A cluster randomized trial with primary care practices at the unit of randomization, comparing the effects of the screening program (with subsequent targeted interventions in newly detected diseases) compared to usual care on the number of previously unrecognized chronic diseases and on patient outcomes and costs. See Figure
for the study scheme.
Study population and recruitment
Patients aged 65
years or over are selected from the electronic medical file of the participating general practitioners (GPs). All participants are recruited by GPs from the catchment area of the ‘network Utrecht care for elderly (NUZO)’. In the Netherlands all citizens are registered with a GP, irrespective of (co)treatment by a specialist, including patients living in a home for the elderly, but except those living in a nursing home or hospice. Our sample can therefore be considered as a representative sample of community dwelling patients aged 65
year or over.
To identify frail elderly we will use the criteria of the ‘transition project Elderly Care Mid Utrecht (Om-U)’; i.e. men or women who are chronically prescribed five or more different types of drugs during the last year, and/or have three or more chronic or life-threatening diseases.
Patients already known with an established diagnosis of COPD (according to the GOLD-criteria [11
]) AND also known with established HF (i.e. diagnosed by a cardiologist, backed-up by an echocardiogram [4
]) are excluded. Patients who are unable to travel to the surgery of the GP or with severe cognitive problems are also excluded.
The participating GP sends a letter to all eligible frail patients, including an information letter about the study. Patients willing to participate fill out two short questionnaires concerning dyspnoea and exercise tolerance. Those who scored positive on reduced exercise tolerance or had a ≥2 score on the Medical Research Counsel (MRC) dyspnoea scale [12
] are included in the study (see Additional file 1
for an overview of the inclusion criteria).
Written informed consent is obtained at the first appointment before any study procedure is undertaken.
Primary care practices are randomly allocated to either the diagnostic screening with subsequent targeted intervention strategy or care as usual. Randomization is considered necessary because we aim to quantify the clinical effectiveness of the strategy (test results followed by targeted treatment initiated by the GP). Cluster randomization is used to avoid contamination of GPs and participants between the two groups, which can occur if randomization is performed at an individual participant level.
Randomization is performed by a computer-based program, using the minimization method. Minimization is based on the size of the general practice (number of enlisted men and women) and ensures that the two arms are balanced in the number of participants.
We estimated that with our diagnostic strategy at least 15% of the population under study would be diagnosed with newly detected HF and/or COPD. These numbers are based on previous diagnostic studies showing that among elderly with COPD 20% had previously unknown HF [6
To detect a 10% difference between the 2 arms in diagnoses of HF and/or COPD, using an alpha of 0.05, a power of 0.80 and an intracluster correlation coefficient of 0.05[13
], 301 participants per arm are needed. Considering a drop-out of 10% we aim to include 400 participants in each arm (total 800). We calculated that around 35 general practices should participate to recruit 800 participants.
At the first appointment, all participants fill out a questionnaire of the ‘Dutch National Care for the Elderly Program (ZonMw-NPO)’, i.e. the Minimal Data Set (MDS)[14
] on functionality and a questionnaire on general health related quality of life, i.e. the short form-36 (SF-36) [15
Subsequently, participants receive detailed information on the study procedures in the diagnostic screening strategy arm or care as usual arm, depending on the random allocation of their GP practices.
After six months the electronic medical files of all participants are searched to document changes in medication status, new diagnoses and use of care during the past six months and participants are asked to fill out again the MRC dyspnoea, MDS and SF-36 questionnaires.
Screening strategy and care as usual
The diagnostic screening includes the following tests: standardized questionnaire on medical history, symptoms and current drug use (participants are asked to bring their medication containers), physical examination with special attention for the heart and lungs, blood tests, electrocardiogram (ECG), spirometry with pre- and post-bronchodilator measurements and echocardiography including tissue Doppler imaging.
The physical examination is performed by a trained physician in a standardized manner. Venous blood samples are taken and after centrifugation specimens of plasma and cells will be stored at −70 degrees Celsius. The blood tests include measurements of plasma B-type natriuretic peptide measurements (NT-pro-BNP), glucose levels, creatinin levels (with a calculated estimated glomerular filtration rate (eGFR) according to MDRD (modification of diet in renal disease)), haemoglobin and Thyroid-Stimulating Hormone (TSH) levels. A standard 12-lead ECG is recorded by a trained employee of the GPs’ laboratory (Saltro) and classified according to the Minnesota coding criteria by a single cardiologist, blinded to all other test results [17
The spirometry is also performed by a trained employee of Saltro before and 20 minutes after administration of ipratropiumbromide inhalation with an inhalation chamber, and subsequently read and interpreted by a pulmonologist.
Echocardiography is performed with a mobile echocardiography (Vivid-i) by a trained and experienced cardiac sonographer from Saltro. All echocardiographic images will be stored and interpreted by a single cardiologist who is blinded to all other data.
