Visit Frequency and Hypertension

Abstract

The best return visit interval to achieve blood pressure control is currently unknown. This study investigates the relationship between return visit interval and percent change in blood pressure. We reviewed a cohort of hypertensive patient charts from two large, urban family practice offices. Four hundred twenty‐nine patients with 7910 intervals showed a mean return visit interval of 79.5 days. Blood pressure control occurred during 34.5% of office visits. Pearson's r correlation coefficients between return visit interval and percent change in systolic and diastolic blood pressure demonstrated a small but statistically significant correlation. Shorter return visit intervals were associated with better percent changes in blood pressure. The return visit interval may be a simple and useful tool to improve management of hypertension.

“It is difficult to imagine a medical decision that is more common, has a greater impact on subsequent utilization of resources, and has less of an empirical foundation than the decision made at the end of every outpatient encounter: ‘When should this patient be seen again?’“ ¹

The visit frequency can be measured from the aggregate of a patient's office appointments. These appointments may have arisen for a variety of reasons occurring during each visit and specific to each patient. For example, if a patient's blood pressure (BP) is under control, then the patient may not need to be seen for a longer period of time. On the other hand, if a patient has uncontrolled BP, then medications may be changed, and a more prompt follow‐up appointment may be scheduled.

This commonly understood model of visit frequencies suggests that the return visit interval (RVI) is a result of individual factors for each encounter. These factors, which determine when a patient should (and does) return, are numerous. They have been the subject of multiple investigations, yet the factors studied have not been able to fully explain the variability in RVIs. ² , ³ , ⁴ , ⁵ , ⁶

Physicians and patients view follow‐up appointments as central to the management of chronic disease and not merely the result of other factors. Therefore, the follow‐up interval would partially be an independent variable that is not fully determined by other factors. This paradigm changes the question from “What factors influence visit interval?” to “Does altering the visit interval change the outcome of health or disease?”

BACKGROUND

This question seems basic, but there is limited scientific evidence to answer it. One situation where this question has been explored is the interval for monitoring warfarin levels. The conclusions from the warfarin‐monitoring studies are simple and profound. ⁷ , ⁸ If the patient is seen more, then more costs are incurred. If the patient is seen less, then less therapeutic warfarin levels are achieved. These same principles may be true for hypertension. The advantage of continuity of care in hypertension was shown in a 2004 study by Ornstein et al. ⁹ Continuity of care provides additional opportunities to reinforce and fine‐tune the treatment plan, as well as to educate the patient.

Studies of RVIs in hypertension have shown a wide range of opinions and practices: anywhere from 1 week to 1 year. ³ , ⁴ , ⁵ , ⁶ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ Several hypertension expert panels have made recommendations about appropriate RVIs.

In 1980, The Joint National Committee (JNC) on Detection, Evaluation, and Treatment of High Blood Pressure recommended follow‐up “as indicated at reasonable intervals…a few weeks to a few months or longer.” ¹⁷ The recommendations were based on committee opinion without references. The 1984 JNC report published a detailed table for follow‐up criteria based on specific BPs without references cited. ¹⁸ In 1992, the Canadian Medical Society ¹⁹ adopted guidelines similar to the JNC's, again without citing references. By 1994 the American Heart Association had adopted the same clinical guidelines, referencing the above sources. In 1997 the JNC VI recommended a 1‐2 month follow‐up after initiation of therapy and a 3–6 month interval after BP is controlled. ²⁰ The JNC 7 report ²¹ included a subsection in the text titled “Follow‐up and Monitoring” that stated:

…most patients should return for follow‐up and adjustment of medications at approximately monthly intervals until the BP goal is reached. More frequent visits will be necessary for patients with stage 2 hypertension or with complicating comorbid conditions…. After BP is at goal and stable, follow‐up visits can usually be at 3‐ to 6‐month intervals.

Again, no references were cited.

Hypertension studies have shown that over 50% of patients diagnosed with hypertension do not have their BP under control. ²⁰ , ²² , ²³ , ²⁴ Reasons investigated for this lack of BP control include poor patient compliance with medical recommendations, unrecognized causes of secondary hypertension, insufficient medication and/or doses, and “refractory” hypertension; however, no one has established whether RVIs affect BP control. This investigation addresses the association between RVI and change in BP.

This study hypothesizes that shorter RVIs following visits with uncontrolled BP are associated with greater percent decreases in BP, and that longer RVIs (within reason: less than 18 months) following visits with controlled BP are not associated with greater increases in BP.

METHODS

The Advocate Health Care Institutional Review Board approved this study on May 30, 2002.

