SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References

Antihypertensive agents are one of the most commonly prescribed classes of medications in the country, but patient adherence rates are low. To better understand why rates are low, the authors used data from the 2005 HealthStyles survey and found that among the 1432 respondents who received prescriptions for antihypertensive medications, 407 (28.4%) reported having difficulty taking their medication. ``Not remembering'' was the most common reason reported (32.4%), followed by cost (22.6%), having no insurance (22.4%), side effects (12.5%), other reasons (12.3%), not thinking there is any need (9.3%), and having no health care provider (4.7%). In a multivariate model, younger age, lower income, having mental function impairment, and having had a blood pressure check more than 6 months earlier were factors significantly associated with reporting difficulty taking prescribed antihypertensive drugs. Control of hypertension is a significant public health issue, and alleviating barriers to medication adherence should be a major goal toward hypertension management.

Hypertension affects more than 30% of the United States population aged 20 years and older.1 More than 44 million office visits were made by patients with a primary diagnosis of hypertension in 2005.2 Hypertension is a well-known risk factor for cardiovascular disease, cerebrovascular disease, and renal disease.3 Antihypertensive agents are the second most commonly prescribed class of medications, exceeded only by antidepressants.2 Despite the widespread use and proven efficacy of this class of drugs, only 65% of patients with hypertension were treated between 1999 and 2004.4 A recent Harris Poll survey in 2007, however, reported that as many as 90% of hypertensive patients are now receiving some medication.5 The proportion of individuals with hypertension who have their blood pressure controlled increased from 29.2% in 1999–2000 to 32.5% in 2001–2002 and 36.8% in 2003–2004 and based on the recent survey to about 50%.4–6 However, despite the positive trend, the current overall proportion is still probably less than the goal of 50% set by the Healthy People 2010 initiative.7

One of the reasons for this lack of treatment and control is a low rate of patient adherence to medication. In a review of studies using electronic monitoring to measure rates of antihypertensive prescription adherence, the adherence rate among the 17 studies examined ranged from 39% to 93%, with an average of 73% of patients taking their prescribed doses.8 Further, long-term persistence of treatment adherence has been shown to decrease substantially over time, and among those with a new diagnosis of hypertension, it can decrease to <50% over a period of 4.5 years.9 A variety of reasons for this lack of adherence have been described. They include patients’ demographics, adverse effects of medication, high cost, having to take multiple doses per day, lack of access to health insurance or providers, and presence of mental problems (eg, depression, substance abuse, cognitive impairment).8,10–16 Another important factor may be low motivation by patients to take their medication due to the lack of symptoms associated with hypertension.

Whatever the reason, increasing patient adherence to medication regimens has been shown to lead to overall cost savings because of lower nondrug costs such as hospitalization.17 Hence, increasing adherence is beneficial to both the patient and the U.S. health care system and should be a major focus for public health and clinical practice.

Interventions to alleviate the burden of nonadherence will not be able to focus on the appropriate issues surrounding the problem until the reasons for nonadherence to antihypertensive medication regimens are better understood. Most of the studies addressing reasons for nonadherence are based on data from clinical trials, clinic-based surveys,12,18,19 or population-based studies analyzing possible barriers based on medical records or pharmacy claims.20 We could not find publications of population-based survey studies analyzing self-reported reasons for not adhering to prescribed medication regimens. For this purpose, we examined the 2005 HealthStyles survey to further characterize the reasons for antihypertensive medication nonadherence.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References

Study Population

The HealthStyles survey is an annual questionnaire mailed out to U.S. adults aged 18 years and older and is administered by Porter Novelli, a public relations company. A larger mail survey, ConsumerStyles, was first mailed to a stratified random sample of 20,000 potential respondents from a consumer mail panel comprising 450,000 households managed by Synovate, Inc. The sample was stratified by region, household income, population density, residents’ age, and household size to match the distribution of these characteristics in the U.S. general population. A low-income/minority supplement (n=3000) was included to provide adequate representation of these groups. The ConsumerStyles survey was completed by 12,639 persons, yielding a response rate of 63%. A total of 6168 HealthStyles surveys were then sent out to a representative sample of ConsumerStyles respondents. Of these, 4819 persons completed the HealthStyles surveys, yielding a response rate of 78%. The overall response rate was 49%. The HealthStyles survey is used in health communication planning by several organizations, including the US Centers for Disease Control and Prevention, an organization that also contributes to the design and administration of this survey.

