Validating the Selection Process of Healthy Control Group Based on Apparently Healthy Volunteers

Introduction

Correct interpretation of laboratory tests is a major concern for physicians and medical laboratories. Accurately validated reference intervals (RIs) for each quantitative test is one of the main criteria for medical decision. Reference values (RV) are used to define the dispersion of variables in healthy individuals. RVs first introduced as a philosophy, have gained worldwide acceptance as one of the most influential tools in laboratory medicine to help in the clinical management process [1, 2, 3]. RI is defined as the interval between two reference limits (these included).

Reference subjects are generally assumed to be ‘‘healthy’’; however, health is relative and lacks a precise and quantifiable definition. Therefore, reference individuals are selected using ‘‘well-defined criteria’’ i.e., inclusion and exclusion criteria, which approximate health. Such criteria should be defined specifically, according to the goals of the study, and may differ from one study to another. Strictly derived reference ranges are critical for differentiating healthy from diseased individuals and constitute the foundation of our contemporary methodology to making the diagnosis of clinical disorders.

Defining a healthy subject is not easy. Diverse criteria underlying the concept of wellness can be implied. The Royal College of Physicians has defined the healthy volunteer as an “individual who is not known to suffer any significant illness relevant to the suggested study, who should be within the normal range of body measurements. In addition, the mental state of healthy volunteers is such that he is able to understand and give valid consent to the study [4]. The EMEA guideline also proposes a general definition of healthy volunteer for studies aimed at assessing pharmacokinetics: “healthy, adult volunteers, in well-defined and controlled conditions” [5]. This definition implies that the selection of healthy volunteers is conducted by enrolling subjects without relevant pathologies and with organ functions, such as heart, liver and kidney, in the normal range.

Such general definitions of healthy volunteer, allow wide margins of discretion. A control group may include individuals similar to the trial group in all features that affect the results except for the (treatment/intervention) of interest. On the basis of comparability to the target persons or the persons at risk, controls are carefully selected. This group is critical to determine a treatment or intervention, also aiding in the assessment of efficacy and safety. A control group distinguishes results produced by the treatment or intervention of interest from those caused by other factors, for example normal course of disease [6]. In designing a clinical trial, the choice of control group is always a serious decision, because the choice affects the inferences derived from the trial [7].

Material and Methods

A total of 217 apparently healthy adult subjects were enrolled in the current study during the period of 3 months extending from March to June 2016. The sample was randomly selected from apparently healthy individuals accompanying patients attending the outpatient clinic in a tertiary referral hospital. A systematic random sample of 4 subjects was selected daily from the list of clinic attendants during the 5 working days of each week. The targeted patient was approached and his accompanying adult was asked for his consent to participate in the study. If the targeted subject had no accompanying adult, that subject did not fulfill the inclusion criteria, or the consent was denied, then the next in the list was approached. The age of participants ranged between 18 and 69 years with a mean ± SD of 32 ± 13 years.

Information related to the health status were recorded. Anthropometric measurements (height, weight) were assessed and blood presser was measured using a standard mercury sphygmomanometer. Body temperature was measured using electronic thermometer. Medical history of hypertension, diabetes, chronic diseases and risk factors such as surgical operation and previous admission into hospital was also looked for.

A sample of venous blood (10 mm) was collected in the morning after an overnight fasting, and then subdivided into tow portions. The first one is used for the measurements of hematological indices by using EDTA was as an anticoagulant for preforming complete blood picture test, whereas the second part of blood samples were centrifuged in plane tubes at 3000X rpm for 15 min. Aliquots were separated and prepared for storage (−20°C) until further analysis.

Lipid profile including Total cholesterol (TC), triglycerides, and high -density lipoprotein cholesterol (HDL-C) were measured with commercially available enzymatic colorimetric kits (from biolabo, France). Serum low density lipoprotein LDL-cholesterol (LDL-C) was calculated according to the Friedewald’s formula. This equation calculates the concentration of LDL-C based upon the presence of total cholesterol, HDL and triglyceride levels. In addition to biochemical related tests for serum alanine transferase, urea and creatinine. LDL=total cholesterol−HDL− (triglycerides/5).

