Forward Stepwise Multivariate Regression Analysis Research Articles

A retrospective analysis of 24-month real-world glucose control for children and adolescents with type 1 diabetes using the MiniMed™ 670G insulin pump.

The peer review history for this article is available at https://publons.com/publon/10.1111/dom.14934. Data available on request due to privacy/ethical restrictions Appendix S1. Supporting Information. Table S1. Results of the correlation analysis. %CGM, proportion of time when the continuous glucose monitoring worked; TIR, time in range; GRI, glycaemia risk index; SG, sensor glucose; %CV, glycaemic variability; %HCL, proportion of time when the hybrid closed loop was enabled; TDD, total daily dose. Level of significance: * P ≤ 0.001; # P ≤ 0.01; ° P < 0.05 Table S2. Results of the forward stepwise multivariate regression analysis. TIR, time in range; %CGM, proportion of time when the continuous glucose monitoring worked; %HCL, proportion of time when the hybrid closed loop was enabled; TDD, total daily dose; GRI, glycaemia risk index. Figure S1. The upper part displays the boxplot HbA1c during the 24-month study. The lower part displays the percentage of patients with time spent with sensor glucose below 70 mg/dL (TBR) < 4%, within 70 and 180 mg/dL (TIR) > 70%, and above 180 mg/dL (TAR) < 25% at each timepoint of the study. Figure S2. Rate of basal and premeal insulin during the study. The TDD was 0.91 ± 0.26 IU/kg/d at baseline, 0.90 ± 0.24 IU/kg/d at 6 months, 0.90 ± 0.20 IU/kg/d at 12 months, 0.90 ± 0.22 IU/kg/d at 18 months, and 0.90 ± 0.22 IU/kg/d at 24 months. No statistical differences were found. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

C7 sagittal vertical axis is the determinant of the C5–C7 angle in cervical sagittal alignment

Background ContextPrevious studies have indicated that the T1 slope correlates with cervical lordosis. In contrast, the specific impact of the C7 sagittal vertical axis (C7SVA) on cervical lordosis remains unknown. PurposeThis study aimed to investigate the specific role of C7SVA in cervical lordosis. Study Design/SettingThis was a retrospective radiographic study. Patient SampleForty-eight consecutive patients who underwent lateral standing radiography of the entire spine were retrospectively reviewed. Outcome MeasuresRadiographic parameters included occipito (Oc)-C7, Oc–C2, C2–C7, C2–C4, and C5–C7 angles; T1 slope; C7SVA; T1 pelvic angle (TPA); pelvic incidence; pelvic tilt; and sacral slope. MethodsThe radiographs of 96 consecutive patients who underwent lateral standing radiography of the entire spine in June 2015 in our hospital were retrospectively reviewed. Patients having cervical deformities, having undergone cervical fusion, and under 18 years of age were excluded. A total of 48 Asian patients (14 men and 34 women; mean age, 54.6 years) were eligible. Pathologies included scoliosis, myelopathy, thoracolumbar deformity, and spondylosis. Spearman rank correlation coefficients were used to examine correlations between the parameters. The relationship between C5–C7 lordosis and the radiographic parameters was calculated using the forward stepwise multivariate regression analysis. The authors do not have financial associations relevant to this article. ResultsC7SVA correlated with the Oc–C7 (r=0.42) and C2–C7 (r=0.50) angles. However, the correlation coefficient was smaller than that between the T1 slope and Oc–C7 (r=0.83) or C2–C7 (r=0.76) angles. When the C2–C7 angle was divided into C2–C4 and C5–C7 angles, C7SVA correlated with the C5–C7 (r=0.63) angle but not with the C2–C4 angle. The correlation coefficient between the C5–C7 angle and C7SVA was higher than that between the C5–C7 angle and T1 slope (r=0.53) or the C5–C7 angle and TPA (r=0.60). Using radiographic parameters and age, multiple regression analysis revealed that only C7SVA affected the C5–C7 angle. ConclusionsC7SVA was the only radiographic parameter that affected the C5–C7 angle. Both T1 slope and C7SVA are key to the shape of the cervical sagittal alignment. The results of this study can be a starting point to improve our understanding of cervical sagittal alignment.

