Abstract
Background: Inpatient gout flare is a common problem which could lead to longer length of hospital stay and higher hospital cost. However, factors associated with inpatient gout flare are not well understood.
Objectives: This thesis has three objectives: (1) to identify the predictors of inpatient gout flare in people with comorbid gout, (2) to explore the relationship between the predictors of inpatient gout flare and the length of hospital stay, and (3) to develop and validate a clinical tool to identify people who are at high risk of developing gout flare during hospital stay.
Methods: Three studies were conducted to address each of the objectives. The first study collected data from a retrospective cohort of people with comorbid gout admitted to three hospitals in the Wellington region in 2017. Fifty-two candidate variables were explored, with inpatient gout flare (yes/no) as the dependent variables. A prediction model was built using clinical knowledge-guided variable selection followed by logistic regression with shrinkage. The second study used data from a population-based cohort of people with comorbid gout admitted to New Zealand public hospitals in 2017. The association between 19 gout flare-related variables and the length of hospital stay was explored using a generalized linear model. In the third and final study, a prediction rule for inpatient gout flare was developed from the set of predictors identified in the first study. The prediction rule was then validated in an independent cohort of hospitalized people with comorbid gout (validation cohort) prospectively recruited from a hospital in Thailand.
Results: The first study (N =625) identified nine predictors of inpatient gout flare: (1) pre-admission serum urate >0.36 mmol/L, (2) tophus, (3) no pre-admission urate-lowering therapy (ULT), (4) no pre-admission gout prophylaxis, (5) ULT adjustment, (6) gout prophylaxis started/increased, (7) diuretic adjustment, (8) acute kidney injury and (9) surgery. In the second study (N =36,047), regular pre-admission ULT and urate testing were found to be associated with a shorter length of hospital stay. Loop diuretics, potassium-sparing diuretics and surgery were found to be associated with a longer length of hospital stay. People with multiple factors associated with longer length of stay were estimated to add at least four days to their hospital stay. In the third study, a prediction rule for inpatient gout flare was developed, containing four items; (1) no pre-admission GOut prophylaxis, (2) no pre-admission ULT, (3) Tophus and (4) pre-admission serum urate >0.36 mmol/L (the GOUT-36 rule). The presence of two or more items indicates that the person is at high risk of developing gout flare during hospital stay. In the validation cohort (N =184), the GOUT-36 rule has a sensitivity of 0.74, specificity of 0.69 and AUC of 0.71.
Conclusion: The thesis identified nine predictors of inpatient gout flare, as well as the association between some of the predictors and the length of hospital stay. The GOUT-36 prediction rule for inpatient gout flare was sensitive, intuitive and user-friendly. All four items in the GOUT-36 rule are assessable on the first day of admission, allowing a very early risk stratification for people with comorbid gout.