PCP indicates primary care physician; PNSL, physician-sent message, no survey link; PSL, physician-sent message, survey link; SNSL, system-sent message, no survey link; SSL, system-sent message, survey link.
Data Sharing Statement
Customize your JAMA Network experience by selecting one or more topics from the list below.
Erdmann M, Edwards B, Adewumi MT. Effect of Electronic Portal Messaging With Embedded Asynchronous Care on Physician-Assisted Smoking Cessation Attempts: A Randomized Clinical Trial. JAMA Netw Open. 2022;5(2):e220348. doi:10.1001/jamanetworkopen.2022.0348
Can electronic outreach to patients via portal messaging combined with asynchronous care overcome physician- and patient-level barriers and increase physician-assisted attempts to quit smoking?
In this randomized clinical trial of 188 smokers, 9.5% who received electronic outreach messaging and interacted with their physician via a smoking cessation survey (asynchronous care) initiated a quit attempt, while only 4.3% who received electronic outreach messaging without asynchronous care initiated a quit attempt. This difference was not statistically significant.
These findings suggest that primary care practices can use electronic outreach technology to reach smokers and remotely connect them to resources by removing physician- and patient-level barriers through asynchronous care.
Despite the substantial health and financial burdens of smoking and the availability of effective, evidence-based interventions in primary care settings, few smokers and physicians use these strategies for smoking cessation.
To evaluate whether electronic outreach to smokers with embedded asynchronous care increases the number of quit attempts and explore the roles of the message sender (ie, primary care physician [PCP] vs health care system) and patient-related characteristics.
Design, Setting, and Participants
This quality improvement randomized clinical trial was designed to measure 2 factors: (1) electronic outreach messaging with and without a survey link to asynchronous care and (2) messaging by a personal PCP or health system. The study was conducted within the electronic health record and portal messaging platform of a large health system in the South Central US. Participants were adult patients 18 years or older who were designated as smokers in their electronic health records. Data were collected from January 13 to February 24, 2020, with participating PCPs surveyed in July 2020.
Portal messages encouraging a quit attempt and offering physician assistance were sent to smokers who were randomly selected and assigned to 1 of 4 conditions (message with or without embedded asynchronous care and PCP or system as sender). Half of the messages contained an invitation to come to clinics and the other half contained a link to access asynchronous care.
Main Outcomes and Measures
The primary outcome was electronic health record–documented quit attempts (1 indicates quit attempt; 0, no quit attempt), which were tracked 30 days after the electronic outreach. Secondary outcomes included physician perceptions of the electronic outreach intervention, using a 5-point scale to assess perceptions of workload, comfort with providing medication from survey information, and further interest in the program 6 months after the intervention.
A total of 188 participants (99 women [52.4%] and 89 men [47.3%]) with mean (SD) age of 55.2 (13.9) years were randomized to 1 of 4 conditions. Group 1 (n = 46) received a message from the PCP without a link to the survey; group 2 (n = 48) received a message from the PCP with a link to asynchronous care in the form of the survey. Group 3 (n = 47) received a message from the health system without a link to the survey; group 4 (n = 47) received a message from the health system with a link to the survey. No statistically significant difference in documented quite attempts was found among the 4 study groups. There was also no statistically significant difference in quit attempts between the group that received the asynchronous care survey link and the group that did not (odds ratio, 2.50 [95% CI, 0.72-8.72]). However, the quit attempt rate for those with asynchronous care offered (9 of 95 [9.5%]) was more than double the quit attempt rate for those with in-person care offered (4 of 93 [4.3%]).
Conclusions and Relevance
This quality improvement randomized clinical trial did not find a statistically significant difference in physician-assisted quit attempts among patients who received electronic with asynchronous care vs those who received outreach alone, regardless of whether the message source was a PCP or a health system. However, the program engaged patients in difficult-to-reach rural areas as well as younger patients.
