evolving approaches to intensive insulin therapy in type 1 diabetes: multiple daily injections,...

Reviews in Endocrine & Metabolic Disorders 2003;4:325–334C© 2003 Kluwer Academic Publishers. Manufactured in The Netherlands.

Evolving Approaches to Intensive Insulin Therapyin Type 1 Diabetes: Multiple Daily Injections, Insulin Pumpsand New Methods of Monitoring

Elizabeth Stephens and Matthew RiddleOregon Health & Science University, 3181 SW Sam Jackson Pk Rd,OP05, Portland, OR 97

Key Words. intensive insulin therapy, multiple daily injections,basal-bolus therapy, continuous subcutaneous insulin infusion

Type 1 diabetes affects approximately 1 million indi-viduals in the United States, whose management hassignificantly changed over the last decades. Initially, fol-lowing the discovery of insulin in the early 1920’s, im-provements centered around the purification of insulinformulations. Later came the development of both short-and longer-acting insulin preparations and more recentlythe availability of the insulin analogues aspart, lisproand glargine, which have added to our ability to deliverphysiologic insulin replacement patterns. However, in-sulin is only one component of therapy. Another criticalelement has been the development of devices for self-monitoring blood glucose (SMBG). As these devices havebecome increasingly accurate and convenient, they haveallowed patients to monitor blood glucose more oftenand accurately, adding a critical component to diabetesmanagement.

Much of the impetus to develop better therapies forthe management of type 1 diabetes stems from an un-derstanding of the importance of blood glucose control.The association between glycemic control and microvas-cular complications was established with the publicationin 1993 of results from the Diabetes Control and Compli-cations Trial (DCCT), the landmark study which demon-strated the benefit of efforts to achieve excellent glucosecontrol to limit retinopathy and nephropathy [1]. Estab-lishment of this relationship has led to significant changesin management. In this review we will discuss the con-cepts behind the current management of type 1 diabetes,beginning with a definition of intensive insulin therapy,and the various ways by which it can be achieved. Wewill then discuss the elements involved in intensive ther-apy, including essential educational components, glucosemonitoring, and the newer monitoring devices that areavailable.

Defining Intensive Insulin Therapy

Before the publication of the DCCT, it was not acceptedthat aggressive insulin therapy to lower blood glucose lev-els would be of benefit to those with type 1 diabetes. At thattime, conventional insulin therapy consisted of a combi-nation of NPH and regular insulin given twice daily. Thiscombination has been termed “non-physiologic” insulintherapy [2], and requires strict scheduling of meals andsnacks to prevent hyperglycemia or hypoglycemia.

Publication of the DCCT clearly changed our thinking.This study showed that those with type 1 diabetes whowere managed intensively to attain an A1c approximately2% lower than those in the conventional group for an aver-age of 6.5 years, had a reduction in the risk of retinopathyby 76%, nephropathy by 56% and neuropathy by 69% [3].These findings documented that conventional therapy wasputting patients at risk for worsening complications andcould not be tolerated.

From the DCCT came the concept of intensive insulintherapy (IIT). This is a recently coined term that impliesaggressiveness in seeking desirable glycemic targets, andalso the sophisticated methods by which to reach thesegoals. Use of this term has been somewhat controver-sial because inclusion of the word insulin indicates onlyone component of the strategy [4]. However it is unde-niable that insulin plays a central role. IIT is also calledflexible insulin therapy or basal-bolus insulin therapy. Alldescribe the use of either multiple daily injections (MDI)of insulin or an insulin pump in an attempt to mimic nor-mal pancreatic function and consequently attain the bestmetabolic control possible [5]. In order to be effectivethis therapy requires a combination of a prandial doseor bolus before meals consisting of either a short-actinginsulin such as regular, or a more rapid-acting insulin,

Corresponding author: Elizabeth Stephens.E-mail: [email protected]

325

326 Stephens and Riddle

such as lispro or aspart, to limit glucose excursions af-ter eating. In addition, IIT requires a long-acting insulinsuch as ultralente or glargine, an intermediate-acting in-sulin, such as NPH or lente, or basal insulin deliveryfrom an insulin pump to manage requirements betweenmeals and overnight [4]. Both insulin pump therapy, alsocalled continuous subcutaneous insulin infusion (CSII),and MDI were used in the DCCT, and both were effectivein controlling blood glucose levels. However the abilityto reach goals was limited by the insulins available at thetime of that study and often resulted in significant hypo-glycemia and hyperglycemia. The possibility of reducing,but not eliminating, these swings in blood sugars has comewith the availability of insulin analogues which can morespecifically and separately fill the roles of basal and bolusinsulin.

While insulin is the centerpiece in intensive therapy, ithas little value if the patient does not learn how to use itor the other critical elements of IIT. These are describedin Table 1. They may seem obvious, but each needs atten-tion in order for the therapy to be effective. Patients mustbe motivated, because intensive therapy requires a greatdeal of work. Glucose monitoring is essential to guidedosage adjustments and assure safety from hypoglycemia.Another critical component that is often misunderstood isthe setting of individualized glycemic targets for both pre-meal and 2-hour postmeal values. For example, in manypatients it is appropriate to recommend premeal blood glu-cose goals of 100 mg/dl, whereas for an individual withhypoglycemia unawareness, a premeal goal of 150 mg/dlwould be more appropriate. We also teach individuals toanticipate that their blood glucose will rise approximately30–50 mg/dl in the 2 hours after eating if they have takenthe appropriate amount of insulin, so they have an un-derstanding of what to expect from postprandial bloodglucose values. To evaluate overall control, the establish-ment of A1c targets is also critical but they too need tobe individualized. For example, it would be reasonableto aim for an A1c of <7% in a healthy 25 year-old with

Table 1. Essential elements to intensive insulin therapy

Components required for IIT

Patient motivation and interest.Establishment of targeted blood glucose levels.Understanding of diet composition, specifically carbohydrate content

and the impact that certain foods will have on blood glucose whichcan ultimately be adjusted by the patient.

