As health care systems struggle to meet increasing demands for services with limited resources, delivering value-based healthcare has become a priority. Since the passage of the Affordable Care Act in 2010, US health care has made great strides in its quest to change from “volume to value.”1 The Centers for Medicare and Medicaid Services have led the charge through the development and testing of alternative payment models to reduce spending without compromising quality.2 Despite this, proponents of improvements in health care value highlight the lack of unified country-wide progress; if these efforts are to promote meaningful change, tested and proven local initiatives must be scaled up to achieve system-wide change.1,3
Pediatric health care costs account for ∼9% of total healthcare expenditures in the United States,4 however, this translates into $300 billion dollars annually - equivalent to half of the United States defense budget.5 Low-value health care services contribute substantially to that cost, with an estimated 1 in 10 children receiving at least 1 low-value service each year.6
Since 2017, we have produced 5 updates on pediatric medical overuse on an annual or biannual basis.7–11 The goal of these updates is to highlight important examples of low-value care and targets for deimplementation. In this update, we identify the most important articles pertaining to pediatric medical overuse from 2021 and 2022.
Methods
Article Selection
Similar to our prior methodologies,7–10 articles were identified by both a manual table of contents review and structured Medline review for original research articles from January 1, 2021 to December 31, 2022. Journals included in the manual table of content review were Annals of Emergency Medicine, Archives of Disease in Childhood, BMJ, JAMA, JAMA Pediatrics, Lancet, The New England Journal of Medicine, and Pediatrics. These journals were selected by consensus among authors after considering impact factor, perceived impact on pediatric healthcare, and historical volume of published medical overuse-related articles as identified by prior reviews.8–11 Titles from each journal’s table of contents were reviewed independently by 2 authors; when it was unclear whether the article was related to overuse, the author reviewed the abstract and/or full text.
The structured Medline review was performed by searching PubMed for pediatric articles during the study period that contained the medical subject heading term “health services misuse” or “medical overuse,” as well as articles with titles containing the words “unnecessary,” “inappropriate,” “overutilization,” or “overuse.” For this update, we restricted our keyword search to article title (rather than title and abstract) because of a high rate of irrelevant article identification in prior updates.10 Articles not constituting original research, non-English articles, articles identified in the manual search, and articles without full text were excluded. These articles were reviewed by 1 author (N.M.) because of the higher volume and increased relevance of the table of contents review from prior reviews, and articles unrelated to pediatric medical overuse were excluded. Articles including both adult and pediatric subjects were considered for inclusion by the authors based on the clinical relevance and magnitude of potential harm to children.
Article Scoring
Three authors scored the articles identified by table of contents review, whereas PubMed articles were scored by 2 authors. Articles were scored based on a previously described rubric which included scores of 1 to 3 in each of 3 domains: strength of methods, magnitude of potential harm, and potential number harmed (Table 1).7–10,12 Before scoring, authors practiced scoring and discussing 5 articles to promote consistent scoring practices. Authors flagged articles as “not overuse” if the article did not constitute pediatric medical overuse after reviewing the abstract or full text. Additionally, authors could flag articles as “impactful” if they felt an article was highly impactful but may score poorly because of the confines of the rubric. Articles flagged by 2 authors (or 1 author for PubMed articles) as “not overuse” were excluded, whereas table of contents articles flagged by 1 author as “not overuse” were deliberated by 2 authors (N.M. and T.H.) for inclusion. Interrater reliability of article total score was assessed using the average intraclass correlation coefficient with corresponding 95% confidence interval (CI).
