PREDICT
: Predicting and Perceiving Pregnancy Risk in Women with Chronic Kidney Disease

Student thesis: Doctoral ThesisDoctor of Philosophy

Abstract

Background:
Pregnancy is associated with accelerated decline in kidney function for women with chronic kidney disease (CKD). These pregnancies are also at greater risk of adverse pregnancy outcomes. Thus, the provision of pregnancy risk information is important to enable informed decisions. Women’s risk perceptions are recognised to impact pregnancy related behaviour, including engagement with prenatal care and adherence to medical advice. It is important, therefore, that women perceive their risk accurately and thus pre-pregnancy counselling is especially important for women with CKD for the provision of risk information. However, it is unknown how women with CKD understand their pregnancy risk and current risk estimates for pregnancy associated progression of CKD and adverse pregnancy outcomes do not allow for individualised risk estimates.

Aim:
The aim of my thesis was to understand women’s perceptions of pregnancy risk with CKD and predict the likelihood of pregnancy-associated progression of CKD and adverse pregnancy outcomes.

Methods:
A systematic review with thematic synthesis explored women’s pregnancy risk perceptions with chronic disease. A content validity study adapted the Perception of Pregnancy Risk Questionnaire (PPRQ) and examined its validity. A cross sectional study investigated women’s perceptions of pregnancy risk and what clinical, social, and psychological factors are associated with perceptions of pregnancy risk and pregnancy intention. Two clinical prediction models were developed using four cohort studies predicting the 1) likelihood of a ≥ 25% postpartum decline in eGFR or initiation of dialysis and 2) an adverse pregnancy outcome defined (preterm birth < 34 weeks and/or small for gestational age (3rd percentile)). A validation study aimed to internally and externally validate both prediction models.

Results:
1. The systematic review included eight papers and identified three themes: Information synthesis; Psychosocial factors and Impact on behaviour. These themes fitted together within an overarching concept of Balancing Act. The themes interrelate to understand risk perceptions in pregnancy with chronic disease.

2. The content validity study (n = 26) reported small amendments to the PPRQ to make it appropriate for measuring women’s pregnancy risk perceptions with CKD. The study reported the PPRQ items and sub-scales are representative, comprehensive and clear. This confirmed that the modified PPRQ is a valid measure of pregnancy risk perceptions in women with CKD.

3. The cross-sectional study (n = 315) reported women with CKD appear to perceive greater pregnancy risk in comparison to uncomplicated pregnancies. A third of women had received pre-pregnancy counselling. Objective CKD severity (estimated glomerular filtration rate (eGFR)) was not associated with perceived pregnancy risk. Clinical characteristics were found to have no association with women’s pregnancy risk perception nor pregnancy intention. Pregnancy risk perception was not associated with pregnancy intention.

4. The prediction model development cohort (n = 368) identified pre-pregnancy eGFR, diabetic nephropathy and proteinuria (albumin: creatinine ratio >70 mg/mmol or protein: creatinine ratio >100 mg/mmol) as predictors for a ≥ 25% postpartum decline in eGFR or initiation of dialysis. This model demonstrated strong performance (ROC = 0.73, 95% CI 0.64 to 0.82). For the prediction of an adverse pregnancy outcome, the identified predictors were chronic hypertension, diabetic nephropathy, and pre-pregnancy eGFR. Together this model reported similar performance (ROC = 0.73, 95% CI 0.66 to 0.80). 

5. The validation study demonstrated strong internal performances, with minimal optimism for both models. The calibration slope in both models indicated slight overfitting. External validation of both models has yet to be performed due to substantial delays beyond my control. Substantial progress has been made to prepare four international cohorts for future external validation work.

Discussion:
Overall, this thesis has demonstrated that women’s perceptions of pregnancy risk and their pregnancy intentions are independent to their objective clinical status. The developed clinical prediction models will improve the provision of risk information to women, to enable informed pregnancy-related decision making. Future research is required to externally validate the models and develop a user interface.
Date of Award1 Dec 2022
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorKate Bramham (Supervisor), Joseph Chilcot (Supervisor) & Yanzhong Wang (Supervisor)

Cite this

'