Medical Errors: Root Cause Analysis

RESEARCH ARTICLE

Impact of Technological Innovation on a Nursing Home Performance and on the Medication-use Process Safety

Chantal Baril & Viviane Gascon & Christel Brouillette

Received: 26 September 2013 /Accepted: 4 March 2014 /Published online: 14 March 2014 # Springer Science+Business Media New York 2014

Abstract Despite the fact that since 1985 the government of Québec increased by 5.75 % on average the amount of money spent on healthcare per year, little improvement was noted. It is obvious that an optimal use of resources is essential to reduce waiting times and provide safer and faster services to patients. The use of new technology can contribute to improve the healthcare system efficiency. Our study aims to assess the impact of a medication distribution technology on 1) the performance of a health and social services center’s pharmacy, 2) the performance of one care unit in a nursing home and on 3) the medication-use process safety. To measure performance we were inspired by the Lean approach. The results show that medication distribution technology is considered as an effec- tive way to significantly detect medication errors, to allow nurses to focus more on patients and pharmacy to react more rapidly to changes in patient medications.

Keywords Medication distribution technology . Hospital pharmacy . Technology implementation . Lean approach .

Performance

Introduction

In recent years, the healthcare system in the province of Québec has been the subject of much criticism related, among

others, to congestion of emergency rooms, long waiting time for a surgery and difficulty to get a family doctor. People often think that simply putting more money and resources in the healthcare system will improve the situation. Despite the fact that since 1985 the govern- ment of Québec increased by 5.75 % on average the amount of money spent on healthcare per year [1], little improvement was noted.

Health and social services in the province of Québec count more than 280 000 workers with a budget of $ 28 billion representing almost 45 % of the Québec Government spending [2]. It is obvious that an optimal use of resources is essential to reduce waiting times and provide safer and faster services to patients. The use of new technology can contribute to improve the healthcare system efficiency. New ways to provide social services and healthcare to patients must be de- veloped by considering continuous improvement ap- proaches based on Lean Manufacturing [3] or Six Sigma [4].

In recent years, the province of Québec has had to deal with major labor shortages in hospital’s pharmacy due to unattractive working conditions, lower remunera- tion than in private pharmacies and an additional training of two years to work in a hospital. This situation is not unique to Québec. Indeed shortages in pharmacists are noticed across Canada and in the United States. Consequently we are witnessing a general movement in hospitals towards better support to these professionals, including the modernization of their current working tools.

In 2007, the Ministry of Health and Social Services of Québec put in place instructions requiring that their health services local networks and hospitals become equipped with medication distribution technology by 2012. The province of Québec is divided into 95 local health services networks. Each

C. Baril (*) : C. Brouillette Industrial Engineering Department, Université du Québec à Trois-Rivières, 3351boul. des Forges, Trois-Rivières, Canada e-mail: [email protected]

V. Gascon Management Science Department, Université du Québec à Trois-Rivières, 3351boul. des Forges, Trois-Rivières, Canada

J Med Syst (2014) 38:22 DOI 10.1007/s10916-014-0022-4

of these local health services networks includes a facility called the health and social services center (HSSC), nursing homes, and for most of them at least one hospital. The HSSC is the basis of the local health services network, since it provides services accessibility, continuity and quality of care to its local population. The objective of these new instructions was to improve quality of care by substantially reducing the number of medical and medication errors, increase personnel productivity [5] and indirectly make hospitals more attractive to pharmacists. The medication distribution technology advo- cated by the HSSCs in Québec combines an automated pack- aging device for the pharmacy and mobile medication dis- pensing carts (MDCs) for care units.

Computerization and automation of the medication-use process are generally considered as a mean to substantially reduce the number of medication errors and ensure that nurs- ing staff can devote more time to patients. A recent study of Association des hôpitaux du Québec [6] concludes that: “A medication-use process not organized and supported by tech- nology is unsafe and unproductive: prescription non-reviewed by the pharmacist; duplication of paper and computer tools for the medication management by pharmacist, nurse and doctor; many manipulations of drugs in crowded work areas, many unnecessary movements, etc. ”. This paper presents how these technological devices were implemented in a local health services network in the province of Québec which includes a HSSC, a hospital and six nursing homes. The HSSC and one of the nursing homes are located in the same establishment. The five other nursing homes are separate institutions and are spread over an area of 10 square kilometers.

Our study aims to assess the impact of this technological innovation on 1) the performance of the HSSC’s pharmacy, 2) the performance of one care unit in a nursing home and on 3) the medication-use process safety. To measure performance we were inspired by the Lean approach. This paper is orga- nized as follows. First, a literature review on the impact of a secured medication-use process, automated dispensing sys- tems and mobile medication dispensing carts on medication errors and efficiency is presented. The use of Lean approaches in pharmacies is also reviewed. Second, the methodological framework is explained, the case study is described and the research hypothesis and variables are summarized. Finally, results and conclusions are discussed.

Literature review

The medication-use process

The medication-use process consists of a series of successive steps performed by different professionals: the prescription is a medical procedure, dispensing a medication is a pharmaceu- tical action and administration of a medication is a nursing or

medical act. The concept of “medication-use process” has become a standard in the province of Québec and abroad. Analysis of the medication-use process focusses either on hospitals of a whole region or country [7], or on a single hospital as a case study [8]. In both situations, the objectives remain to improve quality of services by reducing the number of medication errors and to improve the process efficiency.

