Quality Improvement Increases Pediatric Community Hospital Smoking Cessation Interventions

This QI study was conducted in a community hospital using Cerner as its electronic medical record (EMR). Pediatric inpatient care is provided by pediatric hospitalists 24/7. There is no limit to the number of patients who may be admitted to the pediatric hospital medicine (PHM) service, and the number of beds used for pediatrics on the general medical unit flexes according to local needs. There is no onsite pediatric intensive or intermediate care unit. Nursing care is provided by nurses who flex between adult and pediatric care depending on local needs, with the exception of level II nursery patients, who are cared for exclusively by neonatal nurses. Average daily census over the study period was ∼7 patients/day.

All children admitted to the PHM service were eligible, including those on the general pediatric inpatient ward, newborn nursery, or level II nursery. Patients and families with preferred languages other than English were included in screening. Screening and education were performed through the use of an interpreter in all languages, and written education and referrals were available in both English and Spanish. Patients aged 18 years and older are typically admitted to adult services and were therefore excluded. The retrospective baseline study period was December 2019 through November 2020. The intervention period began in December 2020, and data were reviewed prospectively through June 2021.

Our QI team consisted of 6 of the 7 pediatric hospitalists who regularly provide care to pediatric patients at our community hospital site, including 1 pediatric hospitalist with previous QI expertise and 1 pediatric hospitalist with previous expertise in smoking cessation counseling. An information technology specialist and 2 nurse leaders served as consultants to the project team on an as-needed basis. To identify impactful interventions, we conducted a root cause analysis for our primary outcome, which revealed many improvement opportunities. A benefit-effort analysis was conducted to identify interventions with the highest potential benefit and lowest projected effort.

We undertook 5 interventions based on the analysis described previously. Monthly meetings were held among the QI team to analyze project data, identify new interventions, and review implementation of interventions. In monthly PHM meetings (November 2020–May 2021), education was provided regarding strategies for SHS exposure/tobacco use screening, how to provide family education regarding SHS exposure/smoking cessation, and how to make referrals to the CA Smokers’ Helpline. Specifically, physicians were educated to ask either the patient or the caregivers (if the patient was younger than age 12 years) if either they themselves or a household member smoke or vapes tobacco. If this screen was positive, providers were asked to (1) provide bedside education regarding smoking cessation on both admission and discharge, (2) attach a standardized discharge handout addressing SHS exposure and smoking cessation, and (3) offer a referral to the CA Smokers’ Helpline. Documentation was standardized within EMR (December 2020), including documentation of SHS exposure/tobacco use screening within the “Histories” tab and specification of standardized discharge instruction documents addressing SHS exposure and smoking cessation. Visual reminders were posted at the physician workstations to prompt physicians to ask about SHS exposure and to document patient/caregiver responses to these inquiries (December 2020). Wallet cards were ordered and made available for referral to the CA Smokers’ Helpline (February 2021).

The primary measure was the monthly percentage of pediatric inpatients who were screened for SHS exposure/tobacco use. The numerator was the monthly number of pediatric inpatients with the presence or absence of SHS exposure/tobacco use documented in the physician notes or in the “Histories” tab of the EMR. In our younger patients in whom personal smoking history was irrelevant (i.e., newborn, toddler), we looked for documentation of any SHS exposure within the household. We discussed personal smoking history with patients aged 12 years and older, unless patient history prompted earlier assessment because this this is the typical age that our group completes full Home Education Activities Drugs Suicidality Sex assessments. Personal and caregiver smoking history could not be separately analyzed because this is not demarcated in our EMR, with the only option being “Is there a smoker in the household?” unless the screener decided to write a comment. Unfortunately, EMR changes at our institution were slowed by pandemic priorities. Tobacco use or SHS exposure via e-cigarettes was included, although this could also not be separately analyzed because it was not consistently delineated. Marijuana use was excluded because not all staff documented this as smoke exposure or smoking. The denominator was the total monthly number of pediatric inpatients hospitalized on the PHM service. Secondary measures were the percentage of pediatric inpatients or caregivers who screened positive for smoke exposure and (1) received discharge instructions regarding protecting children from SHS exposure/smoking cessation and (2) received a referral (for either self or caregiver) to the CA Smokers’ Helpline. We also measured length of stay (LOS) as a balancing measure to ensure that any attempts to provide smoking cessation education did not delay discharge. LOS was measured in hours, then converted to days for ease of interpretation. We (E.D., L.L., A.T., M.T., R.N., M.H.) conducted manual chart review to collect these data, in addition to demographic data, including patient age, sex, smoking history, race, ethnicity, and maternal race and ethnicity for newborn patients (Table 1).

