Results of the search
Our initial search yielded 5020 studies, which was reduced to 5004 after deduplication. After title and abstract screening, 758 studies remained. 271 studies were found to be irrelevant. Reasons for exclusions are found in Fig. 1. Through hand searching and citation searching we identified 18 models from 16 studies. After extraction 219 models from 188 studies remained (Fig. 1, see Appendix B-F for included studies).
Settings
Of the 188 overall studies, 29 (15%) were conducted in low-income countries (LICs), 67 (36%) were conducted in lower-middle-income countries (L-MICs), 88 (47%) were conducted in upper-middle-income countries (UMICs), and four (2%) were conducted in multiple countries including a mix of income levels. Ninety (48%) were conducted in sub-Saharan Africa, 34 (18%) in East Asia and the Pacific, 27 (14%) in South Asia, eight (4%) in Europe and Central Asia, four (2%) in the Middle East and North Africa, and 25 (13%) in Latin America and the Caribbean. Among studies that specified where services were delivered, 46 (24%) were conducted in rural settings, 70 (37%) in urban settings, four (2%) in peri-urban or suburban areas, and 31 (16%) in mixed regions. Most studies were conducted at single centers (31%) or on the small to medium scale (47%). The most commonly identified institutions were public (112 studies, 60%) and public with outside support (40 studies, 21%). Other models included private (three studies, 2%) nongovernmental organizations (eight studies, 4%), and faith-based organization (two studies, 1%) (Table 2). Nearly all studies found by this search outlined outpatient services.
Table 2
Number and percentage of included studies by income level, region, delivery area, scale, institution, research role in delivery model, and condition category
Dimension and category
|
Number of studies (N = 188)
n (%)
|
Income group
|
n (%)
|
Low-income country (LIC)
|
29 (16)
|
Lower middle-income country (L-MIC)
|
67 (34)
|
Upper middle-income country (UMIC)
|
88 (48)
|
Combined
|
4 (2)
|
Region
|
n (%)
|
Sub-Saharan Africa
|
90 (47)
|
North Africa & Middle East
|
4 (2)
|
Europe & Central Asia
|
8 (5)
|
South Asia
|
27 (16)
|
East Asia & the Pacific
|
34 (16)
|
Latin America & Caribbean
|
25 (14)
|
Delivery area
|
n (%)
|
Rural
|
46 (25)
|
Urban
|
70 (35)
|
Peri-urban
|
4 (2)
|
Mixed
|
31 (18)
|
Not specified
|
37 (19)
|
Scale
|
N %
|
Single center
|
58(30)
|
Small to medium scale
|
88 (46)
|
Large scale
|
31 (18)
|
National
|
8 (5)
|
Multi-country
|
3(2)
|
Institution
|
n (%)
|
Public
|
112 (62)
|
Public with outside support
|
40(21)
|
Private
|
3 (2)
|
Non-governmental organizations
|
8(3)
|
Faith based organization
|
2 (2)
|
Not specified
|
23 (11)
|
Was the delivery model part of routine care or a research study?
|
n (%)
|
Theoretical research protocols (not yet implemented)
|
17 (10)
|
Pilot and feasibility studies
|
57 (27)
|
Experimental studies
|
30 (17)
|
Imbedded or evaluation
|
84 (47)
|
Condition category
|
n (%)
|
Common chronic NCDs
|
118 (64)
|
Severe chronic NCDs
|
42 (21)
|
Common neuropsychiatric
|
53 (29)
|
Severe neuropsychiatric
|
15 (9)
|
Chronic infection
|
49 (24)
|
Acute infection
|
11 (6)
|
Maternal and child health
|
23 (13)
|
Conditions addressed through “primary health care” not otherwise specified
|
27 (15)
|
Sense organ
|
1 (1)
|
The majority of studies reported being experimental in nature, including untested protocols (17, 9%), pilot studies and feasibility studies (57, 30%), or experimental studies (30, 16%). Eighty-four were considered to be imbedded into the health system, or evaluations of current programs (45%) (Table 2).
Over time, publication of integrated models increased. Our study identified only 16 studies from the ten-year periods between 2000 to 2009 (representing less than three per 1,000,000 studies indexed by Medline during that time), 49 studies from 2010 to 2015 (representing almost 11 per 1,000,000 studies indexed by Medline during that time), 121 studies from 2016 to 2020 (representing almost 27 per 1,000,000 studies indexed by Medline during that time) and two in 2021.
Health system level
Models were categorized by which level of the health system offered services. Thirty-five were mixed across multiple levels. Based on available information in the studies or other sources, we were able to separate 26 of these studies by the level of health system into multiple levels. Thus, one study that reported on a model in the community, health center and secondary facility, would be included in three models. This led to an additional 31 models being included, for a total of 219 across 188 studies. These included community (55, 25%), health center (primary health care) (93, 42%), secondary health facilities (31, 14%), and tertiary hospitals (30, 14%). Ten studies (5%) were in standalone specialty outpatient clinics, which were considered separately.