Ascertainment of Outcomes

Participants were entered into the study if they answered “yes” to the question “Has a doctor ever prescribed medication to help lower your blood pressure?” The main outcome variable, having difficulty in taking antihypertensive medication, was determined from the following question: “Which of the following make it difficult for you to take your high blood pressure medication regularly? (Check all that apply.)”

  • 1
     I don’t always remember to take my medication.
  • 2
     I don’t have insurance that covers my high blood pressure medication.
  • 3
    It costs too much.
  • 4
    I don’t like the side effects.
  • 5
    I don’t have a regular health care provider to prescribe medication for me.
  • 6
    I don’t think I need to take the medicine every day.
  • 7
    There are other reasons.
  • 8
    I have no difficulty in taking my high blood pressure medication regularly.

Participants were determined to not have any difficulty in taking antihypertensive medication if they only responded “yes” to statement 8: “I have no difficulty in taking my high blood pressure medication regularly.” The individual barriers were ascertained by selecting any of the 7 reasons, and respondents could choose more than one.

Independent Variables

The demographic variables studied were sex, age (18–34 years, 35–44 years, 45–54 years, 55–64 years, 65 or older), race (white, black, Hispanic, other), education (less than high school, high school, some college, college graduate or higher), and income (<$15K, $15K–$24.9K, $25K–$39.9K, $40K–$59.9K, ≥$60K). The health status variables were dichotomous variables (yes/no) for having diabetes, congestive heart failure, stroke/transient ischemic attack, any cancer except skin, depression, anxiety disorder, migraines, family history of stroke/transient ischemic attack, and number of prescription medications taken (0, 1, 2, 3, 4–7, ≥8). The health behavior variables were time of last blood pressure check (within past 6 months, more than 6 months earlier) and number of primary care visits (0, 1, 2–4, ≥5).

Data Analysis

Univariate associations were examined by selected patient characteristics for outcomes of interest using the chi-square or Fisher’s exact test, as appropriate. A multivariate logistic regression model was used to further examine the association of reporting difficulty in taking antihypertensive medications with the independent variables listed above and categorized as in Table III. Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were estimated, and 2-tailed P values at .05 were considered to be significant. All statistical analyses were performed with the statistical software package SAS, release 9.1 (SAS Institute, Cary, NC).

Table III.  Association of Selected Patient Characteristics With Reporting Any Difficulty in Taking Antihypertensive Medication—HealthStyles Survey, United States, 2005
CharacteristicNo. (%) Reporting DifficultyAdjusted Odds Ratio (95% Confidence Interval)P Value
Sex
 Female218 (28.2)0.9 (0.7–1.1).38
 Male189 (28.7)REF
Age, y
 18–44125 (42.4)2.6 (1.8–3.7)<.0001
 45–54120 (31.5)1.8 (1.2–2.5)
 55–6479 (23.2)1.2 (0.8–1.7)
 ≥6583 (20.0)REF
Race
 White234 (25.4)REF.16
 Black91 (33.5)1.3 (0.9–1.8)
 Other82 (34.5)1.3 (0.9–1.8)
Education
 High school graduation and under164 (30.3)1.04 (0.8–1.3).77
 Some college and above229 (26.9)REF
Income
 <$25K175 (33.1)1.6 (1.1–2.2).02
 $25K–$59.9K119 (27.8)1.2 (0.9–1.7)
 ≥$60K113 (23.7)REF
Mental function (depression/anxiety/migraines)
 Yes166 (37.5)1.6 (1.2–2.1).0005
 No241 (24.4)REF
Comorbidities (diabetes/congestive heart failure/stroke/cancer)
 Yes133 (26.8)0.9 (0.7–1.2).54
 No274 (29.3)REF
Time of last blood pressure check
 Within past 6 months343 (26.5)0.5 (0.3–0.7).0003
 More than 6 months earlier59 (46.1)REF

