Table 5.

A summary table of the included studies.

Author	Design	Sample	Result
Abdelwahid, Al-Shahrani, Elsaba, and Elmorshedi, 2010	Cross-sectional study	N = 452 (male = 183, female = 269)	This is a cross-sectional study conducted in 2010 in Sharurah by Abdelwahid, Al-Shahrani, Elsaba, and Elmorshedi. The main goal of this study was to assess the referral pattern and identify the appropriateness of the referral letter and consultant‘s feedback. Results indicated that there was lack of important information related to patients' medical history, physical examination, and investigation in the referral letters. Overall, there was a high percentage of referrals; however, the quality of referral letters and feedback reports was poor and needs to be improved.
Al Wadaani and Balaha, 2012	Retrospective study	N = 200	This is a retrospective study and conducted in 2012 in Al Hufuf city. The main objective of this study is to assess the appropriateness of physician responses in medical consultation reports and compare physician responses when using these reports from different levels of health care providers. Results indicated that almost half of referral is not safe for patients especially forms filled out by consultant and resident's physician. The form was semi structured and the authors recommended using structured letter [15]. This study may have information bias, because there were poor documented files as well as all variables available.
Alahmadi and Roland, 2005	A review	Not available	This is a review article that shows an overview of quality of primary care in Saudi Arabia. It was conducted by Al-Ahmadi and Roland in 2005. Overall, there was poor access for chronic disease management [12]. Also, the referral letters almost did not contain important medical information. Furthermore, they were handwritten and sometimes hard to read. There was low percentage of feedback sent from hospitals counting for only 22–39% of patients. As well, feedback reports lacked essential information including details of the advice given (100%), diagnoses (15%), or findings on investigations (21%).
Al-Alfi, Al-Saigul, Saleh, Surour, and Riyadh, 2004	Retrospective study	N = 4628	This retrospective study conducted in 2004 in Qassim region by Al-Alfi, Al-Saigul, Saleh, Surour, and Riyadh [20]. The main purpose of this study was to assess the quality of diabetic care of primary care in a rural town called Al Asyah. Results of this study indicated that there was high percentage of referrals, but the quality of referrals was inadequate. Therefore, it recommended that better coordination between primary and secondary care should be available for patients with type 2 diabetes as patients need services from both levels of care.
Al-Saigul, Abed-Elbast, Sourour, Ramzy, and Al-Alfi, 2007	Cross-sectional	N = 330	This cross-sectional study was conducted by Al-Alfi, Al-Saigul, Abed-Elbast, Sourour, and Ramzy in Buraidah in 2007 [22]. The study's main objective was to evaluate the quality of referral letters and feedback reports written according to the standards of Quality Assurance Manual of Ministry of Health from primary health care centers (PHCCs). Results indicated that there was lack of important medical information in referral letters. However, referral letters were better than feedback reports. The referral rate was very low for only 4%. The quality of referral letters and feedback was poor, and therefore, there should be improvements to the referral system.
Albattal, 2014	Cross-sectional	N = 51	This cross-sectional study was conducted in Riyadh in 2013 by Albattal [11]. It aims to identify the factors that contribute to inappropriate referrals and to suggest a better way to improve the quality of referrals. Therefore, 51 physicians were included in this study to give opinions about inappropriate referral. The sample size was very small in this type of study which increases a chance to be biased. Results indicated that one of the main factors to inappropriate referral was poor GP awareness about secondary care clinics. Another important factor was requesting referral by patient leading to inappropriate referral.
Al-Kaabba et al., 2010	Cross-sectional	N = 14138	This cross-sectional study conducted by Al-Kaabba, Abdalla, Saeed, AlZalabani, and Ahmad Mustafa in 2010 [18]. The main objective was to determine the referral pattern and characteristics of referred patients visiting military family medicine clinics in Tabuk, Saudi Arabia. Results indicated that there was a low percentage of feedback from consultants counting for only 13%. Also, females receive more referrals than male. There was no sample bias in this study because the sample size was big (14138). Overall, the quality of this study is good. Although this study was conducted in the military health sector, the results are applicable for general public health sector.
Ahmed, 2007	Cross-sectional	N = 430	This cross-sectional study was conducted in Gurayat in 2007 by Almoutaz [21]. The main objective of this study was to evaluate and compare the referral forms sent by primary care to diabetic clinics in secondary care with that of the American Diabetes Association. Results of this study indicated that the referral forms used were poor, as they did not reflect a clear picture of the referred patient. The author recommended that referral letters should be special and structured for patients with type 2 diabetes. The sample size was large which decrease the chance to be biased.
Al-Qahtani and Imtiaz, 2004	Retrospective	N = 138484	This retrospective study was conducted in 2004 by Al-Qahtani and Imtiaz [14]. The main objective was to analyse the pattern of referrals and evaluate the effect of clinical audit on the number and type of referrals from the primary care physicians to specialists. From the results, it seems that the authors focused more on the quantity of referrals but not the quality as that was mentioned by authors to decrease the cost. Also, the referral process was based on papers. Likewise, guidelines of referrals were not mentioned if disseminated to primary and secondary level; however, the authors mentioned that there was a decrease in the referral letter for two reasons. One of those reasons was increased awareness of physicians that irrelevant referrals will lead to increased workload.
Leena Baghdadi, 2007	Cross-sectional	49	This study was conducted by Leena and Rabab Baghdadi in the Holy Makkah city in 2007 [19]. The main purpose of this study is to evaluate the referral system from primary health care centers to hospitals and to improve the quality of the referral system and communication between the primary health care and the hospitals. This study has many biases. First of all, it is a cross-sectional study and this type of study should have a big sample. However, this study has only 49 participants. Secondly, the authors did not show the validity of the designed questionnaires. Finally, it has not addressed limitation. Overall, the quality of this paper is weak. However, results indicated that the referral system was poor and needed to be improved.
Khawaja et al., 2009	Retrospective-cross-sectional	N = 4616	This retrospective cross-sectional study was conducted in 2009 in Riyadh by Khawaja et al. [13]. The main objectives were to assess the referral rate of King Khalid University Hospital employees, from the employee's health clinic to specialty care, and to compare the rate of referral among both sexes of Saudi nationals versus expatriates. Results of this study indicated that most frequent reasons for referral were chronic problems (diabetes mellitus, hypertension, and bronchial asthma). Since, the main objective of this study was to decrease the rate of referrals, none of the 12 major referral factors were addressed.
Al-Ghamdi, Al-Turki, Al-Baghli, and El-Zubaier, 2007	Cross-sectional	N = 197681	This cross-sectional study was conducted in Eastern Province in 2007 by Al-Ghamdi, Al-Turki, Al-Baghli, and El-Zubaier [16]. The main objective of this study was to describe a community-based diabetes and hypertension screening campaign, the percentage of screened positive individuals, identified participation rate, and the factors affecting the participation. It includes urban and rural areas in Saudi Arabia. Results indicated that referral was higher from the rural areas than urban. Also, females have more successful referral than males. Primary care has the highest percentage of successful referral followed by other government hospitals. However, the MOH hospitals report very low successful referral.