Having in mind the objectives of improving service quality and process efficiency, l’Association des hôpitaux du Québec [6] felt that each hospital should have a global vision of its medication-use process and undertake its revision. The medication-use process is more complex than it seems, with many activities and tools to support them, and requires equip- ment and information systems, many of which are interfaced. The structure of the medication-use process considered in this paper is presented in Fig. 1.

Since the medication-use process consists of several steps, each step becomes a source of potential errors that can cause risks to the patient. Leape [9] has identified some sources of medication errors in the medication-use process. He found that 39 % of medication errors are associated with the medi- cation prescription being illegible or incomplete. These pre- scriptions are transmitted to the pharmacy by various means (fax, scanner, pneumatic tube or by a person), where they are captured (12 % errors) in a software. Preparing and sending medications (11 % errors) require different information tools and technical equipment such as databases used to validate information and automated packaging devices used for pack- aging medication safely. Medications are then delivered to the care units (38 % errors), where they are stored in the cabinets before being administered to patients.

Automated dispensing systems

Automated dispensing systems are used in pharmacies to pack medications in unit doses before delivering them to care units. Packaging contains information on the medication and the patient and sometimes a printed bar code. These automated dispensing systems eliminate manual tasks associated with filling alveolate blisters and packaging medications. Already in 1998, automated dispensing systems with a bar code reader interfaced with the information system of the pharmacy were considered more efficient than humans to prepare and pack medications [10]. If delivery of medications is made daily and mobile carts are used, automated packaging lightens nurses’ work by reducing handling involved in preparing and admin- istering medications.

Past research involving the use of automated dispensing devices has shown improved medication use in medical and surgical care units with a positive impact on errors related to administration time, omissions, and greater efficacy of work activities [11–13]. Medication error rate decreased from 2.9 % to 0.6 % due to these devices according to Weaver et al. [14]

22, Page 2 of 12 J Med Syst (2014) 38:22

and from 0.84 % to 0.65 % according to Klein et al. [15]. Another pre-post evaluation shows a medication error rate of less than 1 % [16]. In general, it is recognized that the use of automated dispensing systems increase the pharmacist’s effi- ciency as shown in Cohen [16]. Cohen [16] mentions that a saving of 9.5 pharmacy hours/day had been noticed in the Montreal’s Jewish General Hospital after introducing the use of such a system.

In a recent study [17] involving a pre-post intervention design comprising a control medical intensive care unit and a test medical intensive care unit, an automated dispensing system was implemented in one of the care units after a two- month observation period. Data about 1,476 medications ad- ministered to 115 patients was collected. After the implemen- tation of the automatic dispensing system researchers found a significant reduced percentage of total opportunities for error in the test unit compared with the control unit (13.5 % versus 18.6 %, respectively, p<0.05). Nurses were more satisfied with their working conditions and they perceived that they could spend more time with their patients.

In practice, the use of automated devices allows pharma- cists to spend more time on their clinical work and helps to reduce the stock of medications. However according to Hall [37], they contribute to only a slight decrease in worked hours per day/patient.

Medication dispensing carts

Security regarding the administration of medications to pa- tients do not always reach the level it should have and, despite the implementation of a daily medication delivery system in many centers, the risk to give a medication to the wrong

patient or during the wrong time of administration, or to omit recording the dose is high for nurses [18].

Camac et al. [19] observed a reduction in the medication error rate after comparing centralized distribution from nurs- ing stations with decentralized distribution from tablet cases at the patient bedside. Fisher et al. [20] also observed a reduction in the medication error rate after comparing centralized distri- bution with distribution of drugs using mobile carts. These authors claim that when medication preparation and adminis- tration are held closer to the patient, it helps to reduce the causes of interruptions and distractions. Alemanni et al. [21] found that several medical institutions do not use their medi- cation dispensing carts at the patient bedside because of how healthcare is provided to patients and of an insufficient num- ber of carts.

Finally, according to Bennett et al. [22], the appropriate use of medication dispensing carts should lighten the work of nursing personnel if the nursing work organization is planned accordingly.

Lean in pharmacies

In recent years, more and more health facilities have turned to a Lean approach to improve their processes. The Lean method helps in analyzing processes, identifying improvement possi- bilities and eliminating non-value added steps (waste) without having to make complex or expensive changes [3]. Several studies [23–26] demonstrate the benefits of Lean on the health sector and more specifically in pharmacies. Hummer [27] applied Lean to improve quality of care and to increase efficiency in the prescription renewal process. His study shows an increase of 57 % in the number of prescriptions completed per day. Hintzen [28] used the 5S method and

Manual prescriptions

Delivery/fax of prescriptions to the pharmacy

Care units

Transcription in a software

Preparation of medications using an automated

packaging device

Fillingof delivery carts

Delivery of medications to care units

Pharmacy

Medication administration to patients

Medication transfer to the dispensing carts

Care units

Fig. 1 Medication-use process in a HSSC

J Med Syst (2014) 38:22 Page 3 of 12, 22

mapping to improve the pharmacy process. Rearrangement of workstations in “U” and the one-piece flow have reduced waste by 40 %. A decrease in inventory costs of $ 50 000 and a 20 % reduction of the number of expired products was observed. The number of medication errors also decreased by 83 %. Houghton [29] has reduced the prescriptions lead time by 33 % and drug inventory costs by $ 153 000 through the elimination of non-value added steps (waste), the centraliza- tion of medications in the same area and the standardization of the medication classification method in a pharmacy. Sobek [23] used the Lean method to better define the roles of each employee and to improve communication between the phar- macy and the care units. The implementation of a visual card system has reduced the number of missing medications by 40 %.

