non-adherence (p = 0.038), and a 0.36 point (95% CI = − 0.23-0.95) was due to unintentional non-adherence (p = 0.226). Additionally, statistically significant decrease in the proportion of patients with manifested DRPs (p < 0.001) and concerns regarding chronic medicines use (p = 0.029) were revealed.

Conclusion

MUR service in Slovenia improves low medication adherence and is effective in addressing DRPs and concerns regarding chronic medicines use.

Trial registration

ClinicalTrials.gov - NCT04417400; 4th June 2020; retrospectively registered.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-021-06223-8.

Factors influencing home blood pressure monitor ownership in a large clinical trial

Home blood pressure monitor (HBPM) ownership prevalence and the factors that influence it are unclear. This study aimed to investigate factors associated with HBPM ownership among participants in the Treatment in Morning versus Evening (TIME) hypertension study. This study is a sub-analysis of the TIME study, a randomised trial investigating the effect of day-time versus night-time dosing of antihypertensive medication on cardiovascular outcomes in adults with hypertension. As part of the TIME study online registration process, participants were asked to indicate whether they owned an HBPM. A multivariable logistic regression model was constructed to determine factors associated with HBPM ownership. Of 21,104 randomised participants, 11,434 (54.2%) reported owning an HBPM. The mean age of all participants at enrolment was 67.7 ± 9.3 years, 12,134 (57.5%) were male, and 8892 (42.1%) reported a current or previous history of smoking. Factors associated with an increased likelihood of reporting HBPM owned include being male (OR:1.47; 95% CI 1.39–1.56) or residing in a less deprived socioeconomic region (IMD Decile 6–10) (OR:1.31; 95% CI 1.23–1.40). Participants with a history of diabetes mellitus (OR:0.74; 95% CI:0.64–0.86) or current smokers, compared to non-smokers, (OR:0.71; 95% CI:0.62–0.82) were less likely to report owning an HBPM. This study has identified important patient factors influencing HBPM ownership. Further qualitative research would be valuable to identify and explore potential patient-level barriers to engagement with self-monitoring of blood pressure.

Donor Age Amplifies the Detrimental Effects of Cold Ischemia Time on Long-Term Kidney Allograft Survival Independently of the Occurrence of Delayed Graft Function or Early Acute Rejection

OBJECTIVES

We investigated the influence of the interaction between donor age and cold ischemia time on allograft survival in the absence of delayed graft function, early acute rejection, or the combination of both.

MATERIALS AND METHODS

We conducted a retrospective analysis of a cohort of patients first transplanted with living-related and deceased-donor allografts between 2001 and 2016. Predictors included cold ischemia time, donor and recipient age and sex, body mass index, renal replacement therapy duration, cause of end-stage renal disease, HLA class I and II mismatches, panel of reactive antibodies score, donor creatinine concentration, development of delayed graft function, and biopsy-proven acute rejection. The response variable was time until return to renal replacement therapy. Patients who died with functioning allografts were censored at the time of death. Analyses included multivariate Cox proportional hazards regression.

RESULTS

The study included 498 patients followed for median of 4.1 years with median cold ischemia time of 17.0 hours. On multivariate analysis, allograft survival was negatively affected by the cold ischemia time-donor age interaction (P = .026), acute rejection (P = .043), delayed graft function (P = .001), and acute rejection combined with delayed graft function (P = .002). Restricted mean allograft survival times in patients who developed neither delayed graft function nor acute rejection decreased from 13.6 to 8.6 years when cold ischemia time increased from 12 to 36 hours and donor age increased from 30 to 60 years.

CONCLUSIONS

Allograft survival was negatively affected by donor age-cold ischemia time interaction independently of the development of delayed graft function, acute rejection, or their combination.

Mediterranean Style Diet and Kidney Function Loss in Kidney Transplant Recipients

BACKGROUND AND OBJECTIVES

Despite improvement of short-term graft survival over recent years, long-term graft survival after kidney transplantation has not improved. Studies in the general population suggest the Mediterranean diet benefits kidney function preservation. We investigated whether adherence to the Mediterranean diet is associated with kidney outcomes in kidney transplant recipients.

