Adherence and diabetes

In the current issue of the Journal of Diabetes, Ofori-Asenso et al.1 analyze adherence to statins among adults aged ≥65 years in a dataset including 10% of the Australian population. Why is adherence so important and, when all is said and done, so poorly addressed in the treatment of diabetes? Billions of prescriptions are written annually in the US alone,2 of which only 50%–80% are filled at the pharmacy (with a subset never actually obtained by the patient), half or fewer are taken as prescribed and, by some estimates, less than one-quarter are refilled as prescribed.3

Adherence does not only include the use of medications; healthy lifestyle adherence in type 2 diabetes mellitus (T2DM) is associated with a 30% reduction in mortality.4 For T2DM, reported mean medication adherence rates are only between 30% and 40%, with good versus poor adherence associated with a greater risk of mortality and hospitalization.5 Similar benefits have been shown with low-dose aspirin6 and with angiotensin-converting enzyme inhibitors and angiotensin receptor blockers.7

Although adherence can easily be defined, for example as “whether patients take their medications as prescribed (e.g. twice daily), as well as whether they continue to take a prescribed medication”,8 it is not as easily ascertained. Most published studies use pharmacy-based methods in calculating adherence, such as the medication possession ratio (MPR) or percentage of days covered (PDC), calculated as follows:9

where the refill interval is the number of days between the first and last fill.

However, there are issues with all the methods used to assess adherence: medical or pharmacy records are often inaccurate; the methods do not account for waste, stockpiling, or actual doses taken, pharmacy records do not account for out-of-system medication (e.g. samples, medication purchased online etc.); measurements of pill count (or the use of medication event monitoring systems capturing the opening of medication bottles) are not necessarily correlated with actual ingestion; the monitoring of HbA1c, glucose, blood pressure, or lipids is influenced by a multiplicity of factors, so these measures cannot be readily equated with medication use; and questionnaires are often inaccurate, with patients often overestimating their medication use in self-report.10 We have found the questionnaire approach to be internally consistent and to suggest that there are a myriad of patient-specific factors distinguishing those patients with high, moderate, and low levels of adherence and those who do not take prescribed medications, including access to information and opportunities to interact with healthcare providers, medication costs, side effects, and depression and anxiety.11 Others have found similar patient self-reported reasons for non-adherence, including forgetting, running out and not being able to reach the pharmacy, costs, and side effects.12 Socioeconomic status is important, with higher adherence associated with smaller family size, greater educational level, and greater levels of health insurance coverage.13

For glucose-lowering treatment, there is some evidence that adherence to metformin is less than that to sulfonylureas and thiazolidinediones, which, in turn, show lower adherence than that to dipeptidyl peptidase 4 inhibitors (DPP-4i).14 Indeed, an important caveat from a large pharmacy database study of people given second-line treatment after metformin showed that fewer than 10% had definite evidence of having taken metformin in the recommended dosage, and that 28% had no prior claim evidence of having received the medicine at all!15 Certainly, one would expect that greater levels of adherence would lead to greater degrees of improvement in glycemia, and there is evidence that this is true for oral hypoglycemic agents, with a report from the UK Clinical Practice Database Study showing nearly twice as great an improvement in HbA1c in adherent people receiving oral hypoglycemic treatment,16 but the mechanism of this association may not simply reflect the pharmacologic action of the medication, with evidence that psychological factors associated with greater medication adherence may track with glycemic improvement.17 Issues of adherence are particularly great with injected diabetes treatments. In a study of 21 363 people with T2DM prescribed basal insulin, only 33.8% had a PDC >80%, with the resulting greater medication cost being offset by lower rates of hospitalization and emergency room visits.18 Adherence may be even more of an issue with glucagon-like peptide-1 receptor agonists (GLP-1RA), with injection frequency, needle size and local discomfort all important in determining the success of treatment.19 In a study of patients receiving liraglutide, although adherent people had nearly twice as great a reduction in HbA1c, only 34% had 1-year PDC >80%.20 A study comparing 2116 people receiving insulin glargine and 2374 receiving liraglutide showed persistence of 64% versus 49%, respectively, with considerably lesser HbA1c lowering related, in part, to greater gastrointestinal side effects and cost, with the latter having a greater effect on adherence than the lesser likelihood of hypoglycemia and weight gain.21, 22 The lesser adherence to GLP-1RA may explain the gap between efficacy in randomized controlled trials and real-world effectiveness of GLP-1RA versus DPP-4i, with prescribed people in real-world studies showing similar HbA1c lowering with both agents despite the much greater effect of the former in clinical trials.23

