INTRODUCTION
According to the U.S. Food and Drug Administration (FDA), a medical device is any product or equipment that is used to diagnose, cure, treat, or prevent a disease or other health condition. A home healthcare medical device is any product or equipment used in the home environment by persons who are ill or disabled. Examples of some home healthcare devices are ventilators and nebulizers (to help breathing), wheelchairs, infusion pumps, blood glucose meters, and apnea monitors. (Home Healthcare Medical Devices, 2003.) This section of the paper will focus on disease-related home medical devices that are used to monitor a condition and which have a digital interface. Although there are several devices for monitoring diseases, few are intended for home use, and even fewer have a chip or computerized/digital interface. In actuality, the home digital market is dominated by diabetes- and asthma-related devices. Mention will be made of cancer and Alzheimer “devices”, but only briefly, as these do not really meet the definition of home digital devices.
Although these devices have come a long way, they all seem to face portability, data-analysis, and data-manipulation shortcomings. There is a long-standing battle in the computer interface world between the portability vs. limited screen size of portable devices, and the immobility vs. data-presentation flexibility of full-screen PC applications. Medical devices are no exception. However, few of the devices examined here present all the information a user would want on one visualization screen (e.g. graph or chart).
Useful sources for this section of the paper were the Food and Drug Administration (FDA) website, commercial device sites, and academic and industrial research centers. It is perhaps worth noting that the FDA Center for Devices and Radiological Health regulates medical devices to ensure that they meet reasonable safety and effectiveness standards, and FDA approval must be obtained for any device entering the U.S. market. The Administration monitors the quality of these products even after market-entry by inspecting manufacturing facilities, gathering information from the manufacturers, and accepting consumer complaints through the MedWatch system. The FDA also conducts user/usability tests for some devices, and sometimes issues specific interface guidelines. (FDA website, 2002.) All of these facts make the FDA a good starting point for examining home medical devices. Usability and interface issues can also be researched through various online support groups and expert websites, where readers candidly critique existing devices’ functional and design-related problems. Commercial sites are useful for understanding current device capabilities and features, but, being essentially advertising vehicles, they prove less helpful for judging a device’s (or its interface’s) strengths and weaknesses. Industrial and academic initiatives provide the bulk of research in this field.
DIABETES
Background
Diabetes is a chronic disease that affects 6% of the American population (McDonald et. al.). Simply described, diabetes impairs the pancreas' ability to produce or properly use insulin, which the body needs to control the amount of sugar (glucose) in the blood. This causes one’s blood sugar level to increase to dangerous levels. Left untreated, this condition can cause blindness, heart attack, poor circulation, loss of feet or legs to gangrene and amputation, kidney dysfunction, and even death. Interestingly, glucose-monitoring has become the fastest-growing segment of the diagnostic testing market (Waynant, 1998), and it seems to boast the greatest number and variety of home devices.
There are three types of diabetes: Type-1, Type-2, and gestational. In Type-1 diabetes, the pancreas produces little or no insulin. This tends to affect people under the age of 30, and requires insulin injections and constant monitoring of blood glucose levels. In Type-2 diabetes, the body produces either insufficient levels of insulin, or cannot process the insulin that is produced. This form typically affects adults over the age of 40, and does not require insulin injections for control. It does still require constant monitoring of the diabetic’s insulin level. Gestational diabetes affects some women during pregnancy; these women are at greater risk of Type-2 diabetes later in life. (McDonald et. al.)
There is no cure for this disease, but there are a number of things a patient can do to control it. The National Institutes of Health’s Diabetes Care and Complications Trial (DCCT), conducted from 1983-1993, demonstrated that frequent monitoring of blood glucose and insulin levels could prevent many of diabetes’ long-term complications, especially for Type-1 patients (Auxter, 1996). Some of the more common, current ways of detecting and monitoring diabetes are by checking the blood glucose level, the urine glucose (ketone) level, and the A1c hemoglobin level. Related tests check for kidney dysfunction, a condition sometimes caused by diabetes.