Frequent SMBG (generally 4–6 times per day).Use of algorithms to adjust food intake and supplement insulin.Ability to provide feedback through utilization of measures of

glycemic control (hemoglobin A1c).Regular contact and support from diabetes team including physician,

educator and dietitian.

type 1 diabetes, but the target might be higher in someonewith a history of hypoglycemia awareness and prior se-vere hypoglycemic events. Nutrition therapy, specificallycarbohydrate counting, which is the estimation of the car-bohydrate content of each meal, is useful for learning howto adjust insulin for meals. Patients should learn to readfood labels to evaluate serving sizes with the help of a di-etitian with experience in diabetes. Specific carbohydrateto insulin ratios can then be established for patients todetermine how much insulin is required. While this ap-proach is ideal, it is too complex for some patients. Asan alternative, bolus doses of insulin can be distributedevenly with each meal, or can be determined based ondiet history, with adjustments made based on blood glu-cose records. Supplementary insulin scales or correctionfactors are also needed to help patients adjust for higherblood glucose readings before or between meals. Theseare predetermined doses of rapid- or short-acting insulinthat are used when blood glucose values are above goal.A common and conservative supplement for those withtype 1 diabetes is 1unit of fast/rapid-acting insulin for ev-ery 50 mg/dl above their targeted blood glucose. This canthen be adjusted depending on blood glucose readings be-fore and after meals. Finally, putting together all thesecomponents requires a good support system including thediabetes team caring for the patient as well as the patients’family and friends. The complexity of this whole processreinforces the importance of having a diabetes team thatcan offer education in each of the various components,and help the patient and their family and friends adapt toIIT.

There are many benefits to the intensification of insulintherapy. First and foremost is the reduced risk of compli-cations that will accompany improved glycemic control[1]. In addition, the DCCT documented that the use of in-tensive therapy by those with residual insulin productionwas able preserve endogenous beta-cell function as mea-sured by C-peptide levels. This benefit was associated witha further reduction in the development of microvascularcomplications [3], as well as a 65% reduction in the fre-quency of severe hypoglycemia when compared to thoselacking endogenous insulin secretion [6]. Finally the useof basal and bolus therapy either through MDI or insulinpump therapy gives the user greater flexibility in their dailyschedule. Individuals using IIT are able to adjust to spe-cific foods and meals, as well as changes in activity, whichwas not previously possible with conventional regimens.This results in a significant improvement in terms of qual-ity of life that has been documented in prior trials [7,8].

While the benefits of IIT are important, as outlinedabove, there can be adverse effects. The most obviousand common side effect is hypoglycemia. Patients in theintensive arm of the DCCT had ∼2.8 times more frequent

Evolving Approaches to Intensive Insulin Therapy in Type 1 Diabetes 327

hypoglycemia than those in the conventional arm, andmore of these events occurred at night [9]. As averageblood glucose levels decline toward the normal range withthe intensification of therapy, symptoms of hypoglycemiaare initiated at lower glucose levels. Physiologically thisis explained by increased efficiency of uptake of glucoseby the brain, which is desirable to protect function, butmay also lead to impairment of catecholamine release inresponse to low blood glucose readings [10]. This, in com-bination with the defect in glucagon release seen as a partof type 1 diabetes, puts the patient at risk for a severeepisode of hypoglycemia and a progressive inability torecognize hypoglycemia. For this reason, treatment goalsincorporate margins of safety to reduce the risk of hypo-glycemia. In addition, patients and their friends/familiesshould learn how to manage hypoglycemia, including theuse of glucagon.

Weight gain is another common side effect of inten-sive insulin therapy. This is thought secondary to a reduc-tion in glycosuria as well as excessive eating in responseto hypoglycemia. The DCCT documented that those inthe intensive therapy group gained on average 10 pounds(4.5 kg) more than those in the conventional arm [1]. Pro-gression of retinopathy, fortunately usually transitory, hasalso been linked to a rapid correction of hyperglycemia,and may affect 5% of patients [2]. Those at greatest riskappear to already have proliferative retinopathy and anA1c >10% prior to intensification of therapy [11].

Overall, while there are concerns with the intensifica-tion of insulin therapy, in most individuals the benefitsof IIT in terms of reducing the long-term complicationsoutweigh the risks. In the following sections we will dis-cuss the methods for implementing IIT, namely MDI andinsulin pump therapy.

Multiple Daily Injections

Because of the complete lack of endogenous insulin pro-duction, those with type 1 diabetes need a combination ofboth long-acting and short-acting insulin. The long-actinginsulin, or basal insulin, functions to control hepatic glu-cose output between meals and overnight, whereas theshorter-acting or bolus insulin helps to control blood glu-cose excursions after meals. As mentioned above, con-ventional therapy previously consisted of combinationsof intermediate-acting NPH and short-acting regular in-sulin given twice daily, which was only partly effectivebecause both components work as bolus and basal in-sulins. NPH insulin has a prominent peak of action be-tween 4 and 10 hours after injection, which can help withlunchtime control after morning administration, but oftencauses nocturnal hypoglycemia when given before dinner.To improve morning glycemic control and reduce the risk

of nocturnal hypoglycemia, NPH is frequently moved tobedtime. This tactic has been shown to reduce both A1cand nocturnal hypoglycemia [12]. Ultralente has also beenwidely used to improve between-meal insulin needs, butit also has a significant peak of action and is highly vari-able in its profile from day to day [13]. As a result, hy-poglycemia with ultralente can be a major problem [14].Finally, a limitation of regular insulin is the need to inject30 minutes prior to eating in order for its delayed onsetto match the mealtime need, which is inconvenient andimpractical for most patients. Thus, efforts to improvethe success of traditional twice-daily NPH and regularby advancing to three or more injections have not reli-ably led to the achievement of stable, excellent glycemiccontrol.

