Effects of the 2016 CDC opioid prescribing guideline: a scoping review Authors: Katherine E Warren, Sooyeon Kho, Carol Mita, Molly F Franke, Shelly F Greenfield, Scott G Weiner Published: 2026 ABSTRACT Abstract AbstractObjectivesIn March 2016, the Centers for Disease Control and Prevention (CDC) released the CDC Guideline for Prescribing Opioids for Chronic Pain, a set of voluntary recommendations for initiating and managing opioid treatment in the ambulatory setting. This scoping review examined guideline effects on patients, providers and health systems.MethodsA scoping review was conducted with a preregistered protocol. Comprehensive searches of PubMed, Embase and Cumulative Index of Nursing and Allied Health Literature were conducted in April 2025. Reports published between 2016 and 2025 that explored the effects of the CDC guideline were included. No restrictions on language or country of study origin were applied, though all retrieved reports were published in the USA and in English. Two authors independently screened titles, abstracts and full-text reports. Data were extracted by healthcare setting, study aims and design, sample size, study population, participant characteristics and study findings and outcomes. Reports were characterised as empirical studies that evaluated guideline effects or implementation studies that assessed uptake. Study findings were presented descriptively and by evidence maps.ResultsNinety-four studies met the inclusion criteria: 75 empirical studies and 19 implementation studies. Eighty-eight per cent measured changes in opioid prescribing; all but one found significant reductions in at least one prescribing measure, often among people receiving ≥50–90 morphine mg equivalents per day. Effects occurred across specialties and populations, including groups not targeted by the guideline. Studies found increased rates of tapering, with mixed findings on opioid-benzodiazepine coprescription. Legal analyses showed widespread policy adoption at the state level. Implementation studies described expanded risk-mitigation strategies, sometimes beyond guideline text. Few studies reported patient-centred outcomes, participant race or ethnicity or equity measures.ConclusionsThis voluntary federal guideline had significant intended and unintended effects. The guideline was associated with reductions in opioid prescribing among groups targeted and not targeted by its design, with limited evidence on patient outcomes. Future work should prioritise equity-focused patient outcomes to inform implementation of the 2022 CDC guideline.Keywords: Chronic Pain, PAIN MANAGEMENT, Health policy STRENGTHS AND LIMITATIONS OF THIS STUDY This scoping review used a comprehensive, structured search strategy across multiple databases and supplemented with grey literature. This scoping review used a comprehensive, structured search strategy across multiple databases and supplemented with grey literature. The review explored the effects of the 2016 Centers for Disease Control and Prevention guideline, a substantial and growing body of literature with impacts across patients, providers, public health and health systems. The review explored the effects of the 2016 Centers for Disease Control and Prevention guideline, a substantial and growing body of literature with impacts across patients, providers, public health and health systems. The review followed established methodological guidance, including adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews checklist, with a preregistered study protocol to enhance transparency and reproducibility. The review followed established methodological guidance, including adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews checklist, with a preregistered study protocol to enhance transparency and reproducibility. Given the heterogeneity in study designs, populations and outcome measures, the findings were limited to a descriptive and thematic synthesis. Given the heterogeneity in study designs, populations and outcome measures, the findings were limited to a descriptive and thematic synthesis. No formal appraisal of study quality or publication bias was conducted. No formal appraisal of study quality or publication bias was conducted. Introduction Amidst high levels of chronic pain,1 the USA has experienced a concurrent drug overdose epidemic that led to more than 100 000 overdose deaths in 2023 alone.2 In response to the dramatic rise in opioid-related overdose deaths, governmental agencies and professional organisations implemented changes intended to curb inappropriate opioid prescribing. Based on expert guidance and a systematic review of the evidence, the Centers for Disease Control and Prevention (CDC) released the CDC Guideline for Prescribing Opioids for Chronic Pain – 3United States, 2016 (herein called the CDC guideline) in March 2016.3 Between 2015 and 2019, opioid prescriptions fell by more than 26% in the USA.4 A significant body of literature has emerged to examine the role of the CDC guideline in these prescribing trends and its effects on patients, providers and healthcare systems. This scoping review examines findings from this emerging body of research and its limitations and next steps. The voluntary guideline explicitly focused on opioid prescribing for chronic pain, which it defined as pain lasting longer than 3 months.3 The report grouped recommendations into