Case scenario

Natasha, a 23-year-old woman, presents to the pharmacy to speak to the pharmacist following a recent cold. She has nasal congestion, sinus tenderness, dental pain, fever (38.5 °C) and general malaise. Natasha has been using a non-prescription saline sinus flush with no apparent effect. Her friend Molly had similar symptoms months ago and told Natasha a full course of antibiotics solved her problems straight away. A full-time lawyer, Natasha says she must return to work as soon as possible and would like the quickest solution. She asks if antibiotics would be appropriate to alleviate her symptoms. 

Learning objectives

After successful completion of this CPD activity, pharmacists should be able to:

  • Describe the statistical measures used in medical literature, number needed to treat, number needed to harm, number needed to screen and number unnecessarily treated
  • Identify statistical measures used to calculate number needed to treat
  • Interpret statistical measures reported in medical literature to inform practice and patient care
  • Communicate risk to patients using number needed to treat pictographs. 

Competency standards (2016) addressed: 1.5, 2.3, 3.5 ,5.3

Already read the CPD in the journal? Scroll to the bottom to SUBMIT ANSWERS.  

Introduction

As health professionals, pharmacists are inundated with new information from the thousands of articles published every year, creating an ever-expanding knowledge bank. Before we can effectively communicate research evidence to our patients, we must first be familiar with the ways that study results are presented. The number needed to treat (NNT) is a simple representation that compares the efficacy of different therapeutic interventions. The effects of new drugs described in randomised controlled trials (RCTs) and systematic reviews are usually based on dichotomous outcomes such as survival vs death, or cure vs no cure. The probabilities generated from these studies can be used to calculate a number of statistical values (i.e. NNT, relative risks, odds ratios, hazard ratios and absolute risk reductions) that can assist pharmacists to make informed decisions when evaluating treatment benefits or harms. But how exactly do we interpret the NNT? And what do we mean by risks and odds? 

Statistical measures

Risks and odds

The terms ‘risks’ and ‘odds’ are often used interchangeably to define the likelihood or chance of an event happening.1 In statistics, the two terms have specific meanings and are calculated in different ways. As such, they are not synonymous and should not be used interchangeably. 

Risk is a concept that appears more familiar to patients and health professionals.1 It is the probability of an event occurring as a proportion of all possible outcomes.2 If we were to apply risk to rolling a six-sided die, there is a 1 in 6 (16.7%) chance of rolling any particular number. Building on risks, RCTs or cohort studies tend to report relative risks or risk ratios (RR). The RR of an event is the likelihood of its occurrence after being exposed to a risk variable (e.g. drug substances, surgery, other interventions), as compared with the likelihood of its occurrence in a control group (e.g. placebo, gold-standard therapy).3 

Odds is a concept more commonly used to describe gambling stakes.1 It is the probability of an event occurring, divisible by the probability of the event not occurring.2 Hence, the odds of rolling any number from a six-sided die is 1 in 5 (20%). Case–control studies will typically report study results as an odds ratio (OR). The OR is the ratio of the odds of an event occurring in the exposed group, compared to the odds of occurrence in the unexposed group.4

  • RR or OR = 1 means that there is no difference in risk or odds between the groups being compared. 
  • RR or OR >1 means there is an increased risk or odds among those who were exposed compared to unexposed.
  • RR or OR <1 means there is a decreased risk or odds among those who were exposed compared to unexposed.

Number needed to treat

The NNT is defined as the number of patients who need to be treated by an intervention (e.g. drug, therapy, surgery) over a defined time period to achieve one extra beneficial outcome, as compared to an alternative intervention or control. For example, a Cochrane review found aldosterone antagonists (i.e. eplerenone or spironolactone) reduced death in patients with chronic kidney disease requiring dialysis, reporting an NNT of 14 (95% Confidence Interval (CI) 10–21).5 This means that for every 14 patients using aldosterone antagonists in dialysis-dependent chronic kidney disease for approximately 1.2 years, one patient avoids death, compared to placebo.5 The number needed to benefit (NNTB) is used interchangeably with the NNT in some medical texts.

Alongside the NNT, a few other absolute risk figures have been derived from the clinical literature. These include the number needed to harm (NNH), the number needed to screen (NNS), and the number unnecessarily treated (NUT). 

These measures are likewise bound to a specified time period and are compared to a control intervention (placebo or gold-standard therapy).

