Patient Compliance – Sweetening the pill

“Compliance is the degree to which a patient is compliant with the instructions that are given by a healthcare professional and written on the medication label (for example, prescribed dose and time schedule).” (p.8 – I didn’t know that definition before reading the book so it made sense to me to start out with this quote, to make sure people are aware of what this book is about.)

It’s an interesting book with a lot of stuff I didn’t know and/or at the very least hadn’t thought about. A couple of the chapters were quite weak and I basically skipped most of chapter 6, which was written by a pharmaceutical marketing consultant who wrote about branding stuff which I couldn’t care less about – but most of the book was quite good. One of the chapters (chapter 8) very surprisingly included undocumented claims which were to some extent proven wrong in a previous chapter (chapter 3) – it seemed as if the authors of that chapter had not read the previous chapter in question. Here’s what they wrote at the very beginning of their chapter (chapter 8):

“Compliance is important. Better adherence to treatment regimes leads to less healthcare resource utilization overall, as fewer illness recurrence or medication errors leading to side-effects take place.” (p.109)

And here’s what Dr. Dyffrig Hughes told us in chapter 3:

From the studies evaluated, the direction and magnitude of the change in costs and consequences resulting from applying sensitivity analysis to the compliance rate was measured and taken as an indicator of the impact of non-compliance. There was consistency among studies, in that as compliance decreased (whatever the measure), the [health] benefits also decreased […] There is no consistency, however, in the direction of change in costs resulting from changes in compliance [my bold, US] […] Whilst some studies show that costs increase as compliance decreases, others showed the opposite trend. This difference did not appear to be related to the nature of the disease, the measure of non-compliance or the assumptions relating to the health benefits experienced by non-compliers.

And here’s even a figure illustrating this point:


A little more from chapter 3 on the same subject: “The economic evaluations described demonstrate that medical expenditures do not always increase because of poor compliance. However, the limitations in the methodology adopted in many of the studies would suggest that the reported changes in healthcare expenditure may not necessarily be observed in practice. It is difficult, therefore, to predict the true economic impact of non-compliance with drug therapy, particularly as evidence relating to discontinuers is often not reported. It is the case, however, that decisions on optimal treatments, based on economic criteria, are influenced by non-compliance […] Health economic evaluations often fail to include non-compliance with medications. As a significant proportion of evaluations are based on efficacy trials, attention should be given to how their findings might be generalized. In particular, as poor compliance is one of the most important elements responsible for the differences that may exist between the effectiveness and efficacy of an intervention, greater consideration should be given to compliance when generalizing from the results of a controlled clinical trial. An optimal cost-effective treatment strategy chosen on the basis of efficacy data may not be so attractive once real-world compliance figures are taken into account.”

I don’t consider this to be an unforgiveable error in a book like this with a lot of authors writing about different aspects of the problem, but it doesn’t help that the authors of chapter 8 repeat the claim that improved compliance will have cost-saving effects in their conclusion of the chapter as well, and at the very least it doesn’t make them look good to me (a more cautious and tentative approach in the introduction and the conclusion of the chapter would have suited me better). A good editor sh(/w)ould probably have caught something like this.

The efficacy/effectiveness difference he talks about relates to the fact that the results of randomized controlled trials (RCTs) could/should be considered estimates of the health effects related to something close to the ideal treatment scenario, whereas real world implementation (effectiveness) of the treatment in question will often provide patients a sometimes significantly lower health benefit in terms of average treatment effect (or similar metrics), because of differences in the composition of the two groups and the settings of the treatment protocols applied, among other things. RCTs often deliberately try to maximize compliance e.g. by excluding patients who are likely to be non-compliers, and that of course will lead to biased estimates if you apply such estimates to the total patient population. There are many variables affecting how big the potential difference between efficacy and effectiveness may be for a particular drug and they cover that stuff, as well as a lot of other stuff, in the book. Non-compliance rates are much bigger than I’d imagined, but there are a lot of reasons for this that I hadn’t considered. The fact that non-compliance is widespread can be inferred even from the definitions applied in clinical trials:

“ultimately it is the outcome that is important. This might not always require that all doses of a drug are taken. Indeed, in short-term efficacy clinical trials patients who take 80 per cent or more of their medication, based upon pill counts, are usually considered ‘compliant’.” (p.14)

You can fail to take one-fifth of the medicine and still be considered compliant. Indeed as Parkinson, Wei and McDonald put it in their chapter:

“As the reader of this chapter it might be informative to reflect on your own behaviour: can you honestly say that you have always complied fully with every tablet of every prescription and have always finished the course? A very few readers will say yes, with honesty. The reality is that nearly everyone is non-compliant; the variable is the degree of non-compliance.”

