Fasting Hyperglycemia: Somogyi and Dawn Phenomenae

Fasting Hyperglycemia: Somogyi and Dawn Phenomenae

Dr. S.M.Sadikot.
Hon. Endocrinologist,
Jaslok Hospital and Research Centre,
Mumbai 400026

Mr. P.F.A, a male aged 38 years was admitted in a comatose condition around 6 a.m. The accompanying relative informed the resident on duty that due to the fact that the patient regularly got up by 5.30 a.m., efforts had been made to wake him up, and when this failed, he had been urgently shifted to the hospital The only history that the relative could give was that the patient had diabetes, was taking insulin, and the family physician had told them that the blood glucose control of the patient was poor, possibly due to the erratic eating habits.

O/E, the patient was comatose with a pulse of 110/minute, regular but "racy". The BP was recorded as 180/90. The patient felt cold and clammy and his clothes were drenched. The heart sounds were well heard, there was no murmer. The examination of the chest and the abdominal system was normal except for sinus tachycardia, and the examination of a catheter specimen of urine showed the presence of a trace of glucose but no ketones. The random blood glucose estimation done using a bedside glucose meter showed the levels to be 45mg%. Once blood had been collected for other tests, 100c.c. 25% dextrose was slowly infused intravenously followed by a 5% dextrose drip. The patient gradually regained consciousness and in a few hours seemed to be back to normal.

At this time, a further detailed history was elicited. The patient informed us that he had been diagnosed as having diabetes about 2 months previously and had been put on a single pre-breakfast dose of 20 units of an intermediate acting insulin. This had been gradually increased to 32 units and a week back, when his blood glucose had been examined, the fasting blood glucose was found to be 160 mg% whilst the post lunch values were 172 mg%. Surprisingly, his HbA1c levels were in mid to high normal range. In view of the fasting hyperglycemia, the pre-breakfast dose of the insulin had been increased to 36 units, and 3 days later to 40 units. In spite of this, the fasting blood glucose levels could not be controlled, and in fact there was some rise seen. At this time, the patient volunteered the information that he had been feeling quite unwell, and when seen by his doctor had complained of tiredness and early morning headache. For the first time, his BP was found to be slightly high. His tiredness had been attributed to the uncon- trolled diabetes and the headache to the high blood pressure.

The previous morning, he had been asked to take his usual dose of 40 units of the intermediate acting insulin. In addition, he was asked to take 10 units of the same insulin before his evening meal and also asked to halve the bedtime milk. It was hoped that this would decrease the raised early morning blood glucose levels.

This case report illustrates a relatively common dilemma seen when patients are treated with "conventional" insulin regimens. The fasting blood glucose levels tend to be in the unacceptable high range whilst the control throughout the day may be quite fair!

From practical viewpoint, there seem to be three main causes for this:

  1. A simple waning of the insulin effect so that there is not enough insulin at nighttime to maintain control over the blood glucose levels in the early morning period;
  2. The Dawn phenomenon; and
  3. The Somogyi phenomenon;

It is of greater importance to correctly distinguish between the Dawn and the Somogyi phenomenae as the management of these are radically different and a misdiagnosis can lead to quite a catastrophe, as happened to this patient.

When Type 1 patients were closely monitored during intensive conventional therapy or whilst being on continuous subcutaneous insulin infusion (CSII), overnight monitoring of the blood glucose levels showed that the blood glucose levels rose and the insulin requirements increased between 3 a.m. and 7 a.m. and this increase in the insulin requirements and the hyperglycemia that followed was referred to as the "Dawn" phenomenon. Later, this phenomenon was also reported in Type 2 patients and even in people who did not have diabetes.

It is now fairly well established that many hormones undergo a diurnal variation. These changes in hormonal levels occur both, in people with and even in those who do not have diabetes. Soon after the onset of sleep, there is a surge in the secretion of growth hormone. Cortisol shows a significant increase between 2 a.m. and 4 a.m. and reaches peak values shortly after waking up. The catecholamines gradually increase overnight whilst glucagon remains relatively constant. Growth hormone, cortisol and the cathecolamines are all potent insulin antagonists and these changes in the secretion of the antagonists seen during the overnight period could explain the increased insulin requirements and the consequent early morning hyperglycemia.

