Background/Purpose:

While guidelines and cost effectiveness analyses assume limited pathways for RA treatment, there is little known about what occurs in clinical practice. We sought to determine the most common RA treatment pathways over the past decade.

Methods:

Enrolled patients provided historical and at least 1 year of prospective treatment use in an open US observational cohort from 1998 through 2012. Treatment groups included 5 DMARDs – methotrexate (MTX), hydroxychlorine (HCQ), leflunomide (LEF), sulfasalazine (SUL), and other DMARDs (OTD) – and 6 biologics – etanercept (ETA), infliximab (INF), adalimumab (ADA), abatacept (ABA), rituximab (RIT), and other biologics (OTB). We implemented sequential data mining, a technique only recently used in the field of medicine, to uncover the most frequent treatment sequences at 1% support  (at least 1% of patients had the sequence). Frequent sequences can then be used to obtain association rules, a happens-afterstatement of the form A→B, given a “confidence” threshold for the probability of taking drug B after drug A. We constrained sequences to only describe consecutive and new therapeutic changes, which were presented as association rules expressed by support, confidence, lift (confidence of a rule divided by the support of the consequent) and coverage (total probability of transiting to all consequents and not switching). Results were further obtained within contrasted subsets defined by calendar year, age, and disease duration.

Results:  With 11 RA drug groups, there were over 7448 drug pathways of the 10,119 eligible RA patients with an average of 5.6 (4.5) years of observation/patient. After implementing the 1% support and confidence thresholds, only 24 pathways remained. Most participants (54.3%) did not switch therapy and were primarily on MTX (37.1%), HCQ (14.3%), and combination MTX + HCQ (9%). Among switchers, the modal treatment pathway was MTX then switch to another DMARD (9.3% of total) or an anti-TNF biologic (9.1%) (See Table). Before 2006, patients tended to switch from MTX to an anti-TNF biologic whereas during and after 2006, non anti-TNF agents appear. This was similar for high duration patients (>5 years), whereas non anti-TNF biologics never made the use-threshold for low duration patients. The transition from MTX to anti-TNF was more frequent in patients with longer disease duration than for early RA patients. ETA was a common choice for patients <65 years whereas INF was more common for patients >65 years.

Conclusion:

By using sequential data mining in a large dataset, we have estimates of the idiosyncratic and myriad drug pathways taken by US RA patients. While our results provide evidence of the recent ACR RA treatment guidelines in practice, only 38.1% of switching patients were in the modal pathway. Future health models of patient treatment need to account for this heterogeneity.

Table – Support of the Left hand side (LHS) of the rules (L(%)), Confidence (%) and Lift of the right hand side (RHS); probability of staying in the LHS (S(%)) and coverage (C(%))
	L	RHS (Support (%) (Confidence (%)) Lift)								S	C
LHS	(%)	ADA	ETA	HCQ	INF	LEF	MTX	OTD	SUL	(%)	(%)
{ETA}	18	1.5 (8.3) 1.0			1.6 (8.9) 0.6	1.2 (6.9) 0.4	1.1 (6.3) 0.1			56	87
{HCQ}	20					1.3 (6.6) 0.4	3.4 (17.3) 0.4	1.0 (5.1) 0.4	1.1 (5.5) 0.7	54	89
{LEF}	18		2.8 (15.5) 0.9		2.3 (12.8) 0.9					45	74
{MTX+HCQ}	9		1.2 (12.7) 0.7			1.5 (15.4) 0.9				49	77
{MTX}	46	1.5 (3.3) 0.4	3.8 (8.3) 0.5	2.8 (6.1) 0.3	4.2 (9.2) 0.6	4.7 (10.3) 0.6		1.6 (3.4) 0.3	1.3 (2.9) 0.4	53	97
{OTD}	13		1.0 (8.0) 0.4			1.5 (11.5) 0.6	1.2 (9.5) 0.2			50	79
{SUL}	8					1.0 (12.6) 0.7	1.1 (13.7) 0.3			43	69

---

« Back to 2013 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/which-are-the-most-common-disease-modifying-antirheumatic-and-biologic-treatment-pathways-for-rheumatoid-arthritis-patients/