After every investigation, the participant fills out a visual analog scale (VAS) to measure the burden experienced during the investigation (0
not burdening at all, 10
All investigations take place in the patient’s general practice office. Every participant in the diagnostic screening group receives all measurements. About two weeks after the investigations, the GP of the participant in the diagnostic screening arm will receive all results, with (preliminary) diagnoses (presence or absence of HF and/or COPD and of other diseases) including a treatment advice. The GP can then initiates or adjusts the treatment according to current clinical guidelines.
In the care as usual group, the participating GP only receives the answers to the questionnaires about symptoms (answers to the MRC dyspnoea and exercise tolerance questionnaires) from the participants. It is up to the GP to decide whether he/she actively starts any investigations or treatment depending on the answers provided.
HF and COPD
The presence of HF and/or COPD present or absent will be determined six months after the initial investigations by an outcome panel that will evaluate all diagnostic test results and the effect of treatment during six months follow-up.
For the diagnosis of COPD, airflow obstruction with spirometry is a prerequisite and defined as a post-dilatory ratio of the forced expiratory volume in one second (FEV1) to the forced vital capacity (FVC) <0.70 (FEV1/FVC ratio <0.70). To classify the severity of obstruction, the GOLD-criteria [11
] will be applied by the panel.
For the diagnosis of HF, the panel applies the criteria of the European Society for Cardiology (ESC) [4
]. HF is considered present when participants have suggestive symptoms (i.e. breathlessness at rest or on exercise, fatigue, tiredness, ankle swelling) and signs (i.e. pulmonary rales, raised jugular venous pressure, peripheral oedema, laterally displaced apical beat) in combination with objective echocardiographic evidence of cardiac dysfunction at rest. In participants who use diuretics for hypertension, signs of volume overload could be masked and therefore in these participants signs of fluid overload are not obligatory. Participants classified as HF by the panel will further be classified as systolic HF, heart failure with preserved ejection fraction (HFPEF) or right-sided HF (cor pulmonale).
For systolic HF, participants need to have a left ventricular ejection fraction (LVEF) ≤45%. To diagnose HFPEF, there must be echocardiographic evidence of diastolic structural or functional abnormalities (abnormal left ventricular relaxation or diastolic stiffness), left atrial enlargement or left ventricular hypertrophy (LVH) in combination with a LVEF >45% [18
]. Participants with echocardiographically-determined LVH and an indexed left atrial volume of ≥34
or tissue Doppler abnormal mitral inflow or pulmonary venous flow profile are also classified as having diastolic dysfunction. In participants with atrial fibrillation (AF) a complete diastolic assessment is not feasible and therefore we consider an elevated indexed LA volume sufficient to classify diastolic dysfunction in participants with AF. Isolated right-sided HF is defined as increased pulmonary artery pressure (calculated systolic pulmonary artery pressure >40
mmHg) and a LVEF >45%.
The panel also assessed the most likely cause of HF based on the information from the investigations of the diagnostic screening.
Other new diagnoses considered as an explanation of reduced exercise tolerance or dyspnoea are (persisting) asthma, pulmonary restriction, obesity, anaemia, renal dysfunction, thyroid disorders and atrial fibrillation.
For the diagnosis asthma a combination of symptoms (dyspnoea, wheezing, coughing, sputum production), allergy and/or hyperreactivity is needed, and an increase of >12% in forced expiratory volume in one second (FEV1) from pre- to post-dilatory measurements further confirms the diagnosis. The diagnosis of pulmonary restriction is based on a post-dilatory forced vital capacity (FVC) post-dilatory
80% of predicted, in the absence of a diagnoses of COPD.
Obesity is defined as a body mass index (BMI) above 30
Anaemia is considered present if Haemoglobin (Hb) is below 7.0
mmol/l in women and below 8.0
mmol/l in men aged 65 to 69
years. In patients aged 70
years or over anaemia is present if Hb is below 6.9
mmol/l in men and below 6.8
mmol/l in women.
Renal dysfunction is considered present if the calculated, based on the MDRD formula, eGFR is below 90
for men and 80
for women. Severity of renal dysfunction is divided into 4 stages; mild (GFR between 60
and 80 (women) or 90 (men), moderate (GFR between 30 and 59
), severe disease (GFR between 15 and 30
) and kidney failure (GFR below 15
Thyroid disorders are divided into hypo- or hyperthyroidism. When the Thyroid-Stimulating Hormone (TSH) level is below 0.35 or above 5.5 mu/l, Free Thyroxin 4 (FT4) will be measured. Hypothyroid disease is considered present when TSH >5.5 mu/l and FT4 <11 pmol/l. Subclinical hypothyroid disease is when TSH >5.5 mu/l, and FT4 within normal limits (i.e. between 11 and 23 pmol/l). Hyperthyroid disease is considered present when TSH <0.35 mu/l and FT4 >23 pmol/l. Subclinical hyperthyroid disease is a TSH <0.35 mu/l and FT4 within normal limits.
