Current issues of ACP Journal Club are published in Annals of Internal Medicine


Symptoms of hypoglycemia with human versus porcine insulin

ACP J Club. 1992 Mar-April;116:33-4. doi:10.7326/ACPJC-1992-116-2-033

Source Citation

Egger M, Smith GD, Teuscher AU, Teuscher A. Influence of human insulin on symptoms and awareness of hypoglycaemia: a randomised double-blind crossover trial. BMJ. 1991 Sep 14;303:617-21.



To compare symptoms of hypoglycemia experienced by patients with diabetes using human or porcine insulin.


Randomized, double-blind, crossover trial.


Diabetes outpatient clinic in Berne, Switzerland.


44 patients with insulin-dependent diabetes mellitus, using fast-acting and long-acting insulin, performing glucose self-measurement, and having stable glycemic control were selected. 31 patients used a pen injector for 4 injections daily; 13 patients took 2 injections daily. 12 patients had a history of ≥ 2 episodes of hypoglycemic coma within the previous 3 years.


Patients were randomly allocated to receive either human or porcine insulin for 6 weeks, then to switch to the other type, also for 6 weeks. Fingerstick blood glucose measurements were made 3 to 4 times per day.

Main outcome measures

Frequency of hypoglycemic symptoms of sweating, tremor, hunger, restlessness, lack of concentration, confusion, and visual disturbance was documented by patients using a standardized questionnaire completed after each episode of hypoglycemia. Hypoglycemia was defined as a blood glucose level ≤ 2.8 mmol/L.

Main results

There were no differences between the 2 treatment groups for glucose levels measured throughout the day for the 11 932 routine blood glucose measurements made. Hemoglobin A1c levels, fructosamine concentrations, and insulin doses were also similar. A smaller proportion of all recorded blood glucose measurements were ≤ 2.8 mmol/L during treatment with human insulin compared with porcine insulin (6.8% vs. 7.9%; P = 0.02). Symptom questionnaires were completed for 234 (56%) of 417 episodes of hypoglycemia occurring with human insulin treatment and 259 (51%) of 510 episodes with porcine insulin. A mean of 2.6 symptoms were reported per hypoglycemic episode. Lack of concentration, confusion, and restlessness occurred more frequently during treatment with human than with porcine insulin in an analysis based on the questionnaires (all P ≤ 0.015). The frequency of other hypoglycemic symptoms was similar in the 2 groups.


Glycemic control is similar with human and porcine insulin. During hypoglycemic episodes, lack of concentration, confusion, and restlessness occurred more often during treatment with human insulin.

Source of funding: Baxter AG.

Address for article reprint: Dr. A. Teuscher, Diabetes Section, Department of Medicine, University of Berne, Berne, Switzerland.


In a series of publications, this group has suggested that treatment of diabetes mellitus with human compared with animal insulin results in fewer neurogenic (autonomic) warning symptoms of developing hypoglycemia (with a corresponding predominance of neuroglycopenic symptoms) and, therefore, might cause an increased frequency of severe iatrogenic hypoglycemia. This suggestion has caused a great deal of concern among patients and physicians.

The present data indicate that 3 common neurogenic symptoms of hypoglycemia—sweating, tremor, and hunger—were not consistently less frequent during treatment with human insulin. However, lack of concentration, clearly a neuroglycopenic symptom, and restlessness, which cannot be readily classified, were more commonly reported with human insulin (53% vs. 35% and 50% vs. 38% of questionnaires, respectively). This study was not designed to determine if iatrogenic hypoglycemia is more frequent, more severe, or both during treatment with human insulin. The fact that there were fewer episodes of hypoglycemia and of hypoglycemic coma with human insulin does not support that possibility.

Support for the suggestion that treatment with human insulin causes a form of hypoglycemia unawareness has not been forthcoming from other investigators despite a rather extensive effort. Although a few contrary reports can be cited, there is now a substantial body of evidence that the hormonal, symptomatic, and cognitive responses to experimental hypoglycemia produced with human and animal insulin are indistinguishable. Similarly, cross-sectional clinical studies have not disclosed differences in the symptom patterns or in the frequency or severity of clinical hypoglycemia in patients treated with human or animal insulin. Thus, although I cannot reject the hypothesis categorically in the absence of a prospective trial, I find no compelling evidence that treatment of diabetes mellitus with human insulin causes clinically important hypoglycemia unawareness.