Scoring Rubric
Category . | Score . | Criteria . |
---|---|---|
1. Quality of methods | High | Randomized controlled trial (single or systematic review) |
Medium | Cohort study (single or systematic review), poor-quality randomized controlled trial | |
Low | Case-control, case series, cross-sectional, cohort with historical control, or poor-quality cohort study | |
2. Magnitude of potential harm | High | Profound impairment or life-threatening harm |
Medium | Prolonged, nonsevere harm | |
Low | Transient, nonsevere harm or costly | |
3. Number harmed | High | Common clinical scenario and common harm |
Medium | Common clinical scenario and uncommon harm or uncommon clinical scenario and common harm | |
Low | Uncommon clinical scenario and uncommon harm |
Category . | Score . | Criteria . |
---|---|---|
1. Quality of methods | High | Randomized controlled trial (single or systematic review) |
Medium | Cohort study (single or systematic review), poor-quality randomized controlled trial | |
Low | Case-control, case series, cross-sectional, cohort with historical control, or poor-quality cohort study | |
2. Magnitude of potential harm | High | Profound impairment or life-threatening harm |
Medium | Prolonged, nonsevere harm | |
Low | Transient, nonsevere harm or costly | |
3. Number harmed | High | Common clinical scenario and common harm |
Medium | Common clinical scenario and uncommon harm or uncommon clinical scenario and common harm | |
Low | Uncommon clinical scenario and uncommon harm |
Results
Overall, we identified 201 articles pertaining to pediatric medical overuse (Fig 1). For the manual table of content review, 4559 articles were reviewed, and 195 (4%) articles were deemed relevant to pediatric medical overuse (99 from 2021, 96 from 2022). Of these, 37 (19%) were flagged by ≥2 authors as not overuse and excluded. Forty-six (24%) articles were marked by 1 author as not overuse, of which 16 (35%) were excluded after review, resulting in 142 articles considered for inclusion from the manual table of contents review. Forty-six (24%) articles were marked by authors as impactful. The PubMed search returned 267 articles, 118 (44%) of which were reviewed and found to be related to pediatric medical overuse. Of these, 59 (50%) were flagged as not overuse, and 14 (12%) were flagged as impactful. The average intraclass correlation coefficient (95% CI) for all article total scores was 0.77 (0.69 to 0.83), reflecting good interrater agreement.13 Fifty-five top-scoring and impactful articles were discussed and considered for final inclusion. The top 10 articles are summarized below. A full list of the 201 articles with references and scores, including those discussed, can be found in Supplemental Table 2. The scores and scoring rationale for the top 10 articles can be found in Supplemental Table 3.
Articles That Question Routine Practices
Routine Temperature Measurement at Well-child Visits May Lead to Unnecessary Medical Care and Vaccine Deferral
Background
Routine temperature measurements are common in well-child visits. However, the utility of this practice is unknown.
Findings
In a 5-year retrospective cohort study of visits to a network of primary care clinics, temperatures were measured at 155 527 of 274 351 (58.9%) of well-child visits.12 Sixteen of 24 clinics measured temperatures at >90% of visits (labeled “routine measurement clinics”) and 8 of 24 measured temperature at <20% of visits (labeled “occasional measurement clinics”). Antibiotics were more common (adjusted odds ratio: 1.21; 95% CI 1.13 to 1.29), though diagnostic testing was less common (adjusted odds ratio: 0.76; 95% CI 0.71 to 0.82) at routine measurement clinics. Fever was detected at 0.2% (270 of 155 527) of routine measurement clinic visits, and of these, 17.4% were classified as incidental fever. Antibiotic prescription and diagnostic testing were more common at visits with incidental fever than without fever (7.4% vs 1.7%; 14.8% vs 1.2%; P < .001), and vaccines were deferred at 50% of such visits.
Limitations
The categorization of incidental fever was subject to the completeness and accuracy of clinical records. Additionally, this study was performed through a single health network, limiting generalizability.
Implications
Although the proportion of incidental fevers was low, given the frequency of well-child visits, the absolute number of patients affected by incidental fevers is high. Such identification may lead to overuse of healthcare resources and unnecessary vaccine deferral.