Although scientific literature abounds with papers on med- ication distribution technology assessment or on the use of the Lean method to improve pharmacies’ processes, no study has been conducted on the implementation of a technological innovation according to the Lean approach. With a Lean approach, performance is not only evaluated by measuring times, delays or waste but also by considering the importance of value-added and non-value added activities. Our study aims to evaluate how the medication-use process performance can be improved with a new technology through the use of a Lean approach.

Methodological framework

Case study

This research was conducted in a health and social services center (HSSC) in the province of Québec. The HSSC’s phar- macy under study serves on average 710 patients spread over 20 care units in six nursing homes (Table 1).

Nursing Home 4, the pharmacy and the HSSC are all located in the same facility. Medications prepared by the pharmacy are delivered to nursing homes 1, 2, 3, 5 and 6 by truck. Pharmacy activities related to new prescriptions and medication production for nursing homes are carried out by two pharmacists and 7 pharmacy technician assistants (PTA). To produce (prepare and verify) medications, 4 PTAs are needed. The other three PTAs and the pharmacists are assigned to new prescriptions and other tasks.

Before implementing the new technology, medications needed by a patient to cover a 35 day period used to be prepared at once. The method consisted in stuffing every cell of an alveolate blister with only one tablet. Blisters were produced byManrex Ltée. One blister was filled for each type of medication, each patient and every period the medication had to be administered. For instance, if a patient had to take 10 medications per day, 10 different blisters were prepared, each one containing 35 pills (to cover the 35 day period). A phar- macy technician assistant (PTA) prepared the blisters manu- ally and labeled themwith the name of the patient. The blisters were next sorted out by patient and the time the medication needed to be administered. Preparing the medications with the manual method consisted in filling, sealing, identifying and classifying the blisters and putting them on racks. The racks were next verified to make sure that they were prepared adequately. Once these steps were completed, medications were delivered to the care units in nursing homes every five weeks, on Friday afternoon.

Once the medications were delivered at the care units, nurses would check each blister to make sure that there were no errors (dosage or type of medication). Table 2 shows the number of care units receiving their medications weekly over a 5 week period with the manual system.

Table 3 presents the dispensing methods investigated in this study before and after the implementation of the automat- ed medication packaging system.

The medication distribution technology chosen by the healthcare centre under study is an automated packaging device produced by the Amerisource Bergen (FastPak EXP model), for the pharmacy, combined with mobile dispensing carts produced by Capsa Solutions (the artromick series, ACM model), for the care units. The automated packaging device can pack solid medications in multi-dose bags (several different medications) and print on them information (patient’s name, period of med- ication administration, types of medications, their descrip- tions and dose). The automated packaging device contains 320 fixed canisters and 80 interchangeable canisters. The canisters contain medications and must be filled manually by a PTA. To reduce the number of medication errors, it is advised, before filling the canisters, to verify if the med- ication bottle bar code corresponds to its related canister bar code. Preparing the medications with the automated method consists in programming the automated device, preparing the medication tray (for medications not

Table 1 Number of care units by nursing home

Nursing home NH1 NH2 NH3 NH4 NH5 NH6

Number of care units 2 2 3 5 5 3

Number of patients 60 80 105 175 190 110

Table 2 Number of care units served weekly with the manual system

Week 1 2 3 4 5

Number of care units 4 4 4 4 4

Number of patients 145 155 130 145 135

22, Page 4 of 12 J Med Syst (2014) 38:22

included in the automated packaging device), filling the canisters if needed, and pack the medications. The bags are next inspected manually to verify if they contain the appropriate medication and quantity, if the bags are sealed correctly and if the medications are in good condition.

Once the medications are packaged and verified, they are put in the drawers of the delivery carts to be delivered to the care units. At the care units, the drawers are transferred to the dispensing carts. The dispensing carts are locked and can only be unlocked when the nurse or the assistant nurse slips his employee card over the sensor and enters his numerical code. The cart is only unlocked when the nurse gets the medications ready to be administered to patients.

With the implementation of the automated dispensing sys- tem, the pharmacy had for objective to deliver medications to care units located in nursing homes twice a week instead of once every 35 days, without increasing human resources.

Research hypothesis and variables

To assess the impact of this technological innovation on the performance of the HSSC’s pharmacy, of one care unit of a nursing home and on the medication-use process safety, three research hypotheses were considered:

1. Hypothesis 1: Technological innovation improves the pharmacy performance.

2. Hypothesis 2: Technological innovation improves the care unit performance.

3. Hypothesis 3: Technological innovation improves the medication-use process safety.

Data was collected in only one of the 20 care units since they are all quite similar. Indeed all care units use the same dispensing cart (before and after implementing the automated system). They all receive their medications from the same

pharmacy with the same automated packaging device. Moreover the number of patients does not vary much from one care unit to the other with an average of 36 patients and a standard deviation of 4,5 (10 care units have 40 patients, 6 have 30 patients and 4 have 35 patients. The chosen care unit includes 40 patients in a nursing home located outside the establishment where the pharmacy is located.

To validate our hypothesis, we considered one discrete independent variable, several dependent variables (discrete and continuous) and the Lean approach. The discrete inde- pendent variable is technological innovation that is the imple- mentation of a medication dispensing technology through a Lean approach. The discrete and continuous dependent vari- ables measured during this study are used to evaluate the pharmacy performance, one care unit’s performance and the medication use process safety. They are presented in Table 4.

Measurement methods and validity

In this study, the dependent variables were assessed and measured by various means. The methods used and the de- pendent variables measured by these methods are described in this section.

