Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Too much noise
- Roderick E Warren, Paul L Padfield, Consultant Physician and Professor of Hypertension (5 May 2009)
-
A hypothesis in need of testing
- Christopher E Clark (5 May 2009)
-
Blood pressure monitoring after stroke
- Rizaldy Pinzon (22 May 2009)
-
Generalisability to primary care
- Mark R Nelson, Stephen Quinn (29 June 2009)
-
Authors' response
- Andrew Hayen, Katherine Keenan, Bruce C Neal, Les Irwig (21 July 2009)
|
|
|||
|
Roderick E Warren, Specialist Registrar Metabolic Unit, Western General Hospital, Edinburgh EH4 2XU, Paul L Padfield, Consultant Physician and Professor of Hypertension
Send response to journal:
|
Keenan et al have nicely demonstrated the futility of re-assessing blood pressure using a single repeat measurement after a short time interval. For example, if a patient's true systolic BP at baseline is 130mmHg, and hypertension is defined as SBP>140mmHg, a single measurement at 3 months will yield 93 false positives for every correct diagnosis of new-onset hypertension. They find that the ratio of false positives to true positives falls as the interval increases. Their statistical analysis is erudite, but it seems to us that this is because the "true underlying" BP is more likely to have changed over the longer interval, and so the prior probability of new-onset hypertension is greater. However, their estimate of the long-term variability of "true" BP seems remarkably high. In Table 2, the SD at 33 months is almost 12mmHg for systolic BP. This appears to indicate that over 33 months, 32% of patients had an increase or decrease of at least 12mmHg, and 5% had an increase or decrease of at least 24mmHg. Clearly these rates of change cannot be sustained, as it would not be very many years before the BP became incompatible with life. Thus, what the authors suggest is "true" BP must itself be fluctuating. A major difficulty is knowing what BP we are interested in. The most helpful approach we can suggest is to consider the frequency with which the typical clinician is prepared to titrate an antihypertensive drug regimen up or down for the purposes of cardiovascular risk protection, recognising that age-related BP change is only of the order of 1mmHg per year. We believe that no or very few clinicians would aim for daily dose adjustment according to that day's blood pressure. We strongly suspect that most clinicians would ignore some week-to-week variation, recognising that people have "good weeks" and "bad weeks". Perhaps the typical clinician would be interested in the average BP over a month or so? Yet there is good reason to expect BP to fluctuate from month to month; for example, it is affected by the outside air temperature1. We suspect the typical clinician would not prescribe drugs for 6 months of each year to a patient whose "hypertension" regularly settles in summer and returns in winter. It seems that we are interested in the average BP over a year or so, assuming BP levels are at target for an individual. Keenan et al have calculated a theoretical instantaneous variance in BP, and subtracted this from overall variance. By inference, the remaining long-term variance represents meaningful change in BP. We suggest that this approach leaves too much random noise in the long-term estimate, and that meaningful change is what remains after around 12 months' variance is subtracted. Consistent with this, Tables 2 & 3 show little additional variance over 9-33 months. Meaningful long-term BP is probably less variable than Keenan et al have estimated. This means that the signal-to-noise ratio in a single BP measurement is likely to remain poor, even after 33 months. We agree that a policy of delayed reassessment is likely to generate fewer false positives than early reassessment, but argue that a single BP measurement is never adequate. 1. Alperovitch A et al. Relationship Between Blood Pressure and Outdoor Temperature in a Large Sample of Elderly Individuals. The Three-City Study. Arch Int Med 2009; 169: 75-80. Competing interests: None declared |
|||
|
|
|||
|
Christopher E Clark, GP School Surgery, Fore St, Witheridge, Devon, EX168AH
Send response to journal:
|