The advent of rapid-acting insulin analogues has donemuch to improve mealtime insulin delivery. Lispro, thefirst rapid-acting insulin analog, became available in 1996followed by the release of aspart in 2001. Both insulinswere produced by alterations of the amino acid structure ofthe human insulin molecule, reducing the self-associationof insulin monomers and speeding absorption after in-jection. The rapid-acting analogues eliminate the need towait after an injection before eating a meal. Given theirmore rapid onset and shorter time to peak effect, lisproand aspart have been more effective at controlling post-prandial blood glucoses when compared to regular, andare less likely to cause hypoglycemia [15,16]. This effectis illustrated in Figure 1, which summarizes the results of

Fig. 1. Mean values for twenty-four hour profiles of glucose andinsulin after 4 weeks of treatment of 90 type 1 patients with humanNPH at bedtime plus regular human (solid line, filled circles) or aspartinsulin (broken line, open circles) with meals. Aspart treatment resultedin more physiologic mealtime insulin peaks, and reduced postprandialglucose increments. However, late evening hyperglycemia occurredwith aspart, associated with lower basal insulin levels. (Adapted withpermission from Home PD et al. Diabetes Care 1998;21:1904.)


a comparison between aspart and human regular insulin[16]. In the 90 patients with type 1 diabetes studied, theuse of aspart was associated with improved post-prandialblood glucose control when compared to regular insulinwith significantly fewer episodes of severe hypoglycemia.However late evening and nighttime glycemic control wasinferior with the use of aspart, likely reflecting the remain-ing limitations of NPH as a basal insulin.

The release of glargine in 2001 has further added to theeffectiveness of MDI. Glargine is produced by recombi-nant DNA technology that alters human insulin on boththe A and B chains. With this modification, it is solubleat a pH of 4, and when injected into human tissue with anaverage pH of 7.4 forms a microprecipitate that is slowlyabsorbed over 24 hours [17]. Due to this slow absorptiona nearly flat serum insulin profile is noted following in-jection, and experience is proving it to be a superior basalinsulin. When compared with NPH insulin, it has been as-sociated with a significant reduction in fasting blood sug-ars, while also reducing the frequency of hypoglycemiaoccurring during daytime and at night in some [18] but notall studies [19]. Other studies using glargine in type 2 dia-betes have suggested an association with less weight gain,which can be problematic in those with type 1 diabetesintensifying therapy [20]. Despite the benefit in reduc-ing hypoglycemia and frequently reducing fasting plasmaglucose readings, most studies of glargine have not beenshown to significantly reduce A1c in those with type 1diabetes when compared to NPH [21]. However, a recentcomparison of glargine and NPH administered four timesper day (with meals and bedtime), both used with a rapid-acting analogue with meals, documented that althougheach regimen was effective to attain good glycemic con-trol, glargine was better able to reduce mean daily bloodglucose levels and frequency of hypoglycemia as well aslower the A1c [22].

Dosing basal-bolus insulin tends to be more logicalwhen compared to more conventional regimens. Due therequirement that NPH serve as both basal and prandialinsulin, it comprised a larger component of the total dailydose of insulin, generally being approximately two-thirds,while the other one-third was a shorter acting insulin suchas regular insulin. However, requirements for insulin re-placement are now thought to be about 50% for overnightand between-meals and 50% for mealtime needs. This hasled to the use of basal-bolus regimens that consist of one-half of the total daily dose of insulin for basal and one-halffor bolus doses [23]. When changing from another basalinsulin to glargine, it is recommended that the dose bedecreased by 20% in order to avoid hypoglycemia aroundthe time of transition. This dose reduction ends up beingthe appropriate dose for the patient approximately one-third of the time [24]. When using ultralente as the basal

insulin, the doses can be split evenly between morningand evening (each consisting of 25% of total dose of in-sulin). NPH can also be used in this way, with 10% of totaldose being given in the morning and 40% given at bed-time, leading the midday meal to be more dependent onthe rapid acting-analog given at lunchtime. Other meth-ods of utilizing NPH have been advocated, including QIDadministration when given with humalog [25]. This stylehas been shown to be effective as a basal insulin, but istoo complicated for many patients.

When determining bolus insulin doses there are sev-eral options, depending on the preference and ability ofthe patient. Ideally a patient learns to estimate the carbo-hydrate content of meals, and match the dose of insulin tothe grams of carbohydrate consumed. While this approachprovides a great deal of flexibility in terms of meal sizesand frequency, as stated previously, it can be too complexfor some patients. As an alternative strategy, the one-halfof the total daily dose (generally either aspart or lispro)can be distributed evenly between meals or can be basedon the estimated percentage of carbohydrates consumeddaily with meals. Dosages determined by any of thesemethods can be further fine-tuned using results of bloodglucose readings and patient experience [23].

Making the transition from conventional regimens us-ing NPH or ultralente plus regular to regimens based onglargine requires further education and guidance for pa-tients. It is critical to make sure they understand that uti-lization of glargine and bolus insulin will require at least4 injections per day if they eat 3 meals. This is due to theneed for prandial insulin with each meal as well as the factthat the glargine cannot be mixed with any other insulin(due to its formulation at a lower pH) [17]. Another issue isthe similarity in appearance between glargine, aspart andlispro, which are all clear solutions. Because many pa-tients are accustomed to identifying long-acting insulinsby their cloudy appearance, this similarity has led to er-rors in dosing [23]. Finally, utilization of insulin analogsmay be ideal for many patients, but their use can be lim-ited by the increased cost. These differences in cost areillustrated in Table 2. When cost is a significant concern,combinations of regular as prandial insulin with ultralentewith breakfast and dinner or NPH at breakfast and bedtimecontinue to be alternatives for some patients.

Syringes versus Insulin Pens

Insulin can be administered several ways, depending onthe needs of the patient. The most traditional approach isto give insulin by syringe, which is efficient, allows formixing of insulin (if not using glargine) and is less ex-pensive. Newer syringes are available using a very small


Table 2. Costs associated with intensive therapy∗

Delivery system Components Cost

Syringes Box of 100 $23.99

Pens Reusable pen Novopen 3 = $30–32 (retail)Pen needles (box of 100) $26

Pumps Insulin pump Approx $5500Tubing/reservoirs Approx $150–200/month

InsulinBasal Cost per bottle (1000 µ/10 ml)

NPH $25.99Glargine (Lantus) $45.99

BolusRegular $25.99Lispro or Aspart 1 Bottle (1000 µ) = $52.99(Humalog or Novolog) For cartridges for pens = Box of 5

(1500 µ or 1.5 ml) = $108.99

∗Information taken from drugstore.com, walgreens.com and minimed.com (6/03).