Recent studies have shown that it is the growth hormone surge seen soon after a person falls asleep, which seems to be the determining factor in the Dawn phenomenon. Although the increase in the cortisol and the cathecolamines may not by themselves be essential for the Dawn phenomenon, it is quite plausible that they have an added or even synergistic effect. The obvious consequence of the rise in the levels of these insulin antagonist hormones leads to a relative insulin "deficiency" and unless more insulin is secreted by the body (or is administered), hepatic neoglucogenesis would not be finely controlled and more and more glucose would be delivered to the bloodstream leading to early morning hyperglycemia.

The Somogyi phenomenon is a manifestation of hyperglycemia following, possibly unrecognised, hypoglycemia. The clinical picture here, is one of worsening control in the face of increasing insulin doses and is manifested especially by early morning hyperglycemia. Excessive insulin dosage could lead to hypoglycemia, which occurring at night-time when the patient is asleep, may go unrecognised. But the counter-regulatory hormones (cathecolamines, glucagon, growth hormone and cortisol) rise as a consequence of the hypoglycemia and cause the blood glucose to rise. As there does not seem to be "fine tuning" of the magnitude of the counter regulatory responses and because these hormones are also insulin antagonists, there is an often an overshoot of the blood glucose which tend to go much higher than normal. Increasing the amount of insulin in order to counter the fasting hyperglycemia serves to intensify the nocturnal hypoglycemia and increase the counter regulatory response. This may be seen as a further increase in the fasting hyperglycemia in spite of increased insulin doses. Too much of an increase in the insulin dose could of course lead to a level of hypoglycemia from which the counter regulatory hormones may not be able to increase the blood glucose levels and the patient may manifest with frank hypoglycemic coma.

The third condition which commonly leads to a fasting hyperglycemia is a waning of the insulin effect seen overnight. In other words, the amount of insulin in the body decreases overnight and this decreased amount of insulin is not enough to prevent a rise in the blood glucose levels. In fact, at one time, it was thought that the Dawn phenomenon could be caused by an increase in the metabolism of insulin at night so that there would be less "active" insulin available to control the blood glucose levels. But studies in which insulin was infused in patients such that the insulin levels were kept constant throughout the night did not prevent the Dawn phenomenon. Fortunately, the management of the Dawn phenomenon and the problems of waning insulin levels is basically the same and therefore, the main clinical problems is the differentiation between the Dawn phenomenon and the Somogyi phenomenon.

The best way to distinguish between the two would be to estimate the blood glucose levels at different times throughout the night. The patient could do self monitoring of the blood glucose levels at home, or if hospitalised, this could be done by more conventional methods. The estimations should be done at bedtime, at midnight and at 3 a.m.

In patients showing the Somogyi phenomenon, a typical pattern would be normal or near normal, blood glucose levels at bedtime and midnight which decrease appreciably around 3 a.m. It has been shown that a value of around 70 mg% at 3 a.m. would suggest the presence of the Somogyi phenomenon.

In patients manifesting the Dawn phenomenon, the blood glucose levels at bedtime, midnight and at 3 a.m. would tend to be similar and after this time, the blood glucose levels would increase to reach hyperglycemic levels by early morning.

If it is not possible to do the blood glucose estimations, one method that may give a clue to the presence of the Somogyi phenomenon is to do a urine glucose estimations (in a double voided specimen) around 3 a.m. and in the early hours. If the former shows no glucose but the latter is MARKEDLY positive, the patient may be experiencing posthypoglycemic hyperglycemia. There are several other clues that may also point to this diagnosis. Rapid fluctuations in the results of the urine tests showing negative results and then suddenly changing to maximal values. Wide swings in the blood glucose levels which are not related to meal intake is another pointer to the diagnosis. It is also imperative that one should be able to recognise the subtle symptoms and signs of nocturnal hypoglycemia. The patient may complain of enuresis, nightmares, unusual sweating during the night, early morning headaches. A typical patient may also show a raised blood pressure in the early hours which tends to normalise after breakfast!

