Stratified Cox Regression Analysis Research Articles

Serum Lactate Dehydrogenase Is Prognostic for Survival in Patients with Bone Metastases from Breast Cancer: A Retrospective Analysis in Bisphosphonate-Treated Patients

Survival is highly variable in women with bone metastases from breast cancer and prognostic factors are needed. We analyzed data from a phase III trial comparing zoledronic acid (ZOL) with pamidronate in patients with breast cancer and bone metastases to identify variables prognostic for overall survival. Patients who received ZOL (n = 435) with bone marker assessments and complete baseline data were included. Relative risks (RR) of death over 24 months were assessed using a stratified Cox regression analysis. A reduced model was generated using stepwise backward elimination until only significant (P < 0.05) variables remained. Only 5 of 19 variables analyzed remained significantly prognostic for survival in the reduced multivariate model. These included age more than 50 years (RR 1.78-2.53, P ≤ 0.01 for each decade >50 versus ≤ 50); Functional Assessment of Cancer Therapy-General (FACT-G) score less than 65 units (P < 0.05 vs. ≥ 75 units); impaired (PS ≥ 1) versus fully active (PS = 0) Eastern Cooperative Oncology Group (ECOG) performance status (RR 1.74, P < 0.01); prior versus no prior chemotherapy (RR 1.97; P < 0.01), and lactate dehydrogenase (LDH) levels. Lactate dehydrogenase ≥ upper limit of normal (ULN) but < 2 × ULN correlated with a two-fold increased risk of death, and LDH > 2 × ULN correlated with a six-fold increased risk of death versus LDH < ULN (P < 0.0001 for both). Baseline bone marker levels were not significantly correlated with survival after adjustment for other significant covariates. This retrospective analysis shows that LDH levels correlate strongly with survival in patients with bone metastases from breast cancer and confirms the relevance of previously described prognostic factors.

Development and validation of comorbidity index in South Korea

This study aims to develop and validate a new comorbidity index using data from hospitalized patients in South Korea. Retrospective cohort study. Hospital inpatients. Data from 4677 hospitalized patients aged 40-79 who had admitted in a medical center in Korea between September and December in 1997 were individually linked to national mortality data through December 2007. Among them, 3274 patients (70%) were randomly included in the development data set and the other 1403 patients (30%) in the validation data set. Another 3413 liver cancer patients from the same hospital were used to validate the index. Comorbidity index and mortality risk. Based on mortality using stratified Cox regression analyses, comorbidity scores ranging from 1 to 4 were assigned for 20 comorbid conditions. Summation of the scores produced the new comorbidity index (Asan comorbidity index, ACI). C-indices in the Cox regression analyses showed the greatest increase [0.589 (95% confidence interval, 0.568-0.609) in age- and sex-adjusted model; 0.697 (0.678-0.716) in the ACI and 0.664 (0.645-0.684) in the Charlson comorbidity index, respectively] when the ACI was added in the validation models. A newly developed comorbidity index using Korean hospitalized patient data based on the International Classification of Disease, 10th Revision (ICD-10) was valid among both general medical inpatients and liver cancer patients. This index may well be widely used in various health-care settings in Korea where patients' information on health conditions is coded with ICD-10.