(30-gauge), as well as a short needle (only 5/16 inch or8 mm), which are useful for thinner patients. Disadvan-tages of using syringes include the need to carry moresupplies, and to fill and handle syringes publicly, which isbothersome to many with diabetes.

These limitations of syringes have led many with type 1diabetes to switch to insulin pens, which can be more con-venient. Insulin pens can deliver insulin extremely accu-rately leading to fewer insulin dosing errors, especiallywhen using smaller amounts of insulin [26]. Some newerpens also offer large dials with which to determine dosageamount, a benefit for those with impaired vision or dexter-ity. Finally, there are new devices that contain both insulinpens and glucose meters, allowing people to test their glu-cose and give insulin using the same device. However,while they add convenience, insulin pens are more expen-sive than syringes plus vials, a limitation which has slowedadoption of their use, especially in the United States. Com-parative costs for insulin pens and insulin bottles are listedin Table 2.

Insulin Pump Therapy

Insulin pumps are another method of delivering insulinin an effort to achieve targeted A1c goals. This form oftherapy has been available since the late 1970’s, and hasattracted increasing interest and use, with over 200,000individuals with type 1 diabetes utilizing pump therapyworldwide [27]. Insulin pumps are battery-operated de-vices that function by delivering insulin at frequent inter-vals through a catheter inserted in the skin and changedby the wearer every 2–3 days. They are programmed todeliver a basal rate which can be adjusted to meet spe-cific needs, such as an increased rate in the early morning

to counteract the dawn phenomenon. At mealtimes, bolusdoses of insulin can be administered through the pump toprovide insulin to cover carbohydrate content in the food,as determined by the patient [5]. Insulin pumps generallyinfuse either lispro or aspart, but only aspart is approvedby the FDA for use in insulin pumps.

Currently five companies are producing insulin pumps.Since their development, insulin pumps have becomesmaller, easier to use and more durable. All have the abil-ity to deliver multiple basal rates at very low doses (as lowas 0.05 µ/hour for those who are very insulin sensitive),and most can provide boluses over extended periods oftime (so-called “extended” or “square” wave boluses) forhigher fat meals that raise blood sugars for a more extendedduration. Newer pumps also have extensive memory andcan be downloaded for information regarding bolus andbasal history, which can be integrated with monitor read-ings to look at decision-making by the patient and theresults that followed.

Pump therapy has important advantages. It has beenshown to significantly reduce mean blood glucose val-ues and A1c and when compared to MDI in a number ofstudies [28,29]. Some of this benefit may be due the elim-ination in the variability in absorption that exists with useof the longer-acting insulins, which previously had beenestimated to vary from 19–55% [30]. Perhaps as a resultof this more reliable insulin absorption, insulin require-ments are less in those wearing pumps with reductionsbetween 14–17% reported when compared to those usingMDI [28,29]. The ability to alter basal rates and increasethem for morning hyperglycemia is a significant benefitof insulin pump therapy, and is not an option with anyof the available injectable insulins, making the dawn phe-nomenon a prominent indication for pump use. CSII hasalso been shown to be an effective tool to reduce the fre-quency of hypoglycemia in some studies, in which patientspreviously treated with MDI have switched to CSII, withreductions as great as 84% in the first year of CSII and 81%the second year, at similar levels of glycemic control asmeasured by A1c [31]. Finally insulin pump therapy givespeople the greatest flexibility to administer insulin accord-ing to their preferred schedules, allowing more freedomin timing of meals, snacks and exercise [29].

There are some limitations to CSII. Studies prior to thepublication of the DCCT results in 1993 documented anincreased rate of diabetic ketoacidosis (DKA) with CSII,likely due to inadvertent interruption of the insulin in-fusion, obstruction of the catheter or malfunction of thepump [29]. More recent assessments have shown that withproper education, the frequency of DKA is similar be-tween CSII and MDI [32]. Site infections have also beenworrisome, with previous reports documenting an inci-dence rate of 7.3 to 11.3 events per 100 person years [33].


However this is data from 1995, and does not reflect moremodern infusion systems, which may be associated withfewer infections. Cost is another issue, and is shown indetail in Table 2. The high cost often makes this form oftherapy prohibitive, especially in those without insurance.Finally, the effect on microvascular outcomes has not beenassessed in those utilizing pump therapy, making justifica-tion of the cost difficult, especially when compared withuse of MDI. This is especially pertinent with the arrival ofglargine, which has made MDI more effective, so that theadvantages of CSII will need to be verified with furtherstudies.

To get the greatest benefit from insulin pumps, appro-priate candidates must be selected. These should be indi-viduals who are motivated to care for their diabetes, able todocument (generally by record-keeping) and explain howthey are making decisions around their insulin administra-tion, have time to be attentive to diabetes management andeducation and are comfortable with the technology asso-ciated with an insulin pump. Next, education must be pro-vided so that the patient can learn the nuances of treatmentspecific to pump therapy, including frequent monitoring(up to 6–10 times per day when intensifying therapy), car-bohydrate counting, and use of supplemental insulin. Thelast element is teaching the patient to operate the pump.Without all of this education, the potential of CSII will notbe realized, and therefore the costs of this form of therapywill not be rewarded with clinical advantages over MDI.

Glucose Monitoring

Glucose monitoring is a critical element to the attain-ment of glycemic goals with any insulin regimen. Self-monitoring of blood glucose (SMBG) became available inthe early 1970’s, with the development of meters that couldbe carried by patients. Since then, meters have becomemore accurate, compact and user-friendly, making SMBGsomewhat less burdensome. The frequency of monitoringrequired or recommended has never been clearly estab-lished, although most experts advocate at least 3–4 testsper day, focusing on premeal and bedtime values. Morerecently with the availability of rapid-acting insulin ana-logues which have improved postprandial control, enthu-siasm for testing postprandial levels has increased. Suchtesting can determine the adequacy of insulin dosing forthe preceding meal, allowing additional insulin to be takenif needed, and also guiding future decisions. However, ob-jective verification of the benefit of postprandial testing islimited, except in the case of managing diabetes duringpregnancy, when an advantage has been shown [34].