Routine Dental X-rays May Lead to Unnecessary Dental Procedures in Younger Children
Background
Untreated dental caries in deciduous teeth affect around 500 million children world-wide.14 The current screening recommendation for detecting dental caries is frequently a combination of visual inspection and radiographic examinations.15,16 A randomized controlled trial suggested that radiographic examinations do not provide benefit in caries management when compared with visual inspection alone.17
Findings
Two hundred and sixteen Brazilian children aged 3 to 6 years old were randomized to receive dental caries screening by visual inspection alone or visual inspection combined with radiographic examination.17 A secondary analysis of these data focused on proximal and occlusive surfaces of deciduous molars that did not receive operative intervention on initial examination.18 In a decision tree analysis of patients managed initially by visual inspection with radiographic findings considered afterward, radiographs missed 26% (1121 of 4383) of caries requiring intervention, though 98% of these were managed nonoperatively with fluoride. Additionally, false positives were more likely to be diagnosed by radiograph than by visual examination, accounting for 38% (17 of 45) vs 7% (3 of 45) of false positives, respectively. Dental caries were overdiagnosed by radiography (defined as caries discovered, not treated, and without subsequent progression) in 42 of 3835 (1%) molar surfaces without detection by visual examination.
Limitations
The study involved a small sample size from a single institution and was conducted in young children, limiting generalizability to older children or those with permanent teeth.
Implications
Radiographs can miss caries that require intervention and overdiagnose caries that would not need treatment, suggesting routine radiography may be unnecessary during routine dental visits for young children.
Longer Birth Hospitalizations May Not Lead to Better Outcomes
Background
During the coronavirus disease 2019 (COVID-19) pandemic, many elements of postpartum newborn care changed to limit viral transmission. The outcomes of these changes on newborn care, including changes in birth hospitalization length of stay, was unknown.
Findings
In this large observational study that included 202 385 infants, short birth hospitalizations, defined as <2 midnights for vaginal deliveries and <3 midnights for cesarean deliveries, increased from 28.5% to 43.0% during the COVID-19 pandemic.19 This finding persisted after multivariable adjustment (adjusted odds ratio, 2.30, 95% CI: 2.25 to 2.36). Despite this increase in short birth hospitalizations, infant readmission within a week of birth hospitalization discharge did not increase (1.2% prepandemic vs 1.1% during pandemic).
Limitations
Other relevant outcomes, including breastfeeding establishment and parental satisfaction, were not examined.
Implications
Infant-mother dyads may be able to spend less time in the hospital, reducing healthcare burden and allowing families to return home sooner.
Longer Antibiotic Courses May Not be Necessary for Children With Nonsevere Community-acquired Pneumonia
Background
Most national guidelines recommend 7 to 10-day courses of antibiotics for treatment of community acquired pneumonia (CAP), however, these recommendations are based on scant evidence.
Findings
In a systematic review and meta-analysis of 9 randomized clinical trials of 11 143 patients without underlying comorbidities between 2 months and 10 years of age, treatment failure occurred in 12.8% vs 12.6% (risk ratio [RR] 1.01; 95% CI:0.92 to 1.11) of patients randomized to shorter versus longer antibiotic courses, respectively.20 Overall, a 3-day course was noninferior to a 5-day course (RR 1.01; 95% CI:0.91 to 1.12) and a 5-day course was noninferior to a 10-day course (RR 0.87; 95% CI:0.50 to 1.53). In subgroup analyses, noninferiority was met for patients aged 2 months to 5 years (RR 1.01; 95% CI:0.91 to 1.11), however, not for patients older than 5 years (RR 2.07; 95% CI:0.76 to 6.63; Risk Difference 0.15; 95% CI: −0.05 to 0.36).
Limitations
The study included infections by different organisms that may require varying durations of treatment. Definitions of CAP and treatment failure varied by study.
Implications
Shorter course oral antibiotics for nonsevere CAP are noninferior to longer course treatments.
Conventional Care in an Incubator is Worse for Infants With Low Birth Weight Than “Kangaroo Mother Care”
Background
Low birth weight infants born premature or small for gestational age account for 70% of neonatal deaths worldwide. Although kangaroo mother care, defined as skin-to-skin contact and exclusive breastfeeding, is known to be highly effective at preventing neonatal death,21 the safety and efficacy of kangaroo mother care before medical stabilization was unclear.
