Examples of Primary Source and Secondary Source
Evidence-Based and Peer-Reviewed Scientific Literature
  
 

Criteria for Scientifically Establishing Cause-and-Effect:

Scientific research mainly seeks to answer questions of cause and effect. In health care, much of the clinical and epidemiologic research is concerned with what causes a health condition such as disease, pain, or dysfunction, and therefore what might be done to prevent the health condition from occurring in the first place.  Or in cases where a health condition is already present, what interventions can help to return a patient to a more healthy state (i.e. to assist in causing healing to occur), or at the very least slow down or stop a condition from progressing to a worse state. Scientific inquiry is based on "common sense" principles for determining cause-and-effect.  The following Table 1 outlines some of the "Criteria for Causality" that must be met to build a convincing "body of scientific evidence" of a cause-and-effect relationship.  Does the hypothesized cause of the condition (e.g. exposure of the patient to a "risk factor") occur before the observed effect (e.g. the onset of illness).  Are there other possible factors that have also occurred, and that could have contributed to causing the illness?  Does the established scientific knowledge-base or evidence from other previous research convincingly support the hypothesized cause-and-effect relationship?
 
Table 1: Examples of Criteria for Establishing Causality (Cause-and-Effect) in Clinical Research
bulletTime-Order
> Does the hypothesized cause precede the effect?
> Can the time-order of cause-and-effect be determined from the Observational or Quasi-Experimental Research Study, or has time-order been "controlled" in the Experimental Study?
 
bulletSpecificity of Cause-and-Effect Association
> Are there other possible causes for the observed effect?
> Have other possible causative factors been considered or "controlled" in the Research Study?
 
bulletConsistency and Scientific Plausibility
> Consistent results across numerous and varied Clinical or Epidemiologic Research Studies will provide stronger cumulative evidence in support of a postulated cause-and-effect relationship.
> Do the findings from the Clinical and Epidemiological Research (Studies of cause-and-effect at the "person-level") "make sense" relative to what we know from Basic Science Research (e.g. Studies of cause-and-effect at the cellular level)?

 

Primary Source Evidence:

The second table (Table 2, below) titled "Continuum of Study Designs and their Causal Inference" demonstrates the hierarchy of the different types of research studies. This hierarchy is based on the ability of the research study design to satisfy the "Criteria for Causality" (some of which were identified in Table 1 above). The research examples at the bottom of the table (e.g. case studies) are good for beginning the process of developing cause-and-effect research hypotheses, but such studies can only provide very weak evidence since they don't meet many of the Criteria for establishing Causality. Moving from the bottom to the top of Table 2, the rigor of the research studies is increased, more of the causal criteria are met, and a stronger case may be made for drawing conclusions (inferences) about cause-and-effect. At the top of Table 2 is the experimental randomized controlled trial (RCT), which is the strongest study design for meeting the causal criteria, and therefore is the "gold standard" of scientific evidence.
 
Table 2: Continuum of Examples of Study Designs and their Causal Inference
 

Study Designation

(1)

  

(2)

  

(3)

 

 Inference
 

Clinical randomized trials

    

 

   x

 

 Firm

More
convincing
evidence
of cause-
and-effect
 

Community randomized trials

    

 

   x   Firm
 

Cohort

 

   x    

 

 Moderately Firm
 

Case-control

 

   x    

 

 More Suggestive
 

Cross-sectional

 

   x    

 

 Suggestive
 

Time series

x

   x       Suggestive
 

Case series

x

           Speculative
 

Case study

x

        

 

 Speculative
(1) Clinical Observations, (2) Observational and "Quasi-Experimental" Research,
(3) Experimental Research (e.g. Randomized Controlled Trials - RCTs)		 
 
Following are brief descriptions and examples of some of the different types of "Primary Source" research studies listed in Table 2.

Case Report, Case Study, Case Series: In these "clinical observations", a clinician reports "out of the ordinary" findings from their clinical experience with a patient (a "case") or a number of similar patients (a "series"). These reports of patients' clinical histories are typically used to notify the clinical scientific community of a new and noteworthy aspect of a condition, clinical intervention, or adverse reaction. 

Cross-Sectional studies are frequently surveys conducted at one point in time, for instance to determine the prevalence of a certain health condition among a defined population.

Case-Control studies identify a group of "cases" (those with a condition) and a group of "controls" (those without the condition).  The researcher looks back in time and compares the prior history of the "cases" to the "controls", to identify possible causative risk factors that preceded the development of the condition. When enough scientific evidence has been accumulated through case-control studies to support a possible cause-and-effect finding, then it justifies spending the time and resources to fund and conduct more rigorous, more difficult, and more expensive Cohort studies. A Cohort study starts by first identifying individuals with a suspected risk factor and comparing them to individuals without the risk factor. This cohort is followed forward over a period of time to determine whether the presence of the suspected causative risk factor is associated with the development of the condition (the effect).

Randomized Controlled Trials (RCTs) are experimental studies in which the researcher can control various important factors, and thereby systematically and convincingly "rule out" other possible explanations for an observed apparent cause-and-effect relationship. This is the strongest and most rigorous study design for meeting the "Criteria for Causality", which is why it is considered the "Gold Standard" of clinical research.

 

 

Secondary Source Summaries of Evidence:

The "primary evidence" from the research studies described above is periodically "Reviewed" by gathering all of the evidence available on a given topic, and comparing and contrasting the various research studies and the results from those studies. Review articles follow acceptable scientific methods which again make use of the common-sense "Criteria for Causality" to compare and contrast the strength of the different studies for supporting (or refuting) a hypothesized cause-and-effect relationship. Generally, Review articles fall into one of 3 broad categories in order of increasing scientific rigor: the narrative review, the systematic review, and meta-analyses, which are briefly described below.

Narrative Reviews are essentially "essays" wherein the researcher/author focuses on a particular clinical topic, and presents in their opinion what they feel are some of the highlights worth noting from the scientific literature they review. Narrative Reviews are the weakest of the "Secondary Source" research literature, since the methods and the reporting are somewhat subjective, and therefore may be more prone to bias. Articles that report original empirical research studies (described as "Primary Source" evidence above) typically also contain good "narrative overviews" of the "existing evidence" as an introduction and background to the article, to explain and justify why the original research study was conducted.

Systematic Reviews are conducted according to scientific methodologies that are explicit, explicitly stated, and reproducible. For instance, the methods section of a systematic review will describe the literature search in enough detail (e.g. what databases were searched, for what years, search terms employed), that the search strategy could be replicated by another researcher and produce essentially the same yield of literature. The "primary evidence" articles are then systematically and explicitly compared and contrasted along a number of important dimensions of research designs that reflect the "Criteria for Causality", which makes for a stronger and more useful "review of the resulting evidence" than do simple narrative reviews. This "attention to detail" in conducting and reporting the systematic review, helps the readers of the review article (i.e. the clinical and scientific communities) to determine how strong the accumulated evidence appears to be, without relying solely on the opinion of the author.

Meta-Analyses take the systematic review to the highest and most rigorous level of integrating "primary evidence" research, by statistically synthesizing the quantitative results across similar studies. Essentially, meta-analyses pool the data and results from a number of original research studies, with the intent of producing a more precise summary of the existing evidence for (or against) a cause-and-effect relationship.  For example, a meta-analysis may collect results from a number of epidemiologic studies, and statistically combine the data to produce a more precise overall "risk estimate" of the relationship between the causative risk factor and the effect (the disease).

[ Top of Page ] [ Up ] [ Home ]