A few numbers from the book illustrating the extent of the problem:

“reports (for example, Sung et al., 1998) have suggested that only 37 percent of participants take greater than 90 per cent of all doses of statins over a two-year period. […]

[Astma:] When patients were aware of being monitored a majority (60 per cent) were fully compliant, but when unaware the majority had a compliance rate between 30 and 51 per cent (Yeung et al., 1994). […]

Significant levels of non-redemption [of prescriptions], as seen in this study, have subsequently been confirmed within the large UK general practice databases such as GPRD where there is only about 90 per cent concordance between the prescriptions issued by the GP and those recorded as being redeemed at a pharmacy by the UK Prescription Pricing Authority (Rodriguez et al., 2000). […]

Chapman et al. (2005) recently examined compliance with concomitant antihypertensive and lipid-lowering drug therapy in 8406 enrollees in a US-managed care plan […] Less than half of patients (44.7 per cent) were adherent with both therapies three months after medication initiation, a figure that decreased to 35.8 per cent at 12 months. […]

Despite international clinical guidelines recommending lipid-lowering treatment in patients with clinically evident atherosclerotic vascular disease, study after study has documented low treatment rates in this high-risk patient population, thereby creating a clinical practice and public health dilemma (Fonarow and Watson, 2003).
Only about 30 per cent of patients with established CVD and raised serum lipids, and fewer than 10 per cent of individuals eligible for primary prevention, receive lipidlowering therapy. Target total cholesterol concentrations are then achieved in fewer than 50 per cent of patients who do receive such treatment (Primatesta and Poulter, 2000).
Poor patient compliance to medication regimen is a major factor in the lack of success in treating hyperlipidaemia (Schedlbauer et al., 2004). All of the lipid-lowering drugs must be continued indefinitely; when they are stopped, plasma cholesterol concentrations generally return to pretreatment levels (Anon, 1998). […]

Up to half of the patients treated for hypertension drop out of care entirely within a year of diagnosis (ibid. [WHO, 2003b], Flack et al., 1996). […]

Non-compliance comes in many forms: depending on the disease area, as many as one in five patients fail to take the first step of collecting a prescription from the pharmacy. Many patients on short-term medications depart from recommended doses within a day or two of starting treatment. And many of those on longer-term medication may take a break from their medication or vary their dose depending on how they feel. A review of the evidence (Horne and Weinman, 1999) concluded that compliance overall is approximately 50 per cent but varies across different medication regimens, different illnesses and different treatment settings.”

A little more stuff from the book:

“Compliance depends on many factors, including the study population (better in educated compared to disadvantaged patients) type of intervention, duration of treatment, complexity of treatment, real or perceived side-effects and life circumstances (see Table 8.1). The reasons are often patient-specific, multifaceted and can change over time. Demographically, the very young, the very old, teenagers and those taking very complex treatment regimes are the least likely to comply. […]

asymptomatic and chronic diseases needing long-term treatment […] result in poorer compliance; and […] the longer the remission in chronic diseases, the lower the compliance (Blackwell, 1976). […] patient-controlled non-compliance was lower in treatment for diseases in which the relationship between non-compliance and recurrence is very clear, such as diabetes, compared to treatment for diseases in which this relationship is less clear […] Of course, cognitive deficit, helplessness, poor motivation and withdrawal all lead to forgetfulness and passive or structural noncompliance (Gitlin et al., 1989; Shaw, 1986). […] most non-compliance is intentional and results from conscious choices. […]

As a rule, patients cannot be simply classified as compliers or non-compliers. Rather, the level of compliance ranges from patients who take every prescribed dose precisely as directed to those who never do with the typical patient lying between these two extremes. The degree to which patients intend to comply with a regimen can be subdivided into patient-controlled and structural. Patient-controlled factors can be subdivided further into rational behaviour (as seen in patients with Parkinson’s disease who regulate their own dosing) and irrational behaviours (such as self-induced seizures). Structural factors are those beyond the patient’s control, such as impaired memory or difficulty accessing medication (Leppik, 1990). […]

Compliance and adherence to therapy are complex issues with no obvious ‘one size fits all’ solution available. It appears that actively involving patients in treatment decisions, empowering patients with access to medical information and providing ongoing monitoring all contribute to improved compliance and adherence rates. The challenge for health services, however, is to provide these enhanced levels of support cost-effectively.”

The book is a few years old and sometimes you can tell. I was curious along the way about how much things have changed in the meantime. I’m guessing less than would have been optimal.

I should point out lastly that I have made a goodreads profile. I haven’t added a lot of books to my profile yet, but I may decide to use that site actively in the future. At goodreads I gave the book 3 stars, corresponding to an ‘I liked it’ evalution.

April 22, 2013 - Posted by | Books, Data, Diabetes, Economics, Health Economics, Medicine, Pharmacology

No comments yet.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: