Opinion: Appreciating a Clinical Approach to the Evaluation of Nonserious, Laboratory Adverse Events

Clinical Researcher—February 2019 (Volume 33, Issue 2)


Robert Jeanfreau, MD, CPI




Recognizing the ongoing necessity for mitigating bias and improving the quality of reporting randomized controlled trials (RCTs), the SORT Group, comprised of medical journal editors, researchers, and epidemiologists, in 1994 published A proposal for structured reporting of randomized controlled trials. The Standards of Reporting Trials Group. This proposal, known as the SORT statement, consisted of a 32-item checklist and flow diagram to standardize the reporting of RCTs.

Three years later, the SORT Group, in collaboration with the Asilomar Working Group on Recommendations for Reporting of Clinical Trials in the Biomedical Literature, published the Consolidated Standards of Reporting Trials (CONSORT) Statement.{1} The reporting of adverse events (AEs) had also come under closer scrutiny; a revised CONSORT Statement was published in 2001 with the addition of an item about reporting AEs.

When it became apparent that this single addition did not adequately address the importance of AE reporting, the CONSORT Group met again to remedy this shortcoming. The resulting second document on Better Reporting of Harms in Randomized Trials: An Extension of the CONSORT Statement was published in 2004.{2} The most recent revision of the CONSORT Statement was published in 2010. The widely respected CONSORT Statement is currently endorsed by 585 journals, including more than 50% of the core medical journals listed in the Abridged Index Medicus on PubMed.{3}

A like-minded organization, known as Medical Publishing Insights & Practices, is comprised of pharmaceutical companies and the International Society for Medical Publication Professionals (MPIP). It has recommended highlighting AEs of most relevance to practitioners and their patients. To this end, MPIP proposes that “authors develop a ‘clinical relevance’ filter.” The authors further state, “The intent of the ‘clinical relevance’ recommendation is to broaden [AE] reporting beyond what is mandated by regulators and to leverage the clinical experience and expertise of physician investigators to judge which [AEs] should be highlighted.”{4}

Applying Clinical Expertise to AE Evaluation

The clinical expertise of physician investigators has been grossly underutilized in the evaluation of AEs in RCTs. This is, perhaps, most clearly demonstrated in the evaluation of the clinical significance of laboratory data and is, in part, due to a poor understanding of the evaluative process.

The generally accepted view of clinical significance is described as follows:

“An abnormal lab value should be deemed clinically significant if either of the following conditions are met:

  • The abnormality suggests a disease and/or organ toxicity that is new or has worsened from baseline.
  • The abnormality is of a degree that requires additional active management, e.g., change of dose, discontinuation of the drug, close observation, more frequent follow-up assessments, or further diagnostic investigation….

Therefore, a clinically significant lab value is one that indicates a new disease process, an exacerbation or worsening of an existing condition, or requires further action(s) to be taken.”{5}

This viewpoint arises from an interpretation of the U.S. Food and Drug Administration’s (FDA) definition of AE:

“[AE] means any untoward medical occurrence associated with the use of a drug in humans, whether or not considered drug related.”{6}

The term “untoward” is generally taken in this context to mean abnormal or outside the defined normal range. In fact, some protocols formally define clinically significant lab abnormalities based upon the amount of deviation from the normal range.

In clinical practice, however, any finding (laboratory result or physical finding) that differs from the “expected” is considered clinically significant. Unexpected findings are important because they suggest the existence of an underlying disorder. It is reasonable then to consider “unexpected” as being under the umbrella of “untoward.”

Clinically significant lab results may or may not be abnormal. Furthermore, even grossly abnormal results may not be clinically significant. The evaluation of laboratory data involves much more than deciding if an abnormal lab is far enough out of range to be of concern. A lab result is never considered in isolation, but in the context of the patient’s physical examination and available history (ideally including past and present medical problems, social history, family history, medications, and previous laboratory studies).

Putting Theory into Practice: Three Scenarios

Let’s take as an example a two-year study looking at a new diabetic medication for adults. Safety labs, consisting of CBCs, are obtained every four months. The reference lab gives the normal range of the hematocrit as 38.5% to 50%. The MCV normal range is set from 80 to 100. The first lab work shows a hematocrit of 50% with an MCV of 98. Since neither is outside the normal range, neither is reported as abnormal. At four months, the hematocrit is 48% and the MCV 95. These are again normal values. At eight months, the hematocrit is 45% and the MCV is 90. At one year, the hematocrit has fallen to 40% and the MCV is 85.

None of these labs would have been flagged as abnormal; therefore, they would not have been identified as AEs and may not have been evaluated for clinical significance. From a clinical standpoint, however, the lab values show a clear trend of deviating from the “expected” previous levels.

The study’s principal investigator (PI) is not considering one isolated set of lab values, but the trend of the results which clearly indicates a falling hematocrit and MCV. Beyond this, the PI would have been reflecting on the subject’s history—how the patient is 65 with no family history of colon cancer, and how his latest colonoscopy was normal.

The subject was taking no known ulcerogenic medications. The subject was experiencing mild episodes of intermittent nausea, and a review of the investigator’s brochure for the drug under study shows that its most commonly related AE is nausea, occurring in 15% of subjects. A simple differential diagnosis is shown below:

  • Lab error
  • Gastrointestinal (GI) blood loss
  • Anemia of chronic disease
  • Investigational product (IP) (this is always listed in the initial differential diagnosis in a clinical trial)

When initially constructed, the differential diagnoses are not ranked. The second step in the evaluative process, which also follows as an important component of determining causality, is ranking the differential diagnoses by likelihood. Based upon clinical expertise and a review of the available information, the clinician would rank the above diagnoses as follows:

1) GI blood loss, possibly due to the IP causing gastric ulceration, suggested by nausea in the investigator’s brochure.

