Abstract
BACKGROUND AND OBJECTIVES: Infants who live in households experiencing food insecurity are at risk for negative health and developmental outcomes. Despite large numbers of households within our population experiencing food insecurity, identification of household food insecurity during standard clinical care is rare. The objective of this study was to use quality-improvement methods to increase identification of household food insecurity by the second-year pediatric residents working in the Pediatric Primary Care Center from 1.9% to 15.0% within 6 months. A secondary aim was to increase the proportion of second-year pediatric residents identifying food insecurity.
METHODS: A team was formed to identify key drivers thought to be critical to the process of identifying food insecurity during well-child care. This project addressed 5 key drivers and tested interventions based on these drivers over a 6-month period at a hospital-based primary care site that serves ∼15 000 children from underserved neighborhoods. Tests included implementing an evidence-based electronic screen for food insecurity, educational interventions to improve understanding of food insecurity, empowerment exercises targeting clinicians and families, and gaining buy-in and support from ancillary personnel.
RESULTS: Implementation of these changes led to an increase in the identification rate of household food insecurity from 1.9% to 11.2% over the 6 months (P < .01). The proportion of residents identifying food insecurity increased from 37.5% to 91.9% (P < .01).
CONCLUSIONS: Application of quality-improvement methods in a primary care clinic increased ability to effectively screen and positively identify households with food insecurity in this population.
- CCHMC —
- Cincinnati Children’s Hospital Medical Center
- EMR —
- electronic medical record
- PL2 —
- Pediatric Level 2 (second-year pediatric resident)
- PPCC —
- Pediatric Primary Care Center
- PSDA —
- plan-do-study-act
- QI —
- quality improvement
- WCC —
- well-child care
Food insecurity is defined as the lack of access to enough food to fully meet basic nutritional needs because of insufficient resources.1–5 Data suggest that 16% to 22% of US households experience food insecurity. Infants who live in food-insecure households are especially vulnerable to the negative effects of insufficient nutrition, which can lead to negative psychological, behavioral, and cognitive outcomes.6 Families experiencing hunger may experience many social and financial strains, forcing difficult decisions between nutrition and other essential needs.6–9 Most low-income families receiving care at urban pediatric clinics report at least 1 unmet basic need (eg, food, housing, employment), with many reporting several.7 The American Academy of Pediatrics’ Task Force on the Family, 2003, suggests that pediatricians screen, assess, and appropriately refer based on families’ social needs, particularly food insecurity.5,10 Insufficient and inconsistent screening for social risks can result from unreliable or absent standardized screening.11 Confounding this is the fact that many trainees have limited awareness and discomfort discussing basic needs, as most come from middle- to upper-income backgrounds.12,13 In addition, responding to identified risks remains a challenge because of time-constrained visits.7,12,14
In 2010, an institutional review board–approved anonymous survey was performed at the Pediatric Primary Care Center (PPCC) to assess background rates of household food insecurity among families with infants younger than 1 year of age. Food insecurity was defined as a positive screen to the US Department of Agriculture’s validated 6-item indicator for determining household food insecurity.2 A total of 144 families were surveyed over a 4-week period in this unpublished study. Food insecurity was present in 30% of the surveyed households. The social risk screening questions embedded in the electronic medical record (EMR) were reviewed for 800 infants seen for well-child care (WCC) between April and August 2010. Although second-year pediatrics residents (PL2s) documented screening 74% of families for food insecurity, they identified only 1.9% as food insecure.
Without accurate screening and identification of household food insecurity, we believed that we were failing to meet a basic need of the children and families from our clinic. Because effective and reliable identification is necessary before implementing interventions, a quality-improvement (QI) project was initiated with the primary aim to increase the PPCC PL2s’ identification of household food insecurity from 1.9% to 15.0% within 6 months. A secondary aim was to increase the proportion of PL2s identifying food insecurity by 50% in the same 6-month period.
Methods
Setting
Cincinnati Children’s Hospital Medical Center (CCHMC) is a large, urban, pediatric academic medical center. This QI project was performed in the PPCC, the hospital-based outpatient primary-care site, which is the medical home for ∼15 000 children (35 000 visits/year). The patient population is 70% African American, 21% white 1% Hispanic, 1% Asian, 3% multiracial, and 4% other. The payer mix is 80% Medicaid, 10% private insurance, and 10% self-pay. The PPCC has 22 attending physicians (8 full-time equivalents) and multiple ancillary providers. The PPCC is the continuity clinic site for approximately half of the CCHMC pediatric residents. The 24 PL2s with the continuity clinic at the PPCC were the intervention group for this QI project. This cohort of PL2s was demographically representative of our resident population: 67% were women; 96% were between 26 and 30 years old; and 83% self-identified as white, 4% as black/African American, and 13% as other.
Improvement Team
A CCHMC pediatric chief resident led the multidisciplinary team that included attending physicians, residents, dieticians, social workers, legal partners from the PPCC’s Medical-Legal Partnership, and ancillary personnel. Caregivers were interviewed to obtain families’ perspectives and input, given the sensitive nature of household food insecurity. A data analyst and QI consultant provided technical and methodological support.
Planning and Executing the Intervention
PL2s already screened for social risk factors, but there was concern about the quality of this screening owing to lack of case detection. Therefore, the team attempted to increase the accuracy of the screening process by conducting preliminary interviews with key stakeholders, including PPCC staff, families, and the PL2s who most frequently identified social risks. The QI team found significant reasons that were potentially affecting clinic screening. There was variation among providers’ style in taking a social history, despite preexisting standardized prompts in the EMR. Some providers lacked awareness of social strains and others did not feel empowered to offer families lasting solutions when strains were detected. In addition, the food insecurity question embedded in the social history was not validated and was located on the social/environmental screen, not the nutrition screen, where most physicians addressed issues related to feeding and household food supply.
Five key drivers to effectively screen our population for food insecurity were identified: the use of evidence-based criteria to standardize and systematize screening, awareness that food insecurity is a problem in our population, understanding why screening for social risks relates to child health, empowerment of providers and families to feel comfortable discussing sensitive topics, and buy-in from clinic staff (Fig 1).
Key drivers included in our QI initiative.
Our goal was to identify half of the households estimated to be food insecure, which would be 15% of an estimated 30% of the PPCC population. PPCC leadership felt there was sufficient support from social workers, nutritionists, and legal advocates to effectively handle the concerns that could arise if 15% of our families were identified as food insecure.
Several interventions were planned to address the various key drivers of this process. First, interventions to use evidence-based criteria to standardize screening began. A published 2-question screen studied in low-income families with young children was tested, which showed that a positive response to either question had a sensitivity of 97% and specificity of 83% in previous studies.15 After a 2-week trial on paper, data collection indicated that this 2-question evidence-based and validated screen for food insecurity was superior to the existing tool, so it was embedded in the EMR. Next, the location of the food insecurity questions was addressed. After discussion with the improvement team and resident users, it was determined that the nutrition form was a better location for food insecurity questions (Fig 2). After a small-scale test, the questions were permanently moved to the new EMR location.
Nutritional screening questions in the EMR.
The next set of interventions focused on 2 key drivers: awareness that food insecurity is a problem in our population and an understanding of how and why screening for social risks relates to child health. To address these drivers, the team started a series of educational interventions. Education was essential to our intervention group, as all residents were less than halfway through their training. In addition, most of the participating pediatric residents were not raised in families that experienced social strain. A series of interactive large- and small-group multidisciplinary case-based sessions that focused on the negative effects of food insecurity, the prevalence of food insecurity, and the importance of accurate and effective social risk screening was implemented and continued over 6 months.
To empower residents and to address our fourth key driver, many PL2s were observed taking a social history and given one-on-one feedback for both validation and suggestions for individual improvement. This enabled the improvement team to gain additional insight into the variation of residents, as it became apparent that some residents did not ask such questions because they did not know how to react when they received a positive response. Further education on interventions and asking sensitive questions was added to the educational series, including videos on how to conduct interviews. In addition, residents received positive feedback via E-mail after positively identifying a family with food insecurity. Residents who had detected no food insecurity were deemed “low performers” and received feedback to help them overcome their specific barriers.
In addition to attempting to empower residents, families were also encouraged to speak up about social concerns they faced. Posters in exam rooms explained to families the PPCC’s focus on social determinants of health and about the services available. The PPCC began a new initiative with a community partner to provide supplemental infant formula to food-insecure households with children younger than 1 year of age. Bundled with the formula was educational material prepared by the QI team that focused on sustainable budgeting, job, career training, and additional food resources.
Finally, gaining buy-in from the entire PPCC staff was essential. Although the PL2s were the focus of the interventions, improvement efforts ultimately involved the entire clinic. The QI team met with clinic management to integrate this work into the existing flow and mission of the PPCC. Data were shared with all staff to further promote awareness.
