. 2013 Nov 1;91(11):824–833I. doi: 10.2471/BLT.13.118786

Table 3. Experimental studies included in meta-analysis of mid-level health workers’ effectiveness, 1973–2012.

Study	Study design	Health worker comparison	Study characteristics	Study participants	Outcomes
Begley, 2011²¹	Randomized controlled trial	Midwives compared with general practitioners	Intervention: antenatal care from midwives in a maternity unit or an outreach clinic and, if desired, from the woman’s general practitioner. When complications arose, women were transferred to a consultant-led unit according to agreed criteria.	Recipients: pregnant women. (n = 1653; 1101 in the intervention arm and 552 in the control arm)	Health status, including rates of caesarean delivery, episiotomy and epidural anaesthesia
			Control: standard care from a consultant-led unit.
			Health-care setting: general hospital.
			Geographical setting: Australia
Harvey, 1996²²	Randomized controlled trial	Midwives compared with doctors	Intervention: a team of seven midwives provided antenatal and intrapartum care in a hospital and postpartum care in the community.	Recipients: women at a low risk of complications who requested and qualified for the intervention group. (n = 101)	Rates of caesarean delivery, episiotomy and epidural anaesthesia
			Controls: usual care from obstetricians or general practitioners at a variety of city hospitals.
			Health-care setting: primary care.
			Geographical setting: Canada
Hundley, 1994²³	Randomized controlled trial	Midwives compared with consultant gynaecologist	Intervention: patients randomized to the midwife-led unit were seen by a midwife throughout delivery with minimal intervention from other hospital staff. Midwives were responsible for delivering and maintaining care.	Recipients: low-risk pregnant women from the general population. (n = 2844)	Maternal and perinatal morbidity
			Controls: care from a consultant in a delivery unit.
			Health-care setting: medical centre.
			Geographical setting: Aberdeen, Scotland
MacVicar, 1993²⁴	Randomized controlled trial	Midwives compared with doctors	Intervention: team of two midwifery sisters assisted by eight staff midwives provided hospital-based antenatal and intrapartum care and hospital care only after childbirth.	Recipients: women with a low risk of complications. (n = 2304)	Induction, augmentation, intrapartum complications, need for pain relief, perineal status, satisfaction with care, birth weight, Apgar score and maternal and fetal mortality
			Controls: shared antenatal care from general practitioner and midwives and intrapartum care from hospital staff.
			Health-care setting: tertiary-care hospital.
			Geographical setting: Leicester, England
Marks, 2003²⁵	Randomized controlled trial	Midwives compared with general practitioners	Intervention: all women in the study received the basic pattern of care provided by the midwifery service at King’s College London; that is, each woman was seen by a midwife 8 to 12 times during pregnancy, on commencement of labour, at delivery and then as required from a minimum of 10 days postpartum to a maximum of 28 days.	Recipients: women who had had at least one episode of a major depressive disorder, as defined by DSM-III-R criteria, either in the past or during the current pregnancy. (n = 51)	Any antenatal illness, new episode of illness after antenatal registration, any postnatal illness and Edinburgh Postnatal Depression Scale score 4 weeks after giving birth
			Controls: women in the control group received a mixture of care at antenatal clinics after giving birth: either from a general practitioner alone or from a general practitioner and a community midwife. However, none of these included continuity of care.
			Geographical setting: King’s College London, London, England
McLachlan, 2008²⁶	Randomized controlled trial	Midwives compared with doctors	Intervention: care from midwives who attended information sessions on the need to adhere to the clinical practice guidelines of the Royal Women's Hospital.	Recipients: pregnant women. (n = 2307; 1150 in the intervention arm and 1157 in the control arm)	Health status, rates of caesarean delivery, episiotomy and epidural anaesthesia
			Controls: care from doctors.
			Health-care setting: teaching hospital.
			Geographical setting: Australia
Di Napoli, 2004²⁷	Randomized controlled trial	Midwives compared with doctors	Intervention: breastfeeding counselling.	Recipients: mothers who had recently given birth. (n = 605; 303 in the intervention and 302 in the control arm)	Breastfeeding 4 months and 6 months postpartum
			Control: no specific intervention.
