Trends in diet counseling among stroke versus non-stroke patients: Evidence from the NAMCS, 2011–2016

Abstract

Introduction:

Dietary modifications in post-stroke patients facilitated by diet counseling improves post-stroke recovery and stroke recurrence. The extent to which clinicians provide dietary counseling for these patients is unknown.

Methods:

2011 to 2016 National Ambulatory Medical Care Surveys (NAMCS) data was used to assess trends in post-stroke diet consultations by age. Multivariate logistic regression models assessed the likelihood of dietary counseling provision among patients with and without stroke.

Results:

The proportion of patients with stroke aged 60–79 who received diet counseling decreased from 18.2% in 2011 to 5.3%, 11.9%, 8.7%, 13.4%, and 15.2% in 2012 – 2016, respectively. Among patients without stroke aged 60–79, diet counseling rate decreased from 12.9% in 2011 to 7%, 9.5%, 10.5%, 13.5%, and 12% in 2012 – 2016 respectively. Similar trends were observed among patients with and without stroke aged over 80.

Conclusions:

The proportion of patients with and without stroke receiving dietary counseling has remained low over the past half-decade. It is likely multifactorial-- related to clinician knowledge, patients’ receptiveness, and system-level factors of time and reimbursement. Future interventions should explore methods to address barriers to nutrition recommendations for post-stroke patients and patient activation to adopt dietary changes.

Introduction

In the United States (US), stroke is one of the top five most common causes of death¹. In 2018, the Centers for Disease Control and Prevention estimated approximately 7.8 million adults would experience at least one stroke in their lifetime, contributing to morbidity and mortality, especially among older Americans¹. Moreover, behavioral risk factors like smoking, poor diet, and low physical activity contributed to 74.2% of the total stroke burden².

The substitution of healthy dietary components for less healthy versions (e.g., Monounsaturated fatty acids [MUFA] and Poly-unsaturated fatty acids [PUFA]) and adherence to whole patterns of healthy eating (e.g., the Mediterranean diet, the Dietary Approaches to Stop Hypertension [DASH] diet and others) have been shown to reduce the risk factors predisposing individuals to stroke^3,4. Among individual dietary components, a diet with more fruits and vegetables was protective^5,6. High salt intake, red/processed meat, and alcohol (protective with moderate intake) were associated with high stroke burden^7,8. Thus, educating all stroke patients on healthy diets during post-stroke care is essential for efforts to reduce the risk of recurrent stroke and its complications.

The dietary changes that protect from coronary artery disease (CAD) also protect from stroke. In the Framingham prospective study, unsaturated fats were found to be associated with a protective effect on stroke⁹. Similar to CAD, fruit, vegetables, and folic acid may significantly affect stroke. Finally, a moderate intake of alcohol may also be related to a lowering of the risk of stroke. Diet recommendations were suggested for stroke patients in studies including the Lyon Diet Heart Study, the PREDIMED study and another in Finland^10,11,12. In both these interventions, mortality from CAD, cancer, and stroke was reduced by more than 50%. Given the compelling evidence for healthy diet for secondary stroke prevention, the American Heart Association adopted dietary guidelines in 2000 for adults who have had a stroke and recommended clinicians discuss diet with their patients¹³. While healthy lifestyle practices can reduce the risk of recurrent stroke, few stroke survivors are adherent¹⁴. Additionally, the extent to which providers follow guidelines for delivering patient education in the ambulatory setting is not known. Using nationally representative data, we examined the proportions of adults with a history of stroke who received diet counseling from their clinician in routine office visits.

Methods

Data from the National Ambulatory Medical Care Survey (NAMCS) from 2011 to 2016 was utilized in this study. The National Center for Health Statistics (NCHS) oversees the annual cross-sectional collection of NAMCS data. Since 1973, NAMCS has been acquiring healthcare information provided by non-federally employed office-based physicians¹⁵. NAMCS collects data representing the US population accessing ambulatory care by utilizing multistage probability sampling procedures. The response rates ranged from a low of 39.3% to a high of 61.9% across the years 2011–2016^{15,16,17,18,19}. Physicians were asked to complete a patient record form for 30 visits during an indiscriminately assigned period of one week of the year. The patient visit is the unit of analysis. In 2012, the NCHS changed the collection approach for NAMCS to a computerized electronic device. The US Census Bureau provides assigned ground agents to abstract data from the clinical records of selected providers, rather than rely on clinic staff to provide information¹⁰. The NAMCS item nonresponse rates were ≤ 5% with few exceptions each year²⁰. The current analysis was conducted from June to August 2020.