ClinicalTrials.gov Identifier: NCT05172219
Cigarette smoking remains a major cause of preventable death and disability in the US.1-3 More than $300 billion is spent annually on diseases related to smoking and more than $156 billion is lost annually in productivity.4,5 Despite the substantial health and financial burdens of smoking and the availability of effective evidence-based interventions in primary care settings, few smokers and physicians use these strategies for smoking cessation. For example, although 70% of smokers report a desire to quit,6 only 4.7% use medications and/or evidence-based counseling7,8 that have been linked to successful quitting.7,9-11
Barriers to use of practice-based smoking cessation treatment include physician-level factors such as lack of time and workflow constraints and patient-level factors such as financial costs and time away from work.12,13 Although the use of technology could mitigate some of these barriers and engage smokers in cessation treatment, the evidence is quite limited.14-17 In particular, there are few examples of using technology to connect smokers to evidence-based, physician-driven strategies such as the use of medications and behavioral counseling for smoking cessation.18-22
Because previous research has indicated that physician-assisted quit attempts lead to the most successful quit attempts,11,23,24 a primary aim of this study was to determine whether asynchronous connection to primary care physicians (PCPs) via an electronic outreach message through the patient portal system would be associated with an increase in the number of quit attempts. A second primary aim was to determine whether the messages would be more likely to be viewed and quit attempts would increase if the sender was the patient’s PCP rather than the health system. We also assessed whether patient-related characteristics were associated with response to electronic outreach and use of asynchronous care to determine who benefited from the program. Using data from a large health system, we evaluated the effect of an asynchronous electronic outreach program on quit attempts by smokers.
The Oklahoma State University Center for Health Sciences Institutional Review Board granted a quality improvement exemption for nonhuman participant research for this study and a waiver of informed consent to use the present data for research purposes. The study has been registered retrospectively at the request of the editors. As a study intended to compare quality improvement interventions and investigate feasibility of asynchronous care for smoking cessation, it was not apparent at prospective review that it met the definition of a clinical trial. The study followed the Standards for Quality Improvement Reporting Excellence (SQUIRE) and Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines.
The study was conducted by the Center for Health Systems Innovation, Oklahoma State University, Tulsa, in partnership with a large health care system in the South Central US, where data from the electronic health record (EHR) were extracted. The health system provides approximately 400 000 patient encounters per year across 35 primary care clinics staffed by more than 140 PCPs, of whom 10 at 7 clinics participated. The asynchronous questionnaire was developed and linked to the health system’s EHR (Epic Systems Corporation) and the MyChart patient portal platform. Patient eligibility criteria included the following: adults 18 years or older who were designated as smokers in the EHR; patients who had at least 1 in-person visit with their PCP in the prior 12 months; and patients who had a patient portal account. Patients were excluded if they no longer saw the PCP, had a diagnosis for which the outreach program would be insensitive (eg, lung cancer), previously expressed not wanting smoking cessation counseling, or no longer smoked. To ensure equal representation across the 10 physicians, we randomly selected 20 patients under each physician to serve in the sample.
A total of 200 patients were selected based on our inclusion criteria and randomized to 1 of 4 test conditions (50 per group). Of the 200 patients, 12 were removed from the analyses because they had received smoking cessation treatment within 60 days before the study. Recent treatment status was overlooked as an initial exclusion criterion. Because our objective was to measure new quit attempts, these 12 patients were deemed to have already initiated a quit attempt that could not have been influenced by our intervention. This post hoc exclusion decreased the final sample size to 188 patients from 200. Of the 12 in the intervention groups, 7 were in the group without the survey link and 5 were in the group with the survey link.
This quality improvement randomized clinical trial was designed to evaluate smoker responsiveness to an asynchronous electronic outreach program on quit attempts by smokers identified in the large health system and to evaluate the differential associations of opening the message and quit attempts for PCP or health system as the message sender. Justifications for the design are provided in the trial protocol in Supplement 1. After the program, we obtained data on patient-related characteristics of interest to assess associations with quit attempts.
We compared the effects of 4 technology-based smoking cessation messaging strategies on quit attempts by smokers identified in the EHR. Specifically, we used a fully crossed between-participants 2 × 2 (link to survey on smoking cessation guide [yes or no] by message source [PCP or health system generated]) experimental design to which participants were randomized to 1 of 4 intervention groups using a computerized random number generator:
group 1 was sent the message from their PCP without a link to the survey;
group 2 was sent the message from their PCP with the link to the survey (which constitutes the asynchronous care);
group 3 was sent the message from the health system without a link to the survey; and
group 4 was sent the message from the health system with the link to the survey.