The primary measure and balancing measure were analyzed using statistical process control in SPC for Excel (BPI Consulting, LLC, Oklahoma City, OK). For the primary measure, a P chart was selected to monitor variation in proportion of nonconforming items in subgroups of variable size. For the balancing measure of LOS, an X-bar S chart was selected to monitor variation in a continuous variable with varying subgroup size. Upper and lower control limits were defined as greater than or less than 3 σ. Special cause variation was identified by a single point outside of the upper or lower control limit, 2 of 3 points in the outer third of control limits, or by 8 consecutive points above or below average.¹⁵  A break in control limits was introduced only when a change in process was detected. Participant demographics were compared in the baseline and intervention periods using Welch’s 2-sample t test to compare continuous variables and χ² tests for categorical variables. Secondary measures were monitored on run charts, with medians reported and months with subgroup size of 0 removed. We ran a subset analysis on screening rates, separately analyzing pediatric and newborn, as these are discrete units with different nursing staff that contribute to screening (Supplemental Figs 5 and 6). A separate analysis was not performed on our secondary measures as writing discharge instructions and referring to the Helpline was done by pediatric hospitalists who work in both units.

This study was determined to be exempt by our local institutional review board because it was deemed “not human subjects research.”

A total of 1073 patients were included in baseline data collection, and 634 were included in the intervention period (N = 1707). Participant demographic characteristics are shown in Table 1. The majority of participants were in the newborn nursery (94% in baseline and 95% in intervention) and White/Caucasian (75% in both baseline and intervention). Non-Hispanic/Latino (53% in baseline and 56% in intervention) was the most common ethnicity. There was no significant difference in distribution of sex, race, or ethnicity between baseline and intervention time periods.

The primary measure of the total percentage of patients screened for SHS exposure/tobacco use increased from the baseline mean of 14% to 90% in the intervention period (Fig 1; interventions denoted by vertical black markers). A statistical process control chart shows declining screening rates in the baseline period with special cause variation in the intervention period. The mean of the entire postintervention period was 73%. We initially achieved special cause variation between December 2020 and March 2021, with 2 of 3 points within the outer third of the upper control limit, then again achieved special cause variation between April 2021 and June 2021, giving a final calculated mean of 90%. The sustain phase of this final mean is unknown because of limitation by the manual nature of data collection. Breaking down screening rates into pediatric and newborn populations showed an increase in pediatric screening from 77% in the baseline period to 85% in the intervention period (Supplemental Fig 5) and an increase in newborn screening from 5% to 72% (Supplemental Fig 6).

Data from secondary measures are shown in Figs 2 and 3. Of the patients who screened positive for SHS exposure/tobacco use, the median percentage who received written discharge instructions about reducing SHS exposure or smoking cessation increased from 0% in the baseline period to 56% in the intervention period (Fig 2). Similarly, the median percentage of patients or caregivers in homes with SHS exposure/tobacco use that received a referral to the CA Smokers’ Helpline increased from 0% in the baseline period to 17% in the intervention period (Fig 3).