Commercial Devices
Blood Glucose monitoring
Meters Currently on the Market
Most current devices require that the diabetic collect a small blood sample, by pricking various body parts and smudging the blood onto a disposable “test strip”, to determine the level of glucose in the body. The test strips are coated with chemicals (glucose oxidase, dehydrogenase, or hexokinase) that combine with the glucose in the patient’s blood. The strips are inserted into a meter, which typically determines the patient’s glucose level by measuring the amount of electricity or light that passes through the sample. This is a prolific market: there are at least 25 different meters available commercially. (FDA website, 2002.)
These meters vary in many ways, but the attributes to look for are:
Testing speed
Where the blood can be collected
How much blood must be collected
Ability to store test results in memory
Whether high and low glucose values are displayed
Size of meter
Cost of meter
Cost of test strips; whether generic or brand-specific strips can be bought
One highly-touted marketing point is whether the project requires only 2 microliters of blood per test, as opposed to the formerly common 3-4 microliters. This would clearly be of interest to the patient, as less blood-drawing means less pain. The finger is the most common area from which to draw blood, but that becomes quite painful with repeated testing. Hence, many diabetics test their glucose levels less than twice a day, although 4-7 times per day is recommended (Waynant, 1998). There is also the problem that the user is more likely to become infected with other diseases, such as Hepatitus B, from reusing or improperly using the lancet device that is currently common for drawing blood (Polish et. al, 1992). Newer devices allow the user to prick less sensitive body parts, such as the forearm, upper arm, abdomen, thigh, calf, and fleshy part of the hand. The FDA, however, warns that blood in the fingertips shows glucose changes more quickly than blood in other parts of the body, and that further research is needed to determine if other locations provide adequately correct readings. (FDA website, 2002.)
Some products offer software programs that allow results to be downloaded onto a computer or PDA, save as many as the last 450 results, and/or calculate one’s 14/30-day or monthly average. Many link to record-keeping systems, as discussed later. Interestingly, few seem to offer the ability to save a personal high or low, or alert the patient when those thresholds have been passed. None seem to marry the glucose readings with other data points that might prove interesting to the diabetic.
Non-Invasive and Minimally Invasive Monitors
As recently as 2002, a new breed of glucose monitor – the “non-invasive” monitor - hit the commercial market. GlucoWatch by Cygnus is a wristwatch device that continuously, non-invasively, and automatically monitors the user’s glucose level without drawing blood. It can save up to 8,500 glucose readings, which can then be downloaded to a personal computer, emailed to one’s doctor, or printed onto graphs, charts, and statistics. It also allows the user to set high- and low-glucose alerts (issued as an alarm), and enter event markers such as MEAL, EXER (for exercise), etc. to explain why a given reading is high or low. The “Children with Diabetes” website claims that 95% of the Biographer readings are clinically acceptable. However, the manufacturer warns that GlucoWatch is intended to supplement, not replace, blood readings, and that GlucoWatch readings must be compared to finger-stick blood tests to ensure proper calibration. Interestingly, the manufacturer does not indicate what technology is used to take these measurements.
One competitor to the GlucoWatch (in that it is minimally invasive) is Medtronic’s MiniMed Continuous Glucose Monitoring System (CGMS). This device is actually quite invasive, however, as it requires that a physician insert a sensor (catheter) under the patient’s abdominal skin. The sensor reads glucose levels in the interstitial fluid, although it is unclear how that data is collected. This sensor remains under the skin for up to 72 hours, at which point the physician must remove it. Although there is a “monitor” interface into which the patient can enter related information (e.g. the time of meals, medication, and exercise), the patient cannot read the CGMS glucose readings proper – only the physician can. Furthermore, the patient must conduct finger-pricks four times a day, to keep the sensor calibrated. Since the average diabetic is advised to take four finger-pricks per day anyway, it is unclear what added benefit the CGMS provides. There is the added question of how “continuous” or “current” the CGMS readings truly are. Medtronic’s website states that:
“…blood glucose and interstitial fluid glucose levels are essentially equal when blood glucose is not changing rapidly. However, since glucose must enter the central blood supply, pass into peripheral circulation and cross a capillary boundary prior to entering the interstitial fluid, rapidly changing glucose levels create a lag between blood and interstitial fluid measurements. This means that when glucose concentrations are rising in the blood, they are lower in interstitial fluid throughout the lag. Overall, the differences between blood and interstitial fluid glucose levels do not affect the clinical utility of the Glucose Sensor data because they are minor (lags in glucose levels last usually less than 10 minutes) and because the data is analyzed in such a way that minor differences are negligible.” (Medtronic: CGMS FAQ.)
