VARIATION IN PREVENTIVE CARE IN CHILDREN RECEIVING CHRONIC GLUCOCORTICOID THERAPY

Abstract

Objective

To assess preventive care measure prescribing in children exposed to glucocorticoids and identify prescribing variation according to subspecialty and patient characteristics.

Study design

Retrospective cohort study of children initiating chronic glucocorticoids in the Gastroenterology, Nephrology, and Rheumatology divisions at a pediatric tertiary care center. Outcomes included 25-hydroxyvitamin D (25OHD) and lipid testing, pneumococcal polysaccharide (PPV) and influenza vaccination, and stress dose hydrocortisone prescriptions.

Results

701 children were followed for a median of 589 days. 25OHD testing was performed in 73%, lipid screening in 29%, and PPV and influenza vaccination in 16% and 78%. Hydrocortisone was prescribed in 2%. Across specialties, 25OHD, lipid screening, and PPV prescribing varied significantly (p<0.001). Using logistic regression adjusting for specialty, 25OHD testing was associated with older age, female sex, non-Hispanic ethnicity, and lower baseline height and BMI Z-scores (all p<0.03). Lipid screening was associated with older age, higher baseline BMI Z-score, and lower baseline height Z-score (all p<0.01). Vaccinations were associated with lower age (p<0.02), and PPV completion was associated with non-white race (p=0.04).

Conclusions

Among children chronically exposed to glucocorticoids, 25OHD testing and influenza vaccination were common, but lipid screening, pneumococcal vaccination, and stress dose hydrocortisone prescribing were infrequent. Except for influenza vaccination, preventive care measure use varied significantly across specialties. Quality improvement efforts are needed to optimize preventive care in this high-risk population.

Although glucocorticosteroid therapy is common and effective in a number of pediatric conditions, its potent immunosuppressive and metabolic effects result in well-known and sometimes severe side effects or adverse events. These include, but are not limited to, infections, vertebral compression fractures, adrenal suppression, weight gain, and dyslipidemia. Additionally, disease-specific factors increase the risk of infections, accelerated atherosclerosis, and fractures in inflammatory conditions such as systemic lupus erythematosus (SLE).

In recent years, expert panel reports evaluated preventive care in chronic illnesses across the age spectrum. For example, the Endocrine Society recommends 25-hydroxyvitamin D (25OHD) testing for individuals at risk for deficiency, such as those receiving glucocorticoid therapy(1). In the American Academy of Pediatrics report from the Expert Panel on Integrated Guidelines for Cardiovascular Health and Risk Reduction in Children and Adolescents, lipid testing was strongly recommended for children with medical conditions associated with accelerated atherosclerosis, such as inflammatory conditions(2). The Advisory Committee on Immunization Practices and the Infectious Diseases Society of America recommend pneumococcal and yearly influenza vaccination(3). Finally, to prevent adrenal crisis in children receiving chronic glucocorticoid therapy, the Lawson Wilkins Pediatric Endocrine Society suggests parents be instructed in the use of intramuscular hydrocortisone in case of vomiting or severe stress to prevent acute adrenal insufficiency(4). The data supporting these guidelines undoubtedly influenced recently published quality indicators in SLE(5, 6) and inflammatory bowel disease(7), which include process metrics in bone health and cardiovascular risk assessment as well as influenza and pneumococcal vaccination.

Implementing guidelines in practice is often variable. Among pediatricians, lack of awareness and uncertainty as to how to address dyslipidemia may contribute to low reported lipid screening rates(8). In a survey of pediatric rheumatologists, physician-reported bone health monitoring was low despite the recognized risk of vertebral compression fractures(9). The suboptimal rates may be in part due to ambiguity with result interpretation and clear management strategies. Low pneumococcal and influenza vaccination rates were noted in children with cystic fibrosis, human immunodeficiency virus, diabetes mellitus, and liver transplants(10). The prevention strategies unique to children receiving chronic glucocorticoid therapy and the predictors of utilization have not been systematically characterized. Additionally, no studies have simultaneously assessed prescribing across preventive care domains. Understanding current care patterns in this high-risk population will be critical to future quality improvement initiatives. In this retrospective cohort study, we aimed to characterize preventive care utilization related to bone health, cardiovascular risk, infection, and adrenal insufficiency in children receiving chronic glucocorticoid therapy.

