SAS macros for point and interval estimation of area under the receiver operating characteristic curve for non‐proportional and proportional hazards Weibull models

Aims and objectives For prediction of risk of cardiovascular end points using survival models the proportional hazards assumption is often not met. Thus, non‐proportional hazards models are more appropriate for developing risk prediction equations in such situations. However, computer program for evaluating the prediction performance of such models has been rarely addressed. We therefore developed SAS macro programs for evaluating the discriminative ability of a non‐proportional hazards Weibull model developed by Anderson (1991) and that of a proportional hazards Weibull model using the area under receiver operating characteristic (ROC) curve. Method Two SAS macro programs for non‐proportional hazards Weibull model using Proc NLIN and Proc NLP respectively and model validation using area under ROC curve (with its confidence limits) were written with SAS IML language. A similar SAS macro for proportional hazards Weibull model was also written. Results The computer program was applied to data on coronary heart disease incidence for a Framingham population cohort. The five risk factors considered were current smoking, age, blood pressure, cholesterol and obesity. The predictive ability of the non‐proportional hazard Weibull model was slightly higher than that of its proportional hazard counterpart. An advantage of SAS Proc NLP in terms of the example provided here is that it provides significance level for the parameter estimates whereas Proc NLIN does not. Conclusion The program is very useful for evaluating the predictive performance of non‐proportional and proportional hazards Weibull models.     

Comparison of Cox and Gray's survival models in severe sepsis

BACKGROUND: Although survival is traditionally modeled using Cox proportional hazards modeling, this approach may be inappropriate in sepsis, in which the proportional hazards assumption does not hold. Newer, more flexible models, such as Gray's model, may be more appropriate. OBJECTIVES: To construct and compare Gray's model and two different Cox models in a large sepsis cohort. To determine whether hazards for death after sepsis were nonproportional. To explore how well the different survival modeling approaches describe these data. DESIGN: Analysis of combined data from the treatment and placebo arms of a large, negative, sepsis trial. SETTING: Intensive care units at 136 U.S. medical centers. SUBJECTS: A total of 1090 adults aged 18 yrs or older with signs and symptoms of severe sepsis and documented or probable Gram-negative infection. MEASUREMENTS: We considered 27 potential baseline risk factors and modeled survival over the 28 days after the onset of sepsis. We tested proportionality in single-variable Cox analysis using Schoenfeld residuals and log-log plots. We constructed a traditional multivariable Cox model, a multivariable Cox model with time-varying covariates, and a multivariable Gray's model. RESULTS: In single-variable analyses, 20 of the 27 potential factors were significantly associated with mortality, and 10 of 20 had nonproportional hazards. In multivariate analysis, all three models retained a very similar set of significant covariates (two models retained the identical set of nine variables, and the third differed only in that it retained the same nine plus a tenth variable). Four of the nine common covariates had nonproportional hazards. Of the three models, Gray's model best captured these changing hazard ratios over time. CONCLUSION: We confirm that many of the important predictors of mortality in severe sepsis are nonproportional and find that Gray's model seems best suited for modeling survival in this condition.     

Sample-size calculations for the Cox proportional hazards regression model with nonbinary covariates

This paper derives a formula to calculate the number of deaths required for a proportional hazards regression model with a nonbinary covariate. The method does not require assumptions about the distributions of survival time and predictor variables other than proportional hazards. Simulations show that the censored observations do not contribute to the power of the test in the proportional hazards model, a fact that is well known for a binary covariate. This paper also provides a variance inflation factor together with simulations for adjustment of sample size when additional covariates are included in the model. Control Clin Trials 2000;21:552-560     

Computer programs to estimate overoptimism in measures of discrimination for predicting the risk of cardiovascular diseases

