Abstract
AIMS
To study reporting of hypoglycaemia in angiotensin receptor blocker (ARB) users, and to investigate the possibility of confounding.
METHODS
The French pharmacovigilance database was examined for an association between hypoglycaemia and ARBs or other drugs using reports notified between 1996 and 2005. This association was also tested in patients taking or not taking antidiabetic agents (ADAs) using reporting odds ratios (ROR).
RESULTS
Hypoglycaemia was mentioned in 807 of the 174 595 reports entered during the study period. Overall hypoglycaemia was associated with the use of ARBs [ROR 2, 95% confidence interval (CI) 1, 3] and with the use of ADAs (ROR 32, 95% CI 27, 37). Moreover, the use of ARBs was associated with the use of ADAs (OR 7, 95% CI 6, 8). Considering separately reports with and without ADA, the association of ARB use with a higher risk of hypoglycaemia disappeared (OR 0.4, 95% CI 0.2, 0.8 and OR 2, 95% CI 1, 3, respectively).
CONCLUSION
A signal indicating an association between ARB use and hypoglycaemia was found in the French pharmacovigilance database. This signal disappeared after stratification on ADA use, thus suggesting confounding by indication. Moreover, the association between ARB use and hypoglycaemia was negative in ADA users.
WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT
Spontaneous reporting is a valuable way to provide early detection for safety signals related to drug use.
Due to the increasing size of pharmacovigilance databases, data-mining and other automated methods for signal generation are more and more often used.
Even if these methods are very useful, they do not allow, for every particular association, an automated exploration of the multiple sources of confounding.
WHAT THIS STUDY ADDS
An association between angiotensin receptor blockers use and hypoglycaemia was found in the French pharmacovigilance database.
This signal disappeared after stratification on antidiabetic drug use, suggesting confounding by indication.
The association between hypoglycaemia and angiotensin receptor blocker use was actually less than expected in concomitant antidiabetic drug users.
Keywords: angiotensin receptor blockers, confounding, diabetes mellitus, pharmacoepidemiology, pharmacology, pharmacovigilance
Introduction
In the 1990s, sporadic reports raised the hypothesis that angiotensin converting enzyme inhibitors (ACEIs) might cause hypoglycaemia [1–4], seemingly confirmed by several studies [5, 6]. Safety signals mentioning the risk of hypoglycaemia with ACEIs were promulgated. However, the mechanism of ACEI-associated hypoglycaemia was never clearly demonstrated [7]. As ACEIs are generally prescribed in hypertension and could have a nephroprotective effect in diabetic patients, this association could also result from preferential prescribing of ACEIs to diabetic patients [8]. Other studies have seemed to support this [9, 10], although a specific risk with enalapril was suspected [10].
The indications and uses of angiotensin receptor blockers (ARBs) are similar to those of ACEIs. We therefore tested the French pharmacovigilance database for a signal of hypoglycaemia associated with ARBs, using the same methodology as used previously for ACEIs in a similar context [8].
Methods
The study used data from the French pharmacovigilance database from 1996 to 2005. Reports of hypoglycaemia were taken as cases, and other reports in the database as noncases.
The cases and noncases were examined for the presence of antidiabetic agents (ADAs), ARBs, drugs used as negative (diazepam) and positive controls (cibenzoline and disopyramide) for the association with hypoglycaemia [11–14] and drugs used in the same indication as ARBs (ACEIs, calcium antagonists, diuretics, atenolol).
Statistical analysis
Cases and noncases were identified from the spontaneous adverse drug reaction reporting database. Exposure was considered as the presence in a report of the drug of interest, whether or not it was suspected of causing the reaction [8]. For each drug of interest, reporting odds ratio (ROR: ratio of the odds of exposure in reports of cases and noncases) and their 95% confidence intervals (95% CI) were computed [15]. The analysis was first performed in the whole database and then separately in reports with or without mention of ADAs.
Results
Of the 174 595 reports corresponding to the study period, 807 were of hypoglycaemia.
Angiotensin receptor antagonists and other non-antidiabetic drugs and hypoglycaemia (Table 1)
Table 1.