Table 3 Description of the medication dispensing methods

Before/after Illustration of packaging system

Manual dispensing method used before the technological innovation (Alveolate blisters)

Automated dispensing method used after the technological innovation (Bags)

Table 4 Description of dependent variables

Dependent variables related to the pharmacy performance

Number of employees needed to package and verify medications (discrete)

Time between delivery to care units (continuous)

Medication production time: time required to package medications into bags and verify them for one care unit (continuous)

Medication preparation time for one care unit (continuous)

Medication verification time for one care unit (continuous)

Number of medications thrown away by the pharmacy (discrete)

Percentage of value-added (VA) and non-value added (NVA) activities performed by PTAs (continuous)

Percentage of value-added (VA) and non-value added (NVA) activities performed by pharmacists (continuous)

Dependent variables related to care unit performance

Time of nurses’ tasks related to medications (continuous)

Time of nursing assistants’ tasks related to medications (continuous)

Medication distribution time by nursing assistants: time to distribute medications to patients (four times a day: 8 h AM, 12 h PM, 4 h PM, 8 h PM)

Dependent variables related to safety medication-use process

Number of steps required to complete the production of medications including preparation, verification and delivery to care units (discrete)

Total number of medication errors identified by the employees on one care unit (discrete)

Type of dispensing carts: secured or not secured dispensing carts (discrete)

J Med Syst (2014) 38:22 Page 5 of 12, 22

Time study

Time study is used as a work measure technique using a chronometer to measure the time necessary to perform each element of a given task under controlled condi- tions. To ensure validity of our results, the number of observed cycles must be large enough according to the population size. The Bureau International du Travail [30] proposes a table to determine the number of cycles to be timed based on the total number of mi- nutes involved in each cycle. The dependant variables measured with the time study and sample sizes are presented in Table 5.

For the pharmacy performance, the number of cycles cor- responds to the number of care units where the medication production time was measured.

For the care units’ performance, the number of cycles corresponds to the number of working shifts during which the time of nurses and nurse assistants’ tasks related to med- ications were measured.

Nurses’ tasks related to medications are:

& Verification of the blisters & Transcription of the prescription in the patient file and on

the medications administration sheet & Verification of the new medications administration sheets & Registration in the patient’s file & Verification of medications needing additional preparation

(not packaged automatically) & Validation about some medications with the pharmacy & Discussion with doctors about patients’ medications & Preparation and administration of some special drugs (not

packaged automatically)

Nursing assistants’ tasks related to medications are:

& Counting of narcotics & Preparation and administration of some medications out-

side the usual distribution periods & Registration in the patient’s file & Classification of the medications administration sheets

Nursing assistants distribute medications to patients. They thus have to put medications in a small bowl, crush it if necessary, and give it to the patient.

For each of the care units, data was collected for 2 working shifts (day and evening). A T-test was used to verify if differ- ences between the mean cycle timesmeasured before and after implementing the technology are significant.

Work sampling

Work measurement survey is regularly used as a method of observation [31–33]. Each employee in the team is observed and their tasks at different periods of the day are reported on an individual observation checklist. This work measurement survey provides a description of the different tasks performed by the employees. The observation checklist can be inserted in a laptop like Palm. Usually during an eight hour shift em- ployees are observed on a random basis around one hundred times. Validity is ensured by a number of observations large enough to reflect the real situation within a given margin of error. Barnes [34] has developed a table for determining the number of observations needed according to the required accuracy level. In this study, the confidence level adopted is 95 % with an accuracy level of 5 % for both measures. The dependant variables measured with work sampling and sam- ple sizes are presented in Table 6.

Definitions of value-added (VA) and non-value-added (NVA) activities used in this paper are the ones proposed by Liker [3] corresponding to the Lean approach. With the Lean approach, added value is an indicator to measure how an organization can create wealth. An activity performed directly on medications (programming the automated device, prepar- ing the medication tray and filling the canisters) by the PTAs is considered as a value-added activity. For the pharmacists, valued-added activities are prescriptions and consultations with patients. Non-value-added activities are usually activities leading to waste and involving movements, handling, over production, waiting, inventory management or verification activities. For people working in health services those defini- tions of non-value-added activities may be questionable, es- pecially verification activities which they often consider

Table 5 Dependent variables measured with time study and sample sizes required

Pharmacy performance Samples size required to ensure validity Samples size observed before and after

Medication preparation time 20 cycles 40 cycles

Medication verification time 20 cycles 40 cycles

Medication production time 20 cycles 40 cycles

Care units performance Sample size required to ensure validity Sample size observed before and after

Time of nurses’ tasks related to medications 5 cycles 7 cycles

Time of nursing assistants’ tasks related to medications 5 cycles 7 cycles

Medication distribution time 14 cycles 14 cycles

22, Page 6 of 12 J Med Syst (2014) 38:22

necessary. However we feel that our definitions of non-valued activities contribute in finding new ways to evaluate healthcare services’ performance having in mind the philoso- phy of continuous improvement. Observations before and after implementing the technology was performed over two working shifts.

Direct observation

This method involves systematic observation without judg- ment or interpretation of an object (person or group, activity, place, event or situation, for example) with note-taking orga- nized for an orderly return of the observations. In this case the dependant variable corresponds to the type of dispensing cart used on care units. The unsecured dispensing cart (before technology) and the secured dispensing cart (after technology) will be described and compared.

Counting

For some dependent variables, it is simply to calculate some values (employees, tablets, steps, etc.). For the pharmacy, the dependant variables measured by counting are time between delivery (in number of days), number of employees and number of medications thrown in one week. For the medication-use process safety, the dependant variables mea- sured by counting are the number of steps in the production process.

