Table 4 Summary of the results from ‘ sampling ’ studies included in the review (n = 33 studies)

Difficult to locate or reach and access groups (e.g., homeless people living on the streets)

• Snowball/social network or respondent-driven recruitment [33, 57, 60, 62, 79, 84, 102–104, 116],[118, 126, 129]

Frequent change of address or self-identifying, (e.g. GLBT) results in no sampling frame.

• Time-space sampling [38, 104, 113]

• Targeted sampling [34, 79]

• Capture-recapture [95]

• Adaptive sampling [53]

• Partnerships with community groups [32, 40, 44, 47, 54, 60–62, 67, 74],[103, 116, 118, 122, 128]

Low prevalence in population (e.g., Aboriginal people).

• Combination of various data sources as a novel methodology to avoid sampling [62, 97] or supplementing with additional data (e.g. from qualitative research) [44, 47, 54, 60, 67, 103, 113, 122],[128]

• Statistical analysis techniques to population survey data for low-frequency samples [91]

• Internet samples [51]