three primary groups: (1) determining when to start or continue opioids; (2) selecting opioid medication, dosage and duration and (3) addressing the risks and harms of opioid use. When considering the initiation of chronic pain treatment, the CDC guideline recommended maximising nonpharmacologic and non-opioid therapy. For opioid treatment initiation, the guideline recommended immediate—rather than extended—release formulations at the lowest effective dose for the shortest duration necessary (usually 3 days or less and up to 7 days) with regular monitoring of pain, function and adverse effects, including urine drug tests and prescription drug monitoring programmes (PDMPs) to detect inappropriate opioid use or substance use disorders. The CDC guideline recommended caution around increasing the daily dose beyond 50 morphine milligram equivalents per day (MME/day) and recommended avoiding or carefully justifying the decision to increase to 90 or more MME/day. To minimise the risks of overdose, the CDC guideline recommended against concomitant use with benzodiazepines and advised naloxone coprescription for overdose reversal. If the harms of opioid therapy were deemed greater than the benefits, the CDC guideline recommended gradual tapering or discontinuation of opioids.3 A preliminary search for existing scoping or systematic reviews directly related to the effects of the CDC guideline was conducted in PubMed, PROSPERO, JBI Evidence Synthesis, the Cochrane Database of Systematic Reviews, Epistemonikos, and the Center for Open Science registries between 11 April and 15 April 2022. No existing scoping or systematic reviews pertaining directly to the CDC guideline were found at the time of initiating this study. Therefore, this scoping review aimed to investigate the literature that evaluated the CDC guideline’s effects on patients, providers, healthcare systems and health policy and to identify avenues for future research. Methods Details on the scoping review methods and review questions can be found in the published and preregistered protocol (Open Science Framework #N8RG7)5 (see online supplemental appendix C). The review followed the Joanna Briggs Institute methodology for conducting scoping reviews by identifying target study population, concept and context.6 The primary research question was to investigate published studies that seek to directly evaluate the effects of the CDC Guideline. There were five subquestions that examined which studies investigate the effects of the CDC Guideline on (1) provider opioid prescribing and practice, (2) the experience of patients with chronic pain, (3) health behaviours and outcomes and (4) policy, including state and federal laws and regulations, and (5) the geographical and clinical contexts in which those studies were conducted. Ethical review was not required as no human subjects were the object of study. Search strategy Publications discussing the CDC guideline in association with the prescription and use of analgesic opioids were identified by searching PubMed/Medline (National Library of Medicine, National Center for Biotechnology Information), Embase (Elsevier, embase.com), and Cumulative Index of Nursing and Allied Health Literature (CINAHL Complete, EBSCOhost), most recently on 3 April 2025. Controlled subject vocabulary terms (ie, Medical Subject Headings, Emtree and CINAHL subject headings) were included when available and appropriate. All available publication types were included. No language limits were applied, though the databases searched contained primarily English-language reports. A date limit of 2016–2025 was applied to correspond with the period following CDC guideline publication. The searches were developed and carried out by a librarian. The exact search terms used for each of the databases are provided in the supplemental documents (see online supplemental appendix A). In addition, hand searching was conducted on reference lists from selected articles. Author contact was used for known but unavailable sources of evidence. Selection of evidence Using Covidence systematic review software (Veritas Health Innovation Ltd., Melbourne, Australia), two reviewers independently assessed all identified publications for eligibility first by title and abstract, followed by full-text examination. Publications were included if they directly examined the effects of the CDC guideline and were published in or after March 2016. Further clarifications were made to the eligibility criteria following a preliminary review of 25 randomly selected articles. Articles were excluded if they were not specific to the CDC guideline, such as those that evaluated a related state law or reductions in opioid prescribing without explicit study of the CDC guideline (‘Not specific to guideline’ in figure 1). Others were excluded if they included the wrong study design to answer the research questions above, such as those that examined CDC guideline adherence at a single point in time and could not draw conclusions about the effects of CDC guideline publication (‘Wrong study design’ in figure 1). Qualitative studies were exempted from requiring a before-after analysis. Articles were also excluded if they sought to evaluate the clinical validity of a clinical recommendation within the CDC guideline, such as the 50-morphine mg equivalent recommended dose maximum (‘Evaluation of guideline’ in figure 1). For implementation studies, articles that reported on programmes that implemented only a portion of the CDC guideline (eg, naloxone distribution) were excluded. Conference abstracts were included per the JBI methodology to map the breadth and quality of the evidence. If conference abstracts had subsequent full-text journal articles published, only journal articles were reported (‘Abstract with subsequent article’ in figure 1). For opinion or expository pieces, articles were excluded for having no editorial content if they only reported on the CDC guideline publication rather than presenting editorial commentary (‘No editorial content’ in figure 1). Screening proceeded when 75% or greater agreement was achieved between reviewers. Discrepancies between reviewers were resolved via consensus. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Flow Diagram. CINAHL, Cumulative Index of Nursing and Allied Health Literature. Data extraction Study characteristics were extracted by two researchers using a data extraction form in Covidence, which included study author, year and origin; study objective(s); study design; study participant characteristics and number; study population and context (eg, inpatient, outpatient, claims database, hospital or healthcare system); inclusion and exclusion criteria; study findings and outcome type (eg, opioid prescribing, pain or patient functioning) and reported funding and conflicts of interest. For the study population, studies conducted in the emergency department were classified as outpatient (specialist); studies conducted within a state health insurance exchange were classified as a hospital or healthcare system. Studies were divided between empirical studies, which examined the effects of the CDC guideline within a system or population, and implementation studies, which reported on programmes that implemented the CDC guideline with data obtained before and after the intervention. The extracted data were imported into Microsoft Excel for analysis. Editorials were identified but not extracted. Analysis and presentation of results Results were presented as a descriptive summary of the data extracted from included sources during the review process. Simple frequency counts of concepts, populations, characteristics and other data fields were generated and reported in the data charting process. Study findings were summarised, and overarching descriptive themes were identified for both empirical and implementation studies. Evidence maps of findings were generated to examine interactions between the strength of the study design, patient population and study outcomes reported. Although the protocol initially called for study quality evaluation by Critical Appraisal Skills Programme checklist, after consultation with our expert librarian, it was determined that this is not commonly done for a scoping review and thus was not performed. Reporting was guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews checklist.7 Results Study characteristics Figure 1 describes the article identification process. The literature searches identified 2922 publications, of which 2188 titles and abstracts were screened after removing duplicates (n=734). Following initial screening, 305 full-text reports were sought for retrieval, and one was unable to be retrieved. 174 articles were excluded using the criteria outlined in the Methods section. Hand searching the reference lists of selected publications identified seven additional reports. There were 43 editorial articles identified, which were not extracted. After the removal of editorials, 94 reports were included and extracted using a standardised form. Of these, 75 were classified as empirical studies and 19 as implementation studies. While no country of origin or language limits were applied, all identified reports were published in the USA and in English. See table 1 for a summary of included study reports. Table 1Study characteristics of included study reportsEmpirical (n=75)Implementation (n=19)n (%)n (%)Study settingUSA75 (100%)19 (100%)PublicationConference abstract15 (20%)6 (32%)Dissertation0 (0%)2 (11%)Peer-reviewed journal article52 (69%)10 (53%)Research letter or brief report8 (11%)1 (5%)DesignCohort study62 (83%)14 (74%) Interrupted time series19 (25%)1 (5%) Before-after28 (37%)13 (68%) Descriptive temporal trend15 (20%)0 (0%)Qualitative12 (16%)5 (26%)Systematic review1 (1%)0 (0%)Data reportedAge reported*37 (54%)8 (42%)Sex/gender reported*43 (62%)7 (37%)Race/ethnicity reported*27 (39%)5 (26%)PopulationInpatient6 (8%)2 (11%)Outpatient17 (23%)15 (79%)Primary care5 (7%)13 (68%)Medicaid2 (3%)0 (0%)Medicare14 (19%)0 (0%)Veterans Affairs1 (1%)1 (5%)Commercial insurance16 (21%)0 (0%)Claims or administrative database37 (49%)0 (0%)Hospital or healthcare system18 (24%)18 (95%)Other12 (16%)0 (0%)*Total out of 69 studies where age, sex/gender or race/ethnicity were relevant. Study characteristics of included study reports Total out of 69 studies where age, sex/gender or race/ethnicity were relevant. Study design, location and population All identified studies were published in the USA. Among the empirical studies, the most common study designs were cohort studies (n=62, 83%). These studies were further subdivided by study design, with most using a before-after design (n=28, 37%), in which data were examined before and after CDC guideline publication or implementation. Others