There are now at least 10 companies producing glucosemeters for patient use. In an effort to reduce sample sizesrequired, many have moved away from light reflectance

technology to electrochemical methods. This has enabledblood sample size to shrink dramatically to as little as0.3 microliters and the time until results are available toas brief as 5 seconds in some meters. Most meters nowreport values calibrated for plasma readings in order to besimilar to those reported by clinical laboratories. This isimportant to discuss with patients when they switch to anewer meter, since plasma reference meters read 10–15%higher than whole blood calibrated meters [35]. The datastorage capacity of meters has also increased, and manynewer meters can retain information on carbohydrate con-tent in addition to glucose readings, making written log-books less necessary. Recorded data can be downloadedto a computer to help the patient and his provider make ad-justments in therapy. Currently one meter company has ameter attached to a hand-held Palm device, providing greatcapacity and convenience [36]. There are also two separatepump companies that have meter readings into pump datato assist with determining integrated bolus needed. Whileall these improvements add greatly to monitoring, a sig-nificant limitation is the cost of the strips, which average$0.65 each.

Alternative Site Testing

While glucose meters have improved, pain associated withlancing fingertips has continued to be a barrier, impairingadherence to monitoring recommendations. This issue hasled to the development of alternative site testing whichcauses significantly less pain. Currently six meters havebeen approved by the U.S. Food and Drug Administration(FDA) for alternate site testing at the forearm, upper arm,abdomen, thigh and calf. Conveniently, they use very smallblood volumes, between 0.3 and 2.6 microliters requiredand can use blood from the fingertip as well.

However, there are concerns about the accuracy ofthis method. Some patients report discrepancies betweensymptoms of hypoglycemia verified by fingerstick moni-toring but with normal values reported at alternative sites[37]. Studies comparing the Freestyle, Soft-sense and OneTouch meters, showed that in the fasting state, glucoselevels at the forearm and fingertip were similar. Howeverwith rapid changes in glucose levels, changes at the fore-arm were delayed on average about 35 minutes in relationto the fingertip [38]. Bina et al. [39] explored this issueand found a lag in alternative site readings when glucoselevels were fluctuating. They noted that palm and finger-tip values were similar all time points, and rubbing of theskin prior to measurement at alternative sites provided nobenefits. Other studies have verified excellent correlationsbetween fingertip and palm glucose concentrations [40].These findings have led the FDA to advise against using al-ternative sites for monitoring if blood glucose values may


be changing rapidly or if hypoglycemia is a possibility[41].

Continuous Glucose Monitoring

The minute to minute variations in blood glucose val-ues in type 1 diabetes, and fear of hypoglycemia whenintensifying management, have spurred interest in de-veloping continuous monitoring systems. In 2000, theFDA approved the Continuous Glucose Monitoring Sys-tem (CGMS) manufactured by Medtronic/Minimed. Thesystem consists of a Holter-style monitor that is worn onthe belt and connected to a sensor that is placed temporar-ily under the skin, generating an electrical signal whichis proportional to the amount of glucose present in theinterstitial fluid [42]. The system is designed to monitorinterstitial glucose levels every 10 seconds with values av-eraged every 5 minutes, providing readings within a rangeof 40–400 mg/dl for 72 hours [42]. Fingerstick values areentered into the monitor at least 4 times per day in order toestablish a correlation between interstitial fluid and bloodglucose values. Currently readings from the system arenot available to the wearer. Information is downloadedto a computer, generally at a medical provider’s office,and glucose values are calculated automatically for each24 hour cycle. The information can then be reviewed andtherapeutic changes initiated as indicated.

Because of the frequency of sampling and the abilityto wear it continuously, the CGMS has been promotedas a tool to identify unrecognized hypoglycemia, espe-cially at night. Studies have demonstrated good accuracyof the CGMS during hypoglycemia [43] in addition toa high frequency of clinically undetected hypoglycemiautilizing the CGMS technology [44]. Other studies havedemonstrated the utility of CGMS to result in treatmentmodifications that lead to an improvement in hemoglobinA1c [45]. Newer software for downloading has made in-terpretation easier, organizing the data into premeal andpostmeal readings.

However, there is controversy about the reliability ofthis technology. One small study of 7 patients with type 1diabetes evaluated the accuracy of the CGMS system in pa-tients with type 1 diabetes whose average A1c was 6.6%,after questions were raised about the ability of the sensorto measure glucose values at or below the normal range[44]. They showed CGMS results to be lower than simul-taneously measured values done with a Beckman glucoseanalyzer 74% of the time. In addition, nighttime readingswith CGMS were lower by 38% in 6 of 7 patients, a dis-crepancy that was believed to be clinically significant. Inanother study, 11 patients with type 1 and 2 diabetes aswell as without diabetes, wore 2 sensors in free-living sit-uations (rather than laboratory conditions) to evaluate the

reproducibility of the data collected. While overall datapoints were highly correlated (r = 0.84), information pro-vided simultaneously by the two sensors was concordantfor only 65% of the evaluation periods, calling into ques-tion the accuracy of the device [42]. Technical problemswith obtaining complete data over 72 hours has also lim-ited the usability this device, with some studies showingonly 50–60% of readings to be usable after filtering anddownload [46,47]. Finally the inability of the wearer toaccess the information from the sensor limits its utility inacute situations, such as hypoglycemia.

A second sensing device, the GlucoWatch biographer,was approved by the FDA in 2001. This device consists of adisposable sensor which adheres to the skin and a reusablewatch that sits on top of it. A low electrical current is con-ducted through the skin, resulting in migration of glucosemolecules to the sensor surface (iontophoresis). As withthe CGMS, a small current proportional to the interstitialglucose concentration is generated. Calibration requiresonly a single fingerstick value at the end a 2-hour warmup period. The latest version (GlucoWatch G2) monitorsfor up to 13 hours, and provides readings every 10 minuteswhich are displayed to the wearer. The watch also can beprogrammed to alarm for readings that are above or belowa certain value determined by the wearer, or when therehas been a decline in blood glucose of over 35% since thelast reading to help determine whether hypoglycemia isimminent.