Study Population

We restricted our analysis to nonsurgical ambulatory care visits. We identified patients with stroke using clinician-reported patient cerebrovascular disease diagnosis within the patient encounter form’s chronic conditions section. Our method of identifying patients with stroke yielded prevalence estimates aligned with national estimates for each year included in the analysis²¹. All other nonsurgical ambulatory care visits were considered comparison for the patient population with no history of cerebrovascular disease. In a trend analysis, we compared patients with and without history of stroke in age groups 60–79 years old and 80+ years old. We restricted to adults over the age of 60, since nearly three-quarters of all strokes occur in people over the age of 65²². Further, the data for stroke patients below the age of 60 were not as reliable due to small sample sizes leading to large differences between weighted and unweighted estimates²³.

Measures

Field agent representatives, physicians, or clinical staff collected data on clinician-reported dietary counseling during the patient visit by checking a diet counseling indicator box on the NAMCS patient encounter form if provision of such counseling was identified during chart review. We examined a range of provider-, practice-, patient-, and stroke-related factors expected to impact the probability of receiving dietary counseling.

Patient-level characteristics included race/ethnicity, insurance type (private or nonprivate), age, and sex. The type of provider seen is documented on the NAMCS encounter form. We limited our analysis to visits in which the provider is a physician. This limitation is due to a small percentage of adults with a history of stroke being seen by non-physicians in the NAMCS data. We categorized physician specialty as family practice/general practice, Internal Medicine, Neurology, Cardiology, or all other specialties which are not disaggregated in the NAMCS data.

Practice-level characteristics included an indicator variable if the practice is located in a metropolitan statistical area and the geographic region (Northeast, Midwest, South, and West). The NAMCS collects data on several stroke risk factors. We included clinician-reported coronary artery disease (CAD), hypertension, obesity, diabetes, and hyperlipidemia for this analysis.

Statistical Analysis

The unit of analysis was an adult ambulatory care visit. Visit sampling weights (assigned by NAMCS) were used to account for uneven collection probabilities stemming from the NAMCS sample design and nonresponse. Stata, version 16.1, was used in all analyses. Weighted means along with 95% confidence intervals (CIs) are reported for all continuous variables.

National estimates of the total number of visits in the 2011–2016 period were estimated for patients with and without a stroke. Percentages of visits of patients with and without history of stroke in which counseling on diet was offered were then estimated. The incidence of counseling during visits each year of the survey sample was evaluated to assess diet counseling trends over time. Non-overlapping CIs for an α of 0.05 were used to assess for statistical significance.

The cross-sectional data were pooled from 2011 to 2016. The associations between patient, provider-, and practice-level characteristics and year of survey (in 2-year intervals) with the provision of diet counseling during the ambulatory care visit in patients with a history of stroke were assessed using multivariate logistic regression models. We built stepwise models to assess the predictive value of each category of variables. In the sample of patients without a stroke diagnosis, the multivariate models were next expanded to include mutually exclusive categories of no risk factors, CAD only, hypertension only, obesity only, diabetes only, and hyperlipidemia only, and two or more risk factors in order to assess the independent associations between the stroke risk factors with the prescription of dietary consultations.

Data Availability & Research Ethics Approval:

Any data not published within the article is available in a public repository of the CDC²⁴. This study was considered to be exempt by the Emory University Institutional Review Board, since the data are de-identified and have permitted access to public use by the NCHS Ethics Review Board (ERB)²⁵.

