P. F. (Pat) Anderson

Evidence-Based Dentistry Workshop: Sample Search

Population + Intervention + Control + Outcome = PICO


The following two searches of Medline were developed by a dental expert (original version) and a librarian/information retrieval expert (revised version) for a workshop on evidence-based dental education and practice.

Statement of search question


As a candidate for the position of Associate Dean for Academic Affairs in a dental school, the search committee requested that you prepare a position paper on whether the school should revise its curriculum and adopt problem-based learning as a method for teaching and learning at the school. You were asked to prepare a discussion paper on the effectiveness of PBL. You have decided to conduct a systematic review on this topic.

Defining the question

"What are the effects of PBL compared to conventional teaching in health professional curricula on student learning, success in national boards examinations, and critical thinking skills?"

Identify population, intervention, comparison group, and outcomes (PICO)

Search strategy: Inclusion criteria

  1. Include studies conducted in dental, medical, nursing, pharmacy, veterinary, schools or health professional schools (public health).
  2. Include studies written in English.
  3. Include only studies with comparison group or groups or evaluation studies with before-after comparisons.

How to Read This Chart: Legend

Original Search Revised Search


1 - education(E)/ or E, dental (D)/, or E, D, graduate (G)/ or E, G/ or E, medical (M)/ or E, M, G/ or E, M, undergraduate (U)/ or E, nursing (N)/ or E, N, Associate/ or E, N, baccalaureate/ or E, N, diploma programs/ or E, N, G/ or E, pharmacy (P)/ or E, P, G/ or E, professional/ or veterinary, or N education research/ 78588
2 - students (S), D/or S, health occupations/ or S, M, or S, N, or S, P / 19989
3 - exp curriculum/ 29908
4 - 1 or 2 or 3 / 102710
1 - exp Education, professional/ 115414
2 - Education, continuing/ 2977
3 - 1 not 2 / 112437
4 - exp Faculty/ 10549
5 - exp Schools, health occupations/ 19227
6 - Area health education centers/ 128
7 - 5 not 6 / 19099
8 - exp Students, health occupations/ 20151
9 - Students, premedical/ 108
10 - 8 not 9 / 20043
11 - 3 or 4 or 7 or 10 / 136624
NOTE: line 5 searches for all papers involving the types of health professional education specified as being of interest. An alternative approach could have been to "explode" the term professional education and then to list all of the ineligible types of education that appear under this general descriptor, with the word "not", but since there are more ineligible categories, list the eligibles is more efficient. Line 6 searches for papers involving health education students, and line 7 simply identifies all papers with the index term "curriculum or one of its sub-terms. Lines 6 and 7 are added as extra search terms to identify papers that were not indexed specifically as involving health professional education COMMENT: This is the short way. To be authoritative, one should also search each of the key subject headings as a keyword in the title and abstract ("map term"). I will give the whole tree structure for these groups at the end of the message, so that it is clear just who all is included in this group of 4 sets.
NOTE: In your sample search, someone included "exp curriculum." This puzzled me, since in the tree structure, PBL is a subdivision of "curriculum." It seemed odd to me to (in essence) include the same term in two presumably separate concept groups. The actual search results should be a little odd, since you weren't really "crossing" the two concepts.


5 - exp problem-based learning/ 584
6 - "problem-based learning" / 794
7 - "PBL" / 4381
8 - 5 or 6 or 7 / 5025
12 - Problem-based learning/ 598
14 - "PBL".mp. / 4403
15 - 12 or 13 or 14 / 5060
NOTE: the first line searches for all papers indexed under the term "problem-based learning.," The nest line looks for papers with the term /"problem-based learning" in the title or abstract, and the third term looks for the term "PBL" in the title or abstract. COMMENT: Searching the concept as a map term <.mp.> searches the term in the title and abstract. I assume that is what your folks were doing with the terms which were in quotes. Also, the abbreviation search pulled in a great many articles on immune diseases. Like your folks, I ignored this with the assumption that crossing it with the other subject terms would focus it. If it didn't, then we would want to build a search set for the complicating concept and "not" it out.