Electronic outreach constituted a message sent to patients via the patient portal. The message encouraged a quit attempt, offered physician assistance, advised the patient on medication use to control cravings, and offered counseling support in the form of a quitline. Patients in groups 1 and 3 were invited to schedule an in-person appointment with their PCP if they wanted help. Patients in groups 2 and 4 received the message with an embedded link to a tobacco cessation questionnaire to receive PCP assistance asynchronously. If patients clicked through to access the survey, they answered 17 questions that confirmed smoking status, reported safety information for medication selection, described prior quit attempts, and shared treatment preferences. The intervention was asynchronous because it did not involve direct, simultaneous physician-patient interaction (eg, in-person meeting, chat, video conference). After patient responses were submitted, they were stored and forwarded to their PCP. Physicians communicated the care plan and instructed patients via the portal message. If the plan included medication, a prescription was sent to the patient’s EHR-documented preferred pharmacy without an in-person visit. The intervention is summarized in the Appendix in Supplement 1.
Six months after the program, we surveyed participating PCPs to determine their perceptions of the electronic outreach intervention. We used a 5-point scale (1 indicates strongly disagree; 5, strongly agree) to assess perceptions of workload, comfort providing medication from survey information, and further interest in the program.
The primary outcome was EHR-documented quit attempts (1 indicates quit attempt; 0, no quit attempt), which were tracked 30 days after the electronic outreach. Patient data were deidentified by the health system and shared with employees at the Center for Health Systems Innovation who analyzed the data. We also obtained additional patient characteristics to investigate differences in the demographics of technology users, including rural vs urban residence, age, sex, distance from home to PCP, health status (eg, number of chronic diseases or conditions), and payer.
The study was performed from January 13 to February 24, 2020. Messages were sent to patients on January 13, and the asynchronous care was available until January 23, 2020. Quit attempts were tracked for 30 days until February 24, 2020. Participating physicians were surveyed in July 2020 to obtain their opinions about the program.
We used a 2 × 2 (link to survey by message source) experimental design; therefore, 2 factors created 4 treatment conditions. The portal messages were sent from either the PCP or the health system. The asynchronous care survey link was only visible (or not) once a patient opened the message. Thus, to determine whether there were differences in opening the message based on the message source, we estimated multivariable logistic regression with the 4 treatment groups as the independent variables and portal message opening as the dependent variable. Because the participants were nested within PCPs, we controlled for PCP as a dummy-coded variable. We also determined the odds ratios (ORs) for opening the message for the 2 main factors (message source and survey link).
To test whether intervention groups were associated with quit attempts, we conducted multivariable logistic regression with quit attempts as the dependent variable, controlling for PCP. We used χ2 tests to compare quit attempts between groups as well as associated ORs. To assess the main effects of each factor, multivariable logistic regression was used to determine whether either factor had a significant association. To more clearly determine whether the outreach message was associated with quit attempts, quit attempts among those who opened the message only were analyzed.
Multivariable logistic regression was used to determine whether patient-related characteristics were associated with opening messages and quit attempts. Specifically, we analyzed whether rural (vs urban) location, age, sex, distance from home to PCP, health status, and payer were associated with opening the email and quit attempts through asynchronous care. Given that the PCP could covary with distance and rural residence, analyses were conducted with and without controlling for the PCP.
Our criterion for evaluating statistical significance was 2-sided P ≤ .05. Because all data existed within the Epic Systems Corporation system, no data were missing. JMP Pro, version 15.0.0 (SAS Institute Inc), was used for statistical analyses.
A total of 188 participants (99 women [52.4%] and 89 men [47.3%]) with a mean (SD) age of 55.2 (13.9) years were included in the analysis. The CONSORT study flow diagram is presented in the Figure. There were 46 participants in group 1 (message from the PCP without a link to the survey), 48 in group 2 (message from the PCP with a link to asynchronous care in the form of the survey), 47 in group 3 (message from the health system without a link to the survey), and 47 in group 4 (message from the health system with a link to the survey). Multivariable logistic regression with the 4 treatment groups as the independent variables and portal message opening as the dependent variable controlling for PCP resulted in a nonsignificant overall χ212 = 11.93 (P = .45), providing evidence that there were equal odds of opening the message or making a quit attempt across the 4 groups. The OR for whether a patient opened the message from the system (vs from their PCP) was 0.83 (95% CI, 0.46-1.49); the OR for whether a patient opened the message if the message also contained the link to the survey (vs no survey link) was 0.81 (95% CI, 0.45-1.47).
