Oral hypoglycaemic drugs and gastrointestinal symptoms in diabetes mellitus

Transcription

1 Aliment Pharmacol Ther 2001; 15: 137±142. Oral hypoglycaemic drugs and gastrointestinal symptoms in diabetes mellitus P. BYTZER*, N. J. TALLEY*, M. P. JONES* & M. HOROWITZ *Department of Medicine, University of Sydney, Nepean Hospital, Sydney, Australia; and Department of Medicine, University of Adelaide, Royal Adelaide Hospital, Adelaide, Australia Accepted for publication 5 September 2000 SUMMARY Background: Gastrointestinal symptoms are commonly reported as side-effects of oral hypoglycaemic drugs. It may be very dif cult to distinguish between spontaneous and truly drug-related symptoms due to the high background incidence of gastrointestinal symptoms. Gastrointestinal symptoms in diabetic patients have also been linked to factors associated with long-standing disease and suboptimal control. Aim: To explore the association between gastrointestinal symptoms and treatment with oral hypoglycaemic drugs in a large cohort of subjects with type 2 diabetes. Patients and methods: 956 subjects with type 2 diabetes participated in the study. All subjects completed a validated, self-administered questionnaire on gastrointestinal symptoms, diabetes, drug use and various potential risk factors for gastrointestinal symptoms. The association between oral hypoglycaemics and nine gastrointestinal symptom groups was assessed based on logistic regression. Results: 405 of the 956 patients used oral hypoglycaemic drugs. Metformin use was independently associated with chronic diarrhoea (odds ratio 3.08, 95% CI: 1.29±7.36, P < 0.02) and with faecal incontinence (odds ratio 1.95, 95% CI: 1.10±3.47, P < 0.05). Use of sulphonylureas was associated with less abdominal pain, but not with any other gastrointestinal symptom. Conclusions: Troublesome gastrointestinal symptoms do not appear to be caused by oral hypoglycaemics, except for diarrhoea and faecal incontinence, which are strongly and independently associated with metformin use. INTRODUCTION The prevalence of gastrointestinal symptoms amongst people with diabetes appears to be increased compared to the general population (P. Bytzer et al., unpublished data), 1±5 although not all studies agree. 6 In our recent population-based study, all upper and lower gastrointestinal symptoms evaluated were more common in diabetics than in controls (P. Bytzer et al., unpublished data). However, there is substantial controversy as to Correspondence: Dr P. Bytzer, Department of Medicine, Division of Gastroenterology, Glostrup University Hospital, DK-2600 Glostrup, Denmark. peter.bytzer@dadlnet.dk the mechanisms responsible for gastrointestinal symptoms in subjects with diabetes. Gastrointestinal symptoms are commonly reported as side-effects of drugs, including oral hypoglycaemics, 7 particularly metformin 8 and alpha glucosidase inhibitors. 9 However, due to the high background incidence of gastrointestinal symptoms it may be very dif cult to distinguish between spontaneous and truly drug-related gastrointestinal symptoms. Gastrointestinal symptoms in diabetic patients have also been linked to factors associated with long standing disease and suboptimal control. In particular, autonomic neuropathy, 1, 4, 10 poor glycaemic control (P. Bytzer et al., unpublished data) 2 and duration of diabetes 11 have been reported to be important risk Ó 2001 Blackwell Science Ltd 137