Outcome Measures and Methods of Evaluation
The primary outcome measure was the percentage of households identified as food insecure by PL2s during routine WCC. This outcome was chosen because identification during WCC would allow for intervention at the point of care. A secondary outcome measure was the percentage of PL2s who successfully identified an infant living in a food-insecure household. This outcome was chosen to determine if a more standardized screening process would better engage PPCC physicians. Baseline data were collected from 800 charts via EMR review from April through August 2010, specifically looking at WCC encounters completed by the 24 residents followed during this project.
Analysis
The QI team used a series of statistical run charts, which show trends over time, to monitor progress toward improvement. Run charts track the effectiveness of piloted interventions, as well as differentiating common-cause and special-cause variation. Common-cause variation is the typical variation that occurs inherent to a process, but lacks significance in individual high and low numbers. In contrast, special-cause variation results from causes that are from specific circumstance. Our goal in using run charts was to identify if special-cause variation could be attributed to our previously stated interventions, as well as to reduce the inherent variation in process or common-cause variation.
Statistically significant differences between preintervention and postintervention outcomes were assessed. For positive identification of food insecurity, we defined April through August 2010 as preintervention and August 2010 to January 2011 as postintervention. P values were obtained via χ2 testing.
Ethical Issues
Background information regarding PL2s was collected anonymously via online survey methodology in a study approved by the CCHMC institutional review board. This project was conducted within PPCC’s ongoing improvement efforts that respect privacy of all key stakeholders. All patient information reviewed was de-identified and no protected patient information was collected, so as to protect patient privacy.
Results
The demographics and childhood social hardships (history of inadequate housing, food security, use of public benefits, maternal depression, and exposure to domestic violence)16 of the PL2s are listed in Table 1. More than 70% of the PL2s grew up in households with none of these defined hardships. Owing to the anonymous survey, this information was not correlated to determine if childhood social hardships affected resident screening for food insecurity.
Demographics of PL2 Residents (n = 24)
Despite the lack of reported personal experiences with social hardships, the PL2s increased the rate of identification of household food insecurity. The cohort’s baseline identification of food insecurity rate increased from 1.9% to 11.2% during the intervention period (Fig 3). Based on the number of PL2s scheduled in the PPCC on any given week, an average of 43 families (range 23–72) were screened weekly during the intervention period. Therefore, the PL2s increased identification of food insecurity from fewer than 1 to almost 5 families per week. The run chart shows that the efforts to standardize the evidence-based questions, alter the EMR, and educate those most involved in the care processes provided a successful change that was statistically significant (P < .01).
Percentage of households with food insecurity identified by PL2s.
Decreased physician variation occurred as the screening process was standardized. At the beginning of the project, 9 of the 24 PL2s (37.5%) had documented identifying a family experiencing food insecurity. On completion of our process and educational interventions, 22 of 24 residents (91.9%) had detected and documented food insecurity during routine WCC visits. Residents were counted as positive regardless of the number of cases of food insecurity detected during the intervention period, as the performance improvement was variable across residents. The increased number of providers effectively screening for food insecurity was statistically significant (P < .01).
Discussion
QI methodology can enhance rates of identification of household food insecurity in a complex resident-driven primary care practice. Our key drivers included the use of evidence-based criteria to systematize screening through the EMR. Increased awareness about food insecurity and understanding how it affects child health were addressed through education. The empowerment of both resident providers and families to feel comfortable discussing sensitive topics, and gaining buy-in from clinic staff were addressed by increasing awareness of food insecurity and early exploration of potential interventions, such as point-of-care formula distribution. Measurable change required a multiphase approach changing over time to determine optimal processes through feedback, observation, and testing interventions in plan-do-study-act (PDSA) cycles. Because this project had a relatively short 6-month time frame, additional measurement will determine if the results can be sustained.
Household food insecurity was more frequently identified after an evidence-based nutritional screen was embedded in the EMR on the nutritional history form (Fig 3).15 Electronic history taking is becoming more common in the primary care setting.17 The introduction of prompts within the EMR can be effective, time efficient, and easily adaptable if used appropriately. This ability to change EMR-based questions proved important, as both content and location were refined. Further changes could still be necessary, as possibilities to intervene expand.12 A provider-administered screening tool can be an effective way to increase identification of food insecurity in a busy clinic14; however, this requires the clinician to interview in a sensitive and family-centered manner.
