Discussion
In this comprehensive analysis of prevalent and incident AF targets obtained from medical and pharmacy claim databases based on a large US adult population, we show that continued reliance only on outpatient and inpatient claims alone greatly underestimates AF prevalence and incidence in the general population by over 100%. Second, we show the comorbidity profile in our AF patients, with the prevalence of multimorbidity being approx. 10-12% in those aged ≥65 years. Again, reliance only on outpatient and inpatient claims alone greatly underestimates multimorbidity in the AF population. Our proposed methodology can guide future analysis of the quality/cost of care for progressive medical conditions at the population level.
Our findings differ from the United States CDC(8) that found approximately 2% of people age < 65 years have AF, while about 9% of people age ≥ 65 years have AF. Instead, our results indicate that the working population aged < 65 years has a 0.9% prevalence and the Medicare population aged ≥65 had a 12.7% prevalence. The difference could possibly be attributed to computation methodologies and different time spans used (i.e., 2009-2014 for the CDC estimates and the 2016-2019 period in the current study). Piccini et al(2) reported a 7.3% prevalence in a study conducted on a 5% random sample of the Medicare population in the US, relying on the use of 2 outpatient visits or 1 inpatient visit within a 365 day period. The prevalence reported by Piccini et al(2) for 2007 is still lower than the overall 12.7% value reported in the present study. There was an increase in prevalence with the analyzed period (3.7%, 5.7%, 6.8%, and 7.3%, respectively for 1993, 1998, 2003 and 2007). In elderly Germans, Ohlmeier et al(9) reported an increased prevalence with the analysis period 7.7%, 9.4%, 9.8%, and 10.3%, respectively for the years of 2004 to 2007.
Our results emphasize the impact of joining pharmacy and medical claims in the calculation of AF prevalence and incidence rates. Using both types of databases, the overall prevalence rate was 2.2%, being 2.5% and 1.9%, respectively for males and females; the overall incidence rate was 3.3 per 1000 person-years (3.6 and 2.9 per 1000 person-years, respectively for males and females). Our overall AF prevalence was close to the 2.4% reported by Tu et al(4) in Canada using a similar approach, and the 2% and 2.13% prevalence reported for Europe(10) and Germany(11), respectively. It should be noted that the methodologies by Tu et al(4) is similar to that used in this study with the exceptions that the claim system in Canada is somewhat different from that used in the US. Indeed, our incidence rates were lower than those obtained in Germany(11) for the general population (males – 4.4 cases/1000 person-years; females – 3.9 cases/1000 person-years).
Although OAC use and office visits captured the highest number of targets in the identified AF population, they cannot be used as a proxy for identifying the prevalence and incidence of AF cases in the general population. OACs accounted for about half of AF prevalent cases, indicating that about 50% of the remaining population are not on anticoagulants, in line with prior studies (4)(12)(13). Office visits explained 55% and 43%, respectively of the AF prevalent and incident cases, with the remainder of AF population not making office visits.
Although males have higher AF prevalence than females, the present study suggest that older (65 years or older) females (normalized relative to younger population aged less than 65 years) have a 50% higher prevalence ratio than older males (normalized to relative to younger population less than 65 years). This is in line with the incidence rates obtained in the Medicare population for males and females (12.3 and 18.2 per 1000 person-years, respectively). Additionally, there is an increase in prevalence with age for the Medicare population from 65-74 to 75-84 to > 85 years.
Hospital admissions accounted for 4% or less of AF prevalent/incident targets. Furthermore, ER visits captured 14% or less of AF prevalent/incident cases. In a systematic review of validated methods for identifying AF using administrative data, Jensen et al(5) found that 10 of 16 studies used only inpatient data. Office observations captured 8.6% of prevalent AF targets in the Medicare population and 5.3% in the working population or Commercial plan. A similar trend was found for incident AF targets (11.3% and 9.4%, respectively in Medicare and Commercial cohorts). Thus, one cannot rely on ER visits, office observation and hospital admissions as the only sources to establish AF prevalence and incidence targets. There was an even smaller percentage of AF cases identified based on rhythm control (14%), in line with Tu et al(4) and LaPointe et al10. The latter study found among AF patients younger than 65 years that 16% received rhythm control medication and 84% had rate control drugs.
The practical implications of our analysis are worth highlighting. The number of prevalent/incident AF targets doubles when one integrates the use of medical and pharmacy claim databases in comparison to the utilization of outpatient and inpatient healthcare services from medical claim databases. We also show that the multimorbidity is common amongst AF patients (affecting approximately 1 in 10 patients), which puts such patients at risk of cardiovascular and non-cardiovascular mortality and morbidity (14). In particular, for incident AF cases, about 40% of the older population and 30% of the overall population have 4 or more co-morbidities; for prevalent cases, these proportions are 30% and 20%, respectively (fig 2 e and f). Collectively, these results empathize that AF cohorts are at an increased risk of complications such as stroke and bleeding events; hence, there has been an increasing move towards a more holistic or integrated approach to AF management(15). Indeed, integrated care management of AF patients is associated with improved clinical outcomes(16-18).