Familial Effects on Ischemic Stroke

Background— Previous studies on familial risk of ischemic stroke have supported genetic influence on the disease incidence. This study aimed to characterize these familial effects in a nationwide population-based study by taking into account sibling relations, sex of siblings, and age of onset, with respect to ischemic stroke incidence. Methods and Results— Incident ischemic stroke cases identified from the Swedish Hospital Discharge and Cause of Death Registers between 1987 and 2007 were linked to their stroke-free siblings (study participants), forming an exposed sib-pair. Each exposed sib-pair was matched up to 5 unexposed sib-pairs from the Multi-Generation Registry by birth and calendar years. Incident ischemic stroke risk was assessed using hazard estimates obtained from stratified Cox regression analyses. A total of 30 735 exposed and 152 391 unexposed study participants were included in the analyses. The overall risk of incident ischemic stroke when exposed was significantly increased (relative risk, 1.61; 95% confidence interval, 1.48–1.75; P &lt;0.001). Familial risk was higher in full (relative risk, 1.64; 95% confidence interval, 1.50–1.81; P &lt;0.001) than in half (relative risk, 1.41; 95% confidence interval, 1.10–1.82; P =0.007) siblings. Familial risk of early ischemic stroke almost doubled when exposed to early ischemic stroke (relative risk, 1.94; 95% confidence interval, 1.41–2.67; P &lt;0.001). Conclusions— There was a 60% increased risk for ischemic stroke in individuals having a sibling with prior stroke. The familial effect was even higher for full-sibling relations. Familial effects were observed in both male and female individuals, and no differential effects depending on the sex of either of the siblings were found.

Prediction of post-remission survival in acute myeloid leukaemia: a post-hoc analysis of the AML96 trial

The optimum post-remission treatment (PRT) in acute myeloid leukaemia (AML) is still a matter of debate. Consolidation treatments include chemotherapy with high-dose cytarabine, or allogeneic or autologous haemopoietic stem cell transplantation (HSCT). In a post-hoc analysis of the AML96 trial (NCT00180115), our aim was to differentiate groups of patients according to the treatments that would provide them optimum benefit. In the multicentre AML96 trial, 586 patients (aged 15-60 years) with AML--excluding those with t(8;21)--who were in complete remission after double induction treatment were consolidated with allogeneic HSCT, autologous HSCT, or chemotherapy containing high-dose cytarabine in a priority-based and risk-adapted manner. We assessed the association between potentially prognostic variables and overall survival after complete remission by use of a stratified Cox regression analysis. With the significant variables of the resulting model, we developed a PRT score in 452 patients with a complete dataset. This score was then validated by use of data from 407 patients from the AML2003 trial (NCT00180102). Age, percentage of CD34-positive blasts, FLT3-ITD mutant-to-wild-type ratio, cytogenetic risk, and de-novo or secondary AML were identified as independent prognostic factors, and included in the PRT score. The PRT score separated patients in AML96 into three groups: favourable (n=190; 3-year survival 68%, 95% CI 60-74), intermediate (n=198; 49%, 42-56), and unfavourable (n=64; 20%, 12-31). All pair-wise comparisons of two of three PRT score groups were significant in the log-rank test (p<0·0001). Similar results were noted when data from AML2003 were used: 3-year survival for the favourable group (n=265) was 69% (62-76), for the intermediate group (n=114) it was 61% (50-71), and for the unfavourable group (n=28) it was 46% (24-65). The overall comparison between the three risk groups resulted in significantly different survival probabilities (p=0·015). We also analysed response to treatment in AML96 in each of the PRT score groups. In the favourable group, patients given allogeneic HSCT (n=60) had higher survival probabilities (82%, 69-89) than did those given chemotherapy (n=56, 55%, 41-67; p=0·0012) or autologous HSCT (n=74, 66%, 54-76; p=0·044). In the intermediate PRT score group, patients given autologous HSCT (n=69) had the best survival probabilities (62%, 50-72) compared with those given chemotherapy (n=72, 41%, 30-52; p=0·0006) or allogeneic HSCT (n=57, 44%, 31-56; p=0·0045). The PRT score groups could help physicians to tailor treatment for patients with AML and our results lend support to the use of autologous HSCT in patients aged 60 years or younger with an intermediate PRT score. Deutsche Krebshilfe.