Once the diagnosis of the problem is clear, the management does not offer any difficulty. In patients with the Dawn phenomenon, one needs to increase the overnight insulin whilst the reverse seems to be the case with the Somogyi Phenomenon. How this is done would obviously depend on the individual patient.

For the management of the Dawn phenomenon, one or more of the following could be done:

  1. Give a small pre-dinner dose of intermediate acting insulin, if the patient is not already taking one;
  2. If the patient is taking a dose of an intermediate acting insulin before dinner, this could either be judiciously increased or the timing of the injection delayed so that the maximum activity could be seen around 2-4 a.m.;
  3. A small dose of regular insulin could be given before the post-dinner(bedtime) snack;

The purpose of all these manipulations would be to increase the amount of insulin available in the body during the crucial 2-4 a.m. period. It is obvious that if an additional injection is given or the dose of the insulin is increased, corresponding changes must be made in the dose of insulin given at other times in order to prevent daytime hypoglycemia.

The strategies to manage the Somogyi phenomenon could be:

  1. Decrease the dose of the evening insulin injection;
  2. Add, or increase, the amount of dinner or, preferably, the bedtime snack;
  3. Decrease the amounts of intermediate or long acting insulin given during the morning;

The patient, Mr. P.F.A. was obviously manifesting the Somogyi phenomenon in the previous days. There were several clues present which could have pointed to the correct diagnosis. There was a worsening of blood glucose control in spite of increasing the insulin dose as could be seen from the increase in the levels of early morning hyperglycemia. The patient also complained of feeling extremely tired in the morning and of having a headache. His BP was found to be high for the first time after the increase in the insulin doses. An indirect clue could be the mid to high normal levels of HbA1c of unacceptable fasting blood glucose levels. Since HbA1c tends to "average" out the ambient blood glucose levels, his level in context of raised fasting blood glucose levels and "fair" post-meal values should have pointed out to the possibility that the patient may have been undergoing episodes of very low levels of blood glucose at some time during the 24 hours.

Glucose Concentration (mg/100ml)
Whole Blood Plasma
Venous Capillary Venous
Diabetes Melittus
Fasting >/=110 >/=110 >/=126
2 hours post Glucose Load or both >180 >200 >200
Impaired Glucose Tolerance
Fasting(If measured) <110 <110 <126
2 hours post Glucose Load >/=120 & </=180 >/=140 & </=200 >/=140 & </=200
Impaired Fasting Glycemia
Fasting >/=100 & </=110 >/=100 & </=110 >/=110 & </=126
2 hours PG (If measured) <120 <140 <140

When analyzing the results, it becomes apparent that all the test results can be divided into three main categories: people who have diabetes, people who do not have diabetes and an intermediate category which depending on the method of testing is known as having either Impaired Glucose Tolerance (IGT) or Impaired Fasting Glycemia (IFG).

Before the ADA changed the procedures for testing, one had to carry out the full 2 hours GTT. When the results of this are analysed, one could be diagnosed as having IGT. This is an intermediate category between those who have frank diabetes and those whose results fall clearly in the normal range. Some of these patients will later go on to have diabetes, whilst others may not progress to the stage of frank diabetes. Patients falling in this category should be closely followed such that, if they progress to the diabetic stage, this would be diagnosed at an early stage. At the same time, many of the patients who are in the IGT category tend to be overweight, have problems with their lipid levels, are hypertensive and have a higher than acceptable serum uric acid level. I.G.T. is risk factor for the development of macrovascular disease.

Thus even if these patients are not diabetics, they should undergo management for these problems.

If one uses only the fasting values as suggested by the ADA, it would not be possible to diagnose IGT as the 2 hours postglucose values needed for a diagnosis of IGT would not be available. Impaired Fasting Glycemia (IFG) is an entity which has recently been introduced to delineate persons in whom only the fasting blood glucose has been done, but who do not come in the normal or diabetic category. Many of the associated disorders we mentioned as being possibly present in those with IGT are also true for those with IFG. In fact, IFG is felt to reflect a higher average glycemic burden than IGT. It is considered a marker for the development of diabetes and its long term complications.