2 138 P. BYTZER et al. factors for gastrointestinal symptoms in diabetes mellitus. Previous studies on the relationship between gastrointestinal symptoms and oral hypoglycaemic drug use have substantial limitations. The results have been based either on the sporadic reporting of side-effects 7, 9, 12 from clinical trials, or on the prevalence of lower gastrointestinal symptoms in highly selected subgroups of diabetes patients. 13, 14 More importantly, adjustments for other risk factors for gastrointestinal symptoms were not performed in these studies and the reported associations may thus have been spurious. It was the aim of this study to explore the association between gastrointestinal symptoms and treatment with oral hypoglycaemic drugs in a large cohort of subjects with type 2 diabetes. We hypothesized that gastrointestinal symptoms would be more prevalent in patients treated with oral hypoglycaemics, even considering other potential risk factors. SUBJECTS AND METHODS Subjects A total of 956 subjects with type 2 diabetes participated in the study. They were derived from two distinct populations: 1 A consecutive sample of 209 out-patients with diabetes were recruited from clinics that comprise the Nepean Hospital Diabetes Service in Sydney (123 patients) and the Endocrine and Metabolic Unit of the Royal Adelaide Hospital in Adelaide (86 patients). Those who had previous gastrointestinal surgery, dementia or major psychosis, or were unable to complete the questionnaire were excluded. 2 A sample of 892 subjects from the mailing list of Diabetes Australia; the latter contains approximately registered individuals with diabetes and 1800 subjects were initially identi ed. Two separate mailings were carried out 12 months apart. Subjects were classi ed as having type 1 diabetes if their age at diagnosis was < 30 years and they had used insulin ever since they were diagnosed with diabetes. 15 All other diabetic subjects were classi ed as having type 2 diabetes and only these patients (n ˆ 956) were included in the current analyses. Measurements Each individual completed a copy of the Diabetes Bowel Symptom Questionnaire (DBSQ), a self-administered screening instrument which is based on a previously validated US questionnaire. 16, 17 The DBSQ can quantify chronic gastrointestinal symptoms, self-reported glycaemic control, type of diabetes, treatment of diabetes, and duration of diabetes; it is reliable and provides information that is comparable with a physician interview. The questionnaire contains detailed information about upper and lower abdominal symptoms over the prior year. All symptoms that are not completely self-explanatory are anchored to a standard description. The subjects were asked to rate their control of blood glucose levels over the prior year on a ve-point Likert scale which included the following options: excellent control, good control, fair control, poor control, and very poor control. We have shown elsewhere that selfreported glycaemic control correlates strongly with measurement of haemoglobin A1c. 18 The following self-reported complications of diabetes were considered in the multivariate models: peripheral neuropathy, nephropathy, and proliferative retinopathy. The complications were classi ed as present according to the following de nitions: Nephropathy. The patient answered yes to the question: `Did your doctor tell you about any kidney damage or protein in your urine that is a result of your diabetes?' Peripheral neuropathy. The patient answered yes to the question: `Do you suffer from ``pins and needles'' in your feet or hands?' Retinopathy. The patient answered yes to the question: `Are you aware of any eye damage that is a result of your diabetes?' and had received laser therapy. We have shown elsewhere that there was a strong association between peripheral neuropathy and other symptoms suggestive of diabetic neuropathic complications, including abnormal sweating, postural hypotension and impotence in males. 18 Oral hypoglycaemic drug use Use of oral hypoglycaemic drugs were derived from the questionnaire. Patients were classi ed into those taking

