Ethnic disparities in risk of cardiovascular disease, end-stage renal disease and all-cause mortality: a prospective study among Asian people with Type 2 diabetes.

AIM: To study prospectively the ethnic-specific risks of cardiovascular disease, end-stage renal disease and all-cause mortality in patients with Type 2 diabetes mellitus among native Asian subpopulations. METHODS: A total of 2337 subjects with Type 2 diabetes (70% Chinese, 17% Malay and 13% Asian Indian) were followed for a median of 4.0 years. Time-to-event analysis was used to study the association of ethnicity with adverse outcomes. RESULTS: Age- and gender-adjusted hazard ratios for cardiovascular disease in ethnic Malay and Asian Indian subjects were 2.01 (1.40-2.88; P<0.0001) and 1.60 (1.07-2.41; P=0.022) as compared with Chinese subjects. Adjustment for conventional cardiovascular disease risk factors, including HbA1c , blood pressure and lipid profile, slightly attenuated the hazards in Malay (1.82, 1.23-2.71; P=0.003) and Asian Indian subjects (1.47, 0.95-2.30; P=0.086); However, further adjustment for baseline renal function (estimated GFR) and albuminuria weakened the cardiovascular disease risks in Malay (1.48, 0.98-2.26; P=0.065) but strengthened that in Asian Indian subjects (1.81, 1.14-2.87; P=0.012). Competing-risk regression showed that the age- and gender-adjusted sub-distribution hazard ratio for end-stage renal disease was 1.87 (1.27-2.73; P=0.001) in Malay and 0.39 (0.18-0.83; P=0.015) in Asian Indian subjects. Notably, the difference in end-stage renal disease risk among the three ethnic groups was abolished after further adjustment for baseline estimated GFR and albuminuria. There was no significant difference in risk of all-cause mortality among the three ethnic groups. CONCLUSIONS: Risks of cardiovascular and end-stage renal diseases in native Asian subjects with Type 2 diabetes vary substantially among different ethnic groups. Differences in prevalence of diabetic kidney disease may partially explain the ethnic disparities.

All-cause and cause-specific mortality after hip fracture among Chinese women and men: the Singapore Chinese Health Study.

UNLABELLED: All-cause mortality risk persisted for 5 years after hip fractures in both men and women. There may be gender-specific differences in effect and duration of excess risk for cause-specific mortality after hip fracture. INTRODUCTION: To determine all-cause and cause-specific mortality risk in the first 5 years after hip fracture in an Asian Chinese population. METHODS: The Singapore Chinese Health Study is a population-based cohort of 63,257 middle-aged and elderly Chinese men and women in Singapore recruited between 1993 and 1998. This cohort was followed up for hip fracture and death via linkage with nationwide hospital discharge database and death registry. As of 31 December 2008, we identified 1,166 hip fracture cases and matched five non-fracture cohort subjects by age and gender for each fracture case. Cox proportional hazards and competing risks regression models with hip fracture as a time-dependent covariate were used to determine all-cause and cause-specific mortality risk, respectively. RESULTS: Increase in all-cause mortality risk persisted till 5 years after hip fracture (adjusted hazard ratio, aHR = 1.58 [95 % CI, 1.35-1.86] for females and aHR = 1.64 [95 % CI, 1.30-2.06] for males). Men had higher mortality risk after hip fracture than women for deaths from stroke and cancer up to 1 year post-fracture but women with hip fracture had higher coronary artery mortality risk than men for 5 years post-fracture. Men had higher risk of death from pneumonia while women had increased risk of death from urinary tract infections. There was no difference in mortality risk by types of hip fracture surgery. CONCLUSIONS: All-cause mortality risk persisted for 5 years after hip fractures in men and women. There are gender-specific differences in effect size and duration of excess mortality risk from hip fractures between specific causes of death.

Occupation and risk of non-Hodgkin's lymphoma in Singapore.