Association of individual drugs with hypoglycaemia in the French pharmacovigilance database for other drugs (all reports)
| All reports | Hypoglycaemia | ROR* | 95% CI† | ||
|---|---|---|---|---|---|
| All reports | 174 595 | 807 | – | – | – |
| Any ARB‡ | 4 153 | 33 | 2 | 1 | 3 |
| Losartan | 1 421 | 12 | 2 | 1 | 3 |
| Irbesartan | 1 088 | 9 | 2 | 1 | 4 |
| Valsartan | 884 | 6 | 2 | 1 | 3 |
| Candesartan | 624 | 4 | 1.4 | 1 | 4 |
| Telmisartan | 124 | 2 | 4 | 1 | 14 |
| Eprosartan | 12 | 0 | 0 | – | – |
| Diazepam | 677 | 1 | 0.3 | 0.1 | 2 |
| Disopyramide | 218 | 16 | 17 | 10 | 29 |
| Cibenzoline | 180 | 57 | 107 | 78 | 148 |
| Captopril | 1 258 | 22 | 4 | 3 | 6 |
| Enalapril | 1 444 | 17 | 3 | 2 | 4 |
| Atenolol | 1 960 | 19 | 2 | 1 | 3 |
| Nicardipine | 1 393 | 13 | 2 | 1 | 4 |
| Nifedipine | 751 | 6 | 2 | 1 | 4 |
| Nitrendipine | 175 | 3 | 3 | 1 | 10 |
| Diltiazem | 1 612 | 12 | 2 | 1 | 3 |
| Verapamil | 1 032 | 7 | 2 | 1 | 3 |
| Frusemide | 7 839 | 93 | 3 | 2 | 4 |
| Diuretics‡ | 4 612 | 45 | 2 | 1 | 3 |
ROR, reporting odds ratio of association of selected drug with hypoglycaemia, compared with all reports.
95% CI, lower and upper limits of 95% confidence interval for OR.
Diuretics: thiazide and combination diuretics (cicletanine, hydrochlorothiazide, indapamide).
Association with any ARB in the complete database approximately doubled the overall risk of reporting hypoglycaemia. There was no clear difference between the ARBs (Table 1).
Diazepam, chosen as a negative control, was not associated with hypoglycaemia, whereas cibenzoline and disopyramide, chosen as positive controls, were.
Among the drugs sharing indications with ARBs, ACEIs (captopril or enalapril; ROR 3, 95% CI 2, 5), atenolol (ROR 2, 95% CI 1, 3), dihydropyridines (DHP) (ROR 2, 95% CI 1, 3), frusemide (ROR 3, 95% CI 2, 4) and thiazide diuretics (ROR 2.2, 95% CI 2, 3) were all associated with an increased risk of reporting hypoglycaemia in the whole database, whereas diltiazem or verapamil were not (Table 1).
Antidiabetic agents and hypoglycaemia
The ROR for hypoglycaemia with ADAs was 32 overall (95% CI 27, 37), ranging from about 4 for the glitazones (95% CI 0.5, 29) to 35 for insulin (95% CI 29, 43). It was 11 for carbutamide (95% CI 2, 85), 14 for acarbose (95% CI 10, 21), 18 for metformin (95% CI 14, 22), 21 for gliclazide (95% CI 16, 26), 32 for repaglinide (95% CI 21, 50), 46 for glibenclamide (95% CI 38, 56), 49 for glipizide (95% CI 31, 76) and 86 for glibornuride (95% CI 17, 446).
Association of studied drugs with ADAs
All the drugs studied except diazepam (ROR 0.7, 95% CI 0.4, 1) were associated with ADAs. Losartan was associated with ADA with an ROR of 5 (95% CI 4, 6), irbesartan with an ROR of 8 (95% CI 6, 9), captopril with an ROR of 7 (95% CI 6, 9), enalapril with an ROR of 7 (95% CI 6, 9), cibenzoline with an ROR of 4 (95% CI 2, 7), atenolol with an ROR of 4.4 (95% CI 4, 5), diuretics with an R0R of 6 (95% CI 5.5, 7), frusemide with an ROR of 5.4 (95% CI 5, 6) and DHP with an ROR of 6 (95% CI 5, 7).
Drugs other than ADAs and hypoglycaemia, according to ADA status
In reports mentioning ADAs (Table 2), none of the drugs was any longer associated with an increased risk of reporting hypoglycaemia except cibenzoline (ROR 5, 95% CI 2, 16). Actually, ARBs and dihydropyridine calcium channel blockers were associated with a reduced risk of reporting hypoglycaemia (ROR, respectively, 0.4, 95% CI 0.2, 0.8 and 0.5, 95% CI 0.3, 0.9).
