Managing Clinical Heterogeneity: An Argument for Benefit-Based Action Limits

Abstract

The use of reference ranges is well established in medical practice and research. Classically, a range would be derived from the local healthy population and matched in age, gender, and other characteristics to the patients under investigation. However, recruiting suitable controls is problematic and the derivation of the range by excluding 2.5% at each end of the distribution results in 5% of the values being arbitrarily discarded. Thus, the traditional reference range is derived using statistical and not clinical principles. While these considerations are recognized by clinicians, it is often not realized that the application of whole population derived reference ranges to complex pathologies that comprise patient subgroups may be problematic. Such subgroups may be identified by phenotypes including genetic etiology, variations in exposure to a causative agent, and tumor site. In this review, we provide examples of how subgroups can be identified in diverse pathologies and how better management can be achieved using evidence-based action limits rather than reference ranges. We give examples from our clinical experience of problems arising from using the wrong reference ranges for the clinical situation. Identifying subgroups will often enable clinicians to derive specific action limits for treatment that will lead to customized management and researchers a route into the study of complex pathologies.

Long‐term testosterone therapy in type 2 diabetes is associated with reduced mortality without improvement in conventional cardiovascular risk factors

“…In our opinion, many chronic pathologies are heterogeneous, with common phenotypes leading them to be grouped as single diseases . We have previously shown that statins and fibrates exert greater effects in certain dyslipidaemic subgroups .…”

Testosterone replacement therapy: Pre‐treatment sex hormone‐binding globulin levels and age may identify clinical subgroups

“…24,25 The variation in SHBG level raises issues regarding how a therapy that changes body testosterone content might influence the serum level and if so would this have clinical implications. [26][27][28][29] We are unaware of a formal study though the BLAST double-blind placebo-controlled intervention randomized controlled trial (RCT) in men with T2DM found, after 30-week TRT, that SHBG levels decreased significantly in those with pre-treatment total testosterone < 8nmol/L. 17 The significance of a TRT-associated fall in SHBG is unclear though it could be related to data indicating the protein functions as more than a steroid hormone carrier.…”

“…We hypothesized, firstly, that the levels of SHBG found in men with adult-onset TD would show significant inter-individual variability in the degree of change following TRT and, secondly, that the wide range of SHBG levels prior to TRT and expected variation in the extent of change during therapy would demonstrate aspects of patient heterogeneity and possibly help identify subgroups that respond differently to therapy. 26 Accordingly, we first determined whether TRT was associated with change in SHBG over the 30-week study and whether baseline SHBG and age were associated with ∆SHBG. We used these data to identify subgroups based on combinations of median values of baseline SHBG and age, and determine whether TRT differently affected change in relevant variables such as waist circumference, HbA1c and blood pressure in the putative subgroups.…”

“…is further indicated by examining the men stratified into four groups by combinations of high or low values of total and free testosterone; the two groups with the highest values of SHBG demonstrated the worse sexual, physical and psychological function scores 29. Accordingly, we speculated that SHBG and age might modify the effect of therapy on clinical phenotype thereby allowing subgroups to be identified 26. …”

An audit of the measurement and reporting of male testosterone levels in UK clinical biochemistry laboratories

“…It is important that clear interpretation is given and that relevant treatment guidelines are accessible and/or referenced in the laboratory report. Moreover, reference intervals based on population distributions are often misinterpreted as the “normal” range and can lead to confusion in the absence of any treatment guidelines 22 . A number of clinical guidelines have been published in recent years which address the matter of the diagnosis and treatment of hypogonadism in men.…”