OBJECTIVES:

Despite 2011 guidelines in which it is suggested that treatment of acute immune thrombocytopenia purpura (aITP) is not needed for patients without significant bleeding, only 14% of children treated for aITP have bleeding symptoms. Our aim was to decrease the percentage of children with first-episode aITP who were unnecessarily treated by 50% within 12 months of guideline implementation.

METHODS:

An intervention was designed by using the precaution-adoption-process model. A standard-of-practice meeting was organized and focused on clinician readiness for change. After education on current evidence and common cognitive errors, consensus clinical guidelines were created. After implementation, an article in a statewide professional newsletter was published to educate community providers. Unnecessary treatment (UT) was defined as treatment of any patient who only had bruising and/or self-resolving nose bleeds. Statistical process control charts were used to track progress, midline shifts were determined by Nelson’s rules, and hospital costs were derived from administrative billing data.

RESULTS:

One hundred children with aITP were seen from January 2013 to September 2018. UT decreased from 70% to a sustained rate of <30% (P = .008), including a mean of 7% over the past 12 months. The admission rate decreased from 100% to 52% (P = .013), and the total percentage of patients treated decreased from 100% to 48% (P = .016), with both numbers continuing to decline. No adverse bleeding events occurred. An estimated 12 admissions, 4 readmissions, and 5 adverse events were avoided annually.

CONCLUSIONS:

We demonstrated successful improvement in UT of aITP through an educational intervention informed by the precaution-adoption-process model change theory.

Acute immune thrombocytopenic purpura (aITP) is a common condition treated by hematologists that is rarely associated with clinically significant bleeding.1  The American Society of Hematology (ASH) stated that treatment of aITP need not be based solely on platelet counts.2  This recommendation was further supported by a recent randomized controlled trial’s data confirming the safety of observation for low-risk patients who have only bruising and self-limited nose bleeds.3  Despite growing evidence in support of observation, clinicians continue to frequently treat patients who lack bleeding symptoms. There are rare single-center examples of successful reduction in aITP treatment by changing inpatient provider practice.4,5  However, these success stories have not led to broad practice change. Pediatric Health Information System (PHIS) data revealed that >90% of children with aITP presenting to freestanding children’s hospitals are treated despite only 14% having bleeding symptoms. Witmer et al6  found no statistical change in treatment and only a modest decrease in unnecessary bone marrow biopsies and length of stay 3 years after publication of the ASH recommendations.

Our center is a tertiary children’s hospital within a larger academic medical center. We have a pediatric emergency department (ED) with 20 000 visits per year and a dedicated pediatric hematology clinic, which had 7 attending hematologists and 3 fellows during the study period who all see benign hematology, including aITP. Similar to the PHIS data, we observed that our providers were treating almost 100% of aITP patients despite <20% having bleeding symptoms. Our aim was to reduce the rate of unnecessary treatment (UT) of aITP by 50% within 12 months after a multistep single intervention by using a theoretical model for change and by educating our community pediatricians.

Our interventions were designed on the basis of the precaution-adoption-process model (PAPM). PAPM is used to recognize that individuals exist on a continuum of readiness to act in accordance with the proposed change. Interventions should be designed to target individuals along the continuum.7  In our interventions, we targeted clinicians along the PAPM spectrum, from those who were unaware that UT was a problem to those who were aware, had changed their practice, and were interested in engaging in broader practice change (Fig 1). Our intervention was designed to address clinicians at their level of readiness for change.

FIGURE 1

PAPM of the stages of change readiness, the interventions focused on each stage of change readiness, and each provider’s stage before and after the intervention.

FIGURE 1

PAPM of the stages of change readiness, the interventions focused on each stage of change readiness, and each provider’s stage before and after the intervention.

