Examining Effect of Nutrition Interventions to Reduce Hyperphosphatemia in Chronic Kidney Disease: Is Including Non-Randomized Trials A Waste of Time?

Article type
Authors
Rozga M1, Cheng F1, Moloney L1, Handu D1
1Academy of Nutrition and Dietetics
Abstract
Background: Randomized controlled trials (RCTs) are the gold standard for examining efficacy of health care interventions. However, RCTs of nutrition interventions may not always be feasible due to the long periods of time that nutrition may require to affect health outcomes, lack of generalizability and lack of funding for nutrition interventions.

Objectives: To examine if there is a difference in the effect size and certainty of evidence from RCTs only versus from RCTs and non-randomized controlled trials (NRCTs) combined in a nutrition-focused intervention. This systematic review included trials examining the efficacy of phosphorus-specific nutrition counseling provided by a registered dietitian nutritionist (RDN), compared to usual care or an alternative intervention, on serum phosphorus levels in dialyzed individuals with chronic kidney disease (CKD).

Methods: The literature was searched using Medline, Embase, CINAHL, Web of Science and other databases from January 1, 2000 until the search date of November 23, 2019. RCTs or NRCTs were assessed for risk of bias using the ROB 2.0 and ROBINS-I tools and certainty of evidence was determined using the GRADE method. Results were included in meta-analysis, when possible. Results were compared between RCTs alone or RCTs in combination with NRCTs.

Results: Eleven RCTs and three NRCTs met inclusion criteria. Serum phosphorus levels were a primary outcome in all 14 studies. Nutrition interventions resulted in a significant decrease in phosphorus levels in the ten RCTs included in meta-analysis [Mean (95% CI): -0.80 (-1.18 to -0.43) mg/dl; I2=65.5%]. In the RCTs, overall risk of bias was high in two studies and there were some concerns in the remaining nine studies, primarily due to randomization and deviations from the intended interventions. Two of the three NRCTs were included in pooled analysis and, in combination with RCTs, results describe a reduction of -0.83 (-1.14, -0.51) mg/dl (I2 =59.7%) in serum phosphorus levels in the intervention groups. In the three NRCTs, two studies resulted in an overall judgement of serious risk of bias and one in moderate risk of bias, primarily due to confounding and measurement of outcomes. Certainty of evidence was not affected whether or not NRCTs were included in the body of evidence; either way, certainty of evidence was “LOW”, suggesting phosphorus-focused nutrition therapy from an RDN or equivalent reduces serum phosphorus levels.

Conclusions: When examining the effect of phosphorus-focused nutrition therapy on phosphorus levels in individuals with CKD on dialysis, overall effect size and certainty of evidence was not notably affected by including or excluding NRCTs. Therefore, when RCTs are identified a priori, it may save time and effort to consider RCTs without NRCTs. Further research is needed to determine if inclusion of long-term cohort studies improves understanding of the long-term feasibility and effects of nutrition interventions on patient-centered outcomes in more generalizable populations when assessed in tandem with RCTs.