used an interrupted time series methodology (n=19, 25%). Finally, some examined data over multiple time points over a period in which the CDC guideline was implemented, using a descriptive temporal trend (n=15, 20%) analysis. For this final subcategory of cohort study, those that examined uncontrolled cohorts over time had to specify the assessment of the effects of the CDC guideline in their study objectives to be included. Studies were also classified by population, including site of the study (inpatient, outpatient, primary care) and primary insurer (Medicare, Medicaid, Veterans Affairs or commercial insurance). Studies were further identified as drawing primarily from claims or administrative databases, such as the IQVIA data warehouse, or from a single hospital or health system, such as an academic medical centre. Studies could fit into multiple population categories. Most were conducted using claims or administrative databases (n=37, 49%) or within a hospital or healthcare system (n=18, 24%). Most studies, therefore, examined trends at a regional, national or larger health-system level. See online supplemental table 1 for individual empirical study designs, aims and findings. The review identified 19 implementation studies, in which a quality improvement or educational programme was based on the CDC guideline and implemented within a clinical context with data collected before and after implementation. Given the nature of these interventions, they had smaller sample sizes with less robust study designs and shorter follow-up intervals (<12 months). Half were peer-reviewed articles; the remaining were published as conference abstracts, dissertations or theses or research letters. See online supplemental table 2 for included implementation studies. Citations for empirical and implementation studies are included in online supplemental appendix B. Opioid prescribing, tapering and management Most empirical (n=67, 89%) and implementation (n=17, 89%) studies examined opioid prescribing practices as a primary or secondary outcome (see figure 2). Among these, all but one study found a statistically significant reduction in at least one opioid prescribing measure across diverse patient populations and practice settings. One systematic review, reported by conference abstract, identified 23 studies examining the effects of the CDC guideline; all but one found that the CDC guideline was associated with reductions in opioid prescribing by at least one prescribing metric.8 Among the empirical studies, effect sizes ranged broadly. Given the wide-ranging study designs and variable definitions employed by these studies, findings were not pooled. Common outcome variable definitions for prescribing metrics included prescribing rates, duration and total MME. Several studies found greater declines for high-dose prescribing, defined usually as greater than 50 to 90 MME/day. For studies employing interrupted time series designs, most found accelerated slope declines after 2016 with the publication of the CDC guideline. Outcomes assessed in studies examining the CDC Guideline for Prescribing Opioids for Chronic Pain — 3United States, 2016. Ten (13%) empirical and two (11%) implementation studies reported on opioid tapering or deprescription. Among quantitative empirical studies that examined opioid tapering, including among people with long-term opioid therapy,9 all found increased rates of tapering or deprescription with mixed findings around whether sex or race impacted these likelihoods. One legal review found that more than half of US states enacted laws related to opioid tapering, nearly half within 3 years of the CDC guideline publication.10 34 (45%) empirical and 9 (47%) implementation studies examined changes in non-opioid prescribing, including benzodiazepines, gabapentin, naloxone and non-steroidal anti-inflammatory drugs. Multiple empirical studies found increased rates of gabapentin prescribing (n=5) as well as naloxone distribution or coprescription (n=3) after the CDC guideline. Findings regarding opioid-benzodiazepine coprescription were mixed. Most empirical studies reported at least one measure of decreased coprescription (n=13), while others found stable (n=5) or increased (n=1) coprescription. Similar mixed findings were described among implementation studies (n=2). Many implementation studies found increased use of tools designed to recognise and reduce the associated risks of opioid use, such as naloxone distribution, urine drug screening and PDMP use. Multiple studies measured increased uptake of opioid treatment agreements, despite the absence of a formal recommendation around their use within the CDC guideline. Studies that evaluated provider knowledge, attitudes or confidence following guideline-based educational interventions consistently reported improvements in clinician self-reported preparedness to manage opioid prescribing. Specialty-specific findings Thirty-one empirical studies (41%) reported on specialty-specific effects. The most common specialties included were oncology (n=5, 7%), surgery (n=3, 4%), emergency medicine (n=3, 4%) and primary care (n=3, 4%). All empirical studies that examined postsurgical opioid prescribing found decreases in at least one measure of opioid prescribing at discharge, including the daily and total dose as well as the duration of postoperative opioid prescriptions. Multiple