Chronic Kidney Disease Observational Database - Taiwan

An important part of care in chronic kidney disease is an adapted diet. Its most important aspect is protein restriction. The rationale for protein restriction is a reduction of uremic wastes. However, nutritional requirements for protein synthesis limit the maximum extent of protein restriction. To deal with these conflicting targets, a minimum protein intake supplemented with ketoanalogues of amino acids (supplemented very low protein diet, sVLPD) meets the protein needs while reducing uremic waste.

The aim of this explorative, observational study is to determine the effect of sVLPD on eGFR decline compared to a low protein diet (LPD) or no protein restriction.

Data collection The study uses only data from routine health care records. The transfer to the electronic case report form is done by center investigators.

Data entry is monitored monthly for completeness and plausibility. Missing or unusual data will be requested for completion or re-assessment.

In case of high loss to follow up (>10%), low follow up frequency (<90% of patients with <3 visits per year), or greater than 5% missing core data (age, gender, descent, height, weight, history of diabetes and hypertension, blood pressure, serum creatinin, dietary prescription, judgement of compliance), audit visits including source data verification and trainings may be done.

Primary analysis Direct comparison of patients receiving sVLPD or LPD is not meaningful due to the non-interventional design. It is expected that sVLPD patients will have a more advanced stage of CKD, most likely will have higher severity of disease and possibly may have different demographic baseline data. Furthermore, other well-known risk factors for progression of chronic kidney disease like the presence of diabetes mellitus or high blood pressure may affect eGFR decline.

A relevant amount of data is expected to be missing due to the observational nature and the use of data from clinical routine. Furthermore, missing data are unlikely to occur completely at random.

Therefore, missing data will not be imputed but they will be implicitly modeled by a mixed model: mean changes of eGFR from baseline will be analyzed using a restricted maximum-likelihood based repeated measures approach. Analyses will include the fixed, categorical effects of actual treatment, study center, gender, visit time, and baseline variables presence of smoking history, diabetes mellitus, hypertension, and baseline eGFR. Patient will be included as a random factor to the model. Significance tests will use a two-sided α = 0.05.

Secondary analyses Compliance, dietary counselling, use of a nutritional diary, primary diagnosis of CKD, diabetes mellitus, and vegetarian diet will be included to the model described before and analyzed for independent effects or effect modification of the diet.

The approach to the secondary analysis of development of serum urea is similar to the primary analysis.

Cox-regression analysis will be done for time to dialysis initiation or reaching the composite endpoint [>50% eGFR decline or initiation of maintenance dialysis treatment] including the same baseline variables as mentioned for the primary endpoint.