Chemotherapy safety and severe adverse events in cancer patients: Strategies to efficiently avoid chemotherapy errors in in- and outpatient treatment

Authors


Abstract

To enhance the quality and safety in cancer treatment, and in acknowledgement that medical errors occur, we have established 2 error management systems: one monitors chemotherapy errors, the other records all severe adverse events occurring in chemotherapy-treated cancer patients (SAECTx) in in- and outpatient treatment. These error systems have been implemented by our departmental “Clinical Service Center,” a multidisciplinary team which controls all chemotherapy protocols and orders prior to the medication reaching the patient. We performed a prospective cohort study in consecutive cancer patients who received chemotherapies in our department between January 2005 and December 2006. Over this 2-year period, 2,337 patients were treated, with an equal distribution as in- and outpatients: 22,216 consecutive chemotherapy orders were analyzed, of which 83.5% were completely flawless, whereas we detected and corrected medical and administrative errors in 17.1%: in 3.8%, these errors involved the chemotherapy itself, in 4.5% the patient data and in 8.7% missing written informed consent forms. Chemotherapy errors were less frequent in outpatients than inpatients (3.3 vs. 4.5%, respectively). In outpatients, the rate of chemotherapy errors decreased from 4% in 2005 to 2.8% in 2006, but remained stable for inpatients (4.4% 2005 vs. 4.7% 2006). Among a total of 3,792 detected errors, only 3 reached the patient, resulting in an error rate in patients of 0.079%. Therefore, since we detected a substantial number of chemotherapy-related errors and intercepted 99.9%, we recommend our efficient surveillance system as an important safety check, thereby ensuring that chemotherapies are delivered error-free to cancer patients. © 2008 Wiley-Liss, Inc.

Hematologists and oncologists are neither more nor less apt to make medical mistakes than other physicians, but the high toxicity and low therapeutic index of the chemotherapeutic drugs they prescribe make chemotherapy errors potentially harmful. The aim and pressing concern for any physician administering chemotherapy is therefore to administer these agents with the best possible safety and efficiency. This is the more mandatory, since chemotherapy protocols have become very complex, involve numerous comedications and are given to elderly cancer patients with several comorbidities. Albeit all precautions, prescribing and applying chemotherapies are associated with the considerable risk of severe adverse events occurring in chemotherapy-treated cancer patients (SAECTx), and there is the potential of grave medical errors. 1–9 Moreover, drug doses must be calculated according to body size, laboratory results must be taken into account and each medication must be administered at precisely-kept time intervals.

To enhance the safety of chemotherapy administration, it is necessary for hematologists and oncologists, internal medicine practitioners, the pharmacy and all others involved in the chemotherapy treatment of cancer patients, to standardize chemotherapy protocols as best as possible, to have an efficient safety system for all chemotherapy-related issues and a dedicated quality assurance team to detect and effectively prevent medical errors. 1–9 Despite these efforts, errors in chemotherapy orders have been reported to occur with a frequency of at least 4%.3–5, 9 Such errors may involve the chemotherapy itself, comedication, patient data and/or missing informed consent forms to name only some relevant and frequent errors and omissions. Any detected error, especially in the chemotherapy itself or comedication, and how this error was successfully detected and intercepted, should be implemented into guidelines, that aim to prevent or at least efficiently reduce SAECTx.1–5

Such a chemotherapy control system requires close cooperation among medical doctors, nurses and clinical pharmacists, which has been implemented in our department since 1994 with formation of a dedicated chemotherapy team, called “Clinical Service Center” (CSC). Since its formation, this CSC has been proven to be essential for (a) the daily avoidance of chemotherapy errors, (b) the entire patient management and (c) the improvement of cancer care. 1–5 In this analysis, we have prospectively investigated consecutive errors in the ordering and administration of chemotherapies and SAECTx occurring on in- and outpatients wards in our institution between January 2005 and December 2006.

Material and methods

Patient description

We analyzed consecutive chemotherapy errors and SAECTx that occurred in cancer patients in our department. Patient characteristics are given in Table I.

Table I. Entire Adult Cancer Patient Cohort and Their Characteristics 2005 Versus 2006, Including Number of Chemotherapy (CTX) Orders
 Total20052006
  • 1

    The percentages are >100%, since some patients were treated both as in- and outpatients.

  • 2

    Not evaluable if CTx was given as inpatient or outpatient.

  • CTx, chemotherapy.