Table 2.
Association of non-antidiabetic drugs with hypoglycaemia in the French pharmacovigilance database, in reports including or not including antidiabetic agents
| All reports | Hypoglycaemia | ROR* | 95% CI† | ||
|---|---|---|---|---|---|
| Reports including ADAs | 3 469 | 299 | – | – | – |
| ARB‡ | 275 | 11 | 0.4 | 0.2 | 0.8 |
| ACEI§ | 336 | 31 | 1 | 0.7 | 2 |
| Atenolol | 154 | 7 | 0.5 | 0.2 | 1 |
| DHP¶ | 248 | 12 | 0.5 | 0.3 | 0.9 |
| Frusemide | 666 | 48 | 0.8 | 0.6 | 1 |
| Cibenzoline | 13 | 4 | 5 | 2 | 16 |
| Reports not including ADAs | 171 126 | 508 | – | – | – |
| ARBs‡ | 2 234 | 10 | 2 | 0.8 | 3 |
| ACEIs§ | 2 366 | 8 | 1 | 0.6 | 2 |
| Atenolol | 1 806 | 12 | 2 | 1 | 4 |
| DHP¶ | 2 122 | 10 | 2 | 1 | 3 |
| Frusemide | 7 173 | 45 | 2.2 | 2 | 3 |
| Cibenzoline | 167 | 53 | 174 | 124 | 244 |
ROR, reporting odds ratio of association of selected drug with hypoglycaemia (see Table 2).
95% CI, lower and upper limits of 95% confidence interval for OR.
ARBs, angiotensin receptor blockers (losartan or irbesartan).
ACEIs, angiotensin converting enzyme inhibitor (enalapril or captopril).
DHP, dihydropyridines (nifedipine, nicardipine, nitrendipine, nimodipine).
In reports not mentioning ADAs (Table 2), atenolol, frusemide and cibenzoline were associated with reporting of hypoglycaemia, but not ARBs, ACEIs or DHP.
Discussion
A signal supporting an association between the use of ARBs and hypoglycaemia was generated in the French pharmacovigilance database considered as a whole. This signal disappeared after stratification on the presence of ADAs, thus suggesting confounding by indication. In fact, a decreased risk was found of reporting of hypoglycaemia in patients taking ADAs and ARBs.
Previous experience with reporting of hypoglycaemia with ACEIs had raised the possibility of confounding by indication [8], at a time when data-mining tools were not generally available. Such signals can now easily be generated by automated methods that are widely used for the detection of new adverse drug reactions. One important limitation of these methods is that they are not always able to investigate in depth potential confounding biases [16]. None explores systematically the existence of confounding by indication or channelling. This lack of adjustment could lead to false-positive signals that could modify clinical practice in inappropriate ways [17].
Our findings underline that confounding can lead to spurious disproportionality signals and suggest that special attention should be paid to eliminating the existence of such bias when considering a signal generated by automated methods using spontaneous reporting databases. They also underline the need for further research on automated signal generation, in order to develop adjustment methods that would include biases specific to pharmacoepidemiology, such as channelling or protopathic bias.
The authors thank all members of the 31 French pharmacovigilance centres as well as the AFSSaPS for the availability of the data. They also thank Philip Robinson for his help in manuscript preparation.
Competing interest: None declared.
