Education And Debate

Optimising drug treatment for elderly people: the prescribing cascade

BMJ 1997; 315 doi: (Published 25 October 1997) Cite this as: BMJ 1997;315:1096
  1. Paula A Rochon, assistant professor of medicinea,
  2. Jerry H Gurwitz, executive directorb
  1. a Division of Geriatric Medicine, Departments of Medicine and of Preventive Medicine and Biostatistics, University of Toronto, Baycrest Centre for Geriatric Care, 3560 Bathurst Street, North York, Ontario, Canada M6A 2E1
  2. b Meyers Primary Care Institute, 100 Central Street, Worcester, MA 01608, USA
  1. Correspondence to: Dr Rochon


    The most frequent medical intervention performed by a doctor is the writing of a prescription. Because chronic illness increases with advancing age, older people are more likely to have conditions that require drug treatment. Advanced age, frailty, and increased use of drugs are all factors that contribute to a patient's risk of developing a drug related problem. As many as 28% of hospital admissions in the United States of older people are as a result of drug related problems,1 up to 70% of which are attributed to adverse reactions to drugs.1 Creating optimal drug regimens that meet the complex needs of elderly people requires thought and careful planning.

    Inappropriate prescribing is expensive. In a recent study the costs of preventable adverse drug events—namely, injury resulting from a drug related medical intervention—occurring during a stay in hospital were estimated to be $2.8m (£1.75) annually in two large American teaching hospitals.2 The national cost of managing the consequences of inappropriate prescribing remains uncertain. One estimate has put the annual cost of drug related morbidity and mortality in outpatient clinics at $76.6bn.3 Drug related morbidity and mortality is an important area to target both to improve the quality of medical care for elderly people and to reduce the costs of health care for this population.

    A prescriber can do little to modify age related physiological changes in trying to minimise the likelihood that an older person will develop an adverse drug reaction. However, when assessing a patient who is already taking drugs, a doctor should always consider the development of any new signs and symptoms as a possible consequence of the patient's drug treatment. This article will focus on an under-recognised, and largely preventable drug related problem that we have termed the “prescribing cascade.” 4 The prescribing cascade begins when an adverse drug reaction is misinterpreted as a new medical condition. A drug is prescribed and an adverse drug effect occurs that is mistakenly diagnosed as a new medical condition. A new drug is prescribed, and the patient is placed at risk of developing additional adverse effects relating to this potentially unnecessary treatment (fig 1). Drawing prescribers' attention to this disturbing sequence of events may be an important step in minimising the occurrence of preventable adverse drug events associated with suboptimal prescribing decisions.

    Summary points

    The “prescribing cascade” cascade begins when an adverse drug reaction is misinterpreted as a new medical condition

    Another drug is then prescribed, and the patient is placed at risk of developing additional adverse effects relating to this potentially unnecessary treatment

    To prevent the prescribing cascade, doctors should always consider any new signs and symptoms as a possible consequence of current drug treatment

    Before any new drug treatment is started, the need for the drug should be re-evaluated and a non-drug treatment should be considered

    If drug treatment is necessary the lowest feasible dose of the drug should be used and alternative drugs with fewer adverse effects considered

    Non-steroidal anti-inflammatory drugs and starting antihypertensive treatment

    Non-steroidal anti-inflammatory drugs are among the most frequently prescribed drugs to elderly patients. An estimated 10-15% of people aged 65 years or older are prescribed such drugs.5 Their anti-inflammatory properties seem to result from their ability to inhibit cyclo-oxygenase, a critical enzyme in the biosynthesis of prostaglandins.6 Good evidence exists to suggest that prostaglandins have an important role in the modulation of two major determinants of blood pressure: vasoconstriction of arteriolar smooth muscle and control of extracellular fluid volume. The effects of non-steroidal anti-inflammatory drugs are most prominent in patients with existing hypertension.7