METHODS

We used a retrospective cohort design to characterize patterns of screening and preventive care measure prescribing for children exposed to chronic glucocorticoid therapy. Data were extracted from electronic health records in the outpatient care network at The Children’s Hospital of Philadelphia. The Institutional Review Board approved this study.

We retrospectively identified children initiating glucocorticoid between January 1, 2011, and December 31, 2012, with a minimum of 90 days of follow up. Data up to and including September 30, 2013, were used. All outpatient prescriptions for glucocorticoids during the study period were extracted. We created an electronic algorithm to inspect the start and end date of each prescription, the number of refills, the quantity dispensed, and free text instructions to determine the intended treatment duration(11). Using both discrete and free text dose information, we determined the daily dose represented by each prescription. All doses were converted to equivalent milligrams of prednisone (1 mg of prednisone or prednisolone is equivalent to 4 mg hydrocortisone, 0.77 mg methylprednisolone, and 0.16 mg dexamethasone). All children whose glucocorticoid exposure represented at least 15 days of glucocorticoid treatment at a minimum dose of either 0.1 mg/kg/day or 5 mg/day of prednisone, whichever was lower, were included. We excluded patients whose exposure was below the minimum treatment threshold and prescriptions for inhaled or topical steroids, including oral budesonide for inflammatory bowel disease. We also excluded children who received 15 days or more of glucocorticoid treatment between July 1, 2010, and December 31, 2010, to ensure accurate glucocorticoid start date ascertainment within our study period. To validate our automated electronic algorithm, we performed a manual chart review of 98 charts of patients identified as having glucocorticoid exposure of any duration during the study period. Of the 98 patients, 60 met chronic exposure criteria using the automated method. We manually examined patient records to determine the glucocorticoid dose and duration to verify chronic. The manual chart review revealed 93% sensitivity and 87% specificity of the automated identification method for chronic glucocorticoid exposure.

Outcomes

Using data from the EHR we described process and clinical outcomes in the domains of bone health, cardiovascular health and nutrition, immunization status, and adrenal insufficiency management. We first set out to characterize glucocorticoid exposure and its effect on anthropometric measures. Among children 2 years and older, we used reference data from the Centers for Disease Control (CDC) to calculate height and body mass index (BMI, kg/m²) Z-scores(12). We compared baseline and follow-up measurements to characterize changes in height and BMI Z-scores in our cohort with glucocorticoid exposure.

Our main outcomes for the study were process measures. Laboratory data were used to identify children with at least one order for a 25OHD between January 1, 2011, and September 30, 2013. The presence of at least one lab order for a lipid profile during the observation period was considered appropriate screening for dyslipidemia. Immunization records were extracted to determine whether children received pneumococcal and influenza vaccination. Due to the risk of invasive pneumococcal disease, all children at least 24 months old in our cohort should have received at least one dose of 23-valent pneumococcal polysaccharide vaccine (PPV). Documentation of any PPV administration prior to 9/30/2013 was recorded. We assessed PPV administration in the entire cohort as well as a limited analysis in children 2 years and older at baseline, and according to primary in or outside our hospital network. These analyses were performed to ensure that there was no difference between in and out of network patients in the recommended age range for vaccination. Excluding observation time for children before 6 months of age, we assessed the proportion of children receiving at least one dose of seasonal influenza vaccine for each influenza season. Only patients who received primary care in our institution were included for the seasonal influenza vaccination analysis. This is commonly given in the outpatient setting and we were unable to detect administration if given out of our network. We documented orders for home injectable hydrocortisone to treat adrenal insufficiency in emergency cases of severe stress after new chronic glucocorticoid exposure.

To determine the volume of preventive care utilization, we assessed the total number of preventive care measures prescribed in each individual. Influenza vaccination was excluded from this analysis because it was not reliably assessed in patients receiving care outside the hospital system. PPV is more difficult to obtain in the pediatric community setting, and can be provided in subspecialty clinics. Therefore, PPV was more likely to be captured in our electronic record and was included in the analysis.