BACKGROUND: Some epidemiological studies have reported that teachers may be at increased risk of non-Hodgkin's lymphoma (NHL), but results are inconsistent. AIMS: To examine the possible association between occupation and risk of NHL in the Singapore population. METHODS: A hospital-based interviewer-administered case-control study was carried out in five major hospitals in Singapore between April 2004 and December 2008. A complete occupational history, which included all jobs lasting over 1 year since graduation from school, was obtained for each participant. The Singapore Standard Occupational Classification was used for coding all occupations recorded. RESULTS: Eight hundred and thirty controls and 465 NHL cases, comprising B-cell (n = 404, 87%) as well as T- and NK-cell (n = 61, 13%) neoplasms, were recruited. Having ever worked as a teacher was associated with a significantly higher risk of NHL (adjusted OR 2.04, 95% CI 1.12-3.72). Teachers who had taught for ≤10 years had a significantly higher risk of NHL (adjusted OR 2.44, 95% CI 1.11-5.34), but we did not observe an elevated risk for those who reported teaching for >10 years. Among the 31 teachers with NHL, 23% taught in upper secondary schools, with equal proportions (13%) teaching in primary and pre-primary schools, respectively. The remainder taught in other settings. CONCLUSIONS: Teachers come into frequent contact with children and may consequently have higher rates of exposure to common infectious agents. Therefore, the hypothesis of an infective aetiology of NHL may be supported by our findings.

An association between floor vacuuming and dust-mite and serum eosinophil cationic protein in young asthmatics.

UNLABELLED: Although vacuum cleaning is recommended to reduce allergen levels and improve asthma and allergic rhinitis symptoms, some studies suggest it may increase allergen load in homes. We conducted a cross-sectional study to determine if home floor vacuuming was associated with sensitization to dust-mites and cockroaches, and serum eosinophil cationic protein (ECP), a biomarker for atopy, in 102 physician-diagnosed spirometry-confirmed asthmatics. We collected data on floor type, floor cleaning method and frequency, asthma severity, allergy medications used, serum ECP and skin prick tests (SPT) to three dust-mites [Dermatophagoides pteronyssisinus (Der p), Dermatophagoides farinae (Der f) and Blomia tropicalis (Blo t)] and two cockroaches [Periplanata americana (Per a) and Blatella germanica (Bla g)]. Those who vacuumed had increased sensitization to three dust-mites [adjusted ORs (95%CI) = Der p: 26.6 (1.8-405.2); Der f: 44.8 (3.2-620.9); Blo t: 14.1 (1.8-108.1)] but not to cockroaches, adjusted for cleaning frequency and other methods of floor cleaning. Subjects who vacuumed their floor had higher levels of serum ECP than those who did not [adjusted median difference (95%CI): 9.4 (1.1-17.7)], adjusted for use of nasal corticosteroids among those with allergic rhinitis. Vacuuming is associated with increased sensitization to dust-mite allergens and higher serum ECP. PRACTICAL IMPLICATIONS: We found an association between floor vacuuming and increased sensitization to dust-mite allergens and higher levels of an atopy biomarker. Current recommendations to use vacuuming to control allergen exposure and allergic conditions may need to be reconsidered until further studies are performed.

Reduced GFR and albuminuria in Chinese type 2 diabetes mellitus patients are both independently associated with activation of the TNF-alpha system.

AIMS/HYPOTHESIS: The involvement of chronic inflammation in albuminuria and renal function was investigated in a cross-sectional study of 320 type 2 diabetic Chinese patients from the Singapore Diabetes Cohort Study. METHODS: Plasma levels of TNF-alpha and its two cellular receptors and of IL-6 and C-reactive protein (CRP) were measured. A composite TNF-alpha score was extracted using principal component analysis. Multiple linear regression analysis was implemented to evaluate the relationship between log( e ) (ln) albumin:creatinine ratio (ACR) and estimated GFR (eGFR) with the inflammatory variables and other clinical covariates. A Bonferroni correction was applied based on the total number of variables entered into regression analyses. RESULTS: ln ACR was significantly associated with TNF-alpha score independently of eGFR even after a Bonferroni correction. TNF-alpha score was also significantly associated with eGFR independently of ln ACR even after correction for multiple testing. These findings were similar when the individual molecules of the TNF-alpha system were analysed separately instead of using the composite TNF-alpha score. No association was observed for IL-6 and CRP with either renal trait. Diabetes duration was a significant predictor for ln ACR but not eGFR. Conversely, age was significantly associated with eGFR but not ln ACR. CONCLUSIONS/INTERPRETATION: Activation of the TNF-alpha system may potentially exert independent effects on ln ACR and eGFR in type 2 diabetes. Because of the study design, one may also consider the possibility that changes in these renal traits may conversely be responsible for such an inflammatory response.

Prognostic significance of thymidylate synthase, dihydropyrimidine dehydrogenase and thymidine phosphorylase protein expression in colorectal cancer patients treated with or without 5-fluorouracil-based chemotherapy.

