Testing the proportional hazards assumption in cox regression and dealing with possible non-proportionality in total joint arthroplasty research: methodological perspectives and review

Perinatal depression and risk of maternal cardiovascular disease: a Swedish nationwide study

BACKGROUND AND AIMS

Increasing evidence suggests that some reproductive factors/hazards are associated with a future risk of cardiovascular disease (CVD) in women. While major (non-perinatal) depression has consistently been associated with CVD, the long-term risk of CVD after perinatal depression (PND) is largely unknown.

METHODS

A nationwide population-based matched cohort study involving 55 539 women diagnosed with PND during 2001-14 in Sweden and 545 567 unaffected women individually matched on age and year of conception/delivery was conducted. All women were followed up to 2020. Perinatal depression and CVD were identified from Swedish national health registers. Using multivariable Cox models, hazard ratios (HR) of any and type-specific CVD according to PND were estimated.

RESULTS

The mean age at the PND diagnosis was 30.8 [standard deviation (SD) 5.6] years. During the follow-up of up to 20 years (mean 10.4, SD 3.6), 3533 (6.4%) women with PND (expected number 2077) and 20 202 (3.7%) unaffected women developed CVD. Compared with matched unaffected women, women with PND had a 36% higher risk of developing CVD [adjusted HR = 1.36, 95% confidence interval (CI): 1.31-1.42], while compared with their sisters, women with PND had a 20% higher risk of CVD (adjusted HR = 1.20, 95% CI 1.07-1.34). The results were most pronounced in women without a history of psychiatric disorder (P for interaction < .001). The association was observed for all CVD subtypes, with the highest HR in the case of hypertensive disease (HR = 1.50, 95% CI: 1.41-1.60), ischaemic heart disease (HR = 1.37, 95% CI: 1.13-1.65), and heart failure (HR 1.36, 95% CI: 1.06-1.74).

CONCLUSIONS

Women with PND are at higher risk of CVD in middle adulthood. Reproductive history, including PND, should be considered in CVD risk assessments of women.

Demographics and medical burden of osteogenesis imperfecta: a nationwide database analysis

The epidemiological data on osteogenesis imperfecta (OI) in Asia is limited. This study, representing the first comprehensive epidemiological investigation on OI in Taiwan, reveals high medical resource utilization and underscores the importance of early diagnosis to enhance care quality.

INTRODUCTION

This study examines osteogenesis imperfecta, a hereditary connective tissue disorder causing pediatric fractures and limb deformities, using a nationwide database from Taiwan to analyze clinical features and medical burden.

METHODS

The study identified validated OI patients from the Catastrophic Illness Registry in the National Health Insurance Research Database from 2008 to 2019. Demographic data and medical resource utilization were analyzed. A multivariate Cox model assessed the influence of sex, validation age, and comorbidities.

RESULTS

319 OI patients (M/F = 153/166) were identified, with 58% validated before age 20. Prevalence and incidence were 0.8-1.3/100,000 and 0.02-0.09/100,000, respectively, with higher rates in the pediatric demographic. In the study period, 69.6% of the patients had admission history, primarily to pediatric and orthopedic wards. The median admission number was 3, with a median length of stay of 12 days and a median inpatient cost of approximately 3,163 USD during the period. Lower limb fractures were the main reason for hospitalization. 57% of OI patients received bisphosphonate treatment. The leading causes of mortality were OI-related deaths, neurovascular disease, and cardiovascular disease. The median age of validation in the non-survival group was significantly higher compared to the survival group (33 vs. 14 years), and patients validated during childhood required more inpatient fracture surgeries than those validated during adulthood.

CONCLUSION

This study provides comprehensive real-world evidence on the clinical characteristics and high medical resource utilization of OI patients in a low prevalence region like Taiwan. Early diagnosis is crucial for improving care quality and enhancing health outcomes.

The use and abuse of survival analysis and Kaplan-Meier curves in surgical trials

BACKGROUND

Survival analysis based on Cox regression and Kaplan-Meier curves, initially devised for oncology trials, have frequently been used in other contexts where fundamental statistical assumptions (such as a constant hazard ratio) are not satisfied. This is almost always the case in trials that compare surgery with medical management.

METHODS

We review a trial that compared extracranial-intracranial bypass surgery (EC-IC bypass) with medical management (MM) of patients with symptomatic occlusion of the carotid or middle cerebral artery, where it was claimed that surgery was of no benefit. We discuss a hypothetical study and review other neurovascular trials which have also used survival analysis to compare results.

RESULTS

The trial comparing EC-IC bypass and MM did not satisfy the fundamental proportional hazard assumption necessary for valid analyses. This was also the case for two prior EC-IC bypass trials, as well as for other landmark neurovascular studies, such as the trials comparing endarterectomy with MM for carotid stenoses, or for the trial that compared intervention and MM for unruptured brain arteriovenous malformations. While minor deviations may have little effect on large trials, it may be impossible to show the benefits of surgery when trial size is small and deviations large.

CONCLUSION

Survival analyses are inappropriate in RCTs comparing surgery with conservative management, unless survival is calculated after the postoperative period. Alternative ways to compare final clinical outcomes, using for example a fixed follow-up period, should be planned for preventive surgical trials that compare intervention with conservative management.

Negative impact of comorbidities on all-cause mortality of patients with psoriasis is partially alleviated by biologic treatment: A real-world case-control study

BACKGROUND

Cardiovascular comorbidities are believed to cause higher mortality in psoriasis patients. Conversely, systemic therapy may improve overall survival.

OBJECTIVE

To evaluate the impact of different comorbidities and therapy on mortality risk of psoriasis patients in the entire population of Alberta, Canada (population 4.37 million).

METHODS

Cohorts of psoriasis cases (n = 18,618) and controls (ambulatory patients matched 1:3 by age and sex) were retrieved from Alberta Health Services Data Repository of Reporting database within the period 2012 to 2019. Cases were stratified according to Charlson Comorbidity Index, and the type of therapy.

RESULTS

Mortality in psoriasis cohort was significantly higher than in the controls (median age of death 72.0 years vs 74.4 years, respectively). Charlson Comorbidity Index and comorbidities were strong predictors of mortality, in particular drug induced liver injury (hazard ratio 1.8, affective bipolar disease, hazard ratio 1.6, and major cardiovascular diseases. Mortality was lower in patients treated with biologics (hazard ratio 0.54).

LIMITATIONS

Some factors (psoriasis type and severity, response to treatment, smoking, alcohol intake) could not be measured.

CONCLUSIONS

Hepatic injury, psychiatric affective disorders and cardiovascular disease were major determinants of overall survival in psoriasis. Biologic therapy was associated with a reduced mortality risk.

Associations between lipid abnormalities and diabetic retinopathy across a large United States national database

BACKGROUND/OBJECTIVES

While dyslipidaemia has been suggested as a potential risk factor for diabetic retinopathy (DR), previous studies have reported conflicting findings. This study aimed to better characterize the relationship between abnormal serum levels of various lipid markers and the risk of the development and progression of DR.

SUBJECTS/METHODS

This retrospective cohort study utilized a United States national database of electronic medical records. Adults with a history of type 2 diabetes mellitus without type 1 diabetes mellitus were divided into cohorts based on the presence of abnormal serum levels of various lipid markers. Propensity score matching was performed to match cohorts with abnormal lipid levels to those with normal lipid levels on covariates. The cohorts were then compared to evaluate the hazard ratios (HR) of receiving a new DR diagnosis, pars plana vitrectomy, panretinal photocoagulation, vitreous haemorrhage, proliferative diabetic retinopathy, diabetic macular oedema (DMO), and traction retinal detachment.

RESULTS

The database contained 1,126,231 eligible patients (mean age: 60.8 [14.2] years; 46.0% female). Among patients without prior DR, low HDL (HR = 0.94, CI = 0.90-0.98), total cholesterol (HR = 0.88, CI = 0.85-0.91), and high triglyceride (HR = 0.91, CI = 0.86-0.97) levels were associated with a decreased risk of receiving a DR diagnosis. Among patients with preexisting DR, high LDL levels was associated with an increased risk of DMO (HR = 1.42, CI = 1.15-1.75), whereas low HDL levels was associated with a marginally decreased risk (HR = 0.92, CI = 0.85-0.99).

CONCLUSIONS

Elevated levels of markers of dyslipidaemia are inversely associated with the risk of receiving a DR diagnosis, but this relationship is blunted after the onset of DR.

Evaluation of the proportional hazards assumption and covariate adjustment methods in comparative surgical observational studies with time-to-event endpoints

INTRODUCTION

Comparative studies on surgical treatments with time-to-event endpoints have provided substantial evidence for clinical practice, but the accurate use of survival data analysis and the control of confounding bias remain big challenges.