Total no. treated patients2,3371,1601,177
No. of CTx- orders22,21610,88511,331
 Study protocols3,626 (16.3%)2,116 (19.4%)1,510 (13.3%)
 Standard protocols18,590 (83.7%)8,769 (80.6%)9,821 (86.7%)
Inpatients11,480 (63.3%)779 (67.2%)701 (59.6%)
Outpatients11,486 (63.6%)702 (60.5%)784 (66.6%)
No. inpatient CTx orders10,416 (46.9%)5,250 (48.2%)5,166 (45.6%)
No. outpatient CTx orders11,384 (51.2%)5,331 (49.0%)6,053 (53.4%)
Not evaluable CTx orders2416 (1.9%)304 (2.8%)112 (1.0%)
Median patient age (year, range)62 (18–92)  
Female vs. male age (year, range)62 (18–87)/62 (18–92)  
Female vs. male patients (%)36.9/63.1%37.6/62.4%36.7/63.3%

Setting

The study was performed by the CSC, a multidisciplinary team of medical doctors, documentation and medical assistants, and study nurses. The function of this team (in addition to the coordination and implementation of all clinical trials in our department and the documentation of all cancer patients in a tumor-based documentation system) is (1) to control all chemotherapy orders before they are given to patients, (2) to assure that our chemotherapy manual is constantly updated and (3) to give expert advice on all chemotherapy-related questions. 1–5 Our standardized chemotherapy manual, existing since 1994, contains over 300 chemotherapy protocols within the so called “Blue Book,” which is updated daily and used extensively within our entire university through intranet-access.1 Each Blue Book protocol contains information on (a) dose and timing of chemotherapy agents, (b) initiation, inclusion and exclusion criteria [blood counts, creatinine, organ function, etc.], (c) rules for dose reduction, (d) day of treatment continuation, (e) the attending physician responsible for the specific protocol, (f) essential comedication to prevent nausea, organ damage and (g) anti-infective prophylaxis and other notable controls. Our CSC-Blue Book team consists of 1 pharmaceutical assistant, 1 resident and 1 attending physician, who modify our chemotherapy protocols according to international guidelines (e.g. ASCO, EORTC, NCCN) and recent full article publications in peer-reviewed journals.

Data collection

Two error-management-systems [(1) PERMIT = Prescription Error Monitoring = the monitoring of any error in chemotherapy ordering, scored by the CSC-team and (2) MERKE = Monitoring of clinically severe adverse events reported to the CSC through ward physicians] were developed at our institution to optimize chemotherapy safety and cancer care. Data were collected over a 24-month period.

PERMIT

This project was implemented by the CSC which receives 50–100 manual (January to December/2005) and computerized (since January 2006) chemotherapy orders per day and is responsible for checking any irregularities occurring in: (a) ordered substances (e.g. dose, drug), (b) patient data (e.g. height, weight, cycle), (c) missing written informed consent forms for the chemotherapy protocol (electronically scanned and stored) and (d) any other errors before the chemotherapy is issued to the patient. We considered the error distinction a–d important, since previous reports have specifically and solely analyzed “medical errors,” rather than defining and deciphering others. Errors in chemotherapy substances and patients data (a+b) were defined as “medical” and missing written informed consent forms (c) as “administrative errors.” Although medical errors are considered more serious, for formal, legal and administrative reasons, we aimed that also informed consent forms should be as complete as possible.

The CSC task is to check all chemotherapy orders by means of comparing the order with (1) the chemotherapy template within the Blue Book, (2) prior chemotherapy orders of the specific patient and (3) with regard of patient specific comorbidities which ask for dose adjustments (Fig. 1). If the orders are flawless or—if not—have been corrected by the CSC, they are transformed into patient-specific protocol plans according to the Blue Book. 1 These patient-specific charts, comprising all chemotherapy agents and comedication are sent directly to the ward via print order. The clinical pharmacy receives these orders simultaneous to the CSC-team and is equally responsible for detecting any error. The pharmacists check the orders against the Blue Book protocol and pharmacy tool (Zenzy). If any error is detected, a coordinated exchange with the CSC is ensured immediately before the chemotherapy is prepared and sent to the ward. If no errors are detected by the pharmacy, the chemotherapy is prepared, the drugs are clearly labeled and forwarded to the patient. When the chemotherapy agents and the patient-specific chemotherapy plan have been received on the ward, the chemotherapy is administered by qualified personnel who countersign the chemotherapy and comedication on the patient-specific chemotherapy plan. The physicians and nurses use this patient-specific chemotherapy schedule likewise as a documentation chart.1, 2, 5

Figure 1.

