CHI 97 Electronic Publications: Design Briefings
CHI 97 Prev CHI 97 Electronic Publications: Design Briefings Next

Designing a Graphical User Interface for Healthcare Workers in Rural India

Sally Grisedale
Apple Research Labs
Apple Computer, Inc.
1 Infinite Loop, MS 301-3J
Cupertino, CA 95014
408 974-0731
sallyg@taurus.apple.com

Mike Graves
Apple Research Labs
Apple Computer, Inc.
1 Infinite Loop, MS 301-3E
Cupertino, CA 95014
408 974-2756
mgraves@apple.com

Alexander Grünsteidl
Apple Research Labs
Apple Computer, Inc.
1 Infinite Loop, MS 301-3E
Cupertino, CA 95014
408 974-7438
agrun@taurus.apple.com

ABSTRACT

This paper describes the research and development of an interface for a mobile computing device to be used by Auxiliary Nurse Midwives (ANMs) in rural India. We describe the insights of the team from Apple Research Lab (ARL), who have had the privilege of working in a very different culture from the ones they are used to. We show how our observations of the healthcare workers performing their caring and administrative functions informed the design of the user interface. We illustrate how we developed the graphical language, navigational structure and data entry techniques. We provide a summary of the feedback we received from early field trials and some thoughts on the appropriateness of our approach to design in this environment.

KEYWORDS

Human interface design, mobile computing, pen input, soft keyboard, Newton MessagePad, rapid prototyping, India, Rajasthan, Ajmer, Auxiliary Nurse Midwife, healthcare, family welfare.

© Copyright ACM 1997

INTRODUCTION

This morning I am following Padma, the local nurse midwife, on her house calls in Narwar, a farming village in Rajasthan. It is a hot dry day, and the local traffic, consisting of a few stray cattle and bristly pigs, lie slumped in troughs of dust trying to keep cool.

We come to the gate of the Sharma household and are greeted by three women who welcome us. "Namaste" they say, and adjust their veils. The yard outside the house is the center of communal activity. It is formed from compacted earth and adorned with chalk drawings. Blankets are brought for us to sit on. Water is served in metal cups, drawn from a ceramic urn at the side of the house. The women of the household came to live at this house as brides of the Sharma men. Three generations of women are living in the house, each the mother in law of the other.

Today, Padma is performing a routine post natal examination on the newest member of the family, baby Rao, whom she delivered three months ago. Padma reaches into her bag to consult her diary, which she keeps close by her. She talks with the women about Rao, and lays him on a low wooden bed to take a closer look. He has black hair and dark eyes, and on his forehead a crescent moon has been drawn, to protect him from the spirits.

A group of cattle wander into the yard - it is their lunch time, and the women go and tend to them. Padma writes up her notes in a diary. The women return from their chore and sing a song for me. It is a lament about the fate of a young bride who comes to a new household and is ill treated by her new mother in law. She pleads with her husband, ". . . husband I love you more dearly than the jewelry around my neck, please protect me from your family". [2] Diary entry, 23rd March 1995.

THE INDIA HEALTHCARE PROJECT

Into this environment Apple Computer Inc., and the Government of India wanted to introduce Newton MessagePad technology. The goal was to develop and deploy a Newton-based data-capture and viewing prototype, which would provide contextual performance support and health education information to healthcare workers like Padma.

The Opportunities the Project Offered

For Apple Computer Inc., this was a huge opportunity to learn how technology could be designed for an emerging market. It provided a case study for learning about developing solutions for "low-tech" circumstances, to be used by people with minimal exposure to computer based tools, and to apply this understanding to designing products.

For the Government of India (GOI), the Secretary for Health wanted to improve the current administration by improving the accuracy of information collected about the nation's health through its ambitious provision of free basic healthcare for all. The particular program we worked on was called Health and Family Welfare (H&FW), and it focused on the collection of information about the rural population in India by the ANMs. Figure 1 shows the hierarchy of the rural healthcare administration from minister to ANM.



Figure 1: Healthcare administration hierarchy and number of people provided for.

OVERVIEW OF THE HEALTHCARE SYSTEM

The rural population of India is about 630 million people spread among 32 states [1]. The program is administered locally from districts within each state. Further administrative blocks and sectors exist within the district to ensure geographic availability of some basic services across the country. The boundaries and sites of health centers within the district of Ajmer are shown in Figure 2.