hospitals and healthcare systems reported decreased rates of emergency department opioid prescribing. Numerous studies examined variation across prescriber specialties and found reductions in primary care as well as proceduralists and specialists (n=7, 9%). Among the implementation studies, most took place in primary-care practice settings (n=13, 68%). Secondary consequences among special patient populations Significant research found opioid prescribing reductions among populations not specifically targeted by the CDC guideline. This included people with cancer despite language in the CDC guideline that exempted their recommendations for people undergoing active cancer treatment (n=4, 5%). Multiple studies also found reduced opioid prescribing among people with sickle cell disease (n=3, 4%). Studies found increased barriers to opioids for people with sickle cell disease without compensatory access to non-opioid therapy11 as well as data suggestive of higher rates of vaso-occlusive crisis-related emergency department visits or hospitalisations.12 Two studies found reductions in opioid prescribing in palliative and end-of-life care, another population of patients exempted from CDC guideline recommendations. Others found significantly decreased opioid prescribing for acute pain (n=3, 4%), which was not a primary focus of the CDC guideline. Many of these studies discussed the unintended consequences of the CDC guideline, cited by guideline authors ahead of the 2022 guideline revision.13,15 Patient-related outcomes and demographic characteristics Few empirical or implementation studies reported effects on patient-related outcomes, including pain and patient functioning (n=19, 20%) or the patient-provider relationship (n=16, 17%). Of the 29 (31%) studies that examined either patient-related outcome, most studies used qualitative designs (n=11, 38%), and 13 (45%) were implementation studies. Many focused only on uptake of risk-mitigation tools in the patient-provider relationship, such as urine drug screening or opioid treatment agreements. Some examined primarily negative patient-related outcomes among patient groups not targeted by the CDC guideline, such as increases in hospitalisations or emergency department visits for patients with sickle cell disease.12 In one implementation study where patient satisfaction levels were measured as the CDC guideline was implemented, there was an improvement in satisfaction scores with a concomitant reduction in opioid prescribing.16 Qualitative studies often cited increased tensions between patient and provider and frustration from people with chronic pain. One qualitative study found that while patients emphasised the pain-relieving effects of chronic opioid therapy, providers often focused on their risks and limited long-term efficacy when discussing the CDC guideline.17 While a significant number of empirical studies used Medicare (n=14, 19%) or commercial insurance (n=16, 21%) data, only two (3%) drew from patients covered by Medicaid. Across both empirical and implementation studies, approximately half gave details on participant age (n=45, 51%) and sex and/or gender (n=50, 57%). One-third reported participant race and/or ethnicity (n=32, 36%). Only one conference abstract focused directly on the effects of the CDC guideline on racial disparities in opioid treatment despite widespread evidence of differences in opioid prescribing by patient race.18 The effects of the CDC guideline on patient-related outcomes as well as possible effect modification by patient identity were therefore underexplored. Legal and regulatory effects Four studies examined the legal and regulatory effects of the CDC guideline, including a systematic review of state and federal opioid policy immediately after its passage19 and relevant changes to laws around prescribing opioids for cancer20 and acute pain.21 All of them found significant statutory changes, with more than half of the states imposing limits on opioid prescribing, many in the 3 years following the CDC guideline release. Evidence mapping of study outcomes by design and population Study outcomes related to prescribing, including opioids, non-opioid alternatives and tapering, were more likely to use a more rigorous study design, such as interrupted-time series analysis (see figure 3). Patient-related outcomes such as pain or functioning and patient-provider relationship were more likely to be reported in qualitative or uncontrolled before-after or descriptive temporal cohort studies. Additionally, while there were overall fewer studies that focused on Medicaid, Medicare and Veterans Affairs populations, there was only one study that reported on patient-related outcomes in these groups, despite structural factors that might predispose each to a disparate or disproportionate effect on patient- and equity-related outcomes (see figure 4). Number of studies reporting outcome domains by study design. Number of studies reporting outcome domains by study population. Discussion This scoping review outlines a large body of evidence that sought to explore the effects of the CDC guideline publication across diverse patient and provider populations as well as outcomes. Most studies examined changes to opioid and related non-opioid prescribing. The CDC guideline publication was associated with substantial decreases in opioid prescribing among groups targeted and not targeted