This is an interesting paper which adds to the uncertainty of interpretation that a single blood pressure reading enjoys. We already know that there are many potential sources of error in collecting a single reading. However this is really a hypothesis generating paper, that suggests the likely false positive value of a single elevated reading based on statistical estimation. It cannot tell us the actual false positive rates as there were no "gold standard" data available for comparison. Therefore further validation studies would be needed to test this hypothesis before it could truly suggest any evidence based enhancement of six monthly or other regular blood pressure monitoring. Current GP practice in the UK is not to change treatment on the basis of a single elevated reading anyway, but rather to regard it as a trigger for further recordings to try and obtain a true estimate of current blood pressure and determine whether it has truly risen. This paper does not justify an overhaul of existing clinical practice or the QOF. It is worth making sure that errors are minimised by allowing sufficient time for the patient to relax, ensure that measurement is undertaken according to current guidance (1), that the correct arm is chosen for measurement(2), and that the data are recorded accurately without rounding (3). These are all evidence-based changes that can be adopted now, whilst the jury is out. Refs: 1. Williams B, Poulter NR, Brown MJ, Davis M, McInnes GT, Potter JF et al. Guidelines for management of hypertension: report of the fourth working party of the British Hypertension Society, 2004-BHS IV. J Hum.Hypertens. 2004;18:139-85. 2. Clark CE, Campbell JL, Evans PH, Millward A. Prevalence and clinical implications of the inter-arm blood pressure difference: a systematic review. J Hum.Hypertens. 2006;20:923-31. 3. Carey IM, Nightingale CM, DeWilde S, Harris T, Whincup PH, Cook DG. Blood pressure recording bias during a period when the Quality and Outcomes Framework was introduced. Journal Hum. Hypertens. advance online publication,12 March 2009; doi:10.1038/jhh.2009.18 Competing interests: None declared |
|||
|
|
|||
|
Rizaldy Pinzon, Neurologist Bethesda hospital Yogyakarta Indonesia 55224
Send response to journal:
|
The study from Keenan et.al. showed that the result of blood pressure value from long term monitoring is poor. The result is different from the previous study that showed blood pressure monitoring is useful for predicting stroke. Systolic blood pressure has a very strong predictive power for stroke (The Ohasama study, 2006). The other study from Ohkubo et.al. (2000) showed in a likelihood ratio analysis that these ambulatory blood pressures were significantly better related to stroke risk than screening blood pressure, and that daytime blood pressure better predicted stroke risk than did night-time blood pressure. Blood pressure monitoring should be used with caution. After the diagnosis of hypertension, home BP monitoring is a useful supplement, but home BP values should not override clinic BP values. Home BP monitoring should be encouraged in order to provide more information for clinician decision and improve adherence to treatment regimens, but should be discouraged whenever it causes patients anxiety or induces self-modification of the treatment regimen. Home BP monitoring should be used under medical supervision. The issue is more relevant in secondary stroke prevention. Previous study showed that found BP-lowering drug interventions reduced the risk of stroke recurrence. Many believe that "lower is better" as a goal in reducing stroke recurrence risk. This should be closely monitored. Competing interests: None declared |
|||
|
|
|||
|
Mark R Nelson, Senior member Menzies Research Institute, University of Tasmania, Private Bag 33, Hobart 7001, Tasmania, AUSTRALI, Stephen Quinn
Send response to journal:
|
Keenan et al make a well reasoned argument that the ratio of false to true positives at say 6 months is so high that modifying blood pressure management on the strength of one careful reading is likely to be done on the result of a false positive and not a true reading(1). However this is a post hoc analysis of a secondary prevention trial that makes generalising these results to primary care, where most blood pressure is managed, problematic. Here the standard of blood pressure measurement never approaches that of a clinical trial(2). Here also treatment decisions are at the discretion of the physician based on recommended repeated measurements both within and outside the clinic rather than a forced titration schedule and careful single visit measurement(3). The authors do not comment on regression to the mean which is a factor at play here. To quote from one of their own references “..but the average of measurements for an individual person within that population is a biased estimate of his true value. His measured average must be adjusted for random fluctuations in a manner analogous to correcting for regression to the mean” (page 200) (4). We agree with the assertion that a single high BP measure taken at 3 or 6 months after blood pressure has reached a specified target