Schema illustrating the chemotherapy ordering and control process for cancer patients with Clinical Service Center (CSC)- and pharmacy-support. equation image The physician informs the patient on the specific chemotherapy substances, of potential side effects of a scheduled chemotherapy protocol and obtains the patient's oral and written informed consent (IC). equation image The physician sends the electronically scanned written IC and chemotherapy order to the CSC and pharmacy simultaneously that both check the chemotherapy order. If any error is detected by the CSC or pharmacy, a coordinated exchange and error eradication between both control instances with the ordering clinician is ensured before thechemotherapy is prepared and sent to the ward. equation image If the chemotherapy order is completely flawless, the CSC transforms this into a patient-specific protocol and sents this to the ward via print order, and the pharmacy delivers all chemotherapy agents directly to the ward equation image, equation image and equation image. When the chemotherapy plan and agents are received on the ward, the chemotherapy is administered by qualified personnel who countersigns the chemotherapy and comedication on the patient-specific chemotherapy plan.

MERKE

This project aims to record any SAECTx occurring before, during or after the chemotherapy administration. These SAECTx are classified according to (1) unexpected deaths, (2) chemotherapy-induced extravasations, (3) unexpected referrals to the intensive care unit (ICU), (4) unscheduled operations and (5) other SAECTx (miscellaneous), such as any serious and undesirable event related to chemotherapy applications (unexpected side effects). We had extensively discussed these 5 entities within our department which considered 1–5 important to register within the electronic SAECTx-databank. Our error monitoring system was explicitly not used to register known chemotherapy complications, for example infections after chemotherapy treatment, but to avoid preventable complications within our clinical practice, where also comorbid and frail patients are dismissed home as early as possible and SAECTx outside clinical trials are less well recorded.

With occurrence of any SAECTx, it is directly reported by means of a computerized reporting system to the CSC. The SAECTx reports are reviewed monthly by the MERKE team, which consists of physicians, nurses, quality control managers, pharmacists and the CSC team itself to determine possible causes and to identify strategies for their prevention. These strategies are implemented in clinical guidelines, SOPs and safeguards (e.g. Blue Book). 1–5 MERKE reports are also discussed during the attending physicians' conferences and CSC meetings once a month.

Technology of chemotherapy ordering and checking

The department's electronic chemotherapy ordering and prescription (eCOP) system was designed and written by our electronic specialist team. This eCOP system was specifically structured according to our chemotherapy ordering- and prescription-requirements. This system is a web-based database application, written in the hypertext preprocessor (PHP) program and running on the apache web server and SAPDB (database management system from SAP AG) database server. For chemotherapy orders, the ward physician uses this web formatted ordering system through a web browser. This form is stored as an HTML (hypertext markup language) document on the server and is sent automatically via E-mail or fax to the CSC and pharmacy. Upon receipt of this order, the CSC and pharmacy check simultaneously, if this specific order contains the correct information on all chemotherapy agents. If any error is detected, this is immediately reported to the ward physician and instantly corrected in the database application. If the chemotherapy is entirely flawless, the CSC creates a detailed treatment schedule, using a preplanned scheme which is available for each chemotherapy protocol in the Blue Book database. These templates have been generated in our databank through a protocol editor module. Immediately after any order has been transformed into an exact chemotherapy schedule, this patient-individual plan is sent through the network printer directly onto each ward. The ward physicians acknowledge this chemotherapy plan for each patient with their signature on the printouts. Only then, the prescription is regarded as valid and ready for use (Fig. 1). Using our chemotherapy databank, basic patient demographics can also be generated through the hospital's central clinical information system via the health level 7 protocol.

Statistical analysis

Descriptive data analysis was performed using median and ranges except where stated otherwise. The descriptive analysis was executed using our tumor-based documentation system and CSC, pharmacy and administration databank sources. The study and analysis were carried out according to the guidelines of the Declaration of Helsinki and good clinical practice. All patients gave their written informed consent for institutional-initiated research studies and specifically for analyses of clinical outcomes research conforming to our institutional review board guidelines. Data were analyzed as of December 31, 2006.