Results

Pooled results from 2011–2016 show 4,960 patients had a history of stroke, and 244,322 patients did not (Table 1). Among the total sample, 57.1% were female; 76.4% were white, and 9.1% were African American. Hispanic patients comprised 10% of the sample, and 4.5% were reported as other, non-Hispanic races. Most patients (59.4%) had private insurance. Over one-third (35.3%) of the sample patients were age less than 40, 27.2% were between age 40 and 59, 29.3% between age 60 and 79, and 8.2% were over 80 years old. Almost one-fifth (18.1%) of the sample patients had two or more risk factors and 63% had no risk factors. Among the patient visits surveyed, 17% of providers were Family Physicians, 8.6% were Internal Medicine physicians, 3.8% were Cardiologists, 3.1% were Neurologists, and 67.6% belonged to other specialties.

Table 1.

Comparison of Demographic Characteristics by Stroke Status, NAMCS^* 2011–2016^†

Characteristics	All	Stroke	No Stroke
	[N=249282]	(n= 4,960)	(n=244,322)
	n (%)	% ± SD‡	% ± SD‡
Race/Ethnicity
White non-Hispanic	190,461 [76.4%]	75.7± 0.01	70.5± 0.01
Black, non-Hispanic	22,658 [9.1%]	10.5± 0.01	10.2± 0.003
Other, non-Hispanic	11,220 [4.5%]	5±0.01	5.6± 0.004
Hispanic	24,943 [10.0%]	8.7± 0.01	13.7± 0.005
Medical Insurance
Private	148,028 [59.4%]	47.1± 0.01	59.6± 0.005
Nonprivate	101,254 [40.6%]	52.9± 0.01	40.4± 0.005
Age (years)
0–39	88,032 [35.3%]	3.0 ± 0.004	37.2± 0.005
40–59	67,829 [27.2%]	17.1± 0.01	27.4± 0.003
60–79	72,970 [29.3%]	52.6± 0.01	28± 0.003
≥80	20,451 [8.2%]	27.3± 0.01	7.4± 0.002
Sex
Female	142244 [57.1%]	52.0± 0.01	58.1± 0.003
Male	107038 [42.9%]	48.0± 0.01	41.9± 0.003
Geographic Region
Northeast	37,857 [15.2%]	21.6± 0.02	20.4± 0.005
Midwest	65,633 [26.3%]	19.1± 0.01	18.7± 0.005
South	84,793 [34.0%]	36.9± 0.02	37.0± 0.008
West	60,999 [24.5%]	22.4± 0.02	23.8± 0.007
Metropolitan statistical area
Yes	220,924 [88.6%]	91.1± 0.01	91.2± 0.006
No	28,358 [11.4%]	8.9± 0.01	8.8± 0.006
Physician Specialty
General/Family practice	42,303 [17.0%]	19.8± 0.01	21.2± 0.007
Internal Medicine	21,345 [8.6%]	19.7± 0.02	13.3± 0.007
Cardiology	9,388 [3.8%]	14.1± 0.01	3.2± 0.002
Neurology	7,654 [3.1%]	7.9 ± 0.01	1.4± 0.001
Other specialties	168,592 [67.6%]	38.5± 0.02	60.8± 0.008
Risk Factors
None	157,083 [63.0%]	16.5± 0.01	61.6± 0.005
CAD only§	1,718 [0.7%]	1.8± 0.003	0.7± 0.001
HTN only| |	26,133 [10.5%]	2.2± 0.002	1.8± 0.001
Obesity only	5,706 [2.3%]	0.9± 0.002	2.6± 0.001
Diabetes only	5,065 [2.0%]	16.0± 0.01	10.4± 0.002
Hyperlipidemia only	8,531 [3.4%]	4.7± 0.005	3.9± 0.001
Two or more	45,046 [18.1%]	57.8± 0.01	19.1± 0.004

^*NAMCS: National Ambulatory Medical Care Survey.

^†Data are weighted means.

^‡SD: standard deviation.

^§CAD: coronary artery disease

^{| |}HTN: hypertension.