What of lipid-lowering treatment? Adherence to lipid-lowering medication in people with T2DM has been shown to be associated with a reduction in cardiovascular events and mortality, both in primary and secondary prevention.24 So-called statin intolerance may be a factor, brought about, in part, by patient beliefs about side effects, both varying to a large degree from country to country and tracking with each other.25 Although statin prescribing has increased considerably over the years, persistence levels are relatively low, around 60% in South Korea,26 55% in Denmark after 3–5 years,27 and, in the study of Ofori-Asenso et al.1 in the current issue of the Journal of Diabetes, decreasing from 54% at 6 months to 37% at 9 years.

There are several caveats to the assumption that improving adherence will directly affect outcome. A meta-analysis of randomized controlled trials for a variety of conditions showed that those people adherent to active treatment compared with non-adherent people had a pooled mortality odds ratio (OR) of 0.55, but that good adherence to placebo was also associated with a lower risk of mortality, with a pooled OR of 0.56,28 suggesting that adherence may be a marker for a variety of good health behaviors, as well as, perhaps, for differences in beliefs about the efficacy of medical care, socioeconomic status, and depression and anxiety, with all having potential effects on outcome. As mentioned, the use of pharmacy databases may be limited in allowing identification of people taking correct amounts of medication, and certainly the use of 80% as a cut-off for an effective adherence level is arbitrary, and may pertain more to one drug than another, even within a given class.29 The implication is, then, that we need to develop better approaches to improving ascertainment, as well as, of course, better communication skills to fully engage with our patients with diabetes in helping them to take effective medications in an effective manner.

在本期Journal of Diabetes杂志中,Ofori-Asenso等¹在年龄≥ 65岁的数据集(包含了10%的澳大利亚人口)中分析了成年患者对他汀类药物的依从性。为什么依从性如此重要,当我们把一切都已经说明白并且去做了之后,为什么糖尿病患者的治疗效果还如此糟糕?每年仅在美国就有数十亿的处方量²,其中能够到达药房的只有50%-80%(有一部分患者实际上从未取药),患者可能只使用了处方剂量的一半或者更少,据估计,能够按照处方再次取药的患者不到四分之一³。

依从性不仅仅只包括药物的使用;对于2型糖尿病(T2DM)患者来说坚持健康的生活方式最终可使死亡率下降30%⁴。T2DM患者报告的平均药物依从性只有30%-40%,而依从性不佳与依从性良好相比可导致患者的死亡率与入院率风险显著升高⁵。长期使用小剂量阿司匹林⁶与使用血管紧张素转换酶抑制剂或血管紧张素受体阻断剂⁷治疗的相关研究结果也证实了相似的获益。

虽然我们可以很容易地定义依从性,例如“患者是否按照他们的处方来使用药物(例如每日2次),以及他们是否在坚持使用处方药物⁸”,但这些都不容易确定。大多数已经发表的研究都使用基于药学的方法来计算依从性,如计算药物占有率(medication possession ratio,MPR)或者覆盖天数百分比(percentage of days covered,PDC),计算方法如下⁹:

MPR=获得供应的总天数/(补充间隔时间+最后一次补充的时间)

PDC=得到药物的天数/指定时间间隔的天数

其中补充间隔是第一次与最后一次补充之间的天数。

然而,所有用于评估依从性的方法都存在问题:医疗或者药房记录往往不准确;该方法没有考虑到浪费、囤积或者实际使用的剂量,药房记录不能解释系统之外的用药情况(例如样品,及通过网络购药等情况);药丸计数测定(或者使用电子药物监测系统获得开启的药瓶数量)未必与实际摄入的药物剂量有关;监测到的HbA1c、血糖、血压或者血脂经常会受到多种因素的影响,因此,这些测量结果不能简单地等同于药物使用效果;并且问卷调查往往不准确,患者在自我报告时经常高估自己的药物使用情况¹⁰。我们发现问卷调查方法往往都具有内在的一致性,这意味着还有无数的患者特异性因素可以用来区别高、中、低水平的用药依从性以及从未服用处方药物的患者,这些因素包括获取与医疗保健提供者互动的信息和机会、药物成本、副作用以及是否合并抑郁与焦虑¹¹。其他人也发现了类似的患者自我报告的依从性变差的原因,包括忘记用药、药物用完并且不能到药房购药、费用以及副作用¹²。社会经济条件很重要,家庭规模越小、教育水平越高、医疗保险覆盖水平越高则依从性越高¹³。