Background Development of chronic disease risk prediction models has become a growing area of research in recent years. The internal validity of such models is sometimes lower than estimated from the development sample. Overfitting or overoptimism of the developed model and/or differences between the samples are likely causes for this. For modelling of an uncommon outcome, bootstrapping for overoptimism is the preferred method for afterwards shrinking of regression coefficients and the model's discrimination and calibration for overoptimism. However, computer programs for different types of bootstrap validation are not readily available. We developed two SAS macro programs – one for the simple bootstrap that compares the discriminatory performance of the Cox proportional hazards model from the original sample in bootstrap samples; and another (which is more efficient), known as stepwise bootstrap validation, that makes the same comparison but from models developed by variable selection from bootstrap samples in the original sample. These are illustrated through an example from cardiovascular disease (CVD) risk prediction. Methods Two SAS macro programs for Cox proportional hazards model using Proc PHREG were developed for estimating overoptimism in Harrell's C and Somers' D statistics. The computer programs were applied to data on CVD incidence for a Framingham cohort that combined both the original and offspring exams. The risk factors considered were current smoking, diabetes, age, sex, systolic blood pressure, diastolic blood pressure, total cholesterol, high‐density lipoprotein cholesterol, triglycerides and body mass index. Results The degree of overoptimism in both Harrell's C and Somers' D statistics were low. Both these statistics were corrected for overoptimism by subtracting overoptimism from their observed values. Between the two bootstrap validation algorithms, the degree of overoptimism was estimated to be higher for stepwise bootstrap validation. Conclusion The programs are very useful for evaluating the ‘overoptimism corrected’ predictive performance of Cox proportional hazards model.     

Competing risks models and time-dependent covariates

New statistical models for analysing survival data in an intensive care unit context have recently been developed. Two models that offer significant advantages over standard survival analyses are competing risks models and multistate models. Wolkewitz and colleagues used a competing risks model to examine survival times for nosocomial pneumonia and mortality. Their model was able to incorporate time-dependent covariates and so examine how risk factors that changed with time affected the chances of infection or death. We briefly explain how an alternative modelling technique (using logistic regression) can more fully exploit time-dependent covariates for this type of data.     

Computer-aided disease prediction system: development of application software with SAS component language

AIMS: The intricacy of predictive models associated with prognosis and risk classification of disease often discourages medical personnel who are interested in this field. The aim of this study was therefore to develop a computer-aided disease prediction model underpinning a step-by-step statistics-guided approach including five components: (1) data management; (2) exploratory analysis; (3) type of predictive model; (4) model verification; (5) interactive mode of disease prediction using SAS 8.02 Windows 2000 as a platform. METHODS: The application of this system was illustrated by using data from the Swedish Two-County Trial on breast cancer screening. The effects of tumour size, node status, and histological grade on breast cancer death using logistic regression model or survival models were predicted. A total of 20 questions were designed to exemplify the usefulness of each component. We also evaluated the system using a controlled randomized trial. Times to finish the above 20 questions were used as endpoint to evaluate the performance of the current system. User satisfaction with the current system such as easy to use, the efficiency of risk prediction, and the reduction of barrier to predictive model was also evaluated. RESULTS: The intervention group not only performed more efficiently than the control group but also satisfied with this application software. CONCLUSIONS: The MD-DP-SOS system characterized by menu-driven style, comprehensiveness, accuracy and adequacy assessment, and interactive mode of disease prediction is helpful for medical personnel who are involved in disease prediction.     

Bayesian approach to control of amikacin serum concentrations in critically ill patients with sepsis

OBJECTIVE: To compare the predictive performance of a Bayesian program incorporating a population model with and without severity of illness covariates in intensive care unit (ICU) patients with sepsis. DESIGN: The clinical, physiologic, and pharmacokinetic data of 62 patients with sepsis admitted to a tertiary-care center were analyzed retrospectively. The patients were randomly assigned to a active group and a validation group. The model was developed using a three-step approach involving Bayesian estimation of pharmacokinetic parameters, selection of covariates by principal component analysis, and final selection of covariates by stepwise multiple linear regression. The predictive performance of this model was tested in patients from the validation group and compared with that of a general population model without covariates. RESULTS: Regression analysis revealed that the Acute Physiologic and Chronic Health Evaluation (APACHE II) score was the most important determinant for amikacin volume of distribution (1.5 L/kg, APACHE II; r2 = 0.77). For amikacin clearance (CIamik), creatinine clearance (CIcr), positive end-expiratory pressure (PEEP), and use of catecholamines (CAT) were the most important predictors (CIamik = 44.5 + 0.67 CIcr - 1.29 PEEP - 8.34 CAT; r2 = 0.72). The relative mean error (deltaME) and root mean-square error (deltaRMSE) (95% CI) were -0.62 (-1.2 to 0.01) and 3.78 (2.3 to 4.8) mg/L, respectively. Since the 95% CI for deltaRMSE did not include zero, it appears that the model with covariates is significantly improved in terms of precision. CONCLUSIONS: Our results show that, in ICU patients treated with amikacin, it is relevant to consider covariates related to pathophysiologic status and therapeutic measures. Application of a Bayesian program allows improved control of the pharmacokinetic parameters in patients who exhibit rapidly changing physiologic conditions.     

