Partopens at the Point of Care

Evaluating Digital Pen-based Maternal Labor Monitoring in Kenya

Heather Underwood

, John Ong’ech

, Maya Appley

, Sara Rosenblum

, Addie Crawley

S. Revi Sterling

and John K. Bennett

ATLAS Institute, University of Colorado Boulder, Boulder, CO, U.S.A.

Kenyatta National Hospital, Nairobi, Kenya

Keywords: Digital Pens, Maternal Health, ICTD, Kenya, Partograph, Clinical Decision-Support System (CDSS).

Abstract: The goal of the PartoPen system is to enhance the partograph, a paper-based labor monitoring tool intended

to promote timely delivery of quality care by birth attendants in developing countries. The PartoPen digital

pen hardware and software system supports partograph use by providing audio instructions for measuring

and recording labor progress indicators, real-time decision support based on recorded measurements, and

time-based patient-specific reminders for taking measurements. Earlier work found the PartoPen system

effective in nursing classrooms at the University of Nairobi (UoN), Kenya where the PartoPen was used to

support teaching and training of students in maternal labor monitoring procedures. This paper presents the

results of several follow-on studies conducted in the maternity ward of Kenyatta National Hospital (KNH)

in Nairobi. Through these studies, we successively refine our understanding of the benefits of PartoPen use

in this setting. We also identify and discuss the interrelated factors impacting PartoPen adoption and use in

the labor ward at KNH, and review the challenges and opportunities likely to face digital pen deployments

in other healthcare settings.

1 INTRODUCTION

The World Health Organization (WHO) estimates

that 300,000 women die every year due to

pregnancy-related complications, most of which

occur in developing countries (World Health

Organization 2012). Timely and informed labor

monitoring by a skilled attendant can help prevent

many of the main causes of maternal death –

hemorrhage, infection, unsafe abortion, eclampsia,

and obstructed labor (United Nations 2010).

Globally, the WHO promotes the paper partograph

as an effective and cost-efficient tool for monitoring

labor, and preventing obstructed labor and resulting

complications. Used correctly, the partograph

provides decision support that assists in early

detection of maternal and fetal complications during

labor. Especially in rural clinics, early detection

allows transport decisions to be made in time for a

woman to reach a regional facility capable of

performing emergency obstetric procedures.

Despite the positive reports of improved

maternal outcomes resulting from correct partograph

use (Kwast et al. 1994; Mathai 2009; Lavender et al.

2013), several recent studies in Kenya have reported

underuse and incorrect use of the partograph at all

levels of maternity care (Opiah et al. 2012; Qureshi

et al. 2010; Lavender et al. 2011). The well-

documented barriers to partograph use include

partograph shortage, staff shortage, low partograph

knowledge and training, and the perspective that the

partograph is time consuming and redundant (Opiah

et al. 2012). The goal of the PartoPen project is to

mitigate some of the barriers preventing correct and

widespread partograph adoption using an interactive

digital pen, dedicated pen software, and partograph

forms printed with a background dot pattern that is

recognized by the pen. Using only the digital pen

and the existing paper form, the PartoPen addresses

training and resource barriers by providing audio-

based decision support, patient-specific reminders,

and partograph use instructions. Prior PartoPen work

at the University of Nairobi (Underwood, Sterling &

Bennett 2013b) suggests that the PartoPen is

effective in multiple healthcare settings: initial

training, training reinforcement, and use with actual

patients. These results motivated two of the studies

described in this paper. These studies focus on

Underwood H., Ong’ech J., Appley M., Rosenblum S., Crawley A., Sterling S. and K. Bennett J..

Partopens at the Point of Care - Evaluating Digital Pen-based Maternal Labor Monitoring in Kenya.

DOI: 10.5220/0004798600900100

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2014), pages 90-100

ISBN: 978-989-758-010-9

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

populations at two ends of the healthcare spectrum:

nursing students with little training or clinical

experience using the partograph, and nurse

midwives at KNH, who are well-trained and

generally have many years of experience using the

partograph in the labor ward.

Our previous work examined the effect of

PartoPen use on partograph completion in nursing

classrooms with third and fourth year nursing

students. The results of follow-on maternity ward

studies are presented here. In addition, we discuss

various environmental factors that led to different

outcomes in the two studies.

We first summarize the results from the PartoPen

nursing student study, and give an overview of the

technological components of the PartoPen system.

The remainder of the paper discusses the results of

maternity ward studies conducted at KNH in July

and August 2012, with follow-up in July 2013.