Among patients with a history of a stroke aged 60–79, the proportion receiving diet counseling decreased significantly from 18.2% (95% CI: 10.78, 25.83) in 2011 to 5.3% (95% CI: 2.85, 7.84) in 2012 and changed to 11.9% (95% CI: 6.91, 16.96), 8.7% (95% CI: 5.00, 12.33), 13.4% (95% CI: 4.85, 21.87), and 15.2% (95% CI: 7.16, 23.25) in the years 2013, 2014, 2015 and 2016, respectively (Table 1). Among patients without a history of a stroke aged 60–79, the proportion receiving diet counseling decreased significantly from 12.9% (95% CI: 10.46, 15.48) in 2011 to 7% (95% CI: 6.00, 7.94) in 2012 and further changed to 9.5% (95% CI:7.93, 11.13), 10.5% (95% CI:8.53, 12.45), 13.5% (95% CI: 9.91–17.02), and 12% (95% CI: 8.25, 15.81) in the years 2013, 2014, 2015 and 2016, respectively. Among patients with a history of stroke over 80 years, the diet counseling proportions decreased significantly from 14.1% (95% CI: 6.12, 22.14) in 2011 to 8.5% (95% CI: 3.84, 13.12) in 2012 and further changed to 11.7% (95% CI: 5.21, 18.13), 9.7% (95% CI: 4.63, 14.83), 12.1% (95% CI: 2.19, 21.96), and 12.1% (95% CI: −.14, 24.31) for the years 2013, 2014, 2015 and 2016, respectively. Among patients without a stroke history over 80 years, the diet counseling proportions decreased from 11.1% (95% CI: 7.75, 14.53) in 2011 to 4.5% (95% CI: 3.44, 5.64) in 2012 and further changed to 7.5% (95% CI: 5.54, 9.53), 8.8% (95% CI: 6.43, 11.24), 9.5% (95% CI: 5.16, 13.91), and 11.4% (95% CI: 7.55, 15.30) for 2013, 2014, 2015, 2016, respectively (Table 1).

In the pooled cross-sectional analyses for the study period (2011−2016), among patients with stroke, those identified as Hispanic had 1.44 (95% CI: 0.73, 2.82) times higher odds of receiving diet counseling compared to whites (Table 2). Blacks and other non-Hispanic patients had 1.62 (95% CI: 0.88, 2.98) and 1.80 (95% CI: 0.85, 3.82) times higher odds of diet counseling compared to whites. Patients aged less than 40 (OR:0.41; 95% CI: 0.17, 0.99) and 40 to 59 (OR: 0.81; 95% CI: 0.50, 1.32) had lower odds of receiving diet counseling than patients aged over 80. Patients aged 60 to 79 years had no difference in odds of receiving diet counseling as patients aged over 80 (OR: 1.05; 95% CI: 0.71, 1.57). For insurance status, patients with private insurance had 1.5 (95% CI: 1.03, 2.19) times higher odds of receiving diet counseling than patients with non-private insurance.

Table 2.

Odds of Counseling by Patient and Physician Characteristics in Adults with Stroke, 2011–2016

Characteristics	Odds Ratio (95% CI)
Race/Ethnicity
White non-Hispanic	1.00 (ref)
Black, non-Hispanic	1.62 (0.88, 2.98)
Other, non-Hispanic	1.8 (0.85, 3.82)
Hispanic	1.44 (0.73, 2.82)
Medical Insurance
Private	1.5 (1.03, 2.19)
Nonprivate	1.00 (ref)
Age (years)
0–39	0.41 (0.17, 0.99)
40–59	0.81 (0.50, 1.32)
60–79	1.05 (0.71, 1.57)
≥80	1.00 (ref)
Sex
Female	0.83 (0.58, 1.18)
Male	1.00 (ref)
Geographic Region
Northeast	1.33 (0.73, 2.43)
Midwest	1.37 (0.74, 2.55)
South	1.47 (0.81, 2.67)
West	1.00 (ref)
Metropolitan statistical area
Yes	1.84 (1.09, 3.09)
No	1.00 (ref)
Physician Specialty
General/Family practice	1.00 (ref)
Internal Medicine	0.96 (0.53, 1.74)
Cardiology	0.74 (0.42, 1.33)
Neurology	0.14 (0.07, 0.28)
Other specialties	0.33 (0.20, 0.54)
Year
2011–2012	0.79 (0.47, 1.33)
2013–2014	0.80 (0.48, 1.32)
2015–2016	1.00 (ref)