METHODS

This was a survey of surgical studies with survival outcomes published in four general medical journals and five general surgical journals in 2021. The two most concerned statistical issues were evaluated, including confounding control by propensity score analysis (PSA) or multivariable analysis and testing of proportional hazards (PH) assumption in Cox model.

RESULTS

A total of 74 studies were included, comprising 63 observational studies and 11 randomized controlled trials. Among the observational studies, the proportion of studies utilizing PSA in surgical oncology and non-oncology studies was similar (40.9 % versus 36.8 %, P = 0.762). However, the former reported a significantly lower proportion of PH assumption assessments compared to the latter (13.6 % versus 42.1 %, P = 0.020). Twenty-five observational studies (25/63) used PSA methods, but two-thirds of them (17/25) showed unclear balance of baseline data after PSA. And the proportion of PH assumption testing after PSA was slightly lower than that before PSA, but the difference was not statistically significant (24.0 % versus 28.0 %, P = 0.317). Comprehensive suggestions were given on confounding control in survival analysis and alternative resolutions for non-compliance with PH assumption.

CONCLUSION

This study highlights suboptimal reporting of PH assumption evaluation in observational surgical studies both before and after PSA. Efforts and consensus are needed with respect to the underlying assumptions of statistical methods.

Diabetes mellitus is a risk factor for incident chronic kidney disease: A nationwide cohort study

Objective

Diabetes mellitus and chronic kidney disease are multifactorial conditions with multiple etiologies that share similar pathophysiologies. This nationwide cohort study examined the impact of diabetes mellitus on the follow-up development of chronic kidney disease.

Methods

By retrieving the Longitudinal Health Insurance Database 2005, 5121 patients with diabetes mellitus were included in this study and 5121 patients without diabetes mellitus, who were matched according to sex, age, and Charlson comorbidity index made up the control group. The adjusted hazard ratios for chronic kidney disease were calculated using Cox proportional hazards regression analysis. Kaplan-Meier analysis was used to estimate the cumulative incidence of chronic kidney disease rate in the diabetes mellitus and control groups.

Results

After adjusting for sex, age, and Charlson comorbidity index score, the diabetes mellitus group had a 1.380 times higher (95% CI: 1.277-1.492) risk of developing chronic kidney disease than the control group. Further stratified analysis showed that patients with diabetes mellitus had a significantly higher risk of developing chronic kidney disease regardless of their sex, age, and Charlson comorbidity index score, compared to those without diabetes mellitus.

Conclusions

There is a possibility that diabetes mellitus serves as an independent risk factor for chronic kidney disease development. Early screening and monitoring of diabetes mellitus appear to be of great importance in the prevention of chronic kidney disease.

Pediatric traumatic brain injury and early age multiple sclerosis in Finland: A nationwide register-based cohort study

OBJECTIVE

Examine the link between pediatric traumatic brain injury (pTBI) and early-onset multiple sclerosis in Finland.

METHODS

Conducted nationwide register study (1998-2018) with 28,750 pTBI patients (< 18) and 38,399 pediatric references with extremity fractures. Multiple sclerosis diagnoses from Finnish Social Insurance Institution. Employed Kaplan-Meier and multivariable Cox regression for probability assessment, results presented with 95% CI.

RESULTS

Of 66 post-traumatic multiple sclerosis cases, 30 (0.10%) had pTBI, and 36 (0.09%) were in the reference group. Cumulative incidence rates (CIR) in the first 10 years were 46.5 per 100,000 (pTBI) and 33.1 per 100,000 (reference). Hazard ratio (HR) for pTBI was 1.10 (95% CI: 0.56-1.48).Stratified by gender, women's CIR was 197.9 per 100,000 (pTBI) and 167.0 per 100,000 (reference) after 15 years. For men, CIR was 44.6 per 100,000 (pTBI) and 34.7 per 100,000 (reference). In the initial 3 years, HR for female pTBI was 1.75 (95% CI: 0.05-6.32), and between years 3 and 20, it was 1.08 (95% CI: 0.51-1.67). For male patients, HR was 1.74 (95% CI: 0.69-4.39).

SIGNIFICANCE

We did not find evidence of an association between pTBI and early-onset multiple sclerosis 20 years post-initial trauma.

Pyroptosis-Related Gene Signature Predicts Prognosis and Response to Immunotherapy and Medication in Pediatric and Young Adult Osteosarcoma Patients

Purpose

Pyroptosis, a new form of inflammatory programmed cell death, has recently gained attention. However, the impact of the expression levels of pyroptosis-related genes (PRGs) on the overall survival (OS) of osteosarcoma patients remains unclear. This study aims to investigate the impact of the expression levels of PRGs on the OS of pediatric and young adult patients with osteosarcoma.

Patients and Methods

Transcriptome matrix datasets of normal muscle or skeletal tissues from the Genotype-Tissue Expression (GTEx) project and osteosarcoma specimen the National Cancer Institute's (NCI) Therapeutically Applicable Research to Generate Effective Treatments (TARGET) database were used to identify pyroptosis-related genes (PRGs) associated with prognosis. The National Center for Biotechnology Information's (NCBI) GSE21257 dataset was employed to validate the predictive value of the pyroptosis-related signature (PRS). Additionally, reverse transcription polymerase chain reaction (RT-qPCR) experiment was performed in normal and osteosarcoma cell lines.

Results

The study identified 18 differentially expressed PRGs (DEPRGs) between normal muscle or skeletal tissues and tumor samples. Multiple machine learning techniques were used to select PRGs, resulting in the identification of four hub PRGs. A PRS-score was calculated for each sample based on the expression of these four hub PRGs, and samples were categorized into low and high PRS-score level groups. It was confirmed that metastatic status and PRS-score level are independent prognostic predictors. A nomogram model for predicting OS of osteosarcoma patients was constructed. Single-cell RNA-sequencing data display the expression patterns of the hub PRGs. RT-qPCR data results were found to be consistent with the differential expression analysis performed on TARGET and GTEx samples.

Conclusion

The study developed a novel pyroptosis-related gene signature that can stratify pediatric and young adult osteosarcoma patients into different risk groups, thus predicting their response to immunotherapy and chemotherapy.

Patient Self-Assessment of Walking Ability and Fracture Risk in Older Australian Adults

Importance

The relationship between self-reported walking limitation, a proxy of muscle function, and fracture risk has not been investigated.

Objective

To examine the association between a self-reported walking limitation of 1000 m or less and 5-year risk of fracture.

Design, Setting, and Participants

This prospective cohort study compared individuals with various degrees of walking ability limitation at 1000 m (a little limitation and a lot of limitation) and those without limitation (no limitation) accounting for age, falls, prior fractures, and weight. Participants from the ongoing population-based Sax Institute 45 and Up Study were followed from recruitment (2005-2008) for 5 years (2010-2013). Data analysis was conducted from July 2020 to September 2023.

Exposure

Self-reported walking limitation.

Main Outcomes and Measures

Incident fracture and site-specific fractures (hip, vertebral, and nonhip nonvertebral [NHNV] fractures).

Results

Among the 266 912 participants enrolled in the 45 and Up Study, 238 969 were included, with 126 015 (53%) women (mean [SD] age, 63 [11] years) and 112 954 (47%) men (mean [SD] age, 61 [11] years). Approximately 20% reported a degree of limitation in walking 1000 m or less at baseline (39 324 women [24%]; 23 191 men [21%]). During a mean (SD) follow-up of 4.1 (0.8) years, 7190 women and 4267 men experienced an incident fracture. Compared with participants who reported no walking limitations, a little limitation and a lot of limitation were associated with higher risk of fracture (a little limitation among women: hazard ratio [HR], 1.32; 95% CI, 1.23-1.41; a little limitation among men: HR, 1.46; 95% CI, 1.34-1.60; a lot of limitation among women: HR, 1.60; 95% CI, 1.49-1.71; a lot of limitation among men: HR, 2.03; 95% CI, 1.86-2.22). Approximately 60% of fractures were attributable to walking limitation. The association was significant for hip, vertebral, and NHNV fracture and ranged between a 21% increase to a greater than 219% increase.

Conclusions and Relevance

In this cohort study of 238 969 participants, self-reported walking limitations were associated with increased risk of fracture. These findings suggest that walking ability should be sought by clinicians to identify high-risk candidates for further assessment.

Return to work after cancer-the impact of working conditions: A Norwegian Register-based Study

PURPOSE

The purpose of this study is to compare a cohort of cancer survivors with a cohort of cancer-free employees (1) with respect to employment prospects over a 15-year period and (2) with respect to the differential impact of working conditions on employment over this time period.

