Family Practice Vol. 18, No. 1, 33-38
© Oxford University Press 2001
Research in Practice |
Studies of symptoms in primary care
Department of Primary Care, Rehabilitation and Preventive Medicine, University of Marburg, Blitzweg 16, D-35033 Marburg,
a Department of Medicine—Nephrology Charité, Humboldt-University of Berlin, Germany and
b Department of Family and Community Medicine, University of Toronto, Ontario, Canada.
Donner-Banzhoff N, Kunz R and Rosser W. Studies of symptoms in primary care. Family Practice 2001; 18: 33–38.
Received 2 February 2000; Revised 13 June 2000; Accepted 5 September 2000.
 |
Abstract |
|---|
 Top Abstract IntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
Publications on the frequency of defined symptoms in the practice setting, underlying conditions and prognosis have been rare in the past. Also, studies addressing these questions have suffered from several methodological problems. We therefore developed criteria to help investigators improve the quality of study design, implementation and publication. Studies evaluating symptoms in practice can make an important contribution to a more rational approach to diagnostic decision making especially in primary care.
Keywords. Differential diagnosis, prevalence, primary health care, prognosis, signs and symptoms.
|
Introduction |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
Several series of articles dealing with the critical appraisal of the medical literature1–3 have been published. While studies of treatments, side effects, causation, prognosis, diagnostic tests and overviews/meta-analyses have been discussed thoroughly, there is a particular type of study that has not received sufficient attention in the past, i.e. studies examining patients presenting with a defined symptom in health care settings. In particular, the questions that these studies aim to answer are
- (i) How often do patients with the respective symptom present (prevalence/incidence)?
- (ii) What are the underlying conditions and their respective frequencies (differential diagnosis)?
- (iii) What is the prognosis of patients presenting with the symptom?
- (ii) What are the underlying conditions and their respective frequencies (differential diagnosis)?
Research reports studying symptom presentations are especially relevant for primary care. In this setting, symptoms and findings usually have not been evaluated by other health professionals, and practitioners have to consider a wide range of conditions from minor self-limiting disease to life-threatening somatic or psychic disorders. The undifferentiated nature of presentations at the primary care level often precludes any specific diagnosis at all, so that a watchful-waiting approach is justified.4 However, potentially serious and treatable conditions have to be identified reliably. Studies examining defined symptoms can help health professionals adopt a rational and efficient approach to complex diagnostic problems, they inform patient counselling with regard to prognosis and the probability of relevant diseases and they reduce uncertainty in frequently occurring clinical situations. The purpose of this paper is to provide some guidance for investigators to improve study designs and published reports in this important area. We will also discuss potential weaknesses and bias threatening the validity of studies exploring symptoms or clusters of symptoms and their respective aetiologies.5
|
Selecting patients |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
If medical practitioners are asked retrospectively to identify subjects fulfilling study criteria, they tend to remember the severe, bizarre or frustrating rather than the usual cases. To avoid this kind of selection bias,6,7 subjects have to be recruited prospectively.
Busy practitioners tend to overlook eligible patients (under registration bias). Checking against practice records will at least show the proportion of patients missed. While investigating the incidence and outcome of rectal bleeding within a network of Dutch Family Physicians,8 the investigators noticed a smaller number of included patients with the symptom than expected. Therefore, they started a second arm with intense monitoring and a modified study design to reduce the number of under-registered patients. The incidence rate of rectal bleeding calculated from the second arm turned out to be three times higher than that from the original design. Intensive monitoring is therefore required to make recruitment as complete as possible. Authors should report the proportion of non-participants.
Practice records can be used to identify patients whose symptoms have been recorded but who have not been included in a particular study. However, studies should not rely exclusively on retrospective analyses of hospital and/or practice records since routine clinical documentation is not dependable as a reliable sampling method for studies of the kind discussed here.