Compared to men, women had 0.83 (95% CI: 0.58, 1.18) times lower odds of receiving diet counseling, despite women being affected by stroke more than men. Geographically, patients in the South had the highest odds of diet counseling (OR: 1.47; 95% CI: 0.81, 2.67) compared to the West. Patients in the Northeast (OR: 1.33; 95% CI: 0.73, 2.43) and Midwest (OR: 1.37; 95% CI: 0.74, 2.55) also had higher odds of receiving counseling than the West. Patients receiving services in a metropolitan statistical area had 1.84 (95% CI: 1.09, 3.09) times higher odds of receiving diet counseling than patients in a non-metropolitan statistical area.

When comparing physician specialties, there was no difference between Internal Medicine physicians and Family Physicians in the provision of diet counseling [OR: 0.96 (95% CI: 0.53, 1.74)]. Neurologists (OR: 0.14; 95% CI: 0.07, 0.28) were less likely to counsel patients on diet during office visits than Family Medicine physicians. Cardiologists (0.74; 95% CI: 0.42, 1.33) were more likely to counsel patients on diet during office visits than all other specialties aside from Family and Internal Medicine physicians

When adjusting for provider, practice, and patient factors in patients without stroke, the patients who had obesity only (OR: 4.35; 95% CI: 3.60, 5.27), hyperlipidemia only (OR: 2.27; 95% CI: 1.84, 2.79), and two or more risk factors (OR: 2.72; 95% CI: 2.39, 3.10) were more likely to receive diet counseling than the patients with no risk factors. There was no difference in the odds of diet counseling among patients with CAD alone (OR: 0.88; 95% CI: 0.61, 1.28) and hypertension (OR:0.88; 95% CI: 0.76, 1.02) compared to patients with no risk factors (Table 3).

Table 3:

Odds of Receiving Counseling by Stroke Risk Factors in Adults Without Stroke, 2011–2016

Characteristics	No Stroke (n=244,322)	Diet Counseling AOR* (95% CI)
Risk Factors
None	61.6± 0.005	1.00(ref)
CAD only†	0.7± 0.001	0.88(0.61, 1.28)
Diabetes only	1.8± 0.001	1.05(0.83, 1.33)
Obesity only	2.6± 0.001	4.35(3.60, 5.27)
HTN only‡	10.4± 0.002	0.88(0.76, 1.02)
Hyperlipidemia only	3.9± 0.001	2.27(1.84, 2.79)
Two or more	19.1± 0.004	2.72(2.39, 3.10)

^*Odds Ratios Adjusted for race/ethnicity, medical insurance, age, sex, geographic region, metropolitan statistical area, physician specialty, and year.

^†CAD: coronary artery disease (CAD).

^‡HTN: hypertension.

Discussion

Despite patients having a stroke and the data supporting the role of dietary modification in stroke risk, their odds of being counseled remained low and similar to not having a stroke. Further, the counseling rates did not improve over these years. We have observed that wide racial and regional differences in the odds of receiving dietary counseling, with those in metropolitan statistical areas having twice the odds than nonmetropolitan areas. Our analysis also showed that neurologists and physicians of other specialties were less likely to counsel patients on diet during office visits than family medicine physicians.

AHA guidelines echo that a diet high in fruits, vegetables, and whole grains is best in order to prevent future cardiovascular disease events²⁶. Despite the guidelines on the importance of a healthy diet for patients looking to prevent future cardiovascular events there are low counseling rates in patients with a history of stroke. Our results also revealed wide variation in the odds of receiving counseling by patient-, provider-, and practice- level characteristics. Among physician specialties, neurologists had the lowest odds of offering diet counseling, compared to family physicians. This discrepancy could be due to a lack of training in nutrition and behavior modification counseling techniques for secondary stroke prevention, lack of time, or lack of reimbursement. Assessing to understand the large discrepancy in Neurologist consultations compared to other relevant specialties is needed to learn how to properly address the differences seen.