BACKGROUND: Low tumour expression levels of thymidylate synthase (TS), dihydropyrimidine dehydrogenase (DPD) and thymidine phosphorylase (TP) have been linked with improved outcome for colorectal cancer (CRC) patients treated with 5-fluorouracil (5-FU). It is unclear whether this occurs because such tumours have better prognosis or they are more sensitive to 5-FU treatment. PATIENTS AND METHODS: Associations between TS, DPD and TP levels, determined by tissue microarrays and immunohistochemistry, and survival was evaluated in 945 CRC patients according to treatment status. RESULTS: Low TS and DPD expression associated with worse prognosis in stage II [hazard ratio (HR) = 1.69, 95% confidence interval (CI) (1.09-2.63) and HR = 1.92 (95% CI 1.23-2.94), respectively] and stage III CRC patients treated by surgery alone [HR = 1.39 (95% CI 0.92-2.13) and HR = 1.49 (95% CI 1.02-2.17), respectively]. Low TS, DPD and TP associated with trends for better outcome in stage III patients treated with 5-FU [HR = 0.81 (95% CI 0.49-1.33), HR = 0.70 (95% CI 0.42-1.15) and HR = 0.66 (95% CI 0.39-1.12), respectively]. CONCLUSION: Low TS and DPD expression are prognostic for worse outcome in CRC patients treated by surgery alone, whereas low TS, DPD and TP expression are prognostic for better outcome in patients treated with 5-FU chemotherapy. These results provide indirect evidence that low TS, DPD and TP protein expression are predictive of good response to 5-FU chemotherapy.

Estimating sample size in the presence of competing risks - Cause-specific hazard or cumulative incidence approach?

In designing randomised clinical trials involving competing risks endpoints, it is important to consider competing events to ensure appropriate determination of sample size. We conduct a simulation study to compare sample sizes obtained from the cause-specific hazard and cumulative incidence (CMI) approaches, by first assuming exponential event times. As the proportional subdistribution hazard assumption does not hold for the CMI exponential (CMIExponential) model, we further investigate the impact of violation of such an assumption by comparing the results obtained from the CMI exponential model with those of a CMI model assuming a Gompertz distribution (CMIGompertz) where the proportional assumption is tenable. The simulation suggests that the CMIExponential approach requires a considerably larger sample size when treatment reduces the hazards of both the main event, A, and the competing risk, B. When treatment has a beneficial effect on A but no effect on B, the sample sizes required by both methods are largely similar, especially for large reduction in the main risk. If treatment has a protective effect on A but adversely affects B, then the sample size required by CMIExponential is notably smaller than cause-specific hazard for small to moderate reduction in the main risk. Further, a smaller sample size is required for CMIGompertz as compared with CMIExponential. The choice between a cause-specific hazard or CMI model in competing risks outcomes has implications on the study design. This should be made on the basis of the clinical question of interest and the validity of the associated model assumption.

Renal Artery Diameter Is a Surrogate Marker for Kidney Volume in Living Kidney Donors.

Donor kidney volume (KV) is an increasingly important parameter evaluated before living kidney donation; however, KV measurements on computed tomographic (CT) scanning requires a manually intensive process of manual or semiautomatic segmentation of kidneys with interobserver variation. Renal artery diameter (RAD) is an easier marker to measure, and this study aims to investigate the relationship between donor RAD and KV. METHODS: A retrospective review of 77 patients who underwent living donor nephrectomy was conducted. Bilateral KVs were measured based on contrast-enhanced CT scan imaging, and renal artery maximum diameter was measured by direct visualization on the arterial phase of transverse CT sections. RESULTS: On regression analysis, there was a significant association between the right and left RADs and their ipsilateral KVs with a regression coefficient of 7.9 (95% CI, 1.3-14.5; P = .02) and 9.8 (95% CI, 3.3-16.3; P = .004), respectively. Mean total RAD correlated with total KV with a regression coefficient of 9.3 (95% CI, 3.8-14.7; P = .001) and weakly correlated with estimated glomerular filtration rate with a Pearson coefficient of .10. CONCLUSIONS: This study demonstrates that renal artery size is positively associated with KV and may be used as an easily measured surrogate marker for kidney size with its attended implications in living donor transplantation.

An exploration of the applicability of the refined disease risk index and its integration with other independent risk factors for individualized prognostication.