DESIGN, SETTING, PARTICIPANTS, & MEASUREMENTS

We included 632 adult kidney transplant recipients with a functioning graft for ≥1 year. Dietary intake was inquired using a 177-item validated food frequency questionnaire. Adherence to the Mediterranean diet was assessed using a nine-point Mediterranean Diet Score. Primary end point of the study was graft failure and secondary end points included kidney function decline (doubling of serum creatinine or graft failure) and graft loss (graft failure or death with a functioning graft). Cox regression analyses were used to prospectively study the associations of the Mediterranean Diet Score with study end points.

RESULTS

During median follow-up of 5.4 (interquartile range, 4.9-6.0) years, 76 participants developed graft failure, 119 developed kidney function decline, and 181 developed graft loss. The Mediterranean Diet Score was inversely associated with all study end points (graft failure: hazard ratio [HR], 0.68; 95% confidence interval [95% CI], 0.50 to 0.91; kidney function decline: HR, 0.68; 95% CI, 0.55 to 0.85; and graft loss: HR, 0.74; 95% CI, 0.63 to 0.88 per two-point increase in Mediterranean Diet Score) independent of potential confounders. We identified 24-hour urinary protein excretion and time since transplantation to be an effect modifier, with stronger inverse associations between the Mediterranean Diet Score and kidney outcomes observed in participants with higher urinary protein excretion and participants transplanted more recently.

CONCLUSIONS

Adherence to the Mediterranean diet is associated with better kidney function outcomes in kidney transplant recipients.

Interactive effects of testosterone and the androgen receptor CAG repeat length polymorphism on cardiovascular-renal events and mortality in men with diabetes

AIM

Current evidence relating testosterone to cardiovascular disease and mortality is inconclusive. Cellular effects of testosterone are mediated by androgen receptor and longer receptor gene CAG repeat length correlates with reduced transcriptional activity. We investigated the independent and interactive association of total testosterone and CAG repeat length with incident cardiovascular disease (CVD), chronic kidney disease (CKD) and mortality in Chinese men with type 2 diabetes.

MATERIALS AND METHODS

From March 2008 and February 2009, 474 men with diabetes underwent structured clinical assessment including genotyping for CAG repeat length. Patients were followed for new-onset CVD, CKD defined by estimated glomerular filtration rate <60 mL/min/1.73m² , and death until 31 May 2015.

RESULTS

In this cohort (mean age: 58.6 years, disease duration: 15.4 years), CAG repeat number ranged from 11 to 32 with median of 23, and 9.3% had low testosterone. Over follow-up of 5.8 years, 49 (10.3%) men had CVD, 139 (29.3%) had CKD, and 43 (9.1%) died. In multivariate Cox regression adjusted for age, duration of diabetes, and cardiometabolic risk factors, both total testosterone and interaction term of total testosterone × CAG repeat were associated with all-cause death with respective hazard ratios 1.63 (P = 0.002) and 0.98 (P = 0.004). Total testosterone and CAG repeat were not related to incident CVD or CKD.

CONCLUSIONS

Among men with type 2 diabetes, high total testosterone was associated with increased mortality in the presence of shorter CAG repeat length but decreased mortality in those with long CAG repeats.

Hypertension referrals from community pharmacy to general practice: multivariate logistic regression analysis of 131 419 patients

BACKGROUND

The burden of hypertension in primary care is high, and alternative models of care, such as pharmacist management, have shown promise. However, data describing outcomes from routine consultations between pharmacists and patients with hypertension are lacking.

AIM

To identify factors associated with referral of patients from pharmacies to general practice within the first 2 weeks of starting a new antihypertensive medication.

DESIGN AND SETTING

Multivariate logistic regression conducted on data from community pharmacies in England.

METHOD

Data were obtained from the New Medicine Service between 2011 and 2012. Analyses were conducted on 131 419 patients. In all, 15 predictors were included in the model, grouped into three categories: patient-reported factors, demographic factors, and medication-related factors.

RESULTS

Mean patient age was 65 years (±13 years), and 85% of patients were of white ethnicity. A total of 5895 (4.5%) patients were referred by a pharmacist to a GP within the first 2 weeks of starting a new antihypertensive medication. Patients reporting side effects (adjusted odds ratio [OR] 11.60, 95% confidence interval [CI] = 10.85 to 12.41) were most likely to be referred. Prescriptions for alpha-blockers were associated with referral (adjusted OR 1.28, 95% CI = 1.12 to 1.47), whereas patients receiving angiotensin-II receptor blockers were less likely to be referred (adjusted OR 0.89, 95% CI = 0.80 to 0.99).

