Modified box plots depicting the adjusted 25th, 50th (median), and 75th percentiles for age at transfer to first adult office visit (A) and first adult preventive visit (B) for the overall sample and by primary predictors. Multivariable Cox models were used to construct the adjusted survival functions, from which these percentiles were obtained. The inset table on the left describes the sample size for each subgroup and the adjusted median age at transfer (also identified by the thick black bar within each box plot). For the provider network variable, subgroup totals do not sum to the total sample because individuals could contribute person-time to both categories of this time-varying covariate. HMO indicates health maintenance organization; POS, point of service; and PPO, preferred provider organization.
Modified box plots depicting the adjusted 25th, 50th (median), and 75th percentiles for transfer/gap length to first adult office visit (A) and first adult preventive visit (B) for the overall sample and by primary predictors. Multivariable Cox models were used to construct the adjusted survival functions, from which these percentiles were obtained. The inset table on the left describes the sample size for each subgroup and the adjusted median gap length (also identified by the thick black bar within each box plot). For the provider network variable, subgroup totals do not sum to the total sample because individuals could contribute person-time to both categories of this time-varying covariate. HMO indicates health maintenance organization; POS, point of service; and PPO, preferred provider organization.
eFigure 1. Unadjusted Estimates for Timing of Transfer to Adult-Focused Primary Care (Office Visits)
eFigure 2. Unadjusted Estimates for Timing of Transfer to Adult-Focused Primary Care (Preventive Visits)
Wisk LE, Finkelstein JA, Sawicki GS, Lakoma M, Toomey SL, Schuster MA, Galbraith AA. Predictors of Timing of Transfer From Pediatric- to Adult-Focused Primary Care. JAMA Pediatr. 2015;169(6):e150951. doi:10.1001/jamapediatrics.2015.0951
Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
A timely, well-coordinated transfer from pediatric- to adult-focused primary care is an important component of high-quality health care, especially for youths with chronic health conditions. Current recommendations suggest that primary-care transfers for youths occur between 18 and 21 years of age. However, the current epidemiology of transfer timing is unknown.
To examine the timing of transfer to adult-focused primary care providers (PCPs), the time between last pediatric-focused and first adult-focused PCP visits, and the predictors of transfer timing.
Design, Setting, and Participants
Retrospective cohort study of patients insured by Harvard Pilgrim Health Care (HPHC), a large not-for-profit health plan. Our sample included 60 233 adolescents who were continuously enrolled in HPHC from 16 to at least 18 years of age between January 2000 and December 2012. Pediatric-focused PCPs were identified by the following provider specialty types, but no others: pediatrics, adolescent medicine, or pediatric nurse practitioner. Adult-focused PCPs were identified by having any provider type that sees adult patients. Providers with any specialty provider designation (eg, gastroenterology or gynecology) were not considered PCPs.
Main Outcomes and Measures
We used multivariable Cox proportional hazards regression to model age at first adult-focused PCP visit and time from the last pediatric-focused to the first adult-focused PCP visit (gap) for any type of office visit and for those that were preventive visits.
Younger age at transfer was observed for female youths (hazard ratio [HR], 1.32 [95% CI, 1.29-1.36]) who had complex (HR, 1.06 [95% CI, 1.01-1.11]) or noncomplex (HR, 1.08 [95% CI, 1.05-1.12]) chronic conditions compared with those who had no chronic conditions. Transfer occurred at older ages for youths who lived in lower-income neighborhoods compared with those who lived in higher-income neighborhoods (HR, 0.89 [95% CI, 0.83-0.95]). The gap between last pediatric-focused to first adult-focused PCP visit was shorter for female youths than male youths (HR, 1.57 [95% CI, 1.53-1.61]) and youths with complex (HR, 1.35 [95% CI, 1.28-1.41]) or noncomplex (HR, 1.24 [95% CI, 1.20-1.28]) chronic conditions. The gap was longer for youths living in lower-income neighborhoods than for those living in higher-income neighborhoods (HR, 0.80 [95% CI, 0.75-0.85]). Multivariable models showed an adjusted median age at transfer of 21.8 years for office visits and 23.1 years for preventive visits and an adjusted median gap length of 20.5 months for office visits and 41.6 months for preventive visits.
