Risperdal® CONSTA® Needle Detachment. Incidence Rates Before and After Kit Redesign: A Retrospective Study using Electronic Health Records and Natural Language Processing in the Department of Veterans Affairs
verfasst von:
Marsha A. Wilcox, Danielle Coppola, Nicole Bailey, Andrew Wilson, Aaron W. C. Kamauu, Patrick R. Alba, Olga V. Patterson, Benjamin Viernes, Daniel W. Denhalter, Ira D. Solomon, Scott L. DuVall
Janssen received reports of needle detachments for Risperdal® CONSTA® and, in response, redesigned the kit.
Objective
The study objective was to estimate the rate of Risperdal® CONSTA® needle detachments prior to and after the introduction of a redesigned kit.
Methods
This retrospective study used record abstraction in the US Department of Veterans Affairs (VA). The 3 phases included: (1) a pilot study for methods evaluation in a sample of 6 hospitals with previously reported detachments; (2) a baseline study to ascertain the baseline detachment rate; and (3) a follow-up study to ascertain the rate for the redesigned kit. Administrative codes and natural language processing with clinical review were used to identify detachments.
Results
Pilot: we identified a subset of spontaneously reported detachments and several previously unreported events. In the baseline study (original device), from January through December 2013, 22 needle detachments were identified among 47,934 administrations of the drug in a census of administrations in the VA; an incidence of 0.0459%. In the follow-up study (redesigned device), from December 2015 through December 2016, there were 14 reported detachments in 41,819 injections, 0.0335%. This represents a reduction of 27% from the baseline.
Conclusion
This approach enabled us to identify needle detachments we would not have otherwise found (“solicited”). However, it likely resulted in incomplete outcome ascertainment. While this may have resulted in lower overall rates, it did not bias the comparison of the baseline and follow-up studies. The results showed that the redesigned Risperdal® CONSTA® kit reduced the incidence of needle detachment events in the VA.
Antipsychotic medications are given to treat schizophrenia and the manic symptoms of bipolar disorder. Schizophrenia is a chronic, severe, disabling brain disorder that affects approximately 1% of the population worldwide. It is a heterogeneous clinical syndrome with a varying constellation of signs and symptoms, including delusions, hallucinations, disorganized speech and/or behavior, and negative symptoms. Men and women are equally affected. Onset in the teens and twenties produces lifelong impairment in all areas of brain functioning, including social interactions, motivation, cognition, emotions, and senses [1, 2]. Even with treatment, most patients remain at least partially disabled. Bipolar disorder is a lifelong episodic illness that also affects about 1% of the population worldwide. It is characterized by rapid changes in mood and behavior, which are a leading cause of disability among young people. Bipolar I disorder, which is characterized by impairing manic episodes of elevated mood and increased motor drive, also affects men and women equally [2, 3].
Typical antipsychotics target the dopaminergic system, a strategy that hasn’t changed since the 1950s [4]. Atypical antipsychotics, such as risperidone, selectively inhibit dopamine D2 and serotonin receptors. Risperdal® CONSTA® is a formulation of risperidone in a long-acting injection (risperidone LAI) given every 2 weeks that is indicated for the treatment of schizophrenia and schizoaffective disorder as monotherapy and for adjunctive maintenance treatment of bipolar I disorder to delay the occurrence of mood episodes [5, 6]. Long-acting injectable antipsychotics have demonstrated improved adherence, less variability in drug plasma concentrations, and increased frequency of interaction with the treatment team, as they are administered by a clinician [6]. Risperidone LAI was first approved in Germany in 2002, then for schizophrenia in the United States in October 2003, and for bipolar I disorder in May 2009 [5].
Anzeige
Regulatory Background
No drug–device combination system is failsafe. In 2009, Janssen became aware of potential needle detachment issues concerning Risperdal® CONSTA®. Spontaneous reports of needle detachment were received via Janssen’s worldwide spontaneous adverse event reporting and product quality complaint (PQC) systems from health care providers (HCPs) and patients. 439 reports of needle detachments were made to Janssen worldwide in the calendar year 2013. In the same year, 6,738,526 kits were distributed. Thus, the reporting rate for 2013 was 0.0065% or 65 PPM (parts per million).