METHODS

The cancer cohort is retrieved from the Cancer Registry of Norway, while data on the non-cancer cohort are retrieved from register data managed by Statistics Norway. Job exposure matrices were used to remedy the lack of working-conditions information in the register data. We use nearest-neighbor matching to match the non-cancer cohort (the control group) to the cancer-survivor cohort (the treatment group). Cox regression analysis was applied to examine the relationships between working conditions, employment, and cancer. The results are reported separately for mechanical-job exposures and psychosocial exposures, as well as by gender.

RESULTS

Cancer survivors are more likely to experience reduced employment as compared to individuals without a history of cancer. Male cancer survivors in physically demanding occupations have an increased risk of reduced employment after being diagnosed with cancer. This does not apply to female cancer survivors. Regarding the impact of psychosocial exposures on employment, we find no differences over time between cancer survivors and the non-cancer population.

CONCLUSIONS

Male cancer survivors in physically demanding occupations have an increased risk of reduced employment after being diagnosed with cancer, whereas this is not the case for female cancer survivors. Psychosocial exposures do not impact the relative risk of reduced employment over time.

IMPLICATIONS FOR CANCER SURVIVORS

We suggest that return to work after cancer should be considered a process rather than only the re-entry step of resuming work. Thus, it is important to provide long-term support for cancer survivors. We recommend providing more attention to working conditions, particularly in occupations that involve a high level of mechanical-job exposures.

Primary reverse total shoulder arthroplasty for fractures requires more revisions than for degenerative conditions 1 year after surgery: an analysis from the Dutch Arthroplasty Register

BACKGROUND

Although reverse total shoulder arthroplasty (RTSA) is considered a viable treatment strategy for proximal humeral fractures, there is an ongoing discussion of how its revision rate compares with indications performed in the elective setting. First, this study evaluated whether RTSA for fractures conveyed a higher revision rate than RTSA for degenerative conditions (osteoarthritis, rotator cuff arthropathy, rotator cuff tear, or rheumatoid arthritis). Second, this study assessed whether there was a difference in patient-reported outcomes between these 2 groups following primary replacement. Finally, the results of conventional stem designs were compared with those of fracture-specific designs within the fracture group.

MATERIALS AND METHODS

This was a retrospective comparative cohort study with registry data from the Netherlands, generated prospectively between 2014 and 2020. Patients (aged ≥ 18 years) were included if they underwent primary RTSA for a fracture (<4 weeks after trauma), osteoarthritis, rotator cuff arthropathy, rotator cuff tear, or rheumatoid arthritis, with follow-up until first revision, death, or the end of the study period. The primary outcome was the revision rate. The secondary outcomes were the Oxford Shoulder Score, EuroQol 5 Dimensions (EQ-5D) score, numerical rating scale score (pain at rest and during activity), recommendation score, and scores assessing change in daily functioning and change in pain.

RESULTS

This study included 8753 patients in the degenerative condition group (mean age, 74.3 ± 7.2 years) and 2104 patients in the fracture group (mean age, 74.3 ± 7.8 years). RTSA performed for fractures showed an early steep decline in survivorship: Adjusted for time, age, sex, and arthroplasty brand, the revision risk after 1 year was significantly higher in these patients than in those with degenerative conditions (hazard ratio [HR], 2.50; 95% confidence interval, 1.66-3.77). Over time, the HR steadily decreased, with an HR of 0.98 at year 6. Apart from the recommendation score (which was slightly better within the fracture group), there were no clinically relevant differences in the patient-reported outcome measures after 12 months. Patients who received conventional stems (n = 1137) did not have a higher likelihood of undergoing a revision procedure than those who received fracture-specific stems (n = 675) (HR, 1.70; 95% confidence interval, 0.91-3.17).

CONCLUSION

Patients undergoing primary RTSA for fractures have a substantially higher likelihood of undergoing revision within the first year following the procedure than patients with degenerative conditions preoperatively. Although RTSA is regarded as a reliable and safe treatment option for fractures, surgeons should inform patients accordingly and incorporate this information in decision making when opting for head replacement surgery. There were no differences in patient-reported outcomes between the 2 groups and no differences in revision rates between conventional and fracture-specific stem designs.

Heart failure risk in patients with atrial fibrillation treated with catheter ablation vs antiarrhythmic drugs

Background

Atrial fibrillation (AF) increases heart failure (HF) risk. Whereas the risk of HF-related hospitalization and mortality are known in the setting of AF, the impact of AF treatment on HF development is understudied.

Objective

The purpose of this study was to compare HF incidence among AF patients treated with antiarrhythmic drugs (AADs) vs catheter ablation (CA).

Methods

AF patients with 1 prior AAD usage were identified in 2014-2022 Optum Clinformatics database. Patients were classified into 2 cohorts: those receiving CA vs those receiving a different AAD prescription. The 2 cohorts were matched on sociodemographic and clinical covariates using propensity score matching technique. Cox regression model was used to compare incident HF risk in the 2 cohorts. Subgroup analyses were performed by race/ethnicity, sex, AF subtype, and CHA2DS2-VASc score.

Results

After matching, 9246 patients were identified in each cohort (AAD and CA). Patients receiving CA had a 57% lower risk of incident HF than those treated with AADs (hazard ratio [HR] 0.43; 95% confidence interval [CI] 0.40-0.46). Subgroup analysis by race/ethnicity depicted similar results, with non-Hispanic White (HR 0.43; 95% CI 0.40-0.46), non-Hispanic Black (HR 0.46; 95% CI 0.35-0.60), Hispanic (HR 0.53; 95% CI 0.40-0.70), and Asian (HR 0.46; 95% CI 0.24-0.92) patients treated with CA (vs AAD) having significantly lower risk of HF, respectively. The effect size of CA remained significant in subgroups defined by sex, AF subtypes, and CHA2DS2-VASc score.

Conclusion

AF patients receiving CA had ∼57% lower risk of developing HF than those receiving AAD. The lower risk of HF associated with CA vs AAD persisted across different race/ethnicity, sex, AF subtypes, and CHA2DS2-VASc score.

Idiopathic Ileal Ulceration After Intestinal Transplantation

Background

Idiopathic ileal ulceration after intestinal transplantation (ITx) has been discussed infrequently and has an uncertain natural history and relation to graft rejection. Herein, we review our experience with this pathology.

Methods

We retrospectively reviewed 225 ITx in 217 patients with minimum 1 y graft survival. Routine graft endoscopy was conducted up to twice weekly within the first 90 d after ITx, gradually decreasing to once yearly. Risks for ulceration over time were evaluated using Cox regression.

Results

Of 93 (41%) patients with ulcers, 50 were found within 90 d after ITx mostly via ileoscopy; delayed healing after biopsy appeared causal in the majority. Of the remaining 43 patients with ulcers found >90 d after ITx, 36 were after ileostomy closure. Multivariable modeling demonstrated within 90-d ulcer associations with increasing patient age (hazard ratio [HR], 1.027; P < 0.001) and loop ileostomy (versus Santulli ileostomy; HR, 0.271; P < 0.001). For ulcers appearing after ileostomy closure, their sole association was with absence of graft colon (HR, 7.232; P < 0.001). For ulcers requiring extended anti-microbial and anti-inflammatory therapy, associations included de novo donor-specific antibodies (HR, 3.222; P < 0.007) and nucleotide oligomerization domain mutations (HR, 2.772; P < 0.016). Whole-cohort post-ITx ulceration was not associated with either graft rejection (P = 0.161) or graft failure (P = 0.410).

Conclusions

Idiopathic ulceration after ITx is relatively common but has little independent influence on outcome; risks include ileostomy construction, colon-free ITx, immunologic mutation, and donor sensitization.

Prediction of 30-day unplanned hospital readmission through survival analysis

Background and Objective

Unplanned hospital readmissions are a severe and recurrent problem that affects all health systems. Estimating the risk of being readmitted the following days after discharge is difficult since many heterogeneous factors can influence this. The extensive work concerning this problem proposes solutions mostly based on classification machine-learning models. Survival analysis methods could make a better match with the assessment of readmission risk and are yet to become well-established in this field.

Methods

We compare different statistical and machine learning survival analysis models trained with right-censored all-cause hospital admission data with covariates available at the moment of discharge. The main focus is on tree-ensemble regression methods based on the assumption of proportional hazards. These models are more thoroughly evaluated at a 30-day time period after discharge, although the actual prediction could be set to any time up to 90 days.

Results

The mean performance obtained by each of the proposed survival models ranges from 0.707 to 0.716 C-Index and 0.709 to 0.72 ROC-AUC at a 30-day time period after discharge. The model with the lower performance on both metrics was Cox Proportional Hazards, while the model marking the upper end on both ranges is an XGBoost Regression model with a Cox objective function.