			Health-care setting: in the community.
			Geographical setting: Rome, Italy
Rowley, 1995²⁸	Stratified randomized controlled trial	Midwives compared with doctors	Intervention: care from a midwifery team.	Recipients: women attending an antenatal clinic. (n = 405)	Antenatal, intrapartum and neonatal events, satisfaction and costs
			Controls: routine care.
			Health-care setting: teaching hospital.
			Geographical setting: Australia
Small, 2000²⁹	Randomized controlled trial	Midwives compared with a consultant gynaecologist	Intervention: debriefing by midwives who were experienced in talking with women about childbirth, able to listen with empathy to women's accounts and aware of the common concerns and issues arising for women after an operative birth. The content of the discussion was determined by each woman's experiences and concerns and up to 1 hour was made available for each session.	Recipients: women who had operative deliveries were identified from maternity ward records at least 24 hours after delivery by two research midwives. (n = 917)	Maternal depression and overall health status
			Controls: no debriefing by midwives.
			Health-care setting: teaching hospital.
			Geographical setting: Melbourne, Australia
Turnbull, 1996³⁰	Randomized controlled trial	Midwives compared with doctors	Intervention: women received care at midwifery development unit.	Recipients: women with normal healthy pregnancies. (n = 648)	Postnatal care, preparation for parenthood
			Controls: women received traditional care.
			Health-care setting: major urban teaching hospital.
			Geographical setting: Glasgow, Scotland
Waldenström, 2001³¹	Randomized controlled trial	Midwives compared with doctors	Intervention: team midwifery care in a hospital’s public delivery wards 24 hours a day; when no woman cared for by her team was in labour, the midwife cared for women outside her team. Each midwife worked on average one shift per week in the hospital antenatal clinic where she saw only women enrolled in team care.	Recipients: patients who presented to the Royal Women's Hospital, Melbourne. (n = 464)	Satisfaction with care, maternal complication rates, procedures performed and infant outcomes such as neonatal death
			Controls: standard care, mostly from doctors.
			Health-care setting: teaching hospital.
			Geographical setting: Australia
Law, 1999³²	Randomized controlled trial	Midwives compared with general practitioners	Intervention: care exclusively from midwives.	Recipients: pregnant women assessed as being at a low risk on admission to a labour ward. (n = 1050)	Caesarean delivery, normal vaginal delivery, episiotomy, Apgar score
			Controls: care from doctors.
			Health-care setting: teaching hospital.
			Geographical setting: Hong Kong Special Administrative Region
Wolke, 2002³³	Randomized controlled trial	Midwives compared with paediatric senior house officers	Intervention: neonate examination by midwives, which was usually carried out 6 to 24 hours after birth.	Recipients: mothers and neonates. (n = 826)	Maternal satisfaction with examination
			Controls: neonate examination by junior paediatricians.
			Health-care setting: district general hospital.
			Geographical setting: south-east England
Eren, 1983³⁴	Controlled comparative trial	Auxiliary nurse midwives compared with doctors	Intervention: intrauterine device insertion by auxiliary nurse midwives.	Recipients: healthy women. (n = 501 in the Philippines; 250 in the intervention arm and 251 in the control arm); (n = 495 in Turkey; 257 in the intervention arm and 238 in the control arm)	Health status, successful intrauterine device insertion
			Controls: insertion by doctors.
			Health-care setting: teaching hospitals.
			Geographical setting: the Philippines and Turkey
Dusitsin, 1980³⁵	Randomized controlled trial	Mid-level health workers and nurse midwives compared with doctors	Intervention: tubal ligation by mid-level health workers. Controls: tubal ligation by doctors. Health-care setting: operating theatre. Geographical setting: Thailand	Recipients: healthy women. (n = 292; 143 in the intervention arm and 149 in the control arm)	Health status, postoperative complications, difficulties during the operation
Warriner, 2006³⁶	Randomized controlled trial	Midwives and doctor’s assistants compared with doctors	Intervention: manual vacuum aspiration performed by a mid-level health worker, with a follow-up 10 to 14 days later.	Recipients: women who presented for an induced abortion at up to 12 weeks’ gestation were informed of the study and invited to participate. (n = 1160) in South Africa; (n = 1734) in Viet Nam	The primary outcome was abortion complications
			Controls: manual vacuum aspiration performed by a doctor.