The refined disease risk index (DRI) is a powerful prognostic model based solely on the disease type and stage for predicting survival outcomes of various hematological malignancies after allogeneic transplant. Here, we analyzed our series of 690 patients transplanted over the past 15 years, and showed that besides overall survival (OS), the refined DRI is also able to segregate event-free survival and relapse mortality in our cohort of largely Southeast Asian patients with a long and complete follow-up. Stratification by refined DRI remains statistically significant even when broken down by specific diseases each with a smaller number of patients, as well as for a small subset of patients younger than 18 years old, providing a robust model for prognostication. Multivariable analysis shows that refined DRI, age, year of transplant and donor type are independent risk factors for OS. We further demonstrated here that prognostication for a given patient with a specific disease can be made more discriminating by integrating independent risk factors such as age and donor type with the refined DRI. The future development of prognostic system incorporating the refined DRI with patient- and transplant-related risk factors will provide a more precise estimate of transplant outcome.

Prevalence and profile of females at risk of eating disorders in Singapore.

INTRODUCTION: The objective of this study was to assess the prevalence of young females at risk of developing eating disorders (ED) and the associated socio-demographical variables. METHODS: A set of self-administered questionnaires consisting of an Eating Attitude Test (EAT), an Eating Disorder Inventory (EDI) and a socio-demographical questionnaire was administered to 4,461 young females. Based on scores for EAT and/or EDI-drive for thinness (EDI-DT) subscale, subjects were categorised into either "normal" (NM) or "at risk" (AR) of ED. RESULTS: Mean age of the subjects was 16.7 years (range 12-26 years). The ethnic composition was 78.8 percent Chinese, 11.7 percent Malay, 6.6 percent Indian and 3 percent other ethnic groups. Prevalence of AR was 7.4 percent (95 percent confidence interval [CI] 6.7-8.2 percent). Mean EAT and EDI-DT scores for AR were significantly higher than that of NM (EAT: mean difference is 22.1, 95 percent CI 20.7-23.4, p-value is less than 0.0001; EDI-DT: mean difference is 10.9, 95 percent CI 10.5-11.4, p-value is less than 0.0001). Female Malays constituted a significantly larger proportion of AR (20.6 percent) as compared to NM (10.9 percent). AR females are more likely to use Malay as a spoken language at home (prevalence rate ratio 1.70, p-value is 0.001) and to be better educated with completion of General Certificate of Education (GCE) "O" levels. However, the parents of AR females are likely to be less well educated (below GCE "A" levels). CONCLUSION: The prevalence of females at risk of developing ED is 7.4 percent. Malay ethnic group, using Malay language at home and the educational levels of both the subjects and their parents appear to be associated with an increased risk for development of ED.

MELD may not be the better system for organ allocation in liver transplantation patients in Singapore.

INTRODUCTION: The Model for End-Stage Liver Disease (MELD) score is a good predictor of mortality on the liver transplant waiting list and is the current system of organ allocation in the USA. However, a higher MELD may be associated with poorer outcome post-liver transplantation. The aim of this study was to determine if MELD should be implemented as the system for organ allocation for liver transplantation in Singapore. METHODS: There were 46 adult patients who underwent primary liver transplantation at the National University Hospital, Singapore from January 1996 to December 2002. We applied the MELD score to patients who were transplanted and looked for a correlation with survival post-transplant. Patients were followed-up until the most recent visit or death. Survival analysis was performed using Cox regression and Kaplan-Meier method. RESULTS: The mean age at transplant was 52.7 (SD 2.34) years. The majority of the patients transplanted had Hepatitis B (43 percent). The median MELD score at transplantation was 17 (7-42) and the median Child's score was 11 (6-15). There was a significant correlation between pre-transplant MELD and survival at six months (p-value is 0.037, 95 percent confidence interval [CI] is 1.004-1.13) but not at one year (p-value is 0.065, 95 percent CI is 0.99-1.12). There were no differences in the pre-transplant MELD (odds-ratio [OR] 1, 95 percent CI 0.9-1) as well as survival for patients with and without Hepatitis B (OR 0.72, 95 percent CI 0.22-2.35). CONCLUSION: MELD allows livers to be allocated to the patients with the greatest medical urgency but its influence on post-transplant survival should be further clarified so that post-transplant survival is not compromised.

Compensatory Hypertrophy After Living Donor Nephrectomy.