Findings
An open-label multicenter trial in 5 hospitals in Ghana, India, Malawi, Nigeria, and Tanzania randomized 3211 infants with birth weights between 1.0 kg and 1.799 kg to receive either immediate kangaroo mother care (intervention group) or conventional care in an incubator or radiant warmer (control group) until stabilization.22 Infants receiving kangaroo care were secured to the mother’s chest with a binder that maintained the infant’s airway. In the intention-to-treat-analysis, infants that received immediate kangaroo mother care had lower mortality (12.7%) in the first 28 days compared with infants in the control group (15.7%), (relative risk 0.75 [95% CI: 0.64 to 0.89]). These findings were explained by lower rates of hypothermia and sepsis in the intervention group. The trial was stopped early because of the benefits of immediate kangaroo mother care.
Limitations
Blinding was not possible in this trial given the nature of the intervention, which may have resulted in bias in the assessment of secondary outcomes but did not affect the primary mortality outcome. Twenty percent of eligible infants were not enrolled in the study because either the infant or mother was deemed too sick to participate. Results may not be generalizable to developed nations.
Implications
Hospitals in low-resource settings should consider providing immediate kangaroo care to low-birth weight infants before medical stabilization rather than placing them in incubators.
Articles Supporting Judicious Surgery Referral
Gastrostomy Tubes are Frequently Placed in Children for Whom the Benefit is Unclear and in Children for Whom Home Nasogastric Tube Feeding Would be a Superior Option
Background
Findings
A systematic review and meta-analysis of gastrostomy tube decision-making initially identified 900 related articles, from which the authors created 17 recommendations from 58 publications after reviewing data, level of evidence, and risk of bias.25 These recommendations were summarized into 4 themes: (1) there should be strong consideration of long-term home nasogastric feeding before GT placement, particularly for patients who are likely to learn to eat by mouth; (2) GT placement should only be pursued after failure to wean off nasogastric feeding and multidisciplinary team evaluation should occur before and after GT placement; (3) investigation for an underlying medical diagnosis should be performed before GT placement; and (4) do not obtain routine screening upper gastrointestinal contrast studies for asymptomatic patients undergoing GT placement.
Limitations
Studies were largely observational and did not evaluate direct comparisons between GT and nasogastric tube feeds.
Implications
Encouraging long-term nasogastric tube use, reducing the number of GT placements, and decreasing unnecessary radiologic testing may improve resource utilization and reduce potentially harmful interventions. Randomized controlled trials are needed to compare nasogastric and GT feeding outcomes.
Tympanostomy Tubes May Not Reduce the Incidence of Acute Otitis Media
Background
Findings
A multicenter randomized controlled trial included 250 vaccinated children aged 6 to 35 months, of whom 208 (83%) completed a 2-year follow-up.28 Included children had at least 4 episodes of AOM in 12 months and at least 1 episode in the preceding 6 months. Children were randomized to receive tympanostomy tubes within 2 weeks or episodic medical management with 10 days of oral amoxicillin-clavulanate. Ten percent of those in the tympanostomy-tube group did not undergo placement, and 45% of those in the medical-management group ultimately received tympanostomy-tubes because of treatment failure or parental request. In the intention-to-treat analysis, the rate (± SE) of episodes of AOM per child-year during a 2-year period was 1.48 ± 0.08 in the tympanostomy-tube group and 1.56 ± 0.08 in the medical-management group (P = .66). In the per-protocol analysis, the tympanostomy group had 1.47 ± 0.08 episodes per year and the medical management group had 1.72 ± 0.11 episodes per year (RR, 0.82; 95% CI, 0.69 to 0.97).
Limitations
A high rate of children in the medical management group eventually underwent tympanostomy tube placement. Blinding was not feasible given the nature of the interventions. Other outcomes, such as speech and language improvements, were not evaluated.
Implications
Considering the risks of anesthesia and peri-operative complications in young children, serious consideration should be given to medical management rather than tympanostomy placement for recurrent AOM in otherwise healthy children.