3 HYPOGLYCAEMIC DRUGS AND GASTROINTESTINAL SYMPTOMS IN DIABETES 139 sulphonylureas, metformin, acarbose or combinations of these drug classes. In total, 405 of the 956 patients were taking oral hypoglycaemic drugs; 195 patients were taking metformin alone, 188 patients a sulphonylurea drug alone, and two patients acarbose. Fourteen patients were taking both metformin and a sulphonylurea drug, two patients both metformin and acarbose, and four patients both acarbose and a sulphonylurea drug. Of the remaining 551 patients, 358 patients were taking insulin and 193 patients were treated by diet only. At the time of the study, repaglinide and troglitazone were not available in Australia. Gastrointestinal symptoms Patients were classi ed into the following nine gastrointestinal symptom groups, constructed a priori to cover a broad range of gastrointestinal complaints. Unless stated otherwise, a symptom was counted only if reported to occur more than 25% of the time over the past 12 months. Gastrointestinal symptom groups: 1 Frequent abdominal painðabdominal pain of at least moderate intensity occurring at least once a week over the past 12 months. 2 Gastro-oesophageal re ux symptomsðheartburn or acid regurgitation at least once a week. 3 Dysmotility-like dyspepsiaðepigastric pain or discomfort in combination with any upper dysmotility symptom (early satiety, postprandial fullness, bloating, abdominal swelling, nausea (at least 2±3 times per month), vomiting (at least once a month), and retching (at least once a month)). 4 Ulcer-like dyspepsiaðlocalized epigastric pain or discomfort combined with at least two of the following characteristics: night pain, episodic pain, pain relieved by antacids, pain relieved by food or milk, pain provoked by being hungry. 5 Bowel-related abdominal painðabdominal pain or discomfort associated with at least two of three features: relieved with defaecation; onset associated with a change in the frequency of stool; onset associated with a change in stool form. 6 ConstipationÐat least two of the following symptoms: < 3 bowel movements per week, a feeling of anal blockage, manual disimpaction, lumpy or hard stools, a feeling of incomplete evacuation, straining, bran or cereal taken daily because of constipation. 7 DiarrhoeaÐvery loose or watery stools more than 75% of the time and no abdominal pain. 8 Steatorrhoea-like stoolsðvery pale (oily), bulky, smelly bowel movements that were dif cult to ush away. 9 Faecal incontinenceðleakage of bowel movements about once a month or more frequently. Statistical analyses The association between use of oral hypoglycaemics and the proportion of patients with and without a particular gastrointestinal symptom was assessed based on a logistic regression analysis. The following confounders were considered in the models: age (in years), gender, smoking (never/past/current), alcohol consumption (never/1±2 drinks a week/3±6 drinks a week/7±10 drinks a week/11±30 drinks a week/> 30 drinks a week), duration of known diabetes (6 months/ 6 months±2 years/> 2±5 years/> 5±10 years/> 10± 20 years/> 20 years), neuropathy (yes/no), nephropathy (yes/no), proliferative retinopathy (yes/no), current co-medication with any other drug (yes/no) and selfreported glycaemic control (excellent/good/fair/poor/ very poor). Similar analyses were carried out in the subgroups of sulphonylurea drug users (n ˆ 206) and metformin users (n ˆ 211). Due to the low number of acarbose users (n ˆ 8), this drug class was not analysed separately. Results are reported as odds ratios (OR) with 95% con dence intervals (CI). Differences in characteristics between groups were tested using t-test, a Mann± Whitney test or v 2 test, as appropriate. All P-values calculated were two-tailed; the alpha level of signi cance was set at P < RESULTS The patient characteristics, including prevalence rates of gastrointestinal symptoms and diabetic complications, did not differ between the two diabetic populations (out-patients and subjects contacted by mail); therefore, they were analysed together. The patient characteristics are summarized in Table 1. Compared to patients who did not use oral hypoglycaemic drugs, drug users were slightly younger (P < 0.05), they had a