Conclusions and Relevance
Most youths are transferring care later than recommended and with gaps of more than a year. While youths with chronic conditions have shorter gaps, they may need even shorter transfer intervals to ensure continuous access to care. More work is needed to determine whether youths are experiencing clinically important lapses in care or other negative health effects due to the delayed timing of transfer.
The transition from adolescence to adulthood is a critically important developmental period during which youths become independent and learn how to navigate normative adult experiences, including interactions with the health care system. Ensuring continuous access to a primary-care medical home during this transition is vital for addressing acute and chronic health issues, promoting positive health behaviors, and supporting youths’ overall health and well-being.1,2 Guaranteeing continuous access to health care during this life-course transition often involves transferring care from a pediatric-focused to an adult-focused provider, a complicated process that can present numerous challenges to patients and families, especially those with chronic conditions,3- 6 ultimately leading to disruptions in access.
It is well accepted that ensuring a timely, well-coordinated transfer from pediatric- to adult-focused providers is an important component of high-quality health care.7,8 Ideally, discussions with patients, families, and providers about transition should begin in early adolescence, well in advance of actual transfer, which guidelines suggest should occur between 18 and 21 years of age.7,8 These discussions should involve individualized planning, development of self-management skills, and assessment of transition readiness, in addition to identifying an appropriate adult-focused provider and communicating key information about the patient to the receiving provider.7 However, transition preparedness is often inadequately or infrequently discussed, and many youths do not receive recommended transition services,9 particularly those from socioeconomically vulnerable populations.10- 14
To measure and improve the quality of the transfer from pediatric- to adult-focused care, it is necessary to understand the current epidemiology of transfer timing and the factors that influence it. In the present study, we sought to examine the timing of transfer to adult-focused primary care providers (PCPs), the time between last pediatric-focused and first adult-focused PCP visits, and the predictors of transfer timing among youths enrolled in a health plan.
We examined the timing of transfer from pediatric- to adult-focused primary care providers (PCPs), the gap between last pediatric-focused and first adult-focused PCP visits, and the predictors of transfer timing.
Multivariable models showed an adjusted median age at transfer of 21.8 years and an adjusted median gap length of 20.5 months between last pediatric-focused and first adult-focused office visit with a PCP.
Adjusted median age at transfer and adjusted median gap length were longer for preventive visits (23.1 years and 41.6 months, respectively) than for all office visits.
Female enrollees and youths with chronic conditions transferred at younger ages and had shorter gaps than male enrollees and youths without chronic conditions, respectively.
Youths living in lower-income neighborhoods transferred at older ages and had longer gaps than youths in higher-income neighborhoods.
This retrospective cohort consisted of 60 233 adolescents who were enrolled in Harvard Pilgrim Health Care (HPHC), a large not-for-profit health plan with more than 1 million members in commercial plans concentrated in Massachusetts, New Hampshire, and Maine. Members receive care in a variety of settings, including medical groups, community health centers, independent physician practices, and a preferred provider network. Data for our study were obtained from HPHC enrollment and claims data. The HPHC institutional review board approved the study. We did not obtain consent because we used existing secondary claims data in the form of a limited data set, and we obtained a Health Insurance Portability and Accountability Act waiver from the HPHC institutional review board to do so.
The study cohort included 60 233 HPHC members who were enrolled continuously from 16 through at least 18 years of age, and up to 26 years of age or disenrollment, at any point between January 2000 and December 2012. Disenrollment was defined as a lapse in HPHC coverage of more than 2 months. Included participants were required to have had at least 1 pediatric-focused PCP office visit before turning 18 years of age, and participants were excluded if they had office visits before turning 18 years of age from only adult-focused PCPs.