Conditions treated
Conditions named in the studies are summarized into the following condition types: common chronic NCDs, severe chronic NCDs, common neuropsychiatric (NP) conditions, severe NP disorders, maternal and child health (MCH), conditions reported in studies as being covered by “primary health care” (PHC), acute infectious diseases, chronic infectious diseases, and sense organ disorders. The specific conditions included within each of these condition types are described in Supplementary Table 1. Common NCDs were the most common, reported in 118 studies (63%), followed by common NP conditions reported in 53 (28%) studies (Table 2). Conditions included in models stratified by health system level are shown in Table 3.
The condition categories that most often were found to be integrated together were common NCDs integrated into chronic infection programs. In countries with a low HIV prevalence, however, there were only two studies showing models of integrated services for NCDs or NP and chronic infections. In studies conducted in these countries, models frequently showed services for NCDs or NPs being integrated into MCH or PHC services (23 studies, 25%). In mid-range HIV prevalence countries, NCDs and NPs (common and severe) were often integrated into MCH or PHC (16 studies, 53%), but also into chronic infections (12 models, 40%). In high HIV prevalence countries, 35 studies (67%) included NCDs and NP being integrated into chronic infections programs, but only 14 (27%) into MCH and PHC (Supplementary Table 2).
Studies in LICs tended to have a higher percentage of models including more severe conditions. Forty-one percent of studies conducted in LICs included severe NCDs, compared to 25% in L-MICs, and 13% in UMICs. Similarly, 10% of studies in LICs included severe NPs, compared to 7% in L-MICs and 7% in UMICs. The percentage of studies that included common NCDs and NPs was roughly equal across income levels (Supplementary Table 3).
Providers
Many types of primary providers were identified as shown in Table 1. We identified two types of studies that did not have a single primary provider: multi-cadre (different cadres within the same general discipline) and multidisciplinary teams (people from multiple disciplines, for example a mental health counselor and an NCD provider). At the community level, CHWs were the most common primary provider (47%). Mid-level providers were the most commonly reported primary provider in health centers (42%), secondary facilities (48%) and tertiary facilities (30%) (Table 4). Models utilizing multidisciplinary teams instead of a single primary provider were more likely to focus on a single condition such as diabetes or palliative care (Appendix B-F).
Table 4: Number and percentage of primary provider categories reported in study models stratified by health system level
Service types
We identified 15 unique service types across the studies. A complete list is found in Supplementary Table 1.
At all health system levels, health education was found in the majority of models. At the community level, most models reported linkage to higher levels of the health system (69%), health education (65%), and screening (62%). Health promotion (40%), adherence support (38%), and home visits (33%) were also common. At the health center level, in addition to the services found at the community level, initial diagnosis (48%), patient follow-up (54%), medication dispensing (59%) and some medication management (12%) became more common. There was less health promotion (6%), screening (45%), and home visits (11%). At secondary facilities, initial diagnosis (58%), medication dispensing (77%), patient follow-up (62%), and monitoring (52%) were all found in most studies, with a large proportion reporting medication management (42%) (Table 5).
Table 5: Number and percentage of service type categories reported in study models stratified by health system level
Decentralization and task shifting
Decentralization was commonly identified in this review, and was particularly at the community level, where services previously found at health care facilities were decentralized to the community (33%), and health centers (18%), where care was decentralized from hospitals. We also found decentralization in 23% of models at secondary facilities. Similarly, 22% of studies reported task shifting from care teams to lower cadre providers in the community, 37% at health centers, and 39% at secondary facilities (Supplementary Table 4).
Integration type
Among the studies reviewed here, 12% described existing models. The remainder documented changes to existing models including: new care teams providing integrated care (26%), task distribution within existing models (5%), new services integrated into existing models (16%), new conditions integrated into existing models (26%) and new services and conditions integrated into existing models (15%). In the community (31%) and tertiary level (30%), new care delivery teams were the most common model. At health centers (28%) and at the secondary level (35%) new conditions were most commonly integrated into existing models (Table 6).
Table 6: Number and percentage of integration types utilized in each model, stratified by health system level
Other domains
Only one-third of studies (33%) documented the cost of care. A majority of the studies took place in the public/government-run health system and are subject to national policies. Among the studies reporting cost of care, the majority specified that at least some services or medications are offered free of charge, while some reported out-of-pocket payments, including fee-for-service, co-pays, or consultation fees (Supplementary Table 5).
For domains unique to community-based studies, ten studies (18%) were in mobile clinics. The majority of studies (69%) did not report on compensation, but of those that did, nine (16%) reported CHWs as volunteers, and the rest reported compensation as a salary (11%) or as fee for service (4%). In nine (16%) studies, CHWs worked part time, and in seven (13%) they worked full time (Supplementary Table 6).