2) Anemia of chronic disease, possibly due to worsening renal disease from diabetes.

3) Lab error, which seems unlikely considering that a trend was seen.

4) GI blood loss from other etiologies.

At this point, the PI stops the IP for three days and the nausea resolves. The subject is instructed to discontinue the IP until he can be evaluated by a gastroenterologist who, in fact, notes gastric ulcerations at endoscopy. By considering an “unexpected” lab finding as an AE, the PI has identified a potential adverse reaction of the IP.

Let’s consider another example. Another subject in the same study has the very same sets of lab values. This subject is 42 and female. The PI requests that the subject return to the research site after he reviews the third set of labs. The subject informs the PI that she has recently been to the gynecologist for a routine visit. At that visit, the gynecologist expressed concern regarding the hematocrit that the PI had faxed to him. She had explained to the gynecologist that her menses had always been heavy, but that she had stopped the iron supplement due to nausea.

A simple, initial differential diagnosis would have been:

  • Lab error
  • IP
  • Menometrorrhagia
  • GI blood loss
  • Anemia of chronic disease

The differential diagnoses are quickly ranked as follows:

1) Menometrorrhagia

2) GI blood loss

3) Anemia of chronic disease

4) IP

5) Lab error

In this situation, it is possible that the PI may have felt that the diagnosis was quite obvious, and may not have even felt compelled to report the lab as a clinically significant AE.

Even lab results that show a trend toward “improvement” may be considered clinically significant. Let’s take an example of a middle-aged male in the same study. The subject’s initial hematocrit is 39%. The second set of labs again shows a hematocrit of 39%, which is within the normal range. The third set of labs shows a hematocrit of 43%, a deviation from the expected hematocrit of 39%. The PI considers this an unexplained increase in hematocrit that had been previously borderline low but stable.

In trying to develop a differential diagnosis, the PI concludes that he needs more information. The search for additional data is an important, but frequently overlooked aspect of the clinical evaluative process. The coordinator contacts the subject who explains that, during the week before the lab was drawn, he had a GI “bug” with nausea and diarrhea that was “going around his household.”

A simple, initial differential diagnosis would have been:

  • Lab error
  • IP
  • Dehydration and subsequent hemoconcentration

The differential diagnoses are quickly ranked as follows:

1) Dehydration and subsequent hemoconcentration

2) Lab error

3) IP

In this context, the PI decides that the hematocrit of 43% is clinically significant since it was probably indicative of hemoconcentration due to dehydration. He reports the GI symptoms and the hematocrit as AEs and orders a repeat CBC along with a BUN and creatinine.

Considering the Options

As previously stated, grossly abnormal lab values may not be clinically significant. Let’s take a look at two examples.

Although the ALT has a lower limit of normal at 9 U/L, values beneath this level have no clinical significance because “abnormally” low values are not associated with a disease state. For a second example, suppose there is a study being conducted in subjects with pruritus and end-stage renal disease. Screening labs disclose a creatinine of 4. The PI does not consider this to be clinically significant since the subject has known end-stage renal disease and the lab result does not deviate from the expected.

The foregoing examples are not unlikely or contrived, but are very plausible scenarios that illustrate how complicated the process of determining clinical significance can be and how important it is for ensuring subject safety. These examples also underscore how critical it is for the PI to be able to evaluate serial laboratory data for trends.

Because of the way that labs are usually presented in research studies, the PI must locate previous labs and flip back and forth to establish trends in the data. A much more efficient way to accomplish this would be for the central lab to report serial lab data in charts, which would allow the PI to review all the data at a single glance.

It is unlikely that this change in reporting would incur any significant financial expenditures, since most large central labs already possess this ability. Integrating this change into clinical research would be a significant and practical advancement in leveraging the clinical expertise of PIs, thereby improving AE reporting and subject safety.


  1. History: How CONSORT Began. consort-statement.org/about-consort/history
  2. Ioannidis J, Evans S, Gotzsche P, O’Neill R, Altman D, Schulz K, Moher D. 2004. Better reporting of harms in randomized trials: an extension of the CONSORT statement. Ann Intern Med 141:781–8. consort-statement.org/Media/Default/Downloads/Extensions/CONSORT%20Extension%20for%20Harms.pdf
  3. Endorsers: Journals and Organizations. consort-statement.org/about-consort/endorsers
  4. Lineberry N, Berlin J, Mansi B, Glasser S, Berkwits M, Klem C, Bhattacharya A, Citrome L, Enck R, Fletcher J, Haller D, Chen T, Laine C. 2016. Recommendations to improve adverse event reporting in clinical trial publications: a joint pharmaceutical industry/journal editor perspective. BMJ 355:i5078. https://www.bmj.com/content/355/bmj.i5078
  5. Ceh SE. 2009. Documenting clinically significant lab values. J Clin Res Best Pract 5(1):1–4. https://firstclinical.com/journal/2009/0901_CS_Documentation.pdf
  6. Statement of FDA Safety Reporting. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?fr=312.32

Robert Jeanfreau, MD, CPI, (robertjeanfreau@medpharmics.com) is an internist affiliated with multiple hospitals and serves as the president and medical director of Medpharmics in Metairie, La.