Articles Demonstrating Adverse Effects of Pediatric Interventions That Are Commonly Overused
Postoperative Tonsillectomy Mortality Highlights the Importance of Shared Decision Making, Particularly in Children With Medical Complexity
Background
Tonsillectomy is one of the most common surgeries performed in pediatric patients, and children who undergo tonsillectomy often lack evidence-based indications for the procedure.29 Postoperative mortality rates are not well characterized, especially for children with comorbidities.
Findings
A retrospective cohort study from 2005 to 2017 included 504 262 patients under 21 years old.30 The overall unadjusted 30-day postoperative mortality rate was 7 deaths per 100 000 tonsillectomy operations. The adjusted mortality rate for children with chronic complex conditions was significantly higher (117.22 per 100 000) than for those without these conditions (3.87 per 100 000; adjusted rate difference 113.55 [95% CI: 51.45 to 175.64] per 100 000). Rates increased in relation to the number of chronic complex conditions with 1011 deaths per 100 000 operations in children with 3 or more conditions. Though children with complex chronic conditions accounted for 2.8% of tonsillectomies, they accounted for 44% of postoperative deaths.
Limitations
This retrospective study used registry data that were obtained for administrative purposes. Therefore, the appropriateness of surgery could not be assessed and may have underestimated risks from pre-existing conditions. The small absolute numbers of deaths precluded an analysis of potential interactions between covariates.
Implications
The postoperative mortality observed in this study is significantly higher than was previously known. Although the postoperative mortality from tonsillectomy is relatively low, the potential benefits of this elective surgery must be weighed against the risks, especially in children with complex medical conditions.
Proton Pump Inhibitor Use is Associated With Increased Incidence of Childhood Asthma
Background
Findings
A retrospective cohort study of 80 870 matched patients less than 18 years without a prior history of asthma33 found that patients with PPI use had a greater incidence of asthma compared with those without use (21.8 vs 14.0 events per 1000 person-years, hazard ratio 1.57 [95% CI, 1.49 to 1.64]). The risk of asthma was significantly increased across all age groups but those at highest risk of being diagnosed with asthma were children under 2 years (P < .001).
Limitations
Because of the retrospective nature of the study, misclassification of asthma was possible. Indications for PPI use were not available, raising the possibility of confounding by indication.
Implications
PPI use in children may be associated with the subsequent development of asthma and its downstream consequences on health care resource utilization. PPI use in children should be carefully considered and prescribed only when clearly indicated.
Over-the-counter Cough and Cold Medications Continue to Cause Pediatric Deaths
Background
Over-the-counter cough and cold medications are commonly used by caregivers despite warnings against their use.34 Fatalities from exposure to these common over-the-counter medications have not been described since 2008 when these medications underwent labeling changes warning that these products should not be used in young children.
Findings
The Pediatric Cough and Cold Safety Surveillance System conducted a retrospective study of children <12 years exposed to over-the-counter cough and cold medications and reporting at least 1 significant adverse event.35 An expert panel determined the causal relationship between medication exposure and death. Of the 7983 children with adverse events, 188 (2.4%) died. Of the 180 eligible fatalities between 2008 and 2016, 40 (22.2%) were deemed to be related or potentially related to the medication exposure. Of these, 24 patients (60.0%) were <2 years. Of the 31 cases where method of drug administration was reported, a caregiver administered the medication in 28 cases (90.3%). In 7 cases (17.5%), the medication was used intentionally to sedate the child. Diphenhydramine was the index ingredient for 28 of the 40 cases (70.0%).
Limitations
The true prevalence of death may have been underestimated as the data source relies on self-reporting. Similarly, details in reporting may have been insufficient for the expert panel to accurately determine causality and autopsy reports were not available for this study.
Implications
Many children have died of exposure to over-the-counter cough and cold medications despite product labeling changes. It is incumbent on healthcare providers and public health professionals to educate caregivers on the dangers of these medications.