Multivariable logistic regression with quit attempts as the dependent variable, controlling for PCP, resulted in χ23 = 2.52 (P = .47) for the intervention groups, which indicates that there were no statistically significant differences among the 4 groups and no statistically significant differences by PCP. Thus, we examined the main effects to determine whether either of the independent variables alone was associated with quit attempts. There was no statistically significant difference in quit attempts between the group that received the asynchronous care survey link vs the group that did not (OR, 2.50 [95% CI, 0.72-8.72]). As shown in Table 1, the quit attempt rate for the asynchronous care condition (9 of 95 [9.5%]) was more than double the quit attempt rate for the condition without the survey link (4 of 93 [4.3%]). Furthermore, when the analyses were limited to only those 106 patients (56.4% of the sample) who opened the message, outreach without the survey link generated a 7.3% quit attempt rate (4 of 55), whereas outreach with an asynchronous care survey link generated a much higher quit attempt rate of 17.6% (9 of 51).
We ran the same multivariable logistic analysis for the intervention of message source (physician vs system). The χ2 value for message source was not statistically significant, and the quit rates for patients who opened the message were similar to those for patients who received the message from the PCP (6 of 51 [11.8%]) and patients who received the system-generated message (7 of 55 [12.7%]; OR, 1.25 [95% CI, 0.39-4.05]).
In multivariable logistic regression analyzing the association of patient-related characteristics with opening the message, none of the characteristics were statistically significant when controlling for PCP. When not controlling for PCP, rural residence was the only characteristic associated with opening of the message (OR, 0.47 [95% CI, 0.23-0.97]) (Table 2). Again, this outcome likely occurred because PCP covaries with rural residence. The odds of rural patients reading the message (34 of 50 [68.0%]) were greater than those for urban patients (72 of 138 [52.2%]; OR, 2.11 [95% CI, 1.03-4.37]). In multivariable logistic regression to evaluate associations between patient characteristics and asynchronous quit attempts, controlling for PCP, only younger age was significantly associated with quit attempts (OR, 0.88 [95% CI, 0.80-0.97]) (Table 3).
A comparison of patients who made quit attempts with and without asynchronous care is summarized in Table 4. Six patients without chronic disease were in the asynchronous group with the survey link vs 1 patient in the group with no survey link. All quit attempts among the group younger than 45 years (6 of 6 [100%]) were made asynchronously via the survey link. Also, more women made quit attempts, accounting for most asynchronous care users (7 of 9 [77.8%]).
Six of the 10 physicians responded to the follow-up survey (1 indicates strongly disagree; 5, strongly agree). These physicians reported that the workload of this program was manageable (mean [SD] rating, 4.5 [0.5]), that they were comfortable prescribing medication based on the information gathered via the questionnaire (mean [SD] rating, 4.5 [0.8]), and that they were interested in expanding the program (mean [SD] rating, 4.2 [1.3]) and adopting similar programs for other disease states (mean [SD] rating, 4.0 [1.5]).
In this quality improvement randomized clinical trial, we found that (1) emailing portal messages to patients with instructions on how to successfully quit smoking (ie, via medication and counseling) and proactively offering help from their PCPs engaged patients in physician-assisted quit attempts and (2) quit attempts more than doubled when those messages included an embedded link to an asynchronous care survey, although neither finding was statistically significant. It is likely that the asynchronous care was more attractive to patients because they could obtain help and support directly from their PCP (eg, prescription medication) without having to call, schedule an appointment, and take time to attend in-person care. Although personalization has been shown to improve response to addiction intervention,25-27 patients responded similarly to the outreach regardless of whether it originated from their PCP or the health system. We posit that this outcome occurred because patients valued convenient access to physician help via asynchronous care. Interventions such as the present electronic outreach program that connect patients to physician-assisted, evidence-based quit attempts are critical because these attempts are expected to lead to higher rates of successful quit attempts,28 which generate both health cost savings and healthy living years.4-6,28,29 Confidence in the effects of electronic outreach and embedded asynchronous care is supported by the following outcomes: (1) all patients with EHR-documented quit attempts opened the message, (2) all quit attempts among those without a survey link occurred during in-person appointments, and (3) all quit attempts among the group with the survey link occurred asynchronously.