Close modal

We convened an aITP standard-of-practice conference in September 2015, with attendance from most of the pediatric hematology and oncology division. For clinicians in stage 1, we presented the evidence for and against treatment of aITP as well as the ASH treatment guidelines recommending against treatment of low-risk patients. To engage clinicians in stages 2 and 3, we focused on 2 common misconceptions that make UT more likely: (1) overestimation of the clinical benefit of treating aITP and (2) lack of knowledge of errors of commission as an important component of any clinical decision-making framework.8  To address the first misconception, we went through a number-needed-to-treat exercise. Because data revealing that aITP treatment reduced the risk of intracranial hemorrhage were lacking, the group was asked to optimistically estimate the risk reduction from treatment. On the basis of the estimates of the faculty, the number needed to treat was determined to be 330 to 667 admissions for treatment to prevent 1 intracranial hemorrhage. To increase clinician understanding of errors of commission, we reviewed cases from our center that revealed the adverse events associated with treatment with both intravenous immunoglobulin and Rho(D) immunoglobulin. We discussed common adverse events, such as aseptic meningitis after intravenous immunoglobulin treatment and hyper-hemolysis after Rho(D) immunoglobulin treatment. We also reviewed financial toxicity that resulted in these cases, with >$10 000 in unnecessary health care costs in 1 example. After presenting this information, a discussion period occurred to allow clinicians, particularly those in stage 3, to address any concerns or continued misconceptions. Stage 4 (not included in Fig 1) would have included clinicians who decided not to act, whom we did not encounter. Clinicians in stages 5 and 6 were engaged by using the Ikea effect to increase engagement by participating in the creation of a new aITP clinical care guideline. In the Ikea effect, it is suggested that an individual is more likely to engage in change if they are part of the change process.9  No clinicians were in stage 7 (routinely avoiding UT) before the intervention. Two draft guidelines were created during the standard-of-practice conference: 1 that did not include platelet count in the decision to treat and another that allowed for treatment of platelet counts of <20/μL at the clinicians’ discretion. The guideline that did not include platelet count in the treatment decision was unanimously chosen.

After the conference, a period for open comments allowed all involved parties to suggest changes. Modest changes were made to the wording, but no changes were made to the clinical workup or treatment choices. The final treatment guideline was distributed to the clinic, inpatient unit, and ED (Fig 2). Unit education included posting guidelines in clinical areas, e-mail reminders, reeducation at a division meeting, and inclusion of the guideline in our resident manual. The guideline was rolled out in November 2015.

FIGURE 2

aITP treatment pathway: consensus aITP evaluation and treatment algorithm. a Concerning bleeding symptoms include the following: gum bleeding, prolonged nose bleeds, nose bleeds requiring intervention, hematuria, and gastrointestinal bleeds. ANA, antinuclear antibody; CBC, complete blood cell; DAT, direct antiglobulin test; EBV, Epstein-Barr virus; IgA, immunoglobulin A; IgG, immunoglobulin G.

FIGURE 2

aITP treatment pathway: consensus aITP evaluation and treatment algorithm. a Concerning bleeding symptoms include the following: gum bleeding, prolonged nose bleeds, nose bleeds requiring intervention, hematuria, and gastrointestinal bleeds. ANA, antinuclear antibody; CBC, complete blood cell; DAT, direct antiglobulin test; EBV, Epstein-Barr virus; IgA, immunoglobulin A; IgG, immunoglobulin G.

Close modal

During the standard-of-practice conference, clinicians expressed concerns that families would receive conflicting messages from their general pediatricians and our team if they continued to be referred to us for urgent treatment. To address those concerns, after implementation of the guideline, an article in which the evidence behind our practice change was detailed was featured in the American Academy of Pediatrics Virginia Newsletter and published in January 2016. The article was used by community pediatricians to help reframe their expectations of the care that their patients would receive and appropriately counsel families when a diagnosis of aITP was made. An e-mail and verbal reminder of our practice guidelines were sent to our division in April 2017 when a trend toward an increase in UT was observed.

UT was defined as treatment of any low-risk patient diagnosed with aITP at initial presentation to either our ED or clinic or transferred from a referring facility whose symptoms were limited to bruising and/or self-resolving nose bleeds lasting <15 minutes. Children with prolonged nose bleeds, mucosal bleeding, hematuria, hematemesis, bloody stools, or intracranial hemorrhage were excluded from the low-risk group. Social concerns were included in our guideline as a reason to treat; however, because of inconsistency in charting, social factors were not considered as part of the UT definition. Statistical process control charts were used to track progress, and midline shifts were determined by Nelson’s rules. A P-chart was used given that our primary outcome data were binary attribute data. Given that aITP is an infrequent diagnosis, a G-chart was also created to analyze our performance in real time. Patient charges and hospital costs were derived from patient-level administrated billing data. This work met institutional criteria for quality improvement and was exempt from ethics review.

A total of 100 children with initial presentation of aITP were seen in either our ED or clinic or transferred from a referring facility from January 2013 to September 2018. As detailed in Fig 3, the 12-month rolling percentage of unnecessary aITP treatment decreased from 70% to a sustained rate <30% (P = .008), with a mean of 7% over the past 12 months. Improvement in UT was confirmed by a G-chart revealing an increase from 0.3 to 4.3 patients who were appropriately treated per case of inappropriate treatment (Fig 4). The admission rate for children with aITP decreased from 100% to 52% (P = .013) and the total percentage of patients treated decreased from 100% to 48% (P = .016), with both numbers continuing to decline. The rate of treatment-related adverse events decreased from 39% to 22% (P > .05), and readmissions resulting from adverse events decreased from 8% to 0% after implementation (P = .05). No adverse bleeding events occurred. An estimated 12 admissions, 4 readmissions, and 5 treatment-related adverse events were avoided annually. Annual savings from prevented admissions, readmissions, and excess treatment is estimated to be $18 000 to $220 000 in patient charges and $75 000 to $90 000 in hospital costs.