BACKGROUND: Previous studies have shown that kidney volume enhances the estimation of glomerular filtration rate (eGFR) in kidney donors. This study aimed to describe the phenomenon of compensatory hypertrophy after donor nephrectomy as measured on computerized tomographic (CT) scans. METHODS: An institutional Domain Specific Review Board (DSRB)-approved study involved approaching kidney donors to have a follow up CT scan from 6 months to 1 year after surgery; 29 patients participated; 55% were female. Clinical chart review was performed, and the patient's remaining kidney volume was measured before and after surgery based on CT scans. eGFR was determined with the use of the Modification of Diet in Renal Disease equation. RESULTS: Mean parenchymal volume of the remaining kidney for this population (mean age, 44.3 ± 8.5 y) was 204.7 ± 82.5 cc before surgery and 250.5 ± 113.3 cc after donor nephrectomy. Compensatory hypertrophy occurred in 79.3% of patients (n = 23). Mean increase in remaining kidney volume was 22.4 ± 23.2% after donor nephrectomy in healthy individuals. Over a median follow-up of 52.9 ± 19.8 months, mean eGFR was 68.9 ± 12.4 mL/min/1.73 m(2), with 24.1% of patients (n = 7) in chronic kidney disease grade 3. Absolute and relative change in kidney volume was not associated with sex, race, surgical approach, or background of hypertension (P = NS). There was a trend of decreased hypertrophy with increasing age (P = .5; Spearman correlation, -0.12). CONCLUSIONS: In healthy kidney donors, compensatory hypertrophy of the remaining kidney occurs in 79.3% of the patients, with an average increment of about 22.4%. Older patients may have a blunted compensatory hypertrophy response after surgery.

Does pathological fracture affect the rate of local recurrence in patients with a giant cell tumour of bone?: a meta-analysis.

We investigated whether the presence of a pathological fracture increased the risk of local recurrence in patients with a giant cell tumour (GCT) of bone. We also assessed if curettage is still an appropriate form of treatment in the presence of a pathological fracture. We conducted a comprehensive review and meta-analysis of papers which reported outcomes in patients with a GCT with and without a pathological fracture at presentation. We computed the odds ratio (OR) of local recurrence in those with and without a pathological fracture. We selected 19 eligible papers for final analysis. This included 3215 patients, of whom 580 (18.0%) had a pathological fracture. The pooled OR for local recurrence between patients with and without a pathological fracture was 1.05 (95% confidence interval (CI) 0.66 to 1.67, p = 0.854). Amongst the subgroup of patients who were treated with curettage, the pooled OR for local recurrence was 1.23 (95% CI 0.75 to 2.01, p = 0.417). A post hoc sample size calculation showed adequate power for both comparisons. There is no difference in local recurrence rates between patients who have a GCT of bone with and without a pathological fracture at the time of presentation. The presence of a pathological fracture should not preclude the decision to perform curettage as carefully selected patients who undergo curettage can have similar outcomes in terms of local recurrence to those without such a fracture.

Sample size and power calculations for comparing two independent proportions in a 'negative' trial.

Statistical methods for evaluating the adequacy of sample size and power cater mainly to the testing for treatment difference in a 'positive' trial. In biomedical research, the trial can sometimes be postulated as 'negative' to demonstrate that the different treatment groups are statistically equivalent. Herein we describe a computer program to determine the adequacy of sample size and power for comparing two independent proportions in a 'negative' trial. The program is written in MicroSoft QuickBasic Version 4.5, and its executable file is suitable for use as a stand-alone program on a microcomputer.

Comparing a sample proportion with a specified population proportion based on the mid-P method.

A common practice for determining whether a sample proportion is statistically different from a specified population proportion is using the exact probability procedure. However, it produces overly conservative confidence intervals and p-values. A stand-alone executable program is written for applying alternative approaches based on the mid-P and bootstrap methods.

The MELD score may help to determine optimum time for liver transplantation.

BACKGROUND: The model for end-stage liver disease (MELD) score is a good predictor of mortality on the waiting list and short-term survival post liver transplantation. AIM: Our aim was to determine if there is a pretransplant MELD score beyond which liver transplantation is prohibitive. PATIENTS AND METHODS: Forty-six adult patients underwent primary liver transplantation from January 1996 to December 2002. Patients followed to the most recent visit or death underwent survival analysis using Cox regression and Kaplan Meier methods. RESULTS: There was a significant correlation between the pretransplant MELD score and survival at 6 months posttransplant (P=.037 95% CI: 1.004-1.13). Patients with pretransplant MELD score greater than or equal to 32 showed significantly greater mortality compared with those less than 32 (HR 9.18, 95%CI=1.16-72.44). CONCLUSION: Pretransplant MELD may help to determine the optimum time for liver transplantation.