Advantages of this device include the frequent read-ings, which can then be used to track trends in blood glu-cose values, and the alarm features which can be helpfulfor those with frequent hypoglycemia or hypoglycemiaunawareness. Studies of the GlucoWatch have shown itto be safe and to provide adequate correlations with fin-gerstick blood glucose levels in adults. One multicenterstudy of 92 individuals with type 1 or 2 diabetes, stud-ied in controlled clinical environments, showed compara-ble accuracy to other blood glucose monitoring devices,as well as good correlation between the GlucoWatch andfingerstick measurements [48]. The impact of this deviceon outcomes has been less well evaluated. One study of40 children with type 1 diabetes and poor glucose con-trol (A1c > 8%), randomized to GlucoWatch use (4 timesper week for 3 months), showed that its use resulted in asignificant decline in A1c as well as greater detection ofhypoglycemia, especially at night [49].

Disadvantages of the GlucoWatch include discomfortwhen the current is applied as a part of iontophoresis, aswell as the development of a rash that can persist afterthe device is removed. The use of steroid cream in thearea is now commonly recommended with sensor removal.The device will also skip readings if there is interferencewith the connection between the surface of the skin and


the sensor, such as with sweating. Skipped readings arealso common outside of episodes of sweating, with themanufacturer estimating approximately 21% of readingsbeing skipped in regular usage [50]. The readings from thedevice are overall felt to be accurate, but concerns have ledthe manufacturer to recommend that a fingerstick be doneto back up any reading from the GlucoWatch that mightresult in a change in therapy, such as giving or withholdinginsulin. Finally, the device is expensive. The GlucoWatchitself costs $698.00 and each sensor $7.50.

These pioneering glucose sensing devices have signif-icant limitations, but over 25 companies are working todevelop better systems, using non-invasive, minimally-invasive and implantable methods, all intended to provideaccurate information more conveniently than the currentglucose meters. Ultimately the goal is the developmentof a closed-loop system in which glucose levels wouldbe determined accurately, and insulin administered au-tomatically in response to need. A very reliable systemis required because of the potential for erroneous bloodglucose readings to result in inappropriate insulin deliv-ery or omission leading to severe consequences. We alllook forward to such a system with great optimism, for itwould truly revolutionize management in those who areinsulin-dependent.

How far have we come—and what is next?People with type 1 diabetes are not always successful us-ing the methods reviewed above, so the evolution of inten-sive therapy is sure to continue. In coming years we willsee further improvements in insulin delivery, glucose sen-sors, and protection and restoration of beta cells. In timetype 1 diabetes may be increasingly preventable, but whenit does occur future forms of intensive therapy will be ableto restore glycemic patterns far closer to normal than ourcurrent methods. But it is worth looking back a decade ortwo to see the progress we have already made. At entry tothe DCCT, adults with type 1 diabetes had A1c values av-eraging about 9%. The stated aim of the intensive-therapypolicy in the DCCT was to “achieve blood glucose val-ues as close to the normal range as possible [1].” Themethods used, including MDI and CSII with human in-sulins, achieved (with a great deal of effort) A1c valuesaveraging just above 7% throughout the trial. The rate ofhypoglycemia requiring assistance was 62 per 100 patientyears. Less that 5% of those in the intensive treatmentgroup maintained levels consistently within the normalrange.

While A1c in the normal range remains out of reach formost patients today, there are signs we are doing better thanbefore. For example, two recent meta-analysis of trialscomparing CSII with MDI have reached similar conclu-sions: that on average pump therapy can reduce A1c values

Fig. 2. Intensive treatment for 3 months with glargine once-daily atbedtime (open bars) as basal insulin compared with human NPH fourtimes daily (crosshatched bars), while lispro insulin was taken withmeals. Young adults with type 1 diabetes previously managed with NPHfour times daily and lispro with each meal were studied, with 17 in eachgroup. Statistical significance (p < 0.04) vs. baseline (shown as solidbars) and vs. NPH treatment is denoted by asterisks. (Adapted withpermission from Rossetti P, et al. Diabetes Care 2003;26:1490–1494.)

by about 0.4% relative to treatment with multiple injec-tions [28,29]. Whether this difference will persist now thatthe insulin analogs are available is not clear, with compar-isons of intensive treatment with insulin analogs versuswith human insulins showing advantages of the analogs inglycemic control, rates of hypoglycemia., or both together.For instance, Bolli and colleagues compared MDI with hu-man NPH and lispro versus glargine and lispro [22]. Thesubjects in this small (51 patient) study were in good con-trol with NPH plus lispro at entry, with mean A1c closeto 7%. After three months’ treatment those treated withboth a rapid-acting and a long-acting insulin analog hadreduced A1c by 0.4% from this already excellent baselinewhile reducing the rate of hypoglycemia (Fig. 2) Thesefindings must be confirmed by other studies, but theyare consistent with our own experience that current in-tensive methods allow a great many of our adult type 1patients to achieve A1c values less than 7% without unac-ceptable hypoglycemia. It seems clear that the combina-tion of better insulins, better delivery devices, and betterglucose testing methods is having the effect we want: morepatients are successful than a decade ago.

The next step in this evolution may be closer than wehave thought. A leading barrier to success with IIT hasbeen that postprandial hyperglycemia has resisted full cor-rection by bolus insulin injections due to the risk of subse-quent hypoglycemia. Thus, as noted above, we have cometo consider an increase of 30 to 50 mg/dL after a meal tobe expected and acceptable, although clearly not normal.


Renewed study of the physiology of prandial glucose regu-lation is showing that physiologic insulin replacement maynot be sufficient, owing to abnormalities of other hormonalsignals in type 1 diabetes. Notably, the hormone amylinis normally cosecreted by the beta cell, and like insulin isabsent in type 1 diabetes. Replacement of this hormonealong with insulin may add another dimension to inten-sive therapy [51]. Early studies suggest adding injectedpramlintide to injected insulin or CSII can blunt glycemicexcursions, reduce A1c, and in addition improve satietyand control weight [52,53]. If this tactic allows bolus in-sulin dosage to be greatly reduced, risks of hypoglycemiamay be reduced as well.