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References

Overall, 1432 (29.7%) respondents reported that they had ever been prescribed medication for high blood pressure. Of these respondents, 407 (28.4%) reported having difficulty in taking their antihypertensive medication. Table I describes this population according to selected subgroups. Reporting difficulty did not differ by sex but did differ by age (56.8% aged 18–34 to 20.0% aged 65 years or older) and race/ethnicity (25.4% in whites; 33.5% in blacks; 36.8% in Hispanic persons). Significantly more respondents with lower incomes reported difficulty than those with higher incomes, and there was a trend toward individuals with less education reporting more difficulty than those with more education. More respondents with depression (38.8%) than without (25.6%), more respondents with anxiety (43.9%) than without (26.8%), and more respondents with migraines (40.0%) than without (26.6%) reported having difficulty. Persons taking more medications had less difficulty than those taking fewer medications, and persons with more primary care visits had less difficulty than those with fewer. Also, those who had had their blood pressure measured more than 6 months earlier reported more difficulty (46.1%) than those whose blood pressure was measured within the past 6 months (26.5%). Of interest, we found no significant difference between respondents who reported having a blood pressure machine at home and those who did not (data not shown).

Table I.   Percentage of Respondents Reporting Difficulty in Taking Their High Blood Pressure Medication Regularly, by Selected Characteristics—HealthStyles Survey, United States, 2005
Characteristic No. (%)No. (%) Reporting DifficultyP Value
Demographics
 Sex
  Male658 (45.9)189 (28.7).81
  Female774 (54.1)218 (28.2)
 Age, y
  18–3481 (5.7)46 (56.8)<.0001
  35–44214 (14.9)79 (36.9)
  45–54381 (26.6)120 (31.5)
  55–64341 (23.8)79 (23.2)
  ≥65415 (29.0)83 (20.0)
 Race
  White922 (64.4)234 (25.4).004
  Black272 (19.0)91 (33.5)
  Hispanic163 (11.4)60 (36.8)
  Other75 (5.2)22 (29.3)
 Education
  Less than high school graduation145 (10.4)46 (31.7).09
  High school graduation 397 (28.5)118 (29.7)
  Some college491 (35.2)138 (28.1)
  College graduation or more361 (25.9)91 (25.2)
 Income
  <$15K386 (27.0)132 (34.2).0006
  $15K–$24.9K142 (9.9)43 (30.3)
  $25K–$39.9K226 (15.8)64 (28.3)
  $40K–$59.9K202 (14.1)55 (27.2)
  ≥$60K476 (33.2)113 (23.7)
Health status
 Comorbidities
 Diabetes
  Yes359 (25.8)92 (25.6).20
  No1032 (74.2)303 (29.4)
 Congestive heart failure
  Yes81 (5.7)17 (21.0).16
  No1351 (94.3)390 (28.9)
 Stroke/transient ischemic attack
  Yes106 (7.6)34 (32.1).37
  No1289 (92.4)358 (27.8)
 Any cancer (except skin)
  Yes53 (3.7)18 (34.0).36
  No1379 (96.3)389 (28.2)
 Depression
  Yes307 (21.4)119 (38.8)<.0001
  No1125 (78.6)288 (25.6)
 Anxiety disorder
  Yes132 (9.2)58 (43.9)<.0001
  No1300 (90.8)349 (26.8)
 Migraines
  Yes195 (13.6)78 (40.0).0002
  No1237 (86.4)329 (26.6)
 Family history of stroke/transient ischemic attack
  Yes529 (42.3)143 (27.0).57
  No723 (57.7)207 (28.6)
 No. of prescription medications
  064 (4.8)50 (78.1)<.0001
  1138 (10.5)55 (39.9)
  2227 (17.2)64 (28.2)
  3198 (15.0)51 (25.8)
  4–7493 (37.3)115 (23.3)
  >7200 (15.2)40 (20.0)
Health behaviors
 Time of last blood pressure check
  Within past 6 months1292 (91.0)343 (26.5)<.0001
  More than 6 months earlier128 (9.0)59 (46.1)
 No. of primary care visits
  069 (4.9)36 (52.2).0004
  1189 (13.5)58 (30.7)
  2–4819 (58.6)216 (26.4)
  >4320 (22.9)83 (25.9)