REFERENCES
- 1.Mcmurray J, Fraser DM. Captopril, enalapril, and blood glucose. Lancet. 1986;1:1035. doi: 10.1016/s0140-6736(86)91304-8. [DOI] [PubMed] [Google Scholar]
- 2.Rett K, Wicklmayr M, Dietze GJ. Hypoglycemia in hypertensive diabetic patients treated with sulfonylureas, biguanides, and captopril. N Engl J Med. 1988;319:1609. doi: 10.1056/NEJM198812153192417. [DOI] [PubMed] [Google Scholar]
- 3.Washio M, Onoyama K, Makita Y, Fujishima M, Fujimi S. Hypoglycemia associated with the administration of angiotensin-converting enzyme inhibitor in a diabetic hemodialysis patient. Nephron. 1991;59:341–2. doi: 10.1159/000186585. [DOI] [PubMed] [Google Scholar]
- 4.Winocour P, Waldek S, Anderson DC. Captopril and blood glucose. Lancet. 1986;2:461. doi: 10.1016/s0140-6736(86)92169-0. [DOI] [PubMed] [Google Scholar]
- 5.Herings RM, de Boer A, Stricker BH, Leufkens HG, Porsius A. Hypoglycaemia associated with use of inhibitors of angiotensin converting enzyme. Lancet. 1995;345:1195–8. doi: 10.1016/s0140-6736(95)91988-0. [DOI] [PubMed] [Google Scholar]
- 6.Morris AD, Boyle DI, McMahon AD, Pearce H, Evans JM, Newton RW, Jung RT, MacDonald TM. ACE inhibitor use is associated with hospitalization for severe hypoglycemia in patients with diabetes. Diabetes Care. 1997;20:1363–7. doi: 10.2337/diacare.20.9.1363. [DOI] [PubMed] [Google Scholar]
- 7.Oltmanns KM, Deininger E, Wellhoener P, Schultes B, Kern W, Marx E, Dominiak P, Born J, Fehm HL, Peters A. Influence of captopril on symptomatic and hormonal responses to hypoglycaemia in humans. Br J Clin Pharmacol. 2003;55:347–53. doi: 10.1046/j.1365-2125.2003.01771.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Moore N, Kreft-Jais C, Haramburu F, Noblet C, Andrejak M, Ollagnier M, Begaud B. Reports of hypoglycaemia associated with the use of ACE inhibitors and other drugs: a case/non-case study in the French pharmacovigilance system database. Br J Clin Pharmacol. 1997;44:513–8. doi: 10.1046/j.1365-2125.1997.00615.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Akram K, Pedersen-Bjergaard U, Carstensen B, Borch-Johnsen K, Thorsteinsson B. Frequency and risk factors of severe hypoglycaemia in insulin-treated Type 2 diabetes: a cross-sectional survey. Diabet Med. 2006;23:750–6. doi: 10.1111/j.1464-5491.2006.01880.x. [DOI] [PubMed] [Google Scholar]
- 10.Thamer M, Ray NF, Taylor T. Association between antihypertensive drug use and hypoglycemia: a case–control study of diabetic users of insulin or sulfonylureas. Clin Ther. 1999;21:1387–400. doi: 10.1016/s0149-2918(99)80039-3. [DOI] [PubMed] [Google Scholar]
- 11.Goldberg IJ, Brown LK, Rayfield EJ. Disopyramide (Norpace)-induced hypoglycemia. Am J Med. 1980;69:463–6. doi: 10.1016/0002-9343(80)90020-0. [DOI] [PubMed] [Google Scholar]
- 12.Hayashi S, Horie M, Tsuura Y, Ishida H, Okada Y, Seino Y, Sasayama S. Disopyramide blocks pancreatic ATP-sensitive K+ channels and enhances insulin release. Am J Physiol. 1993;265:C337, 42. doi: 10.1152/ajpcell.1993.265.2.C337. [DOI] [PubMed] [Google Scholar]
- 13.Houdent C, Noblet C, Vandoren C, Levesque H, Morin C, Moore N, Courtois H, Wolf LM. Hypoglycemia induced by cibenzoline in the elderly. Rev Med Interne. 1991;12:143–5. doi: 10.1016/s0248-8663(05)81379-7. [DOI] [PubMed] [Google Scholar]
- 14.Ishida-Takahashi A, Horie M, Tsuura Y, Ishida H, Ai T, Sasayama S. Block of pancreatic ATP-sensitive K+ channels and insulinotrophic action by the antiarrhythmic agent, cibenzoline. Br J Pharmacol. 1996;117:1749–55. doi: 10.1111/j.1476-5381.1996.tb15349.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Rothman KJ, Lanes S, Sacks ST. The reporting odds ratio and its advantages over the proportional reporting ratio. Pharmacoepidemiol Drug Saf. 2004;13:519–23. doi: 10.1002/pds.1001. [DOI] [PubMed] [Google Scholar]
- 16.Strom BL. Pharmacoepidemiology. 4. Chichester: Wiley; 2005. [Google Scholar]
- 17.MacDonald TM, Morant SV, Goldstein JL, Burke TA, Pettitt D. Channelling bias and the incidence of gastrointestinal haemorrhage in users of meloxicam, coxibs, and older, non-specific non-steroidal anti-inflammatory drugs. Gut. 2003;52:1265–70. doi: 10.1136/gut.52.9.1265. [DOI] [PMC free article] [PubMed] [Google Scholar]