Conclusions

Our findings indicate that survival models perform well addressing the hospital readmission problem, machine-learning models getting the edge over statistical methods. There seems to be an improvement over classification models when attempting to predict at a 30-day period since discharge, perhaps due to a better handling of cases nearing the 30-day boundary. Some preprocessing steps, such as limiting the observation period to 90 days after discharge, are also highlighted since they resulted in a performance boost.

Impaired Renal Function and Major Cardiovascular Events in Young Adults

BACKGROUND

Cardiovascular (CV) disease in young adults (aged 18-39 years) is on the rise. Whether subclinical reductions in kidney function (ie, estimated glomerular filtration rate [eGFR] above the current threshold for chronic kidney disease but below age-expected values) are associated with elevated CV risk is unknown.

OBJECTIVES

The goal of this study was to examine age-specific associations of subclinical eGFR reductions in young adults with major adverse cardiovascular events (MACEs) and MACE plus heart failure (MACE+).

METHODS

A retrospective cohort study of 8.7 million individuals (3.6 million aged 18-39 years) was constructed using linked provincial health care data sets from Ontario, Canada (January 2008-March 2021). Cox models were used to examine the association of categorized eGFR (50-120 mL/min/1.73 m2) with MACE (first of CV mortality, acute coronary syndrome, and ischemic stroke) and MACE+, stratified according to age (18-39, 40-49, and 50-65 years).

RESULTS

In the study cohort (mean age 41.3 years; mean eGFR 104.2 mL/min/1.73 m2; median follow-up 9.2 years), a stepwise increase in the relative risk of MACE and MACE+ was observed as early as eGFR <80 mL/min/1.73 m2 in young adults (eg, for MACE, at eGFR 70-79 mL/min/1.73 m2, ages 18-30 years: 2.37 events per 1,000 person years [HR: 1.31; 95% CI: 1.27-1.40]; ages 40-49 years: 6.26 events per 1,000 person years [HR: 1.09; 95% CI: 1.06-1.12]; ages 50-65 years: 14.9 events per 1,000 person years [HR: 1.07; 95% CI: 1.05-1.08]). Results persisted for each MACE component and in additional analyses (stratifying according to past CV disease, accounting for albuminuria at index, and using repeated eGFR measures).

CONCLUSIONS

In young adults, eGFR below age-expected values were associated with an elevated risk for MACE and MACE+, warranting age-appropriate risk stratification, proactive monitoring, and timely intervention.

Time to Recovery from COVID-19 and Its Predictors in Patients Hospitalized at Tibebe Ghion Specialized Hospital Care and Treatment Center, A Retrospective Follow-Up Study, North West Ethiopia

Background

Since the end of 2019, the world has been facing a new coronavirus disease 19 (COVID-19), which is considered a global pandemic. COVID-19 is considered a major public health burden due to the uncontrolled morbidity and mortality of the global community. The World Health Organization estimates the recovery time as 2 weeks for patients with mild infection and 3 to 6 weeks for those with serious illnesses. The recovery time and its predictors are not well studied in Ethiopia yet. Therefore, the aim of this study was to estimate time to recovery from COVID-19 and its predictors among COVID-19 patients admitted to Tibebe Ghion Specialized Hospital care and treatment center, North West Ethiopia.

Methods

An institution-based retrospective follow-up study was conducted among 452 COVID-19 patients admitted to Tibebe Ghion Specialized Hospital from March 2020 to September 2021. Simple random sampling using a table of random number generators was used to select study units. Data entry and analysis were performed using EpiData 3.1 and Stata version 14, respectively. Bivariable and multivariable Cox proportional hazard analyses were used to identify predictors of recovery time. An AHR at a 5% level of significance was used to identify significant predictors.

Results

: Among 452 COVID-19 patients, 437 (88%) were recovered, with a median recovery time of 9 days. Recovery time was significantly related to age (AHR = 0.98; 95% CI = 0.97, 0.99), oxygen saturation (AHR = 0.42; 95% CI = 0.31, 0.56), shortness of breath (AHR = 0.65; 95% CI = 0.47, 0.85), disease severity (moderate (AHR = 0.63; 95% CI = 0.47, 0.85) and severe (AHR = 0.32; 95% CI = 0.22, 0.47)), and comorbidities (AHR = 0.67; 95% CI = 0.53, 0.84). Conclusions and recommendations: The overall median recovery time was 9 days. Older age, low oxygen saturation, shortness of breath, disease severity (moderate and severe), history of comorbidities, and high-level of WBC were predictors of delayed recovery time. On the other hand, corticosteroid use significantly shortens the median recovery time of COVID-19 patients. Thus, patients presented with older age, low oxygen saturation, shortness of breath, moderate and severe COVID-19 disease, comorbidities, and increased WBC need to be closely monitoring and followed up by healthcare providers. In addition, there should be special attention during the administration of corticosteroid.

DoSurvive: A webtool for investigating the prognostic power of a single or combined cancer biomarker

We present DoSurvive, a user-friendly survival analysis web tool and a cancer prognostic biomarker centered database. DoSurvive is the first database that allows users to perform multivariant survival analysis for cancers with customized gene/patient list. DoSurvive offers three survival analysis methods, Log rank test, Cox regression and accelerated failure time model (AFT), for users to analyze five types of quantitative features (mRNA, miRNA, lncRNA, protein and methylation of CpG islands) with four survival types, i.e. overall survival, disease-specific survival, disease-free interval, and progression-free interval, in 33 cancer types. Notably, the implemented AFT model provides an alternative method for genes/features which failed the proportional hazard assumption in Cox regression. With the unprecedented number of survival models implemented and high flexibility in analysis, DoSurvive is a unique platform for the identification of clinically relevant targets for cancer researcher and practitioners. DoSurvive is freely available at http://dosurvive.lab.nycu.edu.tw/.

Time-Varying Hazard of Patient Falls in Hospital: A Retrospective Case-Control Study

This study aims to evaluate the association between patient falls and relevant factors and to quantify their effect on fall risk. This is a retrospective case-control study using the secondary data collected from a tertiary general hospital. Study subjects were 450 patients who were admitted to the hospital between January 2016 and December 2020. The prevalence of falls was associated with the fall risk level by the Morse Fall Scale (MFS) and individual status at admission including history of admission, dizziness, sleep disorder, bowel dysfunction, and urinary incontinence. The odds ratios of patient falls were higher in the low-risk group by the MFS score (odds ratio (OR) = 2.61, p < 0.001) and the high-risk group (OR = 5.51, p < 0.001) compared to the no-risk group. The hazard ratio of patient falls was higher in the high-risk group by the MFS score (hazard ratio (HR) = 3.85, p < 0.001). The MFS had a significant explanatory power to predict fall risk. Sleep disorder and urinary incontinence were the significant factors influencing patient falls.

Glutamine and amino acid metabolism as a prognostic signature and therapeutic target in endometrial cancer

INTRODUCTION

Endometrial cancer (EC) is the most common female reproductive system cancer in developed countries with growing incidence and associated mortality, which may be due to the growing prevalence of obesity. Metabolism reprogramming including glucose, amino acid, and lipid remodeling is a hallmark of tumors. Glutamine metabolism has been reported to participate in tumor proliferation and development. This study aimed to develop a glutamine metabolism-related prognostic model for EC and explore potential targets for cancer treatment.

METHOD

Transcriptomic data and survival outcome of EC were retrieved from The Cancer Genome Atlas (TCGA). Differentially expressed genes related to glutamine metabolism were recognized and utilized to build a prognostic model by univariate and multivariate Cox regressions. The model was confirmed in the training, testing, and the entire cohort. A nomogram combing prognostic model and clinicopathologic features was established and tested. Moreover, we explored the effect of a key metabolic enzyme, PHGDH, on the biological behavior of EC cell lines and xenograft model.

RESULTS

Five glutamine metabolism-related genes, including PHGDH, OTC, ASRGL1, ASNS, and NR1H4, were involved in prognostic model construction. Kaplan-Meier curve suggested that patients recognized as high risk underwent inferior outcomes. The receiver operating characteristic (ROC) curve showed the model was sufficient to predict survival. Enrichment analysis recognized DNA replication and repair dysfunction in high-risk patients whereas immune relevance analysis revealed low immune scores in the high-risk group. Finally, a nomogram integrating the prognostic model and clinical factors was created and verified. Further, knockdown of PHGDH showed cell growth inhibition, increasing apoptosis, and reduced migration. Promisingly, NCT-503, a PHGDH inhibitor, significantly repressed tumor growth in vivo (p = 0.0002).