Mortality among Patients with Cleared Hepatitis C Virus Infection Compared to the General Population: A Danish Nationwide Cohort Study

BackgroundThe increased mortality in HCV-infected individuals partly stems from viral damage to the liver and partly from risk-taking behaviours. We examined mortality in patients who cleared their HCV-infection, comparing it to that of the general population. We also addressed the question whether prognosis differed according to age, substance abuse (alcohol abuse and injection drug use) and comorbidity.Methodology/Principal FindingsPatients with cleared HCV-infection were categorized into one of 8 groups according to age (20–39 years or 40–69 years) and patient characteristics (no substance abuse/no comorbidity; substance abuse/no comorbidity; no substance abuse/comorbidity; and substance abuse/comorbidity). For each patient, 4 age- and gender-matched individuals without substance abuse or comorbidity were selected from the general population, comprising a total of 8 comparison cohorts. We analyzed 10-year survival and used stratified Cox Regression analysis to compute mortality rate ratios (MRRs), comparing mortality between the 8 patient groups and the comparison cohorts, adjusting for personal income. Among patients without substance abuse or comorbidity, those aged 40–69 years had the same mortality as the comparison cohort (10-year survival: 95% (95% confidence interval [CI]: 93%–97%), MRR: 1.3 (95% CI: 0.8–2.3)), whereas those aged 20–39 years had higher mortality than the comparison cohort (10-year survival: 93% versus 99%, MRR: 5.7 (95% CI: 2.3–14.0). For both age categories, substance abuse and comorbidity decreased survival and increased MRRs. Patients aged 40–69 years with substance abuse and comorbidity suffered from substantial mortality (MRR: 12.5 (95% CI: 5.1–30.6)).ConclusionsMortality in patients aged 40–69 years with cleared HCV-infection is comparable to individuals without HCV, provided they have no substance abuse or comorbidity. Any substance abuse and/or comorbidity not captured in the registries used for our study could explain the increased mortality in patients aged 20–39 years without documented substance abuse or comorbidity.

Fasting C-Peptide Levels and Death Resulting From All Causes and Breast Cancer: The Health, Eating, Activity, and Lifestyle Study

To examine the association between serum C-peptide, a marker of insulin secretion, measured 3 years after a breast cancer diagnosis, and death resulting from all causes and breast cancer. This was a prospective, observational study of 604 women enrolled onto the Health, Eating, Activity, and Lifestyle (HEAL) Study who were diagnosed with local or regional breast cancer between 1995 and 1998 and observed until death or December 31, 2006, whichever came first. The hazard ratio (HR) for all deaths and deaths owing to breast cancer and 95% CIs for the HR were estimated using multivariable stratified Cox regression analyses. Among women without type 2 diabetes, fasting C-peptide levels were associated with an increased risk of death resulting from all causes and from breast cancer. A 1-ng/mL increase in C-peptide was associated with a 31% increased risk of any death (HR = 1.31; 95% CI, 1.06 to 1.63; P = .013) and a 35% increased risk of death as a result of breast cancer (HR = 1.35; 95% CI, 1.02 to 1.87, P = .048). Associations between C-peptide levels and death as a result of breast cancer were stronger in certain subgroups, including women with type 2 diabetes, women with a body mass index less than 25 kg/m(2), women diagnosed with a higher stage of disease, and women whose tumors were estrogen receptor positive. Treatment strategies to reduce C-peptide levels in patients with breast cancer, including dietary-induced weight loss, physical activity, and/or use of insulin-lowering medications, should be explored.

Survival of cancer patients with prior monoclonal gammopathy of undetermined significance