FORMING A TEAM

The team from ARL made their first field visit to India in 1994 to observe and collect information about the healthcare workers. The team grew from 2 to 6 people in two years. The team is international and has combined expertise in system architecture, user studies, human interface design, programming and management. We had additional help from experts in healthcare, videography, travel, administration and translation within India.

The site for our field visits was designated by the Secretary for Health and Family Welfare. He suggested we start a pilot effort in the state of Rajasthan, in a district called Ajmer. The town itself is a well developed trading center and has a famous Muslim Shrine. The area is surrounded by less well developed rural desert land. The local industry comprises marble mining, fabric design, ceramic production and farming. The local spoken language is predominantly Hindi; this is one of the more prevalent of the sixteen major languages spoken in India today, broken out into 325 regional dialects [4]. Figure 2 shows the location of Ajmer and its administrative divisions, in relation to Rajasthan and India.



Figure 2: Ajmer District, Rajasthan, India.

Our approach was to work intensely with a few healthcare workers before trying to build anything. From watching and listening to the ANMs and their colleagues at the village, sector and district level, we learned about the job of the ANM.

It took us nearly a year before we began to understand what was being asked of us, and how we could possibly be of help by introducing Apple technology into this setting.

THE ROLE OF THE ANM

There are about 350,000 ANMs, all women, working in India today. They are each assigned responsibility for between 3,500 and 9,000 people whom they visit by foot, bicycle or moped. We worked directly with 10 ANMs, mostly from the district of Ajmer.

Not all ANMs share the same work practices, but the duties are basically the same. Her work includes treating minor injuries and ailments, referring people to the local hospital, providing ante and post natal care, vaccinating people, malaria testing and motivating villagers to use contraceptives. Each ANM receives two years training in basic healthcare and hygiene, with refresher courses once they are working in the field.

The people she visits live in widely scattered communities which are fragmented along caste and communal lines. Each ANM covers 3 to 5 villages or hamlets. In the eyes of the community, the ANM is seen as a Government representative whose purpose is primarily to enlist couples of child bearing age onto the family welfare program. The program provides incentives and practical advice on childbirth, ante/pre-natal care, contraception and sterilization options. In households where people rely on having children, preferably males, to bring in income, her advice is not always welcome. The ANMs need to possess good communication skills and the ability to build up a level of trust with people. This is no easy task when often the people she is working with are a different gender, or from a community or caste not of her own.



Fig. 3: Padma with patient



Fig. 4: Padma with records

THE ACT OF RECORD KEEPING

The ANM keeps notes of her activities which she then compiles into a weekly and monthly report. The data she collects about the villages/hamlets include things like the number of local wells and dates on which they have been treated with bleach, the location of ponds and times that DDT is sprayed to keep away potential sources of malaria. She also keeps a list of weddings and gifts.

On a household basis, she records the number of people, the name of the head of the household and the house number. Curiously, the house numbers and visits are often annotated on the wall near the font door by the ANM or her helper the MPW (multi-purpose worker, usually a man). This numbering scheme is unique to each district, and the villagers don't appear to mind having these marks left on their houses.

For every eligible couple, the ANM records the method of contraception used, e.g. Pill, IUD, condom etc., and the name of any Government agency worker who may be assisting with motivating the couple to join an incentive scheme to become sterilized.

For an individual she will record any illnesses, vaccinations, medication given, referral, or operations performed. Records are not kept on a patient basis, rather the action she takes is recorded on the form appropriate to the event. For example, if a malaria slide is taken, it will be recorded in the malaria register.

She monitors all pregnancies, births, infant mortalities and abortions. There is paperwork for both ante and post natal care, and Mother and Child Welfare. She records the number of disposable delivery kits dispensed, the name of the local village birth attendant, the names of people who attend her talks on hygiene and nutrition, the quantity of vitamins she gives out to expectant women, and notes of difficult pregnancies and referrals to the local hospital.

If she attends a birth, she will record the date and the approximate weight and sex of the child. As the infant matures, a record will be kept of any illnesses, vaccinations, and ante-natal check ups. Again, the notes will be kept on separate forms, like the immunization card shown in Padma's hand in Figure 3.

While on her rounds the ANM may carry a "Link Diary" where she notes her house calls. There may be about 9 house calls in a day which covers about 45 people, but this figure varies upon the geographic distance between villages and hamlets. After a day of house calls, she transcribes her notes from the link diary into the bulky register at her office called the sub center (Figure 4).

The sub center also serves as a walk in medical post for the community. The staff keep a basic supply of pills and vitamins, together with simple tools for treating wounds. Although there is usually no telephone, and intermittent electrical supply, there is usually a bed made up for patients, and a box full of condoms outside the center for people to help themselves.

THE IMPORTANCE OF RECORD KEEPING

Something we noticed immediately was that the records she kept about her activities had little bearing on the events which actually took place, at least at the local level we were witnessing. Nor did there appear to be any incentive for the ANM to complete the records accurately, since she received no feedback for doing so, only a reprimand if the paperwork was not handed in on time. The activity of filling in paperwork was abstract with respect to the realities of treating camel bites and issuing contraceptives. Yet it seemed there was no end to the number and variety of records she kept. Was all this record keeping really any use?

"Of the approximate 1.6 million children born in Rajasthan each year, within 12 months of their birth, 0.15 million of these children are dead from disease, malnutrition and in the case of girls, often sheer neglect." [5 ]


Figure 5: 40 years of records

If a Government wants to allocate appropriate funding for healthcare, it needs to know where best to spend the money. With statistics like this for the state of Rajasthan, it wasn't surprising that Health and Family Welfare was a priority. However, it isn't as simple as just looking at healthcare statistics to determine how and where to allocate resources. As Dr. Singh, the deputy chief district medical officer in Ajmer, pointed out. "Health is co-dependent upon the environment and education, and there are some things you can't change over night".

It took a lot of self control on occasion, to focus on what we had been asked to do. We centered our thoughts on what we had seen, and worked from there.

Learning from our Observations

Many of the workers we met were dexterous with multiple form filling and rapid reproduction of figures by hand. The system in many respects does work. However, ANMs don't always complete the forms correctly. This can happen for a number of reasons: they don't have time to complete the entries, don't understand how to fill them in correctly, or the villagers have not been entirely accurate in their descriptions. The onerous quality of the paperwork, duplication of entries, lack of any data verification capabilities, and overwhelming quantity of work contribute to inaccurate and missing data.

One of the reasons there are so many forms and duplication issues is because over the last 40 years, each time there is a change in policy or a new program or scheme implemented, a new set of forms and administrators get set up. The old forms continue to be used, new ones arrive and often without much explanation of how to fill them in. There was clearly an element of organized bureaucratic wheel spinning at every level of the administration. It did make us realize however, that re-designing a new set of paper forms was not going to solve the problem, and neither was it our charter to suggest changing the healthcare system.

Overview of Design Goals

From our observations of the ANMs and our discussions with the Government we came up with a charter to try and support both the ANM and the Government by providing a support tool to the ANM which would enable her to: To meet our charter we needed to integrate the current abstract world of paperwork with the concrete world of camel bites and condoms, and make one reflect the other. We wrote down what it would take to make a support tool which could help the ANM achieve these things. Work on each point occurred at different times, by different people, with differing levels of completion. What follows is a discussion of how we evolved the interface elements from our observations into early paper designs to working prototypes and informal testing in the field.

Transferring from paper to an electronic format

This was not as easy as it sounded. First we needed to collect the existing forms, have them translated into English and plot who filled in what, where the information was sent, and how the data gathering process influenced the work flow of the healthcare workers up to the district level. This took over a year to figure out, and there are still loopholes we can't explain. Getting real data and real forms was important, because concepts for interface designs were not well received by the very practical and dedicated members of the healthcare community. In fact some of our expert helpers in India were shocked that we were prepared to stand and look naive in front of healthcare experts in the interest of learning from them what they thought of our early paper prototypes.

Working with paper meant we could work quickly and get through a rapid number of ideas. The first designs were primarily an exercise in explaining what we understood of the report structure to ourselves. On index cards, we would write the name of the various programs, e.g. immunization, and lay them on the floor. Then we would push the cards around into different arrangements and mimic possible scenarios the ANM might go through, when dealing with a patient.

We then transferred these task sequences onto paper strips, which we folded. The folds could be opened to reveal more detailed information about each of the ANM's activities and their relationship to the various programs she was supposed to report on paper.

During field visits we used picture cards (Figure 6) with drawings of the typical daily activities of an ANM, taken from earlier video observations. The cards were useful props which let us start a dialogue with the ANMs despite the language barrier. We were able to determine the task flow of the ANMs' work in relation to their duties, by asking them to arrange them into the order which made most sense.

The advantage of working this way was that we were able to discuss navigation ideas, like the folds, with the ANMs in the medium they were most used to dealing with, before introducing it again on an unfamiliar medium like the Newton MessagePad.