The 2 primary outcome measures were (1) transfer timing, measured from 16 years of age to first adult-focused PCP visit, and (2) transfer gap, measured from last pediatric-focused PCP visit to first adult-focused PCP visit. These outcomes were determined separately for any type of office visit and for the subset of office visits that were preventive visits. Provider specialty codes from claims data were used to categorize providers as pediatric- or adult-focused PCPs; each provider could have up to 5 specialty provider codes. Pediatric-focused PCPs were defined as those with any of the following provider specialty types, but no others: pediatrics, adolescent medicine, or pediatric nurse practitioner. Adult-focused PCPs were defined as having provider types that see primarily adult patients (internal medicine, adult nurse practitioner, and geriatric medicine), provider types that may see both adult and pediatric patients (family practice, general practice, and family nurse practitioner), or providers with these adult-focused provider types in combination with any of the pediatric-focused provider types (eg, those with both internal medicine and pediatrics). Providers with any specialty provider designation (eg, gastroenterology or gynecology) were not considered PCPs.
Office visits were identified by having one of the following Current Procedural Terminology codes (identifying general and preventive office and outpatient visits): 99201 to 99205, 99211 to 99215, 99241 to 99245, 99384 and 99385, 99394 and 99395; along with one of the following encounter types: outpatient clinic, office, urgent care, or hospital ambulatory visit. Preventive visits were identified as a subset of office visits that had Current Procedural Terminology codes 99384 and 99385 or 99394 and 99395 (identifying comprehensive preventive evaluations), along with outpatient clinic, office, or hospital ambulatory visit encounter types. For analyses using preventive visits, we excluded 5110 participants who did not have any preventive visits with a pediatric-focused PCP before turning 18 years of age.
A participant’s transfer date was defined as the date of first adult-focused PCP visit during his or her enrollment; we constructed 2 distinct transfer dates, one using the first adult-focused PCP office visit of any type and the other using the first adult-focused PCP preventive visit. We excluded 8092 participants from office-visit analyses and 454 participants from preventive-visit analyses whose last pediatric-focused visit was after their first adult-focused visit because we could not determine whether this adult-focused visit represented a permanent transfer. However, we included these participants in sensitivity analyses, and we measured their transfer as the first adult-focused visit with a PCP after the last pediatric-focused visit with a PCP. Results were consistent with the main analyses; thus, only the main analyses are presented.
To measure transfer timing, we determined the time from 16 years of age to the transfer date, censoring participants at disenrollment or at 26 years of age, whichever came first. To measure the transfer gap, we determined the number of months between the last pediatric-focused PCP visit and the first adult-focused PCP visit, with right censoring at disenrollment or at 26 years of age. (With this design, youths who do not have an observable transfer are included and contribute data until they disenroll from the health plan or turn 26 years of age, whichever occurs first. Being 26 years of age was used as a cutoff because health care reform policies extend dependent coverage through 26 years.) We measured transfer timing and transfer gap separately using any office visits and preventive visits only.
Participants’ age, sex, and state of residence were obtained from enrollment files. We linked participants’ 5-digit zip codes to 2000 Census Bureau data to create a binary measure of neighborhood poverty in which a participant’s zip code was defined as a low-income zip code if 20% or more of residents were below the federal poverty level.15,16
To identify adolescents with chronic conditions, we used the Pediatric Medical Complexity Algorithm.17 This algorithm uses International Classification of Diseases, Ninth Revision codes from claims data to classify youths as having a complex chronic condition (a significant chronic condition in >1 body system [eg, diabetes mellitus and depression] or a single condition that is progressive or malignant [eg, cystic fibrosis]), a noncomplex chronic condition (a lifelong condition involving only 1 body system that is not progressive or malignant [eg, asthma]), or no chronic condition. We applied the Pediatric Medical Complexity Algorithm over the 2-year period from 16 to 18 years of age, and we required at least 2 separate claims of a qualifying diagnosis code.
