Modeling non-linear relationships in epidemiological data: The application and interpretation of spline models

Association of Relative Dysglycemia With Hospital Mortality in Critically Ill Patients: A Retrospective Study

OBJECTIVES

Relative dysglycemia has been proposed as a clinical entity among critically ill patients in the ICU, but is not well studied. This study aimed to clarify associations of relative hyperglycemia and hypoglycemia during the first 24 hours after ICU admission with in-hospital mortality and the respective thresholds.

DESIGN

A single-center retrospective study.

SETTING

An urban tertiary hospital ICU.

PATIENTS

Adult critically ill patients admitted urgently between January 2016 and March 2022.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Maximum and minimum glycemic ratio (GR) was defined as maximum and minimum blood glucose values during the first 24 hours after ICU admission divided by hemoglobin A1c-derived average glucose, respectively. Of 1700 patients included, in-hospital mortality was 16.9%. Nonsurvivors had a higher maximum GR, with no significant difference in minimum GR. Maximum GR during the first 24 hours after ICU admission showed a J-shaped association with in-hospital mortality, and a mortality trough at a maximum GR of approximately 1.12; threshold for increased adjusted odds ratio for mortality was 1.25. Minimum GR during the first 24 hours after ICU admission showed a U-shaped relationship with in-hospital mortality and a mortality trough at a minimum GR of approximately 0.81 with a lower threshold for increased adjusted odds ratio for mortality at 0.69.

CONCLUSIONS

Mortality significantly increased when GR during the first 24 hours after ICU admission deviated from between 0.69 and 1.25. Further evaluation will necessarily validate the superiority of personalized glycemic management over conventional management.

Changing trends in clinical characteristics and in-hospital mortality of patients with infective endocarditis over four decades

BACKGROUND

Infective endocarditis continues to be a significant concern and may be undergoing an epidemiological transition.

METHODS

Were studied 1804 consecutive episodes of infective endocarditis between 1978 and 2022. The mean age was 48 ( ± 19), and 1162 (64%) patients were male. Temporal trends in demographic data, comorbidities, predisposing conditions, microorganisms, complications and in-hospital death have been studied over the decades (1978-1988, 1989-1999, 2000-2010 and 2011-2022). The outcomes and clinical characteristics were modeled using nonlinear cubic spline functions.

FINDINGS

Valve surgery was performed in 50% of the patients and overall in-hospital mortality was 30%. From the first to the fourth decade studied, the average age of patients increased from 29 to 57 years (p < 0.001), with significant declines in the occurrence of rheumatic valvular heart disease (15% to 6%; p < 0.001) and streptococcal infections (46% to 33%; p < 0.001). Healthcare-associated infections have increased (9% to 21%; p < 0.001), as have prosthetic valve endocarditis (26% to 53%; p < 0.001), coagulase-negative staphylococcal infections (4% to 11%; p < 0.001), and related-complications (heart failure, embolic events, and perivalvular abscess; p < 0.001). These changes were associated with a decline in adjusted in-hospital mortality from 34% to 25% (p = 0.019).

INTERPRETATION

In the 44 years studied, there was an increase in the mean age of patients, healthcare-related, prosthetic valve, coagulase-negative staphylococci/MRSA infections, and related complications. Notably, these epidemiological changes were associated with a decline in the adjusted in-hospital mortality.

Effects of Miosis on the Visual Acuity Space under Varying Conditions of Contrast and Ambient Luminance in Presbyopia

Background: Presbyopia is an age-related ocular condition, typically affecting individuals aged over 40 years, characterized by a gradual and irreversible decline in the eye's ability to focus on nearby objects. Correction methods for presbyopia encompass the use of corrective lenses, surgical interventions (corneal or lens based), and, more recently, the FDA-approved topical administration of 1.25% pilocarpine. While prior research has demonstrated the efficacy of daily pilocarpine eye drop application in enhancing near visual acuity by increasing the depth of focus leveraging the pinhole effect, limited knowledge exists regarding its influence on visual acuity under varying conditions of contrast and ambient luminance. Methods: This study aims to investigate the impact of these variables on visual acuity, employing the VA-CAL test, among 11 emmetropic and 11 presbyopic volunteers who reported subjective difficulties with near vision. This study includes evaluations under natural conditions with a pinhole occluder (diameter of 2 mm), and subsequent administration of 1% pilocarpine (Pilomann, Bausch + Lomb, Laval, Canada). Results: The VA-CAL results demonstrate the expected, statistically significant effects of contrast and ambient luminance on visual acuity in both emmetropic and presbyopic volunteers. Furthermore, in emmetropic individuals, the application of pilocarpine resulted in a statistically significant reduction in visual acuity. In contrast, presbyopes did not exhibit statistically significant differences in the visual acuity space under either the pinhole or pilocarpine conditions when compared to natural conditions. Conclusions: The pharmacological treatment of presbyopia with pilocarpine eye drops, intended to enhance near vision, does not adversely affect visual acuity in presbyopes. This suggests that pilocarpine may offer a viable alternative for individuals averse to wearing corrective eyewear.