2 BACKGROUND

2.1 PartoPen Software System

Over the past fifty years, a large body of work on

pen-and-paper computing and pen-and-paper user

interfaces (PPUIs) has been developed. More

recently, digital pens have been used and evaluated

for usability and efficiency in the healthcare context

both by patients (Lind et al. 2007; Lind et al. 2008),

and by nurses (Procuniar & Murphy 2008; Estellat et

al. 2008). The key findings in these studies is that

digital pen technology is an intuitive and usable

technology with great potential, but healthcare-

specific digital pen applications must be designed to

meet the actual needs of the user. The Livescribe

(LS) digital pen technology (Livescribe Inc.) used

by the PartoPen system captures pen input and

digitizes paper content by using a unique location

tracking and page identification technique patented

by the Anoto AB group (Anoto).

The current implementation of the PartoPen

system uses the LS Echo digital pen running the

custom PartoPen software that allows the pen to

meaningfully interact with the paper partograph. The

Echo pen has a built in microphone, speaker, and

OLED display. The pen relies on a rechargeable

lithium ion battery, which is advertized to last about

36 hours during normal use. A battery life of 20 to

26 hours has been observed, depending upon the

amount of audio played during use. Pens can store

between 200 and 800 hours of audio, or the

equivalent amount of text data, depending on the pen

model, and all stored data can be downloaded to a

desktop computer using a standard micro-USB

cable.

One of the goals of the PartoPen system is to

enforce birth attendant training on correct use of the

partograph, as this has been cited as a significant

barrier to consistent use of the form. The WHO

partograph user manual, and a local partograph

manual issued to clinics by the Kenyan Ministry of

Health, are the primary resources for partograph

instruction in Kenya. The PartoPen system makes

the instructions found in these manuals accessible

directly from the partograph itself. The PartoPen

uses fixed print “button” regions around the

partograph text to provide verbatim audio recordings

of the instructions found in the partograph use

manuals. Thus, by tapping on these “buttons,”

nurses and nursing students can get short

informational prompts on how to use each section of

the form correctly.

One of the most commonly cited barriers to

partograph use is the inability to interpret the data

plotted on the partograph and take appropriate

action. Nursing students and less-experienced nurses

often plot the data correctly on the partograph, but

fail to derive the meaning of the plotted data, or do

not remember what actions to take based on the data

that they have plotted. The decision support

functionality of the PartoPen addresses these issues

by interpreting plotted data based upon page

location, and providing real-time feedback on the

appropriate actions to take. Currently, the PartoPen

provides decision support in three of the partograph

sections: cervical dilation, liquor/amniotic fluid, and

fetal heart rate.

The labor ward at KNH delivers approximately

1000 babies during the “busy” months from October

to March, or roughly 34 babies every day. On

average there are 4-6 nurses working at a time, and

based on survey data collected at the end of the

PartoPen study, nurses on average are responsible

for 5-7 patients during a day shift and 7-10 patients

during a night shift. The WHO recommends a

maximum ratio of one nurse to three patients to

ensure compliance with partograph completion

protocols. In the survey, nurses nearly unanimously

reported that staff shortage is the most common

reason for low partograph completion rates. While

the PartoPen does not replace nurses or supplement

the shortage of nurses in the labor ward, it does

provide a reminder system intended to help busy and

tired nurses keep track of when patients need

measurements taken.

PartopensatthePointofCare-EvaluatingDigitalPen-basedMaternalLaborMonitoringinKenya

2.2 PartoPen Nursing Student Study

Ninety-five nursing students in their third and fourth

years of study at the UoN School of Nursing

Sciences participated in the study. Students were

asked to complete a partograph worksheet, which

consisted of two patient case studies and two blank

partograph forms printed with the dot pattern. The

students recorded the patient data on the blank

partograph forms as if they were actively monitoring

that patient during labor. In each worksheet, students

received two of three possible patient case studies.

The three case studies represent three possible labor

outcomes. Mrs. A’s data represents an

uncomplicated, timely labor that progresses without

medical intervention. Mrs. B’s data illustrates a case

of prolonged or obstructed labor. Mrs. C’s labor

progression data illustrates an increasing number of

complications, including fetal distress, and

ultimately results in a cesarean section.

The students were first divided into three groups.

Group 1 was the control group, and Groups 2 and 3

were the intervention groups. Group 1 students

completed a partograph worksheet task with a

PartoPen in “silent logging mode,” and received no

instructions on how to use the technology. Group 2

completed the same worksheet task, but used a fully

functional PartoPen in “use” mode. The PartoPen

software in “use” mode for the student pilot has two

main components: instructions and decision support.

