The review concluded that sublobectomy was associated with poorer survival than lobectomy in stage I lung cancer patients, but similar survival in the comparisons of segmentectomy with lobectomy and sublobectomy with lobectomy in patients with tumours 2cm or smaller. Due to potential bias in the review process, limited evidence base and substantial heterogeneity, the reviewers' conclusions should be treated cautiously.

To compare the effectiveness of sublobectomy with lobectomy in patients with stage I non-small-cell-lung cancer.

PubMed was searched for relevant peer reviewed published studies in English up to August 2010; limited search terms were reported. The reference lists of included studies were also searched.

Studies of clinical stage I non-small-cell-lung cancer patients, where the hazard ratio (HR) or relative risk (RR) of either overall survival or cancer-specific survival was calculable, were eligible for inclusion in the review. Letters to the editor or reviews were excluded.

In the included studies, participants had either stage I or stage Ia lung cancer, some participants were diagnosed with tumours 2cm or smaller. Sublobectomy included either wedge resection or anatomic segmentectomy and lobectomy included bilobectomy and pneumonectomy. Studies were published between 1990 and 2009.

The authors did not state how many reviewers selected studies for the review.

The authors did not report whether the studies were assessed for quality.

Data on the log hazard ratio of overall or cancer-specific survival were extracted or calculated from reported data in the studies, with the standard error. Ninety-five percent confidence intervals (CIs) were also calculated.

The authors did not state how many reviewers extracted data for the review.

The results of the outcomes were synthesized in meta-analyses. Summary effect log hazard ratios, with 95% confidence intervals, were calculated using a random-effects model where significant heterogeneity was identified; otherwise a fixed-effect model was used. Subgroup analyses compared the effects of segmentectomy with lobectomy on either overall or cancer-specific survival, and compared sublobectomy with lobectomy on overall survival in patients with tumours 2cm or smaller. Heterogeneity was assessed using the Χ² test (with p<0.05 being considered significant). Publication bias was assessed using Egger's methods (a p value below 0.1 represented significance with 90% confidence intervals).

Twenty-four studies (11,360 participants, ranging from 73 to 1,272) were included in the review. All studies were retrospective, except for one randomised controlled trial (RCT).

Sublobectomy versus lobectomy overall: Lobectomy was significantly associated with improved overall survival (HR 1.26, 95% CI 1.07 to 1.47; significant heterogeneity; 22 studies), improved cancer-specific survival (HR 2.07, 95% CI 1.14 to 3.77; significant heterogeneity; six studies) and improved overall or cancer-specific survival (HR 1.40, 95% CI 1.15 to 1.69; significant heterogeneity).

Sublobectomy versus lobectomy in participants with tumours 2cm or smaller: There was no evidence of a significant difference in overall survival when lobectomy was compared with sublobectomy in participants with tumours 2cm or smaller (HR 0.81, 95% CI 0.39 to 1.71; no significant heterogeneity (p=0.06); five studies).

Segmentectomy versus lobectomy: There was no evidence of a significant difference in either overall survival (HR 1.09, 95% CI 0.85 to 1.40; no significant heterogeneity; eight studies), cancer-specific survival (HR 0.99, 95% CI 0.72 to 1.38; no significant heterogeneity; three studies) or combined overall or cancer-specific survival (HR 1.05, 95% CI 0.87 to 1.28; no significant heterogeneity).

There was no evidence of publication bias in any of the analyses.

For stage I lung cancer patients, sublobectomy was associated with poorer survival than lobectomy, but survival rates appeared similar when segmentectomy was compared with lobectomy and when sublobectomy was compared with lobectomy in patients with tumours 2cm or smaller.

The review addressed a clear research question, supported by broad inclusion criteria. A limited search for studies published in English was conducted of one electronic database with reference lists of retrieved studies, so language and publication bias could not be ruled out and it was possible studies may have been missed. However, formal assessment of publication bias using Egger's methods did not find any evidence that publication bias was likely. The authors did not report how many reviewers selected studies or extracted data, so reviewer error and bias could not be excluded.

The studies were not assessed for quality and were mostly retrospective, so they were not likely to produce reliable results. The authors acknowledged the considerable heterogeneity found in the overall analyses and suggested that variation in tumour size, distribution of histological types, male/female ratio, reasons for sublobectomy and details of the operation may have contributed to the identified heterogeneity; no exploratory sensitivity analyses were made. They also suggested that other details not collected in the studies such as treatment information may have impacted on the results. Due to potential bias in the review process, the possibility of missed studies, the limited reliability of the primary study results and substantial heterogeneity, the reviewers' conclusions should be treated with caution.

Practice: The authors did not state any implications for practice.

Research: The authors did not state any implications for research but they noted that two relevant RCTs were currently ongoing.

Foundation of Ministry of Education of China; Natural Science Foundation of Shandong Province.