As a result, Janssen made incremental improvements to the componentry in order to reduce the risk of needle detachment (Fig. 1). The development of a second-generation kit was initiated in 2011, and consisted of improvements to several device components, testing, and new instructions for use (IFU) (Fig. 2). In 2014, the second-generation kit was approved by regulatory authorities and introduced into commercial distribution in the US in August 2015. As of December 2015, there were no first-generation kits on the market.
Fig. 1
Changes made to the Risperdal® CONSTA® kit
Fig. 2
Instructions for use of the second-generation kit
×
×
Drug Product Kit Device and Improvements
Registered and approved globally, Risperdal® CONSTA® is a controlled release formulation of the drug risperidone. It is injected intramuscularly as a powder (risperidone-containing microspheres) in suspension. Four dose strengths are available in the US: 12.5 mg, 25 mg, 37.5 mg, and 50 mg risperidone per vial. The finished drug product is provided as a kit containing the following components: a vial filled with risperidone extended release microspheres for injection, a prefilled syringe containing 2 mL of diluent for reconstitution of the microspheres prior to injection, an Alaris™ SmartSite® Needle-Free Vial Access Device (manufactured by CareFusion, formerly Cardinal Health) used for vial access and powder reconstitution, a 0.9 mm × 50 mm (2-in) 20G TW gluteal injection needle with a Needle-Pro® Safety Device, and a 0.8 mm × 25 mm (1-in) 21G UTW deltoid injection needle with a Needle-Pro® Safety Device (both injection needles with the Needle-Pro® Safety Device are manufactured by Smiths Medical North America).
The Alaris™ SmartSite® device and the injection needles with the Needle-Pro® Safety Device are all standard medical devices that are already marketed separately by their respective manufacturers but are included in the Risperdal® CONSTA® finished drug product kit specifically for the reconstitution and administration of the drug product. Syringe barrels, OVS (which is a German abbreviation for the Vetter tamper-evident seal) and the rubber stoppers are also externally purchased from qualified suppliers. All primary packaging materials undergo incoming release testing supported by a supplier certificate of analysis.
The Microlance® gluteal needle is directly shipped from Becton, Dickinson and Company Singapore to Smiths Medical (USA) for assembly into the Needle-Pro® Safety Device. The molded Needle-Pro® Safety Device is assembled with the Microlance® gluteal needle and then packaged into a single unit blister. The complete blister is then sterilized with ethylene oxide. A similar process is followed for the Nipro® deltoid injection needle. The Alaris™ SmartSite® needle-free access device is supplied directly in its sterile blister to the final kit assembly packaging site. Incoming release testing of the single components and the needle blister is performed by the Global Pharmaceutical Supply Group (GPSG). The final kit packaging operation is performed at either GPSG’s Ortho–McNeil packaging facility in Raritan, NJ, USA or by a contract secondary packager, Anderson Packaging in Illinois, USA. Figure 1 shows changes made from the original to the second-generation kit.
The Risperdal® CONSTA® kit consists of components that are assembled and reconstituted prior to intramuscular injection. The administration process consists of several steps that are performed to reconstitute the drug, assemble the syringe, administer the drug, and safely dispose of the kit (Fig. 2).
Anzeige
Administrative Codes
Administrative codes are used to identify both the drug and administration (procedure) in electronic health records (EHRs) and medical claims data. The Healthcare Common Procedure Coding System (HCPCS) is a standardized coding system, maintained by the American Medical Association, to ensure consistent and orderly processing of healthcare claims [7]. HCPCS Level I consists of Current Procedural Terminology (CPT) using 5-digit codes to describe procedures and services provided by healthcare professionals.
HCPCS Level II, maintained and distributed by the Centers for Medicare and Medicaid Services (CMS), is used to identify healthcare products, supplies and services that fall outside the scope of CPT codes. J-codes are permanent codes assigned to drugs administered via a route other than oral, and for chemotherapy agents [8]. J-codes were used to identify the drug used in Risperdal® CONSTA® injections.