Group 2 received no training on how to use the

technology. Group 3 received a fully functional

PartoPen in “use” mode and a 15-minute

introduction and demonstration of the PartoPen

system before completing the partograph worksheet

task.

Using an unpaired t-test, the difference between

Group 1 (M=.520, SD=.141) and Group 3 (M=.722,

SD=.089) for the patient case study Mrs. C, a

prolonged labor resulting in a CS, was found to be

significant; t(8)=2.709, p=0.0267. These data

suggest that for more challenging or complex labor

cases, the availability and utilization of the PartoPen

instruction prompts promotes more accurate form

completion.

After each group completed the worksheet task,

students were asked to participate in a short focus

group session. Students unanimously reported that

plotting contractions was one of the most difficult

sections of the partograph, because both duration

and frequency are plotted together using a

combination of bar charts and coloring patterns.

Students also reported unanimously that plotting

descent of the fetal head was particularly

challenging. Difficulties plotting descent of the fetal

head can also be attributed to having to plot on the

same graph as another measurement (cervical

dilation), but may also be due in part to the nursing

school transitioning to a different partograph version

that requires users to plot the descent in increments

of one instead of two, and on the left side of the

graph instead of the right.

The completion results of the ‘contractions’

section of the partograph show improvements in all

three case studies (Mrs. A, B, and C) between

groups that did and did not use the PartoPen. There

was a statistically significant improvement in

contraction plotting on the Mrs. C case study

between Group 1 (M=.513, SD, .232) and Group 3

(M=.803, SD=.139); t(8)=2.399, p=0.0433.

‘Descent of fetal head’ measurements also

showed a significant improvement on the Mrs. C

case study between Group 1 (M=.337, SD=.152) and

Group 2 (M=.585, SD=.162); t(10)=2.699,

p=0.0223.In addition, there was a very significant

improvement on descent plotting on the Mrs. C case

study between Group1 and Group 3 (M=.705,

SD=.137); t(8)=4.028, p=0.0038.

The UoN PartoPen study indicated that PartoPen

use in classrooms can improve students’ ability to

correctly complete a partograph form. The study

results also suggest that significant PartoPen training

is not required to achieve these benefits. A

significant increase in partograph completion and

accuracy was observed with minimal prior training,

due to the intuitive design, push-based functionality,

and the enhancement – rather than replacement – of

the current paper-based system.

3 MATERNITY WARD STUDY

The follow-on PartoPen studies in the maternity

ward at KNH examined the impact of the digital pen

software system on partograph completion by nurse-

midwives monitoring patients during actual labor.

The partograph used in this study is pictured in

Appendix A. The first study was conducted in July-

August of 2012 at Kenyatta National Hospital. The

study was designed as a pre- and post study, which

compared partograph completion rates for

partographs completed in June (without the PartoPen

system) and in August when the PartoPen system

was in use by nurses.

3.1 Methodology

Currently KNH evaluates partograph completion

HEALTHINF2014-InternationalConferenceonHealthInformatics

using a rubric with four options – “complete,”

“incomplete,” “correct,” or “incorrect” – boxes for

each partograph category (fetal heart rate, moulding,

cervical dilation, etc.) Due to the wide range of

variation in how partographs are used and

completed, this basic evaluation rubric does not

correctly capture the actual completeness of the

partograph, or the real usefulness the data recorded

on the partograph. We therefore created a new rubric

that would better assess these measures. This rubric

is built upon the basic tenants of the evaluation tool

used by KNH. The rubric has grading criteria for

each partograph category, including a separate set of

grading criteria for the labor summary printed at the

bottom of each partograph. For each partograph

category there are three grading criteria: (1)

measurements recorded, (2) symbols correct, and (3)

spacing correct. The total possible value for each of

these grading criteria is determined by the time

between patient admission and delivery.

Previous efforts to evaluate partograph

completion required researchers to continuously

observe nurses during labor monitoring to assess

partograph completion (Rotich et al. 2011), or

researchers were required to use a coarse-grained

categorization scheme similar to the original KNH

rubric (Khonje 2012).

The new rubric also has some limitations. Chief

among these is the fact that the time of admission is

used to determine how many measurements are

expected on the partograph form. The time of

admission, however, does not always accurately

reflect when the woman went into active labor or

when the partograph was started.

All of the partographs collected during the study

were first categorized by delivery mechanism –

spontaneous vaginal delivery (SVD) and cesarean

section (CS). The CS deliveries were further

categorized into emergency CS (EmCS) and “other”,

which includes voluntary CS and CS due to previous

CS scars. Deliveries of twins, triplets, or deliveries

lasting less than one hour were noted among the

SVD partographs, but not included in the data

analysis because partographs a) are not designed to

monitor multiple births, and b) do not provide

beneficial monitoring for labors that are less than

one hour in length.