Covariates

We extracted demographic and clinical data to identify predictors of preventive care measure prescribing. Variables included race, ethnicity, sex, age at treatment initiation, in network primary care, insurance status, and primary specialty managing glucocorticoid treatment. Total cumulative glucocorticoid exposure per kilogram was converted into quartiles for analysis.

Statistical analyses

Descriptive analyses were performed to determine means, standard deviations, medians, ranges and distributions for continuous variables. The Student t-test was used to assess group differences for normally distributed data and the Wilcoxon’s rank sum test for non-parametric distributions. Differences in proportions were assessed using the chi-square test. We utilized a two-sided test of hypothesis, and a p-value of <0.05 was considered statistically significant. Univariable logistic regression was performed to identify predictive covariates for each preventive care measure outcome. Multivariable logistic regression analysis was performed for all variables with p-values <0.1 by univariable analysis, adjusting for follow up time to limit misclassification bias and primary specialty to account for differences in divisional practices. Variables with p-values <0.05 in multivariable regression were kept in the final model. Area under the curve and goodness of fit were assessed for all logistic regression models. Analyses were performed using Stata 13.1 (Stata Corp; College Station TX, 2008).

Results

Between January 2011 and December 2012, 701 children (52% female, 66% Caucasian) with a median age of 15 years received at least 15 days of glucocorticoid therapy (Table I). Glucocorticoid exposure quartiles were found to be: Quartile 1 (<21.6 mg/kg), Quartile 2 (21.6 – 48.0 mg/kg), Quartile 3 (48.1 – 90.5 mg/kg) and Quartile 4 (≥90.6 mg/kg). Common disorders included inflammatory bowel disease (n= 256, 37%), minimal change disease or nephrotic syndrome (n=91, 13%), and SLE (n=63, 9%). Additionally, among patients who received a transplant, 62 were followed in the Division of Nephrology and 44 in the Division of Gastroenterology. Primary care was provided within our hospital network in 143 patients (20%) and 533 (76%) carried private insurance.

Table 1.

Characteristics of Pediatric Patients Receiving Glucocorticoid Therapy

Characteristic	Value (n=701)
Baseline age, years (median, IQR)	15 (10,18)
Range	1-21
Follow Up, median days (IQR)	589 (392,797)
Range	90-990
Female, n (%)	362 (52)
Ethnicity, n (%)
Non-Hispanic or Latino	611 (87)
Hispanic or Latino	64 (9)
Unknown	26 (4)
Race, n (%)
White	463 (66)
Black	149 (21)
Asian	25 (4)
Other or unknown	64 (9)
Primary Specialty, n (%)
Gastroenterology	356 (51)
Nephrology	202 (29)
Rheumatology	143 (20)
Glucocorticoid Exposure (median, IQR)
Duration, days	143 (64, 284)
Cumulative prednisone dose, mg	2083 (970, 3975)
Cumulative prednisone dose per mg/kg	48 (22, 91)

Characteristics and Effects of Glucocorticoid Exposure

The median cumulative glucocorticoid exposure was 48 mg/kg (IQR: 22 to 91) over 143 days (IQR: 64 to 284). During the follow up period, we observed the expected glucocorticoid -induced changes in height and BMI. Over the duration of treatment, the median height Z-score decreased from −0.34 to −0.40 SD (n=673, p=0.006), and the BMI Z-score increased from 0.20 to 0.45 SD (n=651, p<0.001).

Variation in Preventive Care Measure Prescribing

25OHD was obtained in 511 (73%), and lipid screening in 205 (29%). PPV was provided in 112 (16%) and stress dose hydrocortisone prescribed in 15 (2%) (Table II). Among the 143 children receiving primary care within the hospital network and who were eligible for influenza vaccination in the 2012-2013 influenza season vaccination was given in 78 (72%).

Table 2.

Preventative Care Prescribing Variation According to Pediatric Specialty

Preventative Measure	Overall	GI	Nephrology	Rheumatology	p
25-hydroxyvitamin D	511 (73)	291 (82)	129 (64)	91 (64)	<0.001
Lipid screening	205 (29)	44 (12)	98 (49)	63 (44)	<0.001
PPV	112 (16)	9 (3)	89 (44)	14 (10)	<0.001
Influenza vaccination	78 (72)	41 (69)	27 (73)	10 (83)	0.616
Injectable hydrocortisone	15 (2)	9 (3)	2 (1)	4 (3)	0.401

Values expressed as n (%). GI: Gastroenterology; PPV: Pneumococcal polysaccharide vaccination. Influenza vaccination data are provided only for the 2012-2013 influenza season and for those receiving primary care in the hospital network.