FIGURE 3

Percent of inappropriate aITP treatment: P-chart indicating the 12-month rolling percentage of children with aITP who received UT. The solid line indicates the mean percent of UT, and dotted lines represent upper and lower control limits at 3 SDs from the mean. AAP, American Academy of Pediatrics.

FIGURE 3

Percent of inappropriate aITP treatment: P-chart indicating the 12-month rolling percentage of children with aITP who received UT. The solid line indicates the mean percent of UT, and dotted lines represent upper and lower control limits at 3 SDs from the mean. AAP, American Academy of Pediatrics.

Close modal
FIGURE 4

G-chart showing the number of appropriately managed aITP cases between each case of IT. AAP, American Academy of Pediatrics; IT, inappropriate treatment; UCL, upper control limit.

FIGURE 4

G-chart showing the number of appropriately managed aITP cases between each case of IT. AAP, American Academy of Pediatrics; IT, inappropriate treatment; UCL, upper control limit.

Close modal

We have demonstrated a significant decrease in UT of children with aITP seen in our center. Compared with national data, this represents a significant improvement. PHIS data reveal little change in practice since the 2011 ASH guidelines were published. Although 2 other centers have reported a similar local practice change, our approach to practice change was unique in several ways. First, we used the PAPM for change, an established framework to assess and address change readiness. It has been used in a variety of health care settings, including global pediatric care delivery10  and bone health,11  but examples of its use remain sparse. The PAPM framework was used to improve the design of the intervention by intentionally engaging clinicians across the PAPM spectrum. Second, within that framework, we recognized an important gap in clinician knowledge of errors of commission in their treatment decisions. Errors of commission are commonly underweighted by clinicians, leading to clinicians misjudging the risks and benefits and ultimately putting patients at risk for overtreatment.8  Additionally, financial toxicity12  is increasingly recognized as an important adverse event that needs to be included in the discussion of errors of commission.

Finally, we added a component of community primary care provider education. Our providers identified the difficulty for parents to accept our recommendation against admitting and treating their children when a trusted community pediatrician urgently sends them to our ED. For some patients, ED referrals remain warranted if they have bleeding symptoms or to rule out more serious causes of thrombocytopenia. In those cases, well-educated community pediatricians are able to accurately council families that they may not require admission or treatment. For many patients, community pediatricians can safely refer patients to our outpatient clinic or even manage children in their clinics with help as needed via our subspecialty consult line. We do not track our consult line call volume or reasons for calls. As such, we believe that our data likely underestimate the effect of our intervention. We did see a trend toward more of our ED referrals and admissions having bleeding symptoms that justify treatment (24% vs 52%), suggesting that some low-risk patients are being managed in the community. It is also important to note that the ASH recommendations do not expressly state that patients should never be treated on the basis of platelet count alone. There are many patient-specific scenarios in which treatment may be reasonable. As a result, decreasing the rate of UT, as we define it, to 0 is not practical nor does it allow for shared decision-making, which we support.

In another important single-center study, Labrosse et al4  also showed a decreased admission rate through standardization of care, but their design and treatment decisions were different in several important ways. First, they discussed using preprinted order sheets to guide management but did not otherwise focus their article on detailing their implementation process. Second, their management differed in 2 important ways. They recommended treating children <3 years of age in their guideline and, as a result, did not see a change in management in that age group. They also recommended steroid treatment of mild bleeding symptoms and of patients with a platelet count of <10 × 109/L. Our division opted to avoid steroids as first-line management and did not set a minimum platelet threshold for treatment. Finally, methodologically, they attempted to apply Buchanan bleeding scores to the patients during the retrospective chart review, whereas we chose a simpler approach to defining UT that mirrored our treatment guideline.4 

There were several barriers to implementation. The most significant barrier was physician past experience. The individual most resistant to change had cared for a patient with intracranial hemorrhage early in their career and was, therefore, most reluctant to change practice. Another barrier was how to navigate the conversation when patients with aITP were sent to the ED having been told that they needed to be treated. We had internal discussions about how to honestly but accurately convey risk but continue to have variable success in those situations. Finally, sustainability is always a barrier to consistent implementation of quality improvement work. We did see a spike in UT corrected by additional education. The culture of treatment of aITP has changed over time; however, electronic tools would likely have expedited the culture change.