Introduction It is unknown whether a prior diagnosis of monoclonal gammopathy of undetermined significance (MGUS) affects cancer survival. Design and methods We linked data on 1652 cases of MGUS diagnosed during 1978–2006 in North Jutland County, Denmark with the Danish Cancer Registry and the National Patient Registry to obtain information on survival of cancer patients with a previous MGUS compared with that of cancer patients without MGUS, matched on cancer type, age, sex and year of diagnosis. Stratified Cox regression analysis was used to compute mortality rate ratios controlling for the matching factors and comorbidity. Results We included 323 cancer patients with previously detected MGUS and 3154 comparison cancer patients without MGUS. The 5-year survival probability was 26.2% (95% CI, 21.2%–31.5%) in cancer patients with MGUS and 30.5% (28.8%–32.1%) in cancer patients without MGUS. The adjusted mortality rate ratio (MRR) was = 0.94 (95% CI, 0.82–1.09). Survival following a diagnosis of multiple myeloma, the cancer site of main interest, did not differ according to a preceding MGUS diagnosis. Among patients with immune-related cancers (liver, cervix, malignant melanoma and non-melanoma skin cancers), a preceding MGUS diagnosis was associated with reduced survival (adjusted MRR = 1.49 (95% CI: 0.96–2.31)). In contrast, for other non-haematological cancers a prior MGUS diagnosis was associated with a lower MRR (0.78 (95% CI, 0.63–0.96)). Conclusion Our study does not indicate that previously detected MGUS is a prognostic factor for cancer survival in general.

Associations of Hyperglycemia and Insulin Usage With the Risk of Cancer in Type 2 Diabetes: The Hong Kong Diabetes Registry

OBJECTIVEInsulin has mitogenic effects, although hyperglycemia may be a risk factor for cancer in type 2 diabetes. It remains uncertain whether use of insulin increases cancer risk because of its effect on cell growth and proliferation or decreases cancer risk because of its glucose-lowering effect.RESEARCH DESIGN AND METHODSA 1:2-matched new insulin user cohort on age (±3 years), smoking status, and likelihood of initiating insulin therapy (±0.05) was selected from a cohort of 4,623 Chinese patients with type 2 diabetes, free of cancer, and naive to insulin at enrollment. Stratified Cox regression analysis on the matched pairs was used to obtain hazard ratios (HRs) of insulin therapy and A1C for cancer risk. A structured adjustment scheme was used to adjust for covariates.RESULTSOf 973 new insulin users, 971 had matched nonusers (n = 1935). The cancer incidence in insulin nonusers was much higher than that in insulin users (49.2 vs. 10.2, per 1,000 person-years, P < 0.0001). After further adjustment for all other covariates with a P value less than 0.3 and nonlinear associations with cancer, A1C was associated with an increased cancer risk (HR per percentage 1.26, 95% CI 1.03–1.55), whereas use of insulin was associated with a decreased cancer risk (HR of insulin users vs. nonusers: 0.17, 0.09–0.32). Consistent results were found in analyses including all 973 insulin users and 3,650 nonusers.CONCLUSIONSIn Chinese patients with type 2 diabetes, hyperglycemia predicts cancer, whereas insulin usage was associated with a reduced cancer risk.

Glycaemic control in newly diagnosed diabetes patients and mortality from ischaemic heart disease: 20-year follow-up of the HUNT Study in Norway

To assess the influence of glycaemic control on long-term mortality from ischaemic heart disease (IHD) in patients with newly diagnosed diabetes. In a large population study in Norway, people > or =40 years with non-fasting glucose > or =8 mmol/L were invited to a fasting glucose test, and if the fasting value was <7 mmol/L, an oral glucose tolerance test was also performed. Among people who were diagnosed with diabetes, 205 patients were followed with annual measurements of HbA1c in order to monitor glycaemic control. Stratified Cox regression analysis was used to compare IHD mortality rates during 20 years of follow-up, with comparison of newly diagnosed diabetes patients and a matched group of 205 individuals without diabetes. Among patients, we also assessed the relation of HbA1c with IHD mortality. After adjustment for potentially confounding factors, IHD mortality in the total diabetes group was substantially higher (HR 1.8, 95% CI, 1.0-3.4) compared with the comparison group. However, the increased risk was particularly high in patients with HbA1c in the highest quartile (HR 4.2, 95% CI, 2.1-8.1). Analysing HbA1c as a continuous time-varying variable showed 30% (HR 1.3, CI 1.1-1.5) higher risk per increment of HbA1c among diabetes patients without known CVD at baseline. Poor long-term glycaemic control is associated with a substantial increase in the risk of dying from IHD in patients with diabetes, whereas in patients with reasonably good control, risk of dying from IHD may not substantially differ from that of people without diabetes.