			Health-care setting: primary care.
			Geographical setting: South Africa and Viet Nam
Warriner, 2011³⁷	Randomized controlled trial	Certified nurses and nurse auxiliary midwives compared to doctors	Intervention: randomized patients were managed by mid-level health workers who had full responsibility for the management of each case. Abortions were performed according to Nepalese health authority guidelines.	Recipients: women seeking a termination early in the first trimester of pregnancy. (n = 542)	The primary outcome was the successful abortion rate
			Controls: similar care from doctors with similar responsibilities.
			Health-care setting: primary care.
			Geographical setting: Nepal
Sanne, 2010³⁸	Randomized controlled trial	Nurses compared with doctors	Intervention: primary care from primary health-care nurses. The experimental nurse monitoring strategy involved doctor-initiated ART, which was monitored by the nurses.	Recipients: HIV-positive patients. (n = 404)	The primary study outcome was the composite endpoint of the various treatment-limiting events that could occur on first-line ART
			Controls: primary care from doctors. The care strategy was consistent with the routine management of patients in the current South African ART programme, which is based on treatment initiated and monitored by a doctor.
			Health-care setting: primary care in two towns.
			Geographical setting: South Africa
Mann, 1998³⁹	Randomized controlled trial	Nurses compared with doctors	Intervention: study 1 evaluated the effectiveness of standardized psychiatric assessment by a practice nurse with feedback of information to a general practitioner. Study 2 evaluated the same assessment by and feedback from a practice nurse combined with nurse-assisted follow-up care. In study 1, nurses interviewed patients and scored them using the Beck Depression Inventory. In study 2, nurses were also responsible for follow-up care.	Recipients: patients diagnosed as depressed by a general practitioner. (n = 577; 158 in study 1 and 419 in study 2)	Change in Beck Depression Inventory scores 4 and 5 and change in the proportion of patients fulfilling DSM-III criteria for major depression, as determined by the nurse assessment interview
			Controls: patients were seen and managed by a general practitioner.
			Health-care setting: clinics.
			Geographical setting: Liverpool, England
Du Moulin, 2007⁴⁰	Randomized controlled trial	Nurses compared with doctors	Intervention: in a cluster-randomized study, women were referred to a continence nurse who, guided by a protocol, assessed the patients and gave advice about therapy, lifestyle and medications. If progress was disappointing, therapy was revised.	Recipients: low-risk pregnant women from the general population. (n = 45)	The primary outcome was the number of incontinence episodes
			Controls: women received usual care.
			Health-care setting: primary care.
			Geographical setting: the Netherlands
Gordon, 1974⁴¹	Randomized controlled trial	Nurse clinicians compared with doctors	Intervention: primary medical care was provided by a nurse clinician.	Recipients: patients who presented to hospital outpatient clinics. (n = 169; 82 in the intervention arm and 87 in the control arm)	Correct diagnosis, frequency of visits and clinical status
			Controls: care was provided by attending physicians in a general medical clinic.
			Health-care setting: primary care.
			Geographical setting: New York, United States
Hemani, 1999⁴²	Prospective randomized controlled trial	Nurses compared with doctors	Intervention: care provided by a nurse practitioner.	Recipients: adult patients. (n = 150)	Specialty care, primary care, emergency and walk-in visits and hospitalizations over a 1-year period
			Controls: care provided by a resident or attending physician.
			Health-care setting: primary care clinic.
			Geographical setting: Veterans Affairs medical centre, United States
Katz, 2004⁴³	Randomized controlled trial (secondary analysis)	Licenced practice nurses compared with medical officers	Intervention: counselling provided by licenced practice nurses.	Recipients: adult smokers. (n = 1221)	Proportion of patients who received counselling as recommended by guidelines
			Controls: counselling provided by medical officers.
			Health-care setting: primary health care.