CONCLUSION

Our work established and validated a glutamine metabolism-related prognostic model that favorably evaluates the prognosis of EC patients. DNA replication and repair may be the crucial point that linked glutamine metabolism, amino acid metabolism, and EC progression. High-risk patients stratified by the model may not be sufficient for immune therapy. PHGDH might be a crucial target that links serine metabolism, glutamine metabolism as well as EC progression.

Associations between modest reductions in kidney function and adverse outcomes in young adults: retrospective, population based cohort study

OBJECTIVE

To study age specific associations of modest reductions in estimated glomerular filtration rate (eGFR) with adverse outcomes.

DESIGN

Retrospective, population based cohort study.

SETTING

Linked healthcare administrative datasets in Ontario, Canada.

PARTICIPANTS

Adult residents (18-65 years) with at least one outpatient eGFR value (categorized in 10 unit increments from 50 mL/min/1.73m2 to >120 mL/min/1.73m2), with no history of kidney disease.

MAIN OUTCOME MEASURES

eGFRs and hazard ratios of composite adverse outcome (all cause mortality, any cardiovascular event, and kidney failure) stratified by age (18-39 years, 40-49 years, and 50-65 years), and relative to age specific eGFR referents (100-110 mL/min/1.73m2) for ages 18-39 years, 90-100 for 40-49 years, 80-90 for 50-65 years).

RESULTS

From 1 January 2008 to 31 March 2021, among 8 703 871 adults (mean age 41.3 (standard deviation 13.6) years; mean index eGFR 104.2 mL/min/1.73m2 (standard deviation 16.1); median follow-up 9.2 years (interquartile range 5.7-11.4)), modestly reduced eGFR measurements specific to age were recorded in 18.0% of those aged 18-39, 18.8% in those aged 40-49, and 17.0% in those aged 50-65. In comparison with age specific referents, adverse outcomes were consistently higher by hazard ratio and incidence for ages 18-39 compared with older groups across all eGFR categories. For modest reductions (eGFR 70-80 mL/min/1.73m2), the hazard ratio for ages 18-39 years was 1.42 (95% confidence interval 1.35 to 1.49), 4.39 per 1000 person years; for ages 40-49 years was 1.13 (1.10 to 1.16), 9.61 per 1000 person years; and for ages 50-65 years was 1.08 (1.07 to 1.09), 23.4 per 1000 person years. Results persisted for each individual outcome and in many sensitivity analyses.

CONCLUSIONS

Modest eGFR reductions were consistently associated with higher rates of adverse outcomes. Higher relative hazards were most prominent and occurred as early as eGFR <80 mL/min/1.73m2 in younger adults, compared with older groups. These findings suggest a role for more frequent monitoring of kidney function in younger adults to identify individuals at risk to prevent chronic kidney disease and its complications.

Leisure-time physical activity, desire to increase physical activity, and mortality: A population-based prospective cohort study

The aim was to investigate associations between leisure-time physical activity (LTPA) and mortality, and associations between desire to increase LTPA and mortality within the low LTPA group. A public health survey questionnaire was sent in 2008 to a stratified random sample of the population aged 18-80 in southernmost Sweden, yielding a 54.1% response rate. Baseline 2008 survey data with 25,464 respondents was linked to cause of death register data to create a prospective cohort with 8.3-year follow-up. Associations between LTPA, desire to increase LTPA and mortality were analyzed in logistic regression models. An 18.4% proportion performed regular exercise (at least 90 min/week, leading to sweating), 23.2% moderate regular exercise (once or twice a week at least 30 min/occasion, leading to sweating), 44.3% moderate exercise (more than two hours walking or equivalent activity/week) and 14.1% reported low LTPA (less than two hours walking or equivalent activity/week). These four LTPA groups were significantly associated with covariates included in the multiple analyses. The results showed significantly higher all-cause, cardiovascular (CVD), cancer and other cause mortality for the low LTPA group but not for the moderate regular exercise and moderate exercise groups compared to the regular exercise group. Both the "Yes, but I need support" and the "No" fractions within the low LTPA group had significantly increased ORs of all-cause mortality compared to the "Yes, and I can do it myself" reference, while no significant associations were observed for CVD mortality. Physical activity promotion is particularly warranted in the low LTPA group.

A proposal to analyze the progression of non-dialytic chronic kidney disease by surrogate endpoints: introducing parametric survival models

Introduction

Chronic kidney disease (CDK) progression studies increasingly use surrogate endpoints based on the estimated glomerular filtration rate. The clinical characteristics of these endpoints bring new challenges in comparing groups of patients, as traditional Cox models may lead to biased estimates mainly because they do not assume a hazard function.

Objective

This study proposes the use of parametric survival analysis models with the three most commonly used endpoints in nephrology based on a case study. Estimated glomerular filtration rate (eGFR) decay > 5 mL/year, eGFR decline > 30%, and change in CKD stage were evaluated.

Method

The case study is a 5-year retrospective cohort study that enrolled 778 patients in the predialysis stage. Exponential, Weibull, Gompertz, lognormal, and logistic models were compared, and proportional hazard and accelerated failure time (AFT) models were evaluated.

Results

The endpoints had quite different hazard functions, demonstrating the importance of choosing appropriate models for each. AFT models were more suitable for the clinical interpretation of the effects of covariates on these endpoints.

Conclusion

Surrogate endpoints have different hazard distributions over time, which is already recognized by nephrologists. More flexible analysis techniques that capture these relevant clinical characteristics in decision-making should be encouraged and disseminated in nephrology research.

Probabilistic survival modeling in health research: an assessment using cohort data from hospitalized patients with COVID-19 in a Latin American city

Probabilistic survival methods have been used in health research to analyze risk factors and adverse health outcomes associated with COVID-19. The aim of this study was to employ a probabilistic model selected among three distributions (exponential, Weibull, and lognormal) to investigate the time from hospitalization to death and determine the mortality risks among hospitalized patients with COVID-19. A retrospective cohort study was conducted for patients hospitalized due to COVID-19 within 30 days in Londrina, Brazil, between January 2021 and February 2022, registered in the database for severe acute respiratory infections (SIVEP-Gripe). Graphical and Akaike Information Criterion (AIC) methods were used to compare the efficiency of the three probabilistic models. The results from the final model were presented as hazard and event time ratios. Our study comprised of 7,684 individuals, with an overall case fatality rate of 32.78%. Data suggested that older age, male sex, severe comorbidity score, intensive care unit admission, and invasive ventilation significantly increased risks for in-hospital mortality. Our study highlights the conditions that confer higher risks for adverse clinical outcomes attributed to COVID-19. The step-by-step process for selecting appropriate probabilistic models may be extended to other investigations in health research to provide more reliable evidence on this topic.

Outcomes and Trends of Treatments in High-Risk Differentiated Thyroid Cancer

OBJECTIVES

To analyze the variant-specific survival benefits and usage patterns of standardized treatment combinations of surgery (S), radioactive iodine ablation (RAI), and thyroid-stimulating hormone suppression therapy (THST) for high-risk differentiated thyroid cancer.

STUDY DESIGN

Retrospective cohort study.

SETTING

National Cancer Database.

METHODS

The 2004-2017 National Cancer Database was queried for patients receiving definitive surgery for high-risk papillary, follicular, or Hurthle cell thyroid cancer. Cox proportional hazards and Kaplan-Meier analyses assessed for treatment-associated survival.

RESULTS

Of 21,076 cases, 18,214 underwent survival analysis with a mean ± SD age of 50.6 ± 17.1 years (71.3% female). When compared with surgery alone, S + RAI was associated with reduced mortality in papillary (hazard ratio [HR], 0.574; P < .001) and follicular (HR, 0.489; P = .004) thyroid cancer. S + RAI + THST was associated with reduced mortality in papillary (HR, 0.514; P < .001), follicular (HR, 0.602; P = .016), and Hurthle cell (HR, 0.504; P = .021) thyroid cancer. In papillary thyroid cancer, S + RAI (91.3%), S + THST (89.2%), and S + RAI + THST (92.7%) were associated with higher 5-year overall survival rates than surgery (85.4%, all P < .001). Papillary thyroid cancer treatments involving THST were associated with higher 5-year overall survival rates than corresponding regimens without THST (all P < .001). In follicular thyroid cancer, S + RAI (73.9%) and S + RAI + THST (78.7%) were associated with higher 5-year overall survival rates than surgery (65.6%, all P < .05). In Hurthle cell thyroid cancer, S + RAI (66.5%) and S + RAI + THST (73.4%) were associated with higher 5-year overall survival rates than surgery (53.7%, all P < .05). On linear regression, THST usage increased by 77.5% (R² = 0.944, P < .001), while RAI usage declined by 11.3% (R² = 0.320, P = .035).

CONCLUSIONS

High-risk differentiated thyroid cancer exhibited varying susceptibilities to different treatment combinations depending on histology, with greatest responses to regimens that included RAI. Physician practices have trended toward decreased RAI and increased THST usage.