is likely to be a false positive is correct. However we would have thought the message is, given the unreliability of a single reading within a short period, that repeated readings are required to verify blood pressure reliably rather than extending the period between measurements. Also, it is a pity that the authors did not present a table on the ratio of false to true positives for observed hypertensive values of BP above the thresholds 140/90mmHg and various review times, for example P(T6<90, O¬6>95)/ P(T6>90, O¬6>95). This information would enable the clinician to better assess the true state of hypertension of a patient. References 1. Katherine Keenan, Andrew Hayen, Bruce C Neal, and Les Irwig Long term monitoring in patients receiving treatment to lower blood pressure: analysis of data from placebo controlled randomised controlled trial BMJ 2009; 338: b1492 2. Nelson MR, Quinn S, Bowers-Ingram L, Nelson JM, Winzenberg T. Cluster randomised controlled trial of oscillometric versus manual sphygmomanometer for blood pressure management in primary care (CRAB). Am J of Hypertens 2009; doi:10.1038/ajh.2009.55 3. National Heart Foundation of Australia (National Blood Pressure and Vascular Disease Advisory Committee). Guide to management of hypertension 2008. 4. Shepard DS. Reliability of blood pressure measurements: implications for designing and evaluating programs to control hypertension. J Chronic Dis 1981;34(5):191-209. Competing interests: None declared |
|||
|
|
|||
|
Andrew Hayen, Senior lecturer School of Public Health, University of Sydney, NSW 2006, Australia, Katherine Keenan, Bruce C Neal, Les Irwig
Send response to journal:
|
We thank all the authors of rapid responses. As several point out, the issue is one of ‘signal’ [how much true increase in BP there is over time] and ‘Noise’ [ short-term within-person variability]. Drs Warren and Padfield have suggested that our estimate of true long-term change in blood pressure may be too large. We agree with Drs Clark, Warren and Padfield that steps should be taken to reduce systematic errors in blood pressure, including the factors they mention, and indeed others, such as the calibration of sphygmomanometers (1). However, as Drs Nelson and Quinn point out - standardisation in randomised trials such as that which provided the data for our paper is likely to be better than in clinical practice - yet there remains substantial within-person variability. Putting together the comments above that we may have overestimated the true variability and underestimated the background within-person variability, the FP:TP ratio in practice may be still poorer than we suggested in our paper. We have suggested less frequent monitoring in our paper as a way of allowing sufficient time between measurements for there to have been more ‘signal [a true change in BP]. Several authors of the rapid responses point out that another approach is to take multiple measurements at clinic visits, or by home or ambulatory BP measurements, to reduce the ‘noise’. That is true and potentially valuable in some patients, but certainly not for all. In table 4 in our paper, we point out that less than 5% of people with a true baseline on-treatment systolic BP of 130mmHg –and only about 1% for Systolic PB of 120mm Hg were estimated to have increased to 140mmHg by 6 months. While we agree that repeat measurements can be used to get better estimates of an individual's true blood pressure, doing the very large number of measurements to detect the small proportion of patients whose BP has truly increased may not be worthwhile. These remarks about long-term-monitoring to detect drift over time as dealt with in our paper also apply to initial response monitoring, which asks the question: what has been the initial response to treatment within a few weeks or a couple of months. For perindopril and indapamide, we have shown that there is likely to be very little variability in response to treatment so that initial response monitoring is not worthwhile (2). References (1) Turner MJ, Irwig L, Bune AJ, Kam PC, Baker AB. Lack of sphygmomanometer calibration causes over- and under-detection of hypertension – a computer simulation study. Journal of Hypertension 2006; 24: 1931-1938 (2) Bell KJL, Hayen A, Macaskill P, Craig JC, Neal BC, Irwig L. Mixed models showed no need for initial response monitoring after starting anti- hypertensive therapy. Journal of Clinical Epidemiology. 2009; 62(6): 650-9 Competing interests: Competing interests: BN has received honoraria from Servier for speaking at scientific meetings. Servier provide research support for the ADVANCE trial of which BN is a management committee member |
|||