Discovery and identification of serum potential biomarkers for pulmonary tuberculosis using iTRAQ‐coupled two‐dimensional LC‐MS/MS

Pulmonary tuberculosis (TB) caused by Mycobacterium tuberculosis is a chronic disease. Currently, there are no sufficiently validated biomarkers for early diagnosis of TB infection. In this study, a panel of potential serum biomarkers was identified between patients with pulmonary TB and healthy controls by using iTRAQ-coupled 2D LC-MS/MS technique. Among 100 differentially expressed proteins screened, 45 proteins were upregulated (>1.25-fold at p < 0.05) and 55 proteins were downregulated (<0.8-fold at p < 0.05) in the TB serum. Bioinformatics analysis revealed that the differentially expressed proteins were related to the response to stimulus, the metabolic and immune system processes. The significantly differential expression of apolipoprotein CII (APOCII), CD5 antigen-like (CD5L), hyaluronan-binding protein 2 (HABP2), and retinol-binding protein 4 (RBP4) was further confirmed using immunoblotting and ELISA analysis. By forward stepwise multivariate regression analysis, a panel of serum biomarkers including APOCII, CD5L, and RBP4 was obtained to form the disease diagnostic model. The receiver operation characteristic curve of the diagnostic model was 0.98 (sensitivity= 93.42%, specificity= 92.86%). In conclusion, APOCII, CD5L, HABP2, and RBP4 may be potential protein biomarkers of pulmonary TB. Our research provides useful data for early diagnosis of TB.

A prospective observational study of skin to subarachnoid space depth in the Indian population.

Background and Aims:A pre-puncture estimate of skin to subarachnoid space depth (SSD) may guide spinal needle placement and reduce complications associated with lumbar puncture. Our aim was to determine (1) The SSD in Indian males, females, parturients and the overall population; (2) To derive formulae for predicting SSD and (3) To determine which previously suggested formula best suited our population.Methods:In this prospective, observational study, 800 adult Indian patients undergoing surgery under spinal anaesthesia were divided into three groups: Males (Group M), females (Group F) and parturients (Group PF). SSD was measured after lumbar puncture. The relationship between SSD and patient characteristics was studied and statistical models were used to derive formula for predicting SSD. Statistical analysis included One-way ANOVA with post hoc analysis, forward stepwise multivariate regression analysis and paired t-tests.Results:Mean SSD was 4.71 ± 0.70 cm in the overall population. SSD in adult males (4.81 ± 0.68 cm) was significantly longer than that observed in females (4.55 ± 0.66 cm) but was comparable with SSD in parturients (4.73 ± 0.73 cm). Formula for predicting SSD in the overall population was 2.71 + 0.09 × Body Mass Index (BMI). Stocker's formula when applied correlated best with the observed SSD. Formulae were derived for the three groups.Conclusions:We found gender-based differences in SSD, with SSD in males being significantly greater than that observed in the female population. SSD correlated with BMI in the parturient and the overall population. Amongst the previously proposed formulae, Stocker's formula was most accurate in predicting SSD in our population.

The Impact of Vascular Resection on Early Postoperative Outcomes after Pancreaticoduodenectomy: An Analysis of the American College of Surgeons National Surgical Quality Improvement Program Database

Several single-center reports suggest that vascular resection (VR) during pancreaticoduodenectomy (PD) for patients with pancreatic adenocarcinoma is feasible without affecting early postoperative mortality or morbidity. Our objective is to review the outcomes associated with VR during PD using a large multicenter data source. A retrospective cohort analysis was performed using the National Surgical Quality Improvement Program Participant User Files for 2005-2009. All patients undergoing PD for a postoperative diagnosis of malignant neoplasm of the pancreas were included. Forward stepwise multivariate regression analysis was used to determine the association between VR during PD and 30-day postoperative mortality and morbidity after adjustment for patient demographics and comorbidities. 3,582 patients were included for analysis, 281 (7.8 %) of whom underwent VR during PD. VR during PD was associated with significantly greater risk-adjusted 30-day postoperative mortality [5.7 % with VR versus 2.9 % without VR, adjusted odds ratio (AOR) 2.1, 95 % confidence interval (CI) 1.22-3.73, P = 0.008] and overall morbidity (39.9 % with VR versus 33.3 % without VR, AOR 1.36, 95 % CI 1.05-1.75, P = 0.02). There was no significant difference in risk-adjusted postoperative mortality or morbidity between those patients undergoing VR by the primary surgical team versus those patients undergoing VR by a vascular surgical team. Contrary to the findings of several previously published single-center analyses, the current study demonstrates increased 30-day postoperative morbidity and mortality in PD with VR when compared with PD alone.

