Scolaris Content Display Scolaris Content Display

Contenido relacionado

Ir a:

Revisiones y protocolos relacionados

Respuestas clínicas Cochrane

Topics

Temas

Outcomes

Relative effect

Quality of evidence

Comments

Health services utilisation

 

‐ In one study (cluster‐RCT), there were differences in 2 of 8 outcomes measured (an absolute increase of 21% and 19% in use of public facilities and uptake of vitamin A)

‐ In the CBA study there was an increase of 20.8% in the number deliveries attended by health personnel. However, there was no effect on the duration of hospital stay or bed occupancy.

‐ The third study showed an immediate increase of more than 130% (+ 144% on daily visits, + 135% on monthly visits), but the effect faded with time.

(3 studies)

Very low

2 observational studies (CBA and ITS)  with high risk of bias, and one cluster‐RCT with high risk of bias

 

Healthcare expenditure

 

Reduction in household health expenditures, although the size of the effect was difficult to estimate (decrease in the range of US$15 to $56 in annualised individual curative care spending)

Low

1 RCT with a high risk of bias

Health outcomes

 

The probability of individuals reporting that they had been sick in the past month was reduced. There was also a decrease in the incidence of diarrhoea in infants. 

Low

1 RCT with a high risk of bias

CBA = controlled before and after study
ITS = interrupted time series
RCT = randomised controlled trial

Figuras y tablas -
Ir a la tabla de la revisión
Table 1. Assessment of risk of bias in included studies

Controlled before and after (CBA) studies

Study ID

Baseline measurement

Baseline characteristics (equivalent control site)

Protection against exclusion or selection bias

Protection against contamination

Reliability of  outcome measures

Appropriate analysis

Overall risk of bias

Notes

Lavadenz 2001

DONE

NOT DONE

NOT DONE

NOT CLEAR

NOT CLEAR

NOT DONE

High risk of bias

No statistical analysis to test the robustness of results; control and intervention sites very different; intervention and control districts are very close, so contamination may have occurred; only facility‐based outcomes

Randomised controlled trial

Study ID

Concealment of allocation

Protection against exclusion bias

Sampling

Appropriate analysis (clustering)

Quality/reliability of the data

Protection against detection bias

Baseline measurement

Protection against contamination

Overall  risk of bias

Notes

Bloom 2006

NOT DONE

DONE

NOT DONE

DONE

DONE

NOT CLEAR

DONE

NOT CLEAR

High risk of bias

Despite major flaws in the study design (problems in the allocation of the contracts, limited number of clusters) this study tries to compensate for many outliers with a meticulous analysis

Interrupted time series

Study ID

Protection against changes

Appropriate analysis

No selection bias in the sample framing

Quality of outcome data

Number of points specified

Intervention effect specified

Detection bias

Overall risk of bias

Notes

Ali 2005

NOT DONE

DONE(re‐analysis)

NOT DONE

NOT CLEAR

DONE

DONE

DONE

High risk of bias

No information provided on concurrent events, nor on the number and type of facilities from which the figures come

Figuras y tablas -
Table 1. Assessment of risk of bias in included studies
Ir a la tabla de la revisión
Table 2. Treatment effects on health service outcomes specified as targets in the contracts

% Fully immunised children aged 12 to 23 months

% Children who received high‐dose vitamin A in the past 12 months

% Women who delivered in the past year and had at least 2 antenatal care visits

% Women who delivered in the prior year with a trained professional

% Women who delivered in the prior year in a facility (public or private)

% Women (with a live child aged 6 to 23 months old) who currently use a contraception method

% Women who gave birth in the prior 24 months and know about modern  contraception methods

% Population who choose a public sector facility when needing a curative care consultation

Baseline measure in all provinces (%)

34

43

9

24

3

13

22

4

Follow‐up measure in all provinces (%)

81

61

35

34

1

23

8

13

Treatment effect of contracting out (change in percentage points)

15

4.17***

26.3

‐12.3

7.4

‐3.8

7.3

28.9***

Notes: this table reports the results from Bloom 2006  regressions with province X year effects. Standard errors are corrected for clustering at the district level. Stars indicate significance under clustering: * at 10%; ** at 5%; *** at 1%.