			Geographical setting: United States
Kinnersley, 2000⁴⁴	Randomized controlled trial	Nurses compared with doctors	Intervention: patients seen by a nurse practitioner. Patients who were prepared to consult either a general practitioner or a nurse practitioner were informed about the study in general terms. Consent was obtained when patients attended the surgery and were told which clinician they would see. All practices had a trained member of staff to manage the study under the supervision of the project research officer.	Recipients: patients seeking a same-day consultation. (n = 652)	Patients’ satisfaction with care, resolution of symptoms and concerns and patients' intentions regarding seeking care in the future
			Controls: patients seen by a general practitioner.
			Health-care setting: primary care.
			Geographical setting: 10 general practices in south Wales and south-west England
Strömberg, 2003⁴⁵^,⁴⁶	Randomized controlled trial	Cardiac nurses compared with doctors	Intervention: care was provided in a nurse-led heart failure clinic, which was staffed by specially educated, experienced cardiac nurses who took responsibility for making protocol-led changes in medication. The first follow-up visit took place 2 to 3 weeks after hospital discharge. During that visit, the nurse evaluated the degree of heart failure and the treatment received, provided education about heart failure and gave social support to the patient and his or her family.	Recipients: patients hospitalized for heart failure. (n = 106)	All-cause mortality and all-cause hospital admission after 12 months
			Controls: patients received usual care from primary care physicians in accordance with current guidelines.
			Health-care setting: primary care.
			Geographical setting: Linköping University, Sweden
Moher, 2001⁴⁷	Cluster randomized controlled trial	Nurses compared with doctors	Intervention: setting up a disease register and a procedure for the systematic recall of patients to a nurse-led clinic (nurse recall group).	Recipients: patients aged 55–75 years. (n = 1906)	Adequate assessment of risk factors
			Controls: setting up a disease register and a procedure for the systematic recall of patients to a general practitioner (general practitioner recall group).
			Health-care setting: 21 general practices.
			Geographical setting: Warwickshire, England
Sakr, 1999⁴⁸	Randomized controlled trial	Nurse practitioners compared with doctors	Intervention: patients were first managed by a nurse practitioner who carried out a clinical assessment. Assessments were transcribed afterwards to maintain blinded conditions. Patients were then assessed by an experienced accident and emergency physician (i.e. a research registrar) who completed a research assessment but took no part in the clinical management of the patients.	Recipients: patients attending an accident and emergency department. (n = 727)	The primary outcome measure was the adequacy of the care provided
			Controls: patients were first managed by a junior doctor.
			Health-care setting: accident and emergency department at a general hospital.
			Geographical setting: Sheffield, England
Smith, 2005⁴⁹	Randomized controlled trial	Respiratory nurse specialists compared with doctors	Intervention: in addition to usual care, patients received a 6-month psychoeducational programme of home visits and telephone calls from a supervised respiratory nurse specialist.	Recipients: patients with a confirmed diagnosis of severe asthma who had been admitted to hospital. (n = 92)	The primary outcome was asthma symptom control
			Controls: patients received usual care from primary care and secondary care clinics, which included an asthma assessment every 3 to 6 months.
			Health-care setting: hospital and primary care asthma clinics.
			Geographical setting: Norfolk and Suffolk, England
Stein, 1974⁵⁰	Randomized controlled trial	Nurse practitioners compared with doctors	Intervention: patients were randomized to a consultation with a nurse practitioner, which included history-taking, physical examination, ordering tests and referrals.	Recipients: patients diagnosed with type-2 diabetes. (n = 23)	The primary outcome was diabetic status
			Controls: patients were managed by an internist.
			Health-care setting: hospital clinics.
			Geographical setting: Gainsville, Florida, United States
Chambers, 1977⁵¹^,⁵²	Interrupted time series	Family practice nurses compared with doctors	Intervention: care was provided by family practice nurses who were employed to provide health care to local residents.	Recipients: patients in all age groups. (n = 1167)	Utilization of primary health-care services and hospital services and economic effect of deploying a family practice nurse
			Controls: care was provided by doctors at a local hospital.
			Health-care setting: primary care in two villages.