Notably, the program reached patients in populations where smoking is concentrated (ie, rural and younger patients). Although it has been reported that rural patients use health technology less often,30,31 in our study rural patients were 2 times more likely to open the message and their use of asynchronous care did not differ significantly from that of urban patients. We found that the asynchronous care users were healthier, significantly younger, and predominantly female. Consistent with national surveys,32 more women made quit attempts. We conclude that outreach programs with embedded, asynchronous care can improve evidence-based quit attempts among persons aged 25 to 44 years. Remotely reaching this age group is important because they use more tobacco (24% higher use)6,33,34 and visit physicians less often.29,35 Furthermore, if these individuals quit smoking by age 35 to 44 years, they will avoid most of the mortality risks caused by smoking.28,36
It is clear from the physician survey that physicians also valued this program, likely because the workload was manageable, and that they provided resources to patients who were willing to attempt quitting. Proactively finding smokers who are interested in physician assistance and asynchronously connecting them with their PCP via outreach is a substantial change in practice that could help overcome physician- and patient-level barriers (ie, time and effort) to collaborative quit attempts.
To our knowledge, our study is the first to evaluate the use of electronic outreach with and without embedded asynchronous care to engage smokers in evidence-based quit attempts. Unlike other physician-based smoking cessation programs, the program led to a high rate of quit attempts (17.6% among those who opened the message) without additional interventions such as motivational interviewers37,38 or specialized cessation clinics39,40 and without expensive mass media campaigns41-43 by using existing technologies. In comparison with previous studies evaluating online smoking cessation programs, our study produced a higher return of quit attempters,44-46 which suggests that motivated smokers welcome the assistance of a PCP. The need for PCP-provided technology-based smoking cessation assistance is clear by the low adherence to evidence-based practices among those who use smartphone apps for cessation.47 State quitline research has also found that smoker engagement increases when nicotine replacement medication is available remotely.48 The present study supports the ability of technology to remotely engage smokers in evidence-based quit attempts, which is timely given COVID-19’s impact on in-person care.39,49,50
Our study findings are limited by the sample size, which was intentional owing to unknown smoker response and a desire to not overwhelm the participating PCPs. In addition, although patients were randomly selected, they were selected among nonrandomized physicians. Future research should further investigate the enthusiasm smokers have for quitting after receiving remote assistance from their PCPs and should also test asynchronously supporting smokers until cessation. Because prescription medication is associated with more successful quit attempts, future research should also investigate whether remote PCP collaborations improve smoking cessation success rates.
The findings of this quality improvement randomized clinical trial, in which EHR data and portal messaging were leveraged to allow PCPs to remotely find and assist smokers via an electronic outreach program, higher rates of collaborative attempts were found in the group of patients offered asynchronous care, although the result was not statistically significant. Messages sent from a PCP (vs a health system) were not associated with higher rates of message opening or numbers of quit attempts. The program engaged individuals who tend to smoke more and face greater barriers to in-person care (ie, rural residents and younger patients). Smokers’ responsiveness to the program suggests that the use of technology could increase the percentage of evidence-based quit attempts.
Accepted for Publication: January 3, 2022.
Published: February 28, 2022. doi:10.1001/jamanetworkopen.2022.0348
Open Access: This is an open access article distributed under the terms of the CC-BY-NC-ND License. © 2022 Erdmann M et al. JAMA Network Open.
Corresponding Author: Mopileola Tomi Adewumi, MBS, MHA, Center for Health Sciences, Oklahoma State University, Tulsa, 1111 W 17th St, Tulsa, OK 74107 (firstname.lastname@example.org).
Author Contributions: Ms Erdmann had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Erdmann, Adewumi.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Erdmann, Adewumi.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: All authors.
Administrative, technical, or material support: Erdmann, Adewumi.
Conflict of Interest Disclosures: Ms Erdmann reported receiving personal fees from Pfizer, Inc. No other disclosures were reported.
Funding/Support: This program was partially funded by Pfizer Inc, in coordination with Center for Health Systems Innovation (CHSI) rural health medical quality project of Oklahoma State University. The health system where this program was implemented received no funds from CHSI or Pfizer Inc.
Role of the Funder/Sponsor: The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Data Sharing Statement: See Supplement 2.
Additional Contributions: William Paiva, MBA, PhD, Ipe Paramel, MHA, Cari Marshall, and Scott Shepherd, DO, Center for Health Systems Innovation, Oklahoma State University, contributed to the conceptualization and development of study materials and assisted in coordination with the health system. Ms Marshall was compensated for her work; the others were not.