    The high prevalence of use of non-steroidal anti-inflammatory drugs among older people emphasises the importance of studying the clinical impact of these drugs on blood pressure in elderly people. To determine whether there is an increased risk associated with starting antihypertensive treatment in older people prescribed non-steroidal anti-inflammatory drugs (1), a case-control study was performed involving patients enrolled in the New Jersey Medicaid programme who were aged 65 years or older.8 Over 9000 patients who were newly started on an antihypertensive drug were compared with a similar number of randomly selected control patients. The adjusted odds ratio for starting antihypertensive treatment for recent users of non-steroidal anti-inflammatory drugs compared with non-users was 1.66 (95% confidence interval, 1.54 to 1.80). The odds ratio increased with increasing daily dose of the anti-inflammatory drug: compared with non-users, the adjusted odds ratio for users of low average daily doses was 1.55 (1.38 to 1.74), for medium dose users was 1.64 (1.44 to 1.87), and for high dose users was 1.82 (1.62 to 2.05). The conclusion was that the use of non-steroidal anti-inflammatory drugs may increase the risk associated with starting antihypertensive treatment in older people. Given the high prevalence of use of non-steroidal anti-inflammatory drugs by elderly people, this association could have important public health implications for the care of older patients.

    Examples of prescribing cascade

    View this table:

    This relation also shows a clear sequence of events where the use of one treatment leads to the start of a second that might have been avoided. Based on the findings of numerous epidemiological and clinical studies that have characterised the adverse consequences of use of non-steroidal anti-inflammatory drugs in older people, recommendations have been made to avoid using these agents when clinically feasible.5 As with other drugs prescribed to elderly patients, the most prudent approach is to limit prescribing non-steroidal anti-inflammatory drugs to situations in which benefits clearly outweigh risks and to use them only after potentially safer alternatives have been tried.11 Because of the multiple adverse effects attributable to these drugs, for some indications (such as osteoarthritis) treatments such as acetaminophen, gentle exercise, and weight reduction may be effective alternatives.12 13 14 When treatment with non-steroidal anti-inflammatory drugs is necessary, the lowest feasible dose should be used for the shortest time required to achieve the desired effect.

    Furthermore, if patients require extended treatment with non-steroidal anti-inflammatory drugs, periodic monitoring of blood pressure is warranted, as such treatment may contribute to newly detected rises in pressure. With recognition of this association between non-steroidal anti-inflammatory drugs and rises in blood pressure, the starting or intensifying of antihypertensive treatment may be avoided.

    Thiazide diuretics and starting treatment for gout

    The development of some degree of hyperuricaemia is a well documented side effect of treatment with thiazide diuretics.15 16 17 18 Population based studies have shown an association between hyperuricaemia and the development of gout. For example, data from the Framingham study document a cumulative incidence of gout of 36% over 12 years in patients with serum uric acid concentrations >476 μmol/l, compared with less than 3% in those with lower concentrations.19 The occurrence of hyperuricaemia that has been induced by thiazide diuretics raises some important issues about the use of these diuretics in elderly people. Ample data show the efficacy of these agents in treating hypertension in elderly patients and in preventing major sequelae such as stroke—data that are absent for many other commonly used antihypertensive drugs.20 21 22 The impact of thiazide diuretics on serum uric acid concentrations, however, raises questions about whether this treatment may precipitate the use of additional drugs.

    This question was recently examined in a retrospective cohort study of 9249 patients enrolled in the New Jersey Medicaid programme aged 65 or older who had been started on a variety of antihypertensive agents.9 None of the patients in the cohort had previously used treatment for gout (allopurinol, colchicine, or uricosuric agent). Follow up extended for up to two years, and exposure to antihypertensive drugs was characterised over this period according to the following categories: thiazide diuretics alone; non-thiazide antihypertensive drugs alone; thiazide diuretics combined with any non-thiazide antihypertensive drug; and no use of antihypertensive drugs. The relative risk for starting treatment for gout was 1.00 (0.65 to 1.53) for non-thiazide antihypertensive drugs alone, 1.99 (1.21 to 3.26) for thiazide diuretics alone, and 2.29 (1.55 to 3.37) for thiazide diuretics combined with any non-thiazide drug. Risk for starting treatment for gout was significantly increased for thiazide doses of 25 mg/day (in hydrochlorothiazide equivalents) or more; no significant increase in risk was seen for lower doses. It was concluded that the use of thiazide diuretics in doses of ≥25 mg/day was associated with a significantly increased risk for starting treatment for gout, relative to antihypertensive regimens that did not include the use of a thiazide diuretic.