4 140 P. BYTZER et al. Table 1. Characteristics of study population with type 2 diabetes mellitus according to oral hypoglycaemic drug use Oral hypoglycaemic drug use Type of drug All patients n ˆ 956 Yes n ˆ 405 No n ˆ 551 Sulfonylureas n ˆ 206 Metformin n ˆ 211 Males, % Age, mean SD Duration of known diabetes > 10 years, % Insulin users, % Current smokers, % Past smokers, % Alcohol users,* % Self-reported glycaemic control, % excellent good fair poor very poor *³ 3 standard drinks a week. shorter duration of known diabetes (P < ), and were less likely to use insulin (P < ) and to report retinopathy (P < ). Compared to users of sulphonylureas, the sub-group of metformin users were more likely to use insulin (P < 0.02) and to report nephropathy (P < 0.05). However, all of these factors were controlled for in the analyses. Any oral hypoglycaemic drug use was signi cantly associated with diarrhoea (OR 2.72, 95% CI: 1.08± 6.84, P < 0.05). This association was con ned to the metformin users (OR 3.08, 95% CI: 1.29±7.36, P < 0.02, Table 2). Metformin use was also signi cantly associated with faecal incontinence (OR 1.95, 95% CI: 1.10±3.47, P < 0.05, Table 2). Use of sulphonylureas was associated with less abdominal pain (OR 0.40, 95% CI: 0.19±0.82, P < 0.02, Table 3), but not with any other gastrointestinal symptom. Eight of the 25 metformin users who reported diarrhoea also had faecal incontinence. In a separate analysis, we examined the independent effects of diarrhoea and faecal incontinence in metformin users after adjusting for age, gender, alcohol, smoking, duration of diabetes, insulin use, neuropathy, nephropathy, Prevalence rates % (95% CI) Metformin use Adjusted* odds Symptoms No (n ˆ 745) Yes (n ˆ 211) ratios with 95% CI Table 2. Prevalence rates of gastrointestinal symptoms in diabetic subjects according to metformin use Frequent abdominal pain 9.0 (6.9±11.1) 8.5 (4.8±12.3) 0.97 (0.55±1.73) Gastro-oesophageal re ux 15.6 (13.0±18.2) 13.7 (9.1±18.4) 0.83 (0.51±1.34) Dysmotility-like dyspepsia 10.5 (8.3±12.7) 11.9 (7.5±16.2) 1.02 (0.61±1.72) Ulcer-like dyspepsia 6.6 (4.8±8.4) 8.5 (4.8±12.3) 1.30 (0.70±2.39) Bowel-related pain 13.7 (11.2±16.2) 13.3 (8.7±17.9) 0.80 (0.50±1.30) Constipation 25.2 (22.1±28.4) 20.9 (15.4±26.3) 0.76 (0.51±1.13) Diarrhoea 1.9 (0.9±2.9) 5.2 (2.2±8.2) 3.08 (1.29±7.36) Steatorrhoea-like stools 16.2 (13.5±19.0) 14.6 (9.8±19.5) 0.81 (0.50±1.30) Faecal incontinence 6.3 (4.6±8.1) 10.4 (6.3±14.6) 1.95 (1.10±3.47)à *Adjusted for age, gender, smoking, alcohol, duration of diabetes, complications of diabetes, self-reported glycaemic control, insulin use and other drug use. P < àp < 0.05.

5 HYPOGLYCAEMIC DRUGS AND GASTROINTESTINAL SYMPTOMS IN DIABETES 141 Table 3. Prevalence rates of gastrointestinal symptoms in diabetic subjects according to use of sulfonylurea drugs Prevalence rates % (95% CI) Sulfonylurea drug use Adjusted* odds Symptoms No n ˆ 750 Yes n ˆ 206 ratios with 95% CI Frequent abdominal pain 10.0 (7.9±12.2) 4.9 (1.9±7.8) 0.40 (0.19±0.82) Gastro-oesophageal re ux 16.0 (13.4±18.6) 12.1 (7.7±16.6) 0.70 (0.42±1.15) Dysmotility-like dyspepsia 11.2 (8.9±13.5) 9.2 (5.3±13.2) 0.77 (0.44±1.34) Ulcer-like dyspepsia 7.1 (5.2±8.9) 6.8 (3.4±10.2) 0.80 (0.41±1.56) Bowel-related pain 14.1 (11.6±16.6) 11.7 (7.3±16.0) 0.76 (0.46±1.24) Constipation 24.1 (21.1±27.2) 24.8 (18.9±30.7) 0.98 (0.67±1.45) Diarrhoea 2.8 (1.6±4.0) 1.9 (0.1±3.8) 0.65 (0.21±1.96) Steatorrhoea-like stools 15.9 (13.2±18.6) 