by the guideline. Studies found these effects were most pronounced among those receiving ≥50–90 MME/day, the thresholds that drew heightened scrutiny ahead of the 2022 guideline revision.14 While widespread decreases in opioid prescribing were documented, the literature remains more limited around the effects on patient-reported outcomes and health-related behaviours, including patient function and pain levels, healthcare engagement, non-prescribed opioid use and overdose. Studies also infrequently reported results by race or ethnicity, limiting overall ability to make inferences about health effects and health equity. Intended and unintended consequences of a voluntary federal guideline Multiple studies demonstrated the incorporation of voluntary guideline recommendations into clinical and public health practice as well as state and federal law, in ways both consistent with and at odds with CDC guideline recommendations. Studies consistently found decreases in average MME and shorter durations of opioid prescribing. Furthermore, there was widespread uptake of risk-mitigation tools, including naloxone distribution, urine drug screening and use of PDMPs. Many sites adopted opioid treatment agreements, even though the guideline did not explicitly recommend this practice, illustrating ‘implementation drift’. Unintended consequences—including rapid or mandatory tapering and restricted access among people with cancer or sickle cell disease—were also described. The CDC’s 2022 revision explicitly shifted away from dosage thresholds and emphasised flexibility and shared decision-making to address such harms.14 Strengths and limitations The scoping review identified a substantial body of literature using a comprehensive, librarian-assisted search strategy of published and grey literature with independent report screening and extraction to improve methodological rigour. The findings suggest important themes as well as next directions that will be important to consider as studies of the revised 2022 guideline are conceived. This review is unable to identify causal effects of the CDC guideline due to the limitations of many study designs (eg, before-after designs without control groups), concurrent policy changes and growing public awareness of the opioid overdose crisis. Outcomes were dominated by prescribing metrics, with limited patient-reported outcomes and behaviours. Heterogeneity in definitions of population, exposure and outcomes limited synthesis. The scoping review methodology was intended to demonstrate the breadth of reported evidence. In keeping with this design, no formal quality appraisal was conducted, which further limits the ability to report causal effects. Inclusion of brief reports and conference abstracts without quality appraisal introduced methodological challenges, including publication, reporting and selection biases, as well as challenges to internal validity. Implications for public health, policy and practice This scoping review suggests the myriad impacts of a voluntary federal guideline. The CDC guideline was associated with intended reductions to high-risk opioid prescribing as well as unintended consequences among some patient groups. It also highlights the importance of including patient-centred and equity-focused outcomes when translating public health policy into practice. Given the breadth of the evidence, a systematic review to pool effect sizes and draw more robust conclusions about the evidence—particularly given the large body of evidence around changes to opioid prescribing—would be informative. As policymakers and healthcare providers prepare to evaluate the effects of the 2022 CDC guideline revision, studies could include greater use of quasi-experimental designs to isolate causal effects, link administrative claims data to electronic health records, mortality and treatment registry data to better capture patient-centred outcomes, disaggregate results by race, ethnicity, sex, geography, disability and clinical indication and use implementation science frameworks to balance fidelity and local adaptation and mitigate ‘implementation drift’. The results of this work could inform the implementation and evaluation of the revised 2022 CDC guideline and provide broader evidence around the consequences of federal policymaking. Supplementary material Acknowledgements An earlier version of this work was presented at the 2023 College on Problems of Drug Dependence (CPDD) annual meeting.22 Footnotes Funding: This work was supported by the National Institute on Drug Abuse (NIDA) grant number F30DA052116 (KW). Prepub: Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-109859). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Ethics approval: Ethical review was not required as no human subjects were the object of study. The Office of Regulatory Affairs and Research Compliance can be reached at orarc@hsph.harvard.edu. The Research Integrity Officer for the Harvard Longwood IRB can be reached at dwchristiani@hsph.harvard.edu. Patient and public involvement: Patients and/or the public were not involved in the design or conduct or reporting or dissemination plans of this research. Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information. References Supplementary Materials Data Availability Statement All data relevant to the study are included in the article or uploaded as supplementary information.