Incident/accident voluntary reporting forms

In the province of Québec, an incident/accident voluntary reporting form must be completed by any employee working in a health center and providing services to users when he/she is the witness of an incident or an accident involving a user [35]. In our study, errors reported on the incidents/accidents voluntary reporting forms, were collected for all nursing homes to perform a comparison analysis before and after implementation of the automated medication packaging sys- tem. With voluntary reporting the number of errors can be underestimated. However this method is not costly and gives some insights on the process within a given period of time. Errors were collected during 13 periods (4 weeks each) before implementing the automated packaging device and during 11 periods (4 weeks each) after the implementation.

AT-test was used to verify if differences between the mean numbers of errors reported before and after implementing the technology are significant. Data was collected directly at the pharmacy and in one care unit before, during and after imple- mentation of the medication dispensing system. Consequently data could be collected on the delivery times.

The first data collection was performed before the imple- mentation of the technological device while the second data collection was conducted nine months after its full implemen- tation. In both cases, the observed sample size exceeds the required sample size which demonstrates the validity of the results presented in the next Section.

Results and discussion

This section presents the results obtained after comparing data collected before and after implementation of the automated medication packaging system and the mobile dispensing carts. These data are used to validate our three hypotheses. Depending on the magnitude of the measured difference, the impact of the automated medication packaging system can be considered very positive (++), positive (+), neutral (=), nega- tive (−) or very negative (−−). If the number of variables whose impact is positive exceeds the number of variables whose impact is negative then the hypothesis is accepted, otherwise it is rejected. To assess whether the impact is sig- nificant, and if possible, a T-test was used with a confidence level of 95 %.

Validation of hypothesis 1

Recall that hypothesis 1 consists in verifying whether the automated medication packaging system increases the phar- macy performance. Table 7 shows the results obtained before and after implementation of the technology.

The use of an automated medication packaging system did not change the number of employees needed to prepare med- ications (Table 7). Therefore the impact of the automated medication packaging system on the pharmacy performance regarding the number of employees is neutral.

However the objective was to increase significantly the delivery frequency to care units from once every 35 days to twice a week. After introducing the technology, the pharmacy is now able to deliver medications twice a week (Table 8) instead of once every 35 days.

Table 6 Dependent variables measured by work sampling and sample sizes required

Pharmacy performance Samples size required to ensure validity Samples size observed

Value-added (VA) and non-value added (NVA) activities for PTAs 1,160 observations 1,349 observations

Value-added (VA) and non-value added (NVA) activities for pharmacists 1,160 observations 1,349 observations

J Med Syst (2014) 38:22 Page 7 of 12, 22

With this short time between deliveries it is now easier to respond rapidly to changes in medications made by doctors, therefore reducing the number of medication returned. For these reasons, the impact of the automated medication pack- aging system on the pharmacy performance measured by the time between deliveries is quite positive

Packaging medication times decreased from 910 min to 38 min, a 96 % reduction (Table 7). The T-test shows that the difference between two means is significant. For this reason, the impact of the automated medication packaging system on the pharmacy performance measured by packaging time is considered positive.

The time required to verify if the number of tablets in the bag corresponds to the information on it, if the bags are properly sealed and if the tablets are in good condition (ver- ification time) decreased from 203 min to 44 min (78 % difference). The T-test shows that the difference between the two means is significant. The impact of the automated medi- cation packaging system on the pharmacy performance mea- sured by the verification time is therefore considered positive. Even though the verification time has reduced greatly with the new technology, the proportion of time spent on this activity compared to the total production time increased from 18 % (203 min./1,113 min.) to 54 % (44 min./82 min.) (Table 7).

The medication production time is the sum of the packag- ing time and the verification time. Overall, the medication production time per care unit diminished with the automated medication packaging system (from 1,113 to 82 min). The impact of the automated medication packaging system on the pharmacy performance measured by the production time is

considered positive. To be able to package and verify medi- cations in these times, several Lean principles were imple- mented: visual control, SMED, and balancing tasks.

Let us note that each pharmacy employee works on a 480 min daily shift (75 min for lunch and coffee breaks leaving 405 min for production). With the manual method, the time available at the pharmacy to produce (prepare and verify) the medications at each care unit is:

405 min:=employee� 4 employees=day� 5 days 4 care units

¼ 2025 min:=care unit

With the automated method, the time available at the pharmacy to produce (prepare and verify) the medication at each care unit is:

405 min:=employee� 4 employees=day� 1 day 10 care units

¼ 162 min:=care unit

However more Lean principles could be considered in order to improve work organisation and to use resources more efficiently (employees and equipment). Indeed 162 min are available at the pharmacy to produce (prepare and verify) the medication needed at each care unit but only 82 min are used (Table 7). Considering verification as a non-valued-added activity, 44 of the 82 min needed to produce the medications are devoted to non-valued-added activities representing 54 % of the production time (Table 7). This leaves only 38 min to value-added activities divided among programming the auto- mated packaging device, preparing the medication trays and filling the canisters if needed (15 min) and packaging the

Table 7 Validation of hypothesis 1

Number of employees

Time between delivery

Mean time per care unit (in minutes) VA-NVA study (% of VA activities)

Quantity of medications thrown

Packaging Verification Production (packaging+verification)

PTAs Pharmacists

Manual method (before the technology)

4 35 days 910 203 1,113 42.1 % 21.1 % 3,112

Automated method (after the technology)

4 3-4 days 38 44 82 19.9 % 18.1 % 2,121

T-Test (p-value)

N/A N/A 0.00* 0.00* 0.00* N/A N/A N/A

Impact on pharmacy performance

= ++ + + + - = +

Student’s t-test : * Significant at 5 % (p-value<0.05)

Table 8 Number of care units served daily with the automated system

Day Monday Tuesday Wednesday Thursday Friday

Number of care units 10 10 Maintenance, cleaning and filling the automated system

10 10

Number of patients 355 355 355 355

22, Page 8 of 12 J Med Syst (2014) 38:22

medications (23 min). Let us note that during the remaining 80 min (162 – 82=80 min), value-added and non-valued- added activities such as working on sterile and non-sterile preparations, managing inventory, answering phone calls, un- pack orders, etc. are performed by the employees.