We created a time-varying covariate to categorize the provider network of the participant’s health plan as either a more restricted provider network (eg, a health maintenance organization [HMO] plan or a tiered network plan) or a less restricted network (eg, a preferred provider organization [PPO] or point-of-service [POS] plan). We also created a fixed variable for whether participants changed from one type of provider network to another in the 12 months prior to the outcome or censoring.
Multivariable Cox proportional hazards regression was used to model transfer timing (ie, time to first visit with an adult-focused PCP) and transfer gap (ie, time from the last pediatric-focused PCP visit to the first adult-focused PCP visit). Models included sex, chronic condition, residence in a low-income neighborhood, provider network, change in provider network in the prior 12 months, and state fixed effects. To account for temporal trends, we included a continuous variable for the calendar year in which the participant turned 16 years of age. Models for transfer gap also adjusted for age at last pediatric-focused PCP visit. Survival functions output from the multivariable Cox models were used to estimate the adjusted median age at transfer and the adjusted median gap length. All analyses were conducted separately for any office visits and preventive visits only.
For the 60 233 youths with eligible pediatric-focused PCP office visits, the mean (SD) enrollment length after 16 years of age was 4.8 (2.2) years. For the 62 761 youths with eligible pediatric-focused PCP preventive visits, the mean (SD) enrollment length was 4.9 (2.2) years. Based on all PCP office visits, 36.9% of the youths in our sample transferred from a pediatric-focused to an adult-focused provider during their enrollment, while 36.0% were censored by disenrollment from the health plan, 0.3% were censored by reaching 26 years of age, and 26.8% were censored by the end of the study period. Among those who transferred during their enrollment, the (unadjusted) mean (SD) age at transfer for office visits was 19.8 (1.7) years. Kaplan-Meier survival plots for unadjusted age at transfer are presented in eFigures 1 and 2 in the Supplement.
Younger age at transfer (ie, faster rate of transfer) was observed for female youths (hazard ratio [HR], 1.32 [95% CI, 1.29-1.36]; Table 1) who had complex (HR, 1.06 [95% CI, 1.01-1.11]) or noncomplex (HR, 1.08 [95% CI, 1.05-1.12]) chronic conditions compared with those who had no chronic conditions. Transfer occurred at older ages (ie, was slower) for youths who lived in lower-income neighborhoods than for those who lived in higher-income neighborhoods (HR, 0.89 [95% CI, 0.83-0.95]). Compared with youths who had no change in plan network, youths who switched from PPO/POS to HMO/tiered network plans in the prior year transferred at younger ages (HR, 1.29 [95% CI, 1.20-1.39]), and youths who switched from HMO/tiered network plans to PPO/POS plans in the prior year transferred at older ages (HR, 0.78 [95% CI, 0.73-0.84]).
The gap between last pediatric-focused to first adult-focused PCP visit was shorter (ie, faster transfer) for female youths than male youths (HR, 1.57 [95% CI, 1.53-1.61]) and youths with complex (HR, 1.35 [95% CI, 1.28-1.41]) or noncomplex (HR, 1.24 [95% CI, 1.20-1.28]) chronic conditions. The gap was longer for youths living in lower-income neighborhoods than for those living in higher-income neighborhoods (HR, 0.80 [95% CI, 0.75-0.85]). The transfer gap was shorter in HMO/tiered network plans vs PPO/POS (HR, 1.05 [95% CI, 1.01-1.09]) and for youths who changed from a PPO/POS to an HMO/tiered network plan in the prior year (HR, 1.18 [95% CI, 1.10-1.27]), but it was longer for youths who changed from an HMO/tiered network plan to a PPO/POS plan in the prior year (HR, 0.82 [95% CI, 0.76-0.88]).
Using adjusted survival curves from the Cox models, we found that the median age at transfer overall was 21.8 years (Figure 1) and the median gap length was 20.5 months (Figure 2). Differences between subgroups in adjusted median age at transfer and adjusted median gap length are presented in Figures 1 and 2, respectively.