As methods for intensive treatment of type 1 diabetescontinue to improve, the leading challenge may shift fromthe scientific and technical arena to the organization ofhealth care. Already a wide gap exists between the treat-ment of persons who have gained access to modern meth-ods of treatment, and those who have not. Providing thetools and education necessary to apply intensive therapyto all persons with type 1 diabetes who need it will be atask for the coming decade.

References

1. DCCT Treatment Group. The effect of intensive treatment of dia-betes on the development and progression of long-term complica-tions in insulin-dependent diabetes mellitus. New England Journalof Medicine 1993;329: 977–986.

2. DeWitt DE, Hirsch IB. Outpatient insulin therapy in type 1 and type2 diabetes mellitus: Scientific review. JAMA 2003;289:2254–2264.

3. The Writing Team for the DCCT/Epidemiology of Diabetes Inter-ventions and Complications Research Group. Effect of intensivetherapy on the microvascular complications of type 1 diabetes mel-litus. JAMA 2002;287:2563–2569.

4. Hirsch IB. Intensive treatment of type 1 diabetes. Medical Clinicsof North America 1998;82:689–719.

5. Stephens E, Ryan-Turek T. Intensive Insulin Therapy, EndocrineSecrets. Hanley & Belfus, Inc., Publishers, 2002.

6. The DCCT Research Group. Effect of intensive insulin therapy onresidual β-cell function in patients with type 1 diabetes in the DCCT:A randomized, controlled trial. Ann Intern Med 1998;128:517–523.

7. Chantelau E, Schiffers T, Schutze J, Hanser B. Effect of patient-selected intensive insulin therapy on quality of life. Patient Educa-tion and Counseling 1997;30:167–173.

8. DAFNE Study Group. Training in flexible, intensive insulin man-agement to enable dietary freedom in people with type 1 diabetes:Dose adjustment for normal eating (DAFNE) randomized controlledtrial. BMJ 2002;325:746–751.

9. The DCCT Research Group. Epidemiology of severe hypo-glycemia in the diabetes control and complication trial. Am J Med1991;90:450–459.

10. Boyle PJ. Hypoglycemia in The Medical Management of DiabetesMellitus. New York: Marcel Dekker, Inc., 2000:527–538.

11. Chantelau E, Kohner EM. Why some cases of retinopathy worsenwhen diabetic control improves. BMJ 1997;315:1105–1106.

12. Fanelli CG, Pampenelli S, Porcellati F, Rossetti P, Brunetti P, BolliGB. Administration of neutral protamine Hagedorn insulin at bed-time versus with dinner in type 1 diabetes mellitus to avoid nocturnal

hypoglycemia and improve control. Ann Intern Med 2002;136:504–514.

13. Scholtz HE, van Niekert, Meyer BH, Rosenkranz B. As assess-ment of the variability in the pharmacodynamics (glucose low-ering effect) of HOE901 compared to NPH and ultralente hu-man insulins using the euglycemic clamp technique. Diabetologica1998;47 (Supp 1):A235.

14. Tunbridge FK, Newens A, Home PD, Davie SN, Murphy M, BurrinJM, Alberti KG, Jensen I. A comparison of human ultralente- andlente-based twice-daily injection regimens. Diab Med 1989;6:496–501.

15. Anderson JH Jr, Brunell RL, Koivisto VA, Pfutzner A, TrautmannME, Vignati L, DiMarchi R. Reduction of postprandial hyper-glycemia and frequency of hypoglycemia in IDDM patients oninsulin-analog treatment. Multicenter Insulin Lispro Study Group.Diabetes 1997;46:265–270.

16. Home PD, Lindholm A, Hylleberg B, Round P, for the UK InsulinAspart Study Group. Improved glycemic control with insulin aspart:A multicenter randomized double-blind crossover trial in type 1diabetic patients. Diabetes Care 1998;21:1904–1909.

17. Levien TL, Baker DE, White JR Jr, Campbell RK. Insulin glargine:A new basal insulin. Ann Pharmacother 2002;36:1019–1027.

18. Ratner RE, Hirsch IB, Neifing JL, Garg SK, Mecca TE, Wilson CAfor the U.S. Study Group of Insulin Glargine in type 1 diabetes: Lesshypoglycemia with insulin glargine in intensive insulin therapy fortype 1 diabetes. Diabetes Care 2000;23:639–643.

19. Raskin P, Halle J-P, Klaff L, Donley D, Bergenstal R, Mecca T. A16-week comparison of the novel insulin analog insulin glargine(HOE901) and NPH insulin used with insulin lispro in patients withtype 1 diabetes. Diabetes Care 2000;23:1666–1671.

20. Rosenstock J, Schwartz SL, Clark CM, Park GD, Donely DW,Edwards MB. Basal insulin therapy in type 2 diabetes: 28-week comparison of insulin glargine (HOE901) and NPH insulin.Diabetes Care 2001;24:631–636.

21. Younis N, Soran H, Boen-Jones D. Insulin glargine: A new basalinsulin analogue. Q J Med 2002;95:757–761.

22. Rossetti P, Pampanelli S, Fanelli C, Porcellati F, Costa E, TorloneE, Scionti L, Bolli GB. Intensive replacement of basal insulin inpatients with type 1 diabetes given rapid-acting insulin analog atmealtimes. Diabetes Care 2003;26:1490–1496.

23. DeWitt DE, Dugdale DC. Using new insulin strategies in theoutpatient treatment of diabetes, clinical applications. JAMA2003;289:2265–2269.

24. Kelly JL, Trence DL, Hirsch IB. Rapid decrease in clinically signif-icant hypoglycemia with insulin glargine. Diabetes 2002;51:A123.

25. Lalli C, Ciofetta M, Del Sindaco P, Torlone E, Pampanelli S,Compagnucci P, Cartechini MG, Bartocci L, Brunetti P, Bolli GB.Long-term intensive treatment of type 1 diabetes with the short-acting insulin analogue lispro in variable combination with NPHinsulin at mealtimes. Diabetes Care 1999;22:468–477.

26. Lteif AN, Schwenk WF. Accuracy of pen injectors versus insulin sy-ringes in children with type 1 diabetes. Diabetes Care 1999;22:137–140.