Table II describes respondents’ specific reasons for having difficulty in taking their blood pressure medications. “Not remembering” was the most common reason reported for having difficulty in taking blood pressure medication (32.4%), followed by cost (22.6%), having no insurance (22.4%), side effects (12.5%), other reasons (12.3%), not thinking there is any need (9.3%), and having no health care provider (4.7%). Female respondents were more likely to report “no need to take the medication” than male respondents. Younger respondents listed not remembering to take their medication as a reason for difficulty more than older respondents. In contrast, older age was associated with listing “having no insurance” as a reason. The portion reporting side effects as a reason for difficulty decreased with age until 65 years and older, when it increased. Excluding those aged 65 years and older, the trend remained nonsignificant (P for trend, .12). Respondents with lower education levels reported having no insurance and having no provider as difficulties more often than those with higher education levels. As anticipated, persons with higher income levels were less likely to report cost, “having no insurance,” and “having no provider” than were those with lower income levels. However, individuals reporting income ≥$60K reported not remembering to take their medication more often than persons in all other income categories. Side effects were most commonly reported as difficulties among patients taking no medications (24.0%) and those taking >7 medications (20.0%) (data not shown). Respondents who had their blood pressure checked more than 6 months earlier and those who had fewer primary care clinic visits reported “having no provider” more often than persons whose blood pressure was checked within the past 6 months and those who had more primary care visits, respectively.

Table II.  Percentage of Respondents Reporting Specific Barriers to Taking High Blood Pressure Medication, by Selected Characteristics—HealthStyles Survey, United States, 2005
 TotalForgettingNo InsuranceCostSide EffectsNo ProviderNo NeedOther
CharacteristicNo. (%)No. (%)P ValueNo. (%)P ValueNo. (%)P ValueNo. (%)P ValueNo. (%)P ValueNo. (%)P ValueNo. (%)P Value
Overall407132 (32.4)91 (22.4)92 (22.6)51 (12.5)19 (4.7)38 (9.3)50 (12.3)
Sex
 Male189 (46.4)64 (33.9) 42 (22.2) 47 (24.9) 25 (13.2) 10 (5.3) 10 (5.3) 20 (10.6) 
 Female218 (53.6)68 (31.2).6049 (22.5)1.0045 (20.6).3426 (11.9).769 (4.1).6428 (12.8).0130 (13.8).37