Results

With 2,337 cancer patients being treated in our department 2005/2006 allowed us to analyze 22,216 consecutive chemotherapy orders (Tables I and II). The number of orders increased by 4.1% between 2005 and 2006, whereas number of patients remained almost stable, with a decrease in inpatient- and increase in outpatient-chemotherapy orders. Similar numbers of patients were treated as in- and outpatients, which made up to more than 100% (Tables I and II), since often patients were initially treated as in- and later as outpatients or vice versa within the same year. Since the number of chemotherapy orders given to in- and outpatients was almost the same, we also compared the error frequency within these two settings. As depicted in Table I, 16.3% of patients were treated within study protocols. Of note, the number of chemotherapy orders within clinical study protocols decreased between 2005 and 2006, although clinical studies remained constant during that time (approximately n = 60 within our department/year). The decline of chemotherapy orders within study protocols was due to the more complex activities related to clinical trials, narrow inclusion and wide exclusion criteria and additional effort needed for patient accrual and study adherence.

Table II. Chemotherapy (CTX) Errors in Cancer Patients: Entire Cohort and 2005 Versus 2006
inline image

The PERMIT project

Of 22,216 chemotherapy orders, 83.5% were entirely flawless. The CSC and pharmacy detected and corrected errors in 17.1% and—most importantly—could thereby avoid minor to highly relevant errors. These errors involved the chemotherapy itself in 3.8%, patient data in 4.5%, and missing written informed consent forms in 8.7%. Errors in the chemotherapy order itself decreased from 2005 to 2006 from 4.2 to 3.6%, respectively (Table II). In order to ascertain, whether this error reduction translated into 2007, we analyzed all chemotherapy orders (10,674) of 2007 also (Supp. Info. Table IB): Flawless chemotherapy orders were found in 8,843 (82.8%), whereas errors in the chemotherapy itself, patient data and missing written informed consent were found in 201 (1.9%), 519 (4.9%) and 971 (9.1%), respectively. The added 2007 data showed that the rate of error-free chemotherapies remained at 83%, that errors in patient data and missing written informed consent forms also remained stable when compared with 2005 and 2006, but that indeed errors in chemotherapy orders substantially decreased further (from 4.2 to 3.6 to 1.9% in 2005, 2006 and 2007, respectively). Also of note was that errors in the chemotherapy itself were less frequent in outpatients than inpatients (3.3 vs. 4.5%, respectively; Table II). For outpatient orders, the chemotherapy errors even decreased from 4 to 2.8% in 2005 versus 2006, respectively, but remained stable for inpatients (4.4 vs. 4.7%; Table II). Errors in patient data decreased for inpatient orders (4.6% in 2005 vs. 3.4% in 2006), whereas these slightly increased for outpatients (4.5% in 2005 vs. 5.5% in 2006, Table II). Administrative errors (missing written informed consents) were higher in in- than outpatients with 14.1 versus 4.1%, respectively (Table II).

The MERKE-project

The summarized findings are depicted in Tables III–V.

Table III. Classification of Severe Adverse Events (SAECTx) Occurring in CTx-Treated Cancer Patients in Relation to Treated Cancer Patients and Given Chemotherapeutic Substances: Entire Cohort and 2005 Versus 2006
SAECTxTotal20052006
Total SAECTx number1748787
 Unexpected death71 (40.8%)38 (43.6%)33 (37.9%)
 Intensive care unit transfer66 (37.9%)39 (44.8%)27 (31.0%)
 Unexpected operations5 (3.8%)2 (2.3%)3 (3.4%)
 CTx extravasation12 (6.9%)5 (5.7%)7 (8.0%)
 Other20 (11.5%)3 (3.4%)17 (19.5%)
Total no. treated patients2,3371,1601,177
Total no of CTx orders22,21610,88511,331
SAECTx incidence/patient/year7.4%7.5%7.4%
SAECTx incidence/order/year0.78%0.80%0.76%
Table IV. Classification of Unexpected Deaths: Entire Cohort and 2005 Versus 2006
Unexpected deathsTotal20052006
  • 1

    Plural entry possible.

  • 2

    Requiring intermittent dialysis.

  • 3

    Liver function parameters >5 fold of normal.

  • 4

    Other causes of death: ileus, GVHD, suicide.