Atrial fibrillation is classified according to the Minnesota coding [17
Functional health is measured with the Minimal Data Set (MDS). The dimensions asked in the MDS are: living situation, perceived health and quality of life, health related quality of life (questions of the index EuroQol Five-Dimensional Questionnaire (EQ-5D_index))[20
], activities of daily living (ADL) functioning (questions from Katz-15[21
]) and health care use (hospital admissions, visits of GP outside office hours, home care and temporally admission to a nursing home).
Health related quality of life
Health related quality of life is measured by the SF-36 and the EQ-5D_index (included in the MDS) questionnaires. The SF-36 is divided into nine subscales: physical functioning, social functioning, limitations in usual role activities due to physical problems, limitations in usual role activities due to emotional problems, bodily pain, change in health, general mental health and vitality health change and general health perception. Scores range from 0 and 100.
The EQ-5D_index is a questionnaire with five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression), which are divided into three degrees of severity, ‘no problem’, ‘some problems’ or ‘major problems’. A single index score can be produced using information from these five dimensions.
Higher scores are associated with a better health related quality of life.
The electronic medical file from the GP is used to collect information about direct costs during the follow-up period of 6
months. Information about GP visits, investigations (i.e. blood tests, ECG, spirometry), outpatient visits, hospitalizations and medication use are collected in both groups. In the diagnostic screening group additionally costs are calculated for all performed test at baseline. Prices will be based on market prices or tariffs.
Final diagnosis by the panel
The final diagnoses are set by a joint panel consisting of three members, a cardiologist, a pulmonologist and a GP. Consensus diagnosis by the panel will be our reference test for the diagnosis of COPD and HF, in case of any discongruence with the classification described before. This is an established method in case an irreproachable reference standard is lacking, as is the case for HF [22
]. In case of no consensus the majority decision will be used. In analogy with earlier studies, the panel will use all available information from the diagnostic and follow-up results: the latter to include the effect of targeted therapy in the decision whether a disease is present. The reproducibility of the panel diagnoses will be evaluated by re-testing a random sample of 10% by the same outcome panel.
The joint panel evaluation is divided in three steps; step 1 history taking, physical examination, blood tests, ECG and spirometry, step 2 adding NT-pro-BNP and step 3 adding echocardiography and follow-up. After step 1 and 2 the panel gives a percentage (0-100%) of their suspicion for the different outcomes. With this, we can evaluate how the panel reaches their final diagnoses (set at step 3) and which investigations add most in reaching this decision.
The above evaluation is also done individually by a GP, a pulmonologist and a cardiologist, but the procedure is changed slightly by dividing step 3 in two steps whereby step 3 includes only echocardiography and at step 4 the follow-up is included and the final diagnoses are made. We will evaluate this individual approach and compare it with the more time consuming joint panel evaluation.
In our single blind study we apply no active measures for blinding the group assignment. The GPs are invited to participate in a study detecting unrecognized HF and COPD. If they agree to participate, their practice is randomized and they receive detailed information. This strategy of selective information is chosen to avoid that GPs in the control group perform all diagnostic tests available in the diagnostic screening program.
Participants will be aware whether their GP practice is allocated to the diagnostic screening arm or to usual care. Researchers are not blinded. There is no risk in observer bias for the outcome, because the questionnaires used to collect outcome measurements are filled out at home by the participants themselves.
The yield of previously unrecognized HF and COPD (and other diseases) will be calculated as a proportion with 95% confidence intervals in both arms of the trial. The absolute difference in yield between the arms will then be calculated together with its 95% confidence interval. We will also calculate how many participants in the diagnostic screening group were initially incorrectly diagnosed with COPD or HF.
The difference in functional health and quality of life between the diagnostic screening and care as usual groups at 6
months will be compared taking into account potential baseline difference of relevant parameters at baseline (ANCOVA), although such differences are expected to be minimal because of the randomization procedure.
Multivariable regression analysis with receiver operating characteristic (ROC) curves and percentages correctly diagnosed participants will be used to analyze which of the test used in the diagnostic screening are most predictive or necessary for setting the diagnosis of HF and COPD in the study group. A multilevel approach is used in the analyses to correct for the fact that we randomized at the GP practice and not at the patient level.
The cost-effectiveness of the diagnostic strategy for detection and treatment of (previously unrecognized) HF and COPD (diagnostic screening group) is evaluated and compared with care as usual. Cost-effectiveness is expressed in terms of cost per case of COPD and per case of HF detected, and in terms of cost per Quality Adjusted Life Years gained (Cost-utility analysis). For measuring direct cost, resource quantities are collected prospectively through the electronic medical files from the GPs. Prices are based on market prices or tariffs for the investigations performed under study. Relevant indirect medical costs are taken into account such as participant time invested and travel costs. As the population under study is older than 65
years of age, productivity losses will not be studied. Sensitivity analysis and multivariate uncertainty analysis are performed, conform current Dutch standards for pharmaco-economic research [23
This study is conducted according to the principles of the current version of the declaration of Helsinki and in accordance with the Dutch law on Medical Research Involving Human Subjects Act (WMO).
Ethics committee approval
The study was approved by the medical ethical committee (METC) of the University Medical Center Utrecht (UMCU), the Netherlands.