We identified significant variation in preventive care measure prescribing across specialties. A greater proportion of patients receiving care for GI conditions had 25OHD testing (p<0.001), but were significantly less likely to have lipid screening (both p<0.001). Those with renal diseases were most likely to receive PPV (p<0.001), though patients with rheumatic conditions were more likely than those with GI conditions to receive PPV (p=0.001). Neither influenza vaccination nor injectable hydrocortisone prescriptions differed between specialties.

No preventive care measures were prescribed in 128 (18%), one preventive care measure in 354 (51%), two in 169 (24%), and three in 49 (7%). Only one patient was prescribed four preventive care measures (0.1%). Those with renal conditions received a higher median number of preventive care measures (2 versus 1 p<0.001).

Predictors of Preventive Care Measure Prescribing

We assessed demographic and treatment-related predictors of preventive care measure prescribing. Univariable and multivariable models are shown in Tables III and IV. Independent predictors of receiving 25OHD testing included older age, primary GI specialty care, lower baseline BMI (all p<0.001), and height Z-scores (p=0.008; Table III). Predictors of lipid screening included older age (p=0.02), longer follow up time (p=0.01), lower baseline height Z-score (p<0.001), and higher baseline BMI Z-score (p<0.001). Screening was significantly more common in nephrology and rheumatology populations (p<0.001; Table III). Influenza vaccination was more commonly prescribed in younger patients (p=0.008; Table IV). Analyses of patients with lipid screening and 25OHD results were performed with small variations in point estimates. The multivariate model and statistically significant associations remained unchanged and in the same direction. Independent predictors of greater PPV prescribing included younger age (p<0.001), non-private insurance (p<0.001), non-white race (p<0.001), higher baseline BMI (p=0.001), in network primary care (p<0.001), highest glucocorticoid exposure quartile (p<0.001), and specialty care in nephrology (p<0.001) or rheumatology (p=0.001). Results did not differ when we limited the analysis of PPV prescribing to children 2 years and older at baseline, but a higher percentage of those receiving primary care inside our hospital network received the vaccination (29% vs 13%, p<0.001). Given the low frequency of prescriptions, predictors of hydrocortisone prescribing were not assessed.

Table 3.

Predictors of 25-OH Vitamin D and Lipid testing in children exposed to chronic glucocorticoid therapy

	25-hydroxyvitamin D				Lipid Screening
Variable	Univariate OR (95% CI)	p	Multivariate OR (95% CI)	p	Univariate OR (95% CI)	p	Multivariate OR (95% CI)	p
Age, years(baseline)	1.11 (1.07, 1.14)	<0.001	1.10 (1.05, 1.14)	<0.001	1.03 (1.00, 1.07)	0.05	1.06 (1.01, 1.11)	0.008
Follow up duration, months	1.05 (1.03, 1.08)	<0.001	1.05 (1.03, 1.08)	<0.001	1.05 (1.02, 1.07)	<0.001	1.03 (1.00, 1.05)	0.03
Sex, Male	0.70 (0.50, 0.98)	0.04	0.55 (0.37, 0.83)	0.004	0.73 (0.53, 1.02)	0.07	--	--
Race, Non-White	0.81 (0.57, 1.15)	0.24	--	--	1.90 (1.35, 2.65)	<0.001	--	--
Hispanic Ethnicity	0.59 (0.34, 1.00)	0.05	0.49 (0.27, 0.92)	0.03	1.63 (0.96, 2.77)	0.07	--	--
Non-Private Insurance	0.70 (0.48, 1.02)	0.06	--	--	1.38 (0.96, 2.01)	0.09	--	--
Height Z-Score, SD (baseline)	0.83 (0.73, 0.95)	0.008	0.79 (0.67, 0.93)	0.004	0.77 (0.67, 0.87)	<0.001	0.70 (0.60, 0.82)	<0.001
BMI Z-Score, SD (baseline)	0.74 (0.64, 0.86)	<0.001	0.83 (0.71, 0.98)	0.03	1.37 (1.19, 1.57)	<0.001	1.25 (1.07, 1.47)	0.006
In Network Care	0.99 (0.65, 1.49)	0.96			1.57 (1.07, 2.32)	0.02	1.86 (1.17, 2.97)	0.009
GC Cumulative Dose Quartile
1	Reference		--	--	Reference	--	--	--
2	1.32 (0.82, 2.13)	0.26	--	--	1.28 (0.80, 2.03)	0.30	--	--
3	1.44 (0.89, 2.35)	0.14	--	--	1.46 (0.93, 2.32)	0.10	--	--
4	0.77 (0.49, 1.21)	0.26	--	--	0.91 (0.57, 1.48)	0.71	--	--
Subspecialty
Gastroenterology	Reference		Reference	--	Reference	--	Reference	--
Nephrology	0.39 (0.27, 0.58)	<0.001	0.39 (0.25, 0.63)	<0.001	6.68 (4.39, 10.16)	<0.001	6.85 (4.28, 10.96)	<0.001
Rheumatology	0.39 (0.25, 0.60)	<0.001	0.28 (0.16, 0.46)	<0.001	5.58 (3.54, 8.82)	<0.001	5.85 (3.52, 9.70)	<0.001