Because most residents did not grow up in poverty, they cannot rely on personal experience. Based on Knowles’ theory on adult learning styles, this suggests a need for education and feedback on these topics18; however, the correlation between residents’ experience with childhood social hardships and their ability to detect food insecurity was not the study focus. Low-performing residents were asked their opinions and feelings about the issue of food insecurity and how they interact with families. After expressing their personal views and receiving feedback from peers on how others approach such issues, many low performers did detect greater numbers of families with food insecurity. The team anticipates that lessons learned by these residents will be sustainable in their future clinical practice.
Several of our interventions had no measurable effect on our outcome (Fig 3). Some PDSA cycles, including social history videos and engaging low performers, resulted in only common-cause variation. Although we do not know the ultimate effect of these interventions, we felt that these 2 interventions were not worth further pursuit. The new program to provide infant formula was announced at the end of the intervention period as a separate project undertaken in parallel to these QI interventions. The formula distribution has not been abandoned even though it did not show evidence of special-cause variation. Because of the brief time frame of this QI project, the formula distribution was not studied long enough to determine its ultimate impact. Because the formula program targets only families with infants, its effect on the overall trend will likely be limited.
Educational interventions are more difficult to sustain than those based on system change, such as implementation of EMR-based templates. There was increased identification with educational interventions, but the duration of these effects is unknown; however, there is a need for education based on the residents’ limited life experience related to social and financial hardships and to enhance their overall training in general pediatrics. Although isolated educational interventions have low reliability and often lack sustainability, owing to turnover and inconsistent application, the continued and repetitious educational curriculum helps address this problem. EMR documentation of food-insecurity questions was used as a proxy for physician behavior, but discrepancies between what was asked and what was documented cannot be determined. Furthermore, it is difficult to know the degree of sensitivity and family centeredness displayed by the residents, which may affect the family’s response. Although the EMR-based template makes documentation easier, it may lead to inaccurate information. Wagner and Hogan,19 evaluating medication data, found that outpatient EMRs may have significant levels of data error. Because of the nature of our measure, it is impossible to learn from failures, as we do not know if a specific household is food insecure but does not wish to share that information, or if our screening process is insufficient. Risk identification is also limited by the possibility of social desirability bias.20 Families may not wish to report household food insecurity for fear that their physician would think less of them or even report them to children’s protective services. A small number of PL2s from a single hospital-based pediatric clinic with excellent ancillary support staff participated, so the ability to generalize may be limited.
Although a significant increase in detection rates was documented, the predetermined goal of detecting 15% of food-insecure families was not achieved. This goal was established knowing the limitations of family report and provider detection on sensitive issues in conjunction with the PPCC’s ancillary staff’s ability to handle intervention on this rate of detection. Families’ potential to underreport food insecurity may mean that screening was effective in identifying a high percentage of families willing to disclose food insecurity. Even though the goal of 15% was not reached, the team believes that the greater than fivefold increase is clinically significant.
Conclusions
QI methodology, in conjunction with multidisciplinary collaboration, can be effectively implemented in a busy pediatric clinic to improve identification of household food insecurity. Targeting a specific provider population and standardizing the screening process can increase effective screening. Interventions that focused on education and evidence-based prompts in an EMR were most effective. Several interventions instituted in PDSA cycles were ineffective and abandoned. As the importance of awareness of food insecurity within a vulnerable population increases, techniques to improve identification become more essential. This QI platform will be used to develop and test a food insecurity intervention, and to expand screening and intervention for other social determinants of health.
Footnotes
- Accepted October 10, 2011.
- Address correspondence to Mary Carol Burkhardt, MD, Cincinnati Children’s Hospital Medical Center, ML 2008, 3333 Burnet Ave, Cincinnati, OH 45229. E-mail: mary.burkhardt{at}cchmc.org
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
References
- Copyright © 2012 by the American Academy of Pediatrics
In this issue
Jump to section
Related Articles
Cited By...
- Food Insecurity Screening Among Families of Children With Diabetes
- What a Summer Nutrition Benefit Program Could Mean for Clinicians and Their Patients
- Suburban Families Experience With Food Insecurity Screening in Primary Care Practices
- Clinicians Perceptions of Screening for Food Insecurity in Suburban Pediatric Practice
- Housing, Transportation, And Food: How ACOs Seek To Improve Population Health By Addressing Nonmedical Needs Of Patients
- Promoting Food Security for All Children
- Addressing Social Determinants of Health at Well Child Care Visits: A Cluster RCT
- Increasing Recruitment Rates in an Inpatient Clinical Research Study Using Quality Improvement Methods
- Forging a Pediatric Primary Care-Community Partnership to Support Food-Insecure Families