Washington University School of MedicineSt. Louis, Missouri

Author's Response

Dr. Cryer states that our hypothesis was that fewer autonomic symptoms would lead to increased frequency of severe hypoglycemia. In fact, as Dr. Cryer also points out, the study was not designed to investigate whether there was a difference in the incidence of severe hypoglycemia. Rather, we attempted to determine the differences in symptom patterns among patients taking human insulin compared with those in patients taking porcine insulin. In addition to the findings that Dr. Cryer mentions, we also found that hypoglycemic episodes of abrupt onset were significantly more frequent (P = 0.02) with human insulin and that there were fewer reports of the autonomic symptom of hunger when results were analyzed by patient rather than by questionnaire. The latter analysis is less clinically relevant and methodologically less sound because some patients with a large number of hypoglycemic episodes did not experience hunger any differently on the 2 types of insulin.

Dr. Cryer also did not mention the considerable effort that was made to classify symptoms into the two categories, autonomic and neuroglycopenic, using cross-tabulations and principal components analysis. These analyses indicated that restlessness should be considered a neuroglycopenic symptom, not as one "which cannot be readily classified."

We disagree with Dr. Cryer's opinion about the implications of our findings: We feel that the recording of fewer hypoglycemic episodes while patients were on human insulin is consistent with the notion that awareness of hypoglycemia is reduced by human insulin. The weight of evidence referred to by Dr. Cryer against the problem of unawareness of hypoglycemia mainly comprises small, laboratory investigations with low power to detect the problem and retrospective cross-sectional surveys, which are susceptible to selection bias (selective transfer of patients to human insulin) and lack a suitable control group. The only previous randomized, double-blind, crossover study on this matter (1) came to the same conclusion as our trial.

Finally, the conclusions given in the abstract merely repeat the main results. The analysis by questionnaire is methodologically less sound because it is not based on independent events. The differences for hunger disappeared because a few patients who did not experience hunger differently on the 2 types of insulin recorded a large number of hypoglycemic episodes.

M. Egger, MD G. Davey-Smith, MD
University CollegeLondon, United Kingdom

Editors' response

We have departed from our usual format of 1 page for the article by Egger and colleagues for several reasons. First, hypoglycemia unawareness among diabetic patients on insulin therapy is emerging as an important problem, but one that is difficult to study. Investigations such as that by Egger and his colleagues are contributing to better definition and measurement of symptoms of hypoglycemia, an important contribution in its own right.

Second, this study, and the earlier one of Berger and colleagues (see above), provide evidence about human insulin that has some patients and physicians upset and manufacturers of human insulin on edge. Human insulin was introduced with randomized controlled trials showing equivalence in blood sugar lowering effects but no evidence of patient benefit in addition to that conferred by animal insulin (unless one believes that freedom from insulin antibodies is clinically important). Larger premarketing studies would have been required to determine whether there was any other benefit or harm from human insulin, but it was perhaps assumed that human insulin was apt to be better than animal insulin. This is now open to question.

Third, it does not seem to us that this question has been resolved. Nevertheless, the positions of experts in the field are becoming polarized, with some claiming that no adverse effect of human insulin has been proved and others claiming that it has. As both the authors and commentator for the article abstracted in this issue of ACP Journal Club agree, there is need for more definitive studies, in which large enough numbers of patients are followed for long enough to observe the relative incidence and characteristics of hypoglycemia severe enough to cause important clinical consequences. Until such studies are done, there does not seem to us to be a sufficient basis for clinical policy.

Fourth, the authors' final remark provides an opportunity to underscore ACP Journal Club's policy against speculation in abstracts. If the conclusions of the authors stated in the original article stray beyond the data provided, we do not report them in the abstract. The commentator, however, may choose to mention them.

The symptoms of hypoglycemia experienced by patients on human insulin differ from those on porcine insulin. Whether this translates into clinically important consequences remains to be determined. Until then, clinicians should be alert for larger studies of the same rigor as that of Egger and colleagues. In the meantime, we do need to be sensitive to the matter of hypoglycemia unawareness, particularly among patients who are tightly controlled and who have been on insulin for many years, whether or not they are on human insulin.

Hertzel Gerstein, MD R. Brian Haynes, MD