However, if continuous tracking of current glucose levels is the goal, then measuring interstitial fluid seems to be a less than perfect way to do that.
Biocontrol also develops and sells a non-invasive device, the Diasensor. This product was being sold in the European Union as of 2003, but had not yet been approved by the FDA for sale in the U.S. The user takes his readings by placing his arm in the printer-sized device for 2 minutes; the device reads the glucose level using infrared light. Results are evaluated by the doctor, however, not the patient; so, Diasensor is evidently intended for professional more than personal use.
Urine Glucose (Ketone) Monitoring
Ketone-measuring devices are not technically personal devices with digital interfaces, so they are somewhat beyond the scope of this paper. However, they are worth mentioning at least in passing, because they measure another common aspect of diabetes: ketones. Ketones are chemicals that the body produces when there is a shortage of insulin in the blood. This shortage causes the body to break down body fat for energy, and ketones are a by-product of burning fat. (FDA website, 2002.) High levels of ketones can lead to diabetic ketoacidosis (high levels of ketones and acid, leading to nausea and abdominal pain and possibly death) and coma. Although ketones are always present in urine, their concentration is usually so low as to be undetectable by routine tests. They are generally present enough to be detected in the urine of Type-1 diabetics, fasting individuals, and some pregnant women in the morning. Hence their use in diabetes testing. Blood ketone tests are generally preferred over urine tests, however. (Goldstein et. al., 1995.) The disadvantages of urine glucose tests include: 1) they will not indicate low glucose levels (below 180 mg/dl), since low levels of glucose are undetectable, 2) the readings change with the volume of urine, and 3) this measurement is more of an average than a precise-in-time reading. (FDA website, 2002.) Home testing kits do exist, however. They generally require dipping a dipstick into the urine to measure the ketone level, and sending the test to the physician for analysis. Blood (serum) ketone level can be assessed in the same manner. No true user interfaces have been developed as yet for this test. (FDA website, 2002.)
A1C Monitoring
Daily blood glucose testing indicates one’s blood sugar level at the time of testing. Since blood glucose levels vary widely during the day, however, a single measurement of blood glucose is a limited indicator of overall diabetes control. The hemoglobin A1c test, by comparison, measures the average blood sugar for the past several months. This test works as follows. In the human body, sugar in the bloodstream attaches itself to the hemoglobin (the part of the cell that carries oxygen) in red blood cells. The sugar remains there for the life of the blood cell (approximately 120 days). The higher the level of blood sugar, the more sugar is attached to red blood cells. The hemoglobin A1c test measures the amount of sugar attached to the hemoglobin, so that it shows the amount accumulated over the life of the blood cell. Hence, it is used to measure the average blood sugar level for the past 2-3 months. A 1% reduction in A1c level is associated with a significant 35% decrease in the likelihood of damage to a diabetic’s eyes, kidney, and nerves, and a 25% decrease in the likelihood of death (McDonald et. al.).
HbA1c testing is typically done in the doctor’s office, or by sending a blood sample to a commercial lab for analysis. The author could find only one meter (Metrika’s A1cNow Monitor) that allows the user to conduct this test at home. However, its value to the user is dubious. It is a disposable, one-use-only device, and it requires refrigeration. It also displays only the current reading in the display window, keeping no history of past readings and making no attempt to interpret the data. This seems like a very small step above sending results to the lab.
Assistive Devices
There are a few devices that cater to the visually impaired of manually disabled. These have magnifiers over the syringe-loading accessory, larger numbers on the insulin syringe, and positioning aids for the syringe and insulin bottle.