The automated packaging device is used only during 230 min (23 min/care unit×10 care units to pack medications) of the 480 min working shift. This represents a 48 % utiliza- tion rate of the automated packaging device. The automated packaging device could therefore be used to pack medications for other healthcare institutions or to help increase the delivery frequency from twice a week to once a day. However this would require reducing the non-value-added activity percent- age and consequently increasing the value-added activity percentage. It shows however that there is still space for more improvement.

The work sampling for PTAs shows a significant decrease of the value-added activity percentage after implementation of the automated medication packaging system. Value-added tasks are tasks related to the packaging of medications. Verifying a medication is considered in our study as a non- value added task. The decrease in the value-added activity percentage can be explained as follows: packaging medica- tions used to be donemanually by a PTA and therefore used to be considered as a value-added activity. Packaging is now an automated task leaving fewer tasks related to medications to the PTAs. It should thus allow PTAs to spend more time on value-added activities. However, with the automated medica- tion packaging system the proportion of time needed to verify the medications has increased. This increases the non-value added activity percentage and therefore reduces the value- added activity percentage. The impact of technology on the pharmacy performance measured by value-added activities performed by a PTA is therefore considered negative.

For pharmacists, the value-added activity percentage is quite similar to what it was before the implementation of the automated medication packaging system because they are not involved in packaging and verifying the medications. This difference of 3 % represents 12 min per working shift which is not significant. The impact of technology on the pharmacy performance measured by value-added activities of pharma- cists is considered neutral.

The number of medications thrown weekly decreased from 3,112 to 2,121 (32 % difference). This is due to the fact that the medications are delivered more frequently (twice a week instead of every 35 days). Therefore if some prescriptions are modified, adjustments can be made more quickly reducing the number of medications thrown away. Thus, the impact of technology on the pharmacy performance measured by the quantity of medication thrown is considered positive.

Since the number of variables on which the technology has a positive impact (5) is higher than the number of variables on which the technology has a negative impact (1), we conclude

that the automated medication packaging system increases the pharmacy performance in terms of number of employees, packaging time and number of medications thrown.

Validation of hypothesis 2

Recall that hypothesis 2 consists in verifying whether the mobile dispensing carts increase the care unit performance. Table 9 presents the results obtained before and after imple- mentation of this technology.

The new automated medication packaging system has allowed the elimination of several verification tasks: blisters, medication administration sheets and narcotics, which ex- plains the decrease in the percentage of time spent on tasks related to medications by nurses. The observed differences are of 15 % for day shifts and 16 % for evening shifts which represent 61 and 65 min respectively. Nurses can now spend this time with patients. The impact of technology on the care unit performance measured by the percentage of time spent on tasks related to medications by nurses is considered positive

For nursing assistants, there was little change in the per- centage of their time spent on tasks related to medications before and after the implementation of the automated medi- cation packaging system. The observed differences are of 5 % for day shifts and 3 % for evening shifts which represent 20 and 12 min respectively. This doesn’t leave enough time to perform other tasks. The impact of the automated medication packaging system on the care unit performance measured by the percentage of time spent on tasks related to medications by assistant nurses is considered neutral.

Technology did not affect the medication distribution time even if the mobile dispensing cart and the packaging task have changed considerably. Thus, the impact of technology on the care unit performance measured by the medication distribu- tion time is considered neutral.

The number of variables with a positive impact (1) being higher than the number of variables with a negative impact (0), it can be concluded that technology increases the care unit performance.

Validation of hypothesis 3

Hypothesis 3 consists in verifying whether the automated medication packaging system and the new dispensing carts makes the medication-use process safer for the patient. Table 10 presents the results obtained before and after the implementation of technology. To validate this hypothesis all care units were taken into account.

The use of the automated medication packaging system allowed eliminating four manual steps in the process of med- ication production. As there are fewer steps, there is less medication handling and therefore less risk of making mis- takes. The impact of the automated medication packaging

J Med Syst (2014) 38:22 Page 9 of 12, 22

system on the medication-use process safety measured by the number of steps in the process is positive.

Before implementing the automated system, 61 medication errors were reported on average over a 4 week period. After the implementation, it rose to 103 errors, a 68 % increase (Table 10). Even though this increase could be perceived as being negative, it shows that the system is more efficient to detect medication errors thus reducing risks for patients.

A study of Baril et al. [36] shows that the automated system contributes to detect medication errors before medications are given to patients, detect if a patient has not receive his med- ication and detect errors whenwrongmedications are assigned to wrong patients. Thus, the impact of the automated medica- tion packaging system on the medication-use process safety measured by the number of medication errors reported is positive.

The new type of mobile dispensing cart used for medica- tion distribution brought noticeable improvements on the care units. Previously, the blisters were placed on the cart in full view and hand of anybody. Now, the carts are always locked when the nurse is not preparing medications. The new cart provides one drawer per patient. To unlock the drawers, the nurse must use his employee card and enter a numeric code. It is thus possible to keep track of who opened the drawers and when. The impact of the technology on the medication-use

process safety measured by the type of mobile dispensing cart is positive.