3.2 Quantitative Results

Initially all of the collected partographs from June

(369) were compared to all of the collected

partographs from August (457). This blanket

analysis of partograph completion rates between

June and August attempted to capture any broad

improvements that may have occurred due to

researcher presence, or a general increase in interest

and attention to the partograph because of the

PartoPen study taking place. These results are

captured in a previous paper (Underwood, Sterling

& Bennett 2013a), and briefly summarized below.

3.2.1 Phase 1 Data Analysis Summary

In the initial data analysis process, the collected

partographs were graded and checked by two pairs

of research assistants according to the new

evaluation rubric. Each partograph received two

scores: a composite completion score and a

summary score. The composite score was calculated

by dividing the number of points received by the

total number of points possible for all three grading

criteria (mark existence, correct mark symbol, and

correct mark spacing) for each partograph section

(fetal heart rate, cervical dilation, etc.). The

summary score reflects the completion percentage

for the partograph summary section at the bottom of

the form, which summarizes the labor (and is usually

completed after a patient delivers).

Based on this initial analysis, there were no

statistically significant improvements in the

composite scores between June and August

partographs. However, there were improvements in

the summary scores for both SVDs and CSs. This

result can be attributed to several possible factors.

First, the partographs used in June were slanted and

blurred due to frequent photocopying, whereas the

partographs used in August were printed

individually (to assure the unique dot pattern on

each form). This made the August partographs

significantly easier to read and, presumably easier to

complete. Second, the improvement in summary

scores was likely a result of the increased awareness

and underscored importance of the partograph that

occurred during the PartoPen study.

The lack of improvement in completion rates for

the overall composite partograph scores in the

presence of the increased focus on the partograph is

likely due to the impact of understaffing.

Understaffing thwarts completing the graphical

portion of the partograph because the ratio of nurses

to patients (often between 1:5 and 1:10) does not

allow for regular half-hour measurements to be

taken for each patient. The PartoPen system cannot

replace trained staff members, and does not directly

address the understaffing barrier to partograph

completion.

The data from the broad comparison of June and

PartopensatthePointofCare-EvaluatingDigitalPen-basedMaternalLaborMonitoringinKenya

August partographs suggest that the PartoPen does

not have an overall impact on partograph

completion, at least not in facilities like KNH, which

have highly trained and experienced, but critically

over-worked, staff.

3.2.3 Phase 2 Data Analysis

After the initial broad data analysis, a more fine-

grained analysis was performed on the PartoPen data

to ascertain if and how the PartoPen functionality

impacted partograph completion rates. Partographs

completed in June were compared to August

partographs that were actually completed with the

PartoPen. The PartoPen was used to complete 48 of

these partograph forms. PartoPens were only given

to nurses at KNH during the study, which excluded

the nursing students who were actively working in

the labor ward as part of their clinical rotation.

Student-completed partographs in August, which

were not completed with a PartoPen, were excluded

from Phase 2 analysis. In addition, many

partographs were only partially completed with the

PartoPen, due to nurse rotations and patient

handoffs. These partially completed partographs

were also excluded from Phase 2 analysis.

The comparison of all the partographs completed

with PartoPens versus the August partographs not

completed with the PartoPens versus all of the June

partographs is represented in the Figure 1. This

histogram illustrates that August partographs

completed with the PartoPens never received a

completion score lower that 25%, whereas both June

and August partographs completed without the

PartoPen did. Additionally, the August partographs

completed with the PartoPen had the highest

Figure 1: A histogram of partograph completion scores for

August partographs completed with a PartoPen, August

partographs not completed with a PartoPen, and June

partographs.

percentage of partographs in the 75-100%

completion range.

4 2013 FOLLOW-UP STUDY

At the conclusion of the PartoPen maternity ward

study, the nurses and hospital staff had successfully

incorporated the PartoPen system into the daily

operations of the labor ward. The system in place at

KNH at the end of the study included 20 PartoPens,

a printer capable of printing the dot pattern

partographs, PartoPen chargers, and extra pen caps

and ink replacements. Nine months after the

completion of the 2012 PartoPen maternity ward

study, a follow-up study was conducted with KNH

nurses to assess the performance and impact of the

PartoPen system.