Figuras y tablas -
Table 2. Treatment effects on health service outcomes specified as targets in the contracts
Ir a la tabla de la revisión
Table 3. Treatment effects on health service outcomes not specified as targets in the contracts

Outcomes not specified as targets

Probability of receiving a treatment for diarrhoea when needed)

Average number of additional antenatal checks received by women (beyond 2)

Baseline measure in all provinces

0.89

0.65

Follow‐up measure in all provinces

1

3

Treatment effect of contracting out (relative change  in outcome measure, accounting for differences in baseline outcomes)

14.4*

0.578

Notes: this table reports the results from Bloom 2006 regressions with province X year effects. Standard errors are corrected for clustering at the district level. Stars indicate significance under clustering: * at 10%; ** at 5%; *** at 1%.

Figuras y tablas -
Table 3. Treatment effects on health service outcomes not specified as targets in the contracts
Ir a la tabla de la revisión
Table 4. Treatment effects (regression results) on final health outcomes

Probability of having reported ill during past month

Probability of having diarrhoea in the past month

Child < 1 alive

Baseline measure in all provinces

0.202

0.35

0.97

Follow‐up measure in all provinces

0.185

0.26

0.97

Treatment effect of contracting out (relative change in outcome measure ‐ accounting for differences in baseline outcomes)

‐0.145

‐0.252

‐0.043

Notes: this table reports the results from Bloom 2006 regressions with province X year effects. Standard errors are corrected for clustering at the district level. Stars indicate significance under clustering: * at 10%; ** at 5%; *** at 1%.

Figuras y tablas -
Table 4. Treatment effects (regression results) on final health outcomes
Ir a la tabla de la revisión
Table 5. Treatment effects on self‐reported care‐seeking behaviour

Proportion of individuals who consulted a health provider when ill (by type of healthcare provider)

(Conditional on consulting a provider), proportion of individuals who visited a type of healthcare provider

None

Unqualified provider

Qualified private provider

Qualified public provider

Unqualified provider

Qualified private provider

Qualified public provider

Average effect size (qualified provider)

Baseline measure in all provinces (%)

82.4

8.5

9.5

0.9

48

44

6

 — 

Follow‐up measure in all provinces (%)

82.6

8.1

8

3.1

44

41

16

 — 

Treatment effect of contracting out (change in percentage points)

11.8

‐9.8

‐9.0*

5.4***

‐20.5

‐10.1

27.9***

70.1***

Notes: this table reports the results from Bloom 2006 regressions with province X year effects. Standard errors are corrected for clustering at the district level. Stars indicate significance under clustering: * at 10%; ** at 5%; *** at 1%.

Figuras y tablas -
Table 5. Treatment effects on self‐reported care‐seeking behaviour
Ir a la tabla de la revisión
Table 6. Treatment effects on health care spending (annualised individual curative care spending)

All individuals

All individuals excluding 0.5% tails#

All individuals excluding those who spent > US$100 last month#

Baseline average healthcare spending in all provinces

$18.76

$10.42

$15.17

Follow‐up average healthcare spending in all provinces

$12.12

$7.51

$9.84

Treatment effect of contracting out (relative change in outcome measure ‐ accounting for differences in baseline outcomes)

‐55.856***

‐22.122**

‐15.608*

Notes: this table reports the results from Bloom 2006 regressions with province X year effects. Standard errors are corrected for clustering at the district level. Stars indicate significance under clustering: * at 10%; ** at 5%; *** at 1%.

# Bloom et al. argue that due to the non‐normality of the distribution of self‐reported healthcare expenditures as well as extreme cases within this portion of the population, average expenditures would be skewed by the upper tail and therefore exclude it.