Survival analyses and their applications in orthopaedics

Survival analyses are a powerful statistical tool used to analyse data when the outcome of interest involves the time until an event. There is an array of models fit for this goal; however, there are subtle differences in assumptions, as well as a number of pitfalls, that can lead to biased results if researchers are unaware of the subtleties. As larger amounts of data become available, and more survival analyses are published every year, it is important that healthcare professionals understand how to evaluate these models and apply them into their practice. Therefore, the purpose of this study was to present an overview of survival analyses, including required assumptions and important pitfalls, as well as examples of their use within orthopaedic surgery.

Higher center volume is significantly associated with lower mortality in trauma patients with shock

INTRODUCTION

Injured patients presenting in shock are at high risk of mortality despite numerous efforts to improve resuscitation. Identifying differences in outcomes among centers for this population could yield insights to improve performance. We hypothesized that trauma centers treating higher volumes of patients in shock would have lower risk-adjusted mortality.

METHODS

We queried the Pennsylvania Trauma Outcomes Study from 2016 to 2018 for injured patients ≥16 years of age at Level I&II trauma centers who had an initial systolic blood pressure (SBP) of <90 mmHg. We excluded patients with critical head injury (abbreviated injury score [AIS] head ≥5) and patients coming from centers with a shock patient volume of ≤10 for the study period. The primary exposure was tertile of center-level shock patient volume (low, medium, or high volume). We compared risk-adjusted mortality by tertile of volume using multivariable Cox proportional hazards model incorporating age, injury severity, mechanism, and physiology.

RESULTS

Of 1,805 included patients at 29 centers, 915 (50.7%) died. The median annual shock trauma patient volume was 9 patients for low volume centers, medium 19.5, and high 37. Median ISS was higher at high volume compared to low volume centers (22 vs 18, p <0.001). Raw mortality was 54.9% at high volume centers, 46.7% for medium, and 42.9% for low. Time elapsed from arrival to emergency department (ED) to the operating room (OR) was lower at high volume than low volume centers (median 47 vs 78 min) p = 0.003. In adjusted analysis, hazard ratio for high volume centers (referenced to low volume) was 0.76 (95% CI 0.59-0.97, p = 0.030).

CONCLUSION

After adjusting for patient physiology and injury characteristics, center-level volume is significantly associated with mortality. Future studies should seek to identify key practices associated with improved outcomes in high-volume centers. Furthermore, shock patient volume should be considered when new trauma centers are opened.

Impact of Point-of-Care Testing on the Management of Sexually Transmitted Infections in South Africa: Evidence from the HVTN702 Human Immunodeficiency Virus Vaccine Trial

BACKGROUND

Alternative approaches to syndromic management are needed to reduce rates of sexually transmitted infections (STIs) in resource-limited settings. We investigated the impact of point-of-care (POC) versus central laboratory-based testing on STI treatment initiation and STI adverse event (STI-AE) reporting.

METHODS

We used Kaplan-Meier and Cox regression models to compare times to treatment initiation and STI-AE reporting among HVTN702 trial participants in South Africa. Neisseria gonorrhoeae (NG) and Chlamydia trachomatis (CT) were diagnosed POC at eThekwini clinic and in a central laboratory at Verulam/Isipingo clinics. All clinics used POC assays for Trichomonas vaginalis (TV) testing.

RESULTS

Among 959 women (median age, 23 [interquartile range, 21-26] years), median days (95% confidence interval [95%CI]) to NG/CT treatment initiation and NG/CT-AE reporting were 0.20 (.16-.25) and 0.24 (.19-.27) at eThekwini versus 14.22 (14.12-15.09) and 15.12 (13.22-21.24) at Verulam/Isipingo (all P < .001). Median days (95%CI) to TV treatment initiation and TV-AE reporting were 0.17 (.12-.27) and 0.25 (.20-.99) at eThekwini versus 0.18 (.15-.2) and 0.24 (.15-.99) at Verulam/Isipingo (all P > .05). Cox regression analysis revealed that NG/CT treatment initiation (adjusted hazard ratio [aHR], 39.62 [95%CI, 15.13-103.74]) and NG/CT-AE reporting (aHR, 3.38 [95%CI, 2.23-5.13]) occurred faster at eThekwini versus Verulam/Isipingo, while times to TV treatment initiation (aHR, 0.93 [95%CI, .59-1.48]) and TV-AE reporting (aHR, 1.38 [95%CI, .86-2.21]) were similar.

CONCLUSIONS

POC testing led to prompt STI management with potential therapeutic and prevention benefits, highlighting its utility as a diagnostic tool in resource-limited settings.

Long-term survival and second malignant tumor prediction in pediatric, adolescent, and young adult cancer survivors using Random Survival Forests: a SEER analysis

Survival and second malignancy prediction models can aid clinical decision making. Most commonly, survival analysis studies are performed using traditional proportional hazards models, which require strong assumptions and can lead to biased estimates if violated. Therefore, this study aims to implement an alternative, machine learning (ML) model for survival analysis: Random Survival Forest (RSF). In this study, RSFs were built using the U.S. Surveillance Epidemiology and End Results to (1) predict 30-year survival in pediatric, adolescent, and young adult cancer survivors; and (2) predict risk and site of a second tumor within 30 years of the first tumor diagnosis in these age groups. The final RSF model for pediatric, adolescent, and young adult survival has an average Concordance index (C-index) of 92.9%, 94.2%, and 94.4% and average time-dependent area under the receiver operating characteristic curve (AUC) at 30-years since first diagnosis of 90.8%, 93.6%, 96.1% respectively. The final RSF model for pediatric, adolescent, and young adult second malignancy has an average C-index of 86.8%, 85.2%, and 88.6% and average time-dependent AUC at 30-years since first diagnosis of 76.5%, 88.1%, and 99.0% respectively. This study suggests the robustness and potential clinical value of ML models to alleviate physician burden by quickly identifying highest risk individuals.

Contraceptive Continuation and Experiences Obtaining Implant and IUD Removal Among Women Randomized to Use Injectable Contraception, Levonorgestrel Implant, and Copper IUD in South Africa and Zambia

Few longitudinal studies have measured contraceptive continuation past one year in sub-Saharan Africa. We surveyed 674 women who had been randomized to receive the three-month intramuscular contraceptive injectable (DMPA-IM), levonorgestrel (LNG) implant, or copper intrauterine device (IUD) during the Evidence for Contraceptive Options and HIV Outcomes (ECHO) trial in South Africa and Zambia and were subsequently followed for two additional years to explore method continuation, reasons for discontinuation, and access to implant and IUD removal services. We also conducted in-depth qualitative interviews with 39 participants. We estimated cumulative discontinuation probabilities using Kaplan-Meier estimates and assessed factors associated with discontinuation using Cox-proportional hazards models. The LNG implant continuation rate over the maximum 44-month study period was 60 percent, while rates for the copper IUD and DMPA-IM were 52 percent and 44 percent, respectively. Reasons for method discontinuation included side effects, particularly menstrual changes, and method stock-outs. Most implant and IUD users who sought removal were able to access services; however, room for improvement exists. In this cohort originally randomized to receive a contraceptive method and attend regular study visits, implants and IUDs continued to be highly acceptable over an additional two years, but facilities should continue to ensure that insertions and removals are available as requested.

Determining the Proportionality of Ischemic Stroke Risk Factors to Age

While age is an important risk factor, there are some disadvantages to including it in a stroke risk model: age can dominate the risk score and lead to over- or under-predictions in some age groups. There is evidence to suggest that some of these disadvantages are due to the non-proportionality of other risk factors with age, e.g., risk factors contribute differently to stroke risk based on an individual's age. In this paper, we present a framework to test if risk factors are proportional with age. We then apply the framework to a set of risk factors using Framingham heart study data from the NHLBI Biologic Specimen and Data Repository Information Coordinating Center to determine if we can find evidence of non-proportionality. Using our framework, we find that a number of risk factors (diastolic blood pressure, total cholesterol, BMI, sex, high blood pressure treatment) may be non-proportional to age. This suggests that testing for the proportionality of risk factors with age should be something that is considered in stroke risk prediction modelling and traditional modelling methods may need to be adjusted to capture this non-proportionality.

Mental health outcomes in parents of children with a cancer diagnosis in Sweden: A nationwide cohort study

BACKGROUND

The diagnosis of paediatric cancer is a crisis for the parents who are the primary caregivers of the affected child. A comprehensive assessment of the longitudinal impact of childhood cancer on parental mental health and the potential sex differences between the parents is lacking. Thus, we aimed to explore the subsequent short- and long-term mental health outcomes among the parents of children with cancer and examine whether the outcomes vary between the mother and father.