Record-Keeping
Although this section of the paper is intended to cover the medical devices themselves, it is worth venturing out into the world of diabetes record-keeping, as that is a fairly well-developed interface arena. There are numerous record-keeping systems for the diabetic to choose from, ranging from downloadable log sheets/spreadsheets, web-based logs, commercial Windows software, freeware Windows software, to the old-fashioned hand-logging system. (Mendosa, 2003.) The log sheets and spreadsheets require manual data-entry, as would be expected. Most of the Web logs and commercial Windows software programs accept downloads of diabetic readings from commercial meters. The most commonly accepted meters were of the Roche, Therasense, LifeScan, and Bayer lines; logically, many of these websites were created and maintained by the parent company (e.g. Roche) as part of their marketing program. The freeware programs tend to require manual data-entry of glucose levels, insulin injections, etc, but there are several that download the patient’s glucose readings from certain meters. Most of the latter accommodate the LifeScan One Touch and Roche Accu-Check meters, with some others covering Abbot’s MediSense meter, and various meters by Bayer and Therasense.
Regardless of their type or source, most of the software programs allow the user to enter (or upload) glucose readings, medication, insulin, exercise, meals, test results, and personal notes in one session. Some allow for the planning and recording of meals; these often hook up to USDA sites to provide nutritional information online. Only a few run on hand-held devices (PDA’s, Pocket PC’s), and these capture only blood glucose readings. (Pocket DiabetiCare, 2004; Accu-Chek Pocket Compass, 2004; Healthmate, 2004.)
Notably, most devices focus only on reproducing one’s measurements in log, chart, or graphical form; very few attempt to do analysis thereon. One system, WinglucoFacts boasts its ability to analyze the diabetic’s data using statistical “pattern recognition” tests. (WinglucoFacts, 2004.) The few other systems that do analysis merely state that they provide statistical and/or graphical analyses of the users’ blood glucose levels and/or insulin dosages, without providing more details. (Diabetes HomeCare, 2004; DIABASS, 2004; HealthDesk OnLine for Diabetes, 2004.)
User Interface Issues
Since 1984, the Food and Drug Administration (FDA) has received many reports of problems with blood glucose meters. A large number of these problems have been attributed to users' failure to maintain the meter properly, use of incorrect operating procedures, or failure to follow the instructions for meter use. (FDA website, 2002.) This suggests that many of the meters’ problems do not involve user interface issues at all, except to ensure that an adequate Help and Tutorial are provided with the meter. This assertion is somewhat contradicted by other FDA studies and by online communities’ device assessments, however.
The FDA suggests that the character size on glucose monitor displays be 5mm, so that diabetics with uncorrected eyesight problems can read the display’s values. (FDA website, 2002.) It is difficult to tell how many meters live up to this expectation without purchasing and examining each one individually, however. The FDA also conducted a human factors study that revealed that blood sampling, which depends on test strips and the user’s ability to draw and place blood thereon, has the greatest potential for error. (FDA website, 2002.) This problem is echoed in one online diabetic community’s assessment of current monitors. One community member observed that individually wrapped strips are so hard to open, that the device often turns itself off before user can unwrap the strip. There were also complaints that the strips are so small, that they tend to get lost in all manner of places in the house. (Road Testing Blood Glucose Meters.) This was further complicated by the fact that wet or residue-laden fingers, room humidity, or air/light-exposure on the strips can cause faulty readings. (FDA website, 2002.) Taken together with the blood-drawing problems mentioned under “Current Devices”, this would seem to indicate that test strips and the blood-drawing technique in general fail the usability test. No amount of interface changes (e.g. displaying test results) can overcome the strips’ poor functionality. Other methods should be refined and marketed.
Web communities with users’ critiques of glucose monitors offer insight into these devices’ user interface issues. Looking at one such community website, users clearly hold that “smaller and faster” is good. Apparently many meters take 2 minutes to do their calculations, and this was deemed too long. There was also the obvious preference to draw less blood, and to not need to wipe the testing strip after each use. Almost all users were pleased to have the option of pricking some body part other than the finger (despite FDA reservations). Providing a countdown while the meter takes a reading was suggested – the kind of audio-visual feedback that interface-design theories encourage. And a few users noted that they would like the meters to measure more than just discrete glucose levels – for example, the meters should also measure ketones, or show average glucose readings. (Road Testing Blood Glucose Meters.)