Locoregional versus General Anesthesia for Open Inguinal Herniorrhaphy: A National Surgical Quality Improvement Program Analysis

Our objective was to study outcomes associated with open inguinal herniorrhaphy performed under locoregional (LR) versus general anesthesia (GA). National Surgical Quality Improvement Program (NSQIP) data from 2005 to 2009 was queried to capture patients undergoing initial unilateral inguinal herniorrhaphy. We excluded patients with incarcerated/strangulated hernia or those undergoing a concomitant procedure. Outcomes were anesthesia and operative times, postoperative admission, and 30-day morbidity. Using the entire NSQIP sample, forward stepwise multivariate regression analysis was used to compare outcomes between patients receiving LR versus GA after adjustment for patient demographics and comorbid diagnoses. Outcomes were also compared for a smaller subgroup of patients propensity-matched for receiving LR anesthesia. A total of 25,213 patients were analyzed (16,282 GA and 8,931 LR). Patients in the LR group had a higher incidence of comorbid illnesses and were more likely to have an American Society of Anesthesiologists classification ≥ 3. Multivariate analyses demonstrated that LR anesthetic is associated with shorter anesthetic and operative times and a lower hospital admission rate. Comparison using a propensity-matched cohort for undergoing LR anesthesia confirms that these patients had significantly shorter anesthesia (32 vs 38 min, P < 0.0001) and operative times (53.3 vs 57.2 min, P < 0.0001), as well as a significantly reduced rate of postoperative admission (5.9% vs 10.9%, P < 0.0001) and 30-day morbidity (0.9% vs 1.3%, P < 0.05). Our analysis of NSQIP suggests that, compared with general anesthesia, the locoregional technique is associated with shorter anesthesia and operative times, reduced need for postoperative hospital admission, and a small but significant reduction in postoperative morbidity.

The effect of vascular reconstruction on early postoperative outcomes after pancreaticoduodenectomy: An analysis of the American College of Surgeons National Surgical Quality Improvement Program database.

153 Background: Several single-center reports have been published suggesting that vascular reconstruction (VR) during pancreaticoduodenectomy (PD) allows an acceptable oncologic outcome in patients with pancreatic adenocarcinoma without affecting early postoperative mortality or morbidity. The objective of our study was to review the outcomes associated with VR during PD using a large multicenter data source. Methods: A retrospective cohort analysis was performed using the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) Participant User Files for 2005-2009. All patients undergoing PD for a postoperative diagnosis of malignant neoplasm of the pancreas were included. Forward stepwise multivariate regression analysis was used to determine the association between VR during PD and 30-day postoperative mortality and morbidity after adjusting for patient demographics and comorbidities. Results: 3582 patients were included for analysis, 281 (7.8%) of whom underwent VR during PD. VR during PD was associated with significantly greater risk-adjusted 30-day postoperative mortality (5.7% with VR vs. 2.9% without VR, Adjusted Odds Ratio (AOR) 2.1, 95% CI 1.22-3.73, p = 0.008) and overall morbidity (39.9% with VR vs. 33.3% without VR, AOR 1.36, 95% CI 1.05-1.75, p = 0.02). Patients undergoing VR required significantly longer operative times, and were more likely to require intraoperative transfusion or early reoperation, than patients not undergoing VR during PD. There was no significant difference in risk-adjusted postoperative mortality or morbidity between those patients undergoing VR by the primary surgical team versus those patients undergoing VR by a vascular surgical team. Conclusions: Contrary to the findings of several previously published single-center analyses, data from ACS-NSQIP suggests that VR significantly increases the risk of 30-day postoperative death or complications after PD. Patients who may require VR during PD will likely benefit from referral to centers with sufficient experience with this procedure.

