Using Low-cost Sensors to Verify a Fully Mixed Peritoneal Dialysis Solution for On-site Production in Low- and Middle-Income Countries

Abstract

There is limited access to renal replacement therapy (RRT) for end stage kidney disease(ESKD) patients in low- and middle-income countries (LMICs). RRT is needed when a patient’s kidneys no longer function properly; the kidneys’ ability to remove waste and excess fluid from a patient’s blood is insufficient. While kidney transplant is a form of RRT, patients in LMICs diagnosed with ESKD are often not able to receive a transplant and instead rely on dialysis, if and when dialysis is even available. Access to dialysis is limited for a number of reasons, including economic factors, patient awareness of the treatment, nephrologist availability, and travel time to healthcare services. Dialysis, when available, is generally hemodialysis (HD) rather than peritoneal dialysis (PD). HD is an RRT treatment performed in a dialysis clinic or hospital three times a week for ESKD patients. An external filter, known as a dialyzer, removes waste and excess fluid in the blood through osmotic and concentration gradients with the assistance of a concentrated fluid, the dialysate. In contrast, PD is a daily RRT treatment for ESKD patients performed in a dialysis clinic, hospital, or patient’s home. Dialysate is added into the peritoneal cavity and the peritoneum, a membrane that lines the abdomen, acts as a natural filter. This research aimed to: (i) understand the current limitations to dialysis access in Nigeria, (ii) define a device for on-site mixing of PD dialysate, (iii) determine if low-cost conductivity sensors and refractometers can be used to predict sufficient mixing. An on-site ethnographic study was conducted in Lagos, Nigeria to understand existing barriers to RRT for ESKD patients. A secondary aim was to gather the perception and opinions of patients on PD as an alternative treatment option to HD. The study consisted of semi-structured interviews of thirty-three patients that were receiving HD treatment at the time. The interviews were recorded and transcribed. The transcripts were then analyzed based on descriptive and emotion codes. The frequency of the overlap between the descriptive and emotional codes suggested that the high cost of treatment, the time spent during treatment, and the time and cost associated with transportation to and from the treatment site were critical barriers to treatment access.When patients were presented with a visual representation and description of peritoneal dialysis, the majority of patients preferred PD over their current HD. The findings from the ethnographic study suggest that increasing access to the PD treatment modality would increase overall access to RRT in Nigeria. To enable PD as a RRT treatment option in Nigeria and other LMICs, a device would have to be created to produce the necessary PD fluid (dialysate) for treatment. Attempts to import the PD fluid have not been economical and frequently result in supply-chain challenges. The proposed device is an on-demand, in-clinic or in-pharmacy PD fluid generation device. The device will consist of three main modules that: (i) create water-for-injection (WFI), (ii) mix the necessary solutes with the WFI, and (iii) bag the solution. Importantly, these steps must occur without the requirement of a clean room. The focus of this thesis is the mixing module. The mixing module must produce PD fluid and verify the composition is as prescribed while using low-cost sensors. The PD fluid for the proposed device consists of a mixture of four salts and one sugar in water. Various conventional and non-conventional mixing methods were considered for the device, but ultimately mechanical batch mixing of the solutes and water with a mixer and impeller in a mixing tank was found to best suit the device design. The initial hypothesis was that conductivity sensors (x2) and a refractometer could be used to verify dialysate composition. Mixing and verification followed these steps: fill the tank with water, add the sugar, mix the solution, verify the sugar concentration with the refractometer, add in the salts to the same solution, and lastly verify the salt concentration with conductivity sensors. An experimental batch mixing and fluid verification system was constructed in laboratory to test the presented hypothesis. The required mixing speed was determined using the Zwietering correlation for suspended solids in a mixing tank. The Onsager/Falkenhagen, de Diego, Shahrouz, and SM 2510A models were used to predict the well-mixed experimental conductivity based on the composition of the dialysate solution. The Onsager/Falkenhagen model was able to most accurately predict the conductivity of the fully mixed PD fluid. The Onsager/Falkenhagen model over-predicts the mean experimental conductivity by 5.05%. A correction factor on the Onsager/Falkenhagen predicted conductivity is recommended for greater accuracy in modeling the experimental conductivity. For the verification of the sugar, a refractometer was used. Refractometers measure the index of refraction which can be converted to °Brix, the sugar content of an aqueous solution. The targeted index of refraction in °Brix is 6.2% less than the experimental reading from the refractometer. When testing the low-cost sensors, it was discovered that the presence of air bubbles due to mixer agitation and the presence of sugar effect the conductivity of the solution. When using such sensors for fluid verification, the mixer should be turned off when taking readings. Because dextrose effects the conductivity of the solution, a correction factor will need to be applied to the conductivity reading before correlating to salt concentration.