Our work has several limitations. First, at the time the work was initiated, we had 7 attending physicians and 3 fellows caring for these patients. Our ability to understand the providers’ stage of readiness for change and our ability to move providers along the continuum was enhanced by close relationships within the provider group. This fact may limit generalizability to larger programs. Second, we are unable to quantify the effect of our community education. Although we have anecdotal evidence that the article was widely disseminated and well-received, we were unable to assess knowledge or readership directly. As such, the relative impact that community provider education, as a component of our intervention, had on our results cannot be ascertained. Finally, our cost analysis is limited. We used patient-level administrative data to determine charges, but in prospectively estimating savings to patients, we assume that the payer mix will remain constant and that payer reimbursement does not change. Hospital savings include only direct costs and do not include costs of paying laboratory technicians. aITP does not occur frequently enough for our intervention to alter staffing.

The first and most important future direction is to decrease the rate of UT further. As mentioned above, a rate of 0 is not our goal; however, UT remains a problem. Potential strategies include a shared decision-making tool to help guide the discussion between the clinician and family about the risks and benefits of aITP treatment. A Choosing Wisely alert may also be built into the medical record when ordering intravenous immunoglobulin or Rho(D) immunoglobulin for patients. Integration into the e-health record would aid in the sustainability of our current UT rates if not drive them lower. We would also like to be able to collect data on phone referrals to better understand the volume of aITP care that has been diverted away from our ED and clinic. Finally, multicenter work is needed to move the needle on unnecessary aITP care nationally.

We demonstrated successful improvement in UT of aITP through an educational intervention informed by the PAPM change theory.

We thank our hematology staff, ED staff, and community pediatricians for their hard work. We also thank Dr Patrick Brady for providing statistical process control expertise.

Dr Schefft conceptualized and designed the study, collected data, conducted the initial analysis, and drafted the initial manuscript; Drs Gowda and Massey supervised the design, contributed to the analysis, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted.

FUNDING: No external funding.

1
Neunert
C
,
Noroozi
N
,
Norman
G
, et al
.
Severe bleeding events in adults and children with primary immune thrombocytopenia: a systematic review
.
J Thromb Haemost
.
2015
;
13
(
3
):
457
464
2
Neunert
C
,
Lim
W
,
Crowther
M
,
Cohen
A
,
Solberg
L
 Jr
,
Crowther
MA
;
American Society of Hematology
.
The American Society of Hematology 2011 evidence-based practice guideline for immune thrombocytopenia
.
Blood
.
2011
;
117
(
16
):
4190
4207
3
Heitink-Pollé
KMJ
,
Uiterwaal
CSPM
,
Porcelijn
L
, et al
;
TIKI Investigators
.
Intravenous immunoglobulin vs observation in childhood immune thrombocytopenia: a randomized controlled trial
.
Blood
.
2018
;
132
(
9
):
883
891
4
Labrosse
R
,
Vincent
M
,
Nguyen
UP
,
Chartrand
C
,
Di Liddo
L
,
Pastore
Y
.
Using a standardised protocol was effective in reducing hospitalisation and treatment use in children with newly diagnosed immune thrombocytopenia
.
Acta Paediatr
.
2017
;
106
(
10
):
1617
1623
5
Schultz
CL
,
Mitra
N
,
Schapira
MM
,
Lambert
MP
.
Influence of the American Society of Hematology guidelines on the management of newly diagnosed childhood immune thrombocytopenia
.
JAMA Pediatr
.
2014
;
168
(
10
):
e142214
6
Witmer
CM
,
Lambert
MP
,
O’Brien
SH
,
Neunert
C
.
Multicenter cohort study comparing U.S. management of inpatient pediatric immune thrombocytopenia to current treatment guidelines
.
Pediatr Blood Cancer
.
2016
;
63
(
7
):
1227
1231
7
Weinstein
ND
.
The precaution adoption process
.
Health Psychol
.
1988
;
7
(
4
):
355
386
8
Schroeder
AR
,
Redberg
RF
.
The harm in looking
.
JAMA Pediatr
.
2013
;
167
(
8
):
693
695
9
Marsh
LE
,
Kanngiesser
P
,
Hood
B
.
When and how does labour lead to love? The ontogeny and mechanisms of the IKEA effect
.
Cognition
.
2018
;
170
:
245
253
10
Weaver
MS
,
Heminger
CL
,
Lam
CG
.
Integrating stages of change models to cast new vision on interventions to improve global retinoblastoma and childhood cancer outcomes
.
BMC Public Health
.
2014
;
14
:
944
11
Elliott
JO
,
Seals
BF
,
Jacobson
MP
.
Use of the precaution adoption process model to examine predictors of osteoprotective behavior in epilepsy
.
Seizure
.
2007
;
16
(
5
):
424
437
12
Nathan
PC
,
Henderson
TO
,
Kirchhoff
AC
,
Park
ER
,
Yabroff
KR
.
Financial hardship and the economic effect of childhood cancer survivorship
.
J Clin Oncol
.
2018
;
36
(
21
):
2198
2205

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.