Prognostic utility of RIFLE for acute renal failure in patients with sepsis

Acute renal failure (ARF) is common among patients with sepsis and increases mortality [1]. However, the prognostic utility of the newly released classification for ARF, entitled RIFLE (Risk, Injury, Failure, Loss, and End-stage kidney disease) [2], in predicting mortality in such patients has not yet been assessed. We sought to evaluate retrospectively the usefulness of RIFLE in predicting mortality in patients with sepsis admitted to the infectious disease intensive care unit (ICU) of our hospital between January 2005 and December 2006. Data were collected from the unit database. Sepsis was classified in accordance with American College of Chest Physicians and the Society of Critical Care Medicine consensus [3]. Patients with chronic kidney disease on dialysis were excluded from the analysis. The Simplified Acute Physiology Score (SAPS) version II was calculated based on the worst variables recorded during the first 24 hours of ICU admission. Mortality was assessed at day 60 [2]. Forward stepwise multivariate regression analysis was employed to evaluate independent predictors of mortality. The goodness-of-fit was tested by means of the Hosmer-Lemeshaw statistic. Area under the receiver operating characteristic curve (AUROC) was employed to evaluate the prognostic accuracy of RIFLE and SAPS II. A two-tailed P < 0.05 was considered statistically significant. One-hundred and eighty-two patients (mean age 56.2 ± 18.56 years; 120 male; 162 white) were analyzed. Fourteen patients had sepsis, 108 patients had severe sepsis and 60 patients had septic shock. Sixty-eight patients (37.4%) had ARF: 11 patients were in class 'risk' (in RIFLE), 21 patients were in class 'injury', and 36 patients were in class 'failure'. The mortality rate was 37.4%, and increased from classes 'normal' to 'failure' (normal, 9.6%; risk, 27.3%; injury, 28.6%; failure, 55%; P < 0.0001). Multivariate analysis including age above 60 years, sex, SAPS II above 50, comorbidity (namely cardiovascular disease), and RIFLE class revealed that SAPS II score above 50 (odds ratio 0.12, 95% confidence interval [CI] 0.04–0.29; P < 0.0001) and class 'failure' (odds ratio 3.59, 95% CI 1.44–8.94; P = 0.006) were independent predictors of mortality. Similarly, SAPS II score above 50 (AUROC 0.778, 95% CI 0.694–0.863; P < 0.0001) and ARF according to RIFLE (AUROC 0.725, 95% CI 0.63–0.81; P < 0.0001) had good prognostic accuracy. In summary, ARF, based on RIFLE criteria, is common in patients with sepsis, and is associated with increased mortality. RIFLE appears to be a useful prognostic tool in such patients.

Changes in Serum Insulin‐Like Growth Factors, Not Leptin, Are Associated With Postnatal Weight Gain in Preterm Neonates

Serum concentrations of conventional nutrition-related proteins, such as albumin, prealbumin, transferrin, and retinol-binding protein, are usually inconsistent with changes in anthropometric measurements in the postnatal period. The aim of this study was to evaluate how reliable growth hormone (GH), insulin-like growth factors (IGFs), IGF-binding proteins (IGFBPs), and leptin, the proteins known to be involved in the regulation of growth, are in reflecting postnatal growth and nutritional status in preterm neonates. Blood samples and anthropometric measurements were collected from 55 preterm neonates (chronological age 30.4 +/- 2.8 weeks) for 4 continuous weeks (weeks 0 to 3). After adjusting for chronological age, body weights and serum IGF-II concentrations were significantly greater and serum transferrin concentrations were significantly lower in weeks 2 and 3 than in week 0 (repeated-measures analysis of variance (ANOVA) and Bonferroni test, p < .05). Forward stepwise multivariate regression analysis showed that change in total IGF-I (week 0 to week 3) was a positive predictor, and changes in insulin and prealbumin were negative predictors of postnatal weight gain. In addition, daily fat intake was a positive predictor of postnatal length increases, and changes in prealbumin, insulin, and GH were negative predictors of postnatal changes in the ponderal index (weight x length(-3)). Changes in GH and IGFBP-2 were negative predictors of changes in head circumference and triceps skinfold thickness, respectively. Serial measurements of serum IGF-I and IGF-II may be useful adjuncts to anthropometric measurements for monitoring postnatal growth and nutritional status in preterm neonates.