			Geographical setting: Ontario, Canada
Caine, 2002⁵³	Randomized controlled trial	Nurse practitioners compared with doctors	Intervention: care was provided by a nurse practitioner.	Recipients: patients aged over 18 years attending a bronchiectasis clinic with moderate or severe disease confirmed by high-resolution computed tomography. (n = 80)	Number of hospital admissions, health-related quality of life, satisfaction with care and compliance with care
			Controls: care was provided by a general practitioner.
			Health-care setting: primary care.
			Geographical setting: Lung Defence Clinic, Papworth Hospital, England
Chambers, 1978⁵⁴	Randomized controlled trial	Nurses compared with doctors	Intervention: families were allocated to nurse-led primary care.	Recipients: families. (n = 868)	Health status
			Controls: families received doctor-led primary care.
			Health-care setting: health-care clinics.
			Geographical setting: Canada
Chinn, 2002⁵⁵	Randomized controlled trial	Dermatology nurses compared with doctors	Intervention: a trained dermatology nurse holding an ENB 393 certificate in dermatology demonstrated techniques for applying medication, gave advice and provided education in a single 30-minute session. Patients were also provided with leaflets from a drug company that did not promote a product and which covered topics in a one-page format.	Recipients: children aged under 16 years visiting a general practitioner for dermatitis. (n = 240)	The primary outcomes were quality of life, as assessed by the Children's Dermatology Life Quality Index and the Infant's Dermatitis Quality of Life Index, and the effect on the family, as assessed by the Family Dermatitis Index
			Controls: patients received usual care from a general practitioner.
			Health-care setting: primary care.
			Geographical setting: Middlesborough, Teeside, England
Cox, 2000⁵⁶	Prospective follow-up study (data analysed as a case–control study)	Nurses compared with doctors	Intervention: patients were seen by a nurse.	Recipients: patients aged over 2 years presenting with a sore throat to a doctor or practice nurse during normal working hours. (n = 188)	Sore throat settled, median number of days for sore throat to settle, number of patients requiring analgesia, reconsultation rate, dissatisfaction rate and recollection of advice about home remedies
			Controls: patients were seen by a doctor.
			Health-care setting: clinics.
			Geographical setting: United Kingdom
D’Eramo–Melkus, 2004⁵⁷	Prospective randomized controlled trial	Registered nurses compared with doctors	Intervention: registered nurses provided information on diabetes care. Dietary habits were addressed by a dietician and other lifestyle modifications were encouraged.	Recipients: low-risk, pregnant, black African women from the general population. (n = 25)	Weight, body mass index, haemoglobin A1c level, knowledge about diabetes, self-efficacy and diabetes-related emotional distress
			Controls: management was provided by doctors.
			Health-care setting: primary care.
			Geographical setting: Pennsylvania, United States
Dierick–van Daele, 2009⁵⁸	Randomized controlled trial	Nurses compared with doctors	Intervention: patients with common complaints were randomized to be seen by a nurse practitioner.	Recipients: patients who attended a medical practice in selected areas for an appointment during the study period. (n = 759)	Patients’ perception of the quality of care, effectiveness of consultation
			Controls: patients with similar complaints were seen by a general practitioner.
			Health-care setting: primary care.
			Geographical setting: the Netherlands
Federman, 2005⁵⁹	Randomized controlled trial	Nurses compared with doctors	Intervention: treatment provided by a nurse.	Recipients: patients with dyslipidaemia, diabetes mellitus or hypertension. (n = 19 660)	Haemoglobin A1c level, low-density-lipoprotein cholesterol level and blood pressure
			Controls: treatment provided by a doctor.
			Health-care setting: primary care.
			Geographical setting: Connecticut, United States
Houweling, 2009⁶⁰	Randomized controlled trial	Nurses compared with doctors	Intervention: care from nurses, who provided treatment to correct glycaemia, blood pressure and lipid profiles according to a protocol.	Recipients: patients with type-2 diabetes. (n = 84)	Haemoglobin A1c level, total cholesterol level, low-density-lipoprotein cholesterol level and total cholesterol to high-density-lipoprotein cholesterol ratio
			Controls: similar care from doctors (i.e. internists).
			Health-care setting: clinics.