Number needed to harm 

The NNH defines the number of patients who need to be treated by an intervention (e.g. pharmacotherapy or medical procedure) for one patient to be harmed. It reflects the adverse effects or undesirable outcomes associated with treatment options. In context, the NNH associated with using aldosterone antagonists for dialysis-dependent chronic kidney disease was 38 (95% CI 26–68), meaning for every 38 patients taking either spironolactone or eplerenone for approximately 1.2 years, one person will develop gynaecomastia as a harmful outcome.5 

Number needed to screen

The NNS refers to the number of patients needed to screen in order to prevent a death or harmful outcome. 

Its applications pertain to areas of epidemiological research; for example, to estimate the number of women needed to screen to prevent one breast cancer death. A study found that 84 women aged 40–84 need to be screened annually (mammography) to save one life from breast cancer.6 

Number unnecessarily treated

The NUT quantifies the number of patients who do not receive the beneficial outcome of treatment.7 This measure is inversely related to the NNT. 

As an example, let us consider patients who are treated with high-dose statins after hospitalisation to prevent secondary myocardial infarction. Assuming that the absolute difference between the treatment and control arms in developing a second heart attack is 20% implies an NNT of 5. This means if five patients need to be treated to prevent one extra myocardial infarction episode, the remaining four will undergo high-dose statin without benefit; such is the concept of the NUT.7

Absolute measures of risk

Absolute risk is simply defined as the chance or probability of any event occurring in a group.8 

The absolute risk reduction (ARR) is the difference between the basic probability of an event occurring in the intervention group (EER) versus its occurrence in the control group (CER) of a RCT.8

ARR = CER – EER

(EER = experimental event group event rate, CER = control group event rate)

It differs from relative risk reduction (RRR), which express the risk difference between the intervention and control groups, as a proportion of the risk in the control group.8              

 RRR = CER – EER

             CER 

Most RCTs preferentially report RRRs over ARRs or NNTs because communication of relative terms tends to overstate the effectiveness of a treatment.9 The CURE study10 highlighted that patients taking a combination of clopidogrel and aspirin had a 20% RRR for heart attack, cardiovascular death or stroke compared to an ARR of only 2.1%. Drug company representatives promoting the clopidogrel/aspirin combination may prefer to quote the 20% risk reduction, obscuring the perceived benefit of the medication(s). Figures like the ARR or NNT present more concrete information about an intervention by incorporating both the baseline risk without treatment and the magnitude of the risk reduction.9 Treatment effectiveness is often perceived to be lower when quoting ARRs or NNTs, which may explain why these values are typically under-reported even among top-cited journals.9,11 

Box 1 – Key resources

  • NPS MedicineWise provides pharmacists with resources including:
    • Practice-specific introduction to risks and their relevance to the patient – Evidence, risk and the patient.12
    • Step-by-step calculations of absolute and relative risks – Interpreting risks and ratios in therapy trials.8
  • Cochrane Training resources for health professionals provide a generalised summary on interpreting outcome data – Chapter 6: Choosing effect measures and computing estimates of effect. At: https://training.cochrane.org/handbook/current 
  • The NNT Group is a physician-run database of evidence-based medicine (EBM) summaries, specialising in use of the NNT to evaluate and rate therapies. At: www.thennt.com
  • Visual Rx is an online EBM application that generates pictographs from raw research
    data (ORs, RRs, event rates and confidence intervals). At:     www.nntonline.net

Calculating the NNT

Many statistical terms are used to measure and showcase the results of therapy trials. An understanding of figures such as event rates, absolute risk, odds ratio, relative risk and risk reductions is crucial to the calculation and interpretation of the NNT (Box 1). These values can help health practitioners at a basic level to determine whether novel treatments are superior to other treatments or placebo. 

The NNT is calculated by taking the reciprocal of the ARR 1/ARR and is rounded up to the nearest whole number (see Table 1). Note that the NNH is likewise calculated in this way but is rounded down to the nearest whole number.

Interpreting the NNT

The NNT appears in the medical literature, and so pharmacists should be confident to interpret it. The NNT is typically quoted as a whole number (NNT = n), but can also be expressed as a simple frequency (1 in n), a comparative measure (n more patients will benefit) or as a pictograph (Cates Plot13).14 When communicating risk to patients, using pictures to visually demonstrate potential treatment outcomes can have more impact than simply quoting an NNT.15 

Time should also be considered in the context of NNTs, RRs and ORs. Risk is spread across time, and a treatment that doubles the ‘risk’ of 10-year post-myocardial infarction survival may be more desirable than a treatment which doubles the same ‘risk’ at 10 months.2 

Pharmacists can develop and use NNTs generated from multiple research papers to compare the effectiveness of different treatments. The quality of evidence should be similar in both cases, accounting for study participants, timeframe and internal validity (blinding, events outside of the study). For instance, high-dose pregabalin for post-herpetic neuralgia (PHN) has an NNT of 3.9 (95% CI 3.1–5.5) for 50% pain reduction and an NNH of 7.1 (95% CI 5.3–11) for adverse effects causing study withdrawal (somnolence, dizziness).16,17 

High-dose gabapentin for the same indication has an NNT of 6.7 (95% CI 5.4–8.7) for 50% pain reduction and an NNH of 30 (20.0–66.0) for adverse effects causing study withdrawal (somnolence, peripheral oedema, gait disturbance, dizziness).16,18 

A low NNT means that more people will receive the potential ‘benefits’ of the treatment. A high NNH means that fewer persons will receive the potential ‘harms’ of the treatment. Pregabalin has a lower NNT compared to gabapentin, suggesting it may be better for PHN pain control. Gabapentin has a greater NNH compared to pregabalin, which may indicate an increased frequency of adverse effects. 

Is the NNT statistically significant?

If there is a statistically significant difference in the outcomes of the treatment and control groups of therapy trials, the observed difference is unlikely to have occurred by chance. The difference is then meaningful to patient care. If NNT is to be incorporated into medical decision-making, it must be a statistically significant outcome. 

There are two main ways in which studies represent statistical significance: p-values and confidence intervals. Either should always accompany the NNT in the clinical literature.

P-values

Standard convention deems a p-value (probability value) of less than 0.05 as statistically significant.19 This means there is a less than 5% probability that the study results occurred by chance alone, and the difference between the two groups is likely caused by the intervention.19 

  • p<0.05 is generally considered as statistically significant.
  • p≥0.05 is generally considered not statistically significant. 

Confidence intervals

How confident can a researcher be that the results for the sample population represent those for the entire population? In research, this is expressed as a confidence interval (CI), which refers to the interval or range in which the true value for the entire population (known as the target population) lies. A confidence level of 95% is usually selected for medical research. A 95% CI means that if we were to replicate any study 100 times exactly, then 95 of the generated confidence intervals would contain the true effect size (mean, relative risk, odds ratio, hazards ratio).19 Hence, a 95% CI of any single study has a 95% probability of the true effect size sitting between the upper and lower ranges given.19 For dichotomous outcomes (i.e. survival vs death, disease vs disease-free), 1 is the null value of the effect size. This contrasts with continuous outcomes (i.e. blood pressure, total cholesterol, height, weight) where the null value is set at 0. If a 95% CI includes the null value, then there is no statistically meaningful difference between the groups. A smaller 95% CI range also indicates greater precision in the point estimate of effect.

  • A 95% CI that does not contain 1 means the effect size is statistically significant (dichotomous data)
  • A 95% CI that does not contain 0 means the effect size is statistically significant (continuous data).

Table 1 – How to calculate the ARR, RRR and NNT

CER = control event rate, EER = experimental event rate

Informing patients about risk

We can use the NNT to present research evidence to patients, with a few considerations. The ‘external validity’ or degree to which the results of studies can be generalised to populations outside of the study is a key parameter.20 It is important to know that if a healthcare intervention works under ideal conditions (i.e. efficacy), it is also effective in populations and settings outside of the clinical study (i.e. effectiveness).20 

Pharmacists should contemplate whether data gathered from study participants is applicable to everyday patients. This involves considering the baseline characteristics (i.e. age, sex, comorbidities), confounding factors, inclusion and exclusion criteria, or type of analysis (i.e. intention to treat, per protocol).20

Studies should also specify a timeframe from which the benefits or risks of treatments are derived. For instance, in the earlier study, women were randomly assigned to receive either oral tamoxifen 20 mg daily or placebo for 5 years. When communicating the evidence of tamoxifen therapy, pharmacists should recognise that treatment outcomes may rely on continual medication compliance over a longer timeframe. The benefits may not be immediately apparent in the short term, and this should be emphasised in discussions of the treatment plan.

Box 2 – Risk pictographs

  • BMJ Infographics is a new initiative by the British Medical Journal which includes reviewed evidence summaries of recently published journal articles in an infographic form. These are not validated decision aids. At: www.bmj.com/infographics 
  • NNT Choice Aids are newly designed decision aids as part of a pharmacist-piloted research project developed at the University of Canberra. 
  • Visual Rx by Dr Chris Cates is an online tool that can generate a visual representation of the risk/benefit likelihoods from raw data. At: www.nntonline.com/visualrx

Using NNT pictographs

Pictographs can be used to communicate the incremental benefits of risk-reducing treatments to patients in ways that are easy to interpret (see Figure 1).21 Icon arrays are derivations of NNTs, where one shape (e.g. smiley faces, human figures) is repeated a number of times and some shapes are altered (change of colour or shape) such that they represent a clear proportion. For example, in Figure 1, the 5-year NNT for rosuvastatin for myocardial infarction, stroke, revascularisation or death was 20 (95% CI 14–34).22 This means that 20 patients need to take rosuvastatin for 5 years in order for one patient to avoid a cardiovascular outcome. One in 20 (5%) of the icons are coloured blue, representing those who are protected at 5 years from a cardiovascular event. The other 95% (19/20) remain grey, representing those who do not receive this benefit after 5 years. Multiple colours or patterns can also be used to describe the risk of developing adverse effects from medications. Pictographs have the potential to improve patients’ understanding of health information, including treatment options and medication adverse effects.21 Studies have also indicated the promising effects of such aids on health-related outcomes (i.e. quality of life and medical decision-making).21 Going forth in their practice, health practitioners may find a number of resources useful to generate their own or utilise ready-made medical infographics (see Box 2).

Case scenario continued

Natasha has acute sinusitis following her acute respiratory tract infection (common cold). COVID-19 and viral rhinosinusitis has been excluded. To provide an evidence-based answer, you look in the Therapeutic Guidelines, which indicates acute bacterial rhinosinusitis is generally self-limiting, and antibiotics make little difference to the course of illness.23 To visually illustrate the risk-benefit comparison of antibiotics for rhinosinusitis to Natasha, you conduct a search using the Cochrane Library database and extrapolate the data from a systematic review on ‘antibiotics for acute rhinosinusitis in adults’.24 The systematic review included 15 trials (n = 3,057) and the meta-analysis had low risk of bias.24 The study reported that antibiotics could shorten time to cure compared to no antibiotics (OR 1.25, 95% CI 1.02–1.54; NNTB 19, 95% CI 10–205; high-quality evidence). There were also statistically significant findings for  adverse effects with antibiotics (OR 2.21, 95% CI 1.74–2.82, NNTH 8, 95% CI 6–12; high-quality evidence).24 

You extrapolate the significant data into a 100-icon pictograph and discuss the findings with Natasha. You explain that if 100 people were to take antibiotics for rhinosinusitis, only five would achieve cure faster than those who did not take antibiotics. This means 95 persons would take antibiotics with no apparent benefit. You explain further that the risk of taking antibiotics for rhinosinusitis is that 13 more people will experience adverse effects with antibiotics compared to those people without antibiotics. In this case, the risk outweighs the benefit.

You counsel Natasha to get rest (stay at home and limit travelling to public spaces), drink lots of water, take paracetamol to reduce fever and manage pain, and use an oxymetazoline nasal spray for congestion over the next 3 days. Symptoms of infective rhinosinusitis generally resolve or reduce in severity within 7–14 days. You also advise Natasha to see her general practitioner if her symptoms do not improve or worsen (i.e. nasal obstruction, purulent discharge, fever) in 5 days, as antibiotics may be more beneficial. 

Conclusion 

With the upsurge of new study trials and research, pharmacists and health professionals are continuously exposed to the mathematical language used to describe treatment effects. Such terminology is often synonymous with other terms and may perpetuate a general feeling of confusion or bafflement towards approaching research statistics. An understanding of absolute risk, relative risk, odds ratio, risks ratio, NNT, NNH and statistical significance is crucial in evaluating whether a new treatment has any advantage over standard therapy or placebo. With many universities adopting statistical analysis or evidence-based medicine in their pharmacy schools, pharmacists are well-equipped with the necessary research appraisal skills to acquire, assess and convey information to patients. As the most accessible primary health providers, there is also opportunity for knowledge translation to occur outside of a single patient encounter, as pharmacists are more likely to see patients on a regular basis. Pharmacists should be prepared to answer questions from prescribers and health consumers about the effectiveness or appropriateness of medications using evidence-based practices and manipulate the data into easier to understand pictographs. 