Relation of cord serum levels of growth hormone, insulin-like growth factors, insulin-like growth factor binding proteins, leptin, and interleukin-6 with birth weight, birth length, and head circumference in term and preterm neonates

OBJECTIVES: Fetal growth process is governed by multiple factors. We investigated the relation of insulin-like growth factors (IGFs), IGF binding proteins (IGFBPs), leptin, and interleukin-6 (IL-6) with intrauterine growth in preterm and term neonates. METHODS: Thirty-eight preterm and 43 term neonates were recruited. Anthropometric measures were recorded and umbilical cord blood samples were collected at birth. RESULTS: Birth weight (BW), birth length (BL), ponderal index, head circumference (HC), and cord serum levels of albumin, prealbumin, retinol-binding protein (RBP), total and free IGF-I, IGF-II, IGFBP-3, acid-labile subunit (ALS), and leptin were significantly lower, whereas levels of IGFBP-1, IGFBP-2, and IL-6 were significantly higher in preterm than in term neonates ( P < 0.05). Total and free IGF-I, ALS, and leptin had significantly positive correlations, whereas IGFBP-2 had a significantly negative correlation, with BW and BL in preterm plus term neonates. Forward stepwise multivariate regression analysis showed that gestational age (GA), IGFBP-2, leptin, and free IGF-I are significant predictors of BW; GA, IGFBP-2, ALS, transferrin, and leptin are significant predictors of BL; and GA and free IGF-I are significant predictors of HC in preterm and term neonates. CONCLUSIONS: Our results suggest that IGF-I, IGF-II, IGFBP-2, ALS, and leptin play important roles in intrauterine growth.

Donor-specific bone marrow cells suppress lymphocyte reactivity to donor antigens and differentially modulate TH1 and TH2 cytokine gene expression in the responder cell population.

Summary: In order to clarify the most reliable risk factor to predict renal outcome, 206 patients with IgA nephropathy were studied for mean period of 9.2 years. the histopathological changes of this disease using light microscopy were divided into four grades (grade 1–4). These grades included glomerular, interstitial and vascular lesions. the cumulative rate of kidney survival progressing to end stage renal failure (ESRF) in all patients was 94% at 5 years, 87% at 10 years and 80% at 15 years after renal biopsy. None of the patients in grade 1 reached ESRF. the cumulative rate of kidney survival in grade 2 was 99% at 5 years, 98% at 10 years and 89% at 15 years after renal biopsy. In grade 3, it was 94% at 5 years, 79% at 10 years and 75% at 15 years. In grade 4 it was 53% at 5 years, 33% at 10 years and 22% at 15 years after renal biopsy. Forward stepwise multivariate regression analysis revealed that, in addition to the histopathological findings, three more risk factors were found to influence actuarial renal survival rate. These factors were: (i) the levels of serum creatinine; (ii) the level of serum albumin; and (iii) the amount of proteinuria at the time of renal biopsy. In parallel studies, forward stepwise multivariate regression analysis isolated three risk factors that influenced the progression of the reciprocal of serum creatinine. These factors were: (i) the levels of total protein; (ii) the degree of our pathological grading; and (iii) the amount of proteinuria. It was concluded that our pathological grading was useful as a prognostic parameter because of its simplicity and availability in routine clinical activities.

Histopathologic Comparison of the Subsets of Lupus Erythematosus

A recent study by Bangert et al suggests that there are quantitative histologic differences that distinguish discoid lupus erythematosus (DLE) and subacute cutaneous lupus erythematosus (SCLE). Utilizing criteria proposed by these authors, we examined 77 biopsy specimens from 63 patients with various forms of lupus erythematosus, but we were unable to predict the correct clinical subset. Using the clinical diagnosis of DLE as a positive reference standard, the sensitivity and specificity for the overall pathologic diagnosis of DLE were 55% and 42%, respectively. Statistically significant histologic factors favoring the diagnosis of DLE over SCLE in the present study were pilosebaceous atrophy, hyperkeratosis, parakeratosis, basement membrane thickening around the follicles, subepidermal edema, and vascular ectasia. These histologic variables were entered into a forward stepwise multivariate regression analysis to determine distinct predictors of DLE vs SCLE. This analysis showed that pilosebaceous atrophy was the only distinct significant predictor of DLE vs SCLE. These results suggest that the histologic differentiation of clinically defined DLE and SCLE cannot be established from the histologic features examined.