			Geographical setting: the Netherlands
Mundinger, 2000⁶¹	Randomized controlled trial	Nurses compared with doctors	Intervention: patients were assessed and followed up by nurse practitioners, who had the same independence as doctors.	Recipients: adult patients. (n = 806)	Health status at consultation and follow-up
			Controls: patients were seen by doctors.
			Health-care setting: primary care.
			Geographical setting: New York, United States
Myers, 1997⁶²	Follow-up study	Nurses compared with doctors	Intervention: care provided by nurses.	Recipients: patients with acute medical conditions. (n = 1000)	Morbidity
			Controls: care provided by doctors.
			Health-care setting: clinics.
			Geographical setting: United Kingdom
Rushforth, 2006⁶³	Randomized controlled trial	Nurses compared with senior house officers	Intervention: preoperative assessment by nurses before day surgery.	Recipients: children. (n = 595)	Ability to detect clinically significant abnormalities
			Controls: assessment by senior house officers.
			Health-care setting: preoperative clinic.
			Geographical setting: England
Sharples, 2002⁶⁴	Randomized, controlled, crossover trial	Nurse practitioners compared with doctors	Intervention: nurse practitioner-led care, during which patients underwent routine tests followed by a consultation that involved clinical assessment and discussion of a management plan.	Recipients: patients aged over 18 years with bronchiectasis. (n = 80)	Forced vital capacity, 12-minute walk test results, number of exacerbations of infection that required intravenous antibiotics, number of hospital admissions, health-related quality of life
			Controls: doctor-led care following a similar procedure.
			Health-care setting: bronchiectasis outpatient clinic.
			Geographical setting: United Kingdom
Shum, 2000⁶⁵	Randomized controlled trial	Nurses compared with doctors	Intervention: care was managed by nurses, who took the patient's history, performed a physical examination, offered advice and treatment, issued prescriptions (which required a doctor's signature) and referred the patient to the doctor when appropriate.	Recipients: patients who requested and were given a same-day appointment by a receptionist. (n = 1815)	Patients’ general satisfaction with care as assessed using a consultation satisfaction questionnaire
			Controls: care was managed by doctors.
			Health-care setting: five general practices.
			Geographical setting: England
Venning, 2000⁶⁶	Randomized controlled trial	Nurse practitioners compared with doctors	Intervention: the patient’s first medical encounter was with a nurse practitioner, who was responsible for history-taking, physical examination, ordering tests, diagnosis and writing prescriptions to be signed by a general practitioner.	Recipients: patients consulting general practices. (n = 503)	Patients’ satisfaction with care, health status, repeat visits within 2 weeks and costs
			Controls: patients were managed by a general practitioner.
			Health-care setting: 20 general practices.
			Geographical setting: England and Wales
Sackett, 1974⁶⁷	Randomized controlled trial	Nurses compared with doctors	Intervention: care provided by nurses.	Recipients: families. (n = 4325)	Health status, patients’ satisfaction with care and costs
			Controls: care provided by doctors.
			Health-care setting: primary care.
			Geographical setting: Canada.
Babor, 2005⁶⁸	Randomized controlled trial	Nurses compared with doctors	Intervention: alcohol screening and brief intervention by nurses.	Recipients: hazardous and harmful drinkers. (n = 3449)	Percentage of patients screened
			Control: alcohol screening and brief intervention by doctors.
			Health-care setting: primary care clinics.
			Geographical setting: United States
McIntosh, 1997⁶⁹	Randomized controlled trial	Nurses compared with doctors	Intervention: brief advice (i.e. 5 minutes) from a physician, followed by two 30-minute sessions with a nurse practitioner using cognitive behavioural strategies.	Recipients: hazardous drinkers. (n = 159)	Reduction in hazardous drinking and problems related to drinking over 12 months of follow-up
			Controls: the same strategy, but with two 30-minute sessions with a physician.
			Health-care setting: family practice clinics.
			Geographical setting: Australia

ART, antiretroviral therapy; DSM-III-R, Diagnostic and Statistical Manual of Mental Disorders, 3rd edition revised; HIV, human immunodeficiency virus.