Total no. unexpected deaths713833
Cause of death
 Cardio-pulmonary failure140 (56.3%)21 (55.3%)19 (57.6%)
 Sepsis116 (22.5%)5 (13.2%)11 (33.3%)
 Bleeding17 (9.9%)5 (13.2%)2 (6.1%)
 Embolism11 (1.4%)1 (2.6%)0 (0%)
 Renal impairment1,22 (2.8%)2 (5.3%)0 (0%)
 Liver impairment1,31 (1.4%)1 (2.6%)0 (0%)
 Other1,44 (5.6%)3 (7.9%)1 (3.0%)
Underlying cancer disease
 Hematological malignancies43 (60.6%)26 (68.4%)17 (51.5%)
 Lung cancer12 (16.9%)9 (23.7%)3 (9.1%)
 Gastrointestinal Cancer1 (1.4%)0 (0%)1 (3.0%)
 Malignant melanoma3 (4.2%)2 (5.3%)1 (3.0%)
 Urogenital cancer4 (5.6%)1 (2.6%)3 (9.1%)
 Other malignancies8 (11.3%)0 (0%)8 (24.2%)
Patients' remission status
 Initial diagnosis14 (19.7%)9 (23.7%)5 (15.2%)
 Progressive disease/ Relapse (PD)42 (59.2%)21 (55.3%)21 (63.6%)
 Complete remission (CR)3 (4.2%)1 (2.6%)2 (6.1%)
 Stable disease (SD)2 (2.8%)2 (5.3%)0 (0%)
 Partial remission (PR)1 (1.4%)1 (2.6%)0 (0%)
 Not noted9 (12.7%)4 (10.5%)5 (15.2%)
Table V. CTx Extravasations: Entire Cohort and 2005 Versus 2006
CTx extravasationsTotal20052006
No. treated patients2,3371,1601,177
No. CTx orders22,21610,88511,331
No. ordered i.v. chemotherapeutic substances41,74319,69722,046
No. CTx extravasations1257
Extravasation incidence/patient/year0.51%0.43%0.59%
Extravasation risk/CTx order0.054%0.046%0.062%
Patient risk for extravasation/given i.v. substance0.029%0.025%0.032%

Deaths

There were 71 unexpected deaths over the 24-month period (3% of treated cancer patients). Their median age was 64.1 years. Most of these patients had progressive disease (PD) and most died of cardio-pulmonary complications or sepsis. The majority of them had hematological malignancies (Tables III and IV).

Extravasations

In 2005, 5 chemotherapy extravasations were documented (Table III): 3 with fluorouracil, 1 with vinorelbine and 1 with paclitaxel. Two of these substances were necrotizising (vinorelbine and paclitaxel). With 1,160 treated patients and 10,885 chemotherapy orders administered in 2005, the extravasation incidence was calculated as 0.43%/patient/year or 0.046%/chemotherapy order, thereby confirming previously reported extravasation incidence rates of 0.1–6.5%. 10 As chemotherapy protocols included more than 1 chemotherapeutical substance on some days, we also calculated the risk associated with each substance. With a total of 19,697 chemotherapeutical substances prepared for intravenous use, the patient risk for an extravasation was 0.025% (Table V). In 2006, 7 extravasations were recorded (gemcitabine, doxorubicin, vinorelbine, etoposide, cisplatin). With 11,331 chemotherapy days and 1,177 patients, the extravasation rates were 0.59 and 0.062%, respectively, and the patient risk 0.032% (Table V).

ICU referrals

Thirty-nine unexpected referrals to the ICU were recorded in 2005 (Table III). The most common ICU referral causes were sepsis (n = 22) or cardio-pulmonary insufficiency (n = 11). Of note, 19/39 (48.7%) patients could be stabilized successfully and transferred back to their hematology-oncology ward. Twenty patients died on the ICU, mainly due to PD, sepsis or cardio-pulmonary complications. In 2006, there were 27 unexpected referrals to the ICU, most common being again cardio-pulmonary insufficiency (n = 10) or sepsis (n = 8).

Operations

Two unexpected operations were necessary in 2005: one because of neutropenic colitis, the other due to a knee joint empyema. Neutropenic complications—with multi-agent chemotherapies—are frequent, also including rare events, such as neutropenic colitis and empyema. Although the incidence of neutropenic colitis is not high, its occurrence should be considered, since the mortality rate is considerable and the increased awareness of this rapidly progressive and potentially fatal disease leads to accurate diagnosis and prompt treatment. In 2006, 3 unscheduled operations were performed: 1 because of a vertebral fracture in an ALL patient under corticosteroid treatment, 1 because of perforated sigma diverticulitis in an AML patient and 1 for intracerebral bleeding in an AML patient receiving low-dose heparin prophylaxis (Table III).

Other (miscellaneous category)

Three adverse events were reported in this category in 2005 (Table III): 1 sinus venous thrombosis in an ALL patient, 1 false-positive evidence of vancomycin-resistant enterococci and 1 posttraumatic osteomyelitis in a healthy bone marrow donor. In 2006, 17 events were reported (Table III), indicating the improved acceptance of SAECTx and the importance of this “miscellaneous category” to report remarkable SAECTx, thereby allowing to further improve patient care. Among them was an allergic reaction after etoposide and one after the first administration of cetuximab. Concerning chemotherapy doses being administered, 3 SAECTx were of note: in one, cytarabine was given within 12 instead of 22 hr (despite the correct chemotherapy schedule being sent to the ward), in another, cytarabine was spilt due to a disconnected tubing system and in the third, the wrong study medication arm (800 mg/m2 matuzumab, instead of the absolute dose of 800 mg) was almost given to the patient. Two patients experienced dislocated fractures, which—due to their clinical symptoms—led to unscheduled diagnostic evaluations, revealing a dislocated femoral fracture in a myeloma patient and a vertebra fracture in an ALL patient. Another notable SAECTx involved a window descend of a disoriented end-stage NSCLC patient with cerebral metastases, which eventually led to his death. Others errors were those 3 SAECTx as described in “preventability of potential chemotherapy errors.” The remaining SAECTx involved less remarkable, albeit unexpected organ dysfunctions, such as renal insufficiency in lymphoma or solid tumor patients. 11 The SAECTx incidence in 2005 and 2006 remained constant: this was per chemotherapy order 0.8 and 0.76%, respectively (Table III). We also checked the completeness of all SAECTx-reports: one was by analyzing all “death letters” of our deceased patients, thereby verifying, whether all unexpected deaths had been reported, the other by comparing our extravasation SAECTx with those documented by the nursing staff through a separate extravasation notification system. In both categories, the concurrence proved to be excellent (100%).

Preventability of potential chemotherapy errors

To determine, whether all errors in chemotherapy orders were detected, we checked the SAECTx-databank for chemotherapy-related events and identified 3 cases (Table VI): the first patient was treated according to the GEMOXCET protocol. 1 This protocol includes gemcitabine and cetuximab on day 1 (d1), oxaliplatin on d2 and again cetuximab on d8. The initial d1+2 were given correctly. On d8, however, instead of cetuximab, d1 and d2 treatment was repeated. The physicians, nurses, pharmacy and the CSC-team did not intercept this error; nevertheless the patient tolerated this chemotherapy dose without any undesirable reactions. The second patient received a different antibody than intended: the physician had ordered cetuximab, the CSC handed out the correct schedule, however, the pharmacy sent rituximab. The incorrect antibody treatment was detected within 5 mins after therapy initiation and the patient did not experience major side effects. In the third case, a patient was scheduled for the TAC (doxorubicin, cyclophosphamide, docetaxel) protocol.1 This CSC protocol was new and used for the first time. Within the protocol, clemastine was missing. The senior physician signed the protocol, not noticing this omittance. The patient developed an allergic reaction which was easily treated with antiallergic medication. These cases represent typical medical errors made by ward physicians, primary nurses or control instances (CSC or pharmacy). Nevertheless, most errors were intercepted effectively: of 22,216 chemotherapy orders, 83.5% (18,544)—as sent—were classified as error-free. More importantly, of 3,792 orders, 3,789 were intercepted (99.921%) and only three SAECTx occurred due to nonintercepted, incorrect chemotherapy orders (0.079%). The CSC team's error rate was thereby 0.014%. In conclusion, of 3,792 severe- to less-severe errors within our 24-month period, only 0.079% reached a patient. The risk for patients to receive an incorrect chemotherapy protocol was thereby very low (0.13%; Table VI).

Table VI. Preventability of Potential CTx Errors
CTx errorsJanuary 2005–December 2006
Total error rate (recognized and unrecognized CTx errors)3,792 (3,789 + 3)
 Avoided errors through CSC initiative3,789 (99.921%) (3,789/3,792 × 100)
 Unprevented errors despite CSC initiative3 (0.079%) (3/3,792 × 100)
CSC error rate of total CTx orders (submitted to the CSC 2005–2006)0.014% (3/22,216 × 100)
Patient risk to encounter a CTx error0.13% (3/2,337 × 100)

Cost considerations and future analyses

Our safety system constitutes of 1 medical assistant responsible for checking all chemotherapy orders and a computerized ordering system, is neither extensive nor unlikely to be adopted by others. Calculating that the medical assistant spends 1/4 of her daily routine to check 50–100 chemotherapy orders, with a salary of 46,000€/year, the cost for this additional checking accounts for 1€/order (22,216 chemotherapy orders/2 years = 92,000/4 = 23,000€ → 23,000€/22,216 = 1€). Considering the costs for the chemotherapy treatment itself and for any complication resulting from preventable errors, the additional 1€ costs—that allow the application of a 0.014% safe chemotherapy—seem well spent. If one considers that an unintentionally overdosed chemotherapy may increase the potential 1% mortality rate that multi-agent chemotherapies do impose on seriously ill cancer patients and that overdosed chemotherapies do also increase the risks of neutropenic complications (the latter accounting for at least 2.700€ additional costs per hospital treatment for example febrile neutropenia at our institution), the 1€ for safely treating cancer patients appears very economical.