Key points

  • Communicating research evidence involves an understanding of risks, odds, statistical terminology and statistical significance.
  • The number needed to treat (NNT) and number needed to harm (NNH) allow an accurate comparison of the risks and benefits of medications.
  • Pictographs can be used to illustrate NNTs and NNHs to patients and better promote shared decision-making.

This article is accredited for group 2 CPD credits. Click submit answers to complete the quiz and automatically record CPD against your record.

SUBMIT ANSWERS

If you do get an enrolment error, please click here

References

  1. Higgins JPT, Thomas J, Chandler J, et al (editors). Cochrane Handbook for Systematic Reviews of Interventions version 6.2 (updated February 2021). Cochrane, 2021. At: training.cochrane.org/handbook
  2. Ranganathan P, Aggarwal R, Pramesh CS. Common pitfalls in statistical analysis: odds versus risk. Perspect Clin Res 2015;6:222–4. https://doi.org/10.4103/2229-3485.167092
  3. Andrade C. Understanding relative risk, odds ratio, and related terms: as simple as it can get. J Clin Psychiatry 2015;76:e857–61. https://doi.org/10.4088/JCP.15f10150
  4. Bushell M. Supporting your practice: evidence-based medicine. Australian Pharmacist 2019;38(3):46–55.
  5. Hasegawa T, Nishiwaki H, Ota E, et al. Aldosterone antagonists for people with chronic kidney disease requiring dialysis. Cochrane Database of Systematic Reviews 2021, Issue 2. https://doi.org/10.1002/14651858.CD013109.pub2
  6. Hendrick RE, Helvie MA. Mammography screening: a new estimate of number needed to screen to prevent one breast cancer death. AJR Am J Roentgenol. 2012;198(3):723–8. https://doi.org/10.2214/ajr.11.7146
  7. Legemate DA, Koelemay MJW, Ubbink DT. Number unnecessarily treated in relation to harm: a concept physicians and patients need to understand. Ann Surg 2016;263(5):855–6.
  8. Scott I. Interpreting risks and ratios in therapy trials. Aust Prescr 2008;31:12–6.
  9. Duncan B, Ables AZ. Do drug treatment poems report data in clinically useful ways? J Fam Pract 2013;62:E1–5.
  10. Yusuf S, Zhao F, Mehta SR, et al. Effects of clopidogrel in addition to aspirin in patients with acute coronary syndromes without st-segment elevation. N Engl J Med 2001;345(7):494–502. https://doi.org/10.1056/NEJMoa010746
  11. Elliott MH, Skydel JJ, Dhruva SS, et al. Characteristics and reporting of number needed to treat, number needed to harm, and absolute risk reduction in controlled clinical trials, 2001–2019. JAMA Intern Med 2021;181(2):282–4. https://doi.org/10.1001/jamainternmed.2020.4799
  12. Neeskens P. Evidence, risk and the patient. Aust Prescr 2007;30:47–50. https://doi.org/https://doi.org/10.18773/austprescr.2007.025
  13. Cates C. Visual Rx. 2021 At: www.nntonline.net/visualrx/
  14. Saver JL, Lewis RJ. Number needed to treat: conveying the likelihood of a therapeutic effect. JAMA 2019;321(8):798–9. https://doi.org/10.1001/jama.2018.21971
  15. Nguyen C, Naunton M, Thomas J, et al. Availability and use of number needed to treat (NNT) based decision aids for pharmaceutical interventions. Exploratory Research in Clinical and Social Pharmacy. 2021;2:100039. https://doi.org/https://doi.org/10.1016/j.rcsop.2021.100039
  16. Mathieson S, Lin C-WC, Underwood M, et al. Pregabalin and gabapentin for pain. BMJ 2020;369:m1315. https://doi.org/10.1136/bmj.m1315
  17. Derry S, Bell RF, Straube S, et al. Pregabalin for neuropathic pain in adults. Cochrane Database Systematic Reviews 2019, Issue 1. https://doi.org/10.1002/14651858.CD007076.pub3
  18. Wiffen PJ, Derry S, Bell RF, et al. Gabapentin for chronic neuropathic pain in adults. Cochrane Database of Systematic Reviews 2017, Issue 6. https://doi.org/10.1002/14651858.CD007938.pub4
  19. Andrade C. The p value and statistical significance: misunderstandings, explanations, challenges, and alternatives. Indian J Psychol Med 2019;41:210–5. https://doi.org/10.4103/IJPSYM.IJPSYM_193_19
  20. Khorsan R, Crawford C. How to assess the external validity and model validity of therapeutic trials: a conceptual approach to systematic review methodology. Evid Based Complement Alternat Med 2014;2014:694804. https://doi.org/10.1155/2014/694804
  21. Wang T, Voss JG. Effectiveness of pictographs in improving patient education outcomes: a systematic review. Health Education Research. 2021;36:9–40. https://doi.org/10.1093/her/cyaa046
  22. Ridker Paul M, MacFadyen Jean G, Fonseca Francisco AH, et al. Number needed to treat with rosuvastatin to prevent first cardiovascular events and death among men and women with low low-density lipoprotein cholesterol and elevated high-sensitivity C-reactive protein. Circ Cardiovasc Qual Outcomes 2009;2:616–23. https://doi.org/10.1161/CIRCOUTCOMES.109.848473
  23. Therapeutic Guidelines. Acute rhinosinusitis. 2019. At: https://tgldcdp.tg.org.au.
  24. Lemiengre MB, van Driel ML, Merenstein D, et al. Antibiotics for acute rhinosinusitis in adults. Cochrane Database of Systematic Reviews 2018, Issue 9. https://doi.org/10.1002/14651858.CD006089.pub5