Outcomes in African-Americans vs. Caucasians Using Thymoglobulin or Interleukin-2 Receptor Inhibitor Induction: Analysis of USRDS Database

Aim: We used the USRDS database to test the hypothesis that graft survival was similar using either rabbit antithymocyte globulin (rATG) vs. interleukin-2 receptor inhibitor (IL2i) in the Prograf era. We further explored the variable of race in the two groups of patients. Methods: We conducted a retrospective cohort study of kidney transplant patients in the USRDS from 2000 through 2005 to compare graft survival (including death) using rATG vs. IL2i with particular reference to outcomes between African-Americans vs. Caucasians. Kaplan-Meier analysis was performed to assess patient and graft survival after transplantation, stratified by recipient induction with rATG versus IL2i. Cox regression analysis was performed to assess adjusted survival after transplantation, assessing whether induction rATG (vs. IL2i) was significant as an interaction term (i.e. an effect modifier) with black race for graft survival. Propensity score analysis was used to address potential confounding by indication. Results: In stratified Cox Regression analysis limited to IL2i, black race was significantly associated with graft loss (adjusted hazard ratio (AHR) 1.17, 95% CI, 1.09–1.26). In analysis limited to rATG induction, black race was not significant (AHR 1.00, 95% CI, 0.92–1.10). We detected a significant interaction between rATG and black race (in comparison with non-black race) for the development of graft loss (AHR, 0.86, 95% CI, 0.76–0.97). Analysis limited to black recipients showed that while use of rATG was not significantly different from IL2i (AHR 0.95, 95% CI 0.87–1.04), the direction of this association was in the opposite direction of non-blacks. Conclusions: Patient and graft survival were similar in African-American and Caucasian recipients of kidney transplantation using either rATG or IL2i. Limitations of the study are the retrospective nature of USRDS data, center-bias in using rATG vs. IL2i and lack of data on steroid dosage. Results of the present study call for a critical review of induction practices.

Recombinant Human Erythropoiesis Stimulating Agents in Cancer Patients: Individual Patient Data Meta-Analysis on Behalf of the EPO IPD Meta-Analysis Collaborative Group

Abstract Background: Erythropoiesis-stimulating agents (ESAs) reduce anemia in cancer patients; however, there are concerns that ESAs increase mortality in some patients. Within the framework of an international collaboration we thus conducted a patient-level meta-analysis to assess the effects of ESAs on mortality in cancer patients. Methods: We performed a meta-analysis, based on the intention-to-treat principle, of cancer patients enrolled in randomized controlled trials comparing epoetin alfa, epoetin beta or darbepoetin alfa plus red blood cell transfusions as needed versus transfusion alone, for prophylaxis or treatment of anemia while or after receiving anticancer therapy. Patient-level data were obtained and analyzed by independent statisticians at two academic departments. Primary endpoints were on-study mortality and overall survival in patients receiving chemotherapy, defined as all patients from studies in which =/&gt; 70% of study population received chemotherapy, and in all cancer patients regardless of anticancer therapy. On-study mortality was defined as death from any cause between date of randomization and 28 days after end of active study phase. Overall survival was defined as death from any cause between date of randomization and longest follow up available. Hazard ratios (HR) and 95% confidence intervals (CIs) were calculated per study and meta-analyzed with fixed-effects and random-effects models. Stratified multivariable Cox-regression analyses were conducted to assess the impact of baseline imbalances. Tests for interactions were used to identify differences in effect across pre-specified subgroups. All analyses were pre-specified in a peer-reviewed protocol (Bohlius et al. 2008). An independent steering committee consisting of clinicians and methodologists agreed on all analyses and interpretations. Independent investigators and representatives from manufacturers contributing data offered advice, but had no decision-making authority. Results: Data from 13,933 cancer patients enrolled in 53 studies were included in the analysis; 38 trials including 10,441 patients used mainly chemotherapy. Including all cancer patients ESAs increased on-study mortality by 17% (HR 1.17; 95% CI 1.06–1.30), with little evidence for a difference between chemotherapy and other trials (p for interaction=0.42), and worsened overall survival by 6% (HR 1.06; 95% CI 1.00–1.12). In the chemotherapy population on-study mortality was increased by 10% (HR 1.10, 95% CI 0.98–1.24) and overall survival was worsened by 4% (HR 1.04; 95% CI 0.97–1.11). Adjusting for known prognostic factors had little effect on the overall estimates. There was no conclusive evidence for effect modification by patient level characteristics such as age, sex, Hb and Hct at baseline, Hb ceiling, type and stage of tumor or study level characteristics (anticancer treatment, ESA treatment schedules, study design and quality) for the outcomes tested. Conclusion: ESA treatment in cancer patients increased on-study mortality by 17% and worsened overall survival by 6%. For patients undergoing chemotherapy the increase was less pronounced, but could not be excluded. In clinical practice, risks of ESAs must be balanced against benefits of ESAs depending on the clinical circumstances of the individual patient.

Usefulness of Microvolt T-Wave Alternans to Predict Outcomes in Patients With Ischemic Cardiomyopathy Beyond One Year

Previous studies have demonstrated that microvolt T-wave alternans (MTWA) screening effectively risk-stratifies patients with ischemic cardiomyopathy. Whether the prognostic utility of MTWA diminishes over 3 years of follow-up remains unknown. In this study, a prospective cohort of 768 patients with ischemic cardiomyopathy (left ventricular ejection fraction <35%) and no previous sustained ventricular arrhythmia was developed, of whom 514 (67%) screened MTWA nonnegative (positive and indeterminate). The mean follow-up period was 18 +/- 11 months. The primary end point was all-cause mortality and appropriate implantable cardioverter-defibrillator shocks. Stratified Cox regression analyses (by implantable cardioverter-defibrillator status) estimated the predictive power of MTWA within each year of follow-up and determined whether this diminished over time. There were 99 deaths (MTWA negative: 21 [8.3%]; MTWA nonnegative: 78 [15.2%]) and 33 appropriate implantable cardioverter-defibrillator shocks (MTWA negative: 3 [4.0%]; MTWA nonnegative: 30 [9.5%]). After multivariate adjustment, a nonnegative MTWA test result was associated with a greater than twofold increased risk for events in each of the 3 years of follow-up (year 1: stratified hazard ratio 2.19, 95% confidence interval 1.10 to 4.34, p = 0.03; year 2: stratified hazard ratio 3.36, 95% confidence interval 1.28 to 8.83, p = 0.01; year 3: stratified hazard ratio 2.06, 95% confidence interval 0.81 to 5.22, p = 0.13). There were no significant interactions between the time periods (year 1 vs year 2: p = 0.47; year 1 vs year 3: p = 0.92). In conclusion, MTWA reliably and consistently predicts mortality and arrhythmic risk throughout the first 2 to 3 years of follow-up. Although these findings need further validation, they suggest that rescreening with MTWA may not need to be performed more frequently than once every 2 years.

Morbidity in Klinefelter Syndrome: A Danish Register Study Based on Hospital Discharge Diagnoses