A review of software for data management, design and analysis of clinical trials.

In clinical trials, subjects are usually entered one at a time, and their responses to treatment monitored sequentially. Regular monitoring of trial progress in the early stages is crucial for accurate reporting of the final results. This paper discusses in detail the principles of quality data management in clinical trials, with specific reference to three clinical data management systems namely, CLINTRIAL, ORACLE CLINICAL and MACRO. All three systems have the essential features for monitoring and processing quality clinical trials data. In terms of functionality, there appears to be no significant advantage of one system over the other. However, to reap the full benefits of sophisticated systems such as these, a good support network and comprehensive training programmes are essential since their day-to-day use demands a high level of technical competence. Basic considerations pertinent to the success of a clinical trial involve not only logistics and data management. Issues relating to the study design are also of primary concern. In this respect, we briefly describe some sample size packages, NQUERY, PEST, POWER, POWER AND PRECISION and SAMPSIZE. Finally, a brief comparison is made with regards to some distinct features of three commonly used statistical packages, namely SPSS, SAS and STATA.

Risk factors for hospital and long-term mortality of critically ill elderly patients admitted to an intensive care unit.

BACKGROUND: Data on long-term outcomes of elderly (≥65 years) patients in ICU are sparse. MATERIALS AND METHODS: Adult patients (n = 1563, 45.4% elderly) admitted over 28 months were analyzed by competing risks regression model to determine independent factors related to in-hospital and long-term mortality. RESULTS: 414 (26.5%) and 337 (21.6%) patients died in-hospital and during the 52 months following discharge, respectively; the elderly group had higher mortality during both periods. After discharge, elderly patients had 2.3 times higher mortality compared to the general population of the same age-group. In-hospital mortality was independently associated with mechanical ventilation (subdistribution hazard ratio (SHR) 2.74), vasopressors (SHR 2.56), neurological disease (SHR 1.77), and Mortality Prediction Model II score (SHR 1.01) regardless of age and with malignancy (SHR, hematological 3.65, nonhematological 3.4) and prior renal replacement therapy (RRT, SHR 2.21) only in the elderly. Long-term mortality was associated with low hemoglobin concentration (SHR 0.94), airway disease (SHR 2.23), and malignancy (SHR hematological 1.11, nonhematological 2.31) regardless of age and with comorbidities especially among the nonelderly. CONCLUSIONS: Following discharge, elderly ICU patients have higher mortality compared to the nonelderly and general population. In the elderly group, prior RRT and malignancy contribute additionally to in-hospital mortality risk. In the long-term, comorbidities (age-related), anemia, airway disease, and malignancy were significantly associated with mortality.

Blood loss in spinal tumour surgery and surgery for metastatic spinal disease: a meta-analysis.

There is currently no consensus about the mean volume of blood lost during spinal tumour surgery and surgery for metastatic spinal disease. We conducted a systematic review of papers published in the English language between 31 January 1992 and 31 January 2012. Only papers that clearly presented blood loss data in spinal surgery for metastatic disease were included. The random effects model was used to obtain the pooled estimate of mean blood loss. We selected 18 papers, including six case series, ten retrospective reviews and two prospective studies. Altogether, there were 760 patients who had undergone spinal tumour surgery and surgery for metastatic spinal disease. The pooled estimate of peri-operative blood loss was 2180 ml (95% confidence interval 1805 to 2554) with catastrophic blood loss as high as 5000 ml, which is rare. Aside from two studies that reported large amounts of mean blood loss (> 5500 ml), the resulting funnel plot suggested an absence of publication bias. This was confirmed by Egger's test, which did not show any small-study effects (p = 0.119). However, there was strong evidence of heterogeneity between studies (I(2) = 90%; p < 0.001). Spinal surgery for metastatic disease is associated with significant blood loss and the possibility of catastrophic blood loss. There is a need to establish standardised methods of calculating and reporting this blood loss. Analysis should include assessment by area of the spine, primary pathology and nature of surgery so that the amount of blood loss can be predicted. Consideration should be given to autotransfusion in these patients.