Figure 6: ANM and Picture Cards

One of the most difficult things for us to grasp about the current reporting system was that all the records were kept on a household basis. The ANM kept track of a group of people in household. No individual records were kept. Households were referred to by the name of the elder male or head of household. Quite likely his name would be the same as that of his neighbors. In one of our early prototypes, using Hypercard, we tried a design which showed the relationship of family members to the head of the household (Figure 7).



Fig. 7: Hypercard mockup of a household overview

In this example there are three families: Canga, Ram Karn and Ram Ratan. The head of household is Ram Karan. The family whose record is about to be opened is Ram Ratan and his wife Mohini. Their children, Bholu the male and Meera the female, are listed separately according to sex.

This arrangement of people made relational sense, but was far too hard for us as interaction designers to continue working with as a model for both representing people and activities. Something had to give.

A leap of design faith was made. We departed from the existing system of form filling by household to one which focused on individuals living in the household. People would be identified by their family name and a standardized house numbering scheme already existing.

We assumed that the software could take care of filtering field data about an individual from their particular record into and cross referencing it with the various programs. This approach would instantly cut down the time spent entering data, avoid duplication, and reduce the margin of error within the data. What we needed now was a graphical language which we could use to make this approach make sense to an ANM's way of working.

Designing the icons

We were uncertain about introducing both new and established active graphical elements (like scroll bars and buttons) in the interface. We didn't know what symbolic representation would mean to an ANM. We tried to be sensitive to the cultural conventions of the ANM and her environment, but since we didn't know what these were, we spent a lot of time worrying about whether we were doing the right thing. For the graphical representation, we took our inspiration from printed literature already used by her in the field. In many of the sub centers we found brightly colored training material the ANM would take with her on her field visits to teach with (Figure 8). The magazines were really compelling from both a cultural and graphical standpoint. There were no naked bodies portrayed, and a diagram of a pregnant woman would be depicted wearing clothes, but with the shape of the fetus drawn over the top.


Figure 8: ANM's magazines

During a visit to the Christian Medical College in Ludhiana, Punjab we discovered some drawings made by ANMs for illiterate birth attendants (Figure 9). This find came after we had made our first drawings. When we saw that the icons were of people, we started to think that we were taking an appropriate route with our own designs.

From these observations we developed a set of icons to represent people of different genders and ages which would correspond to the report structure categories (Figure 10). We also developed icons for each of the programs the ANMs administered. When we tried the icons out on the ANMs we discovered that the icon for malaria, which we drew as a mosquito, didn't mean anything until we explained the connection. The ANMs suggested a blood drop would be more obvious, because it would reflect the action they take when taking a blood sample for diagnosis.



Figure 9: Icons drawn by ANMs in Punjab.



Figure 10: Representing people by gender and age.

The set of icons was designed to be extensible. This meant that when the ANM took an action, this action would be reflected by a subtle change in the graphic. For example here are some changes in status of an eligible couple. Notice that when the wife becomes pregnant a circle appears on her belly, above her sari.



Figure 11: Modal icons reflect state of patient

Developing this extensible graphical language offered the following advantages

Providing a navigational structure

The icons were also the primary means of navigation through what is a hierarchical database of patient records and programs.

The database is organized and represented in five levels, which lead the ANM from the village, to household, to family, to individual levels and then into the various programs. An Example of a typical task flow through the interface by the ANM is shown in Figure 12.



Figure 12: ANM Task flow through MessagePad screens.
When an individual's icon is selected, an extendible list of medical events for that person will be displayed. The list of programs is iconic. A program can be added to the list of medical events to record the progress of, for example a pregnancy, a disease or an immunization scheme.

Active programs have an influence on the representation of the individual's icon. The iconic representations have been designed to give qualitative feedback on the medical status of patients. Some of these variations to the basic female icon are shown in Figure 13.

The small graphical changes provide the ANM with an instant overview of patient status which she can see at the household level. This not only saves her time looking for individual patient records, but it indicates at a glance which individual needs attention.



Figure 13: Status of patient reflected in Icon

We did not expect the ANMs to learn the meaning of this graphical language overnight, but we felt that even without knowing precisely what these graphical augmentations meant, it was fairly clear that some note-worthy event had taken place.

Entering Information

The paper-based records system currently used by the ANMs is organized by healthcare programs. As a consequence, the ANM must enter data redundantly, repeating information from one program's records to the others'. Additionally, she must compile by hand data from the various programs' records once a month into a report to her supervisor. Hence our overall design goals to reduce time spent by the ANM in duplicate record-entry and in preparation and transmission of reports. Electronic data entry would eliminate the need for duplicate entries and would enable automatic report generation.