Cox models for preventive-visit transfer yielded similar results to models for all office visits (Table 2). Based on adjusted survival curves from the Cox models, the median age at transfer was 23.1 years (Figure 1), and the median gap length was 41.7 months (Figure 2).
Among a commercially insured population, we estimate that nearly half of youths transfer from pediatric- to adult-focused primary care later than recommended based on the timing of all office visits with an adult-focused PCP, with even later transfers occurring for preventive visits. Although there are no clear recommendations for the length of time over which such a transfer will occur, findings from our study suggest that potentially extended gaps in office and preventive visits also exist. Moreover, our results demonstrate that there are important differences in transfer timing by sociodemographic and health characteristics that could have important implications for delivery of high-quality health care for adolescents as they progress to adulthood.
Although there is consensus that the specific timing for optimal transfer from pediatric- to adult-focused primary care will vary according to the circumstances of each person, guidelines suggest that this transfer should ideally occur between 18 and 21 years of age,7 and that adolescents are recommended to have yearly preventive visits through 21 years of age.18 Our study documents that the majority of youths transferring to adult-focused providers are not receiving care according to these guidelines; instead, they are transferring later and with wider gaps between preventive visits than recommended. Multiple factors may have contributed to delayed transfer, including insurance barriers, lack of coordinated delivery systems, absence of mechanisms to ensure follow-up, and unfamiliarity with adult systems of care.4,6 The causes and effects of these delays merit additional study.
Notably, we found wide variation in transfer timing depending on whether preventive visits or all office visits were considered. The earlier transfers and shorter gaps for office visits may reflect acute care or introductory visits to an adult-focused PCP, rather than a purposeful transition. Although many youths do not go more than 2 years before making first contact with an adult-focused PCP after their last visit to a pediatric-focused PCP, the majority of youths who had a gap of at least 3 years before a first preventive visit with an adult-focused PCP may be at risk for adverse health and utilization outcomes due to decreased continuity of care.19- 22 The degree to which such gaps are clinically meaningful depend on the patient’s health status and other considerations; however, 2- to 3-year gaps may not be inappropriate for preventive care for healthy young adults,23 whereas a 1-year gap may be too long for those with chronic conditions, such as cystic fibrosis or diabetes.24,25
Female youths in this sample transferred at younger ages and had shorter gaps in care than male youths for both office and preventive visits, which is consistent with findings that women have higher baseline use of health care2,26 and may be more comfortable accessing care than men.27 Moreover, as young women may begin seeking reproductive care during this time, the observed sex differences may be partially explained by gynecologists promoting transfer to an adult PCP for nonreproductive primary care services. Our findings also identified that youths with chronic conditions transferred at slightly younger ages and had slightly shorter transfer gaps. While it is noteworthy that primary care transfers for youths with chronic conditions are timelier than for their healthy counterparts, contrary to other studies,28 it is possible that they should have even shorter transfer gaps to address their health care needs. A limited availability of qualified adult-focused providers comfortable with childhood-onset chronic conditions may be one reason for delayed transfer for these youths.29- 32
In addition, we found that youths living in low-income neighborhoods transferred at older ages and had longer transfer gaps than youths in higher-income neighborhoods. Neighborhood poverty may be a proxy for access barriers related to socioeconomic status or living in an under-resourced setting.33 For youths in lower-income neighborhoods who may be at a particularly increased risk for obesity,34,35 sexually transmitted infections,36 and other negative health behaviors37 or conditions,38,39 socioeconomic disparities in transfer timing and continuity of primary care during this critical period may have important consequences for receipt of needed treatment(s) and long-term health.
Finally, our study suggests that changes in provider network may affect transfer timing because youths who moved from a less (ie, PPO/POS) to more (ie, HMO/tiered) restrictive network transferred at slightly younger ages and with shorter gaps. Limiting choice or assigning a PCP for new enrollees may make for a less overwhelming decision for youths looking to transfer, or may stimulate choosing an adult-focused provider if the prior pediatric-focused PCP is no longer in their network, although we cannot investigate these mechanisms with our data. Ultimately, providers and health plans can assist youths with transitioning by identifying potential appropriate adult-focused, in-network providers available to the patient.