Age, y
 18–3446 (11.3)20 (43.5) 7 (15.2) 7 (15.2) 10 (21.7) 2 (4.3) 3 (6.5) 4 (8.7) 
 35–4479 (19.4)30 (38.0) 14 (17.7) 18 (22.8) 14 (17.7) 1 (1.3) 8 (10.1) 16 (20.3) 
 45–54120 (29.5)46 (38.3) 23 (19.2) 30 (25.0) 10 (8.3) 8 (6.7) 8 (6.7) 17 (14.2) 
 55–6479 (19.4)21 (26.6) 23 (29.1) 22 (27.8) 5 (6.3) 5 (6.3) 7 (8.9) 5 (6.3) 
 ≥6583 (20.4)15 (18.1).000424 (28.9).01415 (18.1).7612 (14.5).093 (3.6).6412 (14.5).188 (9.6).15
Race
 White234 (57.5)72 (30.8) 50 (21.4) 53 (22.6) 26 (11.1) 11 (4.7) 16 (6.8) 23 (9.8) 
 Black91 (22.4)30 (33.0) 21 (23.1) 19 (20.9) 15 (16.5) 3 (3.3) 14 (15.4) 15 (16.5) 
 Hispanic60 (14.7)24 (40.0) 18 (30.0) 18 (30.0) 7 (11.7) 4 (6.7) 6 (10.0) 5 (8.3) 
 Other22 (5.4)6 (27.3).542 (9.1).222 (9.1).233 (13.6).621 (4.5).822 (9.1).137 (31.8).01
Education
 Less than high school graduation46 (11.7)17 (37.0) 16 (34.8) 14 (30.4) 3 (6.5) 5 (10.9) 4 (8.7) 8 (17.4) 
 High school graduation118 (30.0)40 (33.9) 41 (34.7) 33 (28.0) 18 (15.3) 7 (5.9) 8 (6.8) 11 (9.3) 
 Some college138 (35.1)43 (31.2) 18 (13.0) 20 (14.5) 13 (9.4) 3 (2.2) 20 (14.5) 20 (14.5) 
 College graduation or more91 (23.2)29 (31.9).5014 (15.4)<.000122 (24.2).1315 (16.5).352 (2.2).016 (6.6).7911 (12.1).89
Income
 <$15K132 (32.4)37 (28.0) 46 (34.8) 33 (25.0) 13 (9.8) 10 (7.6) 13 (9.8) 15 (11.4) 
 $15K–$24.9K43 (10.6)11 (25.6) 12 (27.9) 15 (34.9) 5 (11.6) 4 (9.3) 3 (7.0) 7 (16.3) 
 $25K–$39.9K64 (15.7)19 (29.7) 16 (25.0) 14 (21.9) 8 (12.5) 3 (4.7) 9 (14.1) 7 (10.9) 
 $40K–$59.9K55 (13.5)15 (27.3) 11 (20.0) 13 (23.6) 9 (16.4) 2 (3.6) 2 (3.6) 6 (10.9) 
 ≥$60K113 (27.8)50 (44.2).0126 (5.3)<.000117 (15.0).0416 (14.2).22 .00311 (9.7).7915 (13.3).84
Depression/anxiety/migraines
 Yes166 (40.8)56 (33.7) 44 (26.5) 36 (21.7) 22 (13.3) 10 (6.0) 16 (9.6) 19 (11.4) 
 No241 (59.2)76 (31.5).6747 (19.5).1256 (23.2).8129 (12.0).769 (3.7).3422 (9.1).8631 (12.9).76
Time of last blood pressure check
 Within past 6 months343 (85.3)117 (34.1) 72 (21.0) 77 (22.4) 40 (11.7) 10 (2.9) 32 (9.3) 37 (10.8) 
 More than 6 months earlier59 (14.7)15 (25.4).1917 (28.8).1814 (23.7).8310 (16.9).268 (13.6).00036 (10.2).8412 (20.3).04
No. of primary care visits
 036 (9.2)5 (13.9) 12 (33.3) 11 (30.6) 5 (13.9) 8 (22.2)   7 (19.4) 
 158 (14.8)20 (34.5) 16 (27.6) 12 (20.7) 5 (8.6) 5 (8.6) 5 (8.6) 11 (19.0) 
 2–4216 (55.0)77 (35.6) 42 (19.4) 46 (21.3) 27 (12.5) 5 (2.3) 20 (9.3) 19 (8.8) 
 >483 (21.1)26 (31.3).1517 (20.5).0720 (24.1).6311 (13.3).75 <.00019 (10.8).0911 (13.3).13