Autologous venous bypass in the therapy of arterial occlusive disease of the femoropopliteal area: long-term results

The outcome of 532 femoro-popliteal vein grafts performed electively during the years 1970 to 1985 for obliterative arterial disease, was analyzed using the documentation-system of the Austrian Society for Vascular Surgery, as well as SAS and BMDP-software on an IBM 4381 computer of the Medical Faculty. The probability of function was estimated according to the Kaplan-Meier method, statistical differences were checked with Breslow's and Mantel's test, the proportional hazards regression model (Cox) was used to elucidate the influence of different risk factors on each own and in combination of each other. In the univariate analysis, the preoperative clinical status was found to be of prognostic significance, but technical details such as intraoperative as well as postoperative arteriography or site of the distal anastomosis were not important. A postoperative coumarine treatment had no demonstrable impact on graft function, but positively influenced the probability of patient survival. Taking into account the factors found to be of significance in the univariate analysis (e.g. set of factors: site of distal anastomosis, diabetic state and smoking habits) were analyzed using the proportional hazards regression model but were found of no major influence. The factors preoperative clinical status, patients age, and coumarine therapy significantly influenced the probability of patient survival, but diabetes mellitus and smoking were found to be not important.     

A prediction model of 1-year mortality for acute ischemic stroke patients

OBJECTIVE: To develop a prediction model for 1-year mortality in patients with acute ischemic stroke, with the model to be at least as useful and accurate as other previously developed prediction models. DESIGN: Retrospective cohort study. SETTING: Neurology department at an Australian tertiary teaching hospital. PARTICIPANTS: Four hundred forty consecutive patients diagnosed with acute ischemic stroke between July 1, 1995, and June 30, 1997. INTERVENTIONS: Two hundred twenty-three (51%) patients were randomly assigned to the derivation sample to develop a prediction model using the Cox proportional hazards model. The model was then validated in a validation sample of 217 (49%) patients.Main Outcome Measure: One-year mortality. RESULTS: Eight clinical predictors were included in the final model: unconsciousness (3 points), dysphagia (7 points), urinary incontinence (9 points), both sides affected (4 points), hyperthermia (4 points), ischemic heart disease (3 points), peripheral vascular disease (3 points), and diabetes mellitus (2 points). Patients with scores of 10 or higher were allocated to the high-risk group, which had a 1-year mortality rate of 76%, compared with a 1-year mortality rate of 8% in the low-risk group. There was no statistically significant difference in terms of sensitivity, specificity, and positive predictive value in the validation sample. CONCLUSION: We developed a predictive model for 1-year mortality in acute ischemic stroke patients. The model is easy to use and is comparable in its accuracy with other predictive models.     

Dynamic assessment of severity of illness in pediatric intensive care

Severity of illness in 293 pediatric ICU patients was assessed by a daily estimate of ICU survival. The probability of nonsurvival was obtained by logistic regression analysis, using physiologic stability index (PSI) values from previous days as time-dependent covariates. Only PSI values from the previous 2 days gave statistically significant predictions of short-term (less than 24 h) outcome. When the prediction model derived from these data was tested prospectively on a separate set of 345 pediatric patients, there was excellent agreement between observed and predicted short-term mortality. Receiver operating characteristic curves for the 345 patients were statistically equivalent to those originally derived for the 293 patients, and this prediction model had significantly (p less than .025) more accuracy than prediction based on admission PSI. These results indicate that this model for daily risk assessment is statistically reliable and objective, as verified against eventual outcome. In the 345 patients, ICU mortality was predicted with 89% sensitivity and 91% specificity. This prediction model may be used to stratify patient groups for clinical studies, or identify very low-risk patients for potential early ICU discharge.     

Time-dependent effect of non-Hodgkin's lymphoma grade on disease-free survival of relapsed/refractory patients treated with high-dose chemotherapy plus autotransplantation

