Skip to main content
. 2016 May 24;13(5):e1002028. doi: 10.1371/journal.pmed.1002028

Table 6. Comparison of the reporting characteristics of systematic reviews in 2004 and 2014.

Characteristic Percent or Median in 2004 and 2014, and Direction of Change between Time Points
All Therapeutic (Cochrane) Therapeutic (Non-Cochrane) Epidemiology Diagnosis/Prognosis Other
“Systematic review” or “meta-analysis” used in the title or abstract 50% ↑ 85% 39% ↓ 33% 68% ↑ 95% 50% ↑ 93% 65% ↑ 97% 35% ↑ 86%
SR protocol mentioned 46% ↓ 26% 98% = 98% 11% ↑ 22% 8% ↓ 4% 4% ↑ 6% 22% ↓ 7%
Eligible publication status
Both published and unpublished studies 41% ↓ 39% 69% ↑ 91% 22% ↑ 41% 18% = 18% 22% ↓ 18% 30% ↓ 24%
Only published studies 23% ↑ 27% 6% ↓ 4% 33% ↓ 28% 47% ↓ 34% 26% ↑ 27% 28% ↑ 38%
Not reported 36% ↓ 34% 26% ↓ 4% 45% ↓ 30% 34% ↑ 49% 52% ↑ 55% 41% ↓ 38%
Eligible languages
All languages considered 37% ↑ 43% 62% ↑ 82% 26% ↑ 40% 13% ↑ 39% 17% ↑ 30% 15% ↑ 17%
English only 16% ↑ 31% 1% ↑ 2% 31% ↑ 37% 26% ↑ 30% 30% ↑ 36% 26% ↑ 45%
Mixed (English and a specific language other than English) 2% ↑ 10% 0% ↑ 2% 3% ↑ 8% 5% ↑ 14% 9% ↑ 15% 4% ↑ 21%
Not reported 45% ↓ 16% 37% ↓ 13% 40% ↓ 15% 55% ↓ 18% 44% ↓ 18% 54% ↓ 17%
Eligibility criteria based on study designs reported 72% ↑ 79% 100% = 100% 63% ↑ 87% 39% ↑ 76% 27% ↑ 52% 55% ↓ 52%
Number of databases searched 3 ↑ 4 4 ↑ 5 2 ↑ 3 2 ↑ 3 2 ↑ 3 3 ↑ 5
Years of coverage reported
Both start and end dates are reported for all databases 69% ↓ 65% 83% ↑ 91% 58% ↑ 66% 55% ↓ 53% 70% ↓ 58% 63% ↑ 66%
Partially—start and end dates are reported for only one of many databases, or only the end date is reported for all databases 16% ↑ 29% 12% ↓ 9% 21% ↑ 29% 13% ↑ 41% 9% ↑ 39% 20% ↑ 21%
Search terms reported
Full Boolean search logic reported for one or more databases 42% ↑ 45% 78% ↑ 98% 18% ↑ 34% 11% ↑ 35% 17% ↑ 39% 28% ↑ 34%
Main index terms (e.g., MeSH) reported 17% ↓ 12% 11% ↓ 0% 22% ↓ 14% 18% ↓ 14% 17% ↓ 15% 22% ↓ 10%
Free text words reported 26% ↑ 46% 9% ↓ 0% 35% ↑ 54% 40% ↑ 55% 48% ↑ 54% 37% ↑ 48%
No search terms reported 12% ↓ 5% 1% ↓ 0% 22% ↓ 1% 26% ↓ 3% 13% ↓ 3% 11% ↓ 7%
Study risk of bias/quality formally assessed 67% ↑ 70% 100% = 100% 49% ↑ 74% 30% ↑ 59% 13% ↑ 67% 42% ↓ 38%
Review flow reported
Completely in PRISMA-like flow diagram 7% ↑ 69% 4% ↑ 51% 11% ↑ 79% 8% ↑ 62% 4% ↑ 79% 7% ↑ 59%
Completely in text/table only 35% ↓ 7% 38% ↓ 11% 35% ↓ 4% 24% ↓ 8% 39% ↓ 6% 44% ↓ 7%
Partially reported 33% ↓ 13% 48% ↓ 16% 19% ↓ 12% 24% ↓ 15% 22% ↓ 6% 22% ↓ 14%
Not reported 31% ↓ 12% 14% ↑ 22% 43% ↓ 5% 53% ↓ 15% 39% ↓ 9% 35% ↓ 21%
Reasons for exclusion of full text articles reported
Reasons for all excluded articles reported in text/table or PRISMA-like flow diagram 48% ↑ 70% 60% ↑ 87% 45% ↑ 69% 23% ↑ 54% 43% ↑ 76% 46% ↓ 45%
Partially—reasons for only some excluded articles reported 40% ↓ 9% 39% ↓ 9% 38% ↓ 8% 50% ↓ 18% 30% ↓ 3% 35% ↓ 3%
Not reported for any articles 17% ↑ 20% 2% ↓ 0% 25% ↓ 19% 32% ↓ 26% 30% ↓ 12% 26% ↑ 52%
Primary outcome specified 51% ↓ 47% 77% ↑ 96% 37% ↑ 48% 25% ↑ 40% 15% ↑ 18% 24% ↓ 3%
Statistical heterogeneity investigated
Using statistical methods or qualitatively assessed (e.g., via narrative discussion) 68% ↑ 69% 93% ↓ 73% 54% ↑ 72% 47% ↑ 70% 45% ↑ 85% 46% ↓ 28%
Using statistical methods when meta-analysis performed 91% ↑ 93% 100% = 100% 83% ↑ 91% 79% ↑ 92% 73% ↑ 96% 92% ↓ 60%
Risk of publication bias assessed (or intent to assess) 23% ↑ 43% 32% ↑ 78% 18% ↑ 39% 18% ↑ 45% 9% ↑ 45% 12% ↓ 3%
Possibility of publication bias discussed/considered in results, discussion, or conclusion 31% ↑ 47% 39% ↑ 64% 31% ↑ 46% 29% ↑ 50% 18% ↑ 55% 14% ↓ 7%
Source of funding of the SR
Non-profit 48% ↓ 47% 65% ↑ 84% 36% ↑ 40% 26% ↑ 41% 26% ↑ 27% 41% ↑ 59%
For-profit 2% ↑ 3% 1% ↓ 0% 5% ↓ 3% 3% ↓ 3% 0% = 0% 2% ↑ 10%
Mixed 6% ↓ 1% 7% ↓ 0% 6% ↓ 1% 5% ↓ 1% 9% ↓ 0% 4% ↓ 0%
Authors specified there was no funding 1% ↑ 13% 0% ↑ 11% 2% ↑ 12% 0% ↑ 15% 0% ↑ 21% 2% ↑ 7%
Not reported 41% ↓ 36% 19% ↓ 4% 50% ↓ 45% 66% ↓ 41% 65% ↓ 51% 50% ↓ 24%