Influenza vaccination data are provided only for the 2012-2013 influenza season and for those receiving primary care in the hospital network. The multivariate model for 25OHD included: Age, Follow-Up Time, Sex, Ethnicity, Baseline Height Z-Score, Baseline BMI Z-Score, Specialty. The multivariate model for lipid screening included: Age, Follow-Up Time, Baseline Height Z-Score, Baseline BMI Z-Score, In Network, Specialty

Table 4.

Predictors of vaccination in children exposed to chronic glucocorticoid therapy

	Influenza Vaccine				Pneumococcal Polysaccharide Vaccine
Variable	Univariate OR (95% CI)	p	Multivariate OR (95% CI)	p	Univariate OR (95% CI)	p	Multivariate OR (95% CI)	p
Age, years(baseline)	0.87 (0.79, 0.97)	0.01	0.87 (0.78, 0.96)	0.008	0.91 (0.88, 0.95)	<0.001	0.94 (0.89, 0.99)	0.02
Follow up duration, months	0.97. (0.91, 1.03)	0.29	0.96 (0.89, 1.03)	0.22	1.03 (1.00, 1.05)	0.05	1.01 (0.98, 1.04)	0.52
Sex, Male	1.82 (0.77, 4.32)	0.18	--	--	1.18 (0.79, 1.76)	0.43	--	--
Race, Non-White	0.89 (0.37, 2.10)	0.78	--	--	2.78 (1.84, 4.20)	<0.001	1.72 (1.03, 2.89)	0.04
Hispanic Ethnicity	0.75 (0.18, 3.21)	0.70	--	--	0.73 (0.34, 1.58)	0.43	--	--
Non-Private Insurance	0.71 (0.30, 1.69)	0.44	--	--	2.20 (1.43, 3.39)	<0.001	--	--
Height Z-Score, SD (baseline)	0.81 (0.59, 1.12)	0.21	--	--	0.96 (0.82, 1.13)	0.62	--	--
BMI Z-Score, SD (baseline)	0.85 (0.63, 1.15)	0.30	--	--	1.37 (1.14, 1.63)	0.001	--	--
In Network Care	N/A	N/A	N/A	N/A	2.90 (1.87, 4.49)	<0.001	3.48 (1.91, 6.35)	<0.001
GC Cumulative Dose Quartile
1	Reference	--	Reference	--	Reference		Reference	--
2	0.34 (0.08, 1.42)	0.14	--	--	1.33 (0.63, 2.83)	0.46	1.18 (0.50, 2.80)	0.70
3	0.47 (0.12, 1.76)	0.26	--	--	2.18 (1.08, 4.39)	0.03	1.68 (0.75, 3.73)	0.21
4	0.5 (0.14, 1.83)	0.30	--	--	5.68 (2.97, 10.86)	<0.001	2.67 (1.24, 5.78)	0.01
Subspecialty
Gastroenterology	Reference	--	--	--	Reference	--	Reference	--
Nephrology	1.33 (0.80, 2.21)	0.27	1.14 (0.43, 3.01)	0.79	30.37 (14.82, 62.24)	<0.001	28.72 (13.34, 61.84)	<0.001
Rheumatology	0.70 (0.36, 1.38)	0.31	2.06 (0.38, 11.05)	0.40	4.18 (1.77, 9.90)	0.001	4.77 (1.92, 11.84)	0.001