Upon returning to the KNH labor ward in May

2013, researchers found that all 20 of the PartoPens

were accounted for, 19 out of the 20 PartoPen were

functional, and over 600 digital partograph records

were present on the PartoPens ranging from

September 2012 to April of 2013 (the printer used to

print dot-patterned partographs failed in April 2013,

and replacement parts were not readily available in

Nairobi). During the May 2013 visit, PartoPen

researchers identified a local printer model

alternative and worked with hospital IT staff to

establish a recurring printer toner order, so as to

remove this responsibility from the already busy

nurses and record-office staff in the labor ward.

4.1 Methodology

During the May 2013 visit, PartoPen researchers

surveyed KNH labor ward nurses regarding their

impressions of PartoPen deployment and use.

Twenty-six nurses completed a paper survey about

the PartoPen and its affect on labor ward operations

and patient care.

The paper survey consisted of nine questions

(see Appendix B); four YES/NO questions, two

Likert scale questions, and three free response

questions. The survey also included basic

demographic information, and a rank-order question

where nurses ranked the importance of the

partograph sections. The surveys were completed

during morning patient handoff. Nurses took an

average of 15 minutes to complete the survey, and

the nurses were not compensated for their time.

HEALTHINF2014-InternationalConferenceonHealthInformatics

4.2 Follow-up Survey Results

Thirteen of the 26 nurses who completed the survey

felt they were ‘experts’ using the PartoPen system.

The majority of the nurses (19 out of 26) used some

combination of partograph information and other

patient information to make decisions about patient

care. The nurses were asked to rank in order of

importance the different sections of the partograph

as they relate to providing quality patient care. Nine

nurses ranked patient name and age as the most

important section of the partograph to complete.

Eight nurses ranked fetal heart rate as the most

important section of the partograph, and seven

nurses ranked the partograph sections sequentially

(i.e., the most important section is the topmost

section of the form, and the least important is the

bottommost portion of the form). One nurse ranked

contraction frequency as the most important, and

one nurse ranked cervical dilation as the most

important section of the partograph to complete. The

responses from the survey suggest that certain

information on the partograph is more useful for

making critical decisions about patient care, which

may indicate that a simplified and restructured form

that highlights these sections (and makes them easier

to complete) could be useful in this setting. Nurses

largely prioritized patient information and fetal heart

rate as the most important portions of the form. In

the PartoPen study, some of the qualitative feedback

received by nurses indicated that using larger boxes

for information entry for these sections considerably

improved the usability and readability of these

critical pieces of information.

The survey also asked nurses to identify if there

are certain kinds of labor or patients who do not

need a partograph. Twenty of the 26 nurses said that

there were patients who do not need a partograph

during labor. Elective cesarean sections, false labors,

and patients who arrive already in the second stage

of labor were the most common responses for labors

that do not require a partograph to monitor labor

progress. Elective cesarean sections are scheduled in

advance and are categorized separately from

emergency C-sections that happen as a result of

complications during labor. Additionally, Kenyatta

National Hospital, as the leading referral hospital,

receives a very high volume of patients who are in

the second stage of labor. Although KNH

administrative policies require that a partograph be

used during all labors without exception, staff

shortages make prioritization necessary when

deciding to begin or continue a partograph for a

patient. Since KNH is a referral hospital, many

patients arrive late in labor in poor condition, and

completing paperwork or a partograph is not the

highest priority of hospital staff. The result is blank

or retroactively completed partographs.

Nurses were also asked to identify patients and

labor types that benefit the most from being

monitored with a partograph. Nurses were allowed

to circle more than one labor type out of SVD, CS,

IUFD, Referral, and ‘Other’. Twenty-three out of 26

nurses said that spontaneous vaginal deliveries

(SVD), which are often categorized as ‘normal’

labors, benefit the most from correct partograph use.

Eleven out of 26 nurses circled CS, emergency

cesarean sections, as benefiting the most from

partograph use, and 7 out of 26 nurses circled

‘Referral’.

The survey asked several PartoPen-specific

questions, including whether the nurses had

observed any changes in the labor ward because of

the PartoPen. This question was included in the

survey to follow up on qualitative observations and

discussions at the end of the 2012 studies that

suggested labor ward nurses were feeling an

increased sense of pride in their job because of the

interest of senior hospital staff, and reliance on labor

ward nurses to explain the project and demonstrate

its functionality. Additionally, only labor ward

nurses were given PartoPens, and this sense of

privilege was mentioned several times by nurses as

rewarding. Twenty-four of the 26 nurses said ‘yes’,

there had been changes in the labor ward because of

the PartoPen. The majority of the changes nurses

described related to the reminder functionality of the

PartoPen. Nurses frequently noted the reminders

being effective for providing more timely care and

making patient care more efficient. Better decisions

and easier chart interpretation were also noted as

significant changes resulting from PartoPen use in

the labor ward.

