The use of artificial intelligence and data analytics in life insurance
The emergence of data analytics and machine learning is providing insurers and reinsurers with new insights into how they drive and monitor their business.
Adherence to long-term treatment is a key strategy in the management of chronic diseases. Yet according to the World Health Organisation (WHO), even in developed markets, only 50% of patients who suffer from chronic diseases adhere to treatment recommendations. According to the US Food and Drug Administration (FDA), adherence to medicines is defined as the extent to which patients take medication as prescribed by their doctors. This involves factors such as getting prescriptions filled, remembering to take medication on time and following the directions given by their doctors. Poor adherence is a critical challenge for the healthcare industry as it is closely associated with suboptimal health outcomes, increased morbidity and mortality and augmented healthcare utilisation and costs. For example, a 2017 retrospective cohort study using a large US claims database explored why clinical outcomes are not keeping pace with the availability of new treatment options. The study found that real-life HbA1c reductions fell far short of those reported in randomised clinical trials (RCTs), with poor medication adherence emerging as the key driver behind the disconnect.1 The Healthcare Effectiveness Data and Information Set (HEDIS) 2019 report shows that only about 40% of commercially insured health maintenance organisation (HMO) patients and 30% of Medicaid patients achieve the HbA1c target goal of less than 7.0%.2 One of the major contributory factors for failure to achieve the target HbA1c goals is poor medication adherence.3
It therefore becomes critical for insurers and healthcare providers to track and monitor medication nonadherence. What strategies and models can we implement that are reliable, practical and cost-effective to accurately measure medication nonadherence? To understand this better we conducted a limited literature search to identify medical publications between January 2015 and December 2020 that focused on best practices adopted by healthcare stakeholders across the globe to measure medication adherence.
The table in Figure 1 summarises different methodologies adopted by healthcare stakeholders to measure medication adherence.
|Method||Description||Key assessment criteria|
|Drug monitoring through lab tests||Concentration of drug/metabolite in body fluids (blood, urine) or hair as a measure of patient’s medication use levels.||Low||Low||High||High|
|Claims or pharmacy data||Pharmacy and insurance claims data, electronic prescription service (EPS), registries.||High||High||Moderate||High|
|Electronic pill box/monitors/ sensors||Electronic Medication Packaging (EMP) devices incorporated into the packaging of a prescription medication. Examples: Electronic pillboxes and ingestible sensors; Medication Events Monitoring System (MEMS).||Moderate||Low||High||Moderate|
|Manual verification of pill box by care giver/nurse||The pill count taken by the patient is calculated by subtracting the number of pills remaining in the bottle from the total number of pills dispensed. Drug adherence rate is then calculated as number of pills taken divided by number of days elapsed since the last dispense.||High||High||High||Low|
|Self-reporting surveys||Examples: Medication Adherence Report Scale (MARS), 4- and 8-item Morisky Medication Adherence Scale (MMAS-4 and MMAS-8), Hill-Bone Medication Adherence (HBMA) scale.||High||High||Moderate||Low|
|Clinician's assessment||Examples: Structured patient interviews, Brief Medication Questionnaire (BMQ)||High||High||Moderate||Low|
We found the multiple methodologies to measure and monitor drug adherence in these studies used by different stakeholders have their own advantages and limitations. For example:
During the literature review we came across some interesting emerging practices to manage this challenge. The table in Figure 2 summarises some advancements in standardising medication adherence measures. For emerging markets where disease management approaches by providers or payer offerings are still at an early stage of development, this highlights the importance of monitoring mechanisms in the programme planning and information collection processes.