Discussion
We have identified and summarized 10 important articles from 2021 to 2022 related to pediatric medical overuse. Several themes emerged from the articles, the first of which involves questioning routine care interventions. There are many interventions clinicians perform that are habit-driven rather than evidence-based. Seemingly harmless interventions such as vital sign monitoring, for example, may have unintended consequences. One article demonstrated that the discovery of incidental fever during routine temperature measurement was associated with vaccine delay and increased antibiotic prescriptions.12 This is consistent with other studies showing that nighttime vital signs and pulse oximetry monitoring for hospitalized children are common without evidence of benefit and may contribute to sleep interruption.36,37 This review also shows the lack of benefit from routine radiographs during dental checkups and potential harm from overusing infant warmers for infants with low birth weight in low-resource settings, rather than providing “kangaroo mother care.”18,22 Clinicians should be mindful of routine medical interventions not rooted in evidence that do not provide a net benefit to patients.
In addition to routine practices, clinicians should question established treatment durations that are not evidence-based. Using the natural experiment brought on by the COVID-19 pandemic, Handley et al showed that a 1-night postpartum hospital stay did not lead to increased complications compared with a 2-night hospital stay.19 Not only would such practice potentially reduce patient hospital bills and healthcare spending, it may also help newborns avoid unnecessary medical interventions from routine assessments by well-meaning medical personnel. A recent Cochrane review similarly suggested that there is likely minimal difference in benefit and harm between early and standard discharge, but suggested further studies are needed.38 Evidence continues to mount that many “standard” intravenous antibiotic durations based on expert opinion may be safely shortened.39–42 In addition to the article herein reviewed,20 several other recent randomized trials have demonstrated a 5-day antibiotic course to be noninferior to a longer antibiotic course for CAP.43–45
This update’s articles also encourage judicious practices for common pediatric surgeries. Surgical interventions carry potential for severe intraoperative and postoperative complications. Intraoperative anesthesia is generally well tolerated in patients without significant comorbidities,46,47 though severe adverse events are not uncommon48 and repeated exposure in young ages may adversely impact development.49 Postoperative complications because of tympanostomy tube placement draw particular attention because of its high surgical frequency within the United States. A large meta-analysis of more than 60 randomized controlled trials found that tympanostomy tube perioperative complications were not uncommon and included: tympanosclerosis (32%), obstruction (7%), granulation tissue (5%), premature extrusion (4%), and chronic perforation (2% to 17%).50 Furthermore, children undergoing surgeries are at risk for receiving unnecessary postoperative opioid prescriptions.51 These nontrivial complication rates of high-frequency surgeries highlight the importance of avoiding unnecessary childhood operations and practicing shared decision making when the risk-benefit ratio is unclear.
Lastly, our review highlighted 3 articles that highlight the adverse effects of pediatric treatments that are commonly overused: namely tonsillectomy, PPIs, and cough and cold medications. In addition to the possibility of postoperative mortality highlighted above, children undergoing tonsillectomy commonly require subsequent medical care, including additional operations or hospital admission because of postoperative hemorrhage (0.2% to 7%) and dehydration.52–54 Evidence also continues to indicate the negative effects of PPI use in infancy, including future asthma risk, fractures, decreased nutrient absorption, and altered microbiome,33,55,56 prompting the American Academy of Pediatrics to add routine PPI use to their Choosing Wisely list of practices physicians and patients should question.57 Additionally, cough and cold medications continue to contribute to pediatric mortality in the United States, despite recommendations against their use.35
Our review is subject to limitations. Article scoring may be limited by subjectivity, which we mitigated by using a standard scoring rubric and through a practice article scoring session. Our manual table of contents search may have led to disproportionate representation from these journals. Lastly, some overuse articles were likely missed because of search strategy limitations.
Conclusions
The articles in this review highlight important areas of focus to improve pediatric healthcare value. Routine practices, including vital sign measurement, antibiotic durations for common illnesses, and hospital durations that are not evidence-based, should be questioned. Additionally, common pediatric surgeries, including tonsillectomy and tympanostomy tube placement, carry significant morbidity and mortality, which should prompt judicious use. As improved healthcare value typically leads to decreased revenue under a fee-for-service payment model, it will be particularly important that more value-based healthcare models are explored and implemented.
FUNDING: No external funding.
CONFLICT OF INTEREST DISCLOSURES: The authors have indicated they have no conflicts of interest relevant to this article to disclose.
Comments