4.3 Secondary Data Analysis

Based on the data from the 2013 surveys, the data

from the 2012 maternity ward study was re-

examined, as follows: First, only the SVD

partographs were included, as the majority of nurses

indicated that SVD patients benefit most from

partograph use. In addition, partograph sections that

nurses deemed most important (i.e., (patient

information and fetal heart rate) were examined

individually.

The SVD partographs were analyzed in three

categories: August SVDs completed with the

PartoPens, August SVDs completed without the

PartopensatthePointofCare-EvaluatingDigitalPen-basedMaternalLaborMonitoringinKenya

PartoPens, and all of the SVDs from June. Using the

same grading and evaluation rubric, these

partographs were analyzed with respect to

completion. The results of this analysis are shown in

Figure 2. Frequency in this histogram is represented

as a percentage of the total number of partographs

present in the sample (37 August partographs

completed with the PartoPens, 206 August

partographs completed without the PartoPens, and

153 partographs completed in June). The histogram

illustrates that August partographs completed with

the PartoPens never received below 25%

completion, and this set had the highest percentage

of partographs in the 75-100% range.

The same set of SVD partographs was then

analyzed, looking specifically at the completion of

the ‘patient information’ and ‘fetal heart rate’

sections. While fetal heart rate completion did not

change significantly between the three groups, a

significant difference was observed in patient

information completion between August PartoPen

SVDs (M=.949, SD=.086) and June SVDs (M=.882,

SD=.152) using a paired t-test (t(188)=2.6178,

p=.0096). This difference may be attributable to

several factors, including the improved readability

and larger space for the patient information fields on

the PartoPen version of the partograph form.

Figure 2: A histogram of SVD-only partograph

completion scores (for August partographs completed with

a PartoPen for spontaneous vaginal deliveries only,

August partographs completed without a PartoPen for

SVDs only, and June partographs for SVDs only).

5 DISCUSSION

The results from the maternity ward studies do not

generally exhibit significant differences in

partograph completion rates between partographs

completed with the PartoPens and those not

completed with the PartoPens. In retrospect, this

result is not surprising. The PartoPen system was

designed to address training barriers that have been

cited as significant obstacles to correct partograph

use. However, the PartoPen system was deployed at

Kenyatta National Hospital, one of the leading

training and teaching facilities in Kenya. KNH has a

highly trained and knowledgeable staff who are less

likely to benefit from the training re-enforcement

aspects of PartoPen use. The other cited barriers to

partograph use, including staff shortages and lack of

supplies, are not directly addressed by the PartoPen

system, thus at KNH, any training reinforcement

benefit the PartoPen provided was overshadowed by

other barriers.

The positive results in the nursing student study

demonstrate that the PartoPen is beneficial for

partograph training for less-trained staff or for

students learning how to use the partograph. In a

controlled environment like a classroom where the

primary focus is on the task of completing a form

rather than delivering a baby, the PartoPen’s training

reinforcement and decision-support functionality are

fully utilized. In the chaotic and understaffed

environment of the labor ward at KNH, the primary

focus is on patients, not on paperwork, thus the

design objectives of the PartoPen system did not

align well with the primary focus of the KNH

nurses.

The next iteration of the PartoPen project will be

deployment at more rural and local levels of

maternity care, where nurse training, rather than

staff numbers and supplies, is the more problematic

issue. The primary contributions of the PartoPen

study in the maternity ward at KNH include nurses’

reflections on PartoPen usability, nurses’

perceptions of useful versus complete partographs,

and initial data on the durability and infrastructure

requirements of the PartoPen system, which can be

used in future deployments of the platform in other

labor wards.

From interviews with the nurses and researcher

observations, the reminders issued by the PartoPen

had the most impact on nurse behavior, although this

impact did not translate into increased partograph

completion, for the reasons described below. The

partograph used in the study was supplemented with

PartoPen Reminder ID boxes at the very bottom of

the form. Nurses were instructed to use these boxes

to record a memorable patient code, such as a

patient’s initials or the room number where the

patient was located. This patient code would be

displayed on the OLED display on the PartoPen

when the reminder for that patient sounded. The goal

HEALTHINF2014-InternationalConferenceonHealthInformatics

of the reminder system was to ensure timely patient

checkups by nurses who are busy, distracted, or

simply have forgotten to check on one of their many

patients. However, when the ratio of nurses to

patients is between 1:7 and 1:10, even if a nurse has

correctly recorded a patient code and receives the

patient’s reminders, she may be assisting with

another labor, checking on another patient, etc.