|Ascertaining Barriers for Adherence (ABC taxonomy)5||Introduction of new adherence terminology by defining three essential components of adherence. These components are:
-Initiation (taking the first dose of the prescribed medication)
-Implementation (taking medication as prescribed)
-Discontinuation (stopping treatment)
|Unified and standardised terminologies for all health conditions will help stakeholders across disciplines to improve the measurement and reporting of medication adherence.|
|Extension of the SNOMED CT terminology5||Systematised Nomenclature of Medicine-Clinical terms (SNOMED CT). An extension of the SNOMED CT terminology to include terms of adherence in the amendment to ISO 13940 (system of concepts to support continuity of care).||SNOMED CT codes to support clinical documentation and allow for standardised collection and reporting of data related to quality measures.|
|ISPOR Medication Adherence Good Research Practices Working Group recommendations6||ISPOR, the Professional Society for Health Economics and Outcomes Research, has developed a set of recommendations to guide future investigators into producing high-quality Initial Medication Adherence (IMA) research||To summarise, compare and evaluate the existing measurement methods used for calculating medication adherence, regardless of disease area, in patients using polypharmacotherapy.|
Choosing and implementing the measure or combination of measures best meeting a particular organization’s requirements will help frame the most competent and personalised intervention to recognise the medication-taking behaviours in patients with chronic diseases. For a payer, using the most appropriate measure would help improve disease management effectiveness and service profiling. Accurate assessment of medication adherence will help in designing both optimal and cost-effective strategies to improve medication adherence. Each measurement approach has certain dependencies for systems, data and/or personnel and may not be feasible in all contexts. Consequently, a healthcare provider or payer would have to carefully consider these approaches to find the most appropriate solution. For some agencies, a multimodal approach would seem to be an appropriate solution.7
According to the US Centers for Disease Control and Prevention (CDC), nonadherence has been identified as the cause of approximately USD 100 billion to USD 300 billion worth of avoidable annual healthcare spending.8,,9 This potential wastefulness can be avoided by carefully investing in strategies, with the availability of data analytics tools, that can assess and improve a patient’s medication adherence. Increased medication adherence can result in improvements in disease control and other aspects of self-management among members.
1Carls GS, Tuttle E, Tan R-D, et al. Understanding the gap between efficacy in randomized controlled trials and effectiveness in real-world use of GLP-1 RA and DPP-4 therapies in patients with type 2 diabetes. Diabetes Care 2017;40:1469–1478. Retrieved June 14, 2021, from https://care.diabetesjournals.org/content/40/11/1469.long.
2National Committee for Quality Assurance. Comprehensive diabetes care. Retrieved June 14, 2021, from https://www.ncqa.org/hedis/measures/comprehensive-diabetes-care/.
3Polonsky WH, Henry RR. Poor medication adherence in type 2 diabetes: recognizing the scope of the problem and its key contributors. Dovepress Open. 2016 ;1299–1307. Retrieved June 14, 2021, from https://www.dovepress.com/poor-medication-adherence-in-type-2-diabetes-recognizing-the-scope-of--peer-reviewed-fulltext-article-PPA.
4Lam WY, Fresco P. Medication Adherence Measures: An Overview. Biomed Res Int. 2015;2015:217047. Retrieved June 14, 2021, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4619779/.
5Kardas et al . The Need to Develop Standard Measures of Patient Adherence for Big Data: Viewpoint. J Med Internet Res 2020;22(8):e18150. Retrieved June 14, 2021, from https://www.jmir.org/2020/8/e18150/PDF.
6Hutchins DS, Zeber JE, Roberts CS, Williams AF, Manias E, Peterson AM, IPSOR Medication Adherence Persistence Special Interest Group. Initial Medication Adherence-Review and Recommendations for Good Practices in Outcomes Research: An ISPOR Medication Adherence and Persistence Special Interest Group Report. Value Health 2015 Jul;18(5):690-699. Retrieved June 14, 2021, from https://www.valueinhealthjournal.com/action/showPdf?pii=S1098-3015%2815%2901844-6.
7Anghel LA, Farcas AM, Oprean RN. An overview of the common methods used to measure treatment adherence. Med Pharm Rep. 2019;92(2):117-122. doi:10.15386/mpr-1201. Retrieved June 14, 2021, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6510353/.
8Cutler RL, Fernandez-Llimos F, Frommer M, Benrimoj C, Garcia-Cardenas V. Economic impact of medication non-adherence by disease groups: a systematic review. BMJ Open. 2018;8(1):e016982. Retrieved June 14, 2021, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5780689/.
9CDC Grand Rounds: Improving Medication Adherence for Chronic Disease Management — Innovations and Opportunities.2017. Retrieved June 14, 2021, from https://onlinelibrary.wiley.com/doi/full/10.1111/ajt.14649.
You can't manage what you can't measure: Medication adherence in chronic disease management
We summarise different methodologies adopted by healthcare stakeholders to measure medication adherence, and offer some advancements in standardizing this critical metric.