Pictograph References: Figure 1 – NNT for the Top 10 PBS/RPBS drugs by total script count

Rosuvastatin Ridker Paul M, MacFadyen Jean G, Fonseca Francisco AH, et al. Number needed to treat with rosuvastatin to prevent first cardiovascular events and death among men and women with low low-density lipoprotein cholesterol and elevated high-sensitivity c-reactive protein. Circ Cardiovasc Qual Outcomes 2009;2(6):616–23.
Atorvastatin Sever PS, Dahlöf B, Poulter NR, et al. Prevention of coronary and stroke events with atorvastatin in hypertensive patients who have average or lower-than-average cholesterol concentrations, in the Anglo-Scandinavian Cardiac Outcomes Trial–Lipid Lowering Arm (ascot-lla): a multicentre randomised controlled trial. Lancet 2003;361(9364):1149–58.
Pantoprazole Pilotto A, Leandro G, Franceschi M. Short- and long-term therapy for reflux oesophagitis in the elderly: a multi-centre, placebo-controlled study with pantoprazole. Aliment Pharmacol Ther 2003;17(11):1399–406.
Esomeprazole Gralnek IM, Dulai GS, Fennerty MB, et al. Esomeprazole versus other proton pump inhibitors in erosive esophagitis: a meta-analysis of randomized clinical trials. Clin Gastroenterol Hepatol 2006;4(12):1452–8.
Perindopril Campbell DJ. A review of perindopril in the reduction of cardiovascular events. Vasc Health Risk Manag 2006;2(2):117–24.
Cefalexin Rosengren H, Heal CF, Buttner PG. Effect of a single prophylactic preoperative oral antibiotic dose on surgical site infection following complex dermatological procedures on the nose and ear: a prospective, randomised, controlled, double-blinded trial. BMJ Open 2018;8(4):e020213.
Metformin O’Connor PJ, Spann SJ, Woolf SH. Care of adults with type 2 diabetes mellitus. A review of the evidence. J Fam Pract 1998;47(5 Suppl):S13–22.
Escitalopram Montgomery S, Hansen T, Kasper S. Efficacy of escitalopram compared to citalopram: a meta-analysis. Int J Neuropsychopharmacol 2011;14(2):261–8.
Amoxicillin Little P, Stuart B, Moore M, et al. Amoxicillin for acute lower-respiratory-tract infection in primary care when pneumonia is not suspected: a 12-country, randomised, placebo-controlled trial. Lancet Infect Dis 2013;13(2):123–9.
Sertraline Cipriani A, Furukawa TA, Geddes JR, et al. Does randomized evidence support sertraline as first-line antidepressant for adults with acute major depression? A systematic review and meta-analysis. J Clin Psychiatry 2008;69(11):1732–42.
Top 10 drugs Top 10 drugs 2019–20. Aust Prescr 2020;43:209.

CASSANDRA NGUYEN BPharm, Dr MARY BUSHELL BPharm (Hons), AACPA, GCTLHE, AFACP, MPS, PhD, LYN TODD BPharm, Dr JACKSON THOMAS BPharm, MPharmSc, PhD, Professor MARK NAUNTON BPharm (Hons), PhD, Head, School of Health Sciences, University of Canberra.

RELATED ARTICLESMORE FROM AUTHOR