Since the number of variables with a positive impact (3) is higher than the number of variables with a negative impact (0), we conclude that the technology increases medication-use process safety.

It is obvious that environmental conditions such as leader- ship, communication, resistance to change and staff involve- ment and leadership may have influenced some of our results [38]. Even though it is difficult to quantify the impact of these conditions, our results show that the total number of variables with a positive impact (9) exceeds without any doubt the one with a negative impact (1).

Conclusion

This research evaluated the impact of technological innova- tions on a pharmacy and a care unit in a nursing home performance and on the medication-use process safety. Our study showed that the implementation of technological inno- vations according to Lean principles can:

& Reduce time between delivery from 35 days to 3–4 days allowing the pharmacy to be more responsive to changes in

Table 9 Validation of hypothesis 2

% of nurses working time related to medications

% of assistant nurses working time related to medications

Medication distribution time (in minutes)

Day shift Evening shift Day shift Evening shift

Manual method (before the technology)

22 % 29 % 31 % 28 % 63

Automated method (after the technology)

7 % 13 % 26 % 25 % 58

T-Test (p-value)

N/A N/A N/A N/A 0.36

Impact on care unit performance + = =

Student’s t Test: * Significative at 5 % (p-value<0.05)

Table 10 Validation of hypothesis 3

Number of steps in the process (mapping)

Total average number of medication errors over 4 week periods

Type of dispensing cart

Manual method (before the technology)

29 61 Unsecured carts

Automated method (after the technology)

25 103 Secured carts

T-Test (p-value)

N/A 0.002* N/A

Impact on medication-use process safety + + +

Student’s t test : * Significant at 5 % (p-value<0.05)

22, Page 10 of 12 J Med Syst (2014) 38:22

medications proposed by doctors and reduce the number of medication returns and the time to manage these returns.

& Reduce the number of steps of the medication production process (prepare and verify) by 14 % inducing less med- ication handling and therefore reducing risks of error.

& Reduce packaging medication time by 44 % with a time between deliveries of 3–4 days.

& Reduce the number of medications thrown by 32 % allowing a better management of medication costs and inventories

& Reduce the percentage of the time spent by nurses on medication related tasks in care units by 68 % on day shifts and by 55% on evening shifts; for nursing assistants the reduction is 16 % on day shifts and 11 % on evening shifts. Nurses can now spend more time with patients.

& Considering the number of errors reported the technology is more efficient to detect medication errors thus reducing risks for patients.

In conclusion, technological innovations evaluated in this study are considered as an effective way to significantly detect medication errors, to allow nurses to focus more on patients and pharmacy to react more rapidly to changes in patient medications. However, this study has shown an increase in the proportion of time allocated to verifying medications. This leaves space for improvement to make the pharmacy more efficient. The use of bar coding could help to reduce this verification time. Moreover going further with a VA-NVA task analysis through a Lean approach could allow more improvement. It could then be possible to produce medica- tions for all care units, to increase flexibility when prescrip- tions are modified or to assign VA tasks to PTAs (working on sterile and non-sterile preparations, carrying out prescriptions, performing dose calculations or managing medication inven- tory in nursing homes). This paper shows that the technolog- ical innovation is the first step towards a safer and more efficient medication-use process.

References

1. Canadian Institute for Health Information, (2009), Tendances des dépenses nationales de santé : 1975 à 2009. 193 pages.

2. Québec ministry of finance, (2010), Plan budgétaire 2010–2011, 408 pages

3. Liker, J., (2009), Lemodèle Toyota: 14 principes qui feront la réussite de votre entreprise, Paris, Pearson Education, France, 392 pages

4. Breyfogle, F.W., (1999), Implementing Six Sigma: smarter solutions using statistical methods, John Wiley & Sons, 7e édition, 689 pages.

5. MSSS - Ministère de la santé et des services sociaux, (2005), Les systèmes automatisés et robotisés utilisés pour la distribution des médicaments dans les établissements de santé au Québec, Rapport et recommandations du groupe de travail, 116 pages.

6. Association des hôpitaux du Québec, (2004), La pratique pharmaceutique dans les établissements du Québec, analyse et recommandations, Comité de travail mixte de l’AHQ–directeurs des services professionnels et chefs de département de pharmacie, 33 pages.

7. Ministère de l’emploi et de la solidarité, (2001), L’informatisation du circuit du médicament dans les établissements de santé – Approche par l’analyse de la valeur : Quels projets pour quels objectifs?, Direction de l’hospitalisation et de l’organisation des soins, ministère délégué à la Santé, ministère de l’Emploi et de la Solidarité, Paris, France, 128 pages.

8. Huet, P., Paris, J.-L., Gourgand, M. et Kouiss, K., (2010), Modèle de connaissance générique du circuit du médicament dans un hôpital, Conférence international de modélisation et simulation -MOSIM’10, Hammamet, Tunisie, 8 pages.

9. Leape, L. L., Bates, D.W., Cullen, D. J., Cooper, J., Demonaco, H. J., Gallivan, T., Hallisey, R., Ives, J., Laird, N., and Laffel, G., Systems analysis of adverse drug events. J Am Med Assoc 274(1):35–43, 1995.

10. Novek, J., Clinical or industry pharmacy ? : Case studies of hospital pharmacy automation in Canada and France. Int J Health Serv 28(3): 445–465, 1998.

11. Borel, J. M., and Rascati, K. L., Effect of an automated, nursing unit- based drug-dispensing device on medication errors. Am J Health Syst Pharm 52:1875–1879, 1995.