Trends and projection in the proportion of (heavy) cannabis use in Germany from 1995 to 2021

AIMS

To measure the current trends of cannabis use in Germany, measure trends in the proportion of heavy cannabis users and estimate future cannabis use rates.

DESIGN

Repeated waves of the Epidemiological Survey on Substance Abuse, a cross-sectional survey conducted between 1995 and 2021 with a two-stage participant selection strategy where respondents completed a survey on substance use delivered through the post, over the telephone or on-line.

SETTING

Germany.

PARTICIPANTS/CASES

German-speaking participants aged between 18 and 59 years living in Germany who self-reported on their cannabis use in the past 12 months (n = 78 678). With the application of a weighting scheme, the data are nationally representative.

MEASUREMENTS

Questions on the frequency of cannabis use in the past 12 months and self-reported changes in frequency of use due to the COVID-19 pandemic.

FINDINGS

The prevalence of past 12-month cannabis users increased from 4.4% [95% confidence interval (CI) = 3.7, 5.1] in 1995 to 10.0% (95% CI = 8.9, 11.3) in 2021. Modeling these trends revealed a significant increase that accelerated over the past decade. The proportion of heavy cannabis users [cannabis use (almost) daily or at least 200 times per year] among past-year users has remained steady from 1995 (11.4%, 95% CI = 7.7, 16.5) to 2018 (9.5%, 95% CI = 7.6, 11.9), but significantly increased to 15.7% (95% CI = 13.1, 18.8) in 2021 during the COVID-19 pandemic. Extrapolating from these models, the prevalence of 12-month cannabis users in 2024 is expected to range between 10.4 and 15.0%, while the proportion of heavy cannabis users is unclear.

CONCLUSIONS

Trends from 1995 to 2021 suggest that the prevalence of past 12-month cannabis users in Germany will continue to increase, with expected rates between 10.4 and 15.0% for the German-speaking adult population, and that at least one in 10 cannabis users will continue to use cannabis heavily (almost daily or 200 + times in the past year).

Incidence and Clinical Risk Factors of Post-Operative Complications following Primary Total Hip Arthroplasty: A 10-Year Population-Based Cohort Study

Background: Total hip arthroplasty (THA) has become a growing treatment procedure for debilitating hip pathologies. Patients experienced post-operative complications and revision surgeries according to large THA registries. To fully understand the short-term and long-term post-operative outcomes following THA, the purpose of this study is to examine the incidence of post-operative complications following primary THA and to examine how this trend has changed over 10 years within community hospitals in the US using large databases. Methods: This study queried the State Inpatient Database (SID) for primary THA between 2006 and 2015. Individual patients were followed forward in time until the first instance of a post-operative complication. The multivariable logistic regression analyses were computed to examine which post-operative complications were independent predictors of pre-operative comorbidities. Results: Median age of patients was 67 years, and 56% of patients were female. Females with avascular necrosis (AVN) as an indication for THA had a 27% higher risk of complication. Females with osteoarthritis (OA) as an indication for THA had a 6% higher risk of complication. Post-operative complications occurred with higher frequencies in the first two months of THA and the highest risks of THA complications within the first 6 months. Conclusion: The most common indication is OA in elders with primary THA. Females and those of black ethnicity showed the greatest risks of THA complications. Data from our large study can be used to understand post-operative complications and readmissions after THA. Our study also provides data on risk factors associated with these complications.

Dose-response associations of Monitor-Independent Movement Summary with health-related fitness in youth

