Intended for healthcare professionals

Letters

Audit of diagnosis and management of hypertension in primary care

BMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7103.314a (Published 02 August 1997) Cite this as: BMJ 1997;315:314

Interpractice variation in prevalence of hypertension is due to inadequate detection

  1. Kevin Allan, Health information managera,
  2. Paul Murphy, MEDICS project facilitatora,
  3. Stephen Singleton, Director of public healtha,
  4. Richard Edwards, Lecturerb
  1. a Northumberland Health Authority, Morpeth NE61 2PD
  2. b Department of Epidemiology and Public Health, Newcastle University Medical School, Newcastle upon Tyne NE1 7RU
  3. c Prescribing Research Group, Department of Pharmacology and Therapeutics, The Infirmary, Liverpool L69 3GF
  4. d Chelsea and Westminster Hospital, London SW10 9NH

    Editor—An interpractice audit of the diagnosis and management of hypertension in primary care found that hypertension remained uncontrolled in over three fifths of hypertensive patients despite the implementation of an intensive audit; this is disappointing.1 Another common deficiency in the management of hypertension—the failure of health care services to detect and diagnose the condition—was not, however, addressed.

    Data from table 1 in the paper can be used to calculate the crude prevalence of hypertension in individual practices. This varied between 1.61% and 5.01% (mean 3.2%) during phase 1 of the study, which suggests that the detection of hypertension varied widely. This is confirmed by data from the MEDICS (morbidity and epidemiology data interchange and comparison scheme) project in Northumberland. Prevalence data for common chronic diseases have been collected from computerised records at 33 practices covering a population of 200 000 people. In March 1995 the crude prevalence of recorded hypertension in adults (aged >15) varied from 3.0% to 13.2% (mean 6.7%)—from 2.6% to 10.9% (mean 5.5 %) in men and 3.4% to 15.6% (mean 7.9%) in women. This variation persisted when data were indirectly standardised for age and expressed as standardised morbidity ratios.

    MEDICS practices should probably be better than most at detecting and recording hypertension because they are all computerised, provide data on the prevalence of hypertension for the health promotion banding scheme, and participate in a project collecting morbidity data. They are also situated in a district where improving the care of patients with hypertension has had a high priority (through initiatives such as district-wide audits of hypertension and the local development and dissemination of best practice guidelines).

    Despite this, MEDICS practices had difficulty detecting and recording hypertension: the mean prevalence of hypertension after direct standardisation for age was 6.0% for men and 7.5% for women, compared with 19.3% and 18.4% respectively in the national health survey (which was standardised to the European standard population).2 One practice, however, increased its recorded prevalence of hypertension by 62% (from 4.4% to 7.1%) over four months by searching computer records for those of patients prescribed antihypertensive drugs or with high blood pressure but no diagnosis of hypertension.

    Good management of hypertension requires systematic identification of cases and structured review and intervention over long periods.3 The evidence of interpractice variability in recorded hypertension, a reduced prevalence of hypertension recorded in practice computerised databases compared with national data, and the ability to identify more hypertensive patients from practice computerised records suggest that future audits should address adherence to best practice in detecting and recording hypertension.

    References

    1. 1.
    2. 2.
    3. 3.

    Definition of uncontrolled blood pressure used in study is unclear

    1. Stuart Barton, Senior lecturer in primary carec,
    2. Mike Cranney, General practitioner-research fellowc
    1. a Northumberland Health Authority, Morpeth NE61 2PD
    2. b Department of Epidemiology and Public Health, Newcastle University Medical School, Newcastle upon Tyne NE1 7RU
    3. c Prescribing Research Group, Department of Pharmacology and Therapeutics, The Infirmary, Liverpool L69 3GF
    4. d Chelsea and Westminster Hospital, London SW10 9NH

      Editor—Mahendra Mashru and Ariel Lant present interpractice audits of the management of hypertension in primary care, performed before and after educational activity.1 Their study contains methodological problems which cast doubt on their conclusions.