12. Shirley, K. L., Effect of an automated dispensing system on medica- tion administration time. Am J Health Syst Pharm 56:1542–1545, 1999.

13. Guerrero, R., Nickman, N., and Jorgenson, J., Work activities before and after implementation of an automated dispensing system. Am J Health Syst Pharm 53:548–554, 1996.

14. Weaver, P.E. et Perini, V.J., (1998), Random sampling for quality assurance of the RxOBOT dispensing system, American Society of Health Systems Pharmacists Mid-Year Clinical Meeting, Las Vegas.

15. Klein, E. G., Santora, J. A., Pascale, P. M., and Kitrenos, J. G., Medication cart-filling time, accuracy, and cost with an automated dispensing system. Am J Hosp Pharm 51(9):1193–1196, 1994.

16. Cohen, E., Serour, S., and Zelovies, M., L’automatisation en l’an 2000. Pharmactuel 33(6):164–167, 2000.

17. Chapuis, C., Roustit, M., Gaelle, B., et al., Automated drug dispens- ing system reduces medication errors in an intensive care setting.Crit Care Med 38:1–7, 2010.

18. Van den Bemt, P. M. L. A., Idzingac, J. C., Robertz, H., Groot, K. D., and Pels, N., Medication Administration Errors in Nursing Homes Using an Automated Medication Dispensing Systems. J Am Med Inform Assoc 16(4):486–492, 2009.

19. Camac, K. J., Fisher, M. J., and Norris, D. E., Medication errors: a comparative study of drug storage sites. Aust J Hosp Pharm 26:34– 37, 1996.

20. Fisher, M., Norris, D., Camac, K., and Hawkshaw, B., A comparison of medication errors between two storage sites. Contemp Nurse 11: 55–59, 2001.

21. Alemanni, J., Touzin, K., Bussières, J. F., Descoteaux, R., and Lemay, M., An assessment of drug administration compliance in a university hospital center. J Eval Clin Pract 16(5):920–926, 2010.

22. Bennett, J., Harper-Femson, L. A., Tone, J., and Rajmohamed, Y., Improving medication administration systems: an evaluation study. Can Nurse 102:35–39, 2006.

23. Sobek Durward k.J.C., (2003), Applying the Toyota Production System to a hospital pharmacy, Industrial Engineering Research Conference, Montana State University.

24. Andersson, S., Armstrong, A., Björe, A., Bowker, S., Chapman, S., Davies, R., et al., Making medicinal chemistry more effective- application of Lean Sigma to improve processes, speed and quality. Drug Discov Today 14:598–604, 2009.

25. Hanuscak, T. L., Szeinbach, S. L., Seoane-Vazquez, E., Reichert, B. J., and McCluskey, C. F., Evaluation of causes and frequency of

J Med Syst (2014) 38:22 Page 11 of 12, 22

medication errors during information technology downtime. Am J Health Syst Pharm 66:1119–1124, 2009.

26. Al-Araidah, O., Momani, A., Khasawneh,M., andMomani, M., Lead- time reduction utilizing lean tools applied to healthcare: the inpatient pharmacy at a local hospital. J Healthc Qual 32:59–66, 2010.

27. Hummer, J. D., Improvement in prescription renewal handling by application of the lean process. Nurs Econ 27:197–201, 2009.

28. Hintzen, B. L., Knoer, S. J., Van Dyke, C. J., and Milavitz, B. S., Effect of lean process improvement techniques on a university hos- pital inpatient pharmacy. Am J Health Syst Pharm 66:2042–2047, 2009.

29. Houghton, P., (2006), Improving pharmacy service, QualityDigest, Octobre.

30. Bureau international du travail, (1996), Introduction à l’étude du travail, 3e edition, Genève, Suisse, 524 pages.

31. Urden, L., and Roode, J. L., Work Sampling : A Decision-Making Tool for Determining Resources and Work Redesign. J Nurs Adm 27(9):34–41, 1997.

32. Jenkins, J. L., and ORTH, D. L., Productivity improvement through work sampling. Cost Eng 46(3):27–32, 2004.

33. Hodgkin, S., Warburton, J., and Savy, P., Using mixed methods to develop and implement a work sampling tool in residential aged care. Int. J Mult Res Approaches 6(1):23–32, 2012.

34. Barnes, R.M., (1980), Motion and Time Study: Design and Measurement of Work, John Wiley & Sons, 789 pages.

35. Ordre des infirmières et des infirmiers du Québec (OIIQ), (2004), L’administration de médicaments : rappel des obligations déontologiques, Le Journal, Vol. 2 No. 2, http://www.oiiq.org/ uploads/periodiques/Journal/vol2no2/ss02.htm, Consulté le 7 février 2012.

36. Baril, C., Gascon, V., St-Pierre, L. and Lagacé, D., (2013), Technology and medication errors: impact in nursing homes, International Journal of Health Care Quality Assurance, Accepted.

37. Hall, K. et Wilgosh, C., (2008), Hospital Pharmacy in Canada: 2007/2008 Report, Toronto: Eli Lilly Canada, Available: http:// www.lillyhospitalsurvey.ca/hpc2/content/rep_2008_toc.asp (accessed 2012 Mar 17)

38. White, L., Key success factors in the implementation of an automated dispensing system in a community pharmacy. Int J Healthc Technol Manag 10(6):393–407, 2009.

22, Page 12 of 12 J Med Syst (2014) 38:22

Journal of Medical Systems is a copyright of Springer, 2014. All Rights Reserved.

Journal of Medical Systems is a copyright of Springer, 2014. All Rights Reserved.