15.8 (10.8±20.8) 1.05 (0.67±1.65) Faecal incontinence 7.7 (5.8±9.7) 5.3 (2.3±8.4) 0.72 (0.36±1.43) *Adjusted for age, gender, smoking, alcohol, duration of diabetes, complications of diabetes, self-reported glycaemic control, insulin use and other drug use. P < retinopathy, current co-medication and glycaemic control. We found that metformin users were clearly differentiated from patients who did not use metformin in terms of diarrhoea (OR 2.47, 95% CI: 1.04±5.58; P < 0.05) and, while less clear, also in terms of faecal incontinence (OR 1.72, 95% CI: 0.98±3.07; P ˆ 0.07). DISCUSSION There is substantial controversy as to the mechanisms responsible for gastrointestinal symptoms in diabetes mellitus. Gastrointestinal symptoms are commonly reported as side-effects to treatment with oral hypoglycaemic drugs. 8, 19 However, gastrointestinal symptoms are very common, both in diabetic subjects and in the community, and a causal relationship is dif cult to prove. Furthermore, previous studies on potential risk factors for gastrointestinal symptoms in diabetes have been con icting; 1, 4, 11, 18, 20 and the majority of studies can be criticised on methodological grounds. The present study included a broad spectrum of diabetic subjects recruited both from the community and from out-patients. We used a validated questionnaire and we have provided evidence that our measures of self-reported glycaemic control and diabetic complications are valid. 18 Since gastrointestinal symptoms may potentially be linked to age, gender and various risk factors including diabetic complications, glycaemic control and other drug use, adjustments were made for these in the multivariate models. We found that metformin use was strongly associated with diarrhoea and faecal incontinence, but not with any other gastrointestinal symptom. Use of sulphonylureas was not associated with any of the nine gastrointestinal symptom groups. Therefore, it is unlikely that oral hypoglycaemics are responsible for other gastrointestinal symptoms in patients with type 2 diabetes. Importantly, our results cannot be interpreted to suggest that oral hypoglycaemics do not cause gastrointestinal symptoms in some patients. Patients experiencing gastrointestinal symptoms while taking oral hypoglycaemics may have stopped this treatment if they suspected that their symptoms were the result of a drug related side-effect. Adverse gastrointestinal effects like diarrhoea, nausea, epigastric discomfort and anorexia have been reported to occur in up to 30% of patients treated with metformin. In most patients, these sideeffects are transient and dose-related. Gastrointestinal side-effects occur mainly during the rst week of therapy, generally disappearing spontaneous as therapy is continued. 19 Since we did not speci cally ask about details of dosing and duration of treatment or about gastrointestinal symptoms associated with previous drug use, such an effect cannot be ruled out. The signi cant association between metformin use and chronic diarrhoea and/or faecal incontinence is not a new nding. In a questionnaire-based survey of 285 diabetic patients, metformin was the commonest cause of diarrhoea and incontinence, which was found to occur in 20% of those on metformin. 13 Recently, Lysy et al. 14 con rmed that metformin was the most common cause of severe diarrhoea in a large survey of 861 diabetes patients. However, these studies did not take other potential risk factors, like diabetic complications, glycaemic control, gender, age, or other drug use into account, and the reported association may thus have