The balancing measure of mean LOS increased slightly from 1.87 days in the baseline period (December 2019–August 2020) to 2.15 days in the intervention period (December 2020–June 2021). There were 2 months of increased LOS in the intervention period that met special cause variation, but, overall, there was not a mean line shift from special cause variation (Fig 4). Of note, there is significant variation in LOS at baseline, as noted on the accompanying S chart (Fig 4), suggesting that the process was not stable even at baseline.

This QI project shows that pediatrician-led interventions can increase screening rates for SHS exposure/tobacco use in a community hospital pediatric unit. In addition, we successfully increased inclusion of standardized patient/caregiver discharge instructions on SHS exposure/tobacco use and Smokers’ Helpline referrals, with no apparent impact on LOS. This study adds to the growing body of literature demonstrating the effectiveness of a QI approach in improving SHS exposure screening, education, and Helpline referrals,^9–13  applied in a community hospital pediatric setting encompassing a broad range of pediatric inpatients, from newborns to adolescents and their caregivers.

Although we did not collect patient address data, our community hospital serves a large proportion of families living in rural areas. Children living in rural areas are more likely to be exposed to secondhand smoke, and individuals who smoke in rural regions tend to smoke more cigarettes per day.^16,17  In addition, adolescents in rural areas tend to begin smoking earlier than their peers living in urban regions. A 2018 Cochrane review of studies aiming to reduce children’s SHS exposure showed that many interventions were successful in some studies but not others, including in-person counseling, motivational interviewing, telephone counseling, and educational interventions. Measures such as biological feedback of a child’s exposure, nicotine replacement therapy, educational home visits, and group sessions did not show a significant reduction in SHS exposure.¹⁴  Although we did not specifically measure smoking cessation as an outcome, we anticipate that our educational and counseling interventions played a role in promoting smoking cessation and reduction of both personal and SHS exposure in our largely rural pediatric population.

We exceeded our aim for referring patients with a positive SHS exposure/tobacco use screen to the CA Smokers’ Helpline, achieving a median of 17%. This was a vast improvement from 0% before our study, but could be further improved given other studies that achieved greater than 70% referral rates.^9,12  To help improve referral rates, we attempted to create an electronic referral system built into our EMR. However, the build was slowed by pandemic priorities and is not yet live at the time of publication. Another future intervention that could help improve referral rates is incorporation of warm handoffs, which have been shown to increase referral completion rates in other settings.¹⁸ 

One barrier to referral was the inability to consent a household smoker for Helpline referral when he or she was not present at the bedside. Other studies have based referral percentage only on successful discussions directly with the smoker, which may partially explain our lower referral rates. This does raise the question of why the person that smokes was not present. Although this was not formally documented, from our patient interactions, we note that parents in our community do not always have jobs that allow remote work or multiple days off, even when a child is ill. Families in our community also often have multiple children and typically rely on at least one of the caregivers to be at home caring for them. Finally, there are many multigenerational households in our community, and the person that smokes is not always a primary caregiver. We addressed this by providing CA Smokers’ Helpline wallet cards for patients/caregivers to give to the person that smoked, but effectiveness of this intervention was difficult to analyze. This does provide an opportunity for expansion of our QI to include interventions such as follow up by outpatient providers or phone calls to caregivers not present.

Although we met our aim for increasing smoke exposure/tobacco use screening rates overall, when broken down into pediatric and newborn, our rates were 10% higher in the pediatric unit (Supplemental Figs 5 and 6). We suspect this may have been due to higher baseline screening rates in the pediatric unit because it is part of the pediatric nursing checklist. Inclusion of this checklist in the newborn nursery may help increase these numbers. In addition, some previous work suggests that an EMR prompt may help further increase screening rates.¹⁹ 