Klinefelter syndrome (KS) is the most prevalent sex chromosome disorder in man; it affects approximately one in 660 men and is a common cause of hypogonadism and infertility. Our current knowledge of morbidity in KS is based on observational studies and case reports and therefore is limited. We used Danish registers to obtain dates of hospital admissions and discharge diagnoses in a cohort of all males diagnosed with KS in Denmark and a randomly selected, age-matched control group. Our cohort consisted of 832 KS subjects and 4033 control subjects, contributing with a total of approximately 100,000 person years. We used stratified Cox regression analysis on main groups of diagnoses. Where significant results were found, subsequent analyses were performed on subgroups of diagnoses. We found a significantly increased risk of being hospitalized among the KS subjects [hazard ratio (HR), 1.69; 95% confidence interval, 1.54-1.86]. The increased admission risk was present in all but one of the main diagnosis groups, with the highest HRs for congenital malformations (HR, 10.7), psychiatric disorders (HR, 3.7), and endocrine and metabolic disorders (HR, 3.2). We compared hospitalization rates before and after the diagnosis of KS and found that the increased rate was present even before the diagnosis of KS. Males suffering from KS experienced an increased hospitalization rate from a variety of disorders. Some are likely to be caused by hypogonadism, and some may be linked to the syndrome per se, whereas others are not readily explained. However, other factors, e.g. socioeconomic, may be involved.

Low Testosterone Levels Predict Incident Depressive Illness in Older Men

Prior studies found that chronic low testosterone levels are associated with an increased risk of depression. We investigated whether low testosterone levels in older men predict depressive illness over 2 years, while controlling for age and medical morbidity. Participants were 748 men, aged 50 years or older, without prior ICD-9-diagnosed depressive illness, with a testosterone level obtained between 1995 and 1997. Measures were age, mean total testosterone levels (low: < or = 2.5 ng/mL), medical morbidity, and incidence and time to depressive illness. Men with low testosterone levels had a greater 2-year incidence of depressive illness (18.5% vs. 10.4%, df = 1, p = .006) and a shorter time to onset of depressive illness (log-rank chi(2) = 8.1, df = 1, p = .004). The unadjusted hazard ratio (HR) for depressive illness in men with low testosterone levels was 1.9 (95% confidence interval [CI] = 1.2 to 3.0, p = .005). After adjustment for age and medical morbidity, men with low testosterone levels continued to have a shorter time to depressive illness (adjusted HR = 2.1; 95% CI = 1.3 to 3.2, p = .002). Due to a significant interaction between age and medical morbidity, we conducted stratified Cox regression analyses and found that low testosterone levels and high medical morbidity or an age of 50 to 65 years were associated with increased depressive illness (p = .002). Low testosterone levels are associated with an earlier onset and greater incidence of depressive illness. Men with low testosterone levels who had high medical morbidity or were aged 50 to 65 years had an increased risk for depressive illness. Further prospective studies are needed to examine the role of testosterone in depressive illness in older men.

Practice Variability in the Management of Subrenal Arterial Stenoses in Seven Belgian Hospitals

This multicentre retrospective study describes the variation of therapeutic options, treatment outcomes and costs for treating subrenal arterial stenoses as observed in daily practice in 1991-99 in seven Belgian hospitals. Data were obtained from clinical record review and from the sickness fund claims database, and included preoperative functional state, presence of acute ischaemia, diabetes and polyvascular disease, state of the lower-leg run-off arteries, anatomical site and type of lesion, type of treatment, result at 30 days and up to 4 years. A total of 441 episodes were studied, but most analyses dealt with a subgroup of 140 lesions in the common iliac up to the superficial femoral artery. The proportion of surgical treatments (as compared to an endovascular or mixed approach) varied from 15% to 81% between the hospitals.In univariate patency analysis, relapse or failure rates at 4 years ranged from 5% for the common iliac artery to 35% for the superficial femoral artery. Polyvascular disease, a poor run-off, multiple stenoses and chronic occlusion were significant risk factors; age and diabetes were not. In the multivariate (stratified Cox regression) analysis, only a location in the superficial femoral artery and a poor preoperative clinical stage were significant risk factors, but type of therapeutic approach was not.The total average cost of treatment was 5,300 荤, of which 15% was contributed by the patient. Surgery was associated with longer stays (median at 11 days) than endovascular treatments (median 1 days), and was 1.9 times more expensive. In conclusion, the results of the present study suggest that a multidisciplinary approach, orienting the patient to the most appropriate therapeutic pathway, could increase both the quality and the cost-effectiveness of the care.