Many of the nurses reported receiving the reminders

but being unable to act on them because they were

already involved with a different patient.

Additionally, the design of the system was not as

helpful to nurses who had their hands busy, as the

patient code was displayed textually on the screen,

and nurses were often unable to stop what they were

doing to look at the pen and read the patient

reminder ID.

6 CONCLUSIONS

The initial objective of the PartoPen maternity ward

studies was to examine the impact of digital pen

technology on partograph completion. This

objective assumed that a primary barrier to

partograph completion was a lack of training and

knowledge on how to complete and interpret the

form. However, the highly skilled staff at KNH did

not lack in training or knowledge, but rather,

suffered from staff and resource shortages, which

the PartoPen was not designed to address. Despite

the disparity between the study goals and observed

study site realities, several important observations

were made that may contribute to future work in this

area.

First, every clinic or hospital has a unique set of

problems, personnel and procedures, which have to

be identified and addressed during both study design

and implementation. The PartoPen maternity ward

study design did not adequately account for the

myriad confounding factors present at KNH,

including under-staffing issues, different birth rates

between months compared, and the presence of

(different groups of) nursing students in the labor

ward during the intervention month, but not the

control month. Unlike the PartoPen nursing student

study design, the maternity ward study was not

designed such that only the affect of the PartoPens

on partograph completion could be measured. In one

analysis, study results were evaluated assuming an

experimental study where nurses were given the

intervention (the PartoPen) and the nursing students

present in the labor ward were the controls. This was

not the ideal study design, as the experimental and

control groups were not well matched in terms of

training, background, or experience. A more

appropriate study design for this environment would

be a paired comparison of individual nurses’

performance on partographs for similar labor types

with and without the PartoPen during comparably

busy shifts.

The study design that was used – a combination

of qualitative and quantitative data collection –

illustrates a disparity between the data from nurse

surveys and research observation and the data from

the partograph completion evaluation. When

surveyed, all of the nurses reported that they

considered partograph information to be important,

and that they relied upon this information.

Interviews with nurses also revealed that nurses

considered the partograph is an essential tool in the

labor ward. However, the low partograph

completion scores, regardless of the PartoPen

intervention, suggest that the partograph was often

under-utilized, filed out retroactively, or filled out

incompletely. This result is not indicative of a lack

of diligence or aptitude, rather a lack of adequate

staffing. Thus, partograph completion rates should

not be routinely equated with quality of care,

particularly at a short-staffed referral facility. It

would therefore be premature to promote the

partograph universally without conducting large-

scale studies on the direct association between

partograph use and maternal and child outcomes,

which account for environmental and social

circumstances unique to the study site.

Second, health informatics interventions,

especially in developing countries, are often

consumed by the technological aspects of the

project. We sometimes fail to recognize the benefit

of addressing immediate and simple issues, which

do not necessarily require technological

intervention. The qualitative feedback received by

nurses indicated that the cleaner PartoPen form with

larger boxes for information entry considerably

improved the usability and readability of the form.

The cleaner form was simple to produce within the

existing workflow and with existing equipment, and

could have been done independently of the PartoPen

project.

Finally, the PartoPens deployed at KNH were

successfully used and sustained for over nine-

months of continuous hospital use. This illustrates

the robustness of the system, as well as a willingness

among nurses to use the PartoPens on a daily basis.

The PartoPen maternity ward study helped identify

the environmental and physical challenges present in

the KNH labor ward, and illustrated both the

PartopensatthePointofCare-EvaluatingDigitalPen-basedMaternalLaborMonitoringinKenya

challenges and opportunities that arise when

deploying a digital pen software system in a

maternity ward setting. The results of this study are

encouraging for the continued and expanded use of

digital pen systems in healthcare, and stress the need

for more in-depth and well-designed studies in this

area.

ACKNOWLEDGEMENTS

We would like to express our sincere appreciation to

the staff and nurse-midwives at Kenyatta National

Hospital, without whose support and participation

this work would not have been possible. This

research was funded in part by the ATLAS Institute

at the University of Colorado Boulder, a National

Science Foundation Graduate Research Fellowship,

and a Bill and Melinda Gates Foundation Grand

Challenges Exploration Grant (OPP1061309).

REFERENCES

Anoto, Anoto Products. Available at: http://anoto.com

(Accessed September 26, 2013).

Estellat, C. et al., 2008. Data capture by digital pen in

clinical trials: A qualitative and quantitative study.

Contemporary Clinical Trials, 29(3), pp.314–323.