Study Design

This study is a retrospective cohort study utilizing outpatient medical records.

Setting

The medical records reviewed were drawn from two urban family practice centers, each averaging 15,000–20,000 office visits per year, staffed by family practice residents and attending physicians. The office visits included nurse visits, but more than 95% were physician visits. These appointments, ranging from 15 to 60 minutes, included all acceptable forms of treatment for hypertension, including pharmaco‐therapy; diet, exercise, and weight loss counseling; relaxation techniques; and patient education. At the time of the study, the two offices had been operating for 18 and 9 years. Both offices served a diverse population of Hispanics, Caucasians, African Americans, and Asians, with wide ranges of educational level and socioeconomic status among their patients.

Inclusion and Exclusion Criteria

Patients were included if they were assigned the diagnosis of hypertension on administrative billing data (based on International Classification of Diseases, Ninth Revision codes 401.1 or 401.9) for an office visit between January and July 2001 and confirmed by record review. They also had to have been a patient at the office for more than 1 year, with at least two office visits. Every clinical visit at both sites routinely included checking the patient's BP. All of the patients’ encounters for their entire time at the practice were included. Missed appointments and cancellations were not included. Intervals were excluded from analysis if they were greater than 540 days, i.e., 18 months.

Data Collected

Data collected included visit dates, BPs, other diagnoses, medication lists, age, sex, duration of time that the patient had been in the practice, number of visits with each particular physician, insurance type, family history of hypertension, smoking history, and alcohol history.

Defining Terms

An RVI was defined as the number of days between consecutive office visits. Each RVI was tagged as a “controlled” interval or “uncontrolled” interval based on the BP recorded at the start of the interval. Controlled BP was defined as systolic BP (SBP) <140 mm Hg and diastolic BP (DBP) <90 mm Hg.

The measured outcome was the percent change in BP across each interval ([final BP ‐ initial BP] / initial BP). This percent change in BP across intervals was then compared with the length of the RVI. The RVIs were dichotomized by controlled or uncontrolled status and further divided by the number of days in the interval. The designated days in the groupings of RVIs, based on published recommendations and commonly used intervals, were 1–30, 31–60, 61–90, 91–120, 121–180, 181–360, and >360 days. ³ , ⁴ , ⁷ , ⁹ , ¹⁶ , ¹⁷

Sample Size Calculation

The sample size calculation was based on patients as the observational unit. To detect a 5 mm Hg change in SBP and DBP using a standard deviation of 16.0 mm Hg and 10.0 mm Hg, respectively, ²⁵ we estimated that we would need approximately 170 subjects to detect a change in SBP and 70 subjects for a change in DBP ²⁶ (β=0.20; power=80%; α=0.05). If one half of the patients had uncontrolled BP, then we would need to have a sample of 340 patients to have adequate power to detect a change in SBP.

Data Analysis

All BP data are reported as mean ± SD. Student t tests on the percent change in the BPs were performed on each group against all intervals longer than that group (e.g., percent change in BP for intervals of 1–30 days were compared with percent change in BP for all intervals >30 days). Separate analyses were done for SBP and DBP. Pearson's r correlation coefficients between RVI and percent change in BP were calculated for uncontrolled/controlled RVIs and for DBP/SBP. A two‐tailed p value of 0.05 was considered statistically significant in all analyses. Analyses were performed with SPSS software (release 11.5, SPSS Inc., Chicago, IL).

RESULTS

Four hundred twenty‐nine charts were reviewed (232 from one office and 197 from the other), representing a total of 8412 office visits. Total number of RVIs was 7983. All RVIs longer than 540 days (n=73) were eliminated from the analysis, for a total of 7910 RVIs. Mean number of visits per patient ± SD was 18.44±18.37.

Table I illustrates demographic data for the entire group. Including all visits, SBP was 141.0±12.5 mm Hg and DBP was 85.5±6.8 mm Hg. For all patients, on their first visit of the series, SBP was 149.3±20.2 mm Hg and DBP was 91.4±12.0 mm Hg. For all patients, on their last recorded visit, SBP was 137.3±18.7 mm Hg and DBP was 82.3±10.8 mm Hg. BP was controlled (<140/90 mm Hg) on 2902 (34.5%) visits. Of the 429 patients, 122 (28.4%) had their BP controlled for more than 50% of their visits.

Table I.