Although there is still some controversy about using only the fasting blood glucose values for diagnosis, it is now felt that this would be acceptable to pinpoint those with diabetes in prevalence studies. The WHO feels that in most cases this should be followed by a full GTT, if feasible.

Most people tested would show "normal" values.


Gestational Diabetes Mellitus (GDM) is defined by abnormal glucose tolerance during pregnancy; the glucose tolerance test is normal before, and which will usually be normal, after pregnancy. Present in around 3-4% of all pregnancies, GDM can be associated with significant morbidity and mortality in the fetus and newborn. Thus, it is important for gestational diabetes to be ruled out in all pregnancies.

Ideally, all pregnant women should be tested to rule out gestational diabetes, but if this is not feasible, all high risk patients must undergo the test.

  1. Women who had GDM during a previous pregnancy.
  2. Women with a first degree relative who is a diabetic.
  3. Women who gave birth to large weight babies in a previous pregnancy.
  4. Women whose newborn, in a previous pregnancy, showed any complication known to be associated as arising from maternal GDM
  5. Women who gave birth to still born babies or infants with congenital abnormalities.
  6. Women with a bad obstetric history, including recurrent fetal wastage, hypertension, eclampsia, hydramnios, etc.
  7. Women with repeated or persistent urinary tract infection.
  8. Women manifesting glycosuria during pregnancy.
  9. Women over the age of 30 years.

The test should be carried out at the time of initial visit and at the start of every trimester; high risk patients may require more frequent testing

Initial screening produce may be done by estimating the fasting glucose levels and the levels 1 hour after an oral dose of 50 gms. of glucose. This test can be carried out in the fasting stage or at any time; in the latter case, only the one hour blood glucose value is taken into consideration for diagnosis.

Patients with a fasting venous whole blood glucose level of more than 80 mg% (venous plasma glucose more than 90 mg%).


a 1 hour post 50 gms. glucose load venous whole blood glucose value greater than 120 mg/% (venous plasma glucose more than 140 mg% require a more comprehensive test.


A "random" venous blood glucose level exceeding 105 mg% (plasma glucose >120 mg%) also merits a more comprehensive test.

The comprehensive test is the same as described for the diagnosis of diabetes in non pregnant persons. BUT, the criteria differ. In addition to blood glucose levels in the diabetic range, values suggestive of IGT, in a pregnant female, should be taken to be diagnostic of gestational diabetes.

Many centres still utilise the O'Sullivan Criteria.

In this test, blood is collected in the fasting stage and then at 1, 2 and 3 hours intervals after an oral load of 100 gms. of glucose.

Plasma Glucose (mg/100ml) Whole Blood Glucose (mg/100ml)
Fasting 105 90
1 hour 190 165
2 hour 165 145
3 hour 145 125

Two of the four values must be met to diagnose GDM

As a matter of interest, about 30-40% of women who are diagnosed as having gestational diabetes will later go on to have full fledged diabetes within 5-10 years of the pregnancy. Thus, they would automatically come into the High Risk category and must have annual check ups as discussed above.

I am often asked whether an estimation of Glycosylated Hemoglobin level can be used as the parameter to diagnose diabetes. Let me make it quite clear that, at the present times it is not possible to accurately make this distinction between those that are normal and those that have diabetes or may be in the IGT or IFG category. Although the estimation of these levels are excellent to give an idea of the overall control of the blood glucose levels in the preceeding 6-8 weeks, they CANNOT be used for the diagnosis of diabetes, in most patients. The argument that a very high glycosylated hemoglobin level would be diagnostic for diabetes can be very easilv countered by the fact that if the glycosylated hemoglobin level were indeed to be so high, then it is apparent that even a random estimate of the blood glucose levels would show it to be in the distinctly "diabetic" range, and this can be done at much less cost to the patient!

Lastly, once a patient has been diagnosed as being a diabetic, he need NEVER undergo a Glucose Test. This test is only for diagnosis. A known diabetic needs to know whether he is under optimal control or not and stressing him with a glucose load will definitely not give an answer to this problem. This is an important aspect as many doctors, including "specialists", have the patient undergo repeated G.T.T's! When a diabetic comes for a follow up with a fresh G.T.T., it is of absolutely no help in his management.