Multivariate logistic regression analysis revealed that the associations between having difficulty in taking blood pressure medications and sex, education, and comorbidities (congestive heart failure, stroke, diabetes, cancer) remained nonsignificant (Table III). Younger adults were more likely to report difficulty than older adults. Compared with the oldest age group (65 years and older), persons aged 18 to 44 years and 45 to 54 years were 2.6 (95% CI, 1.8–3.7) and 1.8 (95% CI, 1.2–2.5) times more likely to have difficulty, respectively. For persons aged 55 to 64 years, the association was nonsignificant. There was, however, an overall trend toward reporting more difficulty with younger age. The association of reporting difficulties with race became nonsignificant after adjusting for the other covariates in the model. The lowest-income group was significantly associated with higher odds of reporting difficulty as compared with highest-income group (OR, 1.6; 95% CI, 1.1–2.2); there appeared to be a trend toward reporting more difficulty with decreasing income. Having a mental health condition was significantly associated with reporting more difficulty in taking medications (OR, 1.6; 95% CI, 1.2–2.1), while a more recent blood pressure check (less than 6 months ago) was significantly associated with less difficulty (OR, 0.5; 95% CI, 0.3–0.7).

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References

In this study of U.S. adults, we found that difficulties in taking antihypertensive medication are common (28.4% of respondents) and the reasons for these difficulties differ by certain population characteristics. Younger respondents had more difficulty than older respondents overall, and it seems that this is mostly due to forgetting to take medications. However, respondents aged 55 years and older were more likely to report having no insurance as a prohibitive factor. Furthermore, older and lower-income respondents were more likely to report barriers related to health care access than younger and higher-income respondents—both in crude associations and after adjusting for age, sex, race, comorbidities, mental status, and time of last blood pressure check. It was of interest to note that very few respondents reported having no health care provider to prescribe the medications; however, less educated and lower-income respondents still were more likely to report this barrier than the more highly educated and higher-income respondents. Black and Hispanic respondents reported a higher prevalence of barriers than whites in the survey overall, although the significance disappeared after adjusting for other demographic, health status, and behavioral factors. None of the specific barriers were associated with race. Individual characteristics such as socioeconomic and health status may play a greater role in predicting difficulty in taking antihypertensive medication than does race.

Differences in barriers to adherence were largely absent with respect to the comorbidities studied in both crude and adjusted analyses. However, those with impairments of mental function (depression, anxiety, or migraine) reported a higher prevalence of having difficulty in taking their antihypertensive medication than those without these conditions. This finding was true overall (in both crude and adjusted analysis) but showed no difference among specific barriers. Providers should pay special attention to patients with mental health disorders, as these patients may find it more difficult to adhere to medication regimens.

Several health behaviors were associated with more difficulty in taking blood pressure medications, such as infrequent clinic visits and infrequent blood pressure checkups. The presence of side effects was a relatively uncommon reason reported for having difficulty in taking antihypertensive medication. However, complaints of side effects were more prevalent at extremes of age—in the young adult group and in the elderly. Also, those not receiving medications and those taking >7 medications reported side effects more commonly than those with numbers of prescribed medications in between. Patients not taking medications may do so in part because of previously experienced side effects or for other reasons. It is possible that a higher prevalence of difficulties due to side effects might have been better assessed by asking a question such as, “At any time in the past, did you or your doctor stop a blood pressure medication because of side effects?”

The strengths of this study are the nationwide sample of respondents, the population-based design of the survey, and the fact that participants were asked directly about reasons that made it difficult for them to take their antihypertensive medications regularly. In addition, the answer options covered a variety of possible reasons, including patient-related, medication-related, and health system–related factors.

The study limitations include the low overall response rate of 49%, which may have led to nonresponse bias. Although the HealthStyles survey is a population survey designed to match the characteristics of the U.S. adult population, survey participants volunteered to participate and thus were not a random sample. However, estimates of the prevalence of various conditions based on HealthStyles survey data have been shown to be similar to corresponding estimates based on Behavioral Risk Factor Surveillance System data and are stable from year to year.21 The survey question pertaining to barriers to adherence, although covering a wide range of factors, may not have been all-encompassing, and the survey may have missed important reasons for having difficulty in taking antihypertensive medications. Next, assessing medication compliance through a patient questionnaire is not the most reliable method: it is greatly prone to patients’ recall bias or overreporting due to social desirability phenomenon, and thus this method is likely to overestimate adherence. However, it has also been shown that other methods, such as refill audits, measures of drug concentration or biomarkers in blood, pill counts, and electronic monitors, are all subject to various limitations as well; many of them are very expensive and not reasonable for use with a large sample of patients.11,18

Perspectives

Control of hypertension is a significant public health issue, and improving patient adherence rates should be one major goal of hypertension management. This, and the problem of determining the most effective medication in at least 30% of patients whose blood pressure is uncontrolled medication is not changed, are 2 reasons for the less than ideal control rates in hypertensive patients.5 Measures are needed to alleviate barriers to medication adherence and eliminate disparities in adherence, as well as more efforts to increase physician adherence to therapy guidelines in order to lower overall morbidity and mortality from this disease. When adherence is high, overall cost savings can be achieved because of lower overall costs when hospitalization and other nondrug costs are taken into account.17 Health care access issues such as insurance and medication cost must be improved, as well as patient-provider factors such as patient education about the benefit of treatment and about side effects and discerning ways to help patients remember to take their medications. In addition, better surveillance systems are needed to monitor the care of hypertensive patients as well as the effectiveness of their disease management. Last, more investigation into sociodemographic disparities in barriers to antihypertensive medication adherence needs to be conducted to further characterize and intervene on these inequalities.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References