Evaluation of time to event outcomes usually is examined by the Kaplan-Meier method and Cox proportional hazards models. We developed a modified statistical model based on histologic grade and other variables to describe the time-dependent outcome for autologous stem cell transplant (autotransplant) performed for non-Hodgkin's lymphoma (NHL) based on histologic grade and other variables. One hundred and fourteen relapsed or refractory NHL patients were treated using BCNU 600 mg/m2, etoposide 2400 mg/m2, and cisplatin 200 mg/m2 IV followed by autotransplant. Median age was 53.5 (range: 25-70) years, 78 patients had aggressive NHL and 36 indolent NHL. Seventy-five patients received involved-field radiotherapy just prior to transplant. At a median follow-up of 33 (range: 3 to 118) months, the estimated 5-year Kaplan-Meier probabilities of overall survival and disease-free survival were 61% and 51%, respectively. Cox proportional hazards model analysis showed that proportionality did not hold for lymphoma grade, indicating that the relationship between the grade and disease-free survival differed over time. By piece-wise Cox model, the relative risk for experiencing relapse or death after 1 year in patients with indolent compared with patients with aggressive NHL was 2.81 (p=0.019) with 95% confidence interval (1.19, 6.65). The time-dependent effect of lymphoma grade on disease-free survival suggests the need for early (within first year) incorporation of novel therapeutic approaches in management of patients with indolent NHL undergoing autotransplant.     

基于长链非编码RNA的表达谱特征构建乳腺癌患者预后的风险模型

目的构建长链非编码RNA(long non-coding RNA,LncRNA)表达特征的乳腺癌患者预后的预测模型。方法分析癌症基因组图谱(the cancer genome atlas,TCGA)数据库1081例乳腺癌患者的转录组测序数据中LncRNA表达图谱及临床特征,对TCGA数据库中112对配对的乳腺癌及正常乳腺组织的转录组测序数据进行差异表达分析和单因素分析筛选得到差异表达且与乳腺癌患者预后显著相关的LncRNA(DELncRNA),利用DEseq2包进行差异表达分析(为减弱批次效应,测序数据已用DESeq函数标准化)。1081例乳腺癌患者被分成两组:训练集(541例)和验证集(540例)。将DELncRNA纳入Cox比例风险回归模型,在训练集中筛选和建立多LncRNA预后模型并对模型进行比例风险假定检验(proportional hazards assumption,PH假定检验),计算多基因风险评分,并基于此将患者分为高风险组和低风险组,采用Kaplan-Meier方法进行生存分析,并用验证集540例患者的数据进行验证。评价该模型在TCGA数据库肺鳞癌和肝细胞肝癌等患者中的预后评估价值。基因集富集分析(gene set enrichment analysis,GSEA)分析LncRNA影响患者生存的具体机制。结果转录组测序分析筛选得到2815个差异表达基因,其中与乳腺癌患者预后显著相关的LncRNA共91个(P<0.05)。利用541例训练集乳腺癌患者的91个DELncRNA表达数据进行Cox回归分析,构建了基于5个LncRNA的Cox比例风险回归模型(训练集AUC=0.746,验证集AUC=0.650):AC004551.1、MTOR-AS1、KCNAB1-AS2、FAM230G和LINC01283,并进行PH假定检验(P=0.388)。K-M生存分析发现,训练集中高风险组的生存明显差于低风险组(中位生存时间:7.049年与12.21年,HR 0.367,95%CI 0.228~0.597,P<0.001),在验证集中高风险组患者生存时间也明显短于低风险组(中位生存时间:7.57年与10.85年,HR 0.412,95%CI 0.214~0.793,P<0.001)。在TCGA其他癌种中也得到相似的预测结果:肺鳞癌(HR 0.604,95%CI 0.383~0.951,P=0.007)及肝细胞肝癌(HR 0.551,95%CI 0.307~0.987,P=0.011)。GSEA结果提示,上述5个LncRNA的表达模式与肿瘤细胞的细胞周期调控有关。结论基于AC004551.1、MTOR-AS1、KCNAB1-AS2、FAM230G和LINC01283表达谱构建的预后模型可用于预测乳腺癌患者的预后,有利于进一步指导临床治疗。     

Survival methods, including those using competing risk analysis, are not appropriate for intensive care unit outcome studies

The preferred analysis for studies of mortality among patients treated in an intensive care unit should compare the proportions of patients who died during hospitalization. Studies that look for prognostic covariates should use logistic regression. Survival methods, such as the proportional hazards model, or methods based on competing risk analysis are not appropriate because prolonged survival among patients that die during their hospitalization does not benefit the patient and, therefore, should not be measured in the statistical analysis.     