METHODS

By combining several Swedish registers, parents of a child (ages 0-14 years) with a cancer diagnosis between Jan 1, 2006, and Dec 31, 2016 were identified. For each parent of children with cancer, up to five mothers or fathers of cancer-free children were randomly selected and matched, respectively. Hospital contacts for any mental health disorders between 5 years before and 7 years after the diagnosis of childhood cancer were retrieved. An interrupted time series negative binomial regression was performed to assess the short- and long-term impact of a childhood cancer diagnosis on the parents' subsequent mental health outcomes.

FINDINGS

16,199 mothers (2852 with a child with cancer and 13,347 without) and 15,708 fathers (2769 with a child with cancer and 12,939 without) were included in this study. Compared with mothers of children without cancer, mothers of children with cancer had higher risks of mental health disorders in the first year after diagnosis (rate ratio [RR] and 95% Confidence Interval [CI], 1.17 (1.03-1.32)), and notably, the adverse impact became more severe over time (RR and 95% CI, 1.36 (1.07-1.74), in the seventh year). For fathers of children with cancer, the risk of mental health disorders was continuously higher compared to matched comparisons (RR and 95% CI, 1.31 (1.01-1.71)).

INTERPRETATION

Our findings suggested that parental mental health was affected continuously by a diagnosis of childhood cancer in their children. In particular, the mother's mental health was affected more severely. Customised psychological services or interventions are highly needed for the parents of children with cancer.

FUNDING

Swedish Research Council, Allmänna Sjukhusets i Malmö Stiftelsen för bekämpande av cancer, Swedish Heart-Lung Foundation, ALF funding from Region Skåne and China Scholarship Council.

Psychiatric comorbidities and their treatment predict buprenorphine continuation among postpartum people with opioid use disorder

Background

Opioid use disorder (OUD) is a growing crisis among pregnant and postpartum people. Psychiatric comorbidities are common, yet how they impact OUD treatment outcomes is not well characterized. The aim of this study was to assess the association of psychiatric comorbidities and receipt of psychiatric treatment with buprenorphine continuation through one year postpartum among a sample of people with OUD.

Methods

A subsample was identified from a larger retrospective cohort of patients receiving buprenorphine for OUD at the time of delivery from an academic medical center between 2017 and 2020. Medical record abstractions were conducted during pregnancy through one year postpartum. Independent variables included any psychiatric diagnosis and postpartum receipt of psychiatric treatment (medication or behavioral health). The primary outcome was week of buprenorphine discontinuation. Cox Proportional Hazard models were used.

Results

Of 138 patients, 71.8% had a psychiatric condition and 35.5% continued buprenorphine for a full year postpartum. Postpartum buprenorphine continuation was associated with (a) Psychiatric co-morbidity (buprenorphine discontinuation HR 0.49; 95% CI 0.29, 0.82), (b) Receipt of psychiatric medications in weeks 39-52 postpartum (buprenorphine discontinuation HR 0.21; 95% CI 0.06, 0.83), and (c) Receipt of behavioral health therapy in weeks 9-38 postpartum (buprenorphine discontinuation HR 0.40; 95% CI 0.18, 0.90).

Conclusion

Our work suggests a dynamic relationship between OUD treatment outcomes, psychiatric comorbidities and receipt of psychiatric treatments through the highly vulnerable postpartum period. Clinicians and researchers alike should work to advance patient-centered engagement in integrated care models tailored for this unique population.

Survival time to complications of congestive heart failure patients at Felege Hiwot comprehensive specialized referral hospital, Bahir Dar, Ethiopia

BACKGROUNDS

Congestive heart failure is a serious chronic condition when the heart's muscles become too damaged and a condition in which one or both ventricles cannot pump sufficient blood to meet the metabolic needs of the body. This study aimed to identify factors affecting the complications time of congestive heart failure patients treated from January 2016 to December 2019 at Felege Hiwot comprehensive specialized referral hospital in Bahir Dar, Ethiopia.

METHODS

A hospital-based retrospective data collection was collected from the medical charts of 218 randomly selected congestive heart failure patients. The Kaplan-Meier curve and the Cox proportional hazards model were used to compare and identify the factors associated with time to complication in patients with congestive heart failure.

RESULTS

The median complication time of congestive heart failure patients was 22 months [95% CI: 21.98-28.01]. About 194 (88.99%) of the patients were complicated. The Kaplan-Meier curve depicts the survival probability of complicated patients decreasing as the complication time increases. The hazard ratios for serum sodium concentration, left ventricular ejection fraction, patients from rural areas, age of patients, serum hemoglobin concentration, and New York heart association classes I, II, and III were given 0.94 [95% CI: 0.90-1.00], 0.74 [95% CI: 0.65-0.85], 0.75 [95% CI: 0.68-0.84], 1.28 [95% CI: 1.12-1.46], 0.89 [95% CI: 0.85-0.94], 0.44 [95% CI: 0.36-0.53], 0.54 [95% CI: 0.47-0.62] and 0.73 [95% CI: 0.65-0.81] respectively, and they are statistically associated with the complication time of congestive heart failure patients.

CONCLUSIONS

The median complication time of congestive heart failure patients was 22 months. This study strongly suggests that healthcare awareness should be strengthened earlier about the potential complications for patients with lower serum sodium concentrations below the threshold and aged congestive heart failure patients to reduce the risk of developing complications.

Two subtypes of cutaneous melanoma with distinct mutational signatures and clinico-genomic characteristics

Background: To decipher mutational signatures and their associations with biological implications in cutaneous melanomas (CMs), including those with a low ultraviolet (UV) signature.

Materials and Methods: We applied non-negative matrix factorization (NMF) and unsupervised clustering to the 96-class mutational context of The Cancer Genome Atlas (TCGA) cohort (N = 466) as well as other publicly available datasets (N = 527). To explore the feasibility of mutational signature-based classification using panel sequencing data, independent panel sequencing data were analyzed.

Results: NMF decomposition of the TCGA cohort and other publicly available datasets consistently found two mutational signatures: UV (SBS7a/7b dominant) and non-UV (SBS1/5 dominant) signatures. Based on mutational signatures, TCGA CMs were classified into two clusters: UV-high and UV-low. CMs belonging to the UV-low cluster showed significantly worse overall survival and landmark survival at 1-year than those in the UV-high cluster; low or high UV signature remained the most significant prognostic factor in multivariate analysis. The UV-low cluster showed distinct genomic and functional characteristic patterns: low mutation counts, increased proportion of triple wild-type and KIT mutations, high burden of copy number alteration, expression of genes related to keratinocyte differentiation, and low activation of tumor immunity. We verified that UV-high and UV-low clusters can be distinguished by panel sequencing.

Conclusion: Our study revealed two mutational signatures of CMs that divide CMs into two clusters with distinct clinico-genomic characteristics. Our results will be helpful for the clinical application of mutational signature-based classification of CMs.

Using machine learning algorithms for predicting cognitive impairment and identifying modifiable factors among Chinese elderly people

Objectives: This study firstly aimed to explore predicting cognitive impairment at an early stage using a large population-based longitudinal survey of elderly Chinese people. The second aim was to identify reversible factors which may help slow the rate of decline in cognitive function over 3 years in the community. Methods: We included 12,280 elderly people from four waves of the Chinese Longitudinal Healthy Longevity Survey (CLHLS), followed from 2002 to 2014. The Chinese version of the Mini-Mental State Examination (MMSE) was used to examine cognitive function. Six machine learning algorithms (including a neural network model) and an ensemble method were trained on data split 2/3 for training and 1/3 testing. Parameters were explored in training data using 3-fold cross-validation and models were evaluated in test data. The model performance was measured by area-under-curve (AUC), sensitivity, and specificity. In addition, due to its better interpretability, logistic regression (LR) was used to assess the association of life behavior and its change with cognitive impairment after 3 years. Results: Support vector machine and multi-layer perceptron were found to be the best performing algorithms with AUC of 0.8267 and 0.8256, respectively. Fusing the results of all six single models further improves the AUC to 0.8269. Playing more Mahjong or cards (OR = 0.49,95% CI: 0.38-0.64), doing more garden works (OR = 0.54,95% CI: 0.43-0.68), watching TV or listening to the radio more (OR = 0.67,95% CI: 0.59-0.77) were associated with decreased risk of cognitive impairment after 3 years. Conclusions: Machine learning algorithms especially the SVM, and the ensemble model can be leveraged to identify the elderly at risk of cognitive impairment. Doing more leisure activities, doing more gardening work, and engaging in more activities combined were associated with decreased risk of cognitive impairment.

