This review found that early initiation of dialysis was associated with an increased risk of death although age, diabetes and comorbidities may partly explain this effect. The review was very large so the conclusions may be reliable but potential for bias in the included observational studies mean that the results should be interpreted with caution.

To investigate the association between risk of death and early initiation versus late initiation of dialysis.

PubMed (1966 to 2011), EMBASE (1974 to 2011), ClinicalTrials.gov and Cochrane Controlled Clinical Trials Register were searched for studies published in English. Search terms were reported. References in the identified studies and relevant review articles were searched.

All observational studies that compared early and late initiation of dialysis with mortality as an outcome were eligible.

Average ages in the studies ranged from 47 to 71 years and 44% (range 32% to 67%) of participants were women. On average between 12% and 82% of participants had diabetes. Average estimated glomerular filtration rates were more than 10.97mL/min/1.73m² in early dialysis patients and less than 5.98mL/min/1.73m² in late dialysis patients.

It appeared that two reviewers performed the study selection and disagreements were resolved by discussion.

The authors did not state whether they assessed study quality.

Odds ratios (OR) for mortality and their associated 95% confidence intervals were extracted. Average ages, numbers with diabetes, levels of estimated glomerular filtration rates and the Charlson Comorbidity Index were extracted for use in subgroup and adjusted analyses.

Two reviewers extracted data. Disagreements were resolved by discussion.

Pooled odds ratios were calculated in a random-effects meta-analysis. The analysis was adjusted for a range of potential confounding factors. Heterogeneity was assessed using the Ι² statistic. Publication bias was assessed using funnel plots and Egger’s test. Subgroup analyses were used to test the effect of age, diabetes and Charlson Comorbidity Index on mortality. Sensitivity analyses were performed to assess the influence of each study.

Fifteen studies (>1,200,000 patients, range 210 to 896,546) were included: one randomised controlled trial, four prospective cohorts and 10 retrospective studies. Follow-up ranged from one to over 10 years.

In an adjusted analysis early initiation of dialysis was associated with an increased risk of death (OR 1.33, 95% CI 1.18 to 1.49; Ι²=96%; 11 studies). For each 5mL/min/1.73m² increase in estimated glomerular filtration rate the odds ratio of death increased by 1.13 (95% CI 1.07 to 1.19; Ι²=83%; four studies).

Older age at initiation of dialysis was associated with an increased risk of death (OR 1.18, 95% CI 1.05 to 1.33; Ι²=100%; seven studies), as was diabetes (OR 1.61, 95% CI 1.38 to 1.87; Ι²=75%; six studies) and medium or high Charlson Comorbidity Index.

There was no evidence of significant publication bias.

Early initiation of dialysis was associated with an increased risk of death. Older age, greater likelihood of diabetes and comorbidities may partly explain this effect.

This review considered an appropriate review question with broad inclusion criteria. A suitable search was conducted. It was limited to English publications so some studies may have been missed but there was no evidence of publication bias. Action was taken to minimise reviewer error and bias. Study quality was not reported but the authors noted that most studies were observational and many had a retrospective design. Such studies have potential for bias due to confounding factors and differences in how the intervention is allocated to patients. A meta-analysis was performed to pool the studies and sought to investigate these potential biases. There was very substantial heterogeneity across the studies, which may limit the generalisability of the overall pooled results.

The overall size of the review was very large so the author’s conclusion may be appropriate but potential for bias in these observational studies mean that the results should be treated with caution.

Practice: The authors recommended that initiation of dialysis should be based on the patient’s condition and not on estimated glomerular filtration rate alone.

Research: The authors suggested that randomised controlled trials were needed to confirm these results using reliable measurements of renal function.

Not stated.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.