The purpose of this study was to determine the dose-response associations between Monitor-Independent Movement Summary (MIMS) units and health-related fitness in youth. The sample comprised US children and adolescents who participated in the 2012 National Youth Fitness Survey (NNYFS; N = 1158, 48.9% female). Health-related fitness domains were assessed using tests of cardiorespiratory endurance (timed maximal and graded treadmill tests), muscular strength (modified pull-up and grip tests), and muscular endurance (plank test). Movement data were collected using wrist-worn ActiGraph accelerometers with raw data processed using MIMS and the calculated metrics of average MIMS/day, Peak 60-min MIMS, and Peak 30-min MIMS. Weighted regression models examined linear associations between MIMS metrics and fitness test scores. Nonlinear associations were examined using weighted spline models with knots placed at the 10th, 50th, and 90th percentiles. Models were adjusted for covariates and fit was examined using the coefficient of determination (R2 ). The strongest adjusted linear relationships included a positive association between MIMS/day (per every 1000 units) and maximal endurance times (b = 5.5 s, p < 0.001) and between Peak 60-min MIMS (per every 10 units) and estimated aerobic capacity (b = 1.7 mL/kg/min, p < 0.001), modified pull-ups (b = 0.7 reps, p < 0.001), and plank test scores (b = 5.0 s, p < 0.001). Linear spline models had slightly higher R2 values (R2 range = 16.9%-74.8%) compared to linear models (R2 range = 15.0%-74.5%). The relationship between MIMS metrics and fitness test scores was best modeled as piecewise linear functions. Although all MIMS metrics associated with cardiorespiratory endurance, Peak 60-min MIMS showed stronger associations with tests of muscular strength and endurance.

The association between proton pump inhibitors and hyperparathyroidism: a potential mechanism for increased fracture-results of a large observational cohort study

Proton pump inhibitors (PPIs) are associated with increased risk of osteoporotic fracture; however, the mechanism is unclear. PPI users taking calcium supplements were more likely to have hyperparathyroidism compared to non-users (OR 1.56, CI 1.08-2.23, p = 0.018). This highlights the importance of monitoring PPI use, especially in older adults.

PURPOSE

Proton pump inhibitors (PPIs) are associated with increased risk of osteoporotic fracture. Hyperparathyroidism may be implicated, but few studies have considered this relationship. This study evaluated the relationship between PPI use and hyperparathyroidism in older adults.

METHODS

Participants were from the TUDA study, a large cross-sectional cohort of older Irish adults. Participants with an estimated glomerular filtration rate (eGFR) < 30 ml/min and serum calcium > 2.5 mmol/l were excluded to avoid hyperparathyroidism due to chronic renal disease and primary hyperparathyroidism. Hyperparathyroidism was defined as a parathyroid hormone (PTH) > 65 pg/ml. Multivariate regression models were used to analyse the relationship between PPI use and hyperparathyroidism.

RESULTS

A total of 4139 participants met the inclusion criteria, of whom 37.8% (n = 1563) were taking PPI medication. PPI use was identified in 41.4% of calcium supplement users and 35.4% of non-calcium supplement users. Overall, compared to non-users of PPIs, those taking PPIs were older (74.8 vs 72.9 years, p < 0.001) and had a higher prevalence of hyperparathyroidism (17.8 vs 11.0%, p < 0.001). In those taking calcium supplements (but not in non-users), PPI use was significantly associated with hyperparathyroidism (OR 1.56, CI 1.08-2.23, p = 0.018) after adjusting for age, sex, body mass index, serum vitamin D, eGFR, timed-up-and-go, dairy intake, medications, and comorbidities.

DISCUSSION

The results are consistent with the hypothesis of PPIs reducing calcium absorption, leading to a rise in PTH which could mediate increased fracture risk. No relationship of PPI use with hyperparathyroidism was observed in non-users of calcium supplements, possibly owing to lower dietary calcium intake. These results highlight the importance of monitoring PPI use, especially in older adults at risk of fracture.

Gender differences in the association between food costs and obesity in Korean adults: an analysis of a population-based cohort

BACKGROUND/OBJECTIVES

Prior studies, mostly conducted in Western countries, have suggested that the low cost of energy-dense foods is associated with an increased risk of obesity. This study aimed to investigate the association between food costs and obesity risk among Koreans who may have different food cost and dietary patterns than those of Western populations.

SUBJECTS/METHODS

We used baseline data from a cohort of 45,193 men and 83,172 women aged 40-79 years (in 2006-2013). Dietary intake information was collected using a validated food frequency questionnaire. Prudent and Western dietary patterns extracted via principal component analysis. Food cost was calculated based on Korean government data and market prices. Logistic regression analyses were performed to investigate the association of daily total, prudent, and Western food cost per calorie with obesity.

RESULTS

Men in the highest total food cost quintile had 15% higher odds of obesity, after adjusting for demographic characteristics and lifestyle factors (adjusted odds ratio, 1.15; 95% confidence interval, 1.08-1.22; P-trend < 0.001); however, this association was not clear in women (P-trend = 0.765). While both men and women showed positive associations between prudent food cost and obesity (P-trends < 0.001), the association between Western food cost and obesity was only significant in men (P-trend < 0.001).

CONCLUSIONS

In countries in which consumption of Western foods is associated with higher food costs, higher food costs are associated with an increased risk of obesity; however, this association differs between men and women.