In assessing a study, one should ask "Are subjects recruited prospectively and are all consecutively presenting patients included?"
|
Inclusion criteria |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
The majority of studies will use a particular symptom as the basic inclusion criterion. Authors must clearly state whether only patients who presented the symptom in a health care encounter were included, or whether patients attending for any reason were questioned systematically about the study problem. The latter approach is essentially a form of case finding which addresses the ‘iceberg’9 of hidden symptoms in the population. Resulting prevalence or incidences can be quite high, e.g. 23% of patients attending GPs' surgeries admitted dizziness when asked systematically.10 However, only a small proportion of sufferers actually discuss their symptom with a doctor. In the case of dizziness, a study has shown that the 3 year incidence of the presented symptom was only 55 per 1000 internal medicine out-patients (new problems).11 Both approaches are useful but do not necessarily lead to the same conclusions.
For a study on symptom presentations, the patient mentioning a particular symptom fulfils the inclusion criterion. Systematic questioning about symptoms, however, requires particular care with regard to threshold definitions. Symptoms such as fatigue and palpitations are experienced by almost everybody on certain occasions. Fatigue is continuously distributed in the population, ranging from tiredness as a temporary and accepted part of life to severe debilitating illness in the form of chronic fatigue syndrome.12 Although categorizations will always be somewhat arbitrary, a reproducible decision has to be made as to whether to include an individual patient or not. This can be based on subjective severity, restricted function or both.
To judge the applicability to their own practice, readers also need to know whether only patients with a new symptom have been included, or whether patients with complaints treated previously were also eligible. In a prospective study of patients complaining of fatigue to their GPs,13 those who had done so previously were included in addition to first-time presenters. Those previously complaining of fatigue had a worse prognosis at 6 months than the rest of the sample. Closely related is the question of how long ago the patient noticed the symptom first. An appropriate presentation of symptom duration should therefore be included, e.g. a histogram for different time categories.
A study of the aetiology of fatigue excluded patients in whom a cause had been identified at the first encounter.14 Whether a patient was included in the study thus depended on the treating clinician's experience or confidence, diagnostic criteria and the availability of laboratory tests. For the reader, the composition of a sample created in this way is almost impossible to judge.
The question to be asked is "Are inclusion criteria well defined?"
|
The population |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
Studies conducted in one centre only may reflect the unique characteristics of a particular institution and/or the population served. This was shown clearly in a survey of fatigued patients in primary care practices in the USA.15 While at the academic centre the proportion of physical diagnoses was 23%, in two community practices 56% of fatigued patients were diagnosed with physical disease. As a consequence, studies should be performed in several centres that reflect the diversity of their populations. The characteristics of participating institutions, e.g. their size, demographic composition of the population served, and the age and educational status of practitioners, should be reported.
Not only the choice of a particular centre but also the setting in general will influence the generalizability of findings. For the symptom of fatigue, it has been shown that patients attending a British tertiary care clinic for their fatigue had a higher socio-economic status, were more likely to hold somatic beliefs about the origin of their fatigue and were more disabled by their symptoms than patients attending primary care facilities.16 Clinicians working at primary care level should exert caution when applying results from tertiary or secondary care studies to their own practice. For studies on the frequency and aetiology of symptoms in practice, this caveat is even more important than for clinical trials of therapeutic interventions.17
The question to be asked is "Is the sample of practices taking part multicentre and of high external validity?"
|
The population denominator |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
In some health care systems, everybody is required to register with a primary care practitioner, e.g. in the UK, or with Health Maintenance Organizations in the USA or Switzerland. Here GPs also act as ‘gatekeepers’ before secondary of tertiary care can be obtained. Under these circumstances, practice prevalences can be close to population prevalences at least for presented symptomatology, since the primary care level controls access to the whole health care system.
However, in other countries, patients have direct access to specialists even for elective problems and can thus by-pass the primary care level, e.g. in Germany and France. In this case, complex utilization patterns may influence estimates of symptom prevalence and disease frequency. For this kind of research question, there is no alternative to using patients attending a specified health care facility as the denominator. This approach, although not allowing conclusions with regard to the population level, may still yield a useful reference for practising health professionals, since they are interested primarily in data applicable to their respective practice populations.
The question is "Is the denominator of the population used in the study defined adequately?"
|
Comparison groups who do not have symptoms |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
To put diagnostic and prognostic findings into perspective, a comparison group of patients not having the symptom concerned is useful. Ridsdale et al.13 assembled a comparison group age-matched to their sample of patients consulting their GP for fatigue. Among the latter, they found twice as many subjects (75%) with a raised score of psychological distress than among controls (34%). Although the proportion of distressed patients among the fatigued would have been useful information per se, the comparison with other attenders provides a clinically relevant point of reference.
Investigators can assemble comparison groups in two ways. Among all patients attending two Canadian family medicine centres for new problems, those who presented with fatigue (13.6%) were identified by a cross-sectional survey.18 They were compared with attenders not complaining of fatigue with regard to variables such as depressive symptoms, somatic symptoms, illness worry and others. The British study mentioned above,13 however, used a case–control design. For each patient recruited prospectively who complained of being ‘tired all the time’, a control subject matched for age and sex was identified from the practice register. As the number of asymptomatic persons in relation to patients is under the control of the researcher, the latter study design cannot be used to investigate the frequency of a particular symptom in practice. However, with groups being of approximately equal size, this design is quite efficient at comparing cases and controls for variables of interest. To avoid unmasking bias, controls have to be treated in exactly the same way as subjects with symptoms.
A comparison group is desirable only with regard to aetiological or prognostic research questions. If the study is focusing on the frequency of a symptom, it can be dispensed with.
The question to be asked is "Does the study design include a comparison group without the symptom under consideration?"
|
Diagnostic procedures |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
To provide valid and reproducible information, all study patients have to undergo the same diagnostic procedures. In addition, diagnostic procedures must be performed and interpreted in a way that minimizes possible bias.
Spurious associations arise if investigations are not applied uniformly to all participants but left to the discretion of individual clinicians. Knottnerus et al.19 demonstrated that doctors were more likely to order blood counts in patients complaining of fatigue than in those who do not. As a result, among the patients diagnosed with anaemia, there were more patients complaining of fatigue than among the practice population. However, in population studies, the association between (low) haemoglobin and fatigue could not be confirmed.20,21 Anaemic subjects and those with normal haemoglobin therefore have similar probabilities of suffering from fatigue. The pseudo-relationship could arise because, among the fatigued ,‘hidden anaemias’ are more likely to be detected by clinicians ordering blood counts more frequently. The only way to avoid this ‘unmasking’ or ‘detection signal bias’22 would have been to submit every patient to the same diagnostic procedure(s).
Within studies on symptoms presented in practice, authors frequently comment on the usefulness of certain diagnostic tests. Patients who presented with transient loss of consciousness at the emergency room of an academic hospital in the USA were evaluated for possible causes.23 The decision as to whether or not to perform an electroencephalogram (EEG) was left to the treating primary physician. Twenty-six EEGs were ordered for patients who were not thought to have a central nervous system (CNS) cause for their syncope. In only three cases (12%) was relevant diagnostic information revealed. Among the group in which a CNS cause was already suspected in the emergency room, the diagnostic yield was much higher (15 of 37 patients, 41%). Although this information is potentially useful for clinicians, the resulting proportion of instances with diagnostic relevant information (‘yield’) depends on the clinicians' test-ordering behaviours. An experienced practitioner asking for laboratory tests or imaging studies in narrowly defined situations only is likely to produce a higher diagnostic yield than a doctor who lacks experience and/or is exposed to financial incentives towards more procedures.24 In addition, whether a diagnostic test once performed has produced relevant information is difficult to define in a reproducible and transparent way.25
A study of patients with palpitations in primary care has shown how transparency and objectivity can be increased.26 Standardized transtelephonic ECG readings were recorded from all patients and centrally analysed by two specialists who were unaware of the findings from the medical history and physical examination. Application of the test (ECG) to all patients reduced subjective influences of test-ordering habits. Analysis by two blinded independent examiners rendered objective interpretation more likely than in the study on syncope discussed in the previous paragraph. However, both studies together highlight a dilemma. An astute clinician can in fact increase the diagnostic yield of a procedure in comparison with its indiscriminate use. Unfortunately, his or her reasoning is almost impossible to standardize in a reproducible way for scientific purposes. Although studies performing a relevant test on each subject are more transparent and reduce bias, they often do not reflect clinical reasoning that makes extensive use of pre-test probabilities applying to an individual patient and tailor diagnostic procedures accordingly.
The matter is compounded further by the wide range of possible aetiologies that have to be considered in symptoms such as fatigue or headache. Given the large number of diagnostic procedures required to cover the full range of explanations, standardization and transparency would require extremely complex algorithms which would render the whole study unrealistic. Instead, reports focusing on a specific question, such as "how likely are patients complaining of chronic cough to have asthma or chronic obstructive pulmonary disease"27 are probably more likely to provide valid answers.
The question to ask is "Are investigations performed in a transparent, uniform and standardized way?"
|
Diagnostic criteria |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
To know which particular conditions are causing the respective symptom is of major interest for readers. Clear and reproducible criteria for relevant conditions must therefore be reported.
Patients identified as suffering from fatigue by a practice survey were followed for 1 year.28 Authors reviewed patients' records and reported the frequencies of ‘physical’, ‘psychological’ and ‘combined’ diagnoses. Since the investigators gave no explicit definitions for these categories, nor criteria for particular diagnoses or severity of disease, the reader cannot judge the validity, reliability or objectivity of these findings, and the actual clinical relevance of the findings remains unclear.
A US research team evaluated a stepwise approach to the diagnosis and treatment in patients suffering from chronic cough.29 Definitive diagnoses were based on patients' responses to specific therapies, e.g. antihistamine–decongestant for post-nasal drip, ranitidine for gastro-oesophageal reflux. However, this approach did not account for placebo or other non-specific effects that might operate independently from diagnoses or diseases.
The question to ask is "Are diagnostic categories adequately defined?"
|
Follow-up |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
Cross-sectional studies provide valid information on the prevalence of a symptom and underlying conditions. The clinician can use this kind of information to set priorities for the initial work-up of an individual patient. Cohort studies, however, provide additional information on prognosis. This may apply to defined disease states and/ or defined symptoms. However, in primary care, with a large number of undifferentiated complaints or disease categories, this often does not apply and follow-up is based on symptoms. As in studies on the prognosis of disease, participants should be at a well-defined and homogenous stage with regard to the symptom of interest.
However, follow-up not only provides prognostic information, but it can also be used as a diagnostic tool by showing additional causes emerging only after a delay. A follow-up period of 1 year was used in the study on rectal bleeding mentioned above8 as the ‘gold standard’ for the diagnosis of cancer. Although they did not standardize investigations such as barium enema or colonoscopy, the investigators assumed that malignancies would not have caused self-limiting bleeding. With regard to progressive disease such as malignancy, a sufficiently long follow-up can thus provide information even from patients who initially decline invasive procedures.
In addition, a prospective design will help to assess the validity of diagnoses. While investigating fatigued patients, Kroenke et al.30 found four patients with subclinical hypothyroidism, three of whom consented to further testing and follow-up. Despite correction of metabolic abnormalities in all three, none reported improvement in their fatigue. Initially it would have been tempting to ascribe these participants' fatigue to their hypothyroidism. Further treatment and follow-up, however, showed the symptom and abnormal laboratory finding to be unrelated. The coincidence of two relatively common states, in this case fatigue and subclinical hypothyroidism, produced a spurious association that would have gone undetected by a cross-sectional study design.
However, as in other prospective study designs, follow-up has to be sufficiently complete. Too high and/or selective drop-out will distort study results.
The question to ask is "Is a cohort suffering from a defined symptom being followed? Is follow-up sufficiently long and complete?"
|
Sample size considerations and statistical analysis |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
Reported frequencies of aetiologies and outcomes are often based on small numbers. Bates et al.31 identified 85 patients with chronic fatigue that could not be explained by known conditions. Twenty-six of these completed a specially designed protocol. Among these, three (11.5%) were found to have relevant somatic findings (hypothyroidism). The 95% confidence interval for the latter32 spans between 0 and 24%, which is hardly an estimate precise enough to guide future investigations in this type of patient. The fact that 1000 patients attending a large academic group practice had to be screened to arrive at these numbers illustrates the difficulties in obtaining numbers large enough for precise quantitative estimates. This is one more argument for conducting studies on symptoms in more than one centre. At any rate, the uncertainty related to relevant estimates should be quantified by confidence intervals. For comparisons between a symptomatic and a control group, relative risk estimates including confidence intervals and appropriate statistical tests should be reported.
The question to ask is "Have the results been statistically evaluated and are they sufficiently precise?"
|
Conclusion |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
"What is the cause of symptom X" is the most common question that primary care practitioners ask regarding patient care.33 As opposed to other publication types, studies answering this question have been uncommon in the past. In addition, many of those that have been published so far suffer from a large number of methodological weaknesses. This is due partly to a lack of funding since manufacturers have no direct interest in the results. However, because of their importance, projects investigating this area should become a priority for researchers and funding bodies alike.
Studies of symptoms in practice: quality criteria check list
- Subjects recruited prospectively
- All consecutively presenting patients included
- Inclusion criteria well defined
- Multicentre practice sample of high external validity
- Denominator defined adequately
- Group of subjects without the symptom under consideration available for comparison
- Investigations performed in a transparent, uniform and standardized way
- Diagnostic categories defined adequately
- Follow-up sufficiently long and complete
- Sample size adequate for precise estimates of important variables and confidence intervals reported
|
References |
|---|
|
TopAbstractIntroductionSelecting patientsInclusion criteriaThe populationThe population denominatorComparison groups who do...Diagnostic proceduresDiagnostic criteriaFollow-upSample size considerations and...ConclusionReferences |
|---|
1 Department of Clinical Epidemiology and Biostatistics, McMaster University Health Sciences Centre. How to read clinical journals: I. Why to read them and how to start reading them critically. Can Med Assoc J 1981; 124: 555–558.[Medline]
2 Guyatt GH, Rennie D. Users' guides to the medical literature. J Am Med Assoc 1993; 270: 2096–2097.[ISI][Medline]
3
Greenhalgh T. Getting your bearings (deciding what the paper is about). How to read a paper. Br Med J 1997; 315: 243–246.
4 Braun RN. Die gezielte Diagnostik in der Praxis, Grundlagen und Krankheitshäufigkeit. Stuttgart: Schattauer, 1957.
5
Richardson WS, Wilson MC, Guyatt GH, Cook DJ, Nishikawa J for the Evidence-based Medicine Working Group. Users' guides to the medical literature. XV. How to use an article about disease probability for differential diagnosis. J Am Med Assoc 1999; 281: 1214–1219.
6 Curry L, Putnam RW. Accuracy of Physicians' Self-reported Practice Profiles. Ottawa: Health and Welfare Canada, 1995.
7
Montaño DE, Phillips WR. Cancer screening by primary care physicians: a comparison of rates obtained from physician self-report, patient survey, and chart audit. Am J Public Health 1995; 85: 795–800.
8
Fijten GH, Muris JWM, Starmans R, Knottnerus JA, Blijham GH, Krebber TFWA. The incidence and outcome of rectal bleeding in general practice. Fam Pract 1993; 10: 283–287.
9 Verbrugge LM, Ascione FJ. Exploring the iceberg. Common symptoms and how people care for them. Med Care 1987; 25: 539–569.[ISI][Medline]
10 Yardley L, Owen N, Nazareth I, Luxon L. Prevalence and presentation of dizziness in a general practice community sample of working age people. Br J Gen Pract 1998; 48: 1131–1135.[ISI][Medline]
11 Kroenke K, Mangelsdorff AD. Common symptoms in ambulatory care: incidence, evaluation, therapy, and outcome. Am J Med 1989; 86: 262–266.[ISI][Medline]
12
Wessely S. The epidemiology of chronic fatigue syndrome. Epidemiol Rev 1995; 17: 139–151.
13 Ridsdale L, Evans A, Jerrett W, Mandalia S, Osler K, Vora H. Patients with fatigue in general practice: a prospective study. Br Med J 1993; 307: 103–106.