As much as possible, we tried to minimize entry of text data by hand. Pick lists on popup menus provide a standard Newton interface element for entering recurrent values, such as the village names, the names of diseases, etc. Still, hand entry of text cannot be entirely eliminated. Persons' names, new village names, and ANMs' notes for example are still entered by hand in our prototype.

The current Newton operating system does not support Hindi handwriting recognition, and acceptable hardware keyboards for the Newton in Hindi are not available. An on screen software keyboard, however, offers considerable flexibility for design. Unlike English and other languages with Roman script, Hindi contains a relatively large number of characters, without case distinctions, and which, in written form, often alter their appearance in accordance with their surrounding characters. This meant we could not fit all the letters onto one soft keyboard, but had to use two. However, the shift convention does not apply in Hindi so we had to find a compromise. Instead, we followed ISCII [3] key layout standards, but provided visibility to the normally shifted (and hence invisible) set of keys via a grayed-out image (Figure 14). Tapping the Newton pen on the grayed-out image brings the second set of keys to the front and makes them active, graying-out the first set. In addition, the keyboard software module includes instructions for detecting and properly rendering character combinations that need to be rendered other than simply as typed.



Figure 14: Hindi soft keyboard

A main problem was the development of a Hindi font for this type of screen size and resolution. A typical Asian font is about a third taller than a Roman font (Figure 15). Most of these fonts have many rounded features and accentuation, which are difficult to represent in a small font at a low resolution. A special font was developed to limit the height of the characters to be deployed in buttons, fields labels, etc.



Figure 15: Devanagari, Hespyndi 15 points

The healthcare data of the population is acquired in one of the seven different Indian scripts. On higher levels in the Indian healthcare system this data is transcribed into English, the official language of the national healthcare system. Our prototype needs to support any of the main languages but we noticed during our field studies that these even may differ greatly from the local dialects as spoken by the ANMs. To overcome this problem we have implemented a tool to select not only between English and the regional language but to customize the interface elements from default choices to variations entered with the help of local supervisors.

LEARNING BY TESTING

The project team has conducted two field studies in which approximately 10 ANMs have learned about, used, and offered feedback on our prototype designs. Our preliminary findings include the following:



Figure 16: Padma entering data on the MessagePad


By careful observation, evaluation, design and testing, we have built a working prototype of a hand held data capture device, which converts the existing paper based reporting structure into a more useful and accurate data capture and information retrieval model. The tedious task of data compilation has been automated to the widest possible extent, providing an instant overview of the latest health care provision status of the area surveyed by the ANM.

QUESTIONS AND DIRECTIONS

A central question for our research into supporting the work of the ANMs has been whether or not a hand held computing device, such as the Newton MessagePad, is appropriate. We cannot yet offer a definitive answer to the question. Optimistically, the device has been well-received by the ANMs. Feedback from the field has been encouraging regarding the acceptance of the hardware itself, data entry methods, records navigation, and data viewing. Essential questions remain regarding how well use of such a device can be integrated by the ANMs into their daily routines. Our prototypes have not yet been tested over extended use, with ANMs relying on them for their day-to-day duties. Such testing will reveal not only integration issues, but also pragmatic maintenance and support issues.

CONCLUSION

We hope that this methodology will enable other developers to create formats to enable the introduction of information technologies in high volume data capture environments for people with little or no exposure to electronic media. By working in these settings and with these populations, we can learn how to make technology which will empower people at large. We learned the importance of suspending judgment and disbelief.

REFERENCES

1. Government of India. Bulletin on Rural Healthcare Statistics in India. Government of India, New Delhi, 1995.

2. Grisedale, Sally (writer and producer), Padma's Story: A day in the life of a Healthcare Worker in Rajasthan, India (video). Apple Computer, 1995.

3. Indian Script Code for Information Exchange ­p; ISCII. Bureau of Indian Standards, New Delhi, 1991.

4. Silveira, D. M., DM Silveira's India Book 1994-95. Classic Publishers Pvt. Ltd., Goa, India, 1995.

5. UNICEF. Rajasthan: An analysis of the Situation of Children and Women. UNICEF, 1991.
CHI 97 Prev CHI 97 Electronic Publications: Design Briefings Next

CHI 97 Electronic Publications: Design Briefings