The multivariate model for influenza vaccine included: Age, Follow-Up Time, Specialty. The multivariate model for pneumococcal polysaccharide vaccine: Age, Follow-Up Time, Race, In Network, GC Cumulative Dose Quartile, Specialty

DISCUSSION

Despite known risks of chronic glucocorticoid therapy in children, we identified deficits across preventive care domains, and significant variation in vaccine administration and screening, vitamin D deficiency and dyslipidemia according to pediatric specialty.

Bone fragility is a common consequence of chronic glucocorticoid therapy during childhood. Among children receiving chronic glucocorticoid therapy for rheumatologic conditions, vertebral compression fractures were observed in 12% over a three-year period, with a peak incidence in the first year(13). At the time our data were collected, dual x-ray absorptiometry (DXA) was recommended by the International Society for Clinical Densitometry as screening to assess for osteoporosis and, consequently, the risk of vertebral fracture. However, the current position statement points out that low bone density may not be predictive of all fractures and therefore does not recommend the isolated use of DXA to identify fracture risk(14). Screening for vitamin D deficiency was more common.

It is well recognized that atherosclerosis begins during childhood, and that chronic inflammatory diseases are associated with accelerated cardiovascular disease(2). In fact, childhood chronic inflammatory diseases are considered “special risk conditions” requiring a risk-stratified approach to the management of known cardiovascular risk factors. For example, children with these conditions have different BMI, blood pressure, low-density lipoprotein, and triglyceride level targets than healthy children. The low proportion of patients receiving lipid screening suggests that comprehensive assessment and management of cardiovascular risk factors in children receiving chronic glucocorticoid therapy is uncommon. A recent study showed that primary care physicians understand the importance of lipid screening during childhood, but only 16% perform universal screening and 50% perform selective screening(8). We found that older age, higher baseline BMI Z-scores, and lower baseline height Z-scores were associated with lipid screening. Similar to primary care physicians, specialists are screening children at a perceived higher risk of dyslipidemia.

Children receiving chronic glucocorticoid therapy are at risk for acute adrenal insufficiency due to hypothalamic-pituitary-adrenal axis suppression. Among other indications, stress dosing is required for periods of vomiting or acute febrile illnesses. Although rare, significant morbidity and even mortality is observed if stress dosing is withheld(15). Endocrinologists routinely prescribe intramuscular hydrocortisone and provide patient education regarding its appropriate emergency use. In our cohort, intramuscular hydrocortisone prescriptions were rare among children treated with chronic glucocorticoid therapy and not followed by an endocrinologist. Given the consequences of acute adrenal insufficiency and the challenges related to consistent stress dosing across care settings, we recently published a clinical pathway to standardize the approach at our institution(16).

Pneumococcal and influenza vaccination data in our cohort provides an important affirmation regarding the effectiveness of quality improvement strategies. The most consistent pneumococcal vaccination occurred in children with renal diseases. During the observation period, the Division of Nephrology performed a quality improvement project to optimize pneumococcal vaccination. Similarly, there is a hospital-wide influenza vaccination quality improvement effort that is likely responsible for the consistently high influenza vaccination use across subspecialties.

Even though our inclusion criteria required a minimum of 15 days of glucocorticoid exposure, the median exposure duration was 143 days. Although conducted in a single health system, our results are consistent with other studies. For example, in a cohort of children with human immunodeficiency virus, cystic fibrosis, liver transplantation and diabetes mellitus, fewer than 25% received pneumococcal vaccination and 21-90% received seasonal influenza vaccination with significant variation between each disease(10). In children with SLE, a high degree of variation existed in preventive care performance at centers in the United States, Brazil, and India.(17) In general, larger centers more consistently fulfilled the SLE quality indicators. Quality metrics have been scrutinized in ambulatory care delivery. When examined in this manner there remains a pervasive underutilization of screening, diagnosis, treatment, and follow up in children with both acute and chronic medical conditions(18). Thus, the variation in preventive care in children receiving chronic glucocorticoid therapy at our institution is likely not a unique problem.