National Drug Codes (NDCs) are universal product identifiers for human drugs and are published by the US Food and Drug Administration (FDA). These unique 3-segment codes describe the drug manufacturer, the specific product, and the package size, and are used on medical claims submitted to public and private payers [9].
The Department of Veterans Affairs (VA) serves as both provider and payer for US veterans. When care is received at a VA facility, the VA is authorized to bill third-party payers (TPPs) and pharmacy benefits managers (PBMs) to recoup the cost of care, including medication administration. The VA uses NDC codes to identify medications dispensed in these claims [10].
Medications prescribed and administered to veterans by VA providers for service-connected conditions are filled almost exclusively at VA pharmacies, and although there is no claim generated from these transactions, the medication data, including the applicable HCPCS and NDC codes, are stored in the VA electronic medical record and are accessible through the VA Corporate Data Warehouse [11].
Spontaneous vs. Solicited Reports
Spontaneous reports are made by a variety of reporters either to the company or to the FDA. Although spontaneous reporting systems are in place and routinely assessed, they do not represent actual rates of risk. There is no real denominator, number of injections in this case, with which to estimate an incidence rate. Further, spontaneous reports likely represent only a small fraction of actual events. The degree to which reporting rates reflect actual event rates is unknown.
Solicited reports, on the other hand, are systematically acquired using methods in which the sampling frame is known. Importantly, there is a reliable denominator with which to calculate the incidence rate. The rate of solicited reports will almost certainly be higher than the rate for spontaneous reports. Wernicke et al. [12] compared spontaneous and solicited reporting rates and showed higher solicited rates. Notably, this study is based on solicited rather than spontaneous reports.
Aims
The primary aim of this study was to compare the needle detachment incidence rate for the original Risperdal® CONSTA® drug–device kit with the rate for the redesigned kit.
Anzeige
The objectives were to:
1.
Determine the feasibility of identifying needle detachments in electronic health records (pilot study)
2.
Estimate the baseline incidence rate of needle detachments for the first-generation Risperdal® CONSTA® drug–device combination kit during the calendar year 2013 (baseline study)
3.
Estimate the incidence rate of needle detachments in the redesigned Risperdal® CONSTA® drug–device combination kit in the period from December 2015 through December 2016, after first-generation kits were no longer in use (follow-up study)
Methods
Design
This was a retrospective comparative cohort study of Risperdal® CONSTA® needle detachments.
This study was conducted in 3 phases: (1) a pilot study to evaluate the methods, (2) a baseline study to establish the needle detachment incidence rate for the original kits, and (3) a follow-up study to establish the incidence rate for the redesigned second-generation kits.
An electronic health record (EHR) review was selected because there is no code for needle detachments in administrative data (i.e., claims). A detachment may be noted in the medical record text field as part of the normal documentation of care.
Anzeige
Participants and Setting
The study was conducted in the US Department of Veterans Affairs among patients who received at least one Risperdal® CONSTA® injection [13]. The pilot study was conducted in the 6 VA hospitals from which reports of detachments had been received (one from each hospital). The baseline study (calendar year 2013) and follow-up study (December 2015 to December 2016) cohorts were patients who had received at least one Risperdal® CONSTA® injection in the entire VA healthcare system.
Electronic health records for this study were obtained from the VA Corporate Data Warehouse (CDW) and analyzed in the VA Informatics and Computing Infrastructure (VINCI) [11, 14].
No primary patient recruitment or direct patient interaction was included in this work. This study received institutional review board (IRB) and other Veterans Affairs organizational approvals before initiation.
Intervention
The Risperdal® CONSTA® kit was redesigned and re-introduced into commercial distribution in the US in 2015. The redesign included improvements to device components and new instructions for use (Fig. 1).
Anzeige
Outcome
The primary outcome was the difference in needle detachment incidence rates comparing the baseline and follow-up cohorts.
Pilot Study Methods
The single goal of the pilot study was to evaluate the feasibility of the method. We compared detachments identified with the natural language processing (NLP)-assisted chart review method (solicited reports) with those obtained via the spontaneous product quality complaint (PQC) reporting system in order to answer the question, “Can we find detachments about which we have been notified?”