6 142 P. BYTZER et al. been spurious. Moreover, this is the rst community diabetes study to demonstrate such an association. The mechanism(s) causing diarrhoea in diabetes patients taking metformin is not clear. Dandona et al. have suggested an increased intestinal motility as a possible mechanism. 13 The strong association with faecal incontinence but not with steatorrhoea-like stools in the present study supports this view. However, no studies have explored the pathophysiology of metformin-associated diarrhoea. In conclusion, troublesome gastrointestinal symptoms are common in patients with type 2 diabetes. These symptoms do not appear to be caused by oral hypoglycaemics, except for diarrhoea and faecal incontinence, which are strongly and independently associated with metformin use. ACKNOWLEDGEMENTS This study was supported by research grants from the National Health and Medical Research Council of Australia and Diabetes Australia. REFERENCES 1 Feldman M, Schiller LR. Disorders of gastrointestinal motility associated with diabetes mellitus. Ann Intern Med 1983; 98: 378±84. 2 Schvarcz E, PalmeÂr M, Ingberg CM, AÊ man J, Berne C. Increased prevalence of upper gastrointestinal symptoms in long-term type 1 diabetes mellitus. Diabet Med 1995; 13: 478±81. 3 Enck P, Rathmann W, Spiekermann M, et al. Prevalence of gastrointestinal symptoms in diabetic patients and nondiabetic subjects. Z Gastroenterol 1994; 32: 637±41. 4 SpaÊngeÂus A, El-Salhy M, Suhr O, Eriksson J, Lithner F. Prevalence of gastrointestinal symptoms in young and middleaged diabetic patients. Scand J Gastroenterol 1999; 34: 1196± Ricci JA, Siddique R, Stewart WF, Sandler RS, Sloan S, Farup CE. Upper gastrointestinal symptoms in a U.S. national sample of adults with diabetes. Scand J Gastroenterol 2000; 35: 152±9. 6 Janatuinen E, Pikkarainen P, Laakso M, PyoÈraÈlaÈ K. Gastrointestinal symptoms in middle-aged diabetic patients. Scand J Gastroenterol 1993; 28: 427±32. 7 Krentz AJ, Ferner RE, Bailey CJ. Comparative tolerability pro les of oral antidiabetic agents. Drug Safety 1994; 11: 223±41. 8 Davidson MB, Peters AL. An overview of metformin in the treatment of type 2 diabetes mellitus. Am J Med 1997; 102: 99± Johansen K. Acarbose treatment of sulfonylurea-treated noninsulin dependent diabetics. A double-blind cross-over comparison of an alpha glucosidase inhibitor with metformin. Diabetes Metab 1984; 10: 219± Mayne N. Neuropathy in the diabetic and non-diabetic populations. Lancet 1965; 2: 1313±6. 11 Ko GT, Chan WB, Chan JC, Tsang LW, Cockram CS. Gastrointestinal symptoms in Chinese patients with type 2 diabetes mellitus. Diabet Med 1999; 16: 670±4. 12 Phillips BB. Managing therapy and adverse effects with antihyperglycemic agents: a focus on metformin and acarbose. Pharm Pract Manag Q 1997; 17: 21± Dandona P, Fonseca V, Mier A, Beckett AG. Diarrhea and metformin in a diabetic clinic. Diabetes Care 1983; 6: 472±4. 14 Lysy J, Israeli E, Goldin E. The prevalence of chronic diarrhea among diabetic patients. Am J Gastroenterol 1999; 8: 2165± World Health Organization Study Group. Prevention of Diabetes. Geneva: World Health Organization, 1994: Talley NJ, Phillips SF, Melton LJI, Wiltgen C, Zinsmeister AR. A patient questionnaire to identify bowel disease. Ann Intern Med 1989; 111: 671±4. 17 Talley NJ, Boyce P, Owen BK, Newman P, Paterson K. Initial validation of a bowel symptom questionnaire and measurement of chronic gastrointestinal symptoms in Australians. Aust N Z J Med 1995; 25: 302±8. 18 Bytzer P, Talley NJ, Young LJ, Hammer J, Jones MP, Horowitz M. Gastrointestinal symptoms in diabetes mellitus are associated with diabetic complications and poor glycemic control. J Am Med Assoc, Submitted. 19 Haupt E, Knick B, Koschinsky T, Liebermeister H, Schneider J, Hirche H. Oral antidiabetic combination therapy with sulphonylureas and metformin. Diabetes Metab 1991; 17: 224± Clouse RE, Lustman PJ. Gastrointestinal symptoms in diabetic patients: lack of association with neuropathy. Am J Gastroenterol 1989; 84: 868±72.