Several limitations should be considered when interpreting our findings. First, our data are from a single, regional private health plan, so these results may have limited generalizability to other regions and insurance types. Specifically, youths with public coverage or lapses in coverage may face additional barriers to accessing timely care from adult-focused providers,28,40 although the Affordable Care Act may mitigate some of these barriers.41 Second, with claims data, we are unable to identify which provider a person considers as their PCP, which may be an issue for women who seek primary care from a gynecologist. Third, it is possible that our study does not capture transfer for youths receiving primary care at college if such visits are not paid by their HPHC insurance, leading to a potential underestimation of transfer prevalence in our study. Fourth, youths may not have an observable transfer during their enrollment period because of loss of dependent coverage prior to transfer, which may have skewed our estimates. However, 65% of youths who never transferred continued to see a pediatric-focused PCP after 18 years of age, which suggests that many of those remaining covered into young adulthood do not transfer during this time period, in which case our sample may accurately reflect population-level trajectories in health care use.
Ensuring a smooth, timely transfer from pediatric- to adult-focused primary care is a crucial component of high-quality health care, especially for youths with chronic conditions. Our study identified that youths who are male, without chronic conditions, or from low-income neighborhoods are vulnerable to later transfer and longer transfer gaps, and that PCP transfer occurs later and with longer gaps for preventive visits compared with all office visits. Although youths with chronic conditions have shorter transfer gaps than those without chronic conditions, continuous care for this population may require even shorter transfer intervals. Ultimately, care transition is a complex, longitudinal process, of which the transfer from pediatric- to adult-focused providers is only one part. More work is needed to study the effect of transfer timing on health outcomes to know whether youths are experiencing clinically important lapses in care, and to determine and promote optimal transfer timing and high-quality transition.
Accepted for Publication: April 1, 2015.
Corresponding Author: Lauren E. Wisk, PhD, Center for Child Health Care Studies, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, 133 Brookline Ave, 3rd Floor, Boston, MA 02215 (firstname.lastname@example.org).
Published Online: June 1, 2015. doi:10.1001/jamapediatrics.2015.0951.
Author Contributions: Dr Wisk had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Wisk, Finkelstein, Sawicki, Toomey, Schuster, Galbraith.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Wisk.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Wisk.
Obtained funding: Finkelstein, Schuster, Galbraith.
Administrative, technical, or material support: Finkelstein, Lakoma.
Study supervision: Finkelstein, Sawicki, Toomey, Schuster, Galbraith.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was generously supported by grant U18 HS020513 from the Agency for Healthcare Research and Quality and the Centers for Medicare and Medicaid Services (Dr Schuster, principal investigator), grant 5T32HS00063-21 from the Agency for Healthcare Research and Quality (Dr Finkelstein, principal investigator), grant K24HD060786 from the National Institute of Child Health and Human Development (Dr Finkelstein, principal investigator), and by the Thomas O. Pyle Fellowship in the Department of Population Medicine.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Previous Presentation: This paper was presented at the Annual Meeting of the Pediatric Academic Societies; May 3, 2014; Vancouver, Canada; the 20th Annual National Research Services Award Trainees Research Conference; June 7, 2014; San Diego, California; the Annual Research Meeting of AcademyHealth; June 8, 2014; San Diego, California; and the 142nd Annual Meeting and Exposition of the American Public Health Association; November 19, 2014; New Orleans, Louisiana.
Additional Contributions: We wish to acknowledge the contributions of Kelly Horan, MPH, for institutional review board, administrative, and project management support. Ms Horan is affiliated with the Department of Population Medicine at Harvard Medical School and the Harvard Pilgrim Health Care Institute; her efforts were supported by grant U18 HS020513 from the Agency for Healthcare Research and Quality and the Centers for Medicare and Medicaid Services.