Outcome Definition
Needle detachment events were identified by examining records of patients who had received at least one Risperdal® CONSTA® injection during the observation window. Both procedure and drug codes were used to identify an injection. This was followed by a NLP-assisted manual review of notes in temporal proximity to the injection.
Administrative Codes
Two sets of codes were used to identify administrations of Risperdal® CONSTA®. The National Drug Codes (NDCs; 11-digit universal product identifiers for human drugs) used were 50458-306-11, 50458-307-11, 50458-308-11, 50458-309-11, and all NDCs for Risperdal® CONSTA® (1 kit in 1 box, 2 mL in 1 syringe, 2 mL in 1 vial, provided in a dose pack containing 12.5 mg, 25 mg, 37.5 mg, and 50 mg doses) [15]. J-codes are used to report injectable drugs that ordinarily cannot be self-administered. In this study, J-2794 (injection, risperidone, long-acting) was used [8].
Natural Language Processing (NLP)-Assisted Chart Review
Natural language processing can be used to identify and extract information at a scale too large to review by hand. The accuracy of NLP varies across clinical concepts, with a higher accuracy obtained for concepts that are concrete, commonly documented, and consistently described [16]. As it was not clear how and when needle detachments would be documented, an NLP-assisted, semi-automated chart review approach was designed as an alternative to a fully automated NLP system. In this approach, potentially relevant information was selected using keywords and phrases that were then presented in a structured format to clinical annotators for review and classification.
Regular Expressions in NLP
Regular expressions (RegEx) are sequences of characters that form a search pattern in text. RegEx is a well-established and widely used method of searching for keywords in text. It was used to represent relevant concepts in a basic root form in order to capture potential conjugations, misspellings, or variants of the word. Two lists of terms were used. If one word from the first list appeared within 25 words of any word from the second list, the statement was flagged for manual review.
Initial Regular Expression
In the pilot study, we employed a single RegEx statement to find all statements in text that could be documenting evidence of detachment or other issues related to an injection (Fig. 3).
Fig. 3
Regular expression terms used in the pilot study
×
For example, one result might include the word “needle” from the first list (found inside the first set of parentheses) and find the word “broke” from the second list (second set of parentheses) within 25 words of “needle.”
Possible detachments were identified through a multi-step process. First, we identified patterns of keywords and phrases likely to describe one of two concepts: (1) the administration of Risperdal® CONSTA® and (2) needle detachments. Each pattern list contained the root forms of words to capture any conjugations, misspellings, or variants in the text. Next, we processed all documents for patients in the cohort that were created on the date of Risperdal® CONSTA® administration, or which contained the words “risperidone,” “risperdal,” “consta,” and “injection” or any known spelling variations of these words. Statements containing at least one word from each concept list were manually reviewed by two clinical annotators. Frequently occurring duplicate statements were reviewed in groups to simplify manual review. If the reviewed statement appeared to be evidence of a PQC, the note was manually reviewed to confirm a PQC event. If any new terms and phrases were found in relevant statements, they were added to the term lists and the process was repeated to discover new statements. Once an instance of a PQC was identified, a detailed chart abstraction of the note with the confirmed PQC was performed to extract additional information about the event and the patient.
Method Refinement
The NLP-assisted chart review was deployed on all notes 15 days prior to and 2 days after each Risperdal® CONSTA® administration in 2013 in the six pilot VA hospitals. Upon receipt of anecdotal information that PQCs could be collected and reported as infrequently as quarterly, we expanded the search window to 90 days prior to the reported PQC date. The word list for each concept was expanded based on the documented events found.
During the NLP-assisted chart review conducted for this study, new relevant words were discovered. The initial RegEx statement was found to be too slow to process a large number of documents efficiently. The word lists were split into two sets of regular expressions and a custom application was developed to identify mentions of these words in clinical notes and to determine if any of them were within 25 words of each other. Table 1 shows the expressions in each of the two sets.
Table 1
Revised keyword lists for the baseline and follow-up studies
Set 1
Set 2
syringe
needle
shot
inject
\binj\b
complication
glut
delt
intramuscular
\bIM\b
butt
risp
consta
vial
plunger
amp
hypodermic
hub
\bbent
wrong
fail
partial
break\b
broke
separat
untoward
disconnect
detach
dislodge
full
missed
damage
crack
Defect
adverse
fault
\bevents?\b
\bissue
problem
complication
disjunction
fracture
breach
spill
leak
drip
clog
push
In practice, the clinical notes were searched for any occurrences of words in the two sets of keywords. If a word from set 1 appeared within 25 words of a word from set 2, the statement was flagged for manual review.
In the regular expression in Table 1, the notation “\b” indicates the beginning or ending of a word, and “?” indicates that the character immediately before it appears 0 or 1 times. So “\bevents?\b” matches “event” and “events,” but does not match “prevent” or “preventing.” The RegEx application was set to case insensitive to capture words that were capitalized and lowercase.
Record Abstraction and Adjudication
A team of clinical annotators performed chart abstraction of all notes identified through the automated approach as containing potential evidence of needle detachment. They also abstracted available relevant information about the detachment, when documented, including patient age, gender, diagnosis, site of administration, lot number, use of a second kit, detachment timing, detachment location on device, dose completeness, and role of the individual who administered injection. There is no requirement to record details about the injection (e.g., location of administration) or the kit (e.g., whether a second kit was used). These details were reported for cases in which they were available. All identified detachments were reported as PQCs in accordance with Janssen standard operating procedures.
A defined number of notes (n = 1000) that were identified as not containing any evidence of a potential needle detachment by the patterns and keywords were also manually reviewed to estimate the sensitivity and specificity of the automated approach. No additional detachments or potentially relevant terms were found in this review.
Baseline and Follow-Up Methods
The same methods were used for the baseline and follow-up studies. Both were conducted in the same data source. As was the case in the pilot study, Risperdal® CONSTA® injections were identified using administrative codes and health care provider notes. NLP was used to identify notes with potential evidence of detachments. Notes were reviewed by two trained clinical annotators independently to determine if potential detachments had been documented. A random sample of 1000 notes not flagged by the automated approach was reviewed to assess the specificity of the method. Data about the injection and patient characteristics were collected for needle detachment events. Figure 4 shows the process for identifying needle detachments in the VA for the baseline and follow-up studies.
Fig. 4
Baseline and follow-up study procedure
×
The baseline and follow-up studies were based on a census of Risperdal® CONSTA® administrations in the Department of Veterans Affairs during the observation windows. That is, we did not draw a sample; we examined every administration of Risperdal® CONSTA® during the specified periods. The observed difference in needle detachment incidence rate between the baseline and the follow-up was the actual difference. No inferential statistics were required because no sampling was done.
Compliance with Ethics Guidelines
All procedures performed in studies involving human participants were in accordance with the University of Utah IRB, the IRB of record for the VA Salt Lake City Healthcare System, and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The University Utah IRB waived the need for informed consent for this study. The study was also reviewed and approved by the VA Salt Lake City Healthcare System Research and Development Committee.
Results
Pilot Study
Spontaneous Reports
There were 100 reports of needle detachments in the Janssen PQC database in the United States during calendar year 2013. Many of the identifiers for the reports were incomplete. Where possible, facilities were identified from addresses using Google searches with all available information. The result was 74 identifiable institutions with a known number of Risperdal® CONSTA® kits for 2013; five of those had more than one PQC report. From these results, we identified 6 sites within the VA system that had reported a needle detachment event (one at each site). Using a NLP-assisted chart review of VA electronic health records, we were able to identify three of the six spontaneously reported detachments and two additional detachments that were not in the spontaneous reports.
Patient Characteristics
As it is a health system for veterans, the VA has distinct patient demographics. In 2013, the overall patient population was largely male (93%). Nearly half (46%) of all VA patients were age 65 or older, with a median age of 62. VA patients were 73% white, 16% black, and 5% Hispanic [17].
The patient population in the baseline study (calendar year 2013) was largely male (91%), consistent with VA demographics. Nearly two-thirds (65%) of the patients receiving Risperdal® CONSTA® were between the ages of 45 and 64, with a median age of 56, slightly younger than the overall patient population. Only 55% were white, compared with 73% of the overall patient population. The majority of the patients (59%) had a diagnosis of both schizophrenia and bipolar disorder. Patients in the follow-up study were slightly older at 57 years, closer to the median in the VA population. The proportions of the patients who were male (91%) and white (56%) were similar to those in the baseline study. The follow-up study had slightly smaller proportions of patients with schizophrenia only (29% vs. 32%) or with schizophrenia and bipolar disorder (55% vs 59%), but slightly higher proportions of patients diagnosed with bipolar disorder only (9.6% vs. 6.5%) or with neither schizophrenia nor bipolar disorder (6.2% vs. 2.5%) (Table 2).
Table 2
Patient characteristics for US Department of Veterans Affairs
Baseline: calendar 2013
Baseline
Follow-up
Follow-up: Dec 2015–Dec 2016
N
%
N
%
3608
100%
3008
100%
Age
< 18 years
0
0.0%
0
0.0%
18–34 years
405
11.2%
308
10.2%
35–44 years
391
10.8%
344
11.4%
45–54 years
890
24.7%
545
18.1%
55–64 years
1431
39.7%
1128
37.5%
65–74 years
425
11.8%
612
20.3%
≥ 75 years
66
1.8%
71
2.4%
< 65 years
3117
86.4%
2325
77.3%
≥ 65 years
491
13.6%
683
22.7%
Median
55.7
57.2
Mean (sd)
53.4 (12.0)
55.0 (12.9)
Gender
Male
3286
91.1%
2740
91.1%
Female
322
8.9%
268
8.9%
Diagnoses*
Schizophrenia only
1146
31.8%
876
29.1%
Bipolar disorder only
234
6.5%
290
9.6%
Schizophrenia and bipolar disorder
2136
59.2%
1655
55.0%
Neither diagnosis
92
2.5%
187
6.2%
Race
White
1973
54.7%
1677
55.8%
Black
1361
37.7%
1097
36.5%
Asian
34
0.9%
32
1.1%
Native Hawaiian or Pacific Islander
40
1.1%
38
1.3%
American Indian or Alaskan native
33
0.9%
25
0.8%
Unknown
167
4.6%
139
4.6%
Ethnicity
Hispanic
291
8.1%
218
7.2%
Unknown
3317
91.9%
2790
92.8%
Census region**
Northeast
595
16.5%
468
15.6%
Midwest
757
21.0%
669
22.2%
South
1582
43.8%
1365
45.4%
West
573
15.9%
438
14.6%
US Territory
101
2.8%
68
2.3%
*Schizophrenia defined as ICD9 diagnosis codes 295.xx and v11.0. Bipolar defined as ICD9 diagnosis codes 296.xx except for 296.90 and 296.99. All codes must have occurred on or before the date of the first administration of Risperdal® CONSTA® in the study period.
**Census region defined based on the zip code of where care was provided during the first administration of Risperdal® CONSTA® in the study period
Risperdal® CONSTA® Administrations
There were 47,934 administrations of the drug to 3608 patients who had at least one clinical note in the baseline; and 41,819 administrations to 3008 patients in the follow-up (December 2015–December 2016). The number of clinical notes examined in the baseline study was 204,116; this number was 159,361 in the follow-up study (Table 3).
Table 3
Risperdal® CONSTA® administrations and health care provider notes
Baseline
Follow-up
N
N patients
N
N patients
Injections
Administrations of Risperdal® CONSTA® with clinical note
47,934
3608
41,819
3008
Clinical notes
Total # of any clinical notes
652,102
3608
490,175
3008
Total # of relevant clinical notes
204,116
3608
159,361
3008
Needle Detachments
Twenty-two (22) needle detachments were identified in the VA system in 2013; 0.046% of the 47,934 Risperdal® CONSTA® administrations. A majority of these needle detachment events were deltoid injections (n = 14), which resulted in a partial (n = 12) or missed (n = 3) dose. In at least half of the cases, a second kit was used. In the ten cases in which a lot number was recorded (12 were missing a lot number), seven unique lot numbers were found (Table 4).
Table 4
Risperdal® CONSTA® needle detachments
Baseline
Follow-up
N
%
N
%
% of injections
22
0.046%
14
0.033%
% of patients
22
0.59%
14
0.45%
Location of administration
Deltoid
14
63.6%
7
50.0%
Gluteal
2
9.1%
1
7.1%
Unknown
6
27.3%
6
42.9%
Lot number
Known
10
7
Unknown
12
54.6%
7
50.0%
2nd kit used?
Yes
11
50.0%
4
28.6%
No/unknown
11
50.0%
5
35.7%
5
35.7%
Detachment timing
Before injection
5
22.7%
4
28.6%
During injection
11
50.0%
5
35.7%
After injection
0
0.0%
0
0.0%
Unknown
6
27.3%
5
35.7%
Detachment location
Needle detached*
14
63.6%
4
28.6%
Vial adapter detached
6
27.3%
5
35.7%
Both detached
0
0.0%
0
0.0%
Unknown
2
9.1%
5
35.7%
Dose completeness
Full
2
9.1%
2
14.3%
Partial
12
54.6%
4
28.6%
Missed
3
13.6%
6
42.8%
Unknown
5
22.7%
2
14.3%
HCP administering dose
Medical doctor
1
4.6%
0
0.0%
Physician assistant
0
0.0%
0
0.0%
Nurse practitioner
0
0.0%
0
0.0%
Nurse
1
4.6%
2
14.3%
Unknown
20
90.9%
12
85.7%
In the follow-up study, 14 needle detachments were identified in 41,819 administrations of Risperdal® CONSTA® from December 1, 2015 through December 31, 2016; 0.033% of the administrations of the drug. This is a reduction of 27% from the baseline rate of 0.046% (Table 5). Half of the detachments occurred during deltoid administrations (n = 7). Ten [10] resulted in a partial (n = 4) or missed (n = 6) dose. Among the seven cases in which a lot number was recorded, all were distinct.
Table 5
Results summary
Detachments
Injections
Proportion
PPM
PPM reduction
% Reduction
Baseline (2013)
22
47,934
0.000459
459
Follow-up (Dec 2015–Dec 2016)
14
41,819
0.000335
335
124
27.0%
None of the 1000 randomly selected notes not initially flagged by the automated approach contained evidence of a detachment.
Limitations
This method was likely subject to incomplete outcome ascertainment due to needle detachments that were not recorded in the medical record. This may have resulted in an incidence rate that is lower than the true rate.
Conclusion
This study was based on solicited rather than spontaneous reports. This is a strength of the present study design that cannot be obtained with spontaneous reports because solicited reports are systematically acquired using methods in which the sampling frame is known. In this case, we searched all occurrences of Risperdal® CONSTA® injections documented in electronic health records for the occurrence of likely needle detachments, rather than relying on spontaneous reports made to the company or the FDA.
While this may be a better representation of the rate than spontaneous reports, it is still likely a conservative estimate resulting in incomplete outcome ascertainment. This may have resulted in incidence rates that are lower than the true rate. However, it would not bias the comparison of the baseline and follow-up because the same methods were used for both.
In the pilot study, we identified a subset of the spontaneously reported detachments and identified several previously unreported events. The most plausible explanation for finding only a subset of the spontaneous reports is that some events had not been included in the clinical notes accompanying the injection. Anecdotal reports from interviews with healthcare providers suggested that only a portion of detachments are recorded in the text field. While it is likely that both the baseline and follow-up studies captured the majority of detachments recoded in the clinical notes, there may have been needle detachments that were undocumented and perhaps unreported; missed by both solicited and spontaneous reporting systems. At present, there is no way to quantify the number of detachments that were missed by this method or that do not appear in spontaneous reports. However, there is no reason to assume that there was bias in the incomplete outcome assessment from the baseline to follow-up studies. So, while the rate is likely underreported, an assessment of the change from baseline to follow-up remains unbiased.
Relevance
The results showed that the redesigned Risperdal® CONSTA® kit reduced the incidence of needle detachment events in the Department of Veterans Affairs. Replication of this study in other data sources will be important for generalizing the findings. These results were obtained using NLP-assisted electronic medical record review, a method that may prove useful in other settings.
Further work may involve identifying needle-stick injuries to health professionals. This was not included in the present study because information about needle-stick injuries to health professionals is generally not recorded in the patient records and requires complex institutional approvals.
