Hello Kaled Nather Taha

In my opinion, participants must be randomly recruited as it will be more heterogeneous and this would help to bring down the difference in serum biomarkers. Also, there are other factors that could contribute to the difference in serum biomarkers in healthy controls such as demographics, genetic factors, or presence of co-morbidities.

One cannot afford to ignore the assay-related factors. Analytical methods can be subjected to both random and systematic measurement errors. The assay specificity is an important factor to be considered. It refers to how well the assay (with its antibodies and other components) can distinguish between its intended analyte and other structurally similar components. Poor specificity can lead to a systematic overestimation of biomarker levels.

In assay development dilution linearity (measurement levels should be proportional to sample dilution), and parallelism (standard reference and serially diluted sample curves should be parallel) need to be controlled. Sensitivity to interference like lipids, hemolysate or antibodies may result in both over- or underestimation of the measured concentrations of biomarkers.

Biomarker measurements are affected by lot-to-lot variability. This can introduce differences both between studies, and within studies. For instance, if many lots of analytical kits are used within a large study.

Preanalytical factors which include time-of-day for sampling, the technique used by phlebotomist, tube-related factors, and errors in tube handling and labelling could also be one of the contributing factors for differences in serum biomarker levels.

Some biomarkers may vary due to normal physiological processes, or in response to diet, stress factors, or other health issues which are not known. Also, some biomarkers may show instability over time. Sampling and storage procedures may also affect biomarker levels. Intra-assay variability and inter-assay variability must be monitored. Assay validation methods must be completed to ensure that an analytical methodology is accurate, specific, reproducible and rugged over the specified range that a target analyte is to be analyzed.

I hope this is helpful.

Best,

Interesting question - can you post the papers?

John W Pickering Thank you very much for your appreciation. This is a testimony I cherish for your admiration for this question, as a result of my reading of 20 research papers on cytokines 10.13.33.35... and I found a significant and significant difference in the concentrations of the healthy group despite using the same ELISA technique.

So I ask this question

I see. Well, I certainly expect that if you use the usual p = 0.05 to judge if there is a difference then with 20 papers you will see a difference. That is just by chance. I no longer use the concept of statistical significance because it can be very misleading. I just report the magnitude of the differences.

One of the difficulties in making this comparison is how papers report the results of biomarkers. Quite often biomarkers, particularly in control groups, are right-skewed. Because of this the mean and standard deviation are not good descriptions of the cohort (better is median and inter-quartile range). As a "rule of thumb" if the standard deviation is more than half of the mean then the distribution is skewed and it is not really possible to compare means.

A second difficulty is the assay itself - what proportion of patients are below the limit of detection? If this is reported, it may be interesting to see if this proportion differs between papers.