Study characteristics
Thirty-one studies fulfilled the inclusion criteria, of which 14 reported on the costs associated with critical care in Tanzania (Table 2). The disease areas covered by eligible studies included; oncology (n=1), trauma (n=4), obstetrics (n=3), neonatal/child care (n=2), HIV (n=1), renal failure (n=1), TB (n=1), malaria (n=3), stroke (n=1) and typhoid (n=1), antimicrobial resistance (n=1) surgery (n=2) respiratory illness (n=1), blood transfusion (n=1), no disease or area specific (n=8). The majority of studies were cross sectional in design (n=20) and qualitatively reported on the location of critical care and critical care services available.
Most studies were conducted in a specific region: the Coast Zone (n = 7), Lake Zone (n = 3), Northern Highland Zone (n = 3) and Southern Highland Zone (n = 3) and Central Zone (n = 1). A zone is a geographical area made up of health regions which are made up of districts. Seven studies were nationally representative (n = 7); two studies didn’t specify a given region. Three studies were carried out in rural settings while the rest were urban or mixed settings. Nine studies specified that they were conducted in in public (government owned) facilities and six specified non-governmental organisations including faith/missionary based hospitals.
Critical care provision in Tanzania
The structure of the health system in Tanzania comprises six levels26 starting with the village health post (lowest level), dispensaries, health centres, district hospitals, regional hospitals and tertiary referral hospitals (the highest level) 26–39. Critical care is usually offered at the hospital level only, that is the district hospital, regional hospitals and referral hospitals 26–31,33–40. In addition, Nicks et al state that district hospitals offer basic forms of critical care while referral hospitals offer critical care in ICUs26. Thirteen of the fifteen studies that reported on critical care provision were carried out in urban settings. Two of the fifteen studies were carried out in both rural and urban settings; no study was exclusively carried out in a rural setting. Based on the literature, we further categorise critical care into “ICU delivered critical care” (otherwise known as ICU based critical care) and “non-ICU based critical care”.
Seven studies reported on ICU delivered critical care offered in tertiary referral hospitals in Tanzania28,29,35,36,38,39 including regional hospitals. For example, Murthy et al state that St. Francis Hospital in Ifakara region had 10 ICU beds and Sekou Toure Regional Referral Hospital in Mwanza had 8 ICU beds33. ICU delivered critical care services included intubation and ventilation. For example, one study found that 54.2% of all ICU patients were intubated or ventilated38. In addition, two studies reported surgical interventions that were conducted on patients cared for in ICU included wound debridement, treatment of fractures, craniotomy, underwater seal drainage, tracheostomy, eye surgery and Thoracotomy38,41. The district hospitals, for which studies were available did not provide ICU delivered critical care 34.
Critical care outside the ICU was reported to be delivered either in operating theatres or in emergency units and general wards. For example, operating theatres at Kilimanjaro Christian Medical Centre (KCMC) offered critical care for cases like trauma and sepsis42. Theatres in lower-level hospitals like district hospitals often offer initial critical care before referring the patient to higher level hospitals that can offer ICU based critical care. Although district hospitals and some regional hospitals were reported not to have ICUs, these facilities were reported to offer critical care in the general wards34. Some hospitals without ICUs designated areas within general wards to cater for the critically ill patients34. Furthermore, Staton et al found that some critical care services including intubation were performed in the emergency units39.
Availability of resources required for critical care
The majority of eligible studies reported on the known resources used for critical care (n=21). However, many studies did not report the extent to which the resources were readily available at a given hospital (n=18). Three studies reported on availability and use of critical care resources and indicate that their availability varies between hospitals and is sometimes unreliable or inconsistent (supplementary 3). Two studies find that there is a scarcity of trained staff in critical care34,37. Further, the availability of critical care equipment was heterogenous with scarcity in the areas of mask and tubing, pulse oximetry, oxygen cylinders, adult and paediatric oropharyngeal airway and PPE eye protection highlighted. In addition, a greater scarcity in paediatric than adult critical care equipment was identified 34,37.
Resource use and cost of resources associated with provision of critical care
Of the 31 studies, 14 reported on the resource use and critical care costs and were appraised for quality using the Adawiyah et al adapted appraisal checklist25 (Supplementary 4). All 14 studies were found to have clear research questions, mentioned the specific costing perspective, included relevant inputs, clearly stated methods for quantifying resources and reported costs. Four studies did not specify the time horizon while four studies partially addressed it. The majority of the studies (n=8) carried out some form of sensitivity analysis. All studies that collected primary data stated sample size but did not mention how it was determined and whether it was sufficient.
Published costs of critical care were reported as economic costs (n = 8) or financial costs (n = 6), using a mix of bottom up (n = 9) and top down (n = 5) approaches. Sources of the costs were local data collected from hospital records, surveys and price lists. Surveys (including key informant interviews) were a common method of data collection (n = 9). Three studies used patient level data on resource use to inform costs (Kimaro43; Sicuri44; Riewpaiboon45). Hospital or laboratory records were used in 6 of the 14 studies. Some studies used both survey and records review as the source of data (n = 6). Two studies used international literature as a source of data (n =2). Unit costs reported included cost per program, cost per patient, cost per visit, cost per patient day, cost per episode, cost per bed day, inpatient cost, total annual cost, cost per case (see table 4).
Although many studies reported costs within differing disease specific (e.g. malaria and anaemia) or general hospital services (e.g. inpatient and outpatient care), no study identified the cost of elements that are specific for the treatment of critical illness. The highest reported average cost per patient was for a patient undergoing caesarean section (USD 471.28)46 . The lowest reported cost per patient was for hospitalisation for stroke in a rural area (USD 12.76)47. A nationally representative study of general hospital services including critical care found that unit costs vary by provider ownership (public, NGO and private) and level of health facility48 (regional vs primary care). Services provided at private hospitals were more expensive than the same services received at public hospitals. NGO hospitals were less expensive than public hospitals.
When looking at cost categories, 2 studies reported costs of equipment, 2 studies reported costs of consumables and 6 studies reported medication cost. Within studies, we found per item cost for equipment (4 items), consumables (3 items) and medication (13 items). The cost of a mannequin used in training was the most expensive piece of equipment reported (USD 76.03)49. In the medication category, tranexamic acid was the most expensive per dose (USD 9.74)50, whereas antimicrobials were the most expensive on a per case basis (USD 30.36)51.
While diagnostics are not part of critical care provision, they form part of the overall patient care and can inform future economic evaluations. Six studies reported on the cost of diagnostics, for example, FBC, serum cryptococcal meningitis test, sputum culture and imaging (see supplementary material).
Three studies reported human resource costs (see table 6). No human resource costs were identified as specifically for ICU or non-ICU critical care. Reporting of human resource cost was heterogenous. One study provided monthly salaries for all personnel with monthly salaries ranging from USD 2721.79 for medical specialists USD 313.34 for nurses52. Another study reported total cost of personnel as a line item. The third study reported unit costs (i.e. cost per patient) treatment for all obstetric complications broken down by staff category. This study found that the most important cost was for an enrolled nurse or midwife (USD 43.62 per patient) and that the lowest unit personnel cost was for an anaesthesiologist (less than 0.01USD per patient)53.