Demographics (N=429)

Age (y) (mean ± SD)	54.7±14.3
Gender (n [%])
Women	271 (63)
Men	158(37)
Family history of hypertension (n [%])
Yes	185 (43)
Unknown	103 (24)
No	82 (19)
Missing data	59 (14)
Alcohol (n [%])
No	259 (60)
Social	137 (32)
Problem	23(5)
Missing data	10(2)
Tobacco (n [%])
No	262 (61)
Yes	102 (24)
Former	56(13)
Missing data	9(2)
Insurance type (n [%])
Health maintenance organization	161 (38)
Medicare	103 (24)
Other	85 (20)
Preferred provider organization	52 (12)
Medicaid	22(5)
None	6(1)

For all visits, the mean RVI was 75.7+81.8 days (median, 45 days). Uncontrolled BP was seen at the start of 5209 intervals. The average RVI for these uncontrolled intervals was 71.2±80.8 days. Controlled BP was seen at the start of 2701 intervals. The average RVI for these controlled intervals was 84.1±82.8 days. The mean RVI was significantly less for uncontrolled BP visits vs. controlled BP visits (p<0.001). Table II summarizes the RVI categories with the number and percent of visits in each group.

Table II.

Number and Percent of Uncontrolled and Controlled Intervals in Each Return Visit Interval (RVI) Category

RVI (Days)	Uncontrolled (n [%])	Controlled (n [%])
≤30	2048 (39.3)	790 (29.2)
31–60	1182 (22.7)	609 (22.5)
61–90	630 (12.1)	392 (14.5)
91–120	453 (8.7)	319(11.8)
121–180	422 (8.1)	254 (9.4)
181–360	380 (7.3)	294 (10.9)
>360	94 (1.8)	43 (1.6)
All	5209 (100.0)	2701 (100.0)

Table III shows the actual SBP and DBP (± SD) at the start of the intervals for uncontrolled and controlled RVIs for each RVI group.

Table III.

Mean Systolic and Diastolic Blood Pressure (BP) at the Start of Each Interval, by Return Visit Interval (RVI)

	Uncontrolled		Controlled
	Systolic BP	Diastolic BP	Systolic BP	Diastolic BP
RVI (Days)	(Mean± SD) (mm Hg)	(Mean± SD) (mm Hg)	(Mean± SD) (mm Hg)	(Mean± SD) (mm Hg)
≤30	147.4±20.3	85.6±10.8	131.5±15.3	80.1±9.4
31–60	146.1±19.8	84.9±10.6	131.6±16.0	80.5±9.8
61–90	144.9±19.1	84.1 ± 10.9	134.1±16.9	81.6±9.3
91–120	144.3±17.9	84.8±10.2	133.8±15.6	81.8±9.6
121–180	145.5±19.0	85.7±10.9	132.2±15.5	81.6±9.1
181–360	148.5±19.7	88.0±19.7	134.1±16.8	83.6±9.5
>360	148.6±19.7	88.1±10.3	136.1±18.7	84.9±9.6

For intervals where BP was uncontrolled, Pearson's r correlation coefficients calculated between RVI and percent change in SBP and DBP were 0.108 (p<0.001) and 0.086 (p<0.001), respectively. For intervals where BP was controlled, Pearson's r correlation coefficients calculated between RVI and percent change in SBP and DBP were 0.066 (p=0.001) and 0.085 (p<0.001), respectively. These Pearson's r correlation coefficients indicate a small but real correlation between RVI and percent change in BP for all categories.

Table IV (uncontrolled intervals) and Table V (controlled intervals) summarize each RVI group by mean percent change in SBP and DBP. Student t tests compare the percent change in BP for the indexed RVI group against all RVIs of greater length.

Table IV.

Mean Change in Blood Pressure (BP) by Return Visit Interval (RVI) for Uncontrolled Intervals (%)*

	Systolic BP		Diastolic BP
RVI (Days)	Change (%)	p Value	Change (%)	p Value
≤30	−4.2±12.1	<0.001	−3.6±12.1	<0.001
31–60	−3.1 ± 12.5	0.001	−2.6±12.6	0.002
61–90	−2.8±12.5	0.003	−1.5±12.2	0.342
91–120	−1.7±11.8	0.153	−1.2±11.8	0.576
121–180	−2.0±11.8	0.002	−1.5±12.8	0.121
181–360	0.64±13.0	0.700	0.03±13.7	0.683
>360	0.07±11.4		−0.61±12.9
*Based on percent change in BP of indexed RVI vs. all RVIs longer than the indexed RVI

Table V.