      The study did not have a control group, yet random allocation to intervention and control groups is required to assess the quantitative impact of education.2 3 Only one partner in each practice formally received the educational message, and the authors relied on unmonitored diffusion of education within the group practices. The educational message may have been weakened further by the major and potentially confusing differences between the three guidelines.

      It is unclear what time period was examined to determine whether three records of blood pressure were present in either the phase 1 or phase 2 audit. The number of readings used to make a diagnosis of hypertension is itself a poor guide to the quality of the diagnosis—for example, for practices that use only the last record of blood pressure to make the diagnosis the presence of two earlier, normal, records is irrelevant.

      The definition of uncontrolled blood pressure used in the study is unclear, defined both as “not lowered to <160/90 mm Hg” and “>160/90 mm Hg.” Patients with a blood pressure of exactly 160/90 mm Hg are handled inconsistently. As significant digit preference exists in the recording of blood pressure4 many patients may have had single readings of 160/90 mm Hg.5 Furthermore, the number of readings used to assess control was not stated. The apparently poor impact of education may merely reflect the effect of digit preference in reducing the responsiveness of the measure of control.

      The timing of the audits is also suspect. Because the initial planning occurred before phase 1, yet may, as the authors state, have contributed to the educational intervention, an unknown improvement may already have occurred before the phase 1 audit. The change presented—that between the phase 1 and phase 2 audits—may be only part of the change produced by the intervention. Furthermore, we do not know the precise interval between the two audits (the audits could have been consecutive in some practices).

      Primary care may be performing better than predicted by the rule of halves, and education may be helpful in producing improvement. Mashru and Lant's study does not really help us to decide.

      References

      1. 1.
      2. 2.
      3. 3.
      4. 4.
      5. 5.

      Authors' reply

      1. Mahendra Mashru, North West Thames regional research fellow in general practiced,
      2. Ariel Lant, Professor of clinical pharmacology and therapeuticsd
      1. a Northumberland Health Authority, Morpeth NE61 2PD
      2. b Department of Epidemiology and Public Health, Newcastle University Medical School, Newcastle upon Tyne NE1 7RU
      3. c Prescribing Research Group, Department of Pharmacology and Therapeutics, The Infirmary, Liverpool L69 3GF
      4. d Chelsea and Westminster Hospital, London SW10 9NH

        Editor—Kevin Allan and colleagues' comments that there is interpractice variation in the prevalence of hypertension must be correct, but the assumption that it is due largely to inadequate detection of hypertension seems an oversimplification. We do, however, agree that good management of hypertension requires systematic identification of cases and structured clinical review and that future audits should address adherence to best practice in detecting and recording hypertension.

        We agree with Stuart Barton and Mike Cranney's suggestion that a control group would have allowed appraisal of the quantitative impact of education by random allocation, and we pointed this out in our paper. We disagree, however, that the quantitative impact of education cannot be determined in the type of study design that we used; the study did not have parallel groups but looked instead at individual practice behaviour before and after clearly defined educational interventions. The educational method researched in our study was based on the concept of the relay of messages by one practice partner to the rest of the clinical team. Our conclusions therefore can relate only to this technique.

        The three nationally recognised guidelines for the detection and management of hypertension do have differing messages. However, we specifically selected domains for our study after consensus was established between us and the participating practices at the outset. Barton and Cranney claim that the paper is unclear about what we defined as hypertension. We stated that the diagnosis depended on three separate readings of blood pressure >160/90 mm Hg. In reality this demanded a set of three abnormal values >160/90 mm Hg. Digit preference is important, and we have reported on it previously.1 We do not think, however, that this factor would have had a significant impact on the overall conclusions of our study.

        We do not agree with Barton and Cranney about the timing of the study as this was precisely controlled. The sequencing of practices entering phase 1 or 2 was rigidly adhered to by careful timetabling to avoid variability.

        We agree with Barton and Cranney's suggestion that education ought to be helpful in producing improvement in the management of blood pressure. We would, however, emphasise the final sentence in our paper: not enough research has been undertaken to determine which particular types of educational method would be most effective in changing clinical behaviour.

        References

        1. 1.

        Log in

        Log in through your institution

        Subscribe

        * For online subscription