It is critical to consider the strength of the recommendations driving our chosen preventive measures. The Infectious Disease Society of America strongly recommends for both pneumococcal and seasonal influenza vaccination in at risk patients and considers the evidence base moderate(3). Although influenza vaccination is more commonly implemented, both recommendations are generally accepted. The Endocrine Society strongly recommends screening for 25OHD deficiency on the basis of a high-quality evidence base(1). Similarly, the American Academy of Pediatrics recommends lipid screening for all children with moderate- or high-risk conditions on the basis of Grade B evidence (eg, randomized controlled trials with minor limitations or overwhelmingly consistent evidence from observational studies)(2). Although there are no specific guidelines for stress dose steroid prescribing, the Lawson Wilkins Pediatric Endocrine Society Drug and Therapeutics Committee published an expert opinion manuscript(4) with specific suggested actions including “Instruct the family on the use of intramuscular hydrocortisone sodium succinate in case of vomiting illness or severe stress.” Although this is not a formal guideline, it addresses a potentially life-threatening and often overlooked patient safety issue(15).

Providers may not prescribe indicated interventions based on knowledge deficits, attitudes, and external barriers(19). For example, knowledge deficits include lack of awareness regarding a patient’s risk or confusion regarding guideline content. Sometimes, a long latency between medication prescribing and adverse events or a low event frequency minimizes perceived risk. Attitude barriers include controversy regarding guideline content. External barriers include a lack of time or physician continuity, language barriers, access to medications or equipment, insurance denials, and patient adherence to recommendations(20). Barriers to implementation are likely multifactorial and both intervention- and population-specific strategies will be required.

An increasingly common method for improving complex care practice is through quality bundles. This strategy has been shown to streamline quality processes in sepsis management,(22) care of central lines, and catheter associated urinary tract infections(23-27). Additionally, electronic health record optimization is an efficient way to implement care bundles and can overcome gaps in patient identification and provider knowledge. Computerized preventative care reminders,(28, 29) clinical decision support (30) and knowledge management(31) have all been shown to identify and improve preventative screening. However, the unique needs of each patient population and the clinical setting should be considered prior to identifying appropriate interventions. For example, patients with steroid-sensitive nephrotic syndrome may be at lower risk for low bone mass relative to other patients with inflammatory conditions associated with chronic glucocorticoid therapy(32, 33). Therefore, differences in diagnostic properties may have influenced the consistent use of screening tests in our patient population. Moreover, the cost versus benefit of interventions should also be considered for each patient.

Our study has limitations. First, we did not find an association between glucocorticoid exposure quartile and higher rates of preventive care measure implementation. It is possible that the cumulative glucocorticoid dose may not be a driving factor for preventive care measure implementation. An alternative explanation may be that other disease-specific factors may be driving preventive care measure use. However, there are clear deficits in pneumococcal vaccination and stress dose steroid prescribing that cannot be explained by disease category. Second, glucocorticoid prescribing is not consistent across specialists. A common practice is to provide a greater number of tablets or volume of medication than is needed which could cause an overestimation of glucocorticoid exposure. However, we performed a manual validation that demonstrated high sensitivity and specificity of our automated health record search. Additionally, patients in the cohort had the expected changes in BMI and height Z-scores characteristic of children receiving glucocorticoid therapy(34). Third, the majority of our cohort received primary care outside our health network and may have had preventive care not captured in our electronic health record. Therefore, we reported influenza data only for children with primary care in network. We were not able to distinguish if lipid screening and PPV administration were higher in network due to standard documentation in the health record rather than attention to preventive care. However, we included all patients in the lipid screening and PPV analysis because these interventions are potentially more likely to occur in the subspecialty setting. Finally, our study only assessed process metrics. Future studies are needed to assess whether implementation of preventive care guidelines improves outcomes in children receiving chronic glucocorticoid therapy.

Abbreviations

25OHD

25-hydroxyvitamin D

DXA

Dual X-ray Absorptiometry

PPV

Pneumococcal Polysaccharide Vaccine