Our current electronic chemotherapy monitoring system has been amended to allow classification of more exact error types, which are categorized according to severity and the institution (CSC, pharmacy or ward) which successfully detects the error. This system allows us (a) to reduce the incidence and severity of chemotherapy medical errors even further—which we have shown with the additional data analysis of 2007 to indeed substantially decrease for chemotherapy orders—(b) to record errors on in- and outpatient wards, (c) to evaluate their frequency and potential harm for a patient and (d) to determine, how errors can best and most efficiently be prevented. The occurring SAECTx are reviewed monthly by our “MERKE team,” whose task is to identify strategies and guidelines for their prevention. These strategies are implemented in clinical guidelines, SOPs and safeguards.

Discussion

Acknowledging the fact that medical errors are not unusual, it is important to have a thorough and efficient monitoring system to prevent, or at least reduce mistakes being made. In the USA, there are 7,000 deaths per year due to medication errors, 6, 7 chemotherapy overdoses are also frequent,12 and adverse drug events are common: all costly causes of injuries during hospitalization.13

To enhance the quality and safety of cancer treatment and reduce treatment costs, we have established 2 error-management systems through our CSC team which reviews all chemotherapy protocols and orders above and beyond the ward unit- and pharmacy-based checks: PERMIT monitors errors in the ordering and administration of all chemotherapies, and MERKE records SAECTx on in- and outpatient wards. Both systems help to enhance the safety of chemotherapy treatment, since via PERMIT, errors can be successfully avoided before they harm a patient, and via MERKE, adverse events related to chemotherapies are identified, which help to implement and modify guidelines to effectively avoid them in the future.

Our analysis included a significant number of patients (2,337), all receiving chemotherapy (22,216 chemotherapy-orders). Compared to a study on medication safety in ambulatory chemotherapy settings by Gandhi et al. 9 (1606 patients, with 31% receiving chemotherapy treatment), our analysis included more patients treated on in- and outpatient wards. Although there is a general trend for ambulatory care, our number of in- and outpatients was similar, which can be attributed to our university center's obligation to treat demanding patients (e.g. elderly, frail and/or “end-stage”-patients, and those receiving second- to x-line therapies). Our analysis revealed that medical errors and SAECTx occur frequently, but can be effectively avoided or minimized by an experienced error-monitoring team. Our results demonstrated further that SOPs are mandatory, not only for chemotherapies on the ward, but also beforehand, when selecting the correct protocol and when ordering it.1, 2 Of note, our error rate for chemotherapy orders decreased from 4.2% in 2005 to 3.6% in 2006. The additional analysis of 10,674 chemotherapy orders in 2007 revealed a further decrease to 1.9%. We attribute this to the implementation of our electronic chemotherapy ordering and to our educational sessions within our institution on the preventability of chemotherapy errors, factors that have been confirmed in previous studies.6–9, 13–15

In contrast to our analysis, former reports have focused on errors relating to the chemotherapy order itself. 9, 14 Our error-management analysis seemed more precise and extensive, since we included errors in patient data and missing written informed consent forms. Scoring these errors led to our total detected error rate to appear higher than those in previous publications.9 Our data indicate, however, that it is important to detect all errors related to chemotherapy issues, since—only if errors are actively pursued—they can be efficiently avoided.

Medical errors have been described in inpatient surgical medication at a rate of 5%. 16 Gandhi et al. described errors in primary care of 8%, in medication orders for adult patients in the ambulatory setting of 3%, and in chemotherapy orders of 4%.9, 14 Compared to those data, our rate in chemotherapy errors was 3.8% and we clearly demonstrated that it is important to detect inpatient and outpatient errors, as the former was even higher than the latter (4.5 vs. 3.3%, respectively). The aforementioned analysis9 involved ambulatory chemotherapy orders alone, since this presents particularly demanding conditions, such as high volume, time pressure and less control. We determined error rates for outpatient and inpatient orders, which showed that for inpatient orders, stressful conditions resemble those in ambulatory settings, such as distraction due to a constantly high work load, similar high volume and time pressure restraints and frequent physician rotations from one ward to the other to comply with the physicians' educational curricula. Furthermore, the inpatient-given chemotherapies were more complex.