U-shaped link of health checkup data and need for care using a time-dependent cox regression model with a restricted cubic spline

We explored risk indicators likely to result in older adults needing certified long-term care in Japan and ascertained whether this relationship forms a U-shaped link. We analyzed a community-based cohort of residents in Kitanagoya City, Aichi Prefecture, Japan. Participants were 3718 individuals aged 65 years and above who underwent health examinations between April 1, 2011 and March 31, 2012. For continuous clinical variables, we applied a time-dependent Cox regression model. Two types of models were applied-a linear and nonlinear model with restricted cubic splines-to assess the U-shaped association. Statistical significance (set at 0.05) for the nonlinearity was tested by comparing the spline and linear models. Among the participants, 701 were certified as needing Level 1 care or higher during a follow-up. Among the continuous clinical variables, the nonlinear model for body mass index, systolic blood pressure, high-density lipoprotein cholesterol, alanine aminotransferase, aspartate aminotransferase, and γ-glutamyl transpeptidase revealed significant U-shaped associations as compared with the linear model in which the outcome was a certification of the need for nursing care. These results provide an important insight into the usefulness of nonlinear models for predicting the risk of such certification.

Inpatient Mortality in Parkinson's Disease

Introduction

Although a majority of the American public prefer to die at home, a large percentage of Parkinson's disease patients die in acute care hospitals. We examine trends in the clinical and demographic characteristics of Parkinson's disease patients who die in a hospital to identify populations potentially vulnerable to unwanted inpatient mortality.

Methods

Patients with Parkinson's disease admitted to a hospital from 2002-2016 were identified from the National Inpatient Sample (n = 710,013) along with their associated clinical and demographic characteristics. The main outcome examined was mortality during inpatient admission. From these data, logistic regression models were estimated to obtain the odds ratios of inpatient mortality among clinical and demographic attributes, and their change over time.

Results

Characteristics significantly associated with increased odds of inpatient mortality included increased age (OR = 1.70 for 55-65, 2.52 for 66-75, 3.99 for 76-85, 5.72 for 86+, all P < 0.001), length of stay ≤5 days (reference; 6 + days OR = 0.37, P < 0.001), white race or ethnicity (reference; Black OR = .84 P < .001, Hispanic OR = 0.91 P = 0.01), male (reference; female OR = 0.93 P < 0.001), hospitalization in Northeast (reference; Midwest OR = 0.78, South 0.84, West OR = 0.82; all P < 0.001), higher severity of illness (moderate OR = 1.50, major OR = 2.32, extreme OR = 5.57; all P < 0.001), and mortality risk (moderate OR = 2.88, major OR = 10.92, extreme OR = 52.30; all P < 0.001). Fitted probabilities overall declined over time.

Conclusion

Differences exist among PD patient populations regarding likelihood of in-hospital mortality that are changing with time. Insight into which PD patients are most at risk for inpatient mortality may enable clinicians to better meet end-of-life care needs.

Misspecification of confounder-exposure and confounder-outcome associations leads to bias in effect estimates

BACKGROUND

Confounding is a common issue in epidemiological research. Commonly used confounder-adjustment methods include multivariable regression analysis and propensity score methods. Although it is common practice to assess the linearity assumption for the exposure-outcome effect, most researchers do not assess linearity of the relationship between the confounder and the exposure and between the confounder and the outcome before adjusting for the confounder in the analysis. Failing to take the true non-linear functional form of the confounder-exposure and confounder-outcome associations into account may result in an under- or overestimation of the true exposure effect. Therefore, this paper aims to demonstrate the importance of assessing the linearity assumption for confounder-exposure and confounder-outcome associations and the importance of correctly specifying these associations when the linearity assumption is violated.

METHODS

A Monte Carlo simulation study was used to assess and compare the performance of confounder-adjustment methods when the functional form of the confounder-exposure and confounder-outcome associations were misspecified (i.e., linearity was wrongly assumed) and correctly specified (i.e., linearity was rightly assumed) under multiple sample sizes. An empirical data example was used to illustrate that the misspecification of confounder-exposure and confounder-outcome associations leads to bias.

RESULTS

The simulation study illustrated that the exposure effect estimate will be biased when for propensity score (PS) methods the confounder-exposure association is misspecified. For methods in which the outcome is regressed on the confounder or the PS, the exposure effect estimate will be biased if the confounder-outcome association is misspecified. In the empirical data example, correct specification of the confounder-exposure and confounder-outcome associations resulted in smaller exposure effect estimates.

CONCLUSION

When attempting to remove bias by adjusting for confounding, misspecification of the confounder-exposure and confounder-outcome associations might actually introduce bias. It is therefore important that researchers not only assess the linearity of the exposure-outcome effect, but also of the confounder-exposure or confounder-outcome associations depending on the confounder-adjustment method used.