Similarly, blacks and other non-Hispanic patients had 1.62- and 1.80-times higher odds of diet counseling compared to whites, respectively. Accounting for other variables, race seems to be a key determinant factor in whether a post-stroke patient receives diet counseling compared to other races. Previous research has also shown wide disparity in the adherence and perceived knowledge of diet in the US-based stroke belt²⁷ on age, sex, race, ethnicity, and education¹⁷. Our study findings may reflect cultural biases by clinicians regarding dietary practices among non-white Americans^28–30. The disparity in diet counseling based on race requires further exploration to understand whether bias in physicians or counseling guidelines is a determining factor.

Among patients with a history of a stroke aged 60–79, the proportion receiving diet counseling remained low through the study period (5.3% - 15.2%) despite the overwhelming evidence to its benefits. While in patients with a history of stroke over 80 years, the diet counseling proportions increased from 8.5% in 2012 to 12.1% in 2016. This highlights that 80+ year old patients are receiving less diet consults compared to their 60–79-year-old counterparts. There is evidence that shows older age groups are less likely to be motivated to follow diet and lifestyle counseling compared to the younger age groups³¹. One possible explanation is that physicians believe that patients over 80 years are less likely to benefit from or adhere to a diet lifestyle change given that they are late in their age. This disparity in the provision of counseling and the knowledge among those that are most vulnerable suggest the need for interventions to address this gap among older Americans.

Our findings demonstrate lower likelihood of counseling when seen by specialists. This may reflect deficient training in lifestyle changes including nutrition and dietary changes, time limitations during clinical encounters and lack of compensation for counseling. Evidence also shows a low rate of counseling is related to a lack of confidence among physicians in their ability to help patients make meaningful dietary changes, limited time, or inadequate educational materials³². These limitations could be significant in non-primary care specialties, where physical functionality may be the primary objective for post-stroke patients.

Strengths

Our study has several strengths. We used a nationally representative sample which provides a national snapshot of dietary counseling rates. The study has a large sample size and was able to measure the trends through the multiple years of data from 2011–2016. Lastly, measuring counseling rates by multiple physician specialties helped in identifying the patterns of counseling in those specialties.

Limitations

The general limitations of the NAMCS data apply to this study as well. The survey response rate decreased over the years, especially in the years 2014 – 2016 by over 10% compared to the previous years. The cross-sectional data design impedes identification of multiple visits of a patient, thus we may not be capturing the true number of patients who have received counseling from their clinician. The retrospective nature of these data could underestimate dietary counseling. The survey does not provide details on the provided diet counseling, including the quality, frequency, extent, and adherence. The change in the survey approach of NAMCS in 2012 from a paper-based survey to an electronic data collection system might have contributed to the significant drop of counseling rates from 2011 to 2012. The large decrease could be due to providers over-reporting the counseling rates before 2012 or recall bias. NAMCS published a quality control process analysis comparing 2012 estimates to the 2010 estimates. NCHS concluded that the significant change in 2012 could be due to either actual practice pattern changes or to the new sample design^23.

Moreover, NAMCS does not have data on characteristics that may influence the likelihood of receiving counseling, including patient characteristics such as education and socioeconomic status, or provider characteristics like sex, race/ethnicity, years in practice. NAMCS data is limited to patients seen in the clinicians’ office and doesn’t account for those that received outpatient nutrition visits. Lastly, our study examined the trends in diet counseling rates only in the population aged over 60 as the study sample for ages less than 60 was small. Hence, the trends are representative of the stroke population of all age groups, however, does represent that age groups that make up a majority of stroke patients.

Conclusions

Despite significant evidence from multiple studies emphasizing the importance of diet counseling among patients with a stroke, there has been little change in counseling rates from 2011 to 2016. Our study sheds light on counseling care gaps in ambulatory visits for patients with stroke, using a nationally representative sample. Current practice is for specialists such as neurologists to rely on primary care physicians for lifestyle counseling, however our data show that patients are not receiving counseling from either the primary care or specialist clinics, likely leading to this major gap in secondary stroke prevention treatment. Future studies should focus on addressing the highlighted provider-, patient-, and practice- level factors associated with the provision of diet counseling to address this critical gap on secondary stroke prevention, especially among the geriatric population since they have the highest morbidity rates due to stroke.

Fig. 1.

National trends in diet counseling among adults with and without stroke, 2011–2016.