Revisions of Medicare reimbursement policy for ambulatory blood pressure monitoring and the role of qualitative analysis

Ambulatory blood pressure monitoring (ABPM) is endorsed by the American College of Cardiology/ American Heart Association Task Force (section 4.3)1 as a reasonable screening tool to detect white coat (elevated office blood pressure, ie, within normal ranges outside the office) and/or masked hypertension (elevated out-of-office blood pressure, ie, within normal ranges inside the office) among untreated patients, as well as white coat and/or masked hypertension effects among treated patients. These endorsements are supported by findings of the superiority of ABPM compared with clinic and home blood pressure assessments regarding both cost-effectiveness2-4 and prognostic value.5-7 Despite the strong evidence base and scientific community endorsement, ABPM was only Medicare-reimbursed for suspected white coat hypertension among untreated patients. Accordingly, in May 2018, the American Heart Association and American Medical Association sent a request to the Centers for Medicare and Medicaid Services (CMS) to revise Medicare reimbursement to reflect these recommendations.8 In July 2019, CMS announced an update for ABPM coverage to include both white coat and masked hypertension effects, among both untreated and treated patients.9 It also made several other revisions: downward adjustment of hypertension thresholds to match updated guidelines,1 inclusion of patients with target organ damage,10 inclusion of ABPM devices storing readings on Bluetooth cloud instead of internally, and updated device validation recommendations.11, 12 The CMS memo cited several lines of evidence favoring revision of ABPM coverage: (a) a review of existing task force guidelines,7 (b) an updated review of current literature evidence, and (c) an evaluation of public comments. CMS collected public comments by opening a web-based message board13 for 30 days (October 2018) to help shape its revision proposal14 and another 30 days (April 2019) for public reactions to the proposal to inform its final revision.9 Thus, it is timely that this issue of the Journal of Clinical Hypertension includes a qualitative analysis of the public comments from October 2018.15 Dixon et al analyzed the 101 public comments received by CMS during this first comment period (ie, the “brainstorm” to help generate the draft, but not the “reaction” to the draft) and reported emergent content themes. They used inductive analysis, in which the investigators generate themes from analyzing the data and then group data under these themes.16 This contrasts with deductive analysis, where the investigators establish themes for grouping the data using a source such as the published literature, before examining the data. A major strength of the inductive approach is that it capitalizes upon the open public nature of the comments without discriminating based on their consistency with “established” knowledge such as evidence-based position stands. This allows the capture of information that is based on non-literature sources: for example, clinical experiences of individual practitioners or patients (rather than standard decision tree algorithms) and operation logistics and finances of individual clinics. The manuscript is restricted in scope to monologue by public commenters with limited attempt to create exchanges to facilitate richer responses through recognized strategies of the Consolidated Criteria for Reporting Qualitative Studies17 such as semi-structured interviews or focus groups. The authors accordingly acknowledge that most items of the criteria do not apply to their manuscript. The open accessibility of the web-based message board13 could potentially have mitigated this limitation since commenters could read and react to each other's responses; however, the authors state in their discussion that they could not distinguish this possibility from commenters independently arriving at the same concerns. Also, the comment board did not include any features to encourage interaction such as thread grouping, reply features, or encouragement to post multiple times (eg, to reply after receiving a reply). Another hindrance to interactive dialogue was the restriction of scope to the first comment period, thus excluding the second period which invited reactions to CMS's proposed decision. In contrast to Dixon et al, CMS9 does not disclose whether they evaluated the comments using an inductive (as in Dixon et al) or a deductive approach. However, it is likely CMS lacked the sensitivity to capture non-literature information that was afforded by Dixon et al's inductive approach because CMS stated a preference for including comments that used supporting literature citations and noted that it considered results of individual physicians and patients to be less helpful. The resulting influence on the results could have been substantial since only 35% of comments used citations.15 On the other hand, a strength held by the CMS analysis is that their scope included both commenting periods. Furthermore, they included not only themes from the comments but also a response to each one justifying its incorporation or dismissal from the final revision decision. Overall, this created a more enriched dialogue than Dixon et al. Despite these methodological differences, there was some convergence of the findings between Dixon et al and CMS. First, both noted that virtually all comments supported the need for a revised reimbursement policy for ABPM. Second, both identified comments requesting expansion of reimbursement to cover white coat and masked hypertension effects, among both untreated and treated patients, and among both adults and children. Third, both identified requests that reimbursement allows ABPM devices storing readings on Bluetooth cloud instead of internally. CMS heeded all the above themes in the new reimbursement guidelines. This convergence supports the validity of both methods, as well as the hearing that CMS gave public comments. On the other hand, the approach by Dixon et al also highlighted several themes not discussed in the CMS memo. First, some comments pertained to the current amount of reimbursement (average $52) being well below the break-even point considering the equipment, software, and personnel required to administer ABPM correctly. Some physicians reported circumstances in which ABPM was medically necessary but financially unfeasible for patients, in which instances they delivered it at a financial loss to their practice. Second, some comments flagged logistical issues that interfere with ABPM access: required referrals to specialists, overburdened physicians, and lack of coverage for non-physicians such as pharmacists to prescribe it. The omission of these themes potentially reveals limitations of the CMS methodological approach. In addition to themes in Dixon et al not addressed by CMS, there were instances of CMS sub-optimally addressing commenters’ concerns. For example, Dixon et al identified a theme centering around the recommendation that nocturnal hypertension should be an indication for ABPM. CMS responded to this comment by noting that ABPM can detect nocturnal hypertension. This is an oversimplification. Isolated nocturnal hypertension (more commonly described as non-dipping hypertension18) can occur without the elevated daytime readings that are required to meet CMS's criteria for masked hypertension. The matter under discussion is the indication for performing ABPM, not result interpretation, and suspicion of isolated nocturnal hypertension is not currently on the list of reimbursable indications for ABPM. If it were added to the list, it could potentially help providers care for patients at high risk of isolated nocturnal hypertension such as those with obstructive sleep apnea19 and/or diabetic neuropathy.20 In sum, Dixon et al's inductive methodology highlights important elements of public opinion that would not be apparent from reading the CMS Memo. Overall, this manuscript and decision memo highlight some general considerations about the use of a public comment process as part of policy briefing. First, greater transparency would be helpful regarding the methodology for the analysis of comments. CMS's methodology may be valid, but its results only partially converged with those of Dixon et al. It would be prudent for us as readers to trace the source of the discrepancies, but CMS does not provide us with the information required to do so. Second, policymaking teams may consider enlisting direct collaboration with the research personnel needed to uphold a transparent and constant methodology. In the present situation, Dixon et al analyzed comments from the first comment period, but while their analysis was under preparation and peer review CMS meanwhile analyzed the same set of comments, issued its proposed policy decision memo, conducted a second comment period for reactions to the policy, analyzed these comments and issued a final policy decision memo. If Dixon et al could have worked directly with the CMS team, then they could have performed both analysis steps and moreover communicated their results directly within the team, and CMS could have used their identified themes in its point-by-point replies in the decision memos. Third, CMS could consider selecting a research methodology that utilizes inductive (as in Dixon et al) rather than deductive analysis so that information based on non-literature sources (eg, clinical experiences of individual practitioners) can be accommodated without compromising scientific integrity. CMS’s public comment process strengthens their policymaking, so I submit these suggestions to improve it further. None.