RATIONALE AND BACKGROUND
Irritable bowel syndrome (IBS) is a chronic, relapsing gastrointestinal condition characterised by abdominal pain, bloating, and changes in bowel habits. Prevalence estimations vary with the diagnostic criteria used, and in the United Kingdom (UK) were estimated between 9.5% and 22%. IBS can be classified according to Rome III criteria on the basis of the stool?s characteristics: IBS predominantly with diarrhoea (IBS-D); IBS predominantly with constipation (IBS-C); and IBS with mixed bowel habits (IBS-M). Approximately one-third of IBS patients have each type of the disease.
The commercialisation of linaclotide (Constella®), a guanylate cyclase-C receptor agonist with visceral analgesic and secretory activities, was approved as the first medicine authorised for the symptomatic treatment of moderate-to-severe IBS-C in adults in the European Union (EU).
Therefore, this study is planned to assess the safety of linaclotide in terms of the risk of severe complications of diarrhoea (SCD) during treatment and other risk factors among patients with IBS-C in Spain.
RESEARCH QUESTION AND OBJECTIVES
This study will occur over two phases. In Phase I, we will validate an algorithm for identifying SCD. In Phase II, we will use either a cohort design (Phase IIa) or a case-control design (Phase IIb) depending on the results the validation study to investigate whether use of linaclotide is associated with an increased risk of SCD.
The specific objectives of the study are:
Phase I: Validation Study
– Estimate the incidence of diarrhoea and identify patients experiencing severe complications of diarrhoea (SCD) among patients with IBS-C
– Validate a proposed diagnostic algorithm for SCD among patients with IBS-C
Phase IIa: Cohort Study
If a PPV>95% is achieved in the validation study, then we will use a cohort design to:
– Estimate the incidence of SCD among IBS-C patients prescribed linaclotide overall and stratified by groups of interest (i.e. patients ?65 years, patients with hypertension, diabetes, or cardiovascular disease diagnostic codes)
– Identify risk factors for SCD, with linaclotide and laxative use being the key exposures of interest
Phase IIb: Case-control Study
If a PPV<95% is achieved in the validation study, then Phase IIb will be executed, in which a matched case-control design will be used to:
- Identify risk factors for SCD, with linaclotide and laxative use being the key exposures of interest
The rationale for this study design is that if the results from the validation study show that the SCD algorithm has a good performance, we can implement this algorithm to identify our outcome of interest within the entire study population and conduct a classic cohort analysis. However, if the algorithm has poor performance, then we will only be able to utilise confidently the data within the validation sample for the implementation of a matched case-control study.
STUDY DESIGN
The final decision on study design will be made following the validation stage. If the PPV is >95% then a cohort study design will be used, otherwise a matched case-control study will be used.
POPULATION
This study will use observational data from and Spain. The study population will be a cohort of IBS-C patients. Patients with less than 12 months of computerised records prior to IBS-C cohort entry date or no follow-up time will not be included.
VARIABLES
Users will be classified as current, never and past users. For Phase IIa (cohort study), these exposures will be defined based on drug prescriptions in a time-updated manner The outcomes of interest are severe complications of diarrhoea including dehydration that requires intravenous rehydration, dehydration that requires oral rehydration with solutions of electrolytes, electrolyte imbalance (potassium and sodium), oliguria, anuria, new onset-thromboembolism, new-onset orthostatic hypotension, new-onset syncope, new-onset dizziness, new-onset vertigo, acute renal failure, hypovolemic shock, hospitalisation due to diarrhoea, stupor, coma, or death.
All variables will be identified using diagnostic and procedure codes from general practitioner (GP) electronic medical records. IBS type and safety outcomes will be validated through a questionnaire to physicians treating the cases and controls included in the study.
DATA SOURCES
This part of the study will utilise data from the Information System for the Development of Research in Primary Care (SIDIAP). SIDIAP is a primary care database that collects longitudinal data from electronic medical records (EMRs) from 274 primary care centres in Catalonia since 2006, representing approximately 12% of the Spanish population. Data from primary care, specialised care, hospitals, and pharmacies are available, as well as patient characteristics like gender and date of birth, GP-diagnosed conditions, GP prescriptions, prescription dosing and size, date of prescription and dispensation, laboratory test results, other procedures, specialist referrals and diagnosis, hospital referrals, hospital procedures and discharge diagnosis, death date, and pregnancy information. GP diagnoses are coded following the ICD-10 codes, and hospital admissions are classified following the International Statistical Classification of Diseases and Related Health Problems, Ninth Revision (ICD-9). Since 2018, SIDIAP no longer allows the possibility to contact physicians for extracting extended additional information on the patient through long questionnaires. However, there is the possibility of a text review of the primary care records. This text review can be used to validate cases of interest and it would be conducted by IDIAP?s clinical researchers. For this study, it is recommended by the data custodians that a review of free-text in the patient?s medical records is conducted to improve the validity of the actual study, in place of the previously planned validation strategy using questionnaires to GPs. This free-text review will be made after an anonymisation process of the clinical records and a natural language processing of the text.
STUDY SIZE
From the linaclotide drug utilisation study we know that the total number of linaclotide users in SIDIAP between 2014 and 2017 was 1,854 patients.
Within the IBS-C cohort, we expect to accrue a non-trivial number of cases (e.g., 94 cases of SCD with an incidence of 0.05% among IBS-C patients) and the same number (or larger) of controls (e.g., that would provide 80% power [or greater] to detect an OR=4 [or larger] associating SCD with a risk factor present in 5% of controls [notice that we expect exposure to linaclotide in approximately 5% of the cohort]).
DATA ANALYSES
Patient characteristics at cohort entry date will be described for the full cohort of patients with IBS-C, as well as incidence rate of diarrhoea.
If the results of the validation of cases and controls are satisfactory (positive predictive value of EMRs in Spain ? 95% and negative predictive value ? 99%), then information from the whole cohort will be used to estimate incidence rates for those groups of patients with IBS-C prescribed linaclotide and those who are potentially at increased risk of SCD and prescribed linaclotide. The crude incidence of diarrhoea and of severe complications of diarrhoea will be described by calculating the proportion of patients with IBS-C experiencing diarrhoea and SCD, respectively, and their associated 95% confidence intervals (CIs). Crude incidence rates of SCD will be also calculated.
Additionally, the hazard ratios of SCD and the exposures of interest (mainly prescription of linaclotide) will be estimated using Cox proportional hazard models.
If the results of the validation of cases and controls in Spain are not satisfactory (positive predictive value of our outcome algorithm in Spain < 95% or negative predictive value < 99%), only the nested case-control study will be performed and ORs of SCD and the same exposures of interest will be estimated using data from cases and matched controls by applying conditional logistic regression analysis.
The goal is to assess SCD among patients who were treated with linaclotide, were ?65 years old, and had history of hypertension, diabetes or cardiovascular disease diagnostic codes (i.e., SCD occurrence is the outcome variable [yes/no] and the independent variables of main interest are whether the patient has a prescription for linaclotide, a prescription for laxatives, the patient is ?65 years old, and the patient has hypertension, diabetes, or cardiovascular disease diagnostic codes, all of them at index date, or not, controlling for potential confounders).
Statistical analyses will be conducted using R statistical software.