Immunological related tests including serum total immunoglobulin Ig (IgM, IgA, IgG) from (LTA, Italia), complement test (C3,C4) from (LTA, Italia) were measured by using immune-diffusion test. Rheumatoid factor by using agglutination test (spinreact, Spain), serum levels of high sensitive C reactive protein (Hs-CRP) and interleukin 1 beta (IL-1 B) were measured by applying the Enzyme Linked ImmunoSorbent Assay (ELISA) technology, from (Labor Diagnostika Nord(LDN), Germany, Diaclone(France).respectivly.

Statistical Analysis

Data were translated into a computerized database and then was examined for errors using range and logical data cleaning methods, and inconsistencies were identified and corrected. Statistical analyses were done using IBMSPSS version 23 computer software (IBM Statistical Package for Social Sciences) in association with Microsoft Excel. Frequency distributions for selected variables were done first.

As shown in table 2, the count of abnormal test values that belongs to a specific test domain was studied. The prevalence rate of apparently healthy individuals with a completely normal lipid profile (composed of 5 test components) was 25.3%, while those with at least two abnormal test components was 34.1%.

Results

The results presented in this research were based on the analysis of 217 apparently healthy control subjects.

Very young adults constituted 43.3% of the sample, while those 50 years and older were 14.3%. Gender composition was almost equal with females comprising 52.5% of the study sample. Slightly more than half of the studied subjects (53.5%) were of normal BMI, while only 14.3% were obese. Smokers were 28.1% of the sample. A positive history of hospital admission and surgical intervention was observed in 27.6% of subjects. In addition, a positive family history of DM, hypertension and RA was observed in 2.8, 15.7 and 16.6% of study sample respectively, table 1.

Discussion

“An event, condition, or characteristic is not a cause by itself as an intrinsic property it possesses in isolation, but as part of a causal contrast with an alternative event, condition, or characteristic” [8, 9, 10]. The Neyman–Rubin causal model also known as the “potential outcome or counterfactual model” established the general framework for using the control or reference group in both observational and experimental types of analytic studies [11, 12]. Referring to this definition of a cause or risk factor one can understand the critical role for selecting a control group in any analytic study. The choice of a control group in any analytic study can be the single most important factor in deciding the outcome of the study. It can change the possible conclusion of a study from no association between an exposure (explanatory variable, risk factor or a possible cause) and an outcome (response variable or disease status) to a strong effect or association.

Researchers in the field of medical or biology sciences often need to enroll a group healthy control subjects in their studies. The term “apparently healthy control” is more suitable than “healthy control”, since any subject who is not currently complaining, has no annoying symptoms or obvious clinical features can be included in The present study used a statistical framework to screen healthy subjects from a group of “apparently healthy” individuals. Some tests provide a simple yes or no (positive or negative, reactive or non-reactive) interpretation. Other tests are quantitative. A typical lab report will provide such test results followed by a reference range [13]. The term "reference range" is preferred over "normal range" because the reference population can be clearly defined. Rather than implying that the test results are being compared with some vague definition of "normal," the reference range means the results are being considered in the most relevant context. By definition, 1 out of 20 (or 5%) results will fall outside the established reference range with specimens taken from a random sample of healthy individuals [14, 15].

The current study evaluated the probability of being qualified as “Healthy Control” with selected laboratory tests within internationally agreed reference range. A total of 34 different tests were done per study participant.

These tests belonged to lipid, biochemistry, hematology and immunology domains. If all the domains are needed to define a healthy control only 13.4% of the apparently healthy controls would qualify for a laboratory verified healthy controls. However most of the analytic studies would require a laboratory verified healthy control in only one domain of tests. More than three quarters of randomly selected apparently healthy controls would qualify for inclusion as hematologically verified healthy controls. This probability decline to slightly more than a half for immunologically verified healthy controls. Finally, a lipid profile or biochemistry verified healthy controls would be obtained with a probability of around two thirds among a random sample of apparently healthy controls. In conclusion it can be said that a very considerable proportion of population who appear to be healthy , are not in reality, accordingly not all apparently healthy controls are qualified as eligible control.