Khonje, M., 2012. Use and documentation of Partograph

in urban hospitals in Lilongwe - Malawi: Health

workers perspective.

Kwast, B. E., Lennox, C. E. & Farley, T. M. M. 1994.

World Health Organization partograph in management

of labor. Lancet, 343, pp.1399–1404.

Lavender, T. et al., 2011. Students’ experiences of using

the partograph in Kenyan labor wards. African Journal

of Midwifery and Womens Health, 5(3), pp.117–122.

Lavender, T., Hart, A. & Smyth, R.M.D., 2013. Effect of

partogram use on outcomes for women in spontaneous

labor at term. The Cochrane database of systematic

reviews.

Lind, L., Karlsson, D. & Fridlund, B., 2007. Digital pens

and pain diaries in palliative home health care:

professional caregivers' experiences. Medical

informatics and the Internet in medicine, 32(4),

pp.287–296.

Lind, L., Karlsson, D. & Fridlund, B., 2008. Patients' use

of digital pens for pain assessment in advanced

palliative home healthcare. International journal of

medical informatics, 77(2), pp.129–136.

Livescribe Inc., Livescribe Echo Smartpen Specs.

Available at: http://livescribe.com (Accessed

September 26, 2013).

Mathai, M., 2009. The partograph for the prevention of

obstructed labor. Clinical obstetrics and gynecology,

52(2), pp.256–269.

Opiah, M. M. et al., 2012. Knowledge and utilization of

the partograph among midwives in the Niger Delta

Region of Nigeria. African journal of reproductive

health, 16(1), pp.125–132.

Procuniar, M. & Murphy, S., 2008. Intuitive information

technology: enhancing clinician efficiency. AMIA

Annual Symposium proceedings / AMIA Symposium

pp.1247–1248.

Qureshi, Z. P., Sekadde-Kigondu, C. & Mutiso, S.M.,

2010. Rapid assessment of partograph utilisation in

selected maternity units in Kenya. East African

medical journal, 87(6), pp.235–241.

Rotich, E. et al., 2011. Evaluating partograph use at two

main referral hospitals in Kenya. African Journal of

Midwifery and Womens Health, 5(1), pp.21–24.

Underwood, H., Sterling, S.R. & Bennett, J., 2013a. The

PartoPen in Practice: Evaluating the Impact of Digital

Pen Technology on Maternal Health in Kenya.

ICTD'13 proceedings.

Underwood, H., Sterling, S.R. & Bennett, J., 2013b. The

PartoPen in Training and Clinical Use: Two

Preliminary Studies in Kenya. 6

International

Conference on Health Informatics/HealthINF2013

proceedings.

United Nations, 2010. The Millennium Development

Goals Report. pp.1–80.

World Health Organization, Maternal mortality. Available

at: http://www.who.int/mediacentre/factsheets/

fs348/en/ (Accessed September 26, 2013).

HEALTHINF2014-InternationalConferenceonHealthInformatics

APPENDIX A

PartopensatthePointofCare-EvaluatingDigitalPen-basedMaternalLaborMonitoringinKenya

APPENDIX B

KNH PartoPen Study – Participant Survey – June 2013

Please provide answers for the following questions:

Age: __________ Gender: __________

1) Did you participate in the PartoPen study in July and August 2012? YES NO

2) What is your level of experience using the PartoPens? (circle a choice below)

1 2 3 4 5

No experience Expert

3) How much do you rely on the partograph to make decisions about patient care? (circle a choice below)

1 2 3 4 5

I don’t use partograph I only use partograph

information at all information

4) Please number the following partograph sections in order of importance from 1 to 24, where 1 is the first

thing you look at on a partograph to make patient care decisions, and 24 is the partograph section that

you need the least to feel confident making patient care decisions. (Partograph sections were listed

below in original survey, but space constraints prevents us from listing them here.)

5a) Are there certain patients that do not need a partograph? YES NO

5b) If yes, what type of patients do not need a partograph?

6a) What kind of labors/births benefit the most from correct partograph use? (circle all that apply)

SVD CS IUFD Referral Other?

6b) For the answers you circled in 6a, please explain why these types of births benefit the most from correct

partograph use.

7a) Have there been any changes in the labor ward because of the PartoPen? YES NO

7b) What are they? How did they affect you?

8a) Have there been any problems with the PartoPen? YES NO

8b) If yes, what are they? How did they affect you?

9) Now that you have used the PartoPens for several months, what would you like the next steps to be in the

project? (Please list any other comments about the PartoPen project here).

HEALTHINF2014-InternationalConferenceonHealthInformatics

100