    Considerable evidence supports the efficacy of low doses of thiazide diuretics in the treatment of hypertension in elderly people.13 14 15 The dose-response relations found in this study support the use of lower doses of thiazide diuretics when treatment is indicated. Although the recommendations of the United States's joint national committee on detection, evaluation and treatment of high blood pressure suggest starting antihypertensive treatment at low doses in all patients,17 thiazide diuretics are commonly started at doses that extend well beyond the low dose range. Low doses of thiazide diuretics—for example, 12.5 mg of hydrochlorothiazide—often produce as large an antihypertensive effect as larger doses, with a reduced risk of metabolic abnormalities. In fact, evidence exists that a dose of hydrochlorothiazide as low as 6.25 mg can be as efficacious in treating hypertension in many older patients, when combined with a low dose of another antihypertensive drug.23 24 25 When hyperuricaemia does occur during treatment with a thiazide diuretic, clinicians should bear in mind that asymptomatic hyperuricaemia alone does not warrant treatment.

    Use of metoclopramide when starting levodopa treatment

    Metoclopramide hydrochloride is widely used in the treatment of gastro-oesophageal reflux, in the management of disorders of gastric emptying including diabetic gastroparesis, and as an antiemetic after chemotherapy. Its antidopaminergic adverse effects, including unwanted extrapyramidal signs and symptoms, have long been recognised. Such drug induced symptoms in older people can be misinterpreted as indicating a new disease or be attributed to the aging process itself. This misinterpretation is particularly likely when the symptoms are indistinguishable from an illness that is seen more often in older people, such as Parkinson's disease.26 27

    A case-control study, again involving patients enrolled in the New Jersey Medicaid programme aged 65 years or older, showed that patients taking metoclopramide were three times more likely to begin using a drug containing levodopa than patients not taking metoclopramide (odds ratio 3.09 (2.25 to 4.26)).10 The risk increased with increasing daily metoclopramide dose: the odds ratio was 1.19 (0.50 to 2.81) for ≤10 mg/day, 3.33 (1.98 to 5.58) for >10 mg/day to 20 mg/day, and 5.25 (1.16 to 8.50) for >20 mg/day (fig 2). In summary, metoclopramide confers an increased risk of starting treatment generally reserved for managing idiopathic Parkinson's disease. Such multiple prescribing may represent the misdiagnosis of Parkinson's disease in patients with drug induced parkinsonian symptoms.

    Fig 2
    Fig 2

    Adverse effects of drugs that are related to dose, leading to prescribing cascade. Top: Odds ratio for starting treatment with antihypertensive drugs based on dose of non-steroidal anti-inflammatory drug (NSAID).8 Centre: Relative risk for starting treatment for gout based on dose of thiazide.9 Bottom: Odds ratios for starting levodopa treatment based on dose of metoclopramide.10


    The prescribing cascade, whereby additional drug treatment is started after a patient develops an adverse reaction to a drug, is largely preventable by carefully considering whether any new medical condition might be the result of an existing drug treatment. The prescription of a new drug specifically to treat an adverse drug effect should be considered the choice of last resort in the care of older patients. More prudent strategies include:

    • Carefully re-evaluating the absolute need for the offending agent;

    • Using non-pharmacological treatment for managing a patient's medical condition;

    • Reducing the dosage of the implicated drug to the lowest feasible dose that is effective in treating a patient's medical condition; and

    • Considering alternative drugs that might be safer in terms of the risk of adverse effects in older people.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    14. 14.
    15. 15.
    16. 16.
    17. 17.
    18. 18.
    19. 19.
    20. 20.
    21. 21.
    22. 22.
    23. 23.
    24. 24.
    25. 25.
    26. 26.
    27. 27.