CONCLUSION

Most patients were followed up by pharmacists without the need for referral. Patient-reported side effects, medication-related concerns, and the medication class prescribed influenced referral. These data are reassuring, in that additional pharmacist involvement does not increase medical workload appreciably, and support further development of pharmacist-led hypertension interventions.

The risk association of plasma total homocysteine with acute myocardial infarction is modified by serum vitamin A

Background

Plasma total homocysteine (tHcy) has been implicated in the development of cardiovascular disease, but the mechanisms remain unclear. Vitamin A (Vit-A) is involved in homocysteine metabolism and we therefore explored the potential interaction between plasma tHcy and serum Vit-A in relation to incident acute myocardial infarction.

Methods

Cox proportional hazards models were used to assess the prospective relationships between tHcy and acute myocardial infarction in 2205 patients from Western Norway undergoing elective coronary angiography for suspected stable angina pectoris. Results are reported as hazard ratio per standard deviation increase in log-transformed tHcy. An interaction term for tHcy × Vit-A was added to multivariate models including age, sex, smoking, apolipoprotein B fasting, statin and aspirin prescription and estimated glomerular filtration rate.

Results

Geometric mean (geometric standard deviation) age of the participants (64.3% men) was 62.3 (1.24) years. Plasma tHcy was higher among participants in the upper versus lower Vit-A tertile. During 7 (2.4) years of follow-up, 15.1% suffered an AMI. A significant association of plasma tHcy with AMI in the total study population was observed. When we stratified the population according to Vit-A tertiles, plasma tHcy was associated with acute myocardial infarction only in the upper Vit-A tertile (hazard ratio per SD: 1.25, 95% confidence interval: 1.04–1.53, pinteraction = 0.03).

Conclusions

The risk relationship between plasma tHcy and acute myocardial infarction was modified by serum concentrations of Vit-A in patients with suspected stable angina pectoris. This finding may clarify the relationship between tHcy and cardiovascular disease.

Big and tall: Does a height premium dwarf an obesity penalty in the labor market?

Previous studies have shown that both height and weight are associated with wages. However, some gaps in our understanding of the relationship between body size and wages remain. For example, given a height premium and an obesity penalty, due to forces working in opposite directions, the current literature is unable to provide clear answers to questions such as whether a tall obese woman or a short healthy weight woman would earn a higher wage premium. Using Australian data and iso-contour wage curves derived from a semi-parametric wage regression model, this paper illustrates the complex nature of the relationship between height, weight and wages and how the nature of these differences depends on gender and age. As adult height is fixed, a key focus of the paper is illustrating for various height ranges whether there are any wage benefits in the labor market to increasing or decreasing one's weight. For individuals aged 25-54 as a whole, I find that there are strong effects of weight reduction at lower ends of the height distribution for females (between 1.50-1.70m) but not for males (<1.65m). For relatively taller men (>1.85m), a wage premium is found for being overweight. For relatively taller women (>1.72m), no penalty for being overweight is discernible.

Visualizing Risk Prediction Models

OBJECTIVE

Risk prediction models can assist clinicians in making decisions. To boost the uptake of these models in clinical practice, it is important that end-users understand how the model works and can efficiently communicate its results. We introduce novel methods for interpretable model visualization.

METHODS

The proposed visualization techniques are applied to two prediction models from the Framingham Heart Study for the prediction of intermittent claudication and stroke after atrial fibrillation. We represent models using color bars, and visualize the risk estimation process for a specific patient using patient-specific contribution charts.

RESULTS

The color-based model representations provide users with an attractive tool to instantly gauge the relative importance of the predictors. The patient-specific representations allow users to understand the relative contribution of each predictor to the patient's estimated risk, potentially providing insightful information on which to base further patient management. Extensions towards non-linear models and interactions are illustrated on an artificial dataset.

CONCLUSION

The proposed methods summarize risk prediction models and risk predictions for specific patients in an alternative way. These representations may facilitate communication between clinicians and patients.