27. Bode BW, Sabbah HT, Gross TM, Fredrickson LP, Davidson PC.Diabetes Management in the new millennium using insulin pumptherapy. Diabetes Metab Res Rev 2002;18(Supp 1): S14–S20.

28. Pickup JM, Mattock M, Kerry S. Glycemic control with continuoussubcutaneous insulin infusion compared with intensive insulin injec-tions in patients with type 1 diabetes: Meta-analysis of randomizedcontrolled trials. BMJ 2002;324:1705–1708.

29. Weissberg-Benchell J, Antisdel-Lomaglio J, Seshadri R. Insulinpump therapy: A meta-analysis. Diabetes Care 2003;26:1079–1087.

30. Scholtz HE, van Hiekerk N, Meyer BH, Rosenkranz B. An assess-ment of the variability in the pharmacodynamics (glucose lowering


effect of HOE901 compared to NPH and ultralente human in-sulins using the euglycemic clamp technique. Diabetologica 1999;42(Suppl 1): 882.

31. Bode BW, Steed RD, Davidson PC. Reduction in severe hypo-glycemia with long-term continuous subcutaneous insulin infusionin type 1 diabetes. Diabetes Care 1996;19:324–327.

32. Pickup J, Keen H. Continuous subcutaneous insulin infusion at 25years. Diabetes Care 2002;25: 593–598.

33. Diabetes Control and Complications Trial Research Group. Im-plementation of treatment protocols in the Diabetes Control andComplications Trial. Diabetes Care 1995;18:361–376.

34. de Veciana M, Major CA, Morgan MA, Asrat T, Toohey JS, LienJM, Evans AT. Postprandial versus preprandial monitoring in womenwith gestational diabetes mellitus requiring insulin therapy. NEJM1995;333:1237–1241.

35. American Diabetes Association. Tests of Glycemia in Diabetes.Diabetes Care 2003;26:S106–S108.

36. US Food and Drug Administration. Therasense, Inc. FreeStyleTrackerTM Diabetes Management System. www.fda.gov 2002.

37. Jungheim K, Kischinsky T. Risky delay of hypoglycemia detectionin glucose monitoring at the arm. Diabetes Care 2001;24:1303–1304.

38. Jungheim K, Koschinsky T. Glucose monitoring in the arm. Riskydelays of hypoglycemia and hyperglycemia detection. DiabetesCare 2002;25:956–960.

39. Bina DM, Anderson RL, Johnson ML, Bergenstal RM, Kendall DM.Clinical impact of prandial state, exercise, and site preparation onthe equivalence of alternative-site blood glucose testing. DiabetesCare 2003;26:981–985.

40. Peled N. Letter to the Editor. Timely detection of hypoglycemia bymonitoring blood glucose levels of samples obtained from the palmof the hand. Diabetes Tech Therap 2002;4:83–85.

41. U.S. Food and Drug Administration, Center for Devices andRadiologic Help. Glucose meters and Diabetes Management:http://www.fda.gov/diabetes/glucose.html#12, updated 5/1/02.

42. Metzger M, Leibowitz G, Wainstein J, Glaser B, Raz I. Re-producibility of glucose measurements using the glucose sensor.Diabetes Care 2002;25:1185–1191.

43. Monsod TP, Flanagan DE, Rife F, Sainz R, Caprio S, Sherwin RS,Tamborlane WV. Do sensor glucose levels accurately predict plasma

glucose concentrations dduring hypoglycemia and hyperinsuline-mia? Diabetes Care 2002;25:889–893.

44. McGowan K, Thomas W, Moran A. Spurious reporting of nocturnalhypoglycemia by CGMS in patients with tightly controlled type 1diabetes. Diabetes Care 2002;25:1499–1503.

45. Bode BW, Gross TM, Thornton KR, Mastrototaro JJ. Continuousglucose monitoring used to adjust diabetes therapy improves glyco-sylated hemoglobin: A pilot study. Diabetes Research and ClinicalPractice 1999;46:183–190.

46. Chase HP, Kim LM, Owen SL, MacKenzie TA, Klingensmith GJ,Murfeldt R, Garg SK. Continuous subcutaneous glucose moni-toring in children with type1 diabetes. Pediatrics 2001;107:222–226.

47. Boullu-Sanchis S, Jeandidier N, Meyer L, Busch M-S, Ott F, PingetM. Use of continuous glucose monitoring system (CGMS) in 23insulin treated diabetic patients: Feasibility, reliability and efficacyfor the diagnosis of unnoticed hypoglycemia (glycemia ≤ mmol/l).Diabetes 2001;50(Suppl 2):A447.

48. Tamada JA, Garg S, Jovanovic L, Pitzer KR, Fermi S, Potts RO,Cygnus Research Team. Noninvasive glucose monitoring: Compre-hensive clinical results. JAMA 1999;282:1839–1844.

49. Chase PH, Roberts MD, Wightman C, Klingensmith G, Garg SK,Van Wyhe M, Desai S, Harper W, Lopatin M, Bartkowiak M, TamadaJ, Eastman RC. Use of the Glucowatch Biographer in children withtype 1 diabetes. Pediatrics 2003;111:790–794.

50. GlucoWatch G2 Product Insert, Sankyo, Cygnus, 2002.51. Edelman SV, Weyer C. Unresolved challenges with insulin therapy

in type 1 and type 2 diabetes: Potential benefit of replacing amylin,a second β-cell hormone. Diab Tech & Ther 2002;4:175–189.

52. Whitehouse F, Kruger DF, Fineman M, Shen L, Ruggles JA, MaggsDG, Weyer C, Kolterman OG. A randomized study and open la-bel extension evaluating the long-term efficacy of pramlintide asan adjunct to insulin therapy in type 1 diabetes. Diabetes Care2002;25:724–730.

53. Levetan C, Want LL, Weyer C, Strobel SA, Crean J, Wang Y, MaggsDG, Kolterman OG, Chandran M, Mudaliar SR, Henry RR. Im-pact of pramlintide on glucose fluctuations and postprandial glu-cose, glucagon, and triglyceride excursions among patients withtype 1 diabetes intensively treated with insulin pumps. DiabetesCare 2003;26:1–8.

evolving approaches to intensive insulin therapy in type 1 diabetes: multiple daily injections,...

Documents