Mean Change in Blood Pressure (BP) by Return Visit Interval (RVI) for Controlled Intervals (%)*

	Systolic BP		Diastolic BP
RVI (Days)	Change (%)	p Value	Change (%)	p Value
≤30	6.0*±13.0	0.015	4.3±12.8	0.002
31–60	6.5±13.8	0.057	4.9±14.5	0.017
61–90	7.6±14.0	0.748	6.8±13.4	0.747
91–120	7.6±13.0	0.612	5.7±12.8	0.201
121–180	6.9±12.7	0.079	5.3±12.7	0.014
181–360	8.8±14.0	0.573	7.9±14.6	0.556
>360	10.1±15.1		9.3±11.3
*Based on percent change in BP of indexed RVI vs. all RVIs longer than the indexed RVI

Visits following uncontrolled BP (Table IV) showed a statistically significantly greater decrease in SBP for shorter interval groups, up to 90 days. For RVIs of 90‐180 days, SBP had a notable percent decrease, but it was not statistically significant vs. longer intervals. For RVIs over 180 days, the mean SBP actually increased.

DBP decreased significantly for “uncontrolled” intervals of less than 60 days. DBP also had a mean percent decrease for RVI groupings of 61–80 days, but it did not reach statistical significance vs. longer intervals. RVIs over 180 days for DBP in the uncontrolled groups either had a percent increase in DBP or a modest decrease.

Similarly, visits following controlled BP (Table V) showed a statistically significantly lesser increase in SBP for shorter interval groups—up to 60 days. For RVIs of 60‐180 days, SBP had a notably lesser percent increase, but it was not statistically significant vs. longer intervals. The trend of greater percent increase in SBP following a longer RVI persisted throughout the series for controlled intervals.

DBP for controlled intervals had a statistically significantly lesser percent increase following intervals of less than 60 days. DBP also had a lower mean percent increase for RVI groupings under 180 days, compared to longer RVI groupings, but it only reached statistical significance for RVIs of 121–180 days vs. longer intervals. RVIs over 180 days in the controlled groups continued to show ever‐greater percent increases in DBP.

DISCUSSION

Consistent with other published data, the results showed the low overall rate of BP control in treated populations. ²⁰ , ²² , ²³ , ²⁴ Also consistent with other published data, ³ , ⁴ , ⁵ , ⁶ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ this study showed the wide range of RVIs. Additionally, our dataset demonstrated a positive correlation between shorter RVIs and percent change in BP.

The results show a small(about 1%, or 1–3mmHg) but clinically significant change in BP across RVI groupings. The Prospective Studies Collaboration, ²⁷ a quantitative overview of 61 cohort studies with more than one million enrolled subjects, demonstrated that small gradients in SBP or DBP account for sizable differences in cardiovascular outcomes. Additionally, published outcome trials of patients at high cardiovascular risk proved that reductions in SBP as small as 1–3 mm Hg decrease the relative risk of stroke by as much as 20%–30%. ²⁸

The observational nature of this cohort study is a significant limitation. First, confounding variables may exist. We tried to minimize their effect by dividing the intervals into controlled and uncontrolled BP groups and by allowing patients to be represented in multiple RVI groupings. Second, while it is intuitively sensible that increased visit frequency may lead to better control of BP, this study cannot establish cause and effect. Nor can these results establish the correct visit frequency.

Despite these concerns, this study does suggest an association between BP change and RVI, and therefore, a need for further inquiry into this association. Chapko et al. ¹ raised concern about the acceptability of randomizing patients to various RVIs. The results of this study should facilitate the acceptance of randomizing hypertensive patients to shorter RVIs vs. usual care.

We designed this study with the sole outcome measure of BP control or improvement; however, we found this measure to be inadequate because BP can be controlled or uncontrolled from visit to visit. We addressed this limitation by reporting our results in terms of intervals. In future studies, continuous BP control should be measured over a set period of time. Additionally, using BP control as the sole outcome minimizes the full benefit of the “treatment” of hypertension—which continues below the target BP and incorporates other health issues besides BP (i.e., lipid testing, weight reduction, exercise promotion, education, and social support).

CONCLUSIONS

RVI may be a proxy for increased intervention (pharmacotherapy or education) or a specific behavior of the physician or patient, or it may act to strengthen the doctor‐patient relationship. The exact reason why a shorter RVI is associated with greater improvement in BP remains to be defined.

This study suggests that the RVI may be a tool in the management of hypertension. Altering the intervals over which we see patients may be a simple but useful method to improve outcomes in hypertension.

Acknowledgments and disclosures: This study was presented as a poster at the North American Primary Care Research Group Annual Conferences, November 17–20, 2002, New Orleans, LA; and October 25–28, 2003, Banff, Alberta, Canada. This study was supported by a research stimulation grant from the American Academy of Family Physicians Foundation (G0207RS). Special thanks to Marina Panopoulos for her assistance in data collection and to Fangxi Zhou for her assistance with data organization.