A distinguishing feature of our QI project is that it was conducted in a community hospital. In a 2021 survey of the pediatric hospital medicine workforce, one-half of respondents worked at a community site, 21.2% worked at both a community and university site, and 21.2% at solely a university site,²⁰  yet the majority of published smoking cessation QI projects are from academic institutions. Studies at academic centers are essential in guiding our evidence-based care, but there are notable differences from community hospitals. Interventions need to be efficient and achievable in a variety of settings. Community hospitalists typically do not have the assistance of residents, clinical educators, or robust research groups to assist with interventions. We also work in multiple settings, including running codes and providing consults in the emergency department, newborn resuscitation in the operating room, and caring for children on an adult medicine floor.^20–23  Smoking cessation QI in community hospitals has the potential for higher continuity because there can be stronger connections to local outpatient pediatrics. Many community hospital institutions, including our own, have outpatient pediatricians that also work inpatient shifts.²³  Also, we often have different processes and workflow than academic institutions, which includes advocating for children in a system geared toward adult care.^21–23  Effecting sustainable change requires clear knowledge of how an institution and its workforce function.

Running a QI study in a community hospital is valuable for evaluating smoking cessation interventions in a workplace environment and patient population that can differ from academic institutions; however, there are also limitations. Similar to other community hospitals, the majority of our patients are newborns, limiting our general pediatric sample size and opportunities for smoking cessation conversations with pediatric patients. These low general pediatric numbers impact our data; for example, missing 1 discharge instruction could greatly decrease the average for that month. Future studies could compile data from multiple community hospitals to further investigate the success of individual interventions specifically for our nonnewborn patients. Including nonnewborn pediatric patients in a community hospital setting also limits control and standardization. In our newborn units, infants are seen by the same pediatric hospitalists and nursing staff; a relatively closed system. Nonnewborn pediatric patients are cared for in a variety of adult oriented units by staff not consistently working with this population. Therefore, our screening question itself could not be completely standardized, and we were unable to separately analyze SHS exposure from tobacco use because this was not consistently documented and is not delineated in our EMR. This is a clear area for future QI interventions that would benefit other community hospitals.

Finally, as a smaller institution, we have limited resources. Although our population is diverse, with a large Urdu-speaking population, CA Smokers’ Helpline wallet cards and discharge instructions are only available in English and Spanish. Our pediatric hospitalist group is also small, and with our work unfunded, we were unable to engage in ongoing manual data collection or in time intensive interventions, such as in-depth discussions about referral to the CA Smokers’ Helpline or personal follow up. Although we exceeded our aim for Helpline referrals, referral percentage down trended toward the end of the study, which may be due in part to these conversations taking valuable time in a busy hospitalist schedule. As a result, we have limited knowledge about the longevity of our impact in these measures. Insufficient time is a known barrier to smoking cessation efforts in pediatric inpatient units^24,25  and notably, previous studies that have achieved greater impact often leveraged more robust resources, such as funding a dedicated research team to conduct smoking assessment and counseling, and ability to perform pilot testing of smoking interventions.^9,12  These limitations offer future opportunities for creative Plan Do Study Act cycles specific to our unique community, such as training other hospital staff in standardized screening and making referrals to the Helpline, partnering with our outpatient colleagues, follow-up investigation of long-term impact, and translating resources into our community’s most common languages.

This study shows that community pediatric hospitalist-led QI can increase SHS exposure/tobacco use screening and interventions to encourage smoke exposure reduction and smoking cessation for inpatient pediatric patients and their caregivers. Future research is needed in community hospitals on tobacco use screening in pediatric patients outside of the newborn unit, implementing clear hospital-wide EMR documentation of SHS exposure versus tobacco use, and Smokers’ Helpline referrals in cases where the smoker is not present. This research would provide another step forward in advocating for children in adult-oriented systems.

The authors acknowledge Xy Losito, MSN, RN, for his assistance with information technology and the electronic medical record; Valerie Stump, RN, for her nursing leadership and insight into the QI process; and Nicole Bettencourt, BS HCA, RCP, RRT, HACP, for her respiratory therapy and performance excellence expertise.