Serum neopterin and soluble interleukin-2 receptor for prediction of a shock state in gram-negative sepsis

This study aimed to investigate the predictive value of neopterin and soluble interleukin-2 (IL-2) receptor for shock occurrence in gram-negative sepsis. We examined 57 patients admitted to an intensive care unit with gram-negative sepsis diagnosed according to preestablished criteria. Blood samples were collected every 24 hours and neopterin and soluble IL-2 receptor were measured by using commercially available test kits. To judge the predictive significance of these analyses the Cox proportional hazards regression model was used. Both neopterin (P < .05) and soluble IL-2 receptor (P < .01) were identified as significant predictors of a shock state, but the prognostic strength of neopterin exceeded that of soluble IL-2 receptor. To further assess if other factors could interfere with the predictive significance of both compounds, we also investigated other clinical and laboratory variables but these candidate predictors did not contribute any additional significant predictive information. The measurement of serum neopterin and soluble IL-2 receptor concentrations has predictability for identifying patients with gram-negative sepsis at risk for progression toward the syndrome of septic shock.     

Survival methods,including those using competing risk analysis,are not appropriate for intensive care unit outcome studies

The preferred analysis for studies of mortality among patients treated in an intensive care unit should compare the proportions of patients who died during hospitalization. Studies that look for prognostic covariates should use logistic regression. Survival methods, such as the proportional hazards model, or methods based on competing risk analysis are not appropriate because prolonged survival among patients that die during their hospitalization does not benefit the patient and, therefore, should not be measured in the statistical analysis.     

Improvement in clinical outcome by early nephrology referral in type II diabetics on maintenance peritoneal dialysis.

OBJECTIVES: To evaluate the influence of early nephrology referral on clinical outcome in type II diabetes mellitus patients on maintenance peritoneal dialysis (PD). DESIGN: This is a retrospective study in a single University Hospital in Taiwan. PATIENTS: This study analyzed the type II diabetic patients entering our PD program from February 1988 to June 2000. Patients that were presented to a nephrologist more than 6 months before starting dialysis were defined as early referrals (ER). Patients were considered late referrals (LR) if they were transferred to the nephrology department within 6 months before initial dialysis. MAIN OUTCOME MEASURES: Patient survival and technique survival curves were derived from Kaplan-Meier analysis and were compared using the Cox-Mantel log rank test. Covariates were analyzed with Cox proportional hazards model. RESULTS: 52 type II diabetic patients were enrolled in this study: 16 in the ER group and 36 in the LR group. Patient survival was better in the ER group than in the LR group [relative risks [exp(coef)] 0.42; 95% confidence interval 0.152-0.666; p < 0.05]. The improved survival in the ER group was independent of age at dialysis, good glycemic control, and residual renal function, as indicated in the multivariate analysis with stepwise regression by Cox proportional hazards model. The ER group was also associated with better technique survival. CONCLUSIONS: These results suggest that early nephrology referral before initiating dialysis is associated with improved long-term clinical outcome in type II diabetics on maintenance PD.     

Cumulative Hazard Analysis of J-Wire Fracture in the Accufix Series of Atrial Permanent Pacemaker Leads

To permit a more complete analysis of J-wire fracture in the Accufix series of atrial permanent pacemaker leads, the time to occurrence of all known fractures and injuries has been redefined relative to the duration of risk exposure, that is, according to the interval of time between implant and occurrence of the event. This redefinition permits application of a cumulative hazards model to the data, which previously has not been explored. Predictors of J-wire fracture can be tested using this method. This also permits parametric curve-fitting for determination of linearity or constancy of risk of events over time. Results: Among 2,063 Multicenter Study (MCS) leads analyzed, 381 fractures of the J-wire were identified. Stratified analysis based on cumulative hazard curves identified a more open shape of the J-wire as predictive of fracture, which supports the results previously reported based on logistic regression analysis. Fitting a Weibull curve to the cumulative hazard of J-wire fracture gives a shape parameter equal to 0.85. This value indicates that the instantaneous hazard of J-wire fracture decreased over time from implant. Conclusions: (1) The cumulative hazard function can be used to examine predictors of J-wire fracture and preliminary findings support the previously identified predictor of J shape; (2) Based on these analyses, the rate of J-wire fracture appears to decrease slightly as time from implant increases.     