14 Sugarman JR, Berg AO. Evaluation of fatigue in a family practice. J Fam Pract 1984; 19: 643–647.[ISI][Medline]
15 Morrison JD. Fatigue as a presenting complaint in family practice. J Fam Pract 1980; 10: 795–801.[ISI][Medline]
16
Euba R, Chalder T, Deale A, Wessely S. A comparison of the characteristics of chronic fatigue syndrome in primary and tertiary care. Br J Psychiatry 1996; 168: 121–126.
17 Yusuf S, Held P, Teo KK, Toretsky ER. Selection of patients for randomized controlled trials: implications of wide or narrow eligibility criteria. Stat Med 1990; 9: 73–86.[ISI][Medline]
18 Cathébras PJ, Robbins JM, Kirmayer LJ, Hayton BC. Fatigue in primary care: prevalence, psychiatric comorbidity, illness behavior, and outcome. J Gen Intern Med 1992; 7: 276–286.[ISI][Medline]
19 Knottnerus JA, Knipschild PG, Sturmans F. Symptoms and selection bias: the influence of selection towards specialist care on the relationship between symptoms and diagnoses. Theor Med 1989; 10: 67–81.[Medline]
20 Elwood PC, Waters WE, Greene WJW, Sweetnam P, Wood MM. Symptoms and circulating haemoglobin level. J Chron Dis 1969; 21: 615–628.[ISI][Medline]
21 Wood MM, Elwood PC. Symptoms of iron deficiency anaemia. A community survey. Br J Prev Soc Med 1966; 20: 117–121.[ISI][Medline]
22 Sackett DL. Bias in analytic research. J Chron Dis 1979; 32: 51–63.[ISI][Medline]
23 Day SC, Cook EF, Funkenstein H, Goldman L. Evaluation and outcome of emergency room patients with transient loss of consciousness. Am J Med 1982; 73: 15–23.[ISI][Medline]
24 Hlatky MA, Lee KL, Botvinick EH, Brundage BH. Diagnostic test use in different practice settings: a controlled comparison. Arch Intern Med 1983; 143: 1886–1889.[Abstract]
25 Guyatt G, Tugwell PX, Feeny DH, Drummond MF, Haynes B. The role of before–after studies of therapeutic impact in the evaluation of diagnostic technologies. J Chron Dis 1986; 39: 295–304.[ISI][Medline]
26
Zwietering PJ, Knottnerus JA, Rinkens PELM, Kleijne AWJ, Forgels APM. Arrhythmias in general practice: diagnostic value of patient characteristics, medical history and symptoms. Fam Pract 1998; 15: 343–353.
27
Thiadens HA, de Bock GH, Dekker FW et al. Identifying asthma and chronic obstructive pulmonary disease in patients with persistent cough presenting to general practitioners: descriptive study. Br Med J 1998; 316: 1286–1290.
28 Valdini AF, Steinhardt S, Valicenti J, Jaffe A. A one-year follow-up of fatigued patients. J Fam Pract 1988; 26: 33–38.[ISI][Medline]
29
Pratter MR, Bartter T, Akers S, DuBois J. An algorithmic approach to chronic cough. Ann Intern Med 1993; 119: 977–983.
30 Kroenke K, Wood DR, Mangelsdorff AD, Meier NI, Powell JB. Chronic fatigue in primary care. Prevalence, patient characteristics, and outcome. J Am Med Assoc 1988; 260: 929–934.[Abstract]
31 Bates DW, Schmitt W, Buchwald D et al. Prevalence of fatigue and chronic fatigue syndrome in a primary care practice. Arch Intern Med 1993; 153: 2759–2765.[Abstract]
32 Gardner MJ, Altman DG. Statistics with Confidence. London: British Medical Journal, 1989.
33
Ely JW, Osheroff JA, Ebell MH, Bergus GR, Levy BT, Chambliss ML, Evans ER. Analysis of questions asked by family doctors regarding patient care. Br Med J 1999; 319: 358–361.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  N. Donner-Banzhoff Limited value of patient history and physical examination in diagnosing deep vein thrombosis in primary care Fam. Pract., October 1, 2005; 22(5): 579 - 579. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||