对于降糖治疗来说,有些证据表明患者对二甲双胍的依从性低于磺脲类药物与噻唑烷二酮类药物,而患者对磺脲类药物与噻唑烷二酮类药物的依从性又低于二肽基肽酶4抑制剂(dipeptidyl peptidase 4 inhibitors,DPP-4i)¹⁴。事实上,一项大型的药房数据库研究调查了患者在使用二甲双胍治疗的基础上给予二线治疗,其结果给了我们一个严重警示,有确切证据表明只有不到10%的患者服用了推荐剂量的二甲双胍,并且有28%的患者既往根本就没有已经使用过药物的索赔证据¹⁵。当然,人们都会期望更高水平的依从性会导致更好的血糖改善,并且有证据表明对于口服降糖药物来说这是真的,一项来自英国临床实践数据库研究(UK Clinical Practice Database Study)的报告表明,在坚持使用口服降糖药物治疗的人群中,HbA1c的降幅几乎达到了两倍¹⁶,但是这种降糖疗效的机制并不能够简单地归功于药物的药理作用,有证据表明,与更高的药物依从性相关的心理因素可能也与血糖的改善有关¹⁷。糖尿病注射治疗用药的依从性问题尤为突出。在一项纳入了21363名接受基础胰岛素治疗的T2DM患者的研究中,只有33.8%的患者PDC > 80%,由于住院率与急诊就诊率的降低,导致更高的用药成本被抵消了¹⁸。患者对胰高血糖素样肽-1受体激动剂(glucagon-like peptide-1 receptor agonists,GLP-1RA)的依从性可能更有问题,注射频率、针头大小以及局部不适这些问题对于成功治疗来说都很重要¹⁹。在一项利拉鲁肽治疗的研究中,虽然依从性高的患者HbA1c降幅是不依从患者的近两倍,但是1年PDC> 80%的患者只有34%²⁰。在一项纳入了2116名使用甘精胰岛素治疗以及2374名使用利拉鲁肽治疗的患者的对照研究中,结果发现坚持治疗的患者比例分别为64%与49%,利拉鲁肽治疗相关的HbA1c降幅小得多,在某种程度上,它的胃肠道副作用更大并且成本也更高,并且利拉鲁肽对依从性的影响更大,但是低血糖与体重增加的可能性却较小^21,22。在比较GLP-1RA与DPP-4i的随机对照试验以及真实世界疗效研究中,对GLP-1RA的依从性较低可以解释两者之间疗效的差距,尽管在临床试验中前者的疗效更大,但是在真实世界研究中却发现处方这两类药物的患者HbA1c降幅相似²³。

降脂治疗的依从性怎么样?目前已经发现T2DM患者坚持使用降脂药物进行一级预防治疗或者二级预防治疗都可以减少心血管事件与死亡率²⁴。所谓他汀类药物不耐受可能是一个原因,在某种程度上,是因为患者对副作用的认识所导致的,两者在很大程度上都会因国家不同而异并会相互影响²⁵。尽管他汀类药物的处方量在过去几年中已经显著增加,但是能够坚持治疗的患者比例还是相对较低,在韩国大约是60%²⁶,经过3-5年后在丹麦只有55%的患者还在坚持治疗²⁷,在本期杂志中,Ofori-Asenso等¹经过研究后发现,从第6个月至第9年能够坚持治疗的患者比例从54%降至37%。

有几个对提高患者的依从性将会直接影响预后的假设提出的警告。有一项研究对在各种条件下进行的随机对照试验进行了meta分析,结果发现坚持使用活性药物治疗的患者与依从性差的患者相比合并后的死亡率风险比(odds ratio,OR)为0.55,但是坚持使用安慰剂治疗的患者死亡率也有明显的下降,合并后的OR为0.56²⁸,这意味着依从性可能是多种良好健康行为的标志物,除此之外,也许是患者对于医疗效果的信念,患者的社会经济地位、抑郁与焦虑程度都各有不同,所有的这些因素都可能影响预后。如上所述,使用药房数据库来鉴定患者是否正在使用正确剂量的药物可能具有其局限性,当然,将80%定义为有效依从性的切点也具有主观性,并且这个切点对于一种药物来说可能比另外一种更适合,即使是同种类别的药物²⁹。那么,这意味着我们需要研究出更好的方法以改善鉴定效果,当然,除此之外,为了帮助糖尿病患者能够有效地使用正确的药物,还需要有更好的沟通技巧来与他们进行紧密联系。