Acknowledgements
This work was made possible by individuals across several groups at Janssen, including Pharmacovigilance, the Compound Development Team, the Anolinx Team, and the VINCI team at the Department of Veterans Affairs (VA). This work was supported using resources and facilities of the VA Informatics and Computing Infrastructure (VINCI), VA HSR RES 13-457. The content of this manuscript does not represent the views of the Department of Veterans Affairs or the United States Government.
Funding
This study and article processing charges were funded by Janssen Pharmaceuticals, Inc. Authors from the VA had full access to the raw data. Other authors had access to data summaries. All authors take responsibility for the integrity of the data and accuracy of the data analyses.
Authorship
All named authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for authorship for this article, take responsibility for the integrity of the work as a whole, and have given their approval for this version to be published.
Disclosures
Marsha A. Wilcox is an employee of Janssen. Danielle Coppola is an employee of Janssen. Ira D. Solomon is an employee of Janssen. Scott L. DuVall has received research grants from AbbVie Inc., Amgen Inc., Anolinx LLC, Astellas Pharma Inc., AstraZeneca Pharmaceuticals LP, Boehringer Ingelheim International GmbH, F. Hoffman-La Roche Ltd, Genentech Inc., Genomic Health, Inc., Gilead Sciences Inc., GlaxoSmithKline PLC, Innocrin Pharmaceuticals Inc., Janssen Pharmaceuticals, Inc., Kantar Health, Merck & Co., Inc., Mylan Specialty LP, Myriad Genetics, Inc., Novartis International AG, PAREXEL International Corporation, Shire PLC within the last three years outside the submitted work. Nicole Bailey was an employee of an Anolinx LLC that received a research grant from Janssen to conduct/participate in this study. They are employed by PAREXEL, which receives research grants from many pharmaceutical and biotechnology companies to conduct/participate in many clinical research projects/studies. Andrew Wilson was an employee of an Anolinx LLC that received a research grant from Janssen to conduct/participate in this study. They are employed by PAREXEL, which receives research grants from many pharmaceutical and biotechnology companies to conduct/participate in many clinical research projects/studies. Aaron W. C. Kamauu was the owner of an Anolinx LLC that received a research grant from Janssen to conduct/participate in this study. They are employed by PAREXEL, which receives research grants from many pharmaceutical and biotechnology companies to conduct/participate in many clinical research projects/studies. Patrick R. Alba, in the last 36 months, had grants from the National Heart, Lung, and Blood Institute, Department of Defense, Amgen Inc., Anolinx LLC, Genentech Inc., Gilead Sciences Inc., Merck & Co., Inc., Novartis International AG, and PAREXEL International Corporation outside the submitted work. Olga V. Patterson, in the last 36 months, had grants from National Heart, Lung, and Blood Institute, Department of Defense, Amgen Inc., Anolinx LLC, Genentech Inc., Gilead Sciences Inc., Merck & Co., Inc., Novartis International AG, and PAREXEL International Corporation outside the submitted work. Daniel W. Denhalter and Benjamin Viernes have nothing to disclose.
Compliance with Ethics Guidelines
All procedures performed in studies involving human participants were in accordance with the University of Utah IRB, the IRB of record for the VA Salt Lake City Healthcare System, and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. The University Utah IRB waived the need for informed consent for this study. The study was also reviewed and approved by the VA Salt Lake City Healthcare System Research and Development Committee.
Data Availability
The datasets for these analyses are not publicly available due to policies and the regulation of data governed by the Department of Veterans Affairs. Future analysis of the data can be coordinated with the VA authors on reasonable request.
Open Access
This article is distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Risperdal® CONSTA® Needle Detachment. Incidence Rates Before and After Kit Redesign: A Retrospective Study using Electronic Health Records and Natural Language Processing in the Department of Veterans Affairs
verfasst von
Marsha A. Wilcox Danielle Coppola Nicole Bailey Andrew Wilson Aaron W. C. Kamauu Patrick R. Alba Olga V. Patterson Benjamin Viernes Daniel W. Denhalter Ira D. Solomon Scott L. DuVall