Natural Language Processing to Improve Prediction of Incident Atrial Fibrillation Using Electronic Health Records

Background Models predicting atrial fibrillation (AF) risk, such as Cohorts for Heart and Aging Research in Genomic Epidemiology AF (CHARGE-AF), have not performed as well in electronic health records. Natural language processing (NLP) may improve models by using narrative electronic health record text. Methods and Results From a primary care network, we included patients aged ≥65 years with visits between 2003 and 2013 in development (n=32 960) and internal validation cohorts (n=13 992). An external validation cohort from a separate network from 2015 to 2020 included 39 051 patients. Model features were defined using electronic health record codified data and narrative data with NLP. We developed 2 models to predict 5-year AF incidence using (1) codified+NLP data and (2) codified data only and evaluated model performance. The analysis included 2839 incident AF cases in the development cohort and 1057 and 2226 cases in internal and external validation cohorts, respectively. The C-statistic was greater (P<0.001) in codified+NLP model (0.744 [95% CI, 0.735-0.753]) compared with codified-only (0.730 [95% CI, 0.720-0.739]) in the development cohort. In internal validation, the C-statistic of codified+NLP was modestly higher (0.735 [95% CI, 0.720-0.749]) compared with codified-only (0.729 [95% CI, 0.715-0.744]; P=0.06) and CHARGE-AF (0.717 [95% CI, 0.703-0.731]; P=0.002). Codified+NLP and codified-only were well calibrated, whereas CHARGE-AF underestimated AF risk. In external validation, the C-statistic of codified+NLP (0.750 [95% CI, 0.740-0.760]) remained higher (P<0.001) than codified-only (0.738 [95% CI, 0.727-0.748]) and CHARGE-AF (0.735 [95% CI, 0.725-0.746]). Conclusions Estimation of 5-year risk of AF can be modestly improved using NLP to incorporate narrative electronic health record data.

Incidence and Predictors of Calf Morbidity and Mortality From Birth to 6-Months of Age in Dairy Farms of Northwestern Ethiopia

The Ethiopian government has initiatives for expanding the commercial and smallholder market-oriented urban and peri-urban dairy production systems to meet the demands for dairy products. However, there have been only limited on-farm studies on the health performance of commercial dairy breeds. The aim of this longitudinal study was to quantify the incidence and identify predictors of calf morbidity and mortality from birth to 6 months of age in urban and peri-urban dairy farms of Northwest Ethiopia. A total of 439 calves aged below 6 months from 174 dairy farms were included in the study. We collected data on 35 potential risk factors to determine their effect on calf morbidity and mortality in the area. Kaplan-Meier survival analysis was used to summarize survival probability. The Cox proportional hazard regression model with shared frailty to account for unmeasured herd-specific heterogeneity was also used to identify and quantify factors associated with time to morbidity and mortality. Among 439 calves enrolled for 6 months of follow-up period, a total of 141 morbidities and 54 mortality events were recorded. This gives an overall morbidity and mortality incidence rates of 64 per 100-calf 6-months at risk (risk rate of 47.3%) and 19 per 100-calf 6-months at risk (risk rate of 17.9%), respectively. Diarrhea was the most frequent calf health problem with a risk rate of 25.2%. It was the cause of death for 33.3% of all the 54 calf deaths. Next to diarrhea, pneumonia ranked second with risk rate of 8.6% and was responsible for death of 12.9% of all the 54 calf deaths. Among 35 potential risk factors, calf age, vigor status at birth, calf breed, colostrum ingestion, and herd size were significant (p < 0.05) predictors of calf morbidity and mortality. The Cox-shared frailty model revealed that the herd frailty component had no significant effect on hazard estimates of the covariates of all-cause morbidity and mortality. This implies that the dairy herds participated in the study were homogeneous in the distribution of unmeasured random effects. In conclusion, the magnitude of calf morbidity and mortality was higher and above economically tolerable level in this study. This could impede the success of Ethiopia's dairy development initiative in general, and the livelihood of smallholder dairy producers in particular. Therefore, educating farmers aimed at mitigating the identified risk factors can reduce calf morbidity and mortality in the study areas.

Body mass index and all-cause mortality in patients with newly diagnosed type 2 diabetes mellitus in South Korea: a retrospective cohort study

OBJECTIVES

The lower risk of death in overweight or obese patients, compared with normal-weight individuals, has caused confusion for patients with diabetes and healthcare providers. This study investigated the relationship between body mass index (BMI) and mortality in patients with type 2 diabetes.

DESIGN

A retrospective cohort study.

SETTING

We established a national population database by merging the Korea National Health Insurance (KNHI) claims database, the National Health Check-ups Database and the KNHI Qualification Database of South Korea.

PARTICIPANTS

A total of 53 988 patients who were newly diagnosed with type 2 diabetes (E11 in International Classification of Diseases, 10th Edition) in 2007, had available BMI data, lacked a history of any serious comorbidity, received diabetes medication and did not die during the first 2 years were followed up for a median of 8.6 years.

PRIMARY OUTCOME MEASURES

All-cause mortality.

RESULTS

The mean BMI was 25.2 (SD 3.24) kg/m², and the largest proportion of patients (29.4%) had a BMI of 25-27.4 kg/m². Compared with a BMI of 27.5-29.9 kg/m² (the reference), mortality risk continuously increased as BMI decreased while the BMI score was under 25 (BMI <18.5 kg/m²: adjusted HR (aHR) 2.71, 95% CI 2.24 to 3.27; BMI 18.5-20.9 kg/m²: aHR 1.94, 95% CI 1.70 to 2.22; BMI 21-22.9 kg/m²: aHR 1.51, 95% CI 1.34 to 1.70; and BMI 23-24.9 kg/m²: aHR 1.14, 95% CI 1.01 to 1.28). For patients aged ≥65 years, the inverse association was connected up to a BMI ≥30 kg/m² group (aHR 0.76, 95% CI 0.59 to 0.98). However, the associations for men, patients aged <65 years and ever smokers resembled a reverse J curve, with a significantly greater risk of death in patients with a BMI ≥30 kg/m².

CONCLUSIONS

This study suggests that, for patients with type 2 diabetes at a normal weight, distinct approaches are needed in terms of promoting muscle mass improvement or cardiorespiratory fitness, rather than maintaining weight status. Improved early diagnosis considering the inverse association between BMI and mortality is also needed.

Prevention paradox between stroke and multiple potential risk factors using data from a population-based cohort study

Previous studies have found the prevention paradox in the association between stroke events and a single specific risk factor, indicating that a population-based strategy may be more effective than a high-risk-based strategy for prevention. We tested the hypothesis that the prevention paradox does not apply when focusing on multiple potential risk factors simultaneously. The study cohort included 9051 individuals from Japan aged 40-89 years. The time-dependent Cox proportional-hazards models were used to identify the primary risk factor associated with stroke onset. We classified participants based on risk factors in two distinct ways: 1) classifying the high-risk group participants according to a single specific risk factor that had a large association with stroke in both sexes and all ages and 2) classifying the high-risk group participants according to 1-3 risk factor(s) including hypertension, hyperglycemia, and/or dyslipidemia. Then, we compared the proportions of the total number of participants who developed stroke in both groups to assess the prevention paradox. We found that hypertension was a primary risk factor for stroke incidence, regardless of sex and age. The percentage of patients with a single specific risk of and developed stroke was 46%-63%, while the percentage of patients with 1-3 risk factor(s) was 71-83%. This finding leads to the conclusion that the prevention paradox does not hold when multiple stroke risk factors were associated, suggesting that a high-risk-based strategy that focuses on patients with multiple risk factors may be more effective in preventing strokes.

Using fractional polynomials and restricted cubic splines to model non-proportional hazards or time-varying covariate effects in the Cox regression model

The Cox proportional hazards model is used extensively in clinical and epidemiological research. A key assumption of this model is that of proportional hazards. A variable satisfies the proportional hazards assumption if the effect of that variable on the hazard function is constant over time. When the proportional hazards assumption is violated for a given variable, a common approach is to modify the model so that the regression coefficient associated with the given variable is assumed to be a linear function of time (or of log‐time), rather than being constant or fixed. However, this is an unnecessarily restrictive assumption. We describe two different methods to allow a regression coefficient, and thus the hazard ratio, in a Cox model to vary as a flexible function of time. These methods use either fractional polynomials or restricted cubic splines to model the log‐hazard ratio as a function of time. We illustrate the utility of these methods using data on 12 705 patients who presented to a hospital emergency department with a primary diagnosis of heart failure. We used a Cox model to assess the association between elevated cardiac troponin at presentation and the hazard of death after adjustment for an extensive set of covariates. SAS code for implementing the restricted cubic spline approach is provided, while an existing Stata function allows for the use of fractional polynomials.