Physician associate/assistant contributions to cancer diagnosis in primary care: a rapid systematic review

Sheringham, Jessica; King, Angela; Plackett, Ruth; Khan, Anwar; Cornes, Michelle; Kassianos, Angelos P.

doi:10.1186/s12913-021-06667-y

BMC Health Services Research

Table 4 Findings from studies comparing PA performance with primary care physicians

From: Physician associate/assistant contributions to cancer diagnosis in primary care: a rapid systematic review

Study: analysis	Main finding: PA performance vs primary care physicians	Patient cohort seen by PAs vs primary care physicians	Covariates included in adjusted analyses
Diagnostic tests ordered
Drennan et al., 2015 [21]: Diagnostic tests (no specific ones specified)	↔ No significant difference Rate ratio 1.08 (0.89–1.30)	- Younger - From different geographical areas - Healthier/lower healthcare use	Age, acuity of presenting problem, sex, practice attendances in the previous 3 months, no. problems, chronic disease registers, socioeconomic deprivation
Hughes et al., 2015 [23]: Imaging	↑ Higher use Adjusted OR 1.34 (1.27–1.42) ^a)	- Younger - higher % female - higher % of white ethnicity - Healthier - From different geographical areas	Patient age group, sex, race, state, urban, comorbidity.
Kurtzman et al., 2017 [26]: Imaging	↔ No significant difference Adjusted OR 1.14 (0.84–1.54)	Similar (age, gender, ethnicity, payer source) to PCPs.	Age, sex, race, ethnicity, payer, metro status, region, reason for visit, health centre type, education, year.
Mafi et al., 2016 [29] - Radiography (in ‘low value’ cases) ^b)	↔ No significant difference 10.2 in PCP vs 11.4 in PAs (alone), p = 0.71 and 9.5% in PAs (shared) p = 0.75	- Younger - From different geographical areas	Patient age, sex, race or ethnicity, comorbidity, symptom acuity, insurance status, urban location, geographic region, year
- CT or MRI (in ‘low value’ cases)	↔ No significant difference 6.0 in PCP vs 9.9 in PAs (alone), p = 0.3 and 6.8% in PAs (shared) p = 0.69	- Younger - From different geographical areas
Referrals to other physicians
Drennan et al., 2015 [21]	↔ No significant difference Rate ratio 0.95 9 (0.63–1.43) p = 0.80	- Younger - From different geographical areas - Healthier/lower healthcare use	Age, acuity of presenting problem, sex, # practice attendances in the previous 3 months, # problems, # chronic disease registers, socioeconomic deprivation
Kurtzman et al., 2017 [26]	↔ No significant difference Adjusted OR 1.17 (0.87–1.56)	PAs saw similar patient profile (age, gender, ethnicity, payer source) to PCPs.	Age, sex, race, ethnicity, payer, metro status, region, reason for visit, health centre type, education, year.
Mafi et al., 2016 [29]: Situations in which referral considered to be low value	↔ No significant difference 8.2 in PCP vs 5.9 in PAs (alone), p = 0.52 and 8.6% in PAs (shared) p = 0.86	- Younger - From different geographical areas	Patient age, sex, race or ethnicity, comorbidity, symptom acuity, insurance status, urban location, geographic region, and year
Screening
Tang et al., 2016 [31]: PSA screening rates for patients with limited life expectancy	↔ No significant difference c) Screening offered in 41.3% of cases by PAs vs 41.5% by PCPs	Not reported	Patient age, race, marital status, income, education, clinician clustering
Outcomes
Brock et al., 2017 [19]: Malpractice reports per 1000 clinicians ^d): diagnosis related claims comprise diagnosis failure or delay in diagnosis	↓ Lower payments Physician median payments ranged from 1.3 to 2.3 times higher than PAs or NPs	No data but differences in breadth of patient acuity proposed as possible explanation for findings.	n/a
Drennan et al., 2015 [21] - Re-consultation within 14 days for the same or a linked problem - Patient Satisfaction ^e)	↔ No significant difference Adjusted rate ratio 1.24 (0.86–1.79), p = 0.25	- Younger - From different geographical areas - Healthier/lower healthcare use	Age, acuity of presenting problem, sex, # practice attendances in the previous 3 months, # problems, # chronic disease registers, socioeconomic deprivation
	↔ No significant difference Adjusted rate ratio 1.00 (0.42–2.36), p = 0.99
Kurtzman et al., 2017 [26]: Re-consultation	↔ No significant difference Adjusted odds ratio 0.77(.52–1.13)	PAs saw similar patient profile (age, gender, ethnicity, payer source) to PCPs.	Age, sex, race, ethnicity, payer, metro status, region, reason for visit, health center type, education, year.

a. Hughes:. In the main analysis, nurse practitioner and PA data were aggregated as APC. In sensitivity analyses: NPs ordered less imaging than PAs (OR, 0.59 [0.53–0.66]); APCs ordered less imaging than PCPs for acute respiratory tract infection (OR, 0.68 [0.51–0.90]); Differences were greater for radiography than non-radiography imaging
b. Mafi: Findings were presented for both hospital and office based primary care settings. 89.9% of the data reflected visits to clinicians in office-based physician practices (data from the NAMC), so these figures are presented. Disaggregated data from supplementary data are presented here. Alone reflects visits to PAs where they saw the patient without a physician; shared reflects consultations where a physician was alongside
c. Tang: Men whose clinician was a physician trainee had substantially lower PSA screening rates than those with an attending physician, nurse practitioner, or physician assistant
d. Brock: Diagnosis malpractice claims, while higher for physicians, comprise a greater proportion of PA than physician claims (53% vs 32%). This result may be partially explained by the presence of surgeons and anaesthesiologists in the physician group, or it may signal where PAs and NPs might be most at risk for error
e. Drennan: other findings comprised: consultation duration was longer for PAs than GPs but costs per consultation were lower

Back to article page

ISSN: 1472-6963

Contact us

General enquiries: journalsubmissions@springernature.com