Current Research
Two MIT students have done research into marrying up glucose test readings with photographs of diabetics’ daily lives, to help the diabetics identify poor health habits. Their web-based tool consolidates these images and tests, color-codes them (warm colors indicate high, blues low, and yellow normal blood sugar levels), and allows the user to pick a day/hour to see the information visualized. (Frost and Smith, 2001.) Although interesting in concept, this may prove too time-consuming for the average person to actually use.
Similar to the MiniMed CGMS meter, researchers at the University of Pittsburgh have developed a tiny glucose sensor for implantation just underneath the skin. The sensor is a half-inch-long plastic tube about the diameter of a piece of thread, and is filled with particles that fluoresce according to blood sugar levels. A monitor fitted with a photometer, which sits atop the skin and above the sensor, measures the light intensity of the fluorescence; this in turn reflects the glucose concentration. (Wong, 2000.) Like the CGMS, this is a very invasive and perhaps unappealing way to monitor diabetes.
Looking forward yet further, there are some very interesting “non-invasive” glucose-monitoring techniques that are being researched. Few of these appear to have made it from the drawing board to the commercial market, but they may have a role to play in the diabetic’s future. Some alternatives to drawing blood are to investigate interstitial fluid (the fluid that surrounds the individual cells in the body), ocular (eye) fluids, and sweat. (Medtronic’s MiniMed CGMS currently monitors interstitial fluid, actually.) Alternate data-gathering techniques include microdialysis, infrared spectroscopy, near infrared spectroscopy, fluid-extraction, and optical methods. The optical methods alone have several subcategories, most of which measure the wavelength, polarization, or intensity of light that is passed through the patient’s skin. The results of these methods still need to be validated, and the instruments calibrated, however, before they can leave the research laboratory and enter the marketplace. (FDA, 2003; Waynant, 1998.)
Directions for Future Research
One question is outstanding: do these meters actually improve the diabetics’ management of their disease? The American Diabetes Association points out:
Although a number of SMBG (self-monitoring of blood glucose) methods store test results and with a computer interface can provide sophisticated analyses of blood glucose data, it is not known whether use of these data management systems yields better glucose control than patient review of results recorded in a logbook. (Goldstein et. al., 1995.)
One could readily argue that the monitors do help, in that they automatically store the results of many months of tests, saving hours of manual data entry. However, none of the monitors appears to go beyond this very simple data-gathering exercise. A worthy goal would be to superimpose the various measurements (e.g. glucose readings, insulin dosages, exercise, ketones, A1c tests) and their averages onto one screen. Of course, the PDA devices face the challenge of how to project data onto small screens – as do all PDA-based devices. However, such multi-valued graphs could be plotted on a PC, but no one appears to have taken this initiative.
In a similar vein, surprisingly few of the meters (or their fellow record-keeping systems) perform analysis of the data collected; instead, they simply mirror the information in another format (bar charts, graphs, etc.). As a matter of fact, none of the portable devices appears to perform analysis on the data, although this may be due to the limitations of a small PDA/Pocket PC screen.
Current research into combining photography with record-keeping to illuminate poor patient habits is provocative (Frost and Smith, 2001), but user studies should be conducted to see whether patients would actually spend the requisite time to enter all these photos and data. At the least, such projects should consider creating a joint monitoring/record-keeping/photographing device that would perform all the functions at once, rather than requiring the user to carry so many individual pieces of equipment.
Of course, the ideal meter would do everything at once: measure the patient’s glucose levels, show the patient’s glucose history (juxtaposed against other relevant data, and supporting multi-data manipulation and analysis), and pump insulin into the patient’s bloodstream when a reading was dangerously low. (Insulin idea is from Wong, 2000.) And, of course, to have this all bundled up in a portable, fast, reliable interface. That is asking for quite a lot, however, and might give rise to lawsuits if the insulin-injection function ever failed. So, such a device is probably decades away.