红细胞分布宽度对ICU急性呼吸衰竭患者预后的评估

目的探讨红细胞分布宽度（RDW）对急性呼吸衰竭（ARF）患者预后的评估价值。 方法提取来源于MIMIC-Ⅲ数据库的数据,根据国际ICD9-CODE诊断编码,查询ARF患者相关信息。用SQL语言提取患者的一般资料、并发症、评分、实验室检查结果等,并以患者28 d病死率、90 d病死率为主要指标,ICU病死率、院内病死率、入ICU时间和院内时间为次要指标。所有研究对象根据受试者工作特征曲线（ROC）最佳截断值分为高RDW组和正常RDW组,比较2组患者的基线数据及临床结局,采用Kaplan-Meier生存曲线分析2组患者28 d、90 d的累积生存率,Cox比例风险回归模型分析ARF患者28 d、90 d的死亡风险,并计算ROC的曲线下面积（AUC）。 结果共纳入6286例ARF的患者,正常RDW组和高RDW组28 d病死率分别为21.3%、33.7%,90 d病死率分别为27.8%、44.8%,倾向性评分匹配（PSM）后2组28 d病死率分别为25.9%、31.9%,90 d病死率分别为33.5%、43.1%,组间比较差异均有统计学意义（P＜0.001）;进行累积生存分析时发现,高RDW组28 d和90 d的累积生存率较正常RDW组均明显降低,差异有统计学意义（P＜0.001）;采用Cox比例风险回归模型进行分析,RDW每升高1%,ARF患者28 d、90 d的死亡风险显著增加。在28 d病死率预测价值上,RDW、序贯器官衰竭评估（SOFA）评分的ROC的AUC无明显差异（0.663 vs 0.662,P=0.926）,但在90 d病死率上,RDW的ROC的AUC高于SOFA评分,差异有统计学意义（0.678 vs 0.651,P=0.003）。 结论RDW升高可能是ARF患者预后的一个有价值的指标。     

Prediction of all-cause mortality by the Minnesota Multiphasic Personality Inventory Optimism-Pessimism Scale scores: study of a college sample during a 40-year follow-up period

OBJECTIVE: To examine a measure of explanatory style, the Optimism-Pessimism (PSM) scale derived from college-entry Minnesota Multiphasic Personality Inventory scores, as a predictor of all-cause mortality. SUBJECTS AND METHODS: A total of 7007 students entering the University of North Carolina at Chapel Hill completed the Minnesota Multiphasic Personality Inventory during the mid-1960s. Of those students, 6958 had scores on the PSM scale and data for all-cause mortality through 2006. Scores on the PSM scale were evaluated as predictors of mortality using the Cox proportional hazards regression model, adjusted for sex. During the 40-year follow-up period, 476 deaths occurred. RESULTS: Pessimistic individuals who scored in the upper tertile of the distribution had decreased rates of longevity (hazard ratio, 1.42; 95% confidence Interval, 1.13-1.77) compared with optimistic individuals who scored in the bottom tertile of the distribution. CONCLUSION: In a model that adjusted only for sex, a measure of optimistic vs pessimistic explanatory style was a significant predictor of survival during a 40-year follow-up period such that optimists had Increased longevity.     

Risk Factors for Incident Peripheral Arterial Disease in Type 2 Diabetes: Results From the Bypass Angioplasty Revascularization Investigation in Type 2 Diabetes (BARI 2D) Trial

OBJECTIVE

The aim of this article was to define risk factors for incidence of peripheral arterial disease (PAD) in a large cohort of patients with type 2 diabetes mellitus (T2DM), overall and within the context of differing glycemic control strategies.

RESEARCH DESIGN AND METHODS

The Bypass Angioplasty Revascularization Investigation in Type 2 Diabetes (BARI 2D) randomized controlled trial assigned participants to insulin-sensitizing (IS) therapy versus insulin-providing (IP) therapy. A total of 1,479 participants with normal ankle-brachial index (ABI) at study entry were eligible for analysis. PAD outcomes included new ABI ≤0.9 with decrease at least 0.1 from baseline, lower extremity revascularization, or lower extremity amputation. Baseline risk factors within the overall cohort and time-varying risk factors within each assigned glycemic control arm were assessed using Cox proportional hazards models.

RESULTS

During an average 4.6 years of follow-up, 303 participants (20.5%) experienced an incident case of PAD. Age, sex, race, and baseline smoking status were all significantly associated with incident PAD in the BARI 2D cohort. Additional baseline risk factors included pulse pressure, HbA_1c, and albumin-to-creatinine ratio (P < 0.05 for each). In stratified analyses of time-varying covariates, changes in BMI, LDL, HDL, systolic blood pressure, and pulse pressure were most predictive among IS patients, while change in HbA_1c was most predictive among IP patients.

CONCLUSIONS

Among patients with T2DM, traditional cardiovascular risk factors were the main predictors of incident PAD cases. Stratified analyses showed different risk factors were predictive for patients treated with IS medications versus those treated with IP medications.