Cross Intervention with Target Population
9 - 4 and 8 / 774
10 - limit 9 to English language / 673
NOTE: Line 9 limits the entries discussing PBL to those the deal with professional education, and line 10 further limits this group to those published in English.
Identify Comparisons
11 - "comparative study" / 866559
12 - exp evaluation studies/ 309113
13 - exp epidemiologic study characteristics/ or epidemiologic research design/ 776118
14 - 10 and (11 or 12 or 13) / 189
15 - limit 10 to (clinical trial, or controlled clinical trial, or meta analysis or randomized controlled trial) / 10
16 - 14 or 15 / 190
16 - Evaluation studies/ 97755
17 - exp Clinical trials/ 109614
18 - exp Program evaluation/ 8813
19 - Reproducibility of results/ 47615
20 - exp Research design/ 119384
21 - Patient selection/ 6405
22 - 20 not 21 / 112979
23 - exp epidemiologic study characteristics/ 637030
24 - Twin studies/ 294
25 - 23 not 24 / 636736
26 - 16 or 17 or 18 or 19 or 22 or 25 / 825797
27 - Curriculum/ 29056
28 - Competency-based education/ 408
NOTE: lines 11-13 search for three types of studies that typically involve comparisons. This search strategy assumes that any comparison involving PBL will mean a comparison of PBL with a traditional educational approach. Line 14 crosses the /group of papers identified in line 10 (i.e., health professional education papers about PBL in English) with the comparative study papers, yielding 189 papers. Line 15 is a similar strategy describing two additional descriptors of types of comparative studies, but the descriptors are applied as a limit to line 10 (for efficiency sake) Line 16 then combines the two groups of comparative studies COMMENT: I did not explode "Evaluation studies" because there were so many irrelevant terms underneath it in the tree structure. Similarly, with the exploded "Research design" set, I "not"-ed out the irrelevant term "Patient selection." Same for "Epidemiologic study characteristics" and "Twin studies."


No step 4 comments from this group's search. COMMENT: I did not see a desired outcome search group in your original set or question statement (although it was implied), so I assumed you were interested in ALL outcomes of the intervention. This was actually kind of nice, because it saved us from having to come up with search terms for attitudes etc.
NOTE: It would be possible to do this, but with such a small set from this search, you might very well end up with nothing. In particular, those more nebulous terms and concepts are at a higher risk of not being captured by the catalogers (an issue we have previously discussed).


This section missing from original search as delivered. 29 - 11 or 27 or 28 / 144085
30 - 11 and 15 and 26 / 160
31 - 29 and 15 and 26 / 169
32 - 31 not 30 / 9
33 - from 32 keep 1-9 / 9
34 - limit 31 to english language / 152
35 - from 34 keep 1-152 / 152
36 - 29 and 15 / 741
37 - limit 36 to (clinical trial or clinical trial, phase i or clinical trial, phase ii or clinical trial, phase iii or clinical trial, phase iv) / 9
38 - limit 36 to controlled clinical trial / 1
39 - limit 36 to meta analysis / 2
40 - limit 36 to randomized controlled trial / 6
41 - 37 or 38 or 39 or 40 / 11
42 - 34 or 41 / 158
43 - from 42 keep 1-158 / 158
ditto. GENERAL COMMENTS I noticed that your folks applied the limits in the middle of the search. As a rule of thumb, I usually prefer to wait until the search is fully constructed to apply any limits. I could have done the two step cross which you folks used:

[(P+I)+C] or [(P+I)+"limits for C"] = Outcomes

I didn't for the same reason I didn't search all the population groups -- time and level of need. You may want to do that. I feel this search probably works adequately. The population group came out a good bit larger than yours, the intervention set was about the same, and the comparision set was probably smaller, althought your sample search doesn't combine those sets into a single statement so we can't be sure. Note that I didn't use the map search for the term "comparison study", which I probably should have done. This is an area where I would like to go back and tweak this section. My final set came out close to yours, although again comparing is difficult since the sample search you gave me had the boggle with the exploded "curriculum." That may very well have skewed the final search a bit. Anyway, here it is, for what its worth. Hope these comments and insights are helpful!

Tree Structure for Explodes

The term "Program Evaluation" deserves a little special attention. In this search, it was included conceptually as a subdivision of "Evaluation Studies", however it is linked to two other terms which are not clearly related to the topic under examination here: "Quality of Health Care" and "Health Care Evaluation Mechanisms." Since the intervention here is educational, technically, both those areas are irrelevant to the matter at hand. So, it is possible to explode the term "Program Evaluation," but to do so most likely offers no significatn advantage. And here is the tree for the questionable term I mentioned. You could use the curriculum term, but would have to be very careful with it, since the explode includes the intervention term.

Curriculum (29056)

Date last modified: March 26, 2000.
Return to Pat's Personal Homepage.
Return to Pat's Pro Pages.
URL of current page: http://www-personal.umich.edu/~pfa/pro/consult/ebdtest.html