Figuras y tablas -
Table 6. Treatment effects on health care spending (annualised individual curative care spending)
Ir a la tabla de la revisión
Table 7. Description of contracts

Study ID

Country

Contractor

Contractee

Nature of the contract

Measures taken by the NGO

Ali 2005

Pakistan

The district government of Rahimyar Khan, the provincial health department

National NGO called PRSP (Punjab Rural Support Program)

Transfer of the control, management and use of buildings, furniture and equipment of 104 Basic Health Units (BHUs) in the district of Rahimyar Khan and transfer of all relevant budgetary provision (salaries including for unfilled posts, medicine, maintenance, building repairs, equipment, etc.) to be able to provide all relevant curative and preventive primary health services

The contract was signed for 5 years

Division of all BHUs into 35 clusters of 3, where 1 doctor is appointed as a team leader

Facilitation of Medical Officers’ mobility through an interest‐free car loan

Increase of doctors’ salaries (nearly threefold) and prohibition on private practice

Bloom 2006

Cambodia

The Ministry of Health of Cambodia

International NGOs ‐ different for each district

Transfer of the control, management and use of all means in district hospitals, sub‐district health centres and health posts to provide all promotive, preventive and curative services (known as the Minimum Package of activities) in the districts

Performance was measured and poor performance could result in sanction and non‐renewal of the contract. 8 performance targets were identified, essentially for child and maternal health

Increase of salaries (significant increase, but unclear how much)

Supervision by expatriate staff (between 0.5 and 3 per district)

Ban on private practice (for 1 of the 2 NGOs)

Firing non‐co‐operative personnel (between 1 and 3)

 

Lavadenz 2001

Bolivia

Federal Ministry of Health and Social Provision, Departmental Government and El Alto Municipality

NGO

Contracting out of the management of the hospital in the first instance (Aug 1999 to Jan 2000) and then of the whole district

Not mentioned

NGO = non‐governmental organisation
Figuras y tablas -
Table 7. Description of contracts
Ir a la tabla de la revisión
Table 8. Outcome measures

Study ID/ Intervention

Health outcomes

Health services utilisation outcomes

Other

Bloom 2006

Self‐reported illness in the past month

Incidence of diarrhoea among young children

Uptake of vitamin A for children, antenatal care, deliveries in a health facility, number of assisted deliveries, use of contraception, reported use of public facilities when sick

Household health expenditures

 

Ali 2005

None

Number of visits at the facility (monthly)

— 

Lavadenz 2001

None

Number of deliveries, bed occupancy rate, average stay duration

 

— 
Figuras y tablas -
Table 8. Outcome measures
Ir a la tabla de la revisión
Table 9. Results based on data from Lavandez 2001

Deliveries

Bed occupancy rate

Average stay duration

 

Intervention district

Comparison district

Relative effect

Intervention district

Comparison district

Relative effect

Intervention district

Comparison district

Relative effect

Effect of contracting out the hospital + insurance scheme (phase 2 ‐ phase1): % change

24.0%

14.5%

9.4%

‐5.4%

16.9%

‐22.3%

‐12.5%

3.7%

‐16.2%

Effect of contracting out the rest of the network (phase 3 ‐ phase 2): % change

13.8%

5.0%

8.8%

24.3%

1.3%

23.0%

4.8%

‐3.6%

8.3%

Effect of insurance + contracting out of hospital and health centres (phase 3 ‐ phase 1): % change

41.1%

20.2%

20.8%

17.6%

18.5%

‐0.9%

‐8.3%

0.0%

‐8.3%

Figuras y tablas -
Table 9. Results based on data from Lavandez 2001
Ir a la tabla de la revisión
Table 10. Results of the re‐analysis of data from Ali 2005

Outcome

Immediate impact (% change compared to month before the intervention)

Impact after 6 months (% change compared to month before the intervention)

Impact after 12 months (% change compared to month before the intervention)

Impact after 18 months (% change compared to month before the intervention)

Number of daily outpatient visits

135%

 

109%

 

84%

 

59%

 

Number of monthly outpatient visits

144%

 

120%

 

97%

 

73%

 
Figuras y tablas -
Table 10. Results of the re‐analysis of data from Ali 2005
Ir a la tabla de la revisión