Our analysis illustrated that very few MERKE events (SAECTx) are the result of an undetected PERMIT (chemotherapy) error (0.079%): only 3 chemotherapy-related errors (mainly due to our intervention system) led to SAECTx, and most of them were efficiently avoided. Gandhi et al. 9 described a retrospective analysis of prescription errors with any medication (chemotherapies included): they observed intercepted errors in 45%, suggesting however, that 55% of these errors may have reached the patient. Our analysis was a prospective analysis whereby medical and administrative errors were predefined, immediately scored—when made—within the error scoring system, and also being corrected instantaneously, either by the CSC- and/or pharmacy-initiative. With our system of checking prescription orders before these reached any patient, we successfully intercepted 99.9% of errors before these could lead to any SAECTx. We succeeded in minimizing the risk for any cancer patient to receive an incorrect chemotherapy treatment (0.013%), thus promoting chemotherapy safety. Without such an effective error-control instance, we postulate that chemotherapy and other medication errors are abundant, often remain undetected and might frequently harm patients.17–22 Therefore, we recommend an additional team (as the CSC-team) to review chemotherapy orders prior to the medication reaching the patient and being administered.

One possible criticism of our detailed analysis is that few chemotherapy errors may have remained undetected: we could only identify those errors reported via our MERKE system, since their notification is based on voluntary terms. Frequent educational sessions are held at our department, aiming to increase the physicians' eagerness to report medical errors, thus all important cases were most likely recorded. We also checked the completeness of SAECTx-reports, verifying that all unexpected deaths and extravasations were fully reported to the CSC and found a consistency of 100%. Another criticism may be that treatment protocols bear important qualitative differences, whereby leukemia protocols are more complex, which is also true for combination chemotherapy versus single agent treatment and oral chemotherapies. We have not separately analyzed these differences, nevertheless defined our task to decipher how many and which errors occur in in- and outpatient cancer treatment. We also intended to alert the reader that—also within “simple” and “intermediately-complicated” chemotherapy protocols—notable errors may occur so that the general awareness—that at least 4% chemotherapy errors will occur with routine prescriptions (that can be efficiently avoided)—should increase. We would also like to point out that oral protocols are often lesser supervised by medical experts within the clinics but are taken at home, so that one might be mistaken arguing that these bear a lesser risk. On the contrary, the clinician often notices that oral chemotherapy has been taken at home in wrong doses, without comedication, for too long or not long enough, so that we consider it important to ensure that all chemotherapies—oral, i.v., complex and less complex ones—bear no prescription error risk for cancer patients and are correctly applied.

In conclusions, for those physicians and departments with frequent and potentially toxic medication, we recommend the use of a computerized chemotherapy ordering system, standardized chemotherapy treatment templates and an additional team to review chemotherapy orders before these reach any patient. We suggest that (1) our CSC surveillance system and SAECTx management may be of benefit to implement into other departments, that (2) assistance with a pharmaceutical and CSC team is complementary and thereby detection of approximately 20% of minor, but also serious chemotherapy-related errors, is ensured and (3) most importantly, that through this added safety check, 99.9% of our chemotherapy is delivered error-free to cancer patients. To our knowledge, this is the first error analysis in a prospective data collection which impressively shows a highly useful system how to detect and intercept nearly all (99.9%) chemotherapy prescription errors before these may harm any patient and lead to SAECTx. Our recommendations seem of great importance and likely to have wide applicability to other potentially harmful medications that various physicians do prescribe.

Acknowledgements

We are grateful to the entire MERKE team for their dedicated and diligent support and enthusiasm, to Mr. H. Schall for data support from his tumor-based patient documentation databank, our administration for providing patient numbers and data sources, our clinical pharmacy team, specifically Drs. Göbel and Lubrich, for their ongoing support, and our electronic specialist-team. We thank Dr. H. Henβ and the anonymous internal reviewer of our department for valuable suggestions and insightful comments, and the entire CSC staff for the dedication in fulfilling error-free chemotherapy orders. We